Integrated Assessment of Agriculture and
Sustainable Development;
Setting the Agenda for Science and Policy
(AgSAP 2009)
10 – 12 March 2009
Hotel Zuiderduin, Egmond aan Zee
The Netherlands
PROCEEDINGS
AgSAP Conference 2009
Egmond aan Zee
The Netherlands
Editorial Committee of these Proceedings:
Martin van Ittersum Coordinator SEAMLESS, Wageningen University, Plant
Production Systems (NL)
Joost Wolf
Wageningen University, Plant Production Systems (NL)
Gon van Laar
Wageningen University, Centre for Crop Systems Analysis
(NL)
Local Organizing Committee
Martin van Ittersum Coordinator SEAMLESS, Wageningen University, Plant
Production Systems (NL)
Floor Brouwer
LEI, Wageningen UR (NL)
Joost Wolf
Wageningen University, Plant Production Systems (NL)
Theo Jetten
Wageningen University, C.T. de Wit Graduate School
PE&RC (NL)
Marcel Lubbers
Wageningen University, Plant Production Systems (NL)
Charlotte Schilt
Wageningen University, Plant Production Systems (NL)
Corresponding address:
AgSAP Office
Plant Production Systems
Wageningen University
P.O. Box 430, 6700 AK Wageningen
The Netherlands
Website:
www.conference-AgSAP.org
E-mail:
office.PP@wur.nl
Suggested citation:
Van Ittersum, M.K., J. Wolf & H.H. Van Laar (Eds), 2009. Proceedings of the Conference on
Integrated Assessment of Agriculture and Sustainable Development: Setting the Agenda
for Science and Policy (AgSAP 2009). Egmond aan Zee, The Netherlands, 10-12 March
2009. Wageningen University and Research Centre, Wageningen, 560 pp.
Printing: GVO, Ede, The Netherlands
Publisher: Wageningen University and Research Centre, The Netherlands
ISBN: 978-90-8585-401-2
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Contents
Preface
xvii
Plenary Sessions
1
Von Braun, J.: Agriculture and sustainable development in developing countries ................... 2
Legg, W.J.: Agriculture and sustainable development in OECD countries: A policy
perspective............................................................................................................................. 4
Van Ittersum, M.K., F. Ewert, E. Andersen, F. Brouwer, T. Heckelei, J. Alkan Olsson, J.
Wery, J.J.F. Wien: Integrated assessment of agricultural systems: On integrated science
and science integration ......................................................................................................... 6
Rotmans, J.: Two decades of Integrated Assessment tools: A promising way forward ............ 8
Pingali, P.: Enhancing developing world agricultural performance: Getting beyond the current
plateau.................................................................................................................................. 11
Theme A: Methodology for integrated assessment
Session A1: Methods for integrated assessment of agricultural systems and land use
13
Helming, K., I. De la Flor Tejero, M.K. Van Ittersum, P. Nowicki, W. Rienks, A. Volkery:
Scenario studies for agriculture, land use and landscape in Europe: A comparison........... 14
Arriaza, M., O. Nekhay: Integrating social preferences and spatial analysis to optimize the use
of marginal agricultural lands.............................................................................................. 16
Berkhoff, K., S. Herrmann: Understanding land use change with the aid of an integrated
modelling framework: A case study on rural south-west China ......................................... 18
Christen, O.: Concept and application of a DLG sustainability standard on the farm level .... 20
Herrero, M., P.K. Thornton, A. Notenbaert, S. Msangi, S. Wood, R. Kruska, J. Dixon, D.
Bossio, J. Van de Steeg, H.A. Freeman, B. Gerard: An integrated assessment of global
changes in crop-livestock systems to 2030: Implications for policy development ............. 22
Jansson, T., M. Adenäuer, M. Kuiper: The CAPRI and GTAP link: Linking a partial and a
global equilibrium model .................................................................................................... 24
Lippe, M., T. Thai Minh, A. Neef, V. Hoffmann, G. Cadisch, T. Hilger: Integration of
interdisciplinary participative research tools: New vistas for land use modelling? ............ 26
Marohn, C., U. Schuler, M. Lippe, Y.Z. Ayanu, C. Hugenschmidt, T. Hilger, G. Cadisch:
Modelling resource distribution in mountainous watersheds in Thailand: An integrated
approach .............................................................................................................................. 28
McRoberts, N., C.F.E. Topp: Construction of a multi-attribute model of arable system
sustainability by elicitation of expert opinion ..................................................................... 30
McRoberts, N., C.F.E. Topp, C. Hawes, R.L. Walker, C.A. Watson, P. Hallett, T. Daniell, M.
Young, G.R. Squire: Gathering baseline data for studying arable system resilience in
Scotland by integrated farm surveys ................................................................................... 32
Meyer, B., R. Grabaum, D. Watson, D. Adams, A. Hood, A. Phillips: Integrating multiple
functions to optimize rural land health ................................................................................ 34
iii
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Santos, R., P. Antunes, P. Beça: Framework for the socio-economic performance assessment
of irrigation schemes ........................................................................................................... 36
Spoelstra, S.F., B. Elzen: Sustainable livestock production: Assessment of farmers
initiatives ............................................................................................................................. 38
Tittonell, P., M.C. Rufino, M.T. Van Wijk, M. Herrero, M. Corbeels, C. Pacini, J. Smith, S.
Karanja, C. Quiroz, P. Mapfumo, J. K’ungu, F. Baijukya, C. Kaizzi, M. Mwale, O.M.
Sanogo, E. Yeboah, D. Nwage, B. Vanlauwe, A. Albrecht, N. De Ridder, K.E. Giller:
Integrated assessment of nutrient management options in smallholder farming systems:
The NUANCES framework ................................................................................................ 40
Van Delden, H., M.L. Paracchini, T. Stuczynski, P. Ciaian, R. Vanhout: The LUMOCAP
PSS, impact assessment of agricultural policies in an integrative context.......................... 42
Wustenberghs, H., E. Wauters, S. Van Passel: Measuring sustainability at a supra-farm level:
Evaluation of methods ......................................................................................................... 44
Session A2: Methods for up- and downscaling in integrated assessments
47
Antle, J.M.: Strategies for addressing spatial and temporal scale in Integrated Assessment of
agricultural systems ............................................................................................................. 48
Britz, W.: EU-wide spatial down-scaling of results of a regionalized agricultural economic
model to analyse environmental impacts ............................................................................ 50
Britz, W., M. Adenäuer: Calibration of regional programming models to exogenous supply
elasticities by a variable cost function: The CAPRI-EXPAMOD example ........................ 52
Elbersen, B.S., R. Jongeneel, M. Kempen, M. Mikk, A. Peepson: Compliance and costs of
compliance: A procedure to obtain estimates at a disaggregated level ............................... 54
Ewert, F., M.K. Van Ittersum, O. Therond, I. Bezlepkina, E. Andersen, T. Heckelei, G.
Flichman, N. Turpin: Complex scaling issues in integrated assessment modelling:
Approaches used and their integration into SEAMLESS-IF............................................... 56
Pérez Dominguez, I., I. Bezlepkina, A. Kanellopoulos, T. Heckelei, A. Oude Lansink, E.
Romstad: EXPAMOD – connecting the farm and market levels........................................ 58
Righi, E., G.C. Pacini, S. Dogliotti, V. Aguerre, W.A.H. Rossing: Farm typology
identification by multi-variate analysis as a method to scale-up results of integrated impact
assessment .......................................................................................................................... 60
Rufino, M.C., J. Dury, P. Tittonell, M.T. Van Wijk, S. Zingore , K.E. Giller: Competing use
of organic resources, climate variability and interactions at village scale in a communal
area of NE Zimbabwe.......................................................................................................... 62
Turpin, N., R. Laplana, D. Kopeva, L. Stapleton, K. Happe, G. Deffuant, F. Brouwer, M.
Burghard: PRIMA, Prototypical Policy Impacts on Multi-functional Activities in rural
municipalities ...................................................................................................................... 64
Turpin, N., O. Therond, F. Ewert, I. Bezlepkina, J. Alkan Olsson, C. Bockstaller, S. Janssen:
Spatial upscaling of on-farm indicators: Concepts to use a scientific knowledge base to
match policymakers’ requirements...................................................................................... 66
Valdivia, R.O., J.M. Antle, J.J. Stoorvogel: Coupling integrated assessment models to market
equilibrium models to analyse economic and agricultural-environment interactions across
different scales..................................................................................................................... 68
Van Bussel, L.G.J., F. Ewert, P.A. Leffelaar, H. Van Keulen: Assessing the effects of data
aggregation using a crop simulation model......................................................................... 70
Van de Giesen, N., T.J. Stomph: Scale effects in Hortonian surface runoff with management
implications ......................................................................................................................... 72
Van Os, J., B.S. Elbersen, J. Kros, W. De Vries, J.P. Lesschen, M. Kempen: Effect of spatial
resolution within the Cross Compliance Assessment Tool: The case of Nitrate Directive. 74
iv
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Zakayo, S.: Risk assessment methodology for pesticide control in the Nil River Basin,
Belgium ............................................................................................................................... 76
Zasada, I., A. Piorr, R. Berges: A generic downscaling approach for scenario driven
modelling outputs from European to regional scale............................................................ 78
Session A3: Integrated assessment of ecosystem services from agriculture
81
Buckwell, A.: Environmental services delivered and deliverable by land managers .............. 82
Arriaza, M., O. Nekhay, M.J. López-Baldovín: Restoration of abandoned agricultural lands
toward umbrella species habitats......................................................................................... 84
Blandford, D., I. Gaasland, E. Vårdal: Examining the degree of jointness between food and
cultural landscape production.............................................................................................. 86
De Blaeij, A., V. Linderhof, N. Polman: The monetary value of combining ecosystemservices in private artificial wetlands .................................................................................. 88
Dhillion, S.S.: Reconfiguration of ecosystem services in agricultural landscapes: Mitigation
and policy concerns for renewable energy development..................................................... 90
Dirkx, G.H.P.: ‘Green services’: Bridging the gap between consumers and producers of
landscape in The Netherlands? ............................................................................................ 92
Elbersen, B.S., R.G.H. Bunce, A. Van Doorn: Quantification of the relationship between
changes in farming and changes in biological diversity: The Atlantic case ....................... 94
Giannoccaro, G., M. Prosperi, J. Berbel: Integrated assessment of multi-functionality of
irrigated agriculture: An application for the Mediterranean regions................................... 96
Pérez-Soba, M., M.L.M. Jones, C. Pacini, M.L. Paracchini: Land Use Functions: An approach
for integrated assessment of the goods and services provided by agriculture in a multifunctional and sustainable context....................................................................................... 98
Pinto-Correia, T., S. Truchet, C. Machado, F. Barroso, P. Picchi, Y. Michelin, N. Turpin, J.P.
Bousset: How do policy options modify landscape amenities? An assessment approach
based on public expressed preferences ............................................................................. 100
Rodrigues-Filho, S., V.M. De Castro, F. Bakker, D. Lindoso, N. Debortoli, M. Bursztyn:
Regional sustainability contrasts in Brazil as indicated by the CompasSus: Compass of
Sustainability ..................................................................................................................... 102
Romstad, E., K. Louhichi, P. Zander: Economic modelling of multi-functionality at the farm
level ................................................................................................................................... 104
Rowe, E.C., S. Smart, C.D. Evans, B.A. Emmett: Predicting combined impacts of land
management and pollution on biodiversity: A dynamic niche occupancy approach ........ 106
Rudrum, D.P., R.A.M. Schrijver, A. Corporaal, T.J. De Koeijer: Cost effectiveness of
biodiversity in agricultural production systems ................................................................ 108
Stoorvogel, J.J., A. Temme, J.M. Antle, L. Claessens, J.M. Schoorl: A novel site-specific
methodology to assess the supply curve of environmental services ................................. 110
Tagliafierro, C., M. Antrop, V. Van Eetvelde, W.G. Hutchinson, A. Longo: Redefining
landscape economics: A proposal to integrate economics approaches and landscape
disciplines .......................................................................................................................... 112
Turpin, N., L. Stapleton, G. Garrod, E. Perret, F. Brouwer: Towards indicators of multifunctionality: Assessing jointness between commodity and non-commodity outputs...... 114
Van der Heide, C.M., F.J. Sijtsma, P. Van Egmond, S. Kruitwagen, A. Van Hinsberg, F.J.
Dietz: Valuation of ecosystem services and decision making: Combining a CBA with
MCA .................................................................................................................................. 116
Van der Werf, W., D.A. Landis, M.M. Gardiner, A.C. Costamagna, J.M. Baveco, F.J.J.A.
Bianchi, N.C. Schellhorn, W. Zhang: Analysing, forecasting and valuating the effects of
landscape change on the ecosystem service of biological pest control ............................. 118
v
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Villa, F., M. Ceroni, K. Bagstad, G.W. Johnson Jr., S. Krivov: Understanding agricultural
ecosystem services with ARIES (ARtificial Intelligence for Ecosystem Services):
Perspectives for assisted policy making ............................................................................ 120
Westhoek, H., A.G. Prins, H. Van Zeijts: Policy options for safeguarding the provision of
public goods and services in intensive and extensive farming systems ............................ 122
Willemen, L., L. Hein, P.H. Verburg: Translating spatial policy into future landscape
functions ............................................................................................................................ 124
Session A4: The organization of tools for impact assessment: Standards or diversity? 127
De Smedt, P.: Impact Assessment practice to support sustainable policy measures in Europe
........................................................................................................................................... 128
Ewert, F., M.K. Van Ittersum, I. Bezlepkina, O. Therond, E. Andersen, H. Belhouchette, C.
Bockstaller, F. Brouwer, T. Heckelei, S. Janssen, B. Jonsson, M.J.R. Knapen, M. Kuiper,
K. Louhichi, J. Alkan Olsson, L. Ruinelli, A.E. Rizzoli, N. Turpin, J. Wery, J.J.F. Wien, J.
Wolf: A methodology for enhanced flexibility of integrated assessment of policy impacts
in agriculture...................................................................................................................... 130
Fricke, K., K. Helming, S. Sieber: Choose complexity: How to bundle knowledge in
Integrated Assessment Tools? Comparing implicit knowledge integration in SENSOR and
PLUREL ............................................................................................................................ 132
Gaillard, G., T. Nemecek: Swiss Agricultural Life Cycle Assessment (SALCA): An
integrated environmental assessment concept for agriculture........................................... 134
Khanh Hoang, L., S. Van Griensven, J. Cools, L. Benedetti, A. Mynett: Linking agricultural
pollution to receiving waters using OpenMI .................................................................... 136
Reidsma, P., H. König, S. Feng, I. Bezlepkina, H. Van Keulen, M.K. Van Ittersum, F.
Brouwer: A methodological framework for sustainability impact assessment of land use
policies in developing countries: Re-using and complementing approaches.................... 138
Therond, O., N. Turpin, S. Janssen, I.N. Athanasiadis, M.J.R. Knapen, C. Bockstaller, J.
Alkan Olsson, F. Ewert, I. Bezlepkina: From models to indicators: Ontology as a
knowledge representation system...................................................................................... 140
Theme B: State-of-the-art of components for integrated systems
Session B1: Large scale integrated policy assessment: Tools and applications
143
Britz, W., T. Hertel: Impacts of EU biofuels directives on global markets and EU
environmental quality: An integrated PE, global CGE analysis ....................................... 144
Bouwman, A.F., A.H.W. Beusen, G. Billen: Modelling global N and P surface balances:
Implementation of the millennium ecosystem assessment scenarios................................ 146
Herrero, M., P.K. Thornton, M. Rosegrant, P. Havlik, S. Msangi, S. Fritz, H. Lotze-Campen,
A. Notenbaert, A. Popp, B. Eickhout, C. Muller, J. Van de Steeg, D. Gerten, T. Sulser, C.
Ringler, S. Wood, H.A. Freeman: Incorporating livestock in global integrated assessments
of land use and agro-ecosystems services ......................................................................... 148
Kuiper, M.: What if not all land is created equal? The role of heterogeneous land when
assessing the impact of trade liberalization on developing countries................................ 150
Morand, H., G. Achuo, R.J. MacGregor, A. Lefebvre: Integrated economic-environmental
modelling for domestic policy development and assessment: The Canadian experience. 152
Müller, M., I. Pérez Dominguez, W. Britz, T. Heckelei: Comparison of econometric
techniques for deriving consolidated databases: The AGROSAM case ........................... 154
vi
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Sauer, T., P. Havlík, U.A. Schneider, G. Kindermann, M. Obersteiner: Agriculture and natural
resources in a changing world: The role of irrigation ....................................................... 156
Smaling, E.M.A., R. Roscoe, J.P. Lesschen, A.F. Bouwman, E. Comunello: The fate of
nitrogen in the Brazilian soybean chain ........................................................................... 158
Van der Straeten, B., J. Buysse, S. Nolte, F.L. Marchand, L. Lauwers, D. Claeys, G. Van
Huylenbroeck: Spatial planning of livestock production and manure abatement ............. 160
Van Meijl, H., T. Jansson, M. Banse, G.B. Woltjer: The impact of modulation: Modelling
first and second pillar CAP policies .................................................................................. 162
Westhoek, H., B. Eickhout, N. Hoogervorst, T. Rood, E. Stehfest, G.B. Woltjer:
Environmental impacts of the European livestock sector on different scales, and prospects
for reduction ...................................................................................................................... 164
Woltjer, G.B.: Agricultural policy and land use .................................................................... 166
Zimmermann, A., T. Heckelei: Farm structural change in European regions ....................... 168
Session B2: Indicators for integrated assessment
171
Binder, C.R., A. Schmid, J.K. Steinberger: Sustainability Solution Space: An indicator based
tool for assessing the sustainability of agricultural systems .................................................. 172
Adamowicz, M., A. Smarzewska: Model and indicators of sustainable development in rural
areas from the local perspective ........................................................................................ 174
Bockstaller, C., N. Turpin, L. Stapleton, M. Van der Heide, O. Therond, T. Pinto-Correia, V.
Voltr, M. Raley, I. Bezlepkina, J.P. Bousset, J. Alkan Olsson, F. Ewert: A structured set of
indicators for integrated assessment of future agri-environmental policies ...................... 176
Bockstaller, C., L. Stapleton, M. Van der Heide, G. Geniaux, S. Bellon, E. Josien, M. Raley,
J. Alkan Olsson: From a disaggregated set of indicators to a synthetic, composite
assessment of sustainability: Paths and pitfalls ................................................................. 178
Delmotte, S., C. Gary, A. Ripoche, J.M. Barbier, J. Wery: Contextualization of on farm exante evaluation of the sustainability of innovative cropping systems in viticulture, using a
multiple criteria assessment tool (DEXi) .......................................................................... 180
Giannoccaro, G., J.M. Pistón, J. Berbel: Assessing the environmental sustainability of the
CAP: A DPSIR framework approach................................................................................ 182
Leip, A., F. Weiss, W. Britz: Agri-environmental nitrogen indicators for EU27 .................. 184
Meul, M., S. Van Passel: Sustainability of Flemish farms: Advising farmers and
policymakers...................................................................................................................... 186
Scholz, J., S. Herrmann: Rural regions in Europe: A new typology based on a multi-sectoral
point of view ..................................................................................................................... 188
Session B3: Cropping systems modelling
191
Keating, B.A., M.J. Robertson, P.S. Carberry, D.P. Holzworth, N.I. Huth: Cropping systems
modelling: Past performance and future prospects ........................................................... 192
Adam, M., H. Belhouchette, E. Casellas, O. Therond, J. Wery: APES, an agricultural
production and externalities simulator evaluated for two main crops in Midi-Pyrenees .. 194
Belhouchette, H., M. Adam, E. Casellas, F. Celette, M. Corbeels, J. Wery: Performances of
two crop models in various conditions: The importance of underlying assumptions ....... 196
Casellas, E., J.E. Bergez, H. Belhouchette, O. Therond, M. Adam, A. Metay, J. Wery: A
methodology for the evaluation and improvement of a generic biophysical Soil-PlantAtmosphere crop model based on ‘Mini-Application’...................................................... 198
vii
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Casellas, E., O. Therond, H. Belhouchette, D. Wallach, R.J. Oomen, G. Russell, F. Ewert, H.
Hengsdijk, M.K. Van Ittersum, M. Adam, J. Wery: Using a cropping system model for
large scale impact assessment in Europe........................................................................... 200
Clavel, L., M.-H. Charron, D. Baudet, J. Soudais, D. Leenhardt: Spatializing a bio-decisional
crop-model to assess the sensitivity of a region to modifications in agricultural practices
........................................................................................................................................... 202
Corbeels, M., E. Casellas, F. Celette, A. Mérot: The SoilC&N module of APES: Simulating
the immobilization-mineralization turnover for predicting short- and long-term C and N
dynamics in the soil ........................................................................................................... 204
Donatelli, M., G. Russell, A.E. Rizzoli: APES: The Agricultural Production and Externalities
Simulator ........................................................................................................................... 206
Erdélyi, É.: The effects of climate change on the phenology of cereal crops in Hungary ..... 208
Fantke, P., A. Kuhn, S. Wagner, R. Friedrich: Estimating external costs from the application
of plant protection products in Europe .............................................................................. 210
Guillaume, S., J.E. Bergez, E. Justes: Modelling water and nitrogen interactions in durum
wheat: A first step towards the conception of joint management strategies for irrigation and
nitrogen fertilization .......................................................................................................... 212
Heinemann, A.B.: Optimization of upland rice breeding programme based on a crop
modelling approach ........................................................................................................... 214
Ines, A.V.M., J.W. Hansen: Extracting useful information from daily GCM rainfall for
cropping system modelling................................................................................................ 216
Jones, J.W., J. Naab, D. Fatondji, K. Dzotsi, S. Adiku, J. He: Uncertainties in simulating crop
performance in degraded soils and low input production systems ................................... 218
Koo, J., S. Wood, Z. Guo: Crop systems evaluation for strategic investment decisions ....... 220
Launay, M., N. Brisson, S. Satger, H. Hauggaard-Nielsen, G. Corre-Hellou, E. Kasynova,
E.S. Jensen, M.J. Gooding, C. Dahlmann, P. Von Fragstein, A. Pristeri, M. Romeo, M.
Monti: A modelling approach to optimize the management of pea-barley intercropping in
Europe................................................................................................................................ 222
Lebonvallet, S., N. Brisson, J.-P. Raffaillac, M. Launay: Establishment of quinoa
(Chenopodium quinoa Willd.) and simulation of its cultivation with the STICS crop
model ................................................................................................................................. 224
Mérot, A., J.E. Bergez, J.C. Mailhol, J. Wery: An integrative modelling approach to simulate
the agricultural system through a combination of a decisional, technical and biophysical
sub-system ......................................................................................................................... 226
Metay, A., C. Gary, E. Casellas, A. Ripoche, K. Kansou, J. Wery: Modelling cover-cropped
vineyards with APES: A modular simulation platform..................................................... 228
Mirsky, S.B., J.R. Teasdale, R.A. Mangum, M.A. Cavigelli, J. Maul, J.T. Spargo: A systems
approach to optimizing agronomic and economic performance of legume-based organic
maize production ............................................................................................................... 230
Oomen, R.J., H. Hengsdijk, O. Therond, J.E. Bergez, G. Russell, E. Casellas, S. Janssen,
M.K. Van Ittersum: Development of generic management rules for crop growth simulation
models................................................................................................................................ 232
Schönhart, M., E. Schmid, U.A. Schneider, P. Liebhard: Crop rotation modelling for
integrated environmental assessment: Case study results and empirical evidence .......... 234
Snow, V.O., M. Lieffering, B.R. Cullen, I.R. Johnson: Implications of heterogeneous urinaryN return for modelling pasture production and N losses from grazed pastures under climate
change................................................................................................................................ 236
Tittonell, P., M. Corbeels, K.E. Giller: The interaction between sub-optimally available
resources for crop production as implemented in the simplified model FIELD ............... 238
Van der Velde, M., F. Bouraoui, P. Prosperi, J.M. Terres, B.S. Elbersen: The effect of
conservation agriculture practices in Europe .................................................................... 240
viii
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Welch, S.M., A. Wilczek, J.L. Roe, B. Moyer, M. Cooper, J. Schmitt, S. Das, P. Koduru,
X. Cai: Toward gene-based crop simulation models for use in climate change studies ... 242
Session B4: Bio-economic farm modelling
245
Flichman, G., K. Louhichi: Some conceptual issues for modelling the interaction between
agriculture and environment through the bio-economic mathematical programming models
........................................................................................................................................... 246
Akkas Ali, M., M. Robiul Alam, M.S.H. Molla, M.A. Momin: Integrated home garden model
for food security, nutrition and poverty alleviation........................................................... 248
Ben El Ghali, M., A. Ridier, C. Kephaliacos, G. Nguyen: Economic impact of voluntary agrienvironmental measures: The case study of a ‘Test-Action’ in a basin of south-western
France ................................................................................................................................ 250
Bocquého, G., F. Jacquet, L. Bamière: An appraisal of risk on perennial energy crops in
France ................................................................................................................................ 252
Goulevant, G., S. De Cara, F. Jacquet, A. Reynaud, M.H. Jeuffroy, P. Lucas, F. Montfort:
Economic analysis of cover crop (white mustard) adoption to reduce take-all disease and
N-leaching ......................................................................................................................... 254
Janssen, S., K. Louhichi, A. Kanellopoulos, P. Zander, G. Flichman, H. Hengsdijk, Hongtao
Li, T. Heckelei, G. Stokstad, N. Borkowski, M. Hecker, A. Oude Lansink, M. Blanco, E.
Meuter, M.K. Van Ittersum: Farming Systems SIMulator: First generic bio-economic farm
model ................................................................................................................................. 256
Lauwers, L., P. Everaert, J. Van Meensel, K. Bryon: Using time driven activity-based costing
data for modelling complex innovations: The case of batch farrowing ............................ 258
Leneman, H., R.A.M. Schrijver: Higher prices, less nature management? Modelling farmers’
decisions on dairy farms .................................................................................................... 260
Louhichi, K., H. Hengsdijk, S. Janssen, G. Bigot, E. Perret, M.K. Van Ittersum: EU dairy
farming in the face of change: An exploration using a bio-economic farm model ........... 262
Mosnier, C., A. Reynaud, J. Agabriel, M. Lherm: Dynamics of suckler cow farms under
stochastic crop yields: A recursive discrete stochastic programming approach ............... 264
Mosnier, C., J. Agabriel, M. Lherm, A. Reynaud: A dynamic bio-economic model to simulate
optimal adjustments of suckler cow farm management to production and markets shocks
........................................................................................................................................... 266
Nidumolu, U.D., M. Lubbers, V. Alary, P. Lecomte, H. Van Keulen: Modelling Nitrogen
balance for a regional scale livestock-pasture system as a discussion support tool .......... 268
Robertson, M.J., A. Moore, A. Bathgate: Linking cropping systems models with farm bioeconomic models to examine the relationship between profit and natural resource
condition ............................................................................................................................ 270
Salgado, P., M. Lubbers, R. Schipper, H. Van Keulen, V. Alary, P. Lecomte: Adoption of
new forage technology: Impact on the socio-economic sustainability of milk production in
Moc Chau, Vietnam (DAIVIE model) .............................................................................. 272
Schönhart, M., T. Schauppenlehner, E. Schmid: Integration of biophysical and economic
models for spatially explicit land use and landscape analysis on farm level ................... 274
Traoré, B., K. Louhichi, B. Rapidel: A bio-economic model to analyse the performance of the
cotton sector in Mali.......................................................................................................... 276
Van Meensel, J., L. Lauwers, G. Van Huylenbroeck, S. Van Passel: Exploring productiontheoretical insights for economic-environmental trade-off analysis ................................. 278
Vayssières, J., M. Vigne, V. Slegten, P. Lecomte: A whole farm simulation model to improve
multi-criteria system efficiency......................................................................................... 280
ix
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Session B5: Integrated assessment methods for regional and landscape levels
283
Stoorvogel, J.J.: Integrated assessment modelling: How to assess regional impacts?........... 284
Bobojonov, I., E. Berg, C. Martius, J.P.A. Lamers, J. Franz: Integrated Assessment of crop
and water allocation under conditions of uncertain water supply in the Khorezm Region of
Uzbekistan, Central Asia ................................................................................................... 286
Brady, M., K. Kellermann: An agent-based model for assessing the impacts of decoupled
agricultural support on landscape values........................................................................... 288
Breukers, A., W. Van der Werf, M. Mourits, A. Oude Lansink: Generic bio-economic
modelling of plant diseases: An exploration ..................................................................... 290
Bulatewicz, T., X. Yang, J.M. Peterson, S. Staggenborg, D.R. Steward, S.M. Welch:
Integration of agriculture, groundwater and economic models using the Open Modeling
Interface (OpenMI): Application to the Ogallala Aquifer................................................. 292
Gary, C., J.M. Barbier, P. Rio, P. Andrieux, J.M. Blazy, X. Louchart, M. Bonin, J.L. Diman,
F. Causeret, H. Ozier-Lafontaine: How to design technical and organizational innovations
to promote sustainable development in catchments with intensive use of pesticides ....... 294
Griffon, S., D. Auclair: Visualizing changes in agricultural landscapes ............................... 296
Kros, J., T.J.A. Gies, J.C. Voogd, R. Smidt, W. De Vries: Integrated analysis of the effects of
agricultural management and spatial planning on environmental quality at landscape scale
........................................................................................................................................... 298
Osterburg, B., T.G. Schmidt: Assessment of nitrogen balances at farm and regional level .. 300
Schleyer, C., I. Theesfeld, O. Aznar: Assessing the institutional compatibility of new policies
........................................................................................................................................... 302
Shahbazi, F., A.A. Jafarzadeh, M.R. Shahbazi: Assessing sustainable agriculture development
using the MicroLEIS DSS in Souma area, Iran................................................................. 304
Streifeneder, T., C. Hoffmann, F.V. Ruffini: Assessing and predicting agro-structural change
in the Alpine Region.......................................................................................................... 306
Uthes, S., C. Sattler, A. Osuch, K. Happe, A. Piorr, P. Zander: Sustainability impact
assessment of EU policies at regional level: An analysis of land use dynamics and
environmental effects with the MEA-Scope tool .............................................................. 308
Wagner, S., O. Beletskaya, U. Kummer, J. Theloke, E. Angenendt, R. Friedrich, J. Zeddies:
Assessment of mitigation measures for air pollutants and greenhouse gases from
agricultural systems in Germany at high spatial resolution............................................... 310
Warnecke, S., M. Biberacher, G. Broll: Regionally optimized animal farm manure transports
in an area with high intensity animal farming systems ..................................................... 312
Session B6: Software infrastructures and tools for integrated assessment
315
Moore, R.V., J.J.F. Wien: Integrated modelling tools, mass collaboration and the
opportunities ahead............................................................................................................ 316
Antle, J.M., J.J. Stoorvogel, R.O. Valdivia: The Trade-off Analysis software: Modelling tools
to support informed policy decision making ..................................................................... 318
Athanasiadis, I.N., S. Janssen, E. Andersen, A.E. Rizzoli, J.J.F. Wien: Granular ontologies for
integrated assessment of agricultural systems ................................................................... 320
Botreau, R., I. Veissier, P. Champciaux, J.-P. Brun, A. Lamadon, P. Perny: A software tool to
support multi-criteria-based assessment of animal welfare............................................... 322
Carlson, J., O. David, J. Ascough, F. Geter, L. Ahuja: The role of the Object Modelling
System (OMS) for integrated assessments of conservation on agricultural land in the
United States...................................................................................................................... 324
x
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Justes, E., P. Lecharpentier, S. Buis, M. Launay, J. Bourges, J. Soudais, P. Clastre, J.E.
Bergez, P. Debaeke, D. Wallach, M. Guérif: OptimiSTICS, a software for STICS crop
model parameter estimation and evaluation which was build up to function with other
dynamical models .............................................................................................................. 326
Kassahun, A.: ProST: A process support tool for supporting water stress mitigation through a
participatory approach ....................................................................................................... 328
Knapen, M.J.R., I.N. Athanasiadis, B. Jonsson, D. Huber, J.J.F. Wien, A.E. Rizzoli, S.
Janssen: Use of OpenMI in SEAMLESS .......................................................................... 330
Meuter, E., S. Janssen, M.K. Van Ittersum: A graphical user interface for a generic bioeconomic farm model (FSSIM)......................................................................................... 332
Nemecek, T., S. Blaser, D. Dux, G. Gaillard: What are the key factors for successful
integration of life cycle assessment into FADN? ............................................................. 334
Salazar, O., A. Joel, I. Wesström: An integrated framework for management of spatial and
temporal data on a drained watershed ............................................................................... 336
Steward, D.R., X. Yang, W.J. De Lange: Groundwater concepts and agriculture: A GIScience
data model study................................................................................................................ 338
Verweij, P.J.F.M., M. Van Eupen, J. Roos-Klein Lankorst, W. Nieuwenhuizen: Tools for
transparent spatial integrated assessments......................................................................... 340
Wallach, D., F. Van Evert, M. Adam: Parameter estimation software for crop models........ 342
Wien, J.J.F., A.E. Rizzoli, M.J.R. Knapen, I.N. Athanasiadis, S. Janssen, L. Ruinelli, F. Villa,
M. Svensson, P. Wallman, B. Jonsson: Architecture of SEAMLESS-IF as framework
application for integrated modelling ................................................................................. 344
Session B7: Databases for integrated assessment of agriculture and rural development
347
Kasnakoglu, H.: Usefulness, feasibility, limitations and sustainability of an integrated
international database on agricultural statistics ................................................................ 348
Andersen, E., B. Baruth, B.S. Elbersen, G.W. Hazeu, S. Janssen, I. Staritsky, I.J. Terluin,
C.A. Van Diepen, D. Verhoog: Integrating farming system, environmental and socioeconomic data: The SEAMLESS database ....................................................................... 350
Hazeu, G.W., E. Andersen, B. Baruth, M.J. Metzger, C.A. Mücher, M. Pérez-Soba, Ch.
Renetzeder, C.A. Van Diepen: European environmental stratifications: an overview ..... 352
Kempen, M., B.S. Elbersen, I. Staritsky, E. Andersen: Detailed spatial distribution of farming
systems in Europe: A basic starting point for the further environmental assessment of
changes in farming and CAP reform ................................................................................. 354
Lavelle, P., X. Arnauld de Sartre, S. Dolédec, B. Hubert, I. Veiga: Need for a theoretical
framework in building an integrated database................................................................... 356
Terluin, I.J., D. Verhoog, F.E. Godeschalk: Some reflections on the design and use of
typologies of EU regions in socio-economic analyses ..................................................... 358
Theme C: Case studies and application of tools and empirical methods
Session C1: Integrated assessment of biomass for biofuel or renewables - trade-offs with
feed and food
361
Petersen, J.-E.: Challenges in modelling environmental bio-energy potential(s) ................. 362
Bocardo, P.J., K. Kondoh, J.C.S. Cruz, D.S. Kimura: Life cycle assessment of the sugarcane
production system in Brasil ............................................................................................... 364
xi
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Coletto, L., A. Brigi, A. Calgaro, S.G. Sandrini, S. Bona: Application of Life Cycle
Assessment methodology to compare sunflower oil and diesel for fulfilling farm energy
needs .................................................................................................................................. 366
Dos Santos Rocha, A., M. Mendes da Silva: Family production systems with dende (Elaeis
guineensis) and the alternative of biodiesel: Analysis in the Cajaíba community,
municipality of Valença (Bahia-Brasil)............................................................................. 368
Guille, T., A. Fallot, F. Jacquet, P. Havlík: The availability of crop residues for energy,
assessing global potentials................................................................................................. 370
Kukk, L., E. Suuster, A. Astover, M. Noormets, H. Roostalu: Soil optimized suitability
analysis for allocation of energy grasses ........................................................................... 372
Lankoski, J., M. Ollikainen: Environmental impacts of agricultural land allocation between
bio-energy crops and food/feed crops .............................................................................. 374
Leip, A., R.W. Howarth, S. Bringezu: Biofuels: The way forward ....................................... 376
Msangi, S., R. Schaldach, C. Kölking, J. Koch, J. Alcamo: Exploring the land, water and
energy linkages in agriculture: An expanded partial-equilibrium approach to modelling
environmental impacts....................................................................................................... 378
Nonhebel, S.: Energy from agricultural residues and the trade-off to the food system ........ 380
Schuler, J., G. Uckert, P. Zander: Assessing the environmental effects of an increasing biogas
production on agricultural farms ....................................................................................... 382
Stehfest, E., G.B. Woltjer, A.G. Prins, B. Eickhout, M. Banse: First and second generation
biofuels up to 2030: Possible scenarios and their environmental impacts ........................ 384
Suuster, E., L. Kukk, A. Astover, M. Noormets: Spatial assessment of abandoned agricultural
land for bio-energy production.......................................................................................... 386
Thorburn, P.J., D. O’Connell, T. Grant: Enhancing the assessment of biofuels feedstock
production through more realistic representation of farming systems .............................. 388
Topp, C.F.E., R.M. Rees: Greenhouse gas emissions from grassland swards: A Scottish
perspective......................................................................................................................... 390
Van der Velde, M., G. Wriedt, F. Bouraoui: Water implications of biofuel crop cultivation
and set-aside suspension in the European Union .............................................................. 392
Van de Ven, G.W.J., C.L.M. De Visser, H. Langeveld, S.C. De Vries, L. Van den Brink: Bioethanol from sugar beet in The Netherlands: Sustainability assessment........................... 394
Van Slycken, V., N. Witters, E. Meers, K. Adriaensen, T. Thewys, J. Vangronsveld, F.M.G.
Tack: Energy maize for phytoremediation of metal-enriched soils and production of
energy: The Campine region, Belgium ............................................................................. 396
Session C2: Modelling adaptation strategies to cope with climate change
399
Meinke, H., R. Nelson: Will global mitigation policy enhance or undermine local adaptation?
........................................................................................................................................... 400
Baethgen, W.E., J.W. Hansen, A.V.M. Ines, A.M. Greene, L. Goddard: Exploring options to
improve adaptation to climate change in crop production of south-eastern South America
........................................................................................................................................... 402
Bezabij, Y.G., R. Sintanyeho, A. Van Griensven: Assessment and predictions of agricultural
drought for the Upper Blue Nile Basin, Ethiopia .............................................................. 404
Boote, K.J., J.W. Jones, G. Hoogenboom, C. Rosenzweig: Pre-requisites for crop models used
to test strategies for adapting to, or mitigating effects of climate change ......................... 406
Claessens, L., J.J. Stoorvogel, J.M. Antle, R.O. Valdivia, M. Herrero: A minimum-data
approach to ex-ante assessment of climate change adaptation strategies in resource-poor
countries ............................................................................................................................ 408
xii
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Crespo, O., S. Hachigonta, M. Tadross: Sensitivity of southern African maize yields to the
definition of onset under conditions of climate change..................................................... 410
De Vries, W., J. Kros, G.J. Reinds, H.J.J. Wieggers, G.L. Velthof, D.A. Oudendag, G.J.
Nabuurs, M.J. Schelhaas, M. Pérez-Soba, W.P. De Winter, J. Van den Akker, M.M.
Bakker, P.H. Verburg, J. Liski, B. Eickhout, A.F. Bouwman: Assessment of nitrogen and
green house gas fluxes at the European scale in response to land use and livestock change
........................................................................................................................................... 412
Dray, A., S. Asseng, P. Perez, S.P. Charles, B. Bates: SimKat: A virtual laboratory to explore
the impact of climate change scenarios on the Western Australian wheat-belt ................ 414
Havlík, P., U.A. Schneider, E. Schmid, H. Böttcher, G. Kindermann, S. Leduc, M.
Obersteiner: GHG mitigation through bio-energy production versus avoided deforestation
and afforestation: A quantitative analysis ......................................................................... 416
Kan, I., A. Kimhi, K. Shraberman: Impact assessment of climate change on agriculture:
Econometric-process analysis under partial equilibrium .................................................. 418
Kim, Chang-Gil: Counter measures of agricultural sector for post-Kyoto Protocol ............. 420
Kimura, S.D., Y. Toma, Z. Mu, R. Hatano: Analysis of land use combination to minimize
global warming potential from farmland and farmland surplus Nitrogen and to maximize
income using eco-balance analysis.................................................................................... 422
Lieffering, M., P. Newton: Modelling the costs of climate change on grazing systems ...... 424
Moriondo, M., R. Ferrise, M. Bindi: Testing adaptation options in a +2 °C warmer climate 426
Strauss, F., E. Schmid, E. Moltchanova: Simulation of climate scenarios and the bioeconomic assessment of different agricultural management systems in the Marchfeld
region................................................................................................................................. 428
Vigne, M., J. Vayssières, J.L. Bochu, P. Lecomte: Energy use efficiency and greenhouse
gases emissions of dairy farms of an isolated territory: Reunion Island from 2000 to 2007
........................................................................................................................................... 430
Walker, S., D. Nanja, P. Gondwe, J. Dimes, et al.: Mujika community climate: Cropping
systems modelling exercise ............................................................................................... 432
Session C3: Application of integrated assessment tools
435
Adenäuer, M., M. Kuiper, K. Louhichi: Effects of trade liberalization on EU agriculture: A
typical macro level application of SEAMLESS-IF ........................................................... 436
Amblard, L., C. Mann, C. Schleyer: Ex-ante institutional compatibility assessment of policy
options: Insights from empirical applications and interactions with policy experts ......... 438
Antunes, P., R. Santos, V. Karadzic: Challenges in the application of multi-criteria assessment
(MCA) tool for the evaluation of irrigation management alternatives: PLEIADeS project
........................................................................................................................................... 440
Belhouchette, H., K. Louhichi, O. Therond, J. Wery, G. Flichman: A crop-farm-indicators
modelling chain to assess farmer’s decision in response to socio-economic scenarios .... 442
Cacovean, H., T. Man, T. Rusu, D.C. Weindorf, I. Fodorean, C. Moldovan, V. Puiu, A. Pop:
Land-Use Management using a soil survey geographic database for the Viisoara-Aiton
hills (Transylvanian Plain-Romania) ..................................................................................................444
Coletto, L., S. Bona, S.G. Sandrini, A. Brigi, G. Florio, P. Sambo: Organic crop rotations
evaluated using LCA (Life Cycle Assessment)................................................................. 446
Colomb, B., F. Angevin, J.E. Bergez, M. Blouin, F. Carpy-Goulard, A. Glandières, L. Prieur:
Using MASC to evaluate the sustainability of cropping systems: Application to multicriteria assessment of organic cropping systems in Southern France ............................... 448
xiii
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Hochman, Z., J. Hunt, B. Long, D.P. Holzworth, A. Whitbread, S. Van Rees, P. DeVoil:
Building crop producers’ capacity to adapt to climate change and climate variability with
Yield Prophet..................................................................................................................... 450
Lescot, J.M., P. Bordenave: A decision support to choose between changes of agricultural
practices: A spatially distributed Costs/Effectiveness assessment framework ................ 452
Lesschen, J.P., J. Kros, G.L. Velthof, D.A. Oudendag, M. Kempen, W. De Vries:
Quantification of the impacts of Cross Compliance measures on European air, soil and
water quality ...................................................................................................................... 454
Neuwirth, J., K. Wagner, H. Janetschek: Assessment of flood prevention contribution of
multi-functional agricultural land: Results of a case study in Lower Austria ................... 456
Palma, J., A.R. Graves, P.J. Burgess, W. Van der Werf, F. Herzog: Silvo-arable agro-forestry
in Europe: Systems for meeting food and industry crop requirements with improved
resource-use efficiency...................................................................................................... 458
Perugini, M., M. Toderi, G. Seddaiu, R. Orsini, G. De Sanctis, P.P. Roggero: Integrated
impact assessment of agro-environmental schemes on soil erosion and water quality..... 460
Oñate, J.J., B.S. Elbersen, P. Martinez Alonso, B. Peco, J. Roos-Klein Lankhorst: Assessing
the effects of Cross Compliance obligations on European biodiversity and landscapes .. 462
Sasaki, H.: Integrated model of paddy rice production and related environmental effects ... 464
Turpin, N., J.P. Bousset, G. Bigot, E. Josien, E. Perret: Can regional policies mitigate the
environmental effects of price raise? A typical application of SEAMLESS framework at
meso level in Auvergne ..................................................................................................... 466
Session C4: Systems analysis informing policy: Case studies from agricultural systems
and rural regions in the face of change
469
Pannell, D.: Systems analysis informing policy..................................................................... 470
Antle, J.M., J.J. Stoorvogel, R.O. Valdivia: Assessing poverty-sustainability trade-offs in
semi-subsistence agricultural systems: A case study of policy options in Machakos, Kenya
........................................................................................................................................... 472
Chen, Z., F. Qu, X. Huang: Performance analysis on the changes of farmland property right
system: An explanation of the diversification of Chinese farmland system in transition . 474
Dillon, E.J., T. Hennessy, S. Hynes: Assessing the sustainability of Irish agriculture.......... 476
Kanellopoulos, A., H. Hengsdijk, S. Janssen, M.K. Van Ittersum: Ex-ante assessment of the
abolishment of the EU set aside policy: Results from a bio-economic farm analysis ...... 478
Lange, A., A. Piorr, R. Siebert, D. Viaggi: Direct and indirect impacts of CAP changes on the
economy of rural areas: Integrated assessment of farm households’ adaptation strategies in
Eastern and Western Germany .......................................................................................... 480
Mack, G., C. Flury, T. Walter: Alpine grazing trends in Switzerland: Economic importance
and impact on biotic communities..................................................................................... 482
Majewski, E., A. Was, H. Belhouchette, K. Louhichi, I. Mouratiadou: Impact Assessment of
policy changes on the arable sector using the FSSIM model: Case study of the
Zachodniopomorskie NUTS2 region ................................................................................ 484
Marchand, F.L., D. Claeys, L. Lauwers, J. Buysse, B. Van der Straeten, G. Van
Huylenbroeck: Introducing flat rate direct payments taking into account social concerns
and ecological non-commodities ....................................................................................... 486
Mouratiadou, I., C.F.E. Topp, G. Russell: Investigating the economic and water quality
effects of the 2003 CAP reform on arable cropping systems: A Scottish case study ....... 488
Thorburn, P.J., J. Brodie, R. Milla, E. Shannon, S. Attard, T. Anderson, A. Davis: Systems
analysis underpinning development of management action targets to meet water quality
objectives in sugarcane production systems...................................................................... 490
xiv
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Havlikova, M., C. Kroeze: Efficiency of environmental policies in agriculture in the Czech
Republic: The case of dairy cattle ..................................................................................... 492
Theme D: Integrated assessment: Science and impact
Session D1: Stakeholder involvement in designing, use and evaluation of integrated
assessment models
495
Pahl-Wostl, C.: The role of models in supporting processes of change towards sustainable
resource governance .......................................................................................................... 497
De Kraker, J., C. Kroeze, P. Kirschner: Models as social learning tools in participatory
integrated assessment ........................................................................................................ 498
De Mey, K., K. D’Haene, M. Meul: Application of an integrated monitoring tool for
sustainable dairy farming in Flanders................................................................................ 500
Duru, M., G. Martin: Analysis of stakeholder-researcher interactions for designing a generic
grassland model suited for a diversity of farmers’ practices ............................................. 502
Gabbert, S., M.K. Van Ittersum, F. Ewert, C. Kroeze, S. Stalpers, J. Alkan Olsson:
Accounting for user’s uncertainty information needs in Integrated Assessment models: The
case of the SEAMLESS Integrated Framework ................................................................ 504
Jonsson, A., J. Alkan Olsson, L. Andersson: Modellers, model results and local stakeholders –
dynamics in combining expert and lay knowledge in locally grounded policy formulation
........................................................................................................................................... 506
Karadzic, V., P. Antunes, R. Santos, A. Osann: Transition towards more adaptive irrigation
management: The impact of innovative technologies. A case study in Caia, Guadiana River
Basin .................................................................................................................................. 508
Lusiana, B., D. Suyamto, R. Mulia, M. Van Noordwijk, G. Cadisch: Prospecting the
behaviour of agro-ecosystems: Experiences in using FALLOW model with natural
resource managers ............................................................................................................. 510
Nicetic, O., E. Van de Fliert, Ho Van Chien, R. Spooner-Hart, Z. O’Leary, D. Rae: Learning
by doing: Designing and conducting impact assessment studies for citrus Farmer Field
Schools in Vietnam............................................................................................................ 512
Reidsma, P., J. Alkan Olsson, I. Bezlepkina, S. Brogaard, J. Queste, R. Fischer, M.K. Van
Ittersum: SEAMLESS training material on concepts and models for Integrated Assessment
of agriculture and sustainable development ...................................................................... 514
Speelman, E.N., L.E. García-Barrios: Serious Gaming as a tool to create sustainability
awareness, discussion and negation among stakeholders in cattle systems, the case of Los
Angelitos............................................................................................................................ 516
Speelman, E.N., J.C.J. Groot, H. Van Keulen, K. Kok, W.A.H. Rossing: Understanding and
modelling socio-ecological networks to support multi-scale governance of sustainable
dairy farming systems........................................................................................................ 518
Taverne, M., O. Therond, J.P. Bousset, J. Alkan Olsson, F. Ewert, M.K. Van Ittersum, J.
Wery: Specifying impact assessment simulations in interaction with end-users: The
example of SEAMLESS-IF............................................................................................... 520
Thorburn, P.J., E. Jakku: Agricultural decision support systems: Tools to facilitate co-learning
for stakeholders and scientists ........................................................................................... 522
Van Paassen, A., C. Barnaud, G. Trébuil: Participatory modelling for Renewable Resource
Management: A critical assessment .................................................................................. 524
Van Paassen, A., N. De Ridder, L. Stroosnijder: Using explorative models to enhance expert
learning about farm context and farmer rationalities in Burkina Faso .............................. 526
xv
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Session D2: Role of integrated assessment for policy making and evaluation
529
Alkan Olsson, J., M.K. Van Ittersum, A.-K. Bäcklund, S. Brogaard, F. Brouwer: Participatory
IA tool development: Lessons learnt in the SEAMLESS project ..................................... 530
Bäcklund, A.-K., J. Alkan Olsson, S. Brogaard, M. Taverne, J.P. Bousset, J. Wery, M.K. Van
Ittersum: Multiple agendas and cultures in a participatory development of computerized IA
tools: The experience of SEAMLESS ............................................................................... 532
Dahal, B.M.: Integrated assessment for the development of Progressive Coordinative Action
(PCA) plan: An approach for sustainable agriculture ....................................................... 534
Fischer, R., I. Theesfeld: Better regulation by the use of Integrated Assessment? SEAMLESS
integrated framework and European policy making ......................................................... 536
Hasnain, N., H. ApSimon, T. Oxley: A multi-criteria approach to environmental policy .... 538
Hernández Jiménez, V.: Exploring receptivity through innovative actions........................... 540
König, B., K. Helming, K. Tscherning, H. König, K. Diehl: Determinants of policy-oriented
research design .................................................................................................................. 542
Padmanabhan, M.: Integrated impact assessment as a new institution: The political
dimensions of modelling-tool use ..................................................................................... 544
Van Delden, H., R. Vanhout: Contribution of the development process to the actual
implementation and use of Integrated Spatial Decision Support Systems ........................ 546
Sieber, S., P.J.F.M. Verweij, K. Müller, K. Tscherning, K. Helming: Design and institutional
embedment of the meta-model Sustainability Impact Assessment Tools (SIAT) ............ 548
König, H., K. Helming, P. Reidsma, L. Zhen, S. Feng, S. Sieber, F. Brouwer: Bridging the
gap between science and decision makers for sustainable development of multi-functional
land use in developing countries ....................................................................................... 550
Authors Index
553
xvi
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Preface
The theme of the AgSAP conference reflects the increased need for integrated research
approaches that assist in assessing agricultural systems at multiple levels and in terms of
economic, environmental and social aspects. The conference presents the state-of-the-art of
scientific approaches to assess agricultural systems in the context of sustainable development,
and aims to propose an agenda for future research and for improved science-policy interaction
in this domain.
The ca. 250 contributions present, evaluate and compare alternative methods and modelling
approaches, applications and policy support options. They focus on the integration and use of
models for linking science and policy, as a method for improving natural resource use, policy
making and policy implementation in agriculture. The conference is structured along four
major themes, each with several sessions:
A. Methodology for integrated assessment – Alternative methods for integrated assessment of
agricultural systems and contributions to rural development are presented. Methods for upand downscaling are discussed.
B. State-of-the-art components for integrated systems – Progress in modelling tools and their
use for supporting policy assessment on agriculture, rural development and sustainable
development. Focus in these sessions is on components rather than on entire systems:
assessment at field, farm, regional, continental or market level; indicators; databases and
software engineering.
C. Case studies and application of tools and empirical methods – The sessions present
empirical applications, using modelling tools, to specific problems, policy areas and public
concerns.
D. Integrated assessment science and impact – Explore the broader societal use and impact of
integrated assessment tools, including the contribution to science-policy interactions.
The conference was initiated by the SEAMLESS integrated project (System for
Environmental and Agricultural Modelling; Linking European Science and Society), funded
by the European Commission under Framework Programme 6. This project (2005–2009;
www.seamless-ip.org) developed an operational modelling framework for integrated
assessment of agricultural systems. The conference was endorsed by the European Society for
Agronomy (ESA), European Federation for Information Technology in Agriculture (EFITA),
International Consortium for Agricultural Systems Applications (ICASA) and the
International Environmental Modelling and Software Society (iEMSs).
The conference is greatly indebted to the session chairs for organizing the 17 parallel sessions
and evaluating the abstracts:
• Johanna Alkan Olsson, Lund University, Sweden
• Erling Andersen, University of Copenhagen, Denmark
• John Antle, Montana State University, USA
• Robert Argent, Bureau of Meteorology, Canberra, Australia
• Daniel Auclair, INRA, Montpellier, France
• Martin Banse, LEI, Wageningen UR, The Netherlands
• Irina Bezlepkina, LEI, Wageningen UR, The Netherlands
• Floor Brouwer, LEI, Wageningen UR, The Netherlands
• Frank Ewert, University of Bonn, Germany
• Guillermo Flichman, IAMM, Montpellier, France
xvii
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Hayo Haanstra, Ministry of Agriculture, Nature and Food Quality, The Netherlands
Jim Hansen, International Research Institute for Climate Prediction, New York, USA
Thomas Heckelei, University of Bonn, Germany
Katharina Helming, ZALF, Müncheberg, Germany
Sylvia Herrmann, Leibniz University, Hannover, Germany
George Hutchinson, Queen’s University Belfast, Northern Ireland
Jim Jones, University of Florida, Gainesville, USA
Haluk Kasnakoglu, Middle East Technical University, Ankara, Turkey
Brian Keating, CSIRO, Sustainable Ecosystems, Brisbane, Australia
Marijke Kuiper, LEI, Wageningen UR, The Netherlands
Ken Moore, Iowa State University, USA
Roger Moore, CEH, Wallingford, UK
Lennart Olsson, Lund University, Sweden
Maria Luisa Paracchini, Joint Research Centre, Ispra, Italy
Pytrik Reidsma, Wageningen University, The Netherlands
Andrea Rizzoli, IDSIA, Lugano, Switzerland
Mike Robertson, CSIRO, Sustainable Ecosystems, Australia
Graham Russell, University of Edinburgh, UK
Stefan Sieber, IPTS, Joint Research Centre, Seville, Spain and ZALF, Germany
Insa Theesfeld, Leibniz Inst. of Agric. Development in CE Europe, Halle, Germany
Pablo Tittonell, CIRAD, Montpellier, France
Nadine Turpin, CEMAGREF, Clermont Ferrand, France
Elske Van de Fliert, The University of Queensland, Brisbane, Australia
Martin Volk, Helmholtz Centre for Environmental Research - UFZ, Leipzig, Germany
Jacques Wery, UMR System (Agro.M-CIRAD-INRA), Montpellier, France
Jan Erik Wien, Alterra, Wageningen UR, The Netherlands
Finally, these Proceedings are available, in time, thanks to the enormous dedication of Joost
Wolf and Gon van Laar. Joost organized the evaluation and made sure that no abstract is
missing and Gon edited the Book of Proceedings. Thank you Joost and thank you Gon for
your professional and kind contributions!
On behalf of the Organization,
Martin K. Van Ittersum
xviii
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Plenary Sessions
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Agriculture and sustainable development in developing countries
Joachim Von Braun
International Food Policy Research Institute,
2033 K Street, NW, Washington, DC 20006, USA
Contact: j.vonbraun@cgiar.org
Introduction
The sustainability and advancement of agriculture in low-income countries is hampered by
the complex interaction of new challenges, including the global food and financial crises that
occurred in 2007–08, demand for biofuel production, and climate change. Agricultural
productivity growth supported by significant investments in research and development (R&D)
is crucial for addressing these risk factors and building resiliency. Such investments, however,
have been stagnating since the mid-1990s, and the recent financial crunch has further
constrained the availability of capital for agriculture. A strategic initiative for technological
and institutional innovations at global scale and related investment action is urgently needed
to respond to the challenges.
New and ongoing pressures
The global food crisis in 2007–08 and financial crisis in 2008 have serious implications for
agriculture and poor people in developing countries (Von Braun, 2008). The surge in prices of
food commodities resulted in a major food crisis that stemmed from rising demand for
agricultural products (due to income and population growth), rising energy prices, and
expansion of biofuel production. However, the much needed and fast supply response was
limited to industrialized countries and not forthcoming in the developing regions because of
underinvestment in agriculture, farmers’ limited access to inputs, natural resource constraints,
and weather disruptions (Von Braun et al., 2008). In fact, cereal production in developing
countries excluding Brazil, China, and India fell by 1.6% between 2007 and 2008 (FAO,
2008). Although high food prices provided incentives for policymakers, farmers, and
investors to boost agricultural productivity, the variability of prices was an obstacle to longterm planning. As the financial crisis and economic slowdown unfolded, decreasing demand
for agricultural commodities pushed food prices to lower level. Further, with limited and
more expensive capital, broader plans for agricultural investments in low-income economies
are cut short.
Increasing competition for land and water resources for agriculture as a result of the food
crisis and declining capital for long-term investments caused a revaluation of natural
resources in some countries. In Brazil, for example, farmland prices rose by 16% in 2007
alone as reported by the media and probably dropped back in 2008. In addition, constraints on
capital availability may again lead to overexploitation and degradation of scarce natural
resources, i.e. land and water. Developed water sources are almost fully utilized in many
countries, while agricultural demand for water is expected to increase significantly in the
future (GES, 2008).
The productivity and sustainability of agriculture remains at risk as yields and overall
productivity growth are stagnating. Even before the food crisis hit, average annual cereal yield
growth in developing countries was declining (World Bank, 2007). Total factor productivity –
derived from the ratio of total output growth to total input growth – in the developing world
grew on average by 2.1% per annum from 1992 to 2003; in some regions, the rate of growth
was even lower. Threats to productivity and output growth will also rise in the future as
climate variability and change increases temperatures and the risk of droughts and floods
(Cline, 2007) and yields in developing countries could further decrease (Fischer et al., 2005).
2
Plenary Sessions
Strategic action for technological and institutional change in food and agriculture
A significant increase in the level of current investments is crucial for agriculture to face
challenges and respond to new opportunities. Investments should be made in research and
development (R&D), rural infrastructure, rural institutions, and information monitoring and
sharing.
A recent study by IFPRI shows that if investments in public agricultural research doubled
from US$5 to US$10 billion from 2008 to 2013, agricultural output would increase
significantly and millions of people would emerge from poverty. If these R&D investments
are targeted at the poor regions of the world – Sub-Saharan Africa and South Asia – overall
agricultural output growth would increase by 1.1 percentage points a year and lift about 282
million people out of poverty by 2020. International agricultural research projects with
substantial payoffs for a large number of beneficiaries should be given priority. The centres of
the Consultative Group on International Agricultural Research (CGIAR) have identified
examples of ‘best bets’ in agricultural research (Von Braun et al., 2008b). These ‘best bets’
include programmes to revitalize yield growth in intensive cereal systems in Asia, increase
small-scale fish production, address threatening pests like virulent wheat rust, tackle cattle
diseases such as East Coast Fever, breed maize that can be grown in drought-prone areas, and
scale up bio-fortified food crops that are rich in micronutrients.
The technological change initiatives must be accompanied by institutional innovations for
sustainable actions. These include a comprehensive re-vitalization of agricultural extension
and service systems in small farm agriculture, sound cooperative and contract farming
arrangements that address economy of scale constraints, incentives for sustainable land and
water management, innovative arrangements that include agriculture into climate change
adaptation and mitigation regimes, and market and trade arrangements that facilitate fair and
open trade. The adequate governance of these institutional arrangements at national and
global level poses a major challenge for reaching a sustainable food and agriculture system in
the 21st century.
References
Cline, W.R., 2007. Global warming and agriculture: Impact estimates by country. Center for
Global Development and Peterson Institute for International Economics, Washington, DC.
FAO, 2008. Food Outlook. November, Food and Agriculture Organization of the United
Nations, Rome.
Fischer, G., et al., 2005. Philosophical Transactions of Royal Society B 360: 2067-2083.
GES, 2008. The crisis of water management. Global Economic Symposium 2008 session
handout. Available at http://www.global-economic-symposium.org/ges-2008/downloads/
GES%202008%20Session%. Schleswig-Holstein, Germany.
Von Braun, J., 2008. Food and Financial Crises: Implications for agriculture and the poor.
Brief prepared for the CGIAR Annual General Meeting, Maputo, Mozambique. December
1.
Von Braun, J., et al., 2008a. High food prices: The what, who, and how of proposed policy
actions. IFPRI Policy Brief. Washington, DC: International Food Policy Research Institute.
Von Braun J., et al., 2008b. International agricultural research for food security, poverty
reduction, and the environment: What to expect from scaling up CGIAR investments and
‘best bet’ programs. International Food Policy Research Institute, Washington, DC.
World Bank, 2007. World development report 2008: Agriculture for development.
Washington, DC.
3
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Agriculture and sustainable development in OECD countries:
A policy perspective
W.J. Legg
Trade and Agriculture Directorate,
OECD-Organisation for Economic Co-operation and Development,
2 rue André-Pascal, 75016 Paris, France
Contact: wilfrid.legg@oecd.org
Introduction
In the 30 member countries of the OECD – as well as many non-OECD countries – concern
with sustainable agricultural development has led to increasing attention to the role of policies
in improving environmental performance. Countries are also interested in identifying possible
future environmental problems associated with agricultural activities, and trying to better
understand the effects of different agricultural policy measures on the environment. More
recently there is heightened concern over the effect of external environmental events – in
particular climate change and variability risks – on the agricultural sector.
Many factors influence environmental performance – agro-ecological conditions, farmer
knowledge and behaviour, regulations, market forces, external events such as weather, climate
change and industrial pollution – and many of the effects are site specific and take time to
appear. Moreover, across countries there are very different histories and preferences with
regard to the nature, extent and mix of policy interventions in the agricultural sector.
Agriculture is a sector in which policy plays a significant role. Agricultural policies provide
monetary transfers that influence – directly or indirectly – on what and how much to produce,
where and under what conditions, while environmental regulations require farmers – either at
their own cost or with the aid of subsidies – to adopt certain practices or deliver particular
outcomes determined by governments. Overall, this leads to a complex web of incentives and
disincentives facing farmers, with an equally complex set of multiple environmental effects.
In most OECD countries, the dominant trend in recent years has been the gradual (and
sometimes limited) decoupling of farm support from agricultural commodity production and a
shift towards policy measures that do not require farmers to produce specific commodities in
order to be eligible for support (or any commodities at all); are targeted at specific
environmental objectives; or link environmental and income support measures (crosscompliance). Agricultural and associated trade policy reform in itself will have both positive
and negative impacts on the environment as the production incentives facing farmers change.
The challenge is to determine and move towards the level and mix of agricultural production
and practices that is both economically and environmentally efficient: production may be
economically efficient but does not deliver the ‘right’ amount of environmental outputs (or
vice versa). Finding the best balance between farming profitability, competitiveness and
resource conservation is a complex issue. There is a need to identify the existing market or
policy failures causing the problem, establish the extent to which markets could be created or
policy intervention is need to address the problem, define the most cost-effective policy
approaches and choice of instruments, and implement, monitor and evaluate policies.
Essentially, policymakers in most OECD countries have applied at least two broad sets of
objectives and instruments in this area – one relates to agricultural production and farm
incomes, the other relates to environmental performance. Given the integrated nature of the
4
Plenary Sessions
relationship between agriculture and the environment (joint agriculture and environmental
production, and dynamic interactions) such that targeting one set of objectives inevitably
leads to consequences for the achievement of the other, the appropriate policy instruments
cannot be evaluated in isolation. There are consequences for agricultural production where
environmental regulations constrain production. Payments that only compensate for the extra
costs involved in the provision of environmental services could be production-neutral.
Work in the OECD aims to help inform policy makers to design and implement effective
policy measures (that achieve desired objectives), which are also efficient (giving best value
for money with least distortion to production and trade), and which will thus contribute to
sustainable agricultural development. The main conclusions from the work thus far – which is
based on a mix of conceptual, empirical and modelling studies – are the following:
• An overall improvement in the environmental performance of agriculture, but masking a
•
•
•
•
•
•
•
number of severe local and regional problems, while future global pressures on land and
water resources, including climate change, will be significant.
While environmental improvements have resulted from agri-environmental policy
measures, in a number of OECD countries unconstrained commodity production-linked
support policies are still pulling in the opposite direction.
Environmental improvement in agriculture has involved costs that would be lower in the
absence of commodity production-linked support measures, which provide incentives to
adopt environmentally harmful practices and input use, and expand commodity production
to environmentally sensitive land.
Although environmental cross-compliance conditions associated with commodity
production-linked payments to farmers are mitigating some environmental pressures, they
are not necessarily the most effective or efficient ways of reducing environmental
pressures,
Research is underway to analyse the effectiveness and efficiency of different agrienvironmental policy measures in order to identify the policies and market actions that
would achieve the same or better environmental outcomes at lower cost.
Establishing reference levels to distinguish the conditions under which farmers should be
accountable for the environmental costs they impose on society (polluter-pays-principle)
from those under which farmers should be rewarded for the provision of non-remunerated
environmental benefits they provide to society (provider-gets-principle) is crucial.
Agro-ecological conditions and public preferences vary across and within countries, and a
variety of different policy measure, market creation and voluntary initiatives, and
technological developments are appropriate to deal with environmental concerns.
The current financial, economic and food price crises significantly increase the challenge
to produce food sustainably, in particular given longer term demographic, income and
climate change trends.
Reference
OECD, 2008. Environmental performance of agriculture in OECD countries since 1990.
Paris, France.
5
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated assessment of agricultural systems:
On integrated science and science integration
M.K. Van Ittersum, F. Ewert, E. Andersen, F. Brouwer, T. Heckelei, J. Alkan Olsson,
J. Wery, J.J.F. Wien
Plant Production Systems, Wageningen University,
P.O. Box 430, 6700AK Wageningen, The Netherlands
Contact: martin.vanittersum@wur.nl
Introduction
The recent fluctuations in prices of agricultural commodities have distinct impacts on
agricultural systems at global, continental, regional, farm and field level. They not only affect
the economy of agricultural systems, but can also have substantial environmental and social
implications. These price fluctuations underpin two main requirements for ‘integrated
assessment (IA) of agricultural systems’ in general: IA must provide integrated analysis at
multiple scales and of economic, environmental and social factors. The dynamics in food
prices and, for instance, the rise and decline in popularity of using biomass for biofuel also
point at a third requirement: research tools must be flexible, as issues can change within short
time periods, and research tools fit for only one purpose can become outdated rapidly.
These three requirements challenge research for integrated assessment of agriculture and
point at the need for generic and flexible tools. This contribution will illustrate some of the
advances achieved over the past four years in a large integrated and EU-funded research
project (SEAMLESS), involving 30 institutes and a large group of scientists with an
agronomic, environmental, economic, information technology or sociological background
(Van Ittersum et al., 2008). We present key methodological features of a computerized
framework for integrated assessment of agricultural systems. The paper demonstrates an
application, and reflects on some research challenges for interdisciplinary science. Finally, it
discusses the positioning of science-based IA frameworks and derived information in the
science-policy interface.
A component-based framework for integrated assessment
The SEAMLESS Integrated Framework (SEAMLESS-IF) allows integrated, ex-ante
assessments of agricultural and agri-environmental policies and technologies across a range of
scales, from field/farm to region and the European Union. SEAMLESS-IF integrates
relationships and processes across disciplines and scales which are conceptualized following
the paradigm of hierarchy theory (Ewert et al., 2009). The relationships and processes at
different levels of organization are modelled in so-called components. These components
include a modular, biophysical simulation model calculating agricultural production and
externalities at field level (APES); a bio-economic farm model quantifying the integrated
agricultural, environmental and socio-economic aspects of farming systems (FSSIM); and an
agricultural sector model (CAPRI) providing information on supply-demand relationships.
Various scaling methods have been used to link information from one level to another or to
simulate the feedbacks between levels of organization and processes. The framework uses a
European data base with data on soils, weather, farming systems, agro-management, prices
and sectoral accounts as well as a library containing indicators for economic, environmental,
social aspects organized in an indicator framework. Institutional indicators are included and
organized in a specific component for institutional compatibility assessment (PICA). The
conceptual linkage of components is facilitated through the use of ontologies and the
technical linkage through the use of the Open Modelling Interface (OpenMI). Two
applications have been used to test and improve the framework. One simulates impacts of
international trade liberalization proposals on the European Union and a second case study
6
Plenary Sessions
analyses the consequences of environmental policies at a regional and farm level. Additional
components, for instance focusing on landscape quality or on global market interactions have
been developed in the project or are available and can be integrated into SEAMLESS-IF.
Reflections on integrated science
The developed integrated framework follows one of the possible methodological pathways for
integrated assessment. The method focuses on integration of standalone components that are
strong in simulating specific processes and relationships, including crop and livestock
production and externalities, farm responses and supply-demand relationships. A benefit of
this approach is that it allows to structure the development of integrated assessment tools in
relatively independent components and to benefit from advances of science focusing on
specific parts of the system. All details in each of the components may not be needed for a
specific application but do provide a degree of flexibility needed for a broad range of
applications (Ewert et al., 2009).
A key question is whether this approach allows an adequate system representation for
specific problems, and does capture the most relevant feedback mechanisms and interactions
which may occur at the interface of subsystems, e.g. between crops and livestock, between
different fields and landscapes, or between farms and markets. Here, further testing remains to
be done with interesting science. A particular challenge of this research method is the high
data demand. Three routes or combinations of them are available: (1) using statistical
sampling procedures; (2) developing science-based rules to ‘generate’ certain crucial but
missing data; (3) using available pan-European data.
Structuring research tools in components also allows structuring the workforce and
European and international collaboration. The teams developing components can work
relatively independent and may consist of specialists with sufficient so-called T-shaped, that
is integrative, skills (Bouma, 1997). At the same time, a team of adequate size needs to have
the necessary conceptual and technical skills for integration and linkage of components. The
relatively flexible linkage of components puts high demand on state-of-the-art information
technology (IT) which is not present in all teams. The experience of SEAMLESS showed the
crucial importance of having a team with unprecedented interdisciplinary skills and which is
motivated to invest in IT.
Reflections on integration of science and policy
Despite the advances in developing generic and flexible research tools in SEAMLESS and
many other research projects, science cannot take for granted that its results will be heard and
used in the policy domain. It remains very important to understand the aims and role of
integrated analysis within decision-making processes for which science-based information is
just one relevant ingredient. Classifications and ideotypes, of foresight studies (Van Ittersum
et al., 1998), of the role of science in policy and politics (Pielke Jr, 2007) and of the
institutional interfaces of science and policy in so-called boundary arrangements (Sterk et al.,
2009) can assist in defining and contrasting the precise aims of integrated assessment and in
further working on an effective science-policy interface.
References
Bouma, J., 1997. Geoderma 78: 1-12.
Ewert, F., et al., 2009. Environmental Science & Policy. (in review)
Pielke Jr, R.A., 2007. The Honest Broker. Making Sense of Science in Policy and Politics.
Cambridge University Press, 188 pp.
Sterk, B., et al., 2009. Land Use Policy 26: 434-442.
Van Ittersum, M.K., et al., 1998. Agricultural Systems 58: 309-330.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
7
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Two decades of Integrated Assessment tools: A promising way forward
Jan Rotmans
DRIFT: Dutch Research Institute For Transitions,
Erasmus University Rotterdam, P.O. Box 1738,
3000 DR Rotterdam, The Netherlands
Contact: Rotmans@fsw.eur.nl
Integrated Assessment (IA) is the multi- or interdisciplinary process of structuring knowledge
elements from various scientific disciplines in such a manner that all relevant aspects of a
social problem are considered in their mutual coherence for the benefit of decision-making.
IA has a heuristic character, being a quest in triplicate: for causalities, coherence and
commonalties. IA is therefore an iterative, reciproque process, where integrated insights from
the scientific community are conveyed to the decision-making community, and experiences
and insights from decision-makers are taken account of in the integrated analyses. Although
active participation is no prerequisite, more and more people in the IA-community are
convinced of the vital importance of involvement of stakeholders in the IA-process. It is
generally acknowledged that involvement of non-scientific and practical knowledge and
expertise, valuation and preferences in the form of direct involvement of actors will enrich the
process of Integrated Assessment.
In general, two types of Integrated Assessment methods can be distinguished: analytical
methods and participatory methods. While analytical methods are often rooted in natural
sciences, participatory methods, also labelled as interactive or communicative methods, stem
from social sciences. The group of analytical methods is reasonably well-defined and
basically includes model analysis, scenario analysis and risk analysis. Their commonality is
that they provide analytical frameworks for representing and structuring scientific knowledge
in an integrated manner. The group of participatory methods, however, involves a plethora of
methods, varying from expert panels, delphi methods, to gaming, policy exercises and focus
groups. They have in common that they aim to involve non-scientists as stakeholders in the
process, where the assessment effort is driven by stakeholder-scientist interactions.
Although it is hard to generalize, it is acknowledged that over the last two decades tools for
IA have evolved significantly: from supply-driven to more demand-driven, from monodisciplinary to multi-disciplinary, from technocratic to participatory, from objective to
subjective, from certainty to uncertainty and from predictive to explorative. As Wynne &
Shackley (1994) put it: we used to build truth machines, but now we build heuristic tools.
Within this changed IA-context we now face the ultimate challenge is to perform IA for
sustainable development, this is what we call Integrated Sustainability Assessment (ISA), not
to be confused with Sustainability Impact Assessment (SIA). Sustainable development is a
complex, multi-dimensional phenomenon, with a breadth and depth that cannot be fully
covered by the current portfolio of IA-tools. We, therefore, need a new generation of ISAtools, that is rooted in a new paradigm and that can handle the complexity and multiplicity of
sustainable development, in terms of multiple scales, multiple domains and multiple
generations. To build-up a new generation of IA-tools requires time, resources and a clear
investment strategy, it is proposed here to follow a two-track strategy: find new ways to use
the current portfolio of IA-tools as efficiently and effectively as possible, while at the same
time developing the next generation of IA-tools for sustainable development, called ISAtools.
8
Plenary Sessions
References
Rotmans, J. & H. Dowlatabadi, 1998. Integrated assessment of climate change: Evaluation of
models and other methods. In: Eds S. Rayner & E. Malone, Human choice and climate
change: An international social science assessment. Batelle Press, USA.
Rotmans, J., 2006. Integrated Assessment 6(4): 35-57.
Weaver, P. & J. Rotmans, 2006. International Journal of Innovation and Sustainable
Development 1(4): 284-303.
Wynne, B. & S. Shackley, 1994. Environmental models: Truth machines or social heuristics.
The Globe 21, 6-8.
9
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
10
Plenary Sessions
Enhancing developing world agricultural performance:
Getting beyond the current plateau
Prabhu Pingali
Deputy Director, Agriculture Development,
Bill and Melinda Gates Foundation
Contact: Prabhu.Pingali@gatesfoundation.org
Developing world agriculture has witnessed unprecedented levels of growth over the past four
decades yet serious problems remain. While food supply outpaced population growth rates,
hunger and poverty persist in large parts of the developing world, with particular
concentration in sub-Saharan Africa. Technology induced productivity growth has generally
not reached beyond the ‘bread basket areas’, causing a widening welfare gap with the lagging
regions, even in emerging economies. Moreover, ‘Green Revolution’ productivity gains have
shown clear signs of stagnation and potential reversal due to intensification induced resource
degradation, resource competition and the emerging consequences of climate change.
Technology solutions for addressing the problems of lagging regions, especially drought
tolerance, as well as, correcting for resource degradation are hampered by the changing locus
of innovation from the public to the private sector. Incentives for enhancing productivity
growth have been hampered by the general decline in public investments for the agricultural
sector, as well as, a continuing policy bias towards urban areas. The problems of the
developing country agriculture are as complex today as they were four decades ago, if not
more so. However, we are better positioned to address these problems today than in the past
due to a vastly expanded knowledge base and a set of viable solutions. What’s needed is
better ground level information on the constraints to productivity growth and an enhanced
ability to adapt solutions to these problems. Agriculture can be an effective pathway out of
rural poverty, only if there is a long term and sustained commitment by the state to
productivity enhancement, there are no quick wins.
11
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme A
Methodology for integrated assessment
Session A1: Methods for integrated assessment of
agricultural systems and land use
Session organizers:
Irina Bezlepkina
LEI, Wageningen UR, The Netherlands
Stefan Sieber
IPTS, Joint Research Centre, Seville, Spain
ZALF, Germany
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Scenario studies for agriculture, land use and landscape in Europe:
A comparison1
Katharina Helming, Ignacio De la Flor Tejero, Martin K. Van Ittersum, Peter Nowicki,
Willem Rienks, Axel Volkery
Leibniz-Centre for Agricultural Landscape Research (ZALF), Germany
Contact: khelming@zalf.de
Introduction
Land use and agriculture are key human activities, which, through the exploitation of natural
resources, foster socio-economic development and alter structures and processes in the
landscape. At the European level, the Sustainable Development Strategy stresses the need for
real integration of economic, environmental and social issues across policy areas. In
particular, land use policies aim to promote sustainability pathways of natural resources use
and rural development through the decoupling of economic growth from environmental
degradation while supporting social cohesion. In recent years, foresight and ex-ante
assessment studies of land use and agriculture have emerged that place land use decisions into
the logical chain of driving forces and impacts. The purpose of these studies was to provide
decision support to land use and agricultural policy making through the anticipation of
possible effects of land use and agricultural change scenarios on sustainable development
targets. While similar in general purpose and approach, the studies differed in their approach
to the spatio-temporal systems and in their use of driving forces, scenarios, modelling
approaches, etc. The objective of this paper was to conduct a comparative analysis of selected
forecasting studies for land use change in Europe, namely SEAMLESS (Van Ittersum et al.,
2008) SENSOR (Helming et al., 2008), EURURALIS (Eickhout & Prins, 2008), SCENAR
2020 (Nowicki et al., 2007), and PRELUDE (EEA, 2007).
Methods
The DPSIR (Drivers, Pressures, States, Impacts, Responses) was used as a general framework
for comparison. Although originally focused on the environment and the effects of humans on
it, the DPSIR framework is easily applicable to the cases of land use and agriculture and helps
structuring complex analytical chains. Based on the DPSIR chain and integrating further
structuring elements, seven parameters were selected to compare the analytical design of the
assessment studies. These were: (1) purpose of the project, (2) spatial and temporal scale
(grain and extend), (3) driving forces considered, (4) scenario design, (5) modelling approach,
(6) impact analysis (indicator frameworks), (7) sustainability interpretation.
Results
Although considerably different in size, complexity and purpose, a comparison of the studies
using the above described criteria proved possible and revealed a number of similarities. All
projects aimed at providing ex-ante assessments of future development trends as affected by
human decisions. They were based on scenario studies and used indicator systems to address
environmental, social and economic impacts. Despite the similar purposes, each study
revealed its particular strengths and unique features: SCENAR2020 and EURURALIS, two
meta models, were particularly dedicated to the policy environment of agriculture and rural
development in Europe. For the case of SCENAR, the high degree of policy relevancy was
also manifested with the startling short time span of less than one year they required for the
analyses. PRELUDE, an expert system, was based on a very sophisticated approach to
stakeholder involvement in scenario design, which also allowed for visionary, antithetic
1
Keynote presentation
14
Session A1: Methods for integrated assessment of agricultural systems and land use
Table 1. Foresight and assessment studies included in the comparative review.
Time
Spatial System
Issue
Purpose
Study
horizon
SCENAR
2020
Europe at NUTS2/3
and regional clusters
PRELUDE
2035
Europe at 10min grid
Landscape and
environment
EURURALIS 2030
Europe at NUTS2/3
and regional clusters
Agriculture and rural
sustainability
SEAMLESS
SENSOR
Agriculture and rural
areas
Nested system from
farm level to
Agricultural Systems
2015–2025
regional, national,
European and Global
Land use (agriculture,
Europe at NUTS2/3 tourism, forestry, nat.
2025
and regional clusters conservation, transport,
and energy infrastructure)
Future of EU
agriculture and the
rural economy
Visions on future
landscape
developments
Foresight rural areas
Integrated
assessment of
agricultural systems
Ex-ante impact
assessment of
European policies
anticipations of possible future landscape developments. It was particularly useful to trigger
societal debate on future development targets of the rural environment. SEAMLESS and
SENSOR involved far more resources than the other projects allowing for differentiated and
complex approaches to the assessments and to the development of fully dynamic modelling
systems. SEAMLESS, involving the coupling of comprehensive standalone models,
emphasised on sophisticated scaling approaches in integrated modelling leading from detailed
simulations of single farming systems up to global interactions in agricultural decision
making. SENSOR with its meta-model SIAT emphasised on a cross-sectoral approach
integrating a variety of land use sectors. It also developed new approaches to sustainability
interpretation of environmental, social and economic impacts. Both projects developed impact
assessment tools for further use in policy making.
References
Nowicki, P., et al., 2006. SCENAR 2020. European Commission, Directorate-General
Agriculture and Rural Development, Brussels.
EEA, 2007. PRELUDE. Technical report No. 9. Copenhagen, Denmark.
Eickhout, B. & A.G. Prins, 2008. EURURALIS 2.0, Wageningen UR and Netherlands
Environmental Assessment Agency (MNP) Bilthoven, The Netherlands.
Helming, K., et al., 2008 Sustainability impact assessment of land use changes. Springer
Berlin, Heidelberg.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
15
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrating social preferences and spatial analysis to optimize the use of
marginal agricultural lands1
M. Arriaza, O. Nekhay
Centro IFAPA ‘Alameda del Obispo’, Apartado 3092,Córdoba, Spain
Contact: manuel.arriaza@juntadeandalucia.es
Introduction
The land management of olive plantations in Southern Spain is analysed from a threefold
point of view: economic, social and environmental sustainability. The economic approach
addresses the viability of the farming activities, the social criterion aims to prevent population
loss in rural areas, whereas the environmental analysis involves the consideration of the
reduction of soil erosion, the improvement of biodiversity, the control of fire risk and the
provision of quality agricultural landscapes. The study analyses how the type of management
of the olive grove (conventional, integrated and organic) or its abandonment contributes to the
achievement of these objectives.
First, we asked to the local and non-local population to weigh these three aspects of
sustainability via a classical Analytic Hierarchy Process (AHP) questionnaire. Second, a
group of experts weighed how the different land management contributes to these objectives.
Finally, the Geographical Information Systems (GIS) provide the territorial integration of
both weighing exercises. For each objective, the best management alternative is selected for
each pixel, then the weight of the objective in the pixel determines which one dominate the
rest.
The case study is the olive (Olea europaea L.) plantations of the mountain areas of
Andalusia (Spain). These plantations are located in highly sensitive places from an
environmental point of view and, after the decoupling of the subsidies, a large percentage of
them face the risk of abandonment, therefore, the study aims to optimize the type of land
management and to spot the most suitable places for restoration. The municipality of Montoro
is located in the province of Cordoba in Southern Spain. The territory enjoys typical
Mediterranean continental climate conditions with irregular precipitation distribution during
the year (less than 600 mm yr–1). Municipality of Montoro represents a variety of agricultural
ecosystems (pasture, olive groves and annual crops) and forest/shrub natural vegetation near
agricultural areas. Its 58,103 hectares are divided into olive plantations (27.5%), arable crops
(6.6%), forest and natural vegetation (37.5%), natural park (20.9%), Mediterranean pasture
(6.7%) and other crops (0.8%).
Methods
The AHP technique uses expert judgments as inputs for weighing alternatives (Saaty, 1980).
In our study, expert knowledge determines the relative importance of each criterion of the
optimizing function. Empirical studies that have used multi-criteria evaluation methods for
the solution of spatial problems include that of Malczewski (1999), which brought together
two approaches developed much earlier: Multi-Attribute Utility Theory (MAUT) and the use
of Geographical Information Systems (GIS) as a platform for representing the spatial
dimension of the problems. A large number of studies have since adopted Malczewski’s
approach, including Thirumalaivasan et al. (2003), Ayalew et al. (2005) and Neaupane &
Piantanakulchai (2006), this last dealing with different fields of landscape assessment process.
6
The utility function to be optimized takes the form: : Un,g= ∑ Agi ·Pi ·Fni ; where n represents each
i =1
1
This research has been financed by INIA through the research project RTA04-086.
16
Session A1: Methods for integrated assessment of agricultural systems and land use
pixel of the territory (10×10 m), g is the type of management or abandonment, Agi represents
the importance of the type of management g for objective i, Pi is the weight that Society
assigned to objective i and Fni represents the potential suitability/risk (between 0 and 1) of
pixel n for objective i. Then, the highest U determines the optimum alternative in the final
map.
Results and discussion
The maps in Figures 1–3 show the sequence of the analysis:
Figure 1. Initial agricultural
land use.
Figure 2. Aggregation of
maps.
Figure 3. Optimized land use.
From a methodological point of view, the use of the ideal mode of AHP avoids the bias that
arises from the weighting of elements in each layer when the number of elements differs in
individual layers. Consideration of negative priorities instead of their transformation into
small positive priorities also gives the model more internal consistency and produces more
accurate results that are in accordance with the preferences revealed by the experts. Also, the
data requirements of this approach are less rigorous than those of classical statistical models
based on historical data. The sensitivity analysis indicates that the model is stable on the basis
of the results of the four alternative scenarios considered. The simulation carried out in the
study identifies the edges of major agricultural areas (mostly olive groves), areas of natural
vegetation and areas adjacent to Natural Park with oaks as being most suitable for wildlife
habitat restoration. These results have similarities to those obtained by other researchers on
biodiversity, based on either individual or groups of species.
References
Ayalew, L., et al., 2005. Engineering Geology 81: 432-445.
Malczewski, J., 1999. GIS and Multicriteria Decision Analysis. John Wiley & Sons, NY.
Neaupane, K.M. & M. Piantanakulchai, 2006. Engineering Geology 85: 281-294.
Saaty, T.L., 1980. The analytic hierarchy process: Planning, priority setting, resource
allocation. McGraw-Hill, NY.
Thirumalaivasan, D.M., et al., 2003. Environ. Model. Soft. 18(4): 645-656.
17
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Understanding land use change with the aid of an integrated modelling
framework: A case study on rural south-west China
K. Berkhoff, S. Herrmann
Department of Environmental Planning, Leibniz University Hannover,
Herrenhaeuser Strasse 2, 30419 Hannover, Germany
Contact: berkhoff@umwelt.uni-hannover.de
Introduction
Cash crop production and rapid economic growth have generated huge impacts on the natural
resources, ecosystems and livelihoods of ethnic minorities in the ecologically fragile upland
areas of the Naban river catchment. The region is located in the south of Yunnan province in
China and is part of the Indo-Birma hotspot of biodiversity. The production of natural rubber
dominates the regional economy and increasingly replaces traditional, diversified and
sustainable land use practices. This situation requires an integrated approach to land use
planning. The Sino-German research project LILAC1 supports this process by a modelling
framework dealing with the multiple impacts of rubber cultivation: loss of biodiversity,
increasing income, farmers’ dependency on the market, etc. By ‘integrated modelling
approach’, we mean the coordinated application of an agro-economic, ecological and social
model which altogether interact with a land allocation model via defined interfaces (no
dynamic coupling). We call this approach the NabanFrame modelling framework.
Due to the strong social implications of changes in land use the major focus is laid on the
appropriate definition and integration of the social drivers initiating and underlying the
current land use pattern. The presentation describes how the land allocation model
(CLUENaban) is parameterized to model the status quo of land use in the study area. For model
calibration, narrative interviews with individual stakeholders are used. They were conducted
in the study area with regard to knowledge exchange and the adoption of innovations (Aenis
et al., 2008).
Modelling framework (NabanFrame)
The objective of the modelling approach is to provide policy support for land use planning;
the addressees are decision makers of the nature reserve and regional planning authorities.
NabanFrame consists of three disciplinary models describing the issues of ecology, economy
and sociology and further the land allocation model CLUENaban which has an integrating
function. NabanFrame follows a three steps approach of data preprocessing and land use
demand negotiation, then land allocation (ruled by CLUENaban) and finally the evaluation of
the resulting land use maps (see Figure 1). In all three steps information of the disciplinary
models is considered. E.g., in the pre-processing phase yearly demand for the land use types
is adjusted between agro-economic and ecological demand. In the next step of NabanFrame,
all models deliver location factors for the land allocation module. Finally, they are applied to
conduct the impact analysis in the concluding post-processing step.
Incorporation of social information into CLUENaban
The land use allocation model CLUENaban is based on the CLUE-S model (Verburg et al.,
2002). CLUE-S has been designed to simulate land use change using empirically quantified
relations (regression analysis) between land use and its driving factors combined with the
modelling of competition between land use types (dependent on location suitability,
neighbourhood setting, conversion elasticity and a demand-related iteration variable).
1
‘Living Landscapes China’, a project supported by the German Federal Ministry of Education and Research
under promotional reference: 0330797A
18
Session A1: Methods for integrated assessment of agricultural systems and land use
Landscape index
f rom FRAGS TATS
Pre-Processing
Lan d use map
Forest
Rubber
Shrubland
Bamboo
Grassland
Paddy rice
Field
Water
Remote sensing
data
(Landsat,
IKONOS)
Agro-economic
demand
Location factors
Spatial policies
Conversion elasticity
To be considered:
Const ant demand rice paddies (policy)
S t ate quot a system f or rubber
Permit for commercial t imber cut ting
A utonomous change
No rubber > 1200 m
Land allocation
algorithm
Neighbourhood
settings
Traditio nal land use
sequen ces relat ed
t o eth nic group
Location specific
preference addition
Incentives (e. g. f or t eak)
Rubber spreading
S ocial mo del
Rubber spreading
Specific land
use conversion
trajectories
Conversion settings
From regression
Corridors/ barriers
(interconnecte d biot opes)
Special biot opes
Protected
habitat areas
Core zone – no use
Buf fe r zone – litt le anthr. use
E xperiment al zone – anth r. use
Location suitability
Yearly demand
per land use type
Agro-econ. m odel
Spat ial rest rictions
relat ed t o ethnicity
Elevation
S lope
E uclidia n dist ance to villages
Euclidian distance to city
Accessabilit y
P opulat ion density
To tal labour force
Negot iation
Ecol ogical mo del
CLUENaban
Farm type
I ncludes e.g. :
A vailable labour
Farm st ructure
Crop t ypes
Soil qual ity
Climat e
Geolog y
E rosion
Water availa bility
Ecological
demand
NabanFrame
Status quo scenario
2001-2007
Land t enure (ownership/
lease; stat e/private)
Mobilit y level
E thnic group
Innovation level
Family st ructure/net works
De gree of subsist ence
Succession t raject ories
A nalysis of
historic land
conversions
Post-Processing
Evaluation of simulated
land use maps, impact on
Social
factors
Ecology
Biodiv. ind icat ors
Species
abundances
Agroeconomy
Figure 1. The NabanFrame modelling framework, operationalized for the status quo scenario.
Colors refer to model inputs (light grey: social model, medium grey: ecological model, dark
grey: agro-economic model). Empty boxes are not yet operationalized, white boxes are in the
responsibility of the land allocation model.
It is assumed that locations are assigned to the land use type with the highest total probability.
In the presentation, the setup of CLUENaban will be illustrated as it was chosen for modelling
the status quo of land use, focusing on the incorporation of social issues into the modules of
CLUENaban. Data from questionnaires has been referenced to spatial entities in the study area
before it is integrated in CLUENaban. The simulation runs from 2001 to 2007 (referred to as
“status quo”) for which a detailed land use map has been generated from IKONOS 2 remote
sensing data. In contrast to the questionnaire information, the narrative interviews serve the
purpose of calibrating the parameterization of the status quo scenario: Can the processes and
decisions reported in the interviews be recovered in the modelled land use map?
Modelling scenarios of land use change
The understanding of the relevant processes that drive land use change in the study area is an
indispensable condition for the subsequent modelling of scenarios. Two scenarios for the
study area will be modelled by the NabanFrame modelling framework: a scenario of
sustainable development and a business as usual scenario; the time horizon of the scenarios
will be the years 2007 until 2020. The land use maps resulting from the scenario modelling
exercise as well as the results of the impact analysis conducted by the agro-economic, ecological and social model provide the basis for discussion with the regional decision makers.
References
Aenis, T., et al., 2008. Land-use related innovations and knowledge flows (LILAC subproject
SOSI-A). Unpublished document.
Verburg, P.H., et al., 2002. Environmental Management 30: 391-405.
19
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Concept and application of a DLG sustainability standard on the farm level
O. Christen
Institute of Agricultural and Nutritional Sciences,
Chair of Agronomy and Organic Farming,
University of Halle-Wittenberg, 06099 Halle, Germany
Contact: olaf.christen@landw.uni-halle.de
Introduction
To assess practical farms according to the principles of sustainable farming requires a
comprehensive approach taking into account environmental, economic and social aspects of
production. In general such an assessment is possible on different levels, starting at the crop
or crop rotation level, the farm level, or the regional level. In our approach we focus on the
farm level, because the farm level is relevant for all major decisions regarding varieties, input,
crop rotation.
In cooperation with University Freising-Weihenstephan and the German Agricultural
Society (DLG) we have developed an assessment system on the farm level based on the
computer model REPRO, which has been developed at the University of Halle-Wittenberg.
With this model it is possible to quantify all major energy and material flows on the farm and
on the field level as well as relevant information on economic effects of the different
decisions taken on the farm. The aim of this project is to establish a standard for a
certification scheme run by the DLG. In this paper we will focus on the ecological indicators
and the results on a number of pilot farms (Deumelandt & Christen, 2008).
Materials and methods
The assessment of sustainability on the farm level is mainly based on the use of pressure
indictors, which have been established in a number of previous projects (Heyer et al., 2003;
Rücknagel et al., 2006). The following indicators are calculated: Nitrogen-balance,
phosphorus-balance, SOM-balance, energy intensity, GHG-emissions, pesticide-index, soil
compaction, soil erosion, biodiversity, proportion for conservation agriculture. All
calculations for the ecological indicators are conducted with the software REPRO, which
allows to normalize the different scales to
figures between 0 and 1 (see Figure 1). The
threshold for a sustainable system is 0.75 in
this system. A farm is rated as sustainable
if all three areas average above 0.75. The
basis for all calculations is based on
husbandry data as well as details on soil
and climate conditions.
Results and discussion
The results for the five pilot farms are
given in Table 1. On average all five farms
fulfilled the criteria set in this project,
however, single indicators on some farms
were well below the standard of 0.75.
Especially the pesticide index and the
proportion of area for nature conservation
projects were below 0.75 in a number of
Figure1. Function to normalize N-balance.
cases.
20
Session A1: Methods for integrated assessment of agricultural systems and land use
Table 1. Normalized indicators for the five pilot farms.
Indicator
Scale
Farm A Farm B
N-balance
kg ha–1
0.82
0.93
–1
P-balance
kg ha
1.00
0.79
SOM-balance
kg C ha–1
0.70
1.00
GHG
CO2 equ GJ–1
0.71
0.66
Energy intensity
MJ GE-1
1.00
1.00
Pesticide index
0.47
0.14
Biodiversity
0.90
0.63
Conservation
0.35
0.55
Erosion
t ha–1 yr–1
0.89
Soil compaction
0.98
0.88
Average
0.77
0.75
Farm C
0.97
0.85
0.93
0.66
1.00
0.59
0.62
0.29
1.00
0.88
0.78
Farm D
1.00
0.70
1.00
0.49
1.00
1.00
0.89
0.36
0.85
0.75
0.80
Farm E
0.88
0.88
1.00
0.85
1.00
0.00
0.70
0.42
0.98
0.75
Further work is now required to include a greater number of farms in the assessment and to
aggregate the different indicators in various ways based on their environmental relevance.
References
Deumelandt, P. & O. Christen, 2008. Sugar Industry 133: 9-18.
Heyer, W., et al., 2003. Agriculture, Ecosystems and Environment 98: 453-461.
Rücknagel J., et al., 2006. Soil & Tillage Research 92(1-2): 213-220.
21
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
An integrated assessment of global changes in crop-livestock systems to
2030: Implications for policy development
M. Herrero, P.K. Thornton, A. Notenbaert, S. Msangi, S. Wood, R. Kruska, J. Dixon,
D. Bossio, J. Van de Steeg, H.A. Freeman, B. Gerard
International Livestock Research Institute, P.O. Box 30709-00100, Nairobi, Kenya
Contact: m.herrero@cgiar.org
Introduction
The world’s population is predicted to increase by 50% over the next quarter of a century to
reach 9 billion by 2030. During this period, and if the livestock revolution fully materializes,
in developing countries there is likely to be a rapid increase in demand for livestock products,
driven by increasing urbanization and rising incomes. On top of this, the impacts of a range of
driving forces such as water availability, climate change, technological innovations on
smallholder crop and livestock production may be substantial. The result of these drivers is
that smallholder farms will inevitably change. The challenge is to ensure that the resourcepoor, mixed crop-livestock, smallholder sector, which currently provides the majority of milk
and meat in the tropics, is able to take advantage of the opportunity to meet the increased
demand for these products. To do so the sector will need to intensify, but at the same time it is
vital that this does not compromise household food security, sustainable natural resource
management or rural livelihoods. This study attempts to find policy alternatives for the above
by using a range of coupled integrated assessment models and spatial disaggregation methods.
Methods
The framework for the study was based on the framework for the Millennium Ecosystem
Assessment (MA, 2005) and subsequently used for other major assessments like the IAASTD
(2008). We used the IMPACT-Water model (Rosegrant et al., 2005) and a spatially
disaggregated crop-livestock systems classification (Kruska et al., 2003) for looking at
alternative scenarios of change in mixed crop-livestock systems. We built upon the results of
the IAASTD (2008). The scenarios we used were the reference scenario, which tries to mimic
business as usual conditions of growth in agriculture, incomes, population and others.
Additionally we investigated what would be the consequences of increased demand for
biofuels, increase expansion of irrigation to produce more food and feed and what would
happen if the demand for livestock products increased.
Results and discussion
The results of this study have several policy implications and are present in full in (Herrero et
al., 2008). Some of the main findings are as follows:
- Mixed intensive systems in the developing World are under significant pressures. These
pressures are larger in some systems than in others but are all caused by the rising demands of
the human population and its income shifts and rates of urbanization. For example, mixed
intensive systems in South Asia are reaching a point where production factors are seriously
limiting production as land per capita decreases significantly. Significant trade-offs in the use
of resources (land, water, nutrients) exist in mixed systems, especially as the demands for
biomass for food, feed and energy increase.
- Important productivity gains could be made in the more extensive mixed rainfed areas.
Resources constraints in some land-based mixed intensive system are reaching are a point
where livestock production and environmental degradation may have deleterious impacts on
humans. In more extensive systems, with less pressure on the land yield gaps of crops and
livestock are still large. Pro-poor policies and public investments in infrastructure will be
required to create system of incentives, reduce transaction costs and improve risk
22
Session A1: Methods for integrated assessment of agricultural systems and land use
Figure 1. Conceptual framework for studying the impacts of drivers of crop-livestock systems
change (Adapted from MA, 2005).
management in these systems. Integration of production in these systems to supply agroecosystems services (feeds, food, etc) to the more intensive systems should be promoted.
- The livestock revolution – at least from ruminants – could potentially exclude the poor in
terms of the benefits of consumption of meat. If green fodders became scarce because of land
and water shortages, and more grains were fed to ruminants to match production, this is likely
to increase the prices of animal products further, thus bypassing the abilities of the poor to
consume more milk and meat. This would present significant challenges in mixed systems,
particularly in Asia.
References
Herrero, M., et al., 2008. Drivers of change in crop-livestock systems and their potential
impacts on agro-ecosystems services and human well-being to 2030. International
Livestock Research Institute, Nairobi, Kenya, 101 pp.
IAASTD, 2008. The International Assessment of Agricultural Science and Technology for
Development. Island Press, Washington DC.
Kruska, R.L, et al., 2003. Agricultural Systems 77: 39-63.
MA, 2005. The Millennium Ecosystem Assessment. Ecosystems and Human Well-being:
Scenarios, Volume 2. Island Press, US, 2005, 560 pp.
Rosegrant, M.W., et al., 2002. World Water and Food to 2025: Dealing with Scarcity. IFPRI,
Washington DC. http: //www.ifpri.org/ pubs/books/ water2025book. htm.
23
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The CAPRI and GTAP link:
Linking a partial and a global equilibrium model
T. Jansson, M. Adenäuer, M. Kuiper
Nussallee 21, D-53115 Bonn, Germany
Contact: marcel.adenaeuer@ilr.uni-bonn.de
Introduction
Many of the most interesting current research questions in economics span issues from the
very small and detailed to the very large and abstract, or require the interdisciplinary study of
relations between fields that are causally wide separated and generally not studied together:
What are the implications of general trade liberalization on the physical environment in
certain rural regions? Or what do regional differences in policy implementation imply at the
global level? Such questions stress, in the absence of “The Great Model For Everything”, the
trade-off between generality and depth, i.e. between covering many aspects of the system
studied and details of certain interesting components. This dilemma may be resolved by
linking specialized models to exploit their different strengths, as done within SEAMLESS
(Van Ittersum et al., 2008). This paper refines the link between the general equilibrium model
GTAP (offering an economy-wide perspective; Hertel, 2004) and the partial agricultural
model CAPRI (offering a detailed model of the agricultural production and policies, Britz et
al., 2007) which was developed in Jansson et al. (2008b). We aim to illustrate the potential of
the linked system by analysing a multi-lateral trade reform, where results are computed for
regional nitrate surpluses and agricultural incomes in the EU as well as welfare effects for the
world. The results will indicate the interdependence between regional and global economies
and regional physical impacts. This level of analysis is not possible with either model by
itself. We firmly believe that this kind of analysis will be increasingly important to address
future research questions.
Methods
The linking approach relies on using the variables and equations of the models as they are,
introducing the link by shocking parameters already present in the models. This is similar to
Jansson et al. (2008a) but contrasts with for example Grant et al. (2006), which replace the
original GTAP equations by iso-elastic approximations in the form of supply functions. By
maintaining the original model equations, maintenance of the core models is greatly
facilitated, and the links can be adopted for other versions of CAPRI and GTAP with less
effort. In addition, auxiliary facilities like data exploration tools and the welfare computations
in GTAP will keep functioning as in the stand alone applications. The link works with an
iterative exchange of input parameters to each model as visualized here:
D P:
( P ,S , D ,T )
C o n te n t s o f D G:
W
I n p u t p r ice ve c to r f o r C A P R I
M
C o n su m e r e xp e n d itu r e
S H IF T
C o n te n t s o f D P:
P
A g . in p u t - a n d o u tp u t p rice in d ic e s
S
A g g r e g a te su p p ly o f a g r icu ltu r e
D
A g g r e g a te d e m a n d fo r a g ricu ltu r e
T
T ra d e f lo w s o f a g ricu ltu r a l g o o d s
CA P R I
GTAP
S H I F T : P ro g ra m t o co m p u te sh o c k f o r G T A P.
DG:
(W , M )
24
Session A1: Methods for integrated assessment of agricultural systems and land use
Starting by solving GTAP (bottom left), we particularly solve for the price vector W
containing prices of agricultural intermediate inputs, capital and labour, and the vector M of
consumer expenditures per country (aggregate). Those data are written to the dataset DG.
Next, CAPRI is solved, using W and M as exogenous variables (parameters). CAPRI
computes for the aggregate agricultural sector price indices of output P per region, total
supply S, demand D disaggregated into human consumption, processing consumption and
intermediate demand by agriculture itself, and trade flows T. This is written to the dataset DP.
Finally, the program SHIFT computes shocks for GTAP by shocking the agricultural
producers of GTAP so that they, in a partial setting, would replicate the outcome of CAPRI,
and similar for consumption and trade of agricultural goods.
Results and discussion
To assess the value-added of the linked system of CAPRI and GTAP, we analyse a simplified
trade liberalization scenario consisting of a reduction in tariffs according to a tiered formula
described in the Doha draft modalities for agriculture released on February 8, 2008 by the
WTO. Comparison with the results from stand-alone versions of the models provides a
benchmark to assess the value-added of linking the models. We limit the scenario to
agricultural liberalization only to gauge the effect of economy-wide results provided by
GTAP. Since CAPRI is an agricultural sector model it will not account for the impact of
liberalization of non-agricultural trade, which would be included in a more realistic Doha
scenario. This would however obscure the effects of the just the economy-wide feedbacks to a
change in the agricultural sector. Detailed results are not yet available.
References
Britz, W., et al., 2007. Description of the CAPRI Modeling System, Final report of the
CAPRI-DYNASPAT project; Universität Bonn, available on http://www.ilr1.unibonn.de/agpo/rsrch/capri/Final_Report_Model_Description.pdf
Grant, J.H., et al., 2006. Extending general equilibrium to the tariff line. Paper prepared for
the Ninth Annual Conference on Global Trade Analysis, June 15-17, Addis Ababa,
Ethiopia.
Hertel, T.W., 2004. Global trade analysis. Cambridge University Press, UK.
Jansson, T., et al., 2008a. Cross sector land use modelling framework. In: Eds K. Helming, et
al., Sustainability impact assessment of land use changes. Springer, Berlin.
Jansson, T., et al., 2008b. Getting the best of both worlds? Linking CAPRI and GTAP for an
economywide assessment of agriculture. Paper presented at the Eleventh Annual
Conference of the Global Trade Analysis Project (GTAP) “Future of Global Economy”,
Helsinki, June 12-14, 2008.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
25
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integration of interdisciplinary participative research tools:
New vistas for land use modelling?
M. Lippe, T. Thai Minh, A. Neef, V. Hoffmann, G. Cadisch, T. Hilger
Institute for Plant Production and Agroecology, Universität Hohenheim, 70599 Stuttgart
Contact: Melvin.Lippe@uni-hohenheim.de
Introduction
Land use has been generally considered a local environmental issue, but it is becoming a force
of global importance (Foley et al., 2005). The complexity of land-use systems calls for multidisciplinary analyses (Mulder, 2001). Modelling is one of the methods in the portfolio of tools
and techniques available to unravel the dynamics of land use systems A prerequisite to the
development of realistic models of land use change is the identification of the most important
drivers of change (Verburg et al., 2006). Factual knowledge and analytical techniques have to
be combined with local knowledge and subjective perceptions of the various stakeholder
groups (Pahl-Wostl, 2002).
Methods
An interdisciplinary team of socio-economic and ecological researchers was formed to
integrate participative research tools to derive datasets for the FALLOW (Van Noordwijk,
2002) model. The uplands cropping area of Ban Put, Yen Chau District, Son La Province,
Northwest Vietnam was chosen as study site, and a list of model input parameters was
selected as prerequisite for the data collection process. Open and semi-structured interviews
were conducted with local stakeholder groups, e.g., young (18–35) and old (36–65), male and
female farmers, and village key persons. Information on land use history, plot soil fertility
status, farmer management and current problems of the upland land use systems were
collected. The interviews concept aimed to promote participation of local stakeholders by
using different tools, such as a 3D topographic map, drawing and ranking cards, developing a
resource flowchart and uplands seasonal calendar. The approach was designed based on an
information feedback loop: 1st a data analysis is required after each data collection loop; 2nd
results of previous loop are taken as basic knowledge and used as an entry point of the
following loop; 3rd information is crosschecked with stakeholders such as local authorities,
key farmers, farmers of young/old/male/female groups; and finally 4th data generated during
these processes are validated by all local stakeholders during a synthesis discussion. It was
assumed that by interviewing different stakeholder groups a validation of derived information
and thus a consistency assessment could be achieved. The derived datasets were then used to
calibrate the model accordingly. Three scenarios were defined in line with the study findings
to test the FALLOW model modus operandi.
Results and discussion
In general, land use change on the village level revealed the following situation: 1st expansion
of cropping area close to the village and lower elevation and inclination levels to mid-slope
and hilltop positions, 2nd land tenure changed from cooperate to individual management
systems, 3rd traditional swiddening systems changed to permanent annual cropping, 4th
individual plot management continuously intensified by improved tillage methods, use of
chemical fertilizers and cropping of hybrid varieties, and mayor problems of the upland
cropping systems are soil erosion and declining soil fertility.
Following the stage of model calibration, each scenario was set for a time period of 25 and
50 years, respectively. Figure 1 presents the output maps of scenario “transition swiddening to
permanent cropping systems”. The initial land use presents the major cropping area close to
the village area, where only a small part of the area is used for cropping. Beside agroforestry
26
Session A1: Methods for integrated assessment of agricultural systems and land use
areas close to village, the remaining area is in a stage of fallow or old secondary forest. This
image changes after 25 years, where more land has been opened for cropping or agroforestry
purposes. The overall forest area declined, although a pioneer forest develops in the lower
right corner. After 50 years, the agroforestry areas almost vanished and cropping is the
dominating land form. Remaining old secondary and primary forest areas are due to an initial
calibration setup to keep forested areas on the steeper slopes of the study area.
Initial land use
After 25 years
After 50 years
Crop
Pioneer, Young secondary forest , Old secondary forest , Primary forest,
Aroforestry pioneer,
Agroforestry early production stage,
Agroforestry late
production stage,
Agroforestry post-production stage
Figure 1. Outputs of model scenario ‘transition swiddening to permanent cropping systems’.
The application of an interdisciplinary participative research approach was useful to address
different stakeholder perceptions of the upland ecological and socio-economical system. The
combination of qualitative tools together with secondary datasets helped to close to gap of
lacking information, as especially in the context of Northern Vietnam, data collection and
land use analysis is particularly challenging (Castella & Verburg, 2007). By this means,
different scenario input datasets could be derived to test the model modus operandi. The
FALLOW model was able to generate similar land use trends as described by local
stakeholders. Furthermore, results of an in-situ land use survey in summer 2008 (not
presented here) indicate similar findings of cropping intensification. Rather the intensification
timeframe between model scenario run and reality seem to differ, and in this sense, land use
intensification even accelerated in a shorter time then modelled. As the FALLOW model is a
conceptualized tool, specific land use intensity consequences, e.g., soil erosion could not be
presented as indicated in farmer interviews. For this purpose, a rather more specific
biophysical spatial model approach is appropriate, and may in combination with the basic
FALLOW algorithms a useful integration of biophysical and socio-economic driving forces.
References
Castella, J.C. & P.H. Verburg, 2007. Ecol. Mod. 202: 410-420.
Foley, J.A., et al., 2005. Science 309: 570-574.
Mulder, M., 2001. Integrated Assessment 2: 73-78.
Pahl-Wostl, C., 2002. Integrated Assessment 3(1): 3-14.
Van Noordwijk, M., 2002. Ecol. Mod. 149: 113-126.
Verburg, P.H., et al., 2006. Landscape Ecol. 21: 1171-1183.
27
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Modelling resource distribution in mountainous watersheds in Thailand:
An integrated approach
C. Marohn, U. Schuler, M. Lippe, Y.Z. Ayanu, C. Hugenschmidt, T. Hilger, G. Cadisch
Inst. for Agroecology and Plant Production in the Tropics and Subtropics,
University of Hohenheim, Garbenstr. 13, 70593 Stuttgart, Germany
Contact: marohn@uni-hohenheim.de
Introduction
During the past 30 years, upland watersheds in the region of Chiang Mai, NW Thailand, have
been subject to multiple changes in land use. Due to political interventions, ethnic preferences
and market demand, some major trends were the replacement of poppy, a lychee boom and,
recently, a surge of cut-flower cultivation in greenhouses.
To allow the ex-ante assessment of effects of such dynamics on upland – lowland resource
competition, on nutrient flows (erosion but also deposition) or dispersal of contaminants, a
Land Use Change Impact Assessment Tool, acronym LUCIA, has been developed.
LUCIA aims at a generic and process-based simulation of the most relevant processes of
small watersheds at affordable run time.
Model development is part of the interdisciplinary special research programme Sustainable
Landuse and Rural Development in Mountainous Regions of Southeast Asia (Uplands
Program); findings and scenarios of other subprojects are integrated.
Methods
Model concept
The spatially explicit dynamic LUCIA model
runs on PCRaster (Van Deursen, 1995),
using a 25 m grid and daily time step.
LUCIA builds on existing validated models:
Hydrological concepts have been adapted
from GenRiver (Van Noordwijk et al., 2003),
plant growth is based on CGMS-WOFOST
(Supit, 2003), while dynamic bulk density
and succession of natural vegetation follow
the FALLOW approach (Van Noordwijk,
2002). Soil fertility is oriented by a strongly
simplified CENTURY approach (Parton et
al., 1988). The hydrologic and plant growth Figure 1. Schematic coupling of LUCIA
modules will be dynamically linked to an components.
existing Multi-Agent System (Berger et al.,
2006), which simulates economic factors and decision-making (Figure 1). Land use decisions
will then be made on the basis of yield and other biophysical outputs and lead to a land use
map that is fed back into LUCIA.
Parameterization
The model has been parameterized for the small (7 km2) watershed of Mae Sa Noi. Soil and
land use maps were generated from on-site measurements by Schuler (2008) and satellite
imagery, while secondary data were taken for plant parameterization. Weather data were
obtained from twinned subprojects in the Uplands Program. Parameters referring to soil and
land use classes are entered into a spreadsheet, which allows basic calculations outside the
model. Parameterization files, look-up tables and time series are then exported through a
macro.
28
Session A1: Methods for integrated assessment of agricultural systems and land use
First scenarios and future work
Model development is still
ongoing and plausibility is being
tested prior to validation on
measured data. As an example of
interlinkage between land use and
hydrology, Figure 2 shows
groundwater discharge under
different vegetation types: Deeprooting plants consume larger
quantities of soil water so that
percolation into groundwater is
less compared to flows under
shallow-rooting
vegetation.
Groundwater recharge responds
to rainfall events shown on the
Figure 2. Groundwater discharge under different land
right
axis. As groundwater
uses.
discharge is by definition
proportional to groundwater stocks, rainfall is indirectly reflected by groundwater discharge.
Future work will conceptually concentrate on the implementation of an erosion module based
on the GUEST approach (Yu et al., 1997), out- and upscaling and the option to couple single
modules in and out on demand. Technically, dynamic coupling to the MP-MAS model,
transfer functions for soil hydraulic properties (based on WaNuLCAS; Van Noordwijk &
Lusiana, 1999) and a user-friendly interface to determine outputs will be emphasized.
References
Berger, T., et al., 2006. Mathematical Programming-based multi-agent systems to simulate
sustainable resource use in agriculture and forestry. Internal document, University of
Hohenheim, Stuttgart, Germany.
Parton, W.J., et al., 1988. Biogeochemistry 5: 109-131.
Schuler, U., 2008. Towards regionalisation of soils in Northern Thailand and consequences
for mapping approaches and upscaling procedures. Doctoral dissertation, University of
Hohenheim, Germany.
Supit, I., 2003. Updated system description of the WOFOST crop growth simulation model as
implemented in the crop growth monitoring system applied by the European Commission.
Treemail Publishers, Heelsum, The Netherlands.
Van Deursen, W.P.A., 1995. Geographical Information Systems and Dynamic Models:
Development and application of a prototype spatial modelling language. PhD dissertation,
Utrecht University, NGS 190.
Van Noordwijk, M., 2002. Ecological Modelling 149: 113-126.
Van Noordwijk, M. & B. Lusiana, 1999. Agroforestry Systems 45: 131-158.
Van Noordwijk, M., et al., 2003. Spatial variability of rainfall governs river flow and reduces
effects on landuse change at landscape scale: GenRiver and SpatRain simulations. Paper
presented at MODSIM Conference, Australia.
Yu, B., et al., 1997. Aust. J. Soil Res. 37: 13-31.
29
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Construction of a multi-attribute model of arable system sustainability by
elicitation of expert opinion
N. McRoberts, C.F.E. Topp
Land Economy & Environment Research Group, SAC, Edinburgh,
Edinburgh, EH9 3JG, UK
Contact: neil.mcroberts@sac.ac.uk
Introduction
In 2006, the Scottish Government introduced a new procedure for funding research in
agriculture, biological sciences and land use. A number of new 5-year research workpackages
replaced the existing funding streams, with work more closely aligned to policy questions and
additionally aimed specifically at developing inter-institute, interdisciplinary research. Several
of the workpackages deal explicitly with issues of resilience and sustainability in agricultural
systems, and one of these is concerned specifically with arable and mixed arable production
systems which dominate the land use pattern on the eastern side of the Scottish mainland,
between Inverness in the north and Berwick (on the English border) in the south. The aim of
the workpackage is to identify the factors which confer resilience and sustainability on arable
systems along the familiar set of environmental, economic and social dimensions and to
propose management approaches for such systems so that resilience and sustainability can be
maintained. Over the whole life of the project empirical studies and systems simulation
modelling (see McRoberts et al. (2009); Topp & Rees (2009)) will be used to address these
questions but it was recognized at the start of the project that there would be a lag time before
the project was able use such methods to assess policy options and make projections for
sustainable land use. However, since the workpackage is built on previous research and
utilizes approximately 30 scientists, covering a range of disciplines, there was already
considerable knowledge about many of the key issues before any new research was done. To
help structure the research and to provide an interim mechanism for examining policy options,
a multi-attribute model (MAM) of arable system sustainability was constructed by eliciting
expert opinion from the research team. The resulting model is being used to establish a set of
future scenarios which will be tested in more detail once the quantitative modelling
framework is available.
Methods
The MAM was constructed using the DEXi modelling tool (Bohanec, 2008) available from
the Joseph Stefan Institute, Slovenia. An iterative process knowledge elicitation process was
used to gather information from the research team and construct the MAM for arable system
sustainability from the information. An initial seminar session was used to explain to the
whole research team what the aims of the exercise were. The team is composed of a number
of natural sub-groups which specialize in discipline-oriented research; i.e. in ecology,
economics, and social sciences. These sub-groups were asked to discuss among themselves
(i.e. not in conjunction with the other groups) what they already knew about factors which
would lead to either an increase or decrease in sustainability. The groups were allowed to
represent this information in any way they chose, but had been introduced to the hierarchical
structure of DEXi models as one possible approach. A round of discussions was held with
each group to allow us to gain an understanding of their representations of knowledge and an
intial MAM was constructed following these discussions. This MAM was then shown to the
groups, in a second round of discussions and suggestions for changes, additions and
improvements were incorporated to produce the final first operational version of the MAM.
This version of the model is currently being evaluated by the research team as a whole, and
joint decisions are being made about how the different branches of the model would be
30
Session A1: Methods for integrated assessment of agricultural systems and land use
weighted in the overall assessment of sustainability and the initial set of policy options that
will be evaluated.
Results and discussion
Illustrative sections of hierarchical MAM in DEXi are shown in Figure 1. The model contains
89 attributes of which 43 are basic attributes (i.e. they are the end-points of branches in the
hierarchical model) and 46 are aggregate attributes, composed from two or more underlying
more basic attributes.
(a)
(b)
Figure 1. (a) DEXi model main branches
showing main split between ecological and
human attributes of sustainability. (b) The
human system branch has been expanded
to show it contains both social and
economic attributes.
The MAM has proved useful in helping to bridge inter-disciplinary gaps in the research team
by allowing different sub-groups to evaluate their own research objectives more clearly
against the project as a whole, and by providing the opportunity for joint discussions about the
relative importance of the different facets of sustainability. These discussions act as a useful
precursor of the sorts of questions that we, the model developers, can expect policymakers
and stakeholders to raise when the model is deployed in practice to evaluate potential policy
options. In common with previous efforts to use MAMs in the analysis of cropping systems
(Bohanec et al., 2004) we have found that the relationships among real-world entities and
processes in the knowledge base were highly cyclical, suggesting that an alternative model
building approach based on a network structure might be more appropriate. A variety of well
validated methods exist for developing such models and offer a further set of tools for
integrated analysis of agricultural and social systems (e.g., Ozesmi & Ozesmi, 2004). A key
issue in the use of MAMs and similar methods is the careful documentation of the mechanism
by which the relationships represented in the knowledge base are translated into the rules or
weights in the model. This is also true for the construction of formal simulation models, but
since MAMs and other knowledge-based models are often viewed as being more subjective
than their mathematical counterparts, it can be an important issue in user-acceptance to have a
transparent process of model construction.
References
Bohanec, M., 2008. DEXi: Program for Multi-Attribute Decision Making, User’s Manual,
Version 3.00. Joseph Stefan Institute, Ljubljana.
Bohanec, M., et al., 2004. Informatica 28: 387-392.
McRoberts, N., et al., 2009. AgSAP Conference 2009, these proceedings.
Ozesmi, U. & S. Ozesmi, 2004. Ecol. Mod. 176: 43-64.
Topp, C.F.E. & R.M. Rees, 2009. AgSAP Conference 2009, these proceedings.
31
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Gathering baseline data for studying arable system resilience in Scotland
by integrated farm surveys
N. McRoberts, C.F.E. Topp, C. Hawes, R.L. Walker, C.A. Watson, P. Hallett, T. Daniell,
M. Young, G.R. Squire
Land Economy & Environment Research Group, SAC, Edinburgh,
Edinburgh, EH9 3JG, UK
Contact: neil.mcroberts@sac.ac.uk
Introduction
In 2006, the Scottish Government introduced a new procedure for funding research in
agriculture, biological sciences and land use. A number of new 5-year research workpackages
replaced the existing funding streams, with work more closely aligned to policy questions and
additionally aimed specifically at developing inter-institute, interdisciplinary research. Several
of the workpackages deal explicitly with issues of resilience and sustainability in agricultural
systems, and one of these is concerned specifically with arable and mixed arable production
systems which dominate the land use pattern on the eastern side of the Scottish mainland,
between Inverness in the north and Berwick (on the English border) in the south. The aim of
the workpackage is to identify the factors which confer resilience and sustainability on arable
systems along the familiar set of environmental, economic and social dimensions and to
propose management approaches for such systems so that resilience and sustainability can be
maintained. Over the whole life of the project empirical studies and systems simulation
modelling (see McRoberts et al. (2009); Topp & Rees (2009)) will be used to address these
questions. Modelling will allow possible future configurations for land use to be explored in
support of questions being raised by policymakers. Empirical data, however, plays a key role
in the research. First, it allows the models to be parameterized so that they correctly reflect the
local conditions of Scottish arable production. Secondly, data collected now (together with
previously collected data) provide a baseline against which to measure change in the future.
Finally, the data allows an empirical analysis to be made of potential trade-offs which might
exist between different policy objectives, providing a means of checking whether model
predictions aimed at identifying satisfactory solutions for multiple goals are feasible. This
paper describes on-farm survey work which has been undertaken to collect such data in
Scotland and briefly reports on some of the findings.
Methods
The survey was constructed using a targeted sampling approach intended to produce a data set
with a wide variation in crop inputs, production methods geophysical and climatic conditions.
This range of conditions was met in the survey by including farms with certified organic
status, farms which belong to a recognized body which promotes integrated production
methods (LEAF), and farms, covering a wide range of sizes, that did not belong to either of
the other categories. As far as possible, on each farm a cereal crop and a non-cereal, break
crop were selected and a range of biophysical data were gathered following agreed
standardized protocols. The main elements of these surveys were: an assessment of the
within–field flora; the soil seed bank; the composition of the marginal flora of the field and
data on the biological and physical properties of the soil. A two-part farm management survey
was also conducted: participating farmers were asked to provide rotational history and general
farm management information at the time of the initial ecological survey, and were
subsequently asked for information specifically on crop inputs for the crops included in the
survey, to allow gross margins to be calculated for each crop.
32
Session A1: Methods for integrated assessment of agricultural systems and land use
Results and discussion
A wide range of analyses have been carried out on the survey data for 2007/8. Figure 1 shows
an illustrative output from the analyses in which different clusters of fields have been
identified based on their rotational history.
spring crops, no grass
0.6
spring crops, with grass
winter crops, no grass
no potatoes
winter crops, with grass
0.4
PCO2 (17%)
0.2
no veg
with veg
with potatoes
0
-0.2
no veg
with veg
-0.4
-0.6
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
PCO1 (27.6%)
Figure 1. Classification of Scottish arable fields on the basis of presence or absence of
different classes of crop in their rotational history.
As indicated in the introduction the aim in carrying out such analyses is to gain a perspective
on where the Scottish arable system currently is so that we can make objective statements
about its future trajectory and its resilience. The study of resilience requires us to address
concepts which ecologists and economists can seem share in studying the behaviour of
complex systems. A sustainable system or population can be defined as one in which there is
no unchecked trend and for which deviations from the trend are bounded within limits. This
leads to a requirement that resilient systems are those in which variables which ‘measure’ the
system return to the above behaviour after being forced outside the bounds around the trend.
So, for instance, an ideal or sustainable seed bank population (defined by the abundance of a
range of species or functional types) is one which provides food resource for a wide variety of
plant species but not to the detriment of crop yield. It is resilient if it can be returned to the
ideal state if it temporarily moved above or below that state; e.g., through extremely intense
or lax management, respectively. The empirical analyses are allowing us to gain an
understanding of the ranges that such indicators as seed bank size take across a range of
production situations. We are also using the data to examine the extent to which intensive
cereal production is associated with declines in floral biodiversity and soil health. Generally,
in a non-resilient system, measured variables are likely to display increasingly large
departures from their reference values. Measuring such departures in a time series of data
could be used as a method to test for resilience and (by extension) sustainability. Royama
(1996) discusses this issue in an ecological context and Pearce (1998) suggests the
applicability of the method both to ecological and economic variables.
References
Pearce, D.W., 1998. In: Environmental indicators for agriculture. Vol. 2: Issues and Design,
OECD, Paris, France, pp. 29-42.
Royama, T., 1996. Analytical population dynamics (2nd ed.). Chapman & Hall, London.
McRoberts, N. & K. Topp, 2009. AgSAP Conference 2009, these proceedings.
Topp, C.F.E. & R.M. Rees, 2009. AgSAP Conference 2009, these proceedings.
33
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrating multiple functions to optimize rural land health
B. Meyer, R. Grabaum, D. Watson, D. Adams, A. Hood, A. Phillips
School of Spatial Planning,
Dortmund University of Technology,
August-Schmidt-Straβe 10, 44227 Dortmund, Germany
Contact: burghard.meyer@tu-dortmund.de
Introduction
As the results of the Millennium Ecosystems Assessment has shown, there is an obvious and
urgent need for integrated land use planning to help provide optimal outcomes for the health
of rural landscapes. Land use problems in rural landscapes involve complex trade-offs
between multiple, and sometimes competing, objectives. These land use decision problems
require new kinds of decision methodologies which can integrate these multiple objectives
and provide optimal solutions.
Methods
The MULBO (Multi-criteria Landscape Assessment and Optimization) framework has been
specifically developed to help guide these complex multiple objective decisions. In this
current project MULBO has been trialled in the Lake Tyrrell Basin of Northern Victoria,
Australia (Meyer & Grabaum, 2008). Lake Tyrrell is situated within the Mallee region and
covers an area of around 400,000 hectares. The Mallee is a semi-arid region which supports
broad-scale cropping on sandy low nutrient soils, where traditional agriculture has had a high
impact on native fauna and flora, and where naturally saline groundwater and wind erosion
present significant threats to both agricultural production and biodiversity.
The MULBO process includes the following steps (a) goal determination across multiple
management criteria, (b) function analysis on the basis of GIS, (c) function assessment, (d)
scenario formulation and (e) land use compromise optimization to calculate land use scenarios
(Figure 1).
In this project, major land management goals within the Mallee were identified on the
basis of regional and local management objectives and input from regional land managers,
representatives from community environmental programs and farmers. Major landscape
health problems in the region were identified, as were the major management objectives to be
optimized. These objectives included indicators for (a) farm income, (b) salinity risks, (b)
wind erosion and (d) habitat connectivity.
Inputs to the MULBO process include a range of data sets from a variety of sources. Based
on an initial consultation, these data sets were processed to provide a new decision geography.
The resulting data sets take the form of qualitative landscape value surfaces classified from
low to high.
The land use optimization software LNOPT 2.0 was used to integrate the spatial data sets
and provide suggested optimal land use patterns providing a range of compromise solutions
for land use distribution in the Lake Tyrrell Basin Region. Different land use scenarios were
calculated by adjusting weights for each indicator.
Results and discussion
The application of the MULBO framework in Victoria’s Mallee region has shown the
framework’s ability to integrate a range of qualitative and quantitative information in support
of complex land use decision making. The framework’s ability to integrate spatial data from a
range of sources and its open, flexible, structure make it a highly flexible decision tool to
support sustainable solutions to rural land health. In particular, MULBO was able to support
34
Session A1: Methods for integrated assessment of agricultural systems and land use
the engagement of important stakeholders through a deliberate and clear decision
methodology.
Figure 1. Optimized spatial outputs from the MULBO process.
Reference
Meyer, B.C. & R. Grabaum, 2008. Landscape Research 33(2): 155-179.
35
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Framework for the socio-economic performance assessment of irrigation
schemes
Rui Santos, Paula Antunes, Pedro Beça
CENSE/ECOMAN - Ecological Economics and Environmental Management Group,
Department of Environmental Sciences and Engineering,
Faculty of Sciences and Technology, New University of Lisbon,
Campus de Caparica, 2829-516 Caparica, Portugal
Contact: rfs@fct.unl.pt
Introduction
The sustainable use of water in agriculture is a global policy priority of vital importance,
particularly considering an increasing demand for water and competition among uses
resulting in growing scarcity. Irrigated agriculture deals with very complex issues, such as,
resource preservation, environmental compatibility, economic feasibility, technological
knowledge and social acceptance. Several potential economic, social and environmental
impacts are generated and endured by diverse water users, which are more significant in
regions where water is scarce (e.g., IEEP, 2000; Bouwer, 2002; Qadir et al., 2003; Wichelns
& Oster, 2006; Kassam et al., 2007). Social impacts can be related with employment, public
health, security (sometimes associated to conflicts), gender balance and livelihood, while
some relevant environmental impacts can be related to water depletion, water quality
degradation, impact on sensitive habitats and soil erosion. Economic impacts result, for
instance, from changes in water prices, costs of irrigation technologies and infrastructures and
water productivity, as well as other indirect effects on the productivity, competitiveness and
attractiveness of agriculture and rural systems.
This paper proposes a framework for the socio-economic performance assessment of
irrigation schemes, which can be applied in different contexts for irrigation schemes and
farms, allowing for the characterization of existing systems and the comparison of alternative
management options. The innovation is related with the selection of a set of economic and
social indicators (e.g., Bos et al., 2005; Molden et al., 1998) and their combination using a 3tiered flexible approach, designed in order to allow for application in different contexts. The
three levels of assessment adopted, which range from a general level to an intermediate and
advanced levels, are built considering differences in the assessment scope, information needs
and stakeholders’ involvement.
Methods
The proposed assessment framework considers a 3-tiered approach, which allows for the
characterization of the irrigations schemes in different levels, requiring distinct levels of
information detail and research resources. For example, the general level includes only basic
descriptors of the irrigation scheme, whereas the advanced level includes also more social and
economic indicators and is detailed per farm and per crop. In some cases, the development of
a CBA study may be considered at this level. The main stages in the framework
implementation include: (a) description of the irrigation scheme; (b) definition of the
assessment objectives, scope and time horizon; (c) identification of information sources, main
stakeholders, management issues and actors to be involved; (d) selection of the adequate level
of assessment; (e) selection of relevant indicators as well as of needed complementary
evaluation tools; (f) data collection, computation of indicators and application of
complementary tools; (g) validation and comparison of the results with other irrigation
schemes and regional and national information.
It is provided a set of indicators to be selected for each of the three levels of assessment,
considering, simultaneously, the inputs, outputs and impacts of irrigation, which facilitates a
36
Session A1: Methods for integrated assessment of agricultural systems and land use
comparative analysis between diverse regions, different irrigation systems and management
alternatives, as well as the analysis of trends in time and in space. The proposed indicators are
focused not only in water use and agricultural production, but also in social, economical and
environmental aspects associated with irrigated agriculture.
Results and discussion
This framework allows a quantitative and qualitative assessment that can be applied in
different contexts for irrigation schemes and farms, allowing for the characterization of
existing systems, the comparison of alternative management options, as well as an analysis of
trends in time and in space. The proposed indicators are focused not only in water use and
agricultural production, but also in social, economical and environmental aspects associated
with irrigated agriculture. The quantitative results of the assessment process integrated with
the qualitative inputs from stakeholders’ involvement, contribute to the evaluation of global
irrigation scheme performance as well as to the identification of the factors that are limiting
performance such as structural, social or management issues.
The application of the framework to a Portuguese case study is in progress (Caia irrigation
scheme, with an area of 9000 ha). Some difficulties in the process are related with data
availability, but the involvement of the stakeholders in the evaluation process contributes to
mitigate this problem. Results obtained so far confirm the contribution of the proposed
framework to increase the capacity to evaluate irrigation schemes performance, for both water
managers and farmers.
This work points the way towards further developments in the following aspects associated
to this type of assessment:
(1) evaluate the net effects that a change can have in the irrigation scheme;
(2) the inclusion of objective performance standards for each indicator, and
(3) the use of additional information such as key descriptors of the irrigation scheme
characteristics that can help in the interpretation and comparison of results.
Selected References
Bouwer, H., 2002. Journal of Irrigation and Drainage Engineering 128: 193-202.
Bos, M.G., et al., 2005. Irrigation and drainage performance assessment: practical guidelines.
CABI Publishing, UK.
Kassam, A., et al., 2007. Irrigation Science 25: 185-188.
IEEP, 2000. The Environmental impacts of Irrigation in the European Union. Institute for
European Environmental Policy (IEEP), London.
Molden, D.J., et al., 1998. Indicators for comparing performance of irrigated agricultural
systems. Research Report 20. Colombo, Sri Lanka: International Water Management
Institute.
Qadir, M., et al., 2003. Agricultural Water Management 62: 165-185.
Wichelns, D. & J.D. Oster, 2006. Agricultural Water Management 86(1): 114-127.
37
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Sustainable livestock production: Assessment of farmers initiatives
S.F. Spoelstra, B. Elzen
Animal Sciences Group, Wageningen UR, P.O. Box 65,
8200 AB Lelystad, The Netherlands
Contact: sierk.spoelstra@wur.nl
Introduction
Dutch livestock production has become highly efficient in terms of output per unit input, but
is also heavily criticized on sustainability issues including deprived animal welfare,
environmental burdening and outbreaks of contagious diseases. As a response, in 2007 the
Dutch government set policy targets of 5% and 100% sustainable livestock production at the
farm level for 2011 and 2023, respectively. These goals represent major challenges in
defining what sustainable farming is and in developing effective policy measures to support
the required transition (Rotmans, 2003) towards sustainable livestock production. Setting a
specific target of 5% for 2011 assumes that sustainability can be defined at the farm level and
that by choosing proper criteria the number of farms that satisfy the definition can be
assessed. A farm, however, is embedded in a wider socio-technical system and sustainable
development requires changes on a variety of dimensions to lead to a transition. Various new
approaches will have to be tested and learned about in practice to be able to asses what may
work in practice. In this perspective change initiatives at the farm level (called ‘farmers
initiatives’ in this paper) should not just be ‘counted’ to establish their contribution to the
short-term policy goal but be assessed as part of a learning process towards a broader
sustainable livestock production system (the long-term policy goal).
Methods
To monitor achievement of the short-term policy goal, criteria were developed to assess the
number of sustainable farms. These criteria emphasized improved animal welfare while
environmental issues should at least comply with present legislation. This was put into
practice by using existing certification schemes for organic agriculture (EKO-label),
environmental assurance scheme (Milieukeur-label) and compliance with rules for tax
reduction, including green financing. These criteria should allow the ‘counting’ of the
progress in the number of ‘sustainable’ livestock farms in the years up to 2011.
Such a method of counting, however, does not provide a proper assessment tool for a
transition towards a sustainable system. Theories on system innovation and transition
management emphasize development of innovation pathways in a multi-actor and multi-level
learning process to result eventually in a shift to a socio-technical regime of the livestock
production sector. To assess the progress of such developments we need interactive forms of
assessment, which, as Van der Sluijs (2002) wrote, are at the core of integrated assessment.
Such an assessment method should focus on the potential for learning about sustainability.
From this perspective farmers initiatives should be assessed for their potential to contribute
to learning about the (in-)possibilities of such multi-dimensional change processes. In such an
approach sustainability is also understood as a multi-dimensional and dynamic phenomenon.
This adds a considerable complexity to the process which requires learning, on the one hand
because of uncertainty on which of the present initiatives may contribute to sustainable
development in the broader sense and on the other hand because unforeseen trade-offs may
emerge.
To assess the potential of farmers initiatives to contribute to sustainable development we use
four major criteria:
38
Session A1: Methods for integrated assessment of agricultural systems and land use
1. Presence of a strong vision. Visions, e.g., of an integral design of a sustainable farming
system (Groot Koerkamp & Bos, 2008) can play an important role in systems innovation.
They make sustainability operational, give direction to development and can mobilize
stakeholders from various domains.
2. Institutional change. As Roep et al. (2003) argued, technical innovation in farming should
be accompanied by institutional innovation to achieve ‘effective reformism’.
3. Potential business model. Three earning models for an integral form of sustainable farming
(1) The costs of livestock produce is cost effective and gives a competitive advantage in
the market. (2) The costs of livestock produce are higher but the additional costs can be
earned in the market, e.g., by sustainability-certified supply chains. (3) Costs of livestock
produce are higher, but as part of sustainable farming additional functions providing new
sources of income are added (e.g., energy and fertilizer production).
4. Possibility to mitigate excess entrepreneurial risks. Farmer’s initiatives with a potential to
contribute to sustainable development are accompanied by uncertainties and costs which
could jeopardize continuity of the farm (Meijer, 2008).
Results and discussion
Our main argument is that, whereas the assessment of sustainable livestock farming systems
is presently based on quality assurance schemes, it should be based on an assessment of the
potential of farmer’s initiatives to contribute to a sustainable development path. This implies
seeing these initiatives as part of a learning and experimentation strategy to achieve
sustainable development in the broader sense and should be based on insights from systems
innovation analysis. The approach of ‘Strategic Niche Management’ offers a good starting
point to carry out such an assessment. (e.g., Schot & Geels, 2008). Preliminary experiences
based on ideas of strategic niche management with smart experimentation with dairy farmers
initiatives (Roep et al., 2007) were used to corroborate and refine our approach. As a next
step will apply the method by selecting a number of dairy farmers initiatives to obtain a
coherent portfolio allowing learning for sustainable dairy farming including aspects as
welfare, green house gases, energy use, economy and labour conditions.
Thus, the approach not only looks at the potential of single initiatives to asses their
potential to contribute to sustainable development but also analyses the cohesion between
initiatives to obtain synergy in learning and also compares the learning achieved in these
initiatives with future visions and explicit demands of various relevant actors.
References
Groot Koerkamp, P.W.G. & A.P. Bos, 2008. Neth. J. Agric. Sci. 55: 113-138.
Meijer, I., 2008. Uncertainty and entrepreneurial action. PhD thesis, University of Utrecht.
Roep, D., et al., 2003. Neth. J. Agric. Sci. 51: 95-217.
Roep, D., et al., 2007. Slim experimenteren in de veehouderij. Report 26, Animal Sciences
Group, Wageningen UR.
Rotmans, J., 2003, Transitiemanagement, sleutel voor een duurzame samenleving. Van
Gorcum, Assen.
Schot, J. & F.W. Geels, 2008. Technology Analysis & Strategic Management 20: 537-554.
Van der Sluijs, J.P., 2002. In: Ed. T. Munn, Encyclopedia of global environmental change,
Volume 4. John Wiley & Sons, Chichester, UK, pp. 250-253.
39
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated assessment of nutrient management options in smallholder
farming systems: The NUANCES framework
P. Tittonell, M.C. Rufino, M.T. Van Wijk, M. Herrero, M. Corbeels, C. Pacini, J. Smith,
S. Karanja, C. Quiroz, P. Mapfumo, J. K’ungu, F. Baijukya, C. Kaizzi, M. Mwale,
O.M. Sanogo, E. Yeboah, D. Nwage, B. Vanlauwe, A. Albrecht, N. De Ridder, K.E. Giller
Plant Production Systems, Wageningen University, P.O. Box 430, 6700AK Wageningen
Contact: Pablo.Tittonell@cirad.fr
Introduction
Options for integrated management of nutrient resources and soil fertility by smallholder
farmers in Sub-Saharan Africa (SSA) must be targeted (i) to the specific context in which
farming takes place, (ii) to particular types of households pursuing different livelihood
strategies, and (iii) to spatio-temporal ‘niches’ within the farming system. The NUANCES
framework (Nutrient Use in Animal and Cropping systems – Efficiencies and Scales;
www.africanuances.nl) is a methodological approach to ex-ante assessment of the feasibility,
impact and trade-offs of implementing technologies for strategic nutrient management (SNM)
in small-holder farming systems in the short and long terms, with a focus on processes taking
place at the farm rather than the plot scale (Giller et al., 2006). The framework consists of
several methodological steps organized in two major ‘loops’ (Figure 1), and the various steps
are articulated using the ‘DEED’ approach:
1. Describe, current production systems and their problems;
2. Explain, current farmers’ decisions on resource allocation and their consequences;
3. Explore, options for agro-technological improvement in face of possible future scenarios;
4. Design, new management systems that contribute to a sustainable intensification.
C
stu ase
di e
s
rm
F a e ys
v
sur
40
Tr
a
an de -o
a l y ff s
s is
In
te
pa nsi
th fica
wa ti
ys o n
Methodological steps
Current
Alternative
(Design)
(Explore)
Current
System
systems
System
Field (surveys, experiments) and
described
Sc
described
an enar
al y i o
desktop (databases, models) tools
si s
are used in combination. The
diversity of households within a
community is analysed through
‘Long loop’
Entry
socio-economic surveys and categopoints
Synthesis
Prototypes
rization of farm types (farm
&
Best fits
typology) according to main producSimplified
Existing
Sub-system
virtual
tion objectives/ orientation/ resource systems
typologies
systems
constraints. Representative farms
within each farm type are selected
‘Short loop’
as case studies for quantitative
characterization
of
production
Farm
Typology
Dynamic
activities and key resource (cash,
virtual
labour, nutrient) flows. This
systems
(Describe)
(Explain)
information is synthesized as
simplified ‘virtual’ farms.
Figure 1. The NUANCES framework (see text for
At this step key entry points for
explanation)
SNM may be already identified,
thus, closing up the short loop. Virtual farms constitute the basis for the dynamic simulation
of the farm system, coupling soil/crop, grassland, livestock, manure and household models
(e.g., Tittonell et al., 2007; Rufino et al., 2007; Herrero et al., 2007). Modelling is used to
both ‘Explain’ current trade-offs and ‘Design’ alternative strategies for sustainable
intensification. While the first phase of an EU funded project AfricaNUANCES was one of
Session A1: Methods for integrated assessment of agricultural systems and land use
tool development and population of databases, a second phase is being designed (NUANCESDEEDS) to out-scale best-fit SNM options, closing the long loop.
Targeting mineral fertilizers within smallholder farms – An example
The 2006 Fertilizer Summit in Abuja, Nigeria brought together scientists and policymakers
from all over Africa who agreed that fertilizer use should be promoted; the goal of increasing
the average application rates from 10 to 50 kg ha–1 was set. Our explorations using the farmscale model across representative farm types in the Kenya highlands indicate that such
‘blanket’ recommendations lead to dissimilar results across agroecologies, farm types and
fields within individual farms. For example, farms located on sandy soils exhibit poorer
fertilizer use efficiencies than on clayey soils, e.g., with maize responses of 5-10 vs. 30-40 kg
grain per kg N with same rate of fertilizer applied. The home gardens that received continuous
nutrient inputs produce good yields and respond poorly to applied fertilizers; they can be
managed with small ‘maintenance fertilization’ rates. Degraded outfields do not respond to
fertilizers and need long-term rehabilitation through organic matter addition. Model runs
indicate that in such cases investments could be better directed towards improved livestock
productivity and manure handling, to increase the quantity of manure with better quality for
soil amendment.
Table 1. Fertilizer:maize price ratios (kg of maize necessary to pay for 1 kg N fertilizer) for
different commercialization channels and at different times of the year and in western
Kenya (source TSBF, 2007).
Price per kg N fertilizer
Maize sold in 90-kg bags
Before harvest
After harvest
78 KSh (agro dealer)
3.9
9.0
130 KSh (local retailer)
6.5
15.1
1 Euro = 99 KSh (Sep. 2008)
Maize retailed in 2-kg tins
Before harvest
After harvest
2.6
7.8
4.3
13.0
Our understanding of the diversity of households and their accessibility to markets (short loop
in Figure 1) allows us to place these results in context. Poorer farmers in need of cash are
forced to sell their maize immediately after harvest, when maize is abundant on the market
(Table 1). These farmers often buy (locally retailed) maize during the rest of the year and
when they buy fertilizers they do it in small amounts, experiencing the most unfavourable
fertilizer:price ratios. The opposite is true for wealthier farmers, who can buy fertilizers in
bags and stock their maize to wait for better prices before harvesting. Model explorations
suggest that poorer households could benefit from producing fodder crops to sell to livestock
owners instead of producing (poor yielding) maize in most of their farm.
Conclusion
The assessment of SNM options across case studies in 8 countries of SSA has taught us that
production resources should be targeted strategically within diverse, heterogeneous and
dynamic smallholder systems. Ex-ante impact assessment tools such as NUANCES can help
in identifying the most promising options (at farm scale and in the short- and long-term)
before they are promoted among farmers. This is of particular importance nowadays, in face
of the major investments in soil fertility planned for SSA (e.g., www.agra-alliance.org).
References
Giller, K.E., et al., 2006. Agric. Syst. 88: 8-27.
Herrero, M., et al., 2007. Agric. Syst. 92: 240-265.
Rufino, M.C., et al., 2007. Livest. Sci. 112: 273-287.
Tittonell, P., et al., 2007. Field Crops Res. 100: 348-368.
41
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The LUMOCAP PSS, impact assessment of agricultural policies in an
integrative context
H. Van Delden, M.L. Paracchini, T. Stuczynski, P. Ciaian, R. Vanhout
RIKS, P.O. Box 463, 6200 AL Maastricht, The Netherlands
Contact: hvdelden@riks.nl
Introduction
As agriculture covers about half of the EU-territory, the Common Agricultural Policy (CAP)
is a main driver determining land-use structure and landscape quality. It aims at ensuring
adequate market prices, satisfactory income to farmers, food availability and rural
development. The latter includes policy instruments such as agri-environmental schemes
which are designed for achieving sustainability of agri-ecosystems and the preservation of
landscapes.
The LUMOCAP Policy Support System (PSS) aims to assess how different policy
scenarios will impact efficiency in meeting sustainability objectives. It focuses on the
relations between the CAP and landscape changes and emphasizes the spatial and temporal
dimension of this process.
Methods
For any land use model to work properly it is important that it includes the relevant drivers,
processes and characteristics of the land use system. To move beyond the research phase and
provide added value to decision and policy-making, a system needs to connect to the policy
context as well as the policy process and moreover, provide added value to those working
with it.
To capture processes occurring at different spatial levels, the LUMOCAP system includes
sub-models at these different levels (EU-27, country, region, 1000×1000 m cells, 200×200 m
cells). The system utilizes an existing spatially explicit dynamic land use modelling
framework called METRONAMICA. Based on the knowledge of agricultural policies,
driving forces of land use change and end-user requirements, the METRONAMICA model is
adapted and improved to fulfil the requirements of policymakers at the EU, national and
regional level. To capture the interaction between different disciplines and processes
operating on various spatial resolutions emphasize has been put on the inclusion of dynamic
feedback loops between the different model components.
Regarding the context, it is important that such a system incorporates the main drivers and
processes of land use change and is able to simulate the change over time and in space. Very
often, (research) models stop at this point and are unable to translate policy questions under
consideration into model input and provide model results as policy-relevant information. For
an ISDSS, however, it is crucial that the user can analyse the impact of various policy
alternatives on a selected set of policy relevant indictors.
To be able to link to the policy process, it is crucial to have an overview of the
stakeholders involved and the steps that are followed. Based on this information a decision
can be made as to where and how in this process the system should provide support. The
decision has a large impact on the overall development since it requires making trade-offs
based on the desired functionality. If the system is to be used in an interactive workshop
session to scope future developments and brainstorm about possible scenarios, running speed
will be more important than the incorporation of very detailed models, while for a thorough
analysis of possible alternatives further along the policy process, the requirement might be the
opposite.
Since the ISDSS described in this presentation encompasses relatively complex
(integrated) models, the GUI should be able to provide access to two different types of users:
42
Session A1: Methods for integrated assessment of agricultural systems and land use
the policymakers who use the system as part of their policy process and who carry out impact
assessment studies with the model, and the modellers who can update the underlying data and
parameters and possibly even the model equations. The first group mostly benefits from a
GUI that follows the steps of a scenario or impact assessment process, the latter prefers to
look at the system in a more systemic way and values easy access to individual disciplinary
models.
The last crucial element for an ISDSS mentioned in the beginning of this abstract is the
added value that such a system provides to its users. From our experience we have learned
that users value the following benefits of an ISDSS (see also Van Delden & Engelen, 2006):
• An improved understanding of the interaction between the different functions and
processes that shape the region. Analysing the dynamic cause-effect relations of
alternatives enables learning which leads to awareness building;
• Because of the integrated nature, through feedback between models, effects are not only
simulated for the discipline itself. Understanding those impacts can prevent the occurrence
of unexpected and unwanted side-effects after the implementation of new policies;
• Although system development is a time-consuming and expensive task, once the system
has been set-up it provides the possibility to quickly calculate the consequences of
different alternatives;
• Although in every model subjective choices and assumptions will be made, ISDSS provide
a means to objectively measure and evaluate of alternatives in a repeatable way;
• Improved communication between the different sectors and disciplines.
Results and discussion
The final LUMOCAP product is an open-ended, flexible, transparent, PC-based, analytical
system enabling users to interactively enter policy options under a specific set of natural and
socio-economic conditions as external driving forces, to formulate potential land use
scenarios, and to assess the impact of both on the quality of rural landscapes through the
analysis of selected landscape indicators.
Besides model development, special attention is given to the interaction with the end-users
in order to incorporate their requirements into the final system – this is needed to ensure
acceptability of the overall approach and to provide an added value to decision making at
administrations. Therefore, the LUMOCAP system is developed in an iterative process in
which frequent end-user consultations provide information for upgrading and adjusting the
final product.
Training courses have been provided to (1) teach policymakers at the level of the European
Commission and the Member states how to work with the system and (2) assess the
usefulness and usability of the system. Although first reactions are very promising, more
work on data collection, calibration and fine-tuning the system to the exact wishes of
particular organizations is crucial for its actual implementation and use.
Reference
Van Delden, H. & G. Engelen, 2006. In: Eds A. Voinov, et al., Proceedings of the iEMSs
Third Biennial Meeting: ‘Summit on Environmental Modelling and Software’.
International Environmental Modelling and Software Society, Burlington, USA, July 2006.
43
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Measuring sustainability at a supra-farm level:
Evaluation of methods
H. Wustenberghs, E. Wauters, S. Van Passel
Institute for Agricultural and Fisheries Research - Social Sciences Unit
Burg. Van Gansberghelaan 115, Box 2, 9820 Merelbeke, Belgium
Contact: hilde.wustenberghs@ilvo.vlaanderen.be
Introduction
Since the Brundtland report (WCED, 1987) put sustainable development on the policy
agenda, many attempts have been made to put the theoretical concept into practice. This also
entailed the need to monitor progress towards sustainability and hence to ‘measure’ it. So,
over the last 20 years a wide range of sustainability monitoring tools have been developed.
The majority of these efforts focussed on the national level. Well known examples are the
ecological footprint (Wackernagel & Rees, 1995), integrated environmental and economic
(‘green’) national accounts (UN, 2003) or the index of sustainable economic welfare (Daly &
Cobb, 1989). Some efforts focussed on the firm level, such as some eco-efficiency measures
or the sustainable value added (Figge & Hahn, 2004).
Maybe even more than in other sectors, sustainable production is of vital importance in
agriculture, as farming, unlike most other economic activities, forms a part of the ecosystem
rather than being external to it. Sustainability assessments for agriculture at national level
have (partially) been made by drawing up economic and environmental accounts (Atkinson et
al., 2004, Wustenberghs et al., 2004). At the farm level, sustainability can be assessed by a
balanced set of indicators that might be visually integrated (Rigby et al., 2001; Meul et al.,
2008) or by an adjusted sustainable value added (Van Passel et al., 2007).
However, recently, the need has emerged for more regionalized policies that no longer
focus on nationwide measures but on ‘tailor-made’ solutions for relatively small regions or
for (sub)sectors. The Water Framework Directive is a good example of such a policy: while
having common goals for the whole of Europe, catchment basin specific measurements are
encouraged. Another example can be found in rural development policies, where regional
identity has become an important issue. Thus the need for sustainability assessment at an
intermediate level pushes forward.
The goal of this study is to evaluate existing methods for sustainability measurement for
their possibilities of application at the supra-farm level and, if necessary, to formulate
suggestions for the construction of a specifically adapted sustainability monitoring tool.
Methods
Criteria mainly used to assess sustainability monitoring methods are:
• Integration of the three dimensions of sustainability: existing monitoring tools often focus
on only one or two aspects, in general economics and/or ecology. However, the integrated
achievement of economic, environmental and social performance needs to be measured,
while giving equal weight to all three dimensions.
• Possibilities to identify and evaluate both positive and negative externalities: too often
assessment instruments only picture negative externalities (resource depletion or
emissions). Typically though, agriculture also has a number of positive externalities, such
as providing landscape amenities or specific ecosystems. Analogously agriculture can be
very important to social structures in rural areas. A comprehensive assessment of
agriculture’s sustainability can thus only be reached by a method that values both types of
externalities on an equal footing.
• Trade-offs: the goal of integrated sustainability assessment is not in the final figure (e.g., a
multi-dimensional value added, an ecological footprint, ...), but in the elements of this
44
Session A1: Methods for integrated assessment of agricultural systems and land use
•
•
•
•
figure, especially in their mutual dependency and in their evolution in time. A measuring
tool needs to be able to answer questions such as ‘What is the effect of meeting Nitrate
Directive goals on the other aspects of sustainability?’
Transfers: can the methods adequately represent transfers between farms or regions or
between the agricultural sector and the rest of society?
Innovations: can the methods be used for ex-ante evaluation of the introduction of new
technologies or policies?
Data availability and feasibility of calculation: are appropriate data readily available or can
reasonable proxies be derived from existing statistics? Is indicator calculation feasible
within a reasonable time frame? E.g., monitory valuation of non-commodity goods often
requires time-consuming revealed or stated preference techniques (Freeman, 1993).
Unit of measurement: sustainability indicators can be expressed in monetary values,
hectares, indices, etc. Monetizing environmental and social effects has the advantage of
making them directly comparable to economic values, but the implementation is often
cumbersome. ‘Hectarizing’ all effects seems an attractive path, especially for agriculture,
where land is an important production factor (Hubacek & Giljum, 2003). In index methods
selection of the indicators, their relative weights or over-aggregation might be drawbacks.
Results and discussion
To assess sustainability at supra-farm level, diverse methodologies fit with the definitional
diversity of sustainability. In other words, a general framework can be developed, but the
assessment methodology should be tailored to specific research questions and contexts. Based
on our criteria and the evaluation of existing methods (on different levels), we will formulate
suggestions for a general assessment framework that meets some specific needs.
References
Atkinson, G., et al., 2004. Framework for environmental accounts for agriculture. Economics
for the Environment Consultancy (eftec), London.
Freeman, A.M., 1993. The measurement of environmental and resource values. Theory and
methods. Resources for the future, Washington DC, USA.
Daly, H.E. & J.B. Cobb, 1989. For the common good: Redirecting the economy towards
community, the environment, and a sustainable future. Boston Beacon Press, USA.
Figge, F. & T. Hahn, 2004. Ecol. Econ. 48(2): 173-187.
Hubacek, K. & S. Giljum, 2003. Ecol. Econ. 44(2-3): 137-151.
Meul, M., et al., 2008. Agronomy for Sustainable Development 28: 321-332.
Rigby, D., et al., 2001. Ecol. Econ. 39(3): 463-478.
UN, 2003. Integrated Environmental and Economic Accounting. Studies in Methods, Series
F, No.61, Rev.1 (ST/ESA/STAT/SER.F/61/Rev.1).
Van Passel, S., et al., 2007. Ecol. Econ. 62(1): 149-161.
Wackernagel, M. & W. Rees, 1995. Our ecological footprint: Reducing human impact on the
earth. New Society Publishers, Philadelphia.
WCED, 1987. Our Common Future. World Commission on Environment and Development.
Oxford University Press, UK.
Wustenberghs, H., et al., 2004. Monitoring agriculture’s multifunctionality by means of
integrated nation-wide accounting. 90th EAAE Seminar, Rennes, October 27-29, 2004.
45
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme A
Methodology for integrated assessment
Session A2: Methods for up- and downscaling in
integrated assessments
Session organizers:
Frank Ewert
University of Bonn, Germany
Martin Volk
Helmholtz Centre for Environmental Research - UFZ, Leipzig, Germany
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Strategies for addressing spatial and temporal scale
in Integrated Assessment of agricultural systems1
J.M. Antle
Montana State University, Bozeman, MT 59717, USA
Contact: Jantle@montana.edu
Introduction
In this Keynote Address, I argue that integrated assessment models should be designed
primarily to provide information needed to support informed policy decision making. To be
useful, information must be timely and sufficiently accurate to improve the quality of
decisions. These considerations imply that integrated assessment model design should be
demand-driven in the sense that models need to be tailored to address the policy questions of
interest, recognizing that there will always be trade-offs between data availability, model
complexity, and ability to accurately represent processes and interactions between subsystems. There is no one-size-fits-all approach that will work, given the time and other
resource limitations that exist.
In choosing a modelling strategy, one of the major challenges faced by modellers is how to
deal with the fact that data and processes operate at different spatial and temporal scales. In
this Keynote presentation, my goal is to characterize the challenge and discuss one approach
to resolving it.
Spatial and temporal scales in agricultural systems
For my presentation, I define an agricultural system as a complex, human-managed system
intended to provide both market and non-market goods and services. The ‘farm’ is the basic
economic unit which is comprised of a collection of land management units with associated
flows of inputs and outputs (De Jager et al., 1998). As Ewert et al. (2009) and other papers in
this session observe, the farm interacts with markets, and more generally, within a hierarchy
of biophysical and social systems (biological and physical soil and micro-climates, fields,
farms, agro-ecological zones, regional climate, local, national and international markets and
policy regimes). Various physical, biological and human processes and systems operate at
each of these levels in the hierarchy, many on distinct spatial and temporal scales. As one
example, crop growth and pest populations may operate at the level of the cell and crop plant
levels on extremely high temporal resolution, but management processes may operate at the
field scale on daily, weekly or seasonal time steps in conjunction with market and policy
processes that operate at much larger spatial and temporal scales.
Research-driven versus policy-driven modelling strategies
Given these nested levels of complexity, how is the researcher to proceed? Given data and
other resource limitations, it is not currently possible (nor likely to be in our lifetimes) to
create the ‘grand synthesis’ of data and models at all levels covering the entire landscape in
even a relatively small region, not to mention larger regions or the globe. So some
compromises must be made. The question I pose here is how to define a strategy to making
the inevitable trade-offs.
My observation is that, despite giving lip-service to policy, most models are research
driven. By research driven I mean designed to be as true to scientific principles as possible,
and outside of economics, this means process-based in most cases. Research-driven also tends
to mean that the models are motivated by the goal of publication in peer-reviewed scientific
journals. A consequence of the research orientation is that information produced by the
1
Keynote presentation
48
Session A2: Methods for up- and downscaling in integrated assessments
models may or may not be relevant or useful to policy decision makers. For example, it may
not be possible to ‘scale-up’ the results to the geographic regions of interest to policy design.
I shall argue in my Keynote Address that in order to meet the objective of supporting
informed policy decision making, model design should be policy-driven, meaning that
stakeholders should participate in the identification of the key quantifiable indicators that the
models produce, and in the design of technology and policy scenarios to be investigated with
the models. Not coincidentally, this is the approach advocated in Trade-off Analysis
(Stoorvogel et al., 2004). Once these issues have been addressed, the researchers can
determine the appropriate disciplinary components to be incorporated into the agricultural
system model, and the best spatial and temporal scales for each model component to operate,
and how the disciplinary components can be integrated.
Discussion
One conclusion that can be drawn from research on the impact of spatial and temporal scales
is that in some cases, a high-degree of resolution is required to obtain a reasonably accurate
analysis of tillage erosion (e.g., Antle & Stoorvogel, 2006), whereas in the analysis of soil
carbon sequestration, it appears possible to aggregate carbon rates to the regional level and
obtain accurate predictions of carbon sequestration potential (Antle et al., 2007). Likewise, it
appears possible in some cases to significantly simplify data and models and obtain results
sufficiently accurate to inform policy decisions (Antle & Valdivia, 2006). With further
research, there appears to be a good possibility that these kinds of findings can be generalized
so that researchers will know what kinds of models work well for different types of systems
and policy questions.
These considerations suggest that there is no one ‘right’ scale of analysis in integrated
assessment of agricultural systems. Rather, at each hierarchical level, it should be possible to
develop modular frameworks with standard inputs and outputs that can be adapted to different
systems and policy questions, and researchers can choose from this ‘toolbox’ when designing
an analysis to meet the informational needs of stakeholders.
References
Antle, J.M. & J.J. Stoorvogel, 2006. Environment and Development Economics 11(1): 39-58.
Antle, J.M. & R. Valdivia, 2006. Australian Journal of Agricultural and Resource Economics
50(1): 1-15.
Antle, J.M., et al., 2007. Climatic Change 80(1-2): 145-171.
De Jager, A., et al., 1998. Agric. Ecosys. Environ. 71: 37-48.
Ewert, F., et al., 2009. AgSAP Conference 2009, these proceedings.
Stoorvogel, J.J., et al., 2004. Agricultural Systems 80(1): 43-66.
49
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
EU-wide spatial down-scaling of results of a regionalized agricultural
economic model to analyse environmental impacts
Wolfgang Britz
Nussallee 21, D-53115 Bonn, Germany
Contact: wolfgang.britz@ilr.uni-bonn.de
Introduction
Robust simulation responses for economic modelling require observations for estimation or
validation of how agents react to changes in market and policy signals. Most often, at least for
large-scale analysis, solely time series data for larger administrative units provide the necessary variance in signals and responses. For Pan-EU economic analysis, already a sub-national
regional resolution as in CAPRI (Common Agric. Policy Regional Impact model, Britz et al.,
2007) is, therefore, rather unique. With higher spatial variance inside administrative units,
estimates of environmental impacts based on regional averages may be considerably biased
under non-linear dependencies between these impacts and soil parameters, climate or farming
practice. Certain impacts can only be assessed in their proper spatial setting such as the
relation between nitrate leaching and drinking water resources or the effects of land use on
specific habitats. Statistical down-scaling provides, therefore, a bridge between large-scale
economic and environmental analysis in agriculture. The behavioral response regarding crop
shares, yields or animal stocking densities is simulated for administrative units with the
economic model. The downscaling tool then consistently distributes these changes to georeferenced units below the administrative level. Environmental impacts are then analysed with
indicator calculators or simulated with biophysical models at an appropriate spatial resolution.
A work package of the project ‘CAPRI-Dynaspat’ contributed a spatial down-scaling tool to
CAPRI. It established the necessary geo-referenced data bases and developed methodologies
and software to consistently dis-aggregate for the whole EU27 all major results from the
CAPRI modelling system to about 150.000 clusters of 1×1 km grid cells (Leip et al., 2007).
Methods
Firstly, all available data as the soil map, climate data, digital elevation model, land cover
map, or administrative boundaries have been rastered to a 1×1 km resolution for the EU27.
Next, so-called Homogenous Soil Mapping Units (HSMUs) were defined consisting of
clusters of 1×1 km pixel identical in soil, slope, administrative unit and dominant land cover,
based on the assumption that these delineation features are decisive in determining differences
in farming practice below the administrative unit level. Each HSMU covers between one and
several ten thousands 1×1 km cells, capturing spatial variability where it matters. These about
150.000 units were deemed better suitable for the necessary crop share representation for the
about 35 crops covered by CARPI compared to pre-dominant land-cover presentation, and
kept the amount of data cells manageable.
A statistical estimator assigns shares for individual crops and some broad non-agricultural
land cover classes to the HSMUs maximizing the joint posterior density of the shares so that
the area of each HSUM is exhausted and regional hectares for each crop and land use type are
recovered. The necessary a-priori distributions for the shares were derived from locally
widened logit-regressions on soil, climate and relief parameters, using the LUCAS sample as
observations (Kempen et al., 2005). The sub-regional distribution of yields is based on water
and on-water limited potential yields for major crops provided by JRC’s MARS project,
taking into account regional data on irrigation shares and FAO’s irrigation map. Assigning
stocking densities for the about dozen animal activities in CAPRI follows as similar approach
as for crop shares, the a priori distribution are here based on regression on Farm Structure
Survey data at NUTS 2/3 regions.
50
Session A2: Methods for up- and downscaling in integrated assessments
The stocking densities allow estimated organic manure availability per HSMU, using
spatial smoothing to wipe out peaks. Together with the yield estimates, a spatial distribution
of fertilizer application rates is then derived recovering the average regional organic and
mineral fertilizer application rates per crop as estimated by CAPRI. The yields and fertilizer
application rates are a major input data into a spatial explicit link to the biophysical model
DNDC (Li et al., 1994), based on two approaches. The first, more standard one performs
DNDC runs for major crops and representative sites. However, computing time and storage
limitations render that approach only feasible for selected applications. Therefore, a statistical
response surface was derived from DNDC runs with different levels of organic and mineral
fertilizer doses, for a large sample of different soil and climate conditions for each crop and
integrated in the downscaling tool (Britz & Leip, 2008).
Results and discussion
The whole down-scaling procedure is set-up in GAMS code and linked to the overall CAPRI
modelling system. Results can be generated for the base year, for projection results or for
simulation runs. A JAVA-based GUI allows to generate maps and tables from the result set.
The major problem with statistical down-scaling consists in the limited access to georeferenced high-resolution data to estimate and/or validate the down-scaled results against
real-world observations. Validation of the crop shares were performed at the level of subregional administrative units (Elbersen et al., 2006). Especially the shares for important arable
crops were estimated rather convincingly. Differences in grassland share were often
significant, but are already present at the aggregate regional level between the different data
source (land use maps, Farm Structure Survey FSS, land use statistics), and therefore
probably to a large extent not due to the statistical procedure. The approach certainly would
benefit from access to high-resolution data sets as FSS data at municipality level.
Despite the remaining uncertainty due to missing observations, e.g., regarding georeferenced fertilizer application rates per crop, statistical down-scaling has some advantages
over competing methods. Compared to bottom-up approaches which model economic
behavior at a high spatial resolution, it can rely on existing data sources for the economic
modelling part and does not require additional surveys. Due the high cost of the latter, a fullcoverage bottom-up application to the EU level is not probable, asking for some kind
extrapolation to map results for selected spots to all regions. It cannot be excluded that
eventual gains from the fine-tuned and interlinked economic and biophysical modelling for
selected spots are lost in the extrapolation step. Bio-economic bottom-up approaches are
however potentially able to model changes in farming behaviour as switches in tillage
techniques not covered in most top-down approaches. Compared to most existing LUCM
approaches, the CAPRI downscaling tool distributes single crops, and covers further farm
practice parameters as yields, animal stocking densities and fertilizer application rates.
However, it is currently not able to account for neighbouring effects between land use
categories as in classical LUCM, but could be easily linked to such an approach.
References
Britz, W., et al. (Eds), 2007. Final report of the CAPRI-Dynaspat Project. Institute for Food
and Resource Economics, University of Bonn, Germany.
Britz, W. & A. Leip, 2008. NEU-annual meeting PSG2, Ghent 18.02.2008, Belgium.
Elbersen B.S., et al., 2006. Protocols for spatial allocation of farm types, SEAMLESS Report
No. 19, www.SEAMLESS-IP.org, 107 pp.
Kempen, M., et al., 2005. In: Ed. F. Arfini, Modelling agricultural policies: State of the art
and new challenges. Monte Universita, Parma, Italy, pp. 810-831.
Leip, A., et al., 2008. Biogeosciences 5: 73-94
Li, C., et al., 1994. Cycles 8(3): 237-254.
51
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Calibration of regional programming models to exogenous supply
elasticities by a variable cost function: The CAPRI-EXPAMOD example
Wolfgang Britz, Marcel Adenäuer
Nussallee 21, D-53115 Bonn, Germany
Contact: wolfgang.britz @ilr.uni-bonn.de
Introduction
The calibration of programming models to exogenous supply elasticities by setting parameters
of a quadratic cost function is an established idea in the context of ‘Positive Mathematical
Programming’ (Howitt, 1995). Existing approaches (e.g., Helming et al., 2001) refer only to
own price elasticities, and systematically overestimate the supply responsiveness of the
calibrated model by neglecting the effect of changes in dual values. The paper builds on
attempts to estimate variable cost functions of programming models (e.g., Britz & Heckelei,
2000, Heckelei & Wolff, 2003; Jansson, 2007) and proposes an estimator based on an
analytical derivation of the supply response as proposed in Heckelei (2002), including crossprice terms. The calibration which defines a symmetric quadratic matrix of cost function
parameters to recover a given own- and cross price elasticities. It may be seen as a special
case of an estimation characterized by zero degrees of freedom and a specific data generating
process. The approach is motivated by the integration of the large-scale agricultural sector
model CAPRI (Common Agricultural Policy Regionalized Impact) into the SEAMLESS
model chain (Van Ittersum et al., 2008) where supply elasticities derived from simulations
with farm type models and subsequently statistically extrapolated to cover all of Europe
(Bezlepkina et al., 2007) are used in the calibration process for CAPRI’s regional
programming models.
Methods
The basic structure of each CAPRI regional model may be described as
max Z = gm ' x − d ' x − 1 2 x ' Qx where gm are per activity gross margins, and d and Q
x
parameters of a variable cost function, all regional specific, subject to Ax ≥= b [ λ ] , 0 ≤ x . The
vector x comprises levels for about 50 crop and animal production activities in each of the
about 250 regional models in CAPRI x, b is a vector of resource constraints and A the related
coefficient matrix, again region specific. The calibration estimates the matrix Q* which
generates the Jacobian matrix of the activity level equations
H* =
−1
∂x
= Q −1 − Q −1A* ' ( A*Q −1A* ') A*Q −1
∂gm
1
2
H
A*
3
Estimator
4
H*,Q*
5
CAPRI
6
H*
52
which comes close to a matrix H derived
from exogenous elasticities at a given
simplified coefficient matrix A*. The
following graphic gives an overview of the
whole process. In(1), a simplified
coefficient matrix A* is derived which
comprises the major constraints such as
land and fodder balances, set-aside or sales
quotas from the CAPRI regional
programming models and sets feed input
coefficients to constant levels. At the same
time, low and high yield variants for the
crop activities in CAPRI are aggregated.
Session A2: Methods for up- and downscaling in integrated assessments
The gross margins for activity k are defined as
gmk = ∑ okj p j + premk − costsk where o are output coefficient, p are prices, prem are
j
subsidies and costs variable costs per activity. From there, supply effects describing reaction
to output price changes can be derived as
⎛
⎞
∂ ⎜ ∑ olj p j + preml − costsl ⎟
∂xk
∂xk ∂gml
∂xk ⎝ j
⎠ = ∂xk o = H o
=
=
lj
kl lj
∂p j ∂gml ∂p j
∂gml
∂p j
∂gml
∂ ∑ xk oki 0
0
pj
p 0j
∂Si p j
k
=
=
H
o
o
and elasticities defined as ε ij =
. The estimator then
∑
kl ki lj
∂p j Si0
∂p j
Si0
Si0
kl
derives in (3) a matrix Q* along with a matrix H* which deviates from the matrix H as
defined from the elasticities due to differences in structure and parameterization between
CAPRI and FFSIM. In order to ensure regularity, Q* must be positive definite, guaranteed by
a Cholesky decomposition. Q* and an appropriately set vector d calibrate the regional
programming models in (4). The resulting models are then used in the (5) to perform
sensitivity analysis while changing the revenues gm, and the resulting supply effects are in
the final step compared with H.
Results and discussion
For a large-scale calibration exercise involving the 250 regional programming models of
CAPRI and about 50 activities, numerical stability and an acceptable estimation error are
crucial. The latter ensures mutual consistency between prices simulated in CAPRI based on
the supply response of the regional programming models and the simulation behavior of
FFSIM used to analyse environmental or farm type specific impacts at simulated prices.
Compared to existing approaches, we allow for the integration of cross-price effects while
considerably reducing the estimation error. Tests with randomly drawn matrices of supply
elasticities were performed. The errors between the estimated supply effects and the one
derived in sensitivity analysis are in an acceptable range. The major drawback compared to
simplistic calibration approaches is the numerical complexity resulting from the need to
integrate both matrix inversions and a Cholesky decomposition for large matrices in the
estimation process. It leads to considerable computing time. Therefore, parallel estimation of
several regions combined with algorithms to generate suitable starting values are used to
speed up processing.
References
Bezlepkina, I., et al., 2007. EXPAMOD: Component to statistically extrapolate from FSSIM
models to other farm types and regions including aggregation to NUTS2: motivation,
description and prototype, PD3.6.11.2, www.SEAMLESS-IP.org, 28 pp.
Britz, W. & T. Heckelei, 2000. Cahiers d’Economie et Sociologie Rurales 57: 28-50.
Heckelei, T., 2002. Habilitation Thesis, University of Bonn, Germany.
Heckelei T. & H. Wolff, 2003. European Review of Agricultural Economics 30: S. 27-50.
Helming, J.F.M., et al., 2001. In: Eds.T. Heckelei, et al., Wissenschaftsverlag Vauk, Kiel.
Howitt, R.E. 1995. American Journal of Agricultural Economics 77(2): 329-342.
Jansson, T., 2007. Dissertation, Bonn University.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
53
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Compliance and costs of compliance:
A procedure to obtain estimates at a disaggregated level
B.S. Elbersen, R. Jongeneel, M. Kempen, M. Mikk, A. Peepson
Alterra, Wageningen UR,
P.O. Box, 6700 AA Wageningen, The Netherlands
Contact: Berien.elbersen@wur.nl
Introduction
Cross compliance was introduced as part of the CAP Reform of 2003 with Regulation
1782/2003. It involves member states making receipt of the direct payment aid, called the
Single Payment, conditional on farmers meeting two sets of standards. The first, called
Statutory Management Requirements (SMRs), relate to 19 pieces of EU environmental,
public, animal and plant health and animal welfare legislation. The second set, referred to as
standards of Good Agricultural and Environmental Condition (GAEC), relate to the
appropriate management of soils and the minimum maintenance of agricultural land and
features found on that land. These SMRs and GAEC obligations apply to all farmers who
claim the Single Payment, and apply to the entire farm holding.
In order to assess the impacts of cross compliance, insight into the compliance levels as
well as the costs of compliance is needed. More in particular, one would also be able to detect
changes in compliance rate over time in order to determine whether cross-compliance was
successful in increasing the level of compliance with EU regulation. Unfortunately, only
limited information about compliance and its costs is currently publically available. Examples
of recent EU wide studies that provide compliance estimates are the Cross-compliance project
and the study by the Alliance Environment (2007) prepared for DG-Agriculture. But even
these studies, only provide estimates at national level. Since the impact of compliance to
regulation is likely to vary between different locations (depending on type of farming, area
designated as a Nitrate Vulnerable Zone or Natura 2000 area, etc.), still a further
disaggregation of these estimates to regional levels (e.g., Nuts 2/3 regions, Nitrate Vulnerable
Zones) is crucial. The aim of this document is to present a framework that allows for
determining best estimates of compliance and costs of compliance at a disaggregated regional
level (Nuts 2) in terms of farm, livestock and land use shares compliant and non-compliant.
The method is applied to the Nitrate Directive being one of the SMRs in the Cross
Compliance instrument.
Methods
Regulations aim to direct farmer’s behaviour in such a way that certain (minimum) standards
are respected. A farmer has the option to either comply or not comply. In the latter case (s)he
faces the risk to be detected as a non-compliant, and risks that part of the direct payments
might be withdrawn (depending on the significance of the violation). Figure 1 provides a
Enforcement
Adjustment
needed
Not
comply
Non
compliance
Comply
Affected
Farmer
Regulatio
n
Compliance
No adjustment
needed
Nonaffected
farmer
54
Figure 1. Farm affectedness,
farmer
adjustment
and
compliance (Based on Herzfeld
& Jongeneel, 2008).
Session A2: Methods for up- and downscaling in integrated assessments
more refined scheme of the link between a regulatory requirement and compliance. Already
by characterizing farms in being affected or non-affected by an obligation may help to get
more insight into the compliance issue. If data are available at farm group level which could
help to identify which part of the affected farms is likely to need to make adjustments, this
further informs about which part of the farming population is most likely to have lower levels
of compliance.
There are three categories of farms which have a large probability to be compliant: nonaffected farms, affected farms but already exceeding EU standards, and affected farms which
take action in such a way as to become compliant. Here the non-affected farms can be said to
be compliant by definition. With respect to the affected farms the compliance rates are
calculated in the following steps:
1. Identify groups of farms per region (Nuts 2 and NVZ) at no, medium and high risk of not
complying with the Nitrate Directive.
2. Allocation of groups of farms at no, medium and high risk of not complying with Nitrate
Directive to NVZ
3. Calculation of compliance rates per sectoral farm type group using nationally reported
compliance and breaches levels
4. Calculation of the final total compliance rates for the Nitrate Directive obligations
expressed as shares in total holding, livestock and utilized agricultural area shares per
region (Nuts 2 and NVZ).
As for the estimation of the compliance costs specific estimates need to be made per
obligation. According to information from the Cross Compliance project, the Nitrate
Directive is one of the standards for which significant costs at farm level need to be made to
become compliant. Estimation of these cost are based on operational costs (transportation,
handling and spreading costs of surplus manure, labor cost associated with proper registration
and record keeping of manure applications) and costs associated with investments that have to
be made (sufficient manure storage capacity, manure transport and spreading equipment).
Results and discussion
The results show that the overall compliance levels with the Nitrate Directive in 2005 were
already quite high in most EU regions. However, there are several regions where livestock
systems dominate and which coincide with where the Nitrate Vulnerable zones are. In these
regions we still see a very high compliance rate in terms of farm shares reaching above 90%,
but in terms of animal shares the rate is between 70%–90%. These regions therefore have a
significant higher risk for non compliance but also for higher costs of compliance.
References
Alliance Environment, 2007. Evaluation of the application of Cross Compliance as foreseen
under the Regulation 1782/2003. All national reports and synthesis reports prepared for
DG Agriculture.
De Roest, K., et al., 2006. Framework for analysis of the cost of compliance. Deliverable 7,
CROSSCOMPLIANCE project, EU 6th Framework Programme, Priority 8.1 (European
Commission, DG RTD, contract no. SSPE-CT-2005-006489).
Herzfeld, T. & R. Jongeneel, 2008. Economics of Compliance: A review of theories and
application to agriculture. Paper presented at IAMO Forum conference 2008, IAMO, Halle
an der Saale.
Jongeneel, R., et al., 2008. Final Report: Compliance and costs of compliance with standards
and their impact on the EUs external competitiveness. CROSS-COMPLIANCE project,
EU 6th Framework Programme, Priority 8.1 (European Commission, DG RTD, contract no.
SSPECT-2005-006489), 120 pp.
55
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Complex scaling issues in integrated assessment modelling:
Approaches used and their integration into SEAMLESS-IF
F. Ewert, M.K. Van Ittersum, O. Therond, I. Bezlepkina, E. Andersen, T. Heckelei,
G. Flichman, N. Turpin
Institute of Crop Science and Resource Conservation, University of Bonn,
Katzenburgweg 5, D-53115 Bonn, Germany
Contact: Frank.ewert@uni-bonn.de
Introduction
Agricultural systems and associated problems of sustainability and sustainable development
are typically complex. Several methods have been developed to effectively describe and
analyse complexity. It is argued that hierarchy theory can provide a much needed conceptual
framework for developing successful scaling theories and approaches to reach this objective.
Hierarchy theory partitions complex systems into nested levels that share similar temporal and
spatial scales.
In the recently developed integrated assessment (IA) modelling platform SEAMLESS-IF
(Van Ittersum et al., 2008) the concept of hierarchical system has been adopted. SEAMLESSIF integrates relationships and processes across disciplines and scales which are
conceptualized and modelled following the paradigm of hierarchy theory. Several scaling
approaches are used in SEAMLESS-IF which are presented in this paper. Specific emphasis is
on the integration of these different approaches. The progress achieved and the challenges
experienced in modelling complexity within the paradigm of hierarchy theory are discussed.
Scaling methods in SEAMLESS-IF
Different methods have been employed in natural sciences to estimate systems responses
across scales or levels of organization (see Ewert et al., 2006). These include data
extrapolation, data aggregation, model linking, development of summary models, scaling or
aggregation of model parameters. The method chosen depends on the specific objective.
SEAMLESS attempts to capture the biophysical, economic and social, and to some extent the
institutional dimensions of agricultural systems. Modelling efforts consider different levels of
organization from the local field to the globe. Central to the SEAMLESS-IF approach is the
linking of different models across scales and disciplines (Figure 1) to address (parts of) the
complexity of the agricultural system. The core set of processes for which models were
selected refers to the market level (CAPRI), the farming system level (FSSIM) and to the
biophysical processes at the field level (APES) (Figure 1). Additional models can be
considered if required.
The linking of models required a set of other approaches to scale information. These include:
• Generation of coefficients for (static) model linking;
• Development of typologies to define simulation units and to support data sampling;
• Scaling of model parameters;
• Extrapolation and aggregation to transfer data across scales (e.g., from farm to EU)
Due to the complexity of the developed model chain computational extensive dynamic model
links were not considered, only one feedback loop (with a single iteration) is implemented
from FSSIM to CAPRI and back to FSSIM. The link between APES and FSSIM is static in
the form that APES generates coefficients for the FSSIM models.
The complex model structure(s) required simplification of data to consistently apply
models across space. Different types of data are grouped into homogenous classes determined
by the factors that explain most of the variability in the data with respect to specific variables
of interest. The resulting typologies are the basis for up- and down-scaling procedures in
56
Session A2: Methods for up- and downscaling in integrated assessments
SEAMLESS-IF. The following three main typologies have been developed and used so far;
the Agri-environmental Zonation (AEnZ), the farm typology and the administrative (NUTS2)
regions (the latter has not been developed but is used in the project). These typologies are
used to define simulation units to which data are scaled (e.g., aggregation of biophysical input
data per AenZ for APES), to support data sampling (e.g., collection of management activity
data per sample region for FSSIM), to transfer data between models that represent different
scales (e.g., extrapolation of FSSIM outputs to feed CAPRI) and to scale up indicators (e.g.,
aggregation of FSSIM outputs to compute indicators at regional level). As there are spatial
mismatches between typologies, approaches have also been developed to link simulation units
of different typology (e.g., spatial allocation of farm types across AenZs).
CAPRI
Market
Supply
Demand
…
FSSIM
Farm
Type I
Type II
Type n
Field
APES
Field 1
Field 1
Field n
Figure 1. Schematic representation of nested linking of CAPRI-FSSIM and APES within a
hierarchical system.
Discussion and conclusions
We have combined several scaling methods into a coherent modelling framework for IA of
complex problems in agriculture. First experiences in working with SEAMLESS-IF show that
the developed framework provides useful results and advances IA capabilities of earlier
frameworks. Yet, the developed model chain is relatively complex and further simplifications
may improve its usability and transparency. Further modelling work should therefore focus on
developing simplified models for the different parts of the overall system. This can only be
achieved if the most important drivers and processes are identified and understood for well
specified problems. The developed chain (Figure 1) can be of assistance in the process.
References
Ewert, F., et al., 2006. Proceedings of the iEMSs. Burlington, USA, July 2006.
http://www.iemss.org/iemss2006/sessions/all.html.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
57
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
EXPAMOD – connecting the farm and market levels
I. Pérez Dominguez, I. Bezlepkina, A. Kanellopoulos, T. Heckelei,
A. Oude Lansink, E. Romstad
Joint Research Center of the EC,
Institute for Prospective Technological Studies, Sevilla, Spain
Contact: Ignacio.perez-dominguez@ec.europa.eu
Introduction
Farm management models help quantifying changes in land use patterns and agronomic
practices for a given set of prices and policies. These changes impact indicators of multifunctionality like the visual character of agricultural landscapes, biological diversity, and
pollution levels. However, prices are unlikely to remain constant as policies change. Therefore, farm level optimization results based on exogenous prices may no longer be valid.
Market level models, on the contrary, are able to capture the supply and price impacts derived
by policy shocks at the farm and regional level, but are generally not sufficiently detailed for
the calculation of environmental impacts, since they lack differentiated agronomic practices.
By linking farm level and market level models through a newly developed econometric
model, EXPAMOD, we seek to mitigate this weakness and endogenize the price-quantity
response in farm management models. This paper explains how EXPAMOD works, and
applies it to a number of policy scenarios as used in SEAMLESS (Van Ittersum et al., 2008).
The main modelling benefit of our approach is that it enables combining the strong points
of farm management (the detailed linkages to agronomic and natural science models), and
market level models (endogenous prices and closure of the economy). From a policy analysis
perspective the resulting farm type and acreage responses provide a much improved base for
various environmental and landscape modelling exercises.
Methods
Our modelling chain consists of three models. First, FSSIM, a normative mathematical programming model that is specifically fitted for model farm types and regions across Europe.
Second, CAPRI, a comparative static partial equilibrium model of the European agricultural
sector. Third, EXPAMOD, an econometric model that estimates changes in supply responses,
and statistically propagates these responses to out of sample farm-region combinations.
The first step in our model chain is a collection of farm models, FSSIM, for several representative farm types with different exogenous price sets for a baseline and a policy scenario.
Next, the econometric model, EXPAMOD, is used to estimate changes in supply responses
for the exogenous price sets. Changes in relative farm level profits are then used to assign
new weights to the farm types covered by the analysis. The supply changes at the micro level
and the revised weights for the farm types are then used to adjust supply in the market model
CAPRI, so that revised prices are obtained. Finally, these prices are fed back to FSSIM.
A major challenge for connecting the farm and market level is that the number of model
runs at the farm level is limited due to the input data requirements of farm level model,
FSSIM. Therefore, FSSIM is run for a stratified sample of model farms and regions to cover
the main variation in the EU-27 of farm types and agro-climatic zones. Following Andersen et
al. (2007), the selection of sample regions was made at the NUTS2 level, as this is the
minimum disaggregation level for the market models in SEAMLESS. Nevertheless, the major
source of farm type data is only available for FADN regions, i.e. regional classification used
by Farm Accountancy Data Network, so that a mapping between both classifications is
necessary (Janssen et al., 2008). Due to laborious requirements on the data collection, 16
sample regions were targeted as this was judged to be feasible for data collection and
modelling purposes.
58
Session A2: Methods for up- and downscaling in integrated assessments
In FSSIM regional supply at NUTS2 is recovered by aggregating farm type supply using
farm weights from the FADN (see, Wieck & Heckelei (2007) for further details). How well
the farm types selected represent the farm composition within a NUT2 region is an important
issue for EXPAMOD. The weights derived from the observed data are only suitable for the
calibration of the model in the base year, since they refer to FSSIM results. Since
SEAMLESS targets the ex-ante impact assessment of agricultural policies, the projection of
agricultural markets to a baseline period in the future, requires additional assumptions on
technological development, changes in consumer demand, inflation, GDP growth, etc. These
effects are explicitly handled by the CAPRI model and fed back into FSSIM.
The price impacts from supply changes in the farm optimization models generate
information interpreted as ‘pseudo-observations’ for the econometric estimation of
EXPAMOD. The current simulation design implements varying 'one-price-at-a-time'. The
price vector, for each scenario, is kept at the 100% level of the initial price vector obtained
from CAPRI and additional price-quantity vectors for four different price shocks in FSSIM
are considered (–40%, –20%, +20%, and +40% from the initial price). These scenarios
generate information on own and cross price-quantity effects that are reintroduced in the
extrapolation routine of EXPAMOD. In most cases, price changes are likely to be far smaller.
However, sufficient variation of prices is needed to stabilize the estimates of the price-related
coefficients.
Results and discussion
The tests, performed with a flexible functional form, show plausible results and a high
statistical explanatory power. Nevertheless, some poor predictions have been observed for
estimations with a low number of observations and high number of parameters. This should
be easily solved by generating a higher number of pseudo-observations. Additionally, a higher
variance has been observed compared to the data (especially for products under a quota
regime, such as sugar beet) and a closer link of results to the biophysical and farm
management variables would be desirable. Out of sample tests are envisaged to provide
relevant validation of strengths and weaknesses of the statistical extrapolation.
Our approach may also be applicable to scale up non-economic results, such as environmental
impacts. However, further research is needed to refine the method employed for up-scaling
such impacts, in particular where the spatial distribution of impacts matter.
References
Andersen, E., et al., 2007. The environmental component, the farming systems component
and the socio-economic component of the SEAMLESS database for the Prototype 2,
D4.3.5-D4.4-D4.5.4. SEAMLESS integrated project, 195 pp.
Janssen, S., et al., 2008. A European database for policy evaluation and assessment of
agricultural systems. Environmental Science & Policy. (forthcoming)
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
Wieck, C. & T. Heckelei, 2007. Agricultural Economics 36: 203-220.
59
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Farm typology identification by multi-variate analysis as a method
to scale-up results of integrated impact assessment
E. Righi, G.C. Pacini, S. Dogliotti, V. Aguerre, W.A.H. Rossing
Department of Agronomy and Land Management (DISAT), University of Florence,
Piazzale delle Cascine, 18 – 50144 Firenze, Italy
Contact: emilio.righi@unifi.it
Introduction
Farm typology identification within a specific case study area is a means to summarize the
variability among farms in terms of agro-ecological and socio-economic livelihood aspects
into farm groupings or ‘typologies’ and thus obtain insight in the major factors which
distinguish farming systems.
The aim of this paper is to present a methodology to identify farm typologies and indicate
how this method could be linked to model-based integrated impact assessment at regional
level. The method will be presented as part of the description of a case study in South
Uruguay, within the context of the EU project EULACIAS which aims to improve farm
livelihoods in an action-research setting using systems approaches.
Methods
The method starts by defining a complete list of classification variables describing farm
structure and functioning for three general categories of information: Land resources, Socioeconomic aspects and Infrastructure and equipment. Principle Component Analysis (PCA)
was used to compare different sets of key variables and to choose one of them which
produced a meaningful differentiation of the farm samples. All data were standardized by
total to avoid the influence of different units of measure. A Resemblance matrix was then
calculated using the Bray-Curtis method. Farm typologies were generated by overlaying the
results of two Multi-variate Analyses (MVA) techniques: Multi-Dimensional Scaling (MDS)
and Cluster Analysis (CA). An analysis of similarities (ANOSIM) between the farm
groupings was performed to calculate the level of dissimilarity between all pair-wise
combinations. The last step of the methodology was a SIMPER analysis to determine the
contribution to similarity within a group, which supplies information on the most
representative indicators of each group.
The case study area is located in South Uruguay, in the region around Montevideo
(Dogliotti, 2003). The analysis focused on 2373 specialized vegetable production farms. Farm
typology identification was based on Census data of the year 2000 carried out by DIEA
(Directorate of Agriculture and Livestock Economical Inquiries) of the Ministry of
Agriculture of Uruguay. A random subset of 142 farm (5.98% of the total) was taken, due to
calculation and visualization limits of the software PRIMER 6 (Clarke & Warwick, 2001),
which were used for data elaboration.
Results and discussion
In Table 1, the selected key variables are listed and correspondent average abundances, for
each farm grouping that was identified through the overlapping of MDS and CA, are reported.
ANOSIM confirmed what was found by the combination of MDS and CA, since high levels
of dissimilarity were found between all groups with an acceptable significance level. In Table
1, variables identified by the SIMPER analysis as contributing at least 70% to inside-group
similarity are shown. The meaningfulness of the conclusions was verified with local
researchers.
Within the EULACIAS project the farm typologies are used to select representative farms
for the evaluation of consequences of alternative livelihood strategies on income generation
60
Session A2: Methods for up- and downscaling in integrated assessments
and resource use. SIMPER values that identify and describe typologies can be used for the
selection of representative farms and to build average virtual farms aimed at impact
assessment. Representative farms are used for empirical analyses and, in combination with
virtual farms, to calibrate a modelling framework developed for EULACIAS case-studies,
aimed at explorative and design analyses (Groot et al., 2007). Data retrieved from MVA on
farm group extension and composition are used to up-scale farm-level results, obtained
through in-field surveys and simulation models, to study impact of different livelihood
strategies at the regional level (Figure 1).
Table 1. Average abundances of the key variables calculated for the farm groupings identified
for the specialized vegetable production system in South Uruguay. Most representative
variables of each group are marked in bold.
Indicators
Group 1 Group 2 Group 3 Group 4
Area of vegetable crops (ha) / Total area (ha)
0.45
0.56
0.43
0.68
Protected cultivation area (ha) / area of
0.00
0.00
0.76
0.21
vegetable crops (ha)
Family Labour (days×yr–1) / Total labour
0.95
0.94
0.70
0.49
(days× yr–1)
836.97
931.00 2868.80 2308.53
Labour availability per ha (hours×ha–1)
Irrigated area of vegetable crops (ha) / Total
0.01
0.75
0.79
0.55
area of vegetable crops (ha)
*
1.69
2.36
2.22
2.60
Mechanization Level (classes 1 to 5)
*
Mechanization Level is a qualitative variable: five classes were defined according to the
farm mechanization level.
R e g io n a l L e ve l
T ot al p o pu la tio n o f fa r m s
S ca lin g u p o f th e r es u lt s
t o t he re g ion a l le v e l
Fe a sib le s t ra te g ie s
f or fa rm e rs
M u ltiv a ria te a n al ys is
F a rm T yp o log i e s
R e pr e s e nt a tiv e Fa r m s
A v e ra g e V irt u a l Fa rm s
D a ta in p u t
f ro m fie ld re se a rch
M o de l sim u la tio n o n Fa rm
T yp e A ve ra g e Fa rm
Si m u la t io n M od e l
In te g ra t e d Im pa c t A s s e s s m e n t
D e s i g n o f F a rm L e v e l St ra te g ie s
O n -fa r m e va lu a tio n
of s t ra te g ie s
Figure 1. Diagram showing the framework to up-scale results of integrated impact assessment
at farm to regional level using farm typology identification based on MVA.
References
Clarke, K.R. & R.M. Warwick, 2001. Change in marine communities: an approach to
statistical analysis and interpretation (2nd edition). PRIMER-E Ltd.
Dogliotti, S., 2003. Exploring options for sustainable development of vegetable farms in
South Uruguay. PhD thesis Wageningen University, Wageningen, The Netherlands.
Groot, J.C.J., et al., 2007. Agric. Ecosyst. Environ. 120: 58-69.
61
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Competing use of organic resources, climate variability and interactions at
village scale in a communal area of NE Zimbabwe
M.C. Rufino, J. Dury, P. Tittonell, M.T. Van Wijk, S. Zingore , K.E. Giller
Plant Production Systems Group, Wageningen University,
P.O. Box 430, 6700 AK Wageningen, The Netherlands
Contact: mariana.rufino@wur.nl
Introduction
Mixed crop-livestock systems support the livelihood of the largest number of poor people in
Sub-Saharan Africa. These rural people are exposed to a variety of risks such as recurrent
droughts, political instability, failure of markets for inputs and products. Livestock provide
food (milk and meat) and other services to the household, such as animal traction for
cropping, but also fulfil a financial role. In places with low population density, livestock may
feed on grasslands. As population pressure and competition for natural resources increases,
some of the feed for livestock is produced in the cropland, and because of the continuous
cultivation of the land, the removal of nutrients from the soils needs to be compensated by
adding fertilizers, or making use of animal manure. As population pressure and competition
for natural resources increase further, grasslands tend to disappear, more feeds need to be
produced in cropland, and some feed is imported to the farm.
Farm type
LIVSIM
LIVSIM
LIVSIM
Grassland
GrassSIM
LIVSIM
LIVSIM
HEAPSIM
(B)
LIVSIM
LIVSIM
FIELD
(A)
Figure 1. Schematic representation
of the virtual village (A), and of the
integrated tool FARMSIM (B), with
the models FIELD (simulates crop
production and nutrients dynamics
in the soils), LIVSIM (simulates
animal
production
and
reproduction), HEAPSIM (describes
decomposition of manure and other
organic
resources),
GrassSIM
(describes the availability of grass in
the different grazing units).
The objectives of this study were (i) to understand the dynamics of crop-livestock interactions
under climate variability and (ii) to identify opportunities for intensification. To achieve these
objectives, we developed and tested an analytical tool to analyse crop-livestock interactions at
the scale of the village using a communal area of NE Zimbabwe as example.
Methods
We combined information available for the area of study, collected through interviews,
observations, experiments, and literature. We used the NUANCES-FARMSIM modelling
framework (Giller et al., 2006; Figure 1) adapted and tested for the conditions of smallholder
farming in Majonjo, Murewa, NE Zimbabwe. We constructed a simplified ‘virtual’ village
using the farm typology developed by Zingore et al. (2007) which distinguishes four farmer
resource groups (RG) based on cattle ownership, farm size, production orientation, hiring
labour, and food self-sufficiency. Feeding strategies, herding patterns, crop residues, and
manure management were studied during the dry season of 2006 and the rainy season of 2007
(Dury, 2007). Additionally, the communal grasslands were characterized. The tool includes
62
Session A2: Methods for up- and downscaling in integrated assessments
different levels of detail: it simulates crop production at plot scale, grass production for
different grazing units, animal production at individual level, while management decisions at
considered at both farm and village scales by using rules. The most important transfers of
nutrients: from grasslands to cropland, and between different farms within the village
territory, are kept track of by integrating the different scales in which the different models
operate. Climate variability is accounted for by simulating scenarios using data from the
locality, which includes contrasting rainfall series.
Grain production
(t village-1)
Results and discussion
300
200
100
Fraction produced by
non-livestock owners
0
(A)
Average rainfall
Current
management
(B)
Wetter series
(C)
No access to crop residues
from non-livestock owners
Drier series
Targeted fertilisation
(ISFM)
0.5
0.4
0.3
0.2
0.1
(F)
(E)
(F)
0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Time (years)
Figure 2. Simulated grain production for the whole ‘virtual’ village under three management
scenarios (baseline, no access to cattle to crop residues of the non-cattle farmers (RG3 and
RG4), and targeted fertilization), and using three different rainfall series: (A) average series,
(B) a wetter series and (C) a drier series, and the share of the non-cattle farmers grain
production to total production of the whole village for (D) average rainfall series, (E) a wetter
rainfall series and (F) a drier rainfall series.
The interaction between farmers determines who benefits from integration of crop and
livestock. The removal of C by cattle leads to lower crop yields in the poor fields of these
farmers, and has relatively smaller effect on the fields of the cattle owners that receive animal
manure and fertilizers (Figure 2). Rainfall variability intensifies the interactions, when the
start of the rains is delayed, the low availability of crop residues during the dry season may
lead to loss of animals from the herd. In years of good rainfall the removal is relatively
unimportant. Crop-livestock integration at village scale results in concentration of nutrients in
the farms with larger herds and increases dependency of the poorer smallholders on external
inputs, and other types of exchanges within the village such as labour for food, cash or manures.
In the targeted fertilization scenario, fertilizer compensated for the negative effect of the
interactions, though it may be an unrealistic scenario for a smallholder community in
Zimbabwe, certainly under the current economic and political circumstances.
References
Dury, J., 2007. MSc thesis, Wageningen University, 133 pp.
Giller, K.E., et al., 2006. Agric. Systems 88: 8-27.
Zingore, S., et al., 2007. Agric. Ecosyst. Environ. 119: 112-126.
63
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
PRIMA, Prototypical Policy Impacts on Multi-functional Activities in rural
municipalities1
N. Turpin, R. Laplana, D. Kopeva, L. Stapleton, K. Happe, G. Deffuant,
F. Brouwer, M. Burghard
UMR 1273 Métafort, 24, avenue des Landais, BP 50085,
63172 Aubiere Cedex, France
Contact: nadine.turpin@cemagref.fr
Introduction
While the majority of Europe’s population is congregated in cities, rural-designated areas
have nonetheless recorded positive net migration over the past decade. Rural areas are
becoming increasingly differentiated and gradually losing their agricultural specificity. They
now need to support the coexistence of two logical approaches to occupation of their space:
one based on the supply of agricultural and forestry products, the other on the various
demands from local residents and seasonal tourists. Under these conditions, the role of
farming, forestry and tourism industry is evolving; the focus is no longer simply on supplying
market goods while limiting the impacts of this supply on negative external factors but now
also on participating in land development and meeting the multiple expectations of society.
As these expectations regard public goods, public intervention is often needed to
encourage economic actors to supply such public goods. Evaluation of such policies requires
a targeted and systematic approach, and several research efforts currently develop tools that
enable impact assessment for European land use policies and land management practices at
the national and regional scales. The existing tools do not yet allow gaining information at
lower levels on mechanisms that modify the economic structure of the firms (new entrants,
disappearing of firms, new markets, re-organization in industry, local knowledge/expertise,
local opportunities, local geographical constraints and local environmental values).
PRIMA framework aims to develop a method for scaling down the analysis of policy
impacts on multi-functional land-use and on the economic activities from the EU level to the
local scale. Special attention is paid to the structural effects of the policies and on their impact
on the environment quality in the regions. PRIMA considers policies related to sustainable
rural development such as Structural Funds, Cohesion Fund, Pre-accession funds and
EAFRD.
Methods
The objective of PRIMA project is firstly to design and develop micro-simulation and agentbased models, of such dynamics and of the impact of European structural policies at
municipality level, with the involvement of local stakeholders, and secondly to analyse how
such an approach can improve ex-ante policy impact assessment.
To perform such analysis, we set out to design and implement micro-simulation and agentbased models that rely on behaviour rules drawn up at local level from stakeholders'
consultations for the main actors in rural landscapes (e.g., farmers, forest industries, local
consumers, tourism actors), in a set of municipality case studies, drawn from regional case
studies. When facing new driving forces, these actors may adopt new activities, increase the
size of their enterprise, associate themselves with other actors at the local level, re-organize
their production systems in accordance with the accurate industry, or even disappear if the
local context becomes not favourable. Of course, in the case of highly favourable context,
new entrants are liable to appear.
1
PRIMA is a collaborative project, 7th EU Research Framework, theme 6 Environment.
64
Session A2: Methods for up- and downscaling in integrated assessments
The models address the structural evolution of the populations (appearance, disappearing
and change of agents) depending on the local conditions for applying the structural policies
(sphere of influence for processing industries, possibility and tradition of actors' cooperation
at local level, local development potential for the foreseen activities, specific local markets).
Of course, local stakeholders will contribute to validate, in a post-modelling stage, the
aggregated results of such local models, and if those results differ from the ones expected, the
design rules will be modified. The models are designed to be compatible with more usual
European Policy models (SEAMLESS-IF for the agricultural sector, SIAT tools resulting
from the SENSOR project for the forest and tourism sector, GTAP-IMAGE for global
changes and interactions between land use changes and land-related changes on emissions of
greenhouse gases), when simplified to match the main assumptions of these aggregated
models.
Results and discussion
PRIMA starts in November 2008. The paper presents the first insights of the project, with a
focus on agriculture.
65
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Spatial upscaling of on-farm indicators: Concepts to use a scientific
knowledge base to match policymakers’ requirements
N. Turpin, O. Therond, F. Ewert, I. Bezlepkina, J. Alkan Olsson, C. Bockstaller, S. Janssen
UMR 1273 Métafort, 24, avenue des Landais, 63172 Aubiere Cedex, France
Contact: nadine.turpin@cemagref.fr
Introduction
The SEAMLESS-Integrated Framework (S-IF) aims at assessing ex-ante impacts of policy
options and agro-technical innovations on the sustainability of agricultural systems and on
sustainable development at large (Van Ittersum et al., 2008). This assessment is mainly
performed through the use of modelling chains allowing the quantification of a set of
sustainability indicators at multiple scales. SEAMLESS-IF models have been designed to
simulate behaviour of the key hierarchical agricultural systems (field, farm, region, EU and
world). However, despite the wide range of scales covered by these models there can be gaps
between the scale at which model outputs are available (i.e. the model scale) and the scale at
which policymakers’ demand indicators for decision making (i.e. the decision scale).
Accordingly, to meet expectations of policymakers there is a need for procedures changing
the scale of this information from the model to the decision scale (Dalgaard et al., 2004;
Bierkens et al., 2000). This paper presents aggregation concepts and methods that underpin
the scaling capacities with respect to the indicator quantification in S-IF.
Methods
Temporal and spatial aggregation of models outputs to calculate indicators in SEAMLESS_IF
is realized through the definition of appropriate indicator attributes which are capture in the
indicator ontology (Therond et al., 2009). These attributes specify for each indicator the
spatial and temporal resolutions which refers to (the scale of) model outputs and the extents
referring to the decision scales at which at which indicators are demanded. Once this
information is defined the complementary information necessary to manage the aggregation
from the indicator resolution to the indicator extent is associated (tagged) to the indicator
(aggregation algorithms, aggregation weight…).
Results and discussion
The main characteristics of SEAMLESS-IF are its component structure which allows flexible
linkages of independent model components depending on the problem to be addressed. In
addition, components can be added that allow the scaling of data to compute indicators from
model outputs to scales that are relevant for the decision makers. For the spatial up-scaling of
data at the farming system level, information about the spatial allocation of farm types is
important and considered in the proposed scaling approach (Figure 1). Of course, different
scaling factor are used for different indicators (Figure 2), enabling scaling-up of information
on the basis of farm type area for environmental processes, or on the number of farms per
farm type for economic information, or even more complex information as indexes for
income inequalities. For example, for some environmental issues linked to local impact (water
quality, soil erosion, etc.) calculation of a weighted mean by farm size is not relevant, so that
specific indicators like percentage of area with high nitrate leaching have to be developed.
Approaches and results are presented and discussed for different indicators and conclusions
for future work are derived.
66
Session A2: Methods for up- and downscaling in integrated assessments
Figure 1. SEAMLESS farm typology and spatial upscaling issues.
Figure 2. Procedures to scale information for project concerning detailed regions with a
spatial extent of problem assessment corresponding to the region.
References
Bierkens, M.F.P., et al., 2000, Upscaling and downscaling methods for environmental
research. Kluwer Academic Publishers.
Dalgaard, T., et al., 2003, Ecosystems & Environment 100: 39-51.
Therond, O., et al., 2009. AgSAP Conference 2009, these proceedings.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
67
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Coupling integrated assessment models to market equilibrium models to
analyse economic and agricultural-environment interactions across
different scales
Roberto O. Valdivia, John M. Antle, Jetse J. Stoorvogel
Dept. Agricultural Economics & Economics,
Montana State University, Bozeman, MT 59718, USA
Contact: valdivia@montana.edu
Introduction
One of the aims of agricultural research is to develop sustainable agricultural production
systems that balance the often competing goals of profitability, human and environmental
health, and equity. Researchers have been developing different methods to assess the
sustainability of agricultural production systems. On the one hand, methods have been
developed to integrate biophysical and economic models at a disaggregated level with the
objective of capturing the heterogeneity of the physical environment and economic behavior
of farmers (Just & Antle, 1990; Fleming & Adams, 1997; Stoorvogel et al., 2004). On the
other hand, aggregated models based on the construct of ‘representative agent’ have been
widely used in policy decision making. Market equilibrium (ME) models are a good example
of these models. They have been used to evaluate welfare implications of a particular change
(policy, environmental, etc.) using representative data of producer and consumer behaviors.
This implies that aggregated models do not capture the biophysical and economical
heterogeneity that characterizes production systems. Conversely, results from integrated
assessment models that capture this heterogeneity have not been linked to market equilibrium
models. However, several studies have recognized the need to link these two types of models
as an important step in the assessment of agriculture-environment interactions in order to
understand the processes and relationships across different scales (Kayser, 1999). It is clear
from the literature that here is an evident lack of methods that couple site-specific integrated
assessment models with ME models. This study presents a method for coupling a site-specific
integrated assessment model to a ME model.
Methods
The Trade-off Analysis (Stoorvogel et al., 2004) model is an integrated assessment model that
links site-specific biophysical process models and economic decision models and captures the
spatial heterogeneity that characterizes the production system. Results can be statistically
aggregated to a level that is relevant for policymakers and used for welfare and policy
analysis. Resulting trade-off curves are a set of possible equilibrium points associated with
different prices, therefore trade-off curves can be interpreted as generalized heterogeneitybased supply curves that include both market and non-market effects. The proposed linkage of
heterogeneity-based supply curves to the market conditions and their effect on the underlying
spatial distribution is described in Figure 1. The trade-offs among environmental outcomes, E
(e.g., environment quality) and the economic output, Q (e.g. aggregate output) are represented
by the curve T. That trade-off curve is generated by varying a price P1 (e.g., price of Q or an
input price). Each point along T is a possible equilibrium, and corresponds to a point on the
supply curve S. Adding a demand curve such as D0, we obtain a market equilibrium point (a),
which in turn, defines the point b on the trade-off curve as the equilibrium. This point is
associated with a specific spatial distribution, then we can go from market equilibrium back to
the implied spatial distribution of outcomes (Map0). If demand conditions are changed and a
new market equilibrium point, a’ is attained, then this would imply that a different spatial
1
More generally a vector of prices.
68
Session A2: Methods for up- and downscaling in integrated assessments
distribution of outcomes (Map1) is associated with the corresponding equilibrium point (b’) in
the trade-off curve. The key point of this analysis is that we can use site-specific data to
capture the heterogeneity of the population and generate trade-offs among economic and
environmental indicators, estimate market equilibrium and link these results back to the
underlying spatial distributions and measure their effects.
Map0
spatial distribution
E
Map1
spatial distribution
E1
b
E0
b’
T
Q0
Q
Q1
D0
P0
a
D1
P1
Figure 1. Theoretical framework to
link environmental outcomes, market
equilibrium and underlying spatial
distributions.
a’
P
S
Results and discussion
The goal of the empirical work will be to test the modelling methods and to assess the
conditions under which market equilibrium analysis is important in assessment of agricultureenvironment interactions. We will use the case study of a semi-subsistence agricultural
system (Machakos, Kenya) to test policy intervention and technological change scenarios in
the context of market equilibrium and their effects on the underlying spatial distributions.
References
Just, R. & J. Antle, 1990. Am. Econ. Rev. 80: 197-202.
Fleming, R.A. & R.M. Adams, 1997. J. of Envir. Economics and Management 33: 347-358.
Kayser, H., 1999. Assessing Research on the impacts of Climate Change on Agriculture. In:
Global Environmental Change and Agriculture. 1998. Chapter 7.
Stoorvogel, J., et al., 2004. Agricultural Systems 80(1): 43-66.
69
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing the effects of data aggregation using a crop simulation model
L.G.J. Van Bussel, F. Ewert, P.A. Leffelaar, H. Van Keulen
Plant Production Systems Group, Wageningen University, P.O. Box 430,
6700 AK Wageningen, The Netherlands
Contact: Lenny.vanBussel@wur.nl
Introduction
Crop yields are determined to a large extent by the length and timing of their phenological
phases, that are governed by the interactions of genetic properties and environmental
conditions (ambient temperatures and day length), but also by management, especially sowing
date. Future environmental conditions are anticipated to be characterized by higher
temperatures, leading to modifications in crop phenological development patterns, influencing
yields. Information on phenological development of crops under climate change, may be
derived from phenological models such as AFRCWHEAT2, that simulates wheat development
based on thermal time, photoperiod and vernalization (Harrison et al., 2000).
Policy decisions to mitigate or adapt to climate change often refer to larger scales such as
regions or countries. However, crop models like AFRCWHEAT2 are developed for the smaller
field scale. To study climate change impacts, (results from) field-scale models have to be
scaled-up (Rastetter et al., 1992; Harrison et al., 2000; Harvey, 2000; Ewert, 2004). In scalingup model applications, heterogeneity in input data may be encountered, e.g. in temperature or
sowing dates. Large-scale model applications often ignore this variability and use generalized
data. To generate reliable model outcomes and/or estimate their confidence intervals, it is
important to understand and consider the effect of such data generalization.
Approaches for data generalization usually refer to the aggregation of input or output data
(Harvey, 2000; Ewert, 2004). It is expected that with increasing aggregation extreme values will
be averaged out, with implications for the simulation results. Accordingly, the present paper
investigates the impact of aggregating observed sowing dates and temperature on the simulation
of phenological stages of winter wheat (Triticum aestivum) across Germany.
Methodology
Aggregation of input data
Germany has been divided into grid cells of varying size: starting from 10×10 km to 100×100
km with a 10 km interval. Observed sowing dates from the year 1995 for winter wheat were
averaged for grid cells of each size.
Effects of data aggregation on model outcomes
w
In one region, Niedersachsen, an area of
s
w
w
Aggregate
s
s
Model
100×100 km has been identified with ten
w
output data
s
a
observation points for wheat phenology and
Aggregated
weather. The observed range in sowing dates
w
sowing data
s
w
w
was on average 73 days for the considered
s
s
w
Aggregate
s
Model
output data
years from 1984 until 1988 (including a relative
b
warm and a relative cold year). AFRCWHEAT2
Aggregated
w
weather data
s
was run for four different combinations of
w
w
s
s
w
Aggregate
s
Model
aggregating input (sowing date or weather) and
output data
c
output data (Figure 1), for each year. The
Aggregated
w
s
outputs of the runs according to Figure 1b-d
sowing data
w
w
Model
s
s
w
Aggregated
s
have been compared with the output of run a.
d
weather data
Figure 1. Overview of model runs;
w = weather observation (―), s = sowing date observation (---).
70
Session A2: Methods for up- and downscaling in integrated assessments
Results and discussion
Sowing date
(day of year)
(b)
(a)
No data
251 – 260
261 – 270
271 – 280
281 – 290
291 – 300
301 – 310
311 – 320
321 – 330
331 – 345
(c)
Differences (days) with
non-aggregated data
Figure 2. Effect of data aggregation on spatial distribution of sowing dates for grid cell sizes
of: (a) 10×10 km, (b) 50×50 km, and (c) 100×100 km, for the year 1995.
7
6
5
4
3
2
1
0
-1
-2
-3
7
6
5
4
3
2
1
0
-1
-2
-3
(a)
1984
1985
1986
1987
1988
7
6
5
4
3
2
1
0
-1
-2
-3
(b)
1984
1985
1986
1987
1988
(c)
1984
Figure 3. Differences (in days) between non-aggregated data and:
(a) aggregated sowing dates, (b) aggregated weather data, and
(c) aggregated sowing dates and weather data, for a 100×100 km grid,
for the phases listed in the legend.
1985
1986
1987
1988
Terminal spikelet
First awns visible
Beginning of anthesis
Few days before harvest
An increasing grid cell size results in loss of variability and a change in the spatial pattern of
sowing dates (Figure 2). Moreover, data aggregation appears necessary for complete coverage
of the area of interest: data for all grid cells are available only at grid cell sizes with a
minimum area of 50×50 km, considering all the studied sizes.
Effects of input data aggregation on model outcomes are different for sowing date and
weather data (Figure 3). Aggregation of weather data has the strongest effect, particularly on
simulated harvest date that is 2 to 6 days earlier when aggregated weather data are used
compared to point observations.
As aggregation of sowing dates results in only small deviations, the use of an average sowing
date per region, with a maximum area of 100×100 km, seems justified.
References
Ewert, F., 2004. Modelling changes in global regionalized food production systems under
changing climate and consequences for food security and environment. Report, Plant
Production Systems Group, Wageningen University, 55 pp.
Harrison, P.A., et al., 2000. Agricultural and Forest Meteorology 101: 167-186.
Harvey, L.D.D., 2000. Clim. Change 44: 225-263.
Rastetter, E.B., et al., 1992. Ecological Applications 2: 55-70.
71
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Scale effects in Hortonian surface runoff with management implications
Nick Van de Giesen, Tjeerd-Jan Stomph
Delft University of Technology, Stevinweg 1, 2628 CN Delft, The Netherlands
Contact: n.c.vandegiesen@tudelft.nl
Introduction
When rainfall intensity exceeds the infiltration capacity of a soil, Hortonian surface runoff
will occur. This type of surface runoff can mainly be found in (sub-)tropical climates with
heavy rainstorms. Although during such storms water can be observed on the surface throughout the landscape, not all water makes it to the bottom of the slope. Once the rain stops, much
of the water on the surface will infiltrate before reaching the slope bottom. This causes a
reduction in the fraction of rain that runs off. This fraction is called the runoff coefficient. The
longer the slope, the smaller the runoff coefficient becomes. How strong this effect is depends
on slope angle, rainfall intensity and duration, surface roughness, and infiltration capacity of
the soil. In West Africa, on moderate slopes (2%–4%), strong reductions in runoff coefficients
were found with increasing slope length (Van de Giesen et al., 2000). Laboratory and
simulation studies confirmed the field findings and helped explain the relevance of different
factors (Stomph et al., 2001, 2002). Three different runoff regimes can be distinguished,
based on the specific combination of slope and rainfall properties. Through a simple
analytical model and dimensional analysis, the dominant runoff regime can be determined for
a region. In turn, each regime comes with its own runoff and erosion reducing management
measures such as mulching, hedgerows, and riparian zones. Here, a review of the research is
presented, whereby emphasis is placed on the scaling laws that govern different runoff
regimes and the associated different management options.
Methods
The methods consist of three parts. The first part is field experimentation in Côte d’Ivoire
where runoff was measured coming from plots with two different lengths. The short plots
were 0.8 m wide and 1.25 m long. The long plots were 0.8 m wide and 12.0 m long. Runoff
was collected at the bottom of each plots through gutters and oil drums. Runoff was measured
after each of the 30 rainfall events. Figure 1 shows the experimental lay out. Due to different
fallow treatments and slope positions, there was only one true repetition.
Figure 1. Experimental lay-out.
72
Session A2: Methods for up- and downscaling in integrated assessments
To have better control over the different factors that determine runoff, an artificial slope
was built in the laboratory that allowed for runoff measurements under different slope angles
and lengths. The complete set of results helped to develop a model based on the numerical
integration of the kinematic wave equation, using the method of characteristics. This method
also allowed for analytical solutions for simplified rainfall and infiltration patterns, which
aided the dimensional analysis and regime identification.
Results and discussion
Table 1 summarizes experimental results found for the field experimentation in Côte d'Ivoire.
Even a relatively short slope of twelve meters showed a reduction of over 40% in runoff when
compared with the ‘point’ runoff from the short plots. Clearly, runoff from a slope can not be
calculated by simply multiplying the runoff from a point with slope length. This effect has
been observed in many settings. In the literature, this effect is often ‘blamed’ on the spatial
variability in infiltration characteristics. What remains unclear is why this variability would
always lead to a reduction in runoff. The analysis provided here shows that temporal
dynamics alone can account for the observed scale effects. Only under certain circumstances
does spatial variability play a (random) role.
Table 1. Experimental runoff results for short (1.25 m) and long (12.0 m) plots from the M’bé
experimental farm, Bouaké, Côte d’Ivoire. Tot: total runoff (liters); C1: runoff coefficient
short plot; C10: runoff coefficient long plot.
Short
Long
C10/C1
Repetition
Tot (l)
C1
Tot (l)
C10
Set 1, bare fallow/rice
397
0.37
2156
0.21
0.57
Set 2, bare fallow/rice
495
0.46
2801
0.27
0.59
By defining a set of dimensionless variables, the kinematic wave equation was solved for
different rainfall durations and intensities (Van de Giesen et al., 2004). This analysis showed
when scale effects were significant and when not. When scale effects are extreme, only a
small strip along the bottom of the slope will contribute to runoff. In such cases, it will suffice
to manage this riparian zone. In other case, hedgerows and ridges can be used along the slope.
The optimal spacing will depend on the characteristic lengths resulting from slope and rainfall
properties.
References
Stomph, T.J., et al., 2001. Earth Surface Processes and Landforms 26(6): 647-655.
Stomph, T.J., et al., 2002. Earth Surface Processes and Landforms 27(8): 847-855.
Van de Giesen, N., et al., 2000. Hydrological Processes 14(1): 165-175.
Van de Giesen, N., et al., 2004. Agricultural Water Management 72(2): 109-130.
73
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Effect of spatial resolution within the Cross Compliance Assessment Tool:
The case of Nitrate Directive
J. Van Os, B.S. Elbersen, J. Kros, W. De Vries, J.P. Lesschen, M. Kempen
Alterra, Wageningen UR, Droevendaalsesteeg 3,
6708 PB Wageningen, The Netherlands
Contact: Jaap.vanOs@wur.nl
Introduction
The purpose of this paper is to make a description of the spatial scaling problems encountered
in the CCAT project and explain how these were handled and what effects these may have on
the final results. CCAT stands for Cross Compliance Assessment Tool. A Framework 6
research project (EC 44423-CCAT) aiming at the development of a tool to assess the effects
of Cross Compliance on agricultural markets, environment, landscape, biodiversity animal
welfare and public health.
Cross Compliance was introduced as part of the CAP Reform of 2003 with Regulation
1782/2003. It means that farmers in the EU who apply for the Single Payment Scheme, have
to comply to two sets of standards. The first, called Statutory Management Requirements
(SMRs), relate to 19 pieces of EU environmental, public, animal and plant health and animal
welfare legislation, which is applicable to all farmers. The second set, referred to as standards
of Good Agricultural and Environmental Condition (GAEC), relate to the appropriate
management of soils and the minimum maintenance of agricultural land and features found on
that land.
For the integrated assessment of the effects of Cross Compliance two existing core models
are used: the MITERRA model and the CAPRI model. MITERRA is a steady state
environmental model and CAPRI is a dynamic economic model. Both models are linked
through input-output relations and they are joint under one user interface. CCAT is a steady
state model and not affected by temporal resolution. Conversely, this is not the case for the
spatial resolution. Spatial scale of available input data and model calculations are often
different. This means that scale transitions are needed. In this paper the scale problems of
CCAT models encountered during the implementation of the Nitrate Directive in CCAT are
discussed.
Methods
Two steps are followed to explore the scale transitions encountered in the Nitrate Directive
case study in CCAT. First, we will describe the scale of the different input data and the model
calculations and explain which steps were followed to translate them to the appropriate scale
in order to allow for integration. This is a descriptive approach. Secondly, we will focus on
the results of the CCAT calculations concerning the Nitrate Directive and their sensitivity to
up- and downscaling procedures applied in the post model phase.
We will give special attention to the linkage between the different models and the scale
transitions which are necessary to couple the models. In this paper we will focus on prototype
1 of CCAT. Figure 1 gives an overview of the coupling of CAPRI and MITERRA within
CCAT prototype 1.
The CAPRI model (Britz et al., 2008) is a regionalized economic model for agriculture. It
calculates nitrogen flow through the agricultural system in Europe too, embedded in a
regionalized economic model calculating supply of agricultural commodities in Europe within
a global market and a European policy situation. Within the CAPRI modelling system, a
database has been established, which calculated production statistics (crop acreages, animal
population numbers, etc.) at the regional (NUTS2) level.
74
Session A2: Methods for up- and downscaling in integrated assessments
MITERRA-Europe is a deterministic and static N cycling model which calculated N
emissions on an annual basis, using N emission factors and N leaching fractions (Velthof et
al., 2007). The model can be used to assess the effects of measures and policies on the
emissions of ammonia, nitrous oxide (N2O), N oxides (NOx), and methane (CH4) to the
atmosphere, leaching of N (including nitrate) to ground water and surface waters, and on the
phosphorus (P) balance at EU-27 level, country level, and regional (NUTS-2) level.
Results and discussion
Agricultural data and models are often available for administrative regions, such as
municipalities or provinces. In contrast, ecosystems and landscapes cross administrative
boundaries; the scale of these processes is quite different from agricultural activities.
Coupling of all processes in an integrated assessment means that a lot of scale transitions have
to be performed. In these transitions the essential properties of the data have to be maintained.
The results will show to what extent CCAT succeeds in this. The results can be useful for
other integrated assessment projects where different models are coupled and for the
implementation of prototype 2 of CCAT.
Figure 1. Coupling of MITERRA and CAPRI to assess the effects of the N-Directive.
References
Britz, W., et al., 2008. Description of the CAPRI Modeling System, Specific targeted research
project, Ex-ante policy assessment of CAP. Contract nr 501981, University of Bonn,
Germany.
Velthof, G., et al., 2007. Development and application of the Integrated model MITERRAEurope. Service contract European Commission “Integrated measures in agriculture to
reduce ammonia emissions” Contract number 070501/2005/422822/MAR/C1 Task 1.
75
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Risk assessment methodology for pesticide control
in the Nil River Basin, Belgium
S. Zakayo
Department of Hydroinformatics and Knowledge Management,
UNESCO-IHE, P.O. Box 3015,
2601 DA Delft, The Netherlands
Contact: Zakay1@unesco-ihe.org
Introduction
Van Leeuwen & Hermans (1995) defined risk assessment as the process of estimating the
likelihood that a particular event will occur under a given set of circumstances. Hence, risk
assessment for pesticides involves an analysis of information on the environmental fate and
behaviour of chemicals in the environment (i.e. air, water and land) integrated with an
analysis of information on their effects on human beings and ecological systems. Kannan et
al. (2007) indicated that actual risk will vary temporarily and spatially with site-specific
factors such as point-source loads, environmental conditions (e.g., river flows, soil drainage
rates, soil properties) and loss processes. These issues seem to be heavily scale dependent,
e.g., small areas may be very sensitive to direct losses while applying the pesticides, whereas
larger catchments are more affected by the accumulated effect of pesticides in the runoff
water (Holvoet, 2006). Since these factors appear to be difficult to include into generic risk
assessment methodologies, Kannan et al. (2007) suggested that they can be considered in
more sophisticated models like GREAT-ER (Geography-referenced Regional Exposure
Assessment Tool for European Rivers) to predict point-source chemical exposure in rivers
and large number of diffuse-source contaminant transport models.
Methods
In order to achieve the above, modelling of pesticides in question (i.e. atrazine and
isoproturon) serves as a valuable tool in understanding surface water contaminant caused by
pesticides used in agricultural watersheds. Kannan et al. (2006) referred to a few but
increasing number of studies using the Soil and Water Assessment Tool (Arnold et al., 2002)
to model watershed-based pesticide processes. Holvoet (2006) mentioned that an inventory
study on pesticides was conducted in Belgian catchments and the most 20 pesticides were
reported in VMM (2005). Further, detailed information on pesticides application in the Nil
catchment from 1998-2002 is available, which makes research on transport and fate of
pesticides in the area attracting. Holvoet (2006) conducted a study in monitoring and
modelling the dynamic fate and behaviour of pesticides in the Nil catchment.
Despite the complexity in pesticides models, knowledge on transport processes of
pesticides is still incomplete. This study focuses on modelling transport of pesticides and
formulates improved equations in SWAT for pesticides in question (i.e. atrazine and
isoproturon) to represent the dynamic nature of these pesticides characterized by a fast
transport by surface water. A Pareto optimization tool is used to optimize SWAT for better
hydrology and pesticides results. In addition, time series of simulated pesticide concentrations
provided by SWAT are linked with GREAT-ER for providing risk assessment in the Nil
catchment.
Results and discussion
The model results review a diversity in causes of pesticide pollution that relate to the
application practices (Figure 1).
76
30
0
15
25
0
10/04/1998
Total losses
Rainfall (mm)
Application dose as % of total dose per year
Dissolved atrazine load in reach 25 (g)
Session A2: Methods for up- and downscaling in integrated assessments
50
5/05/1998
Runoff losses
Point losses
30/05/1998
Drift losses
Application dose
Rainfall
Figure 1. Predicted load of dissolved atrazine coming from sub-basin 25 during spring 1998,
together with the measured rainfall and the initial pesticide dose (showed as hanging bars on
the secondary axis).
A risk assessment for atrazine and isoproturon enables a more informed environmental impact
assessment whereby information is provided to risk managers and decision makers in a
manner that is understandable and relevant to the decisions being made.
References
Arnold, J.G., et al., 2002, Soil and water assessment tool, theoretical documentation,
Grassland, soil and water research laboratory - Agricultural Research Service, Texas, US.
Holvoet, K., 2006. Monitoring and modelling the dynamic fate behaviour of pesticides in
river Systems at Catchment Scale. PhD thesis, Ghent University, Ghent, Belgium.
Kannan, N., et al., 2006. Environmental Science and Health, Part B 41: 1049-1070.
Kannan, N., et al., 2007. Chemosphere 66: 1336-1345.
Van Leeuwen, C.J. & J. Hermans, 1995. Risk Assessment of Chemicals: An introduction.
Kluwer Academic Publishers, Dordrecht.
VMM, Flemish Environment Agency, 2005. Water- en waterbodemkwaliteit; Lozingen in het
water. Evaluatie saneringsinfrastructuur 2004, 88 pp. (in dutch) http://www.vmm.be
77
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A generic downscaling approach for scenario driven modelling outputs
from European to regional scale
Ingo Zasada, Annette Piorr, Regine Berges
Institute of Land Use Systems and Landscape Ecology, ZALF,
Eberswalder Str. 84, 15374 Muencheberg, Germany
Contact: ingo.zasada@zalf.de
Introduction
Scenarios represent powerful tools and foundations for impact assessments and decisionmaking support at the European level. Rather than predicting and forecasting, the scenario
framework of the FP6 Integrated Project PLUREL refers to possible futures, exploring
development trajectories of driver-pressure-relationships in rural-urban-regions in Europe
(RUR) (Ravetz et al., 2008). In the four different scenarios analysed, the narrative storylines
consider assumptions on drivers related to demography, economy, climate and technology.
These storylines are translated into quantitative socio-economic variables of the econometric
model NEMESIS to calculate (urban) land-use change at NUTS 0 level (member state).
Following the operational approach of the project, focussing on the issue of regional urbanrural-relationships, the responses of different landscape functions on urban sprawl or decline
are assessed. Regression modelling for the impact assessment of land-use changes requires
the lower spatial scale of NUTS 2/3. In addition to the econometric model, locational
characteristics distinguishing different regions within Europe, which are expressed in
determinants influencing (urban) land-use change, need to be taken into consideration. A
downscaling model has been developed which is based on generic and not spatially explicit
approaches. It allows for the adaption of basic response functions to limit the number of
different regression algorithms.
Regional determinants and tacit knowledge
A lot of theoretical and empirical research in the field of regional science is carried out on the
role of regional characteristics as push and pull factors for land use and land use changes.
Especially questions of accessibility (Krugman, 1993) innovative capacity (Acs, 2002),
environmental quality as well as natural and technological hazards and vulnerability
(UN/ISDR, 2004) and the planning and governance regime represent important determinants.
Due to spatial observation and monitoring efforts, e.g., by ESPON, EIS or EUROSTAT,
European-wide typologies and benchmarking assessments of these regional determinants are
available. Even though strong relationships between locational qualities and urban
development are observed, the actual urban growth response is hardly empirically
ascertainable and predictable. Instead, there is broad and resilient knowledge distributed
among researchers, practitioners and policymakers on various spatial levels and fields of
expertise, which is fuzzy structured, non-codified, undirected and hardly quantifiable – socalled tacit knowledge (Polanyi, 1966). On the one hand knowledge on land-use dynamics
evolves throughout the scientific community through research engagement. On the other hand
regional representatives and stakeholder on the local level possess endogenous knowledge on
the basis of regional experience.
Integration of participatory methods in land-use modelling
The newly developed generic regionalization and downscaling procedure consists of various
single modules integrating (i) the high aggregated land-use information (NUTS 0), (ii)
empiric information about regional determinants on NUTS 2 level, (iii) generic land-use
relevance of the particular determinant as well as (iv) the scenario sensitivity of each
determinant. For the estimation of the land-use impact relevance and scenario sensitivity of
78
Session A2: Methods for up- and downscaling in integrated assessments
regional determinants, participatory approaches (expert interview, Delphi process) (Linstone
& Turoff, 2002; Flick, 2005) are applied.
Results and discussion
In a first step an expert judgement among researchers on the relative relevance from a list of
regional determinants has been conducted, which is used for a regional validation. In parallel,
a Delphi process with scenario modellers regarding the scenario sensitivity of the
determinants is carried out and agreed. The results are implemented for the downscaling
procedure to obtain regionally differentiated European maps on NUTS 2 level. Additionally,
single disaggregated determinant values serve as explanatory factors in the regression
modelling approach for the impact assessment (see Figure 1).
Figure 1. Regionalization of scenario-based urban land use change.
References
Acs, Z.J., 2002. Innovation and the growth of cities. Edward Elgar Publishing, Northampton.
Flick, U., 2005. Qualitative Sozialforschung. Eine Einführung. 4 ed. Rowolt Tb.
Krugman, P.R., 1993. Review of International Economics 1: 110-122.
Linstone, H.A. & M. Turoff, 2002. The Delphi Method: 1. Techniques and Applications.
Polanyi, M., 1966. The Tacit Dimension. Doubleday & Co.
Ravetz, J., et al., 2008. Scenario Framework, Report D1.3.2 for Peri-urban land use
relationships (PLUREL). EU 6th framework.
UN/ISDR, 2004. Living with Risk. http://www.unis.unvienna.org/unis/pressrels/2004/
unisinf23.html
79
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme A
Methodology for integrated assessment
Session A3: Integrated assessment of ecosystem services
from agriculture
Session organizers:
Floor Brouwer
LEI, Wageningen UR, The Netherlands
George Hutchinson
Queen’s University Belfast, Northern Ireland
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Environmental services delivered and deliverable by land managers
Allan Buckwell
Country Land and Business Association,
16 Belgrave Square, London SW1X 8PQ, UK
Contact: allan.buckwell@cla.org.uk
Introduction
In the last decade there has been a widening appreciation of the breadth, components and
scale of market failures surrounding the major land management activities of farming,
forestry and certain recreational activities. This has been accompanied by an
acknowledgement that these market failures open a space for public policy action (Braat &
Ten Brink, 2008). However, there are at least four dimensions of difficulty in turning this
appreciation and acknowledgement into operational policy frameworks in the EU. These are,
the motives and objectives of land managers as well as the agricultural policy framework
within which they operate; the different levels of economic development and attitudes to
environment across the EU; the complexity caused by the interaction of positive and negative
externalities; and the lack of agreed conceptual framework for dealing with these issues. The
proposition underlying the paper is that there is a wide gulf between the rhetoric of EU
agricultural policy reform, viz. to move from Pillar 1 to Pillar 2, and the reality.
Methods
This paper will proceed by observation and argumentation rather than following formal
economic or other analytical methods. It will tease out and justify the four difficulties listed.
Private sector farmers, foresters and game managers see themselves predominantly
concerned with a market-based economic activity. A minority, but growing proportion of
them, have appreciated the non-market aspects of their primary activity and in varying ways
have found ways to integrate this in what and how they produce. Meanwhile policy over
many decades has overtly reinforced farmers’ view in focusing single-mindedly on their
primary role. This adds to the scale of the task of reversing policy from focusing on the
market to the non-market outputs.
A critical element of the EU policy mix in finding a way through this policy maze is the
very different historical, political and economic backgrounds of the Member States, and the
scale of the environmental challenges they face. Some summary indicators of the
development status and the environmental indicators of the Member States will be examined
to see if there are predictable clusters of potential attitudes towards moving policy focus from
market to non-market objectives.
In addition to these political-economy considerations, there are genuinely difficult
technical and biological aspects of the environmental services and disservices rendered by
land managers. Farmers and foresters as well as creating and maintaining semi-natural
habitats, ecosystems and cultural landscapes which are now treasured, also degrade the
environment, reducing biodiversity, pollution soil, water and atmosphere. The production
functions of these environmental goods and bads are complex, interactive and dynamic.
Drawing the line between good and bad and devising the most cost effective ways of
incentivizing the good and disincentivizing the bads turns out to be extraordinarily difficult. It
is not clear which is the greater problem conceptual or empirical?
What is the most fruitful framework for pulling these considerations together? The implicit
conceptual model which underpins current EU agricultural/agri-environmental/rural policy is
a (now largely discredited) market model to which some environmental bells and whistles has
been bolted. What would be a better model? Two will be examined: the environmental
accounts approach (Defra, 2008); and the ecosystem services approach (Constanza et al.,
82
Session A3: Integrated assessment of ecosystem services from agriculture
1997; Defra, 2007; EC, 2008). The hope is that this presentation will stimulate suggestions of
other or better approaches.
Discussion
This is not the kind of paper from which to expect concrete research results. The discussion
will put these ideas into the context of the real political choices the EU faces in the coming
two years. The Union is currently drifting. It has failed to agree a Constitution, its latest
Treaty is in abeyance; and it has agreed to review in 2009 all its major policies and the size
and purpose of its budget, yet this review seems not to have stimulated a genuine debate about
the purpose and scope of EU policy. The to the extent that ‘the protection and management of
natural resources’ is thought by academics, by environmental NGOs and by land managers to
be an important part of EU activity, then producing research findings to reflect on the scale
and nature of the policies to do this is a worthwhile activity.
References
Braat, L. & P. Ten Brink, 2008. The cost of policy inaction: The case of not meeting the 2010
biodiversity target. European Commission DG Environment study OJ Ref 2007/S95 –
116033, Wageningen/Brussels.
Constanza, R., et al., 1997. Nature 387 (15 May 1997): 253-260.
Defra, 2007. An Introductory guide to valuing ecosystem services. London.
Defra, 2008. Environmental Accounts for Agriculture, Final Report, conducted by Jacobs,
with Scottish Agricultural College and Cranfield University, SFS0601, London.
EC, 2008. The economics of ecosystems and biodiversity. An interim report, compiled by
Pavan Sukhdev et al., European Communities, Brussels.
83
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Restoration of abandoned agricultural lands toward
umbrella species habitats1
M. Arriaza, O. Nekhay, M.J. López-Baldovín
Centro IFAPA ‘Alameda del Obispo’, Apartado 3092, Córdoba, Spain
Contact: manuel.arriaza@juntadeandalucia.es
Introduction
The present study analyses the suitability of agricultural lands with risk of abandonment to be
restored to suitable habitats for endangered species. As a case study, we focus on the olive
plantations of mountain areas of Southern Spain and the Iberian lynx. The methodology
weighs experts’ judgements about the effects of the landscape elements on this habitat via the
Analytic Hierarchy Process (AHP) and spots the most suitable areas for restoration through
the Geographical Information Systems (GIS).
The main interest of the study relies on the potentiality of the methodology to combine
complex territorial analysis with the biological requirements of endangered species to
facilitate their dispersal.
Methods
The methodology involves three phases: First, an inventory of Iberian lynx habitat
requirements is drawn up. Then, the AHP method is implemented based on the experts’
knowledge. Finally, the Geographical Information Systems technology is used to assess the
potential of the study area for Iberian lynx’s habitat restoration.
Selection of landscape elements related to Iberian lynx habitat requirements
The main causes that have brought the Iberian lynx to the border of extinction are habitat
alterations and removal, the fall in the number of rabbits (the main lynx’s prey), human
activity, such as illegal hunting and traps (Gaona et al., 1998; Delibes et al., 2000).
Analytic Hierarchy Process multi-criteria decision-making technique
There are two specific characteristics that distinguish this method (Saaty, 1980) from the
other multi-criteria methods of this family: (a) the construction of the hierarchy structure of
the problem to be solved, and (b) the pair-wise comparisons made between different criteria
to weigh them with respect to the overall objective.
GIS-aided analysis
The GIS software used as a platform for the representation, management and analysis of the
spatial information was ArcGIS 9.1 and ILWIS 3.4 (Integrated Land and Water Information
System). The operations of spatial multi-criteria evaluation was carried out in ILWIS 3.4.
SMCE module. The input data were: land use map (1999; 1:50,000); aerial monochrome
orthophotos (2001–2002; 1:5000) and colour orthophotos (2005; 1:10,000); olive plantations
productivity maps (2004; 1:25,000); road infrastructure map (1999; 1:25,000).
Results and discussion
According to the results, the most important landscape objects are the natural vegetation
structures (32%), followed by the proximity to asphalted roads (28%), the proximity to the
Natural Park (18%) and the proximity to watercourses (14%). Olive groves with vegetation
cover obtained the lowest weight (8%).
1
This research has been financed by INIA through the research project RTA04-086.
84
Session A3: Integrated assessment of ecosystem services from agriculture
Figure 1. Evaluation of Iberian lynx
habitat restoration potential.
The red spot in the centre represents the urban area and it is regarded as a constraint (noncompensatory criteria). The red lines that pass through the map represent two motorways that
have high levels of traffic. The maximum suitability value recorded in the study area was
0.92, the minimum was 0 and mean value was 0.46 (in a 0–1 scale). The green colour
represents the areas suitable to implement the restoration measures. All situated on the North
of the motorway A-IV. Since the divergences in the opinions of the experts raise a good deal
of uncertainty about the reliability of the result, we decided to perform a sensitivity analysis
of the results.
It is interesting to report some similarities between the results obtained in this study and those
provided by Van der Horst & Gimona (2005), which used multi-criteria spatial analysis to
determine the most suitable territories in agricultural areas for the implementation of action
plans to promote biodiversity. Unlike the present study, these authors combine the
requirements of 15 species as map layers, weighted according to the importance of each
species. However, the results of both studies emphasize the importance of the edge zones of
major agricultural areas, the riparian zones (in our case the natural vegetation) and areas
adjacent to nature pinewoods (in our case Mediterranean forest and shrub lands) as having the
highest potential for biodiversity.
References
Delibes, M., et al., 2000. Nature Environ. 111: 1-45.
Gaona P., et al., 1998. Ecol. Monog. 68: 349-370.
Saaty, T.L., 1980. The analytic hierarchy process: Planning, priority setting, resource
allocation. McGraw-Hill, NY.
Van der Horst, D. & A. Gimona, 2005. Biol. Conserv. 123: 421-432.
85
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Examining the degree of jointness between food and cultural landscape
production
David Blandford, Ivar Gaasland, Erling Vårdal
The Pennsylvania State University,
University Park, PA, USA
Contact: dblandford@psu.edu
Introduction
The main purpose of agriculture is to produce private goods. However, it is well recognized
that agricultural activity renders public goods (cultural landscape) as a by-product, i.e. these
goods are produced jointly (Boisvert, 2001). The degree of jointness is an open question.
Some have assumed that jointness takes the form of a one to one relationship, e.g.,
Prestegaard (2004), others that supply can be completely decoupled. In between these
extremes, different degrees of jointness seem probable, i.e., there should be some possibility
for choosing production techniques that are more geared towards the supply of cultural
landscape than others.
Methods
In this paper, the issue of jointness between food production and cultural landscape is
examined. For the sake of clarity, we set up a model with two representative agricultural
sectors, one for pure plant production (e.g., grain) and one based on animals (e.g., milk
production). Each production sector has a corresponding function for the supply of cultural
landscape. Both the private good and the corresponding cultural landscape is a function of
value added to land, but factor shares in the value added aggregate differ substantially. For
example, a high capital-labor share yields high production levels, but a low cultural landscape
value.
On the demand side, we assume that cultural landscape is a multi-dimensional good,
represented by a utility function where cultural landscape from the two agricultural sectors
enters as arguments. Preferences for variety and the assumption of decreasing marginal
willingness to pay for each landscape type are reflected in the substitution elasticity attached
to this function. Two regions are also included in the model (rural and suburban), allowing for
regional differences in the valuation of landscape. Also, the endowment of land and
productivity for the two agricultural products varies between the two regions.
The model is of a partial equilibrium type, where economic surplus is maximized. On the
supply (of private commodities) side, the two production processes in each region cover a
substantial part of the activity in the Norwegian agriculture sector. The assumption of two
regions is also realistic, since the current agricultural production is divided (appr.) evenly
between suburban and urban areas.
The model is implemented using actual numbers from the Norwegian economy.
Information on the various production processes are taken from official sources, see NILF
(2008). The demand functions of the private goods are based on available demand studies. As
for willingness to pay estimates, they will be based on estimates found in the literature, see
Drake (1992) and Lopez (1994).
Results
The model is used to demonstrate the importance of technology when assessing a shift in
agricultural policy from price support to targeted cultural landscape payments, taking into
account the interplay between supply and demand for different attributes of the cultural
landscape. The focus will be on the consequences of such a policy, i.e. the (expected) decline
in agricultural production in the two regions. Furthermore we will focus on the expected shift
86
Session A3: Integrated assessment of ecosystem services from agriculture
in the use of inputs, from capital intensive to land/labour intensive production. The focus will
also be on where the cultural landscape is upheld, i.e. in the rural versus suburban areas.
References
Boisvert, R., 2001. Joint production in four outputs: Two commodities and positive and
negative externalities. In: Multifunctionality: Toward an analytical framework. Annex 2,
OECD, Paris.
Drake, L., 1992. European Review of Agricultural Economics 19(3): 351-364.
Lopez, R.A., et al., 1994. Land Economics 70(1): 53-62.
NILF (Norwegian Agricultural Research Institute), 2008. Budsjettnemda for jordbruket:
Jordbrukets totalregnskap, 1959-2007. Oslo.
Prestegaard, S.S., 2004. Acta Agriculturae Scandinavica, Section C, Food Economics 1: 151162.
87
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The monetary value of combining ecosystem-services in private artificial
wetlands1
Arianne De Blaeij, Vincent Linderhof, Nico Polman
LEI-WUR, Public Issues Division, P.O. Box 29703,
2502 LS The Hague, The Netherlands
Contact: Arianne.deBlaeij@wur.nl
Introduction
Multi-functional land use and the creation of multiple values in rural areas are creating new
challenges and posing new questions (see also Jongeneel et al., 2008). An artificial wetland is
an example of multi-functional land-use, as it combines at least five different ecosystem
services. The main two reasons for investigating options for creating artificial wetlands in the
Netherlands are the water treatment function of surface water in the form of a reed filter, and
the accompanying biomass production. Additional functions are water storage, the
improvement of biodiversity in the surrounding area by solving the dry-out problems and
recreation2.
Currently, most of the functions supplied by an artificial wetland are public goods without
any market value. Therefore, managing privately owned wetland areas in a multi-functional
way raises the questions whether the public see a role for farmers (landowners) to provide
multi-functional wetland. More in detail, how society values the benefits of these wetlands. In
this paper, we analyse how citizens of the Netherlands value private artificial wetlands. In
particular, we estimate the monetary value of the complete concept of artificial wetlands with
the Contingent Valuation Method (CVM). Additionally, we value the different functions
separately with the Analytical Hierarchical Process method (AHP).
A private multi-functional artificial wetland
Multi-functionality in artificial wetland management is directly, although not exclusively,
linked to the different functions wetlands can fulfil. Wetlands are optimally allocated if they
fulfil the mixture of functions demanded by society. These demands are not constant, but are
influenced by many factors, including changes in income level and population density,
productivity-induced changes in relative price, etc (see also Jongeneel et al., 2008).
According to the Water Frameworks Directive (WFD), the government is responsible for
achieving a good quality status of surface water. Wetlands, whether or not privately owned,
provide next to better water quality, extra water-services to society. The social demand for
private wetlands depends on alternative options to reach objectives derived from national and
international water policies.
Next to this, other policy domains can play a role as well, an example is the aim to realize
lively rural areas. According to European law, it is only allowed to pay for non-statutory
services. Whether a farmer or landowner invests in an artificial wetland depends on the
possibilities of receiving payments for the services delivered. Paying for the non-market
services is an extra source of income next to the future income earned with biomass
production (reed is an agricultural crop).
1
The valuation exercise presented here is part of a technical and economic research project called Waterpark
‘het Lankheet’. The technical research is carried out in wetland’ het Lankheet’ which is located in the eastern
part of The Netherlands. (Meerburg et al., 2008).
2
As ecosystem services are defined as “the capacity of natural processes and components to provide goods and
services that satisfy human needs” (De Groot et al., 2002), recreation can be seen as an ecosystem function as
well.
88
Session A3: Integrated assessment of ecosystem services from agriculture
Valuation method
To determine the value of private artificial wetlands in The Netherlands, we use the
Contingent Valuation Method. Separately valuing the wetland-functions through individual
CV exercises could lead to seriously biased estimations, due to series of biases. With the
Analytical Hierarchical Process (AHP) we are able to decompose the aggregated value of the
artificial wetland into values of its different functions. The AHP method uses a series of pair
wise comparisons between the different the artificial wetland functions to assess the relative
importance of each criterion. We will elicit weights for each of the functions. The cognitive
burden of respondents are reduced because AHP always uses two clear functions comparisons
(Kallas et al., 2007; Moran et al., 2007).
Next to socio-demographic variables, there will be taken into account a number of aspects
such as individuals’ perceptions with respect to the natural environment, the fact that
individuals recreate in agricultural areas, and the fact that individuals will recreate in private
artificial wetlands.
Results and discussion
In The Netherlands, the administrative responsibility of the wetland functions is distributed
over different governmental levels. As an example; the local and regional government
(provinces and municipalities) are responsible for nature-conservation, the national
government for biomass-energy objectives and water boards for water quality (Water
Framework Directive1) and water quantity issues. For policy design it is relevant to know how
citizens value artificial wetlands and its different functions in order to be able to extend or
stimulate the implementation of artificial wetlands in The Netherlands.
References
De Groot, R.S., et al., 2002. Ecological Economics 41: 393-408.
Jongeneel, R.A., et al., 2008. Land Use Policy 25: 81-94.
Kallas, Z., et al., 2007. Journal of Agricultural Economics 58(2): 218-241.
Meerburg, B.G., et al., 2008. Surface water sanitation and biomass production in a
constructed wetland: Pilot Lankheet. Submitted to Ecological Engineering.
Moran, D., et al., 2007. Ecological Economics 63: 42-53.
1
Water Framework Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000
establishing a framework for Community action in the field of water policy.
89
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Reconfiguration of ecosystem services in agricultural landscapes:
Mitigation and policy concerns for renewable energy development
S.S. Dhillion
ENVIRO-DEV, nordstu olsberg, 2510 Tylldalen, Norway
Contact: dhillion@online.no
Introduction
The interactions of agriculture and the environment can be woven in the concept of
agriculture as a sustaining, producing and consuming ecosystem services. Agricultural
landscapes and agriculture provide ecosystem services in the form of food and non-food
services and public goods. In developed nations the main non-food services include
maintaining a healthy environment (air and water), a buffer for climate irregularities, food
security and livelihoods. In developing nations the definition of an agricultural landscape has
many dimensions ranging from large areas dominated by agriculture to those which have
agriculture intimately defused with semi-forest like areas. The latter make up a common
feature in many remote agricultural rural areas of the developing world. Ecosystem services
from these agricultural and agro-forest landscapes bring additional services to light, e.g.,
health security, cultural refuges and cosmological arenas, and a continued supply of food
propagules and land, and more recently encompass carbon sequestration. Development
processes (through international or national projects/aid) continuously bring new dimensions
to these rural areas – and some can be said to increase the number of facets in agricultural
practices and thus even the multi-functionality of land aimed at poverty alleviation. Often the
changes which occur with rural development are gradual and imbibed by rural communities
for better or worse. However changes with development which make create large changes are
those directly linked with the renewable energy project, e.g., the construction of hydropower
dams. This paper is on understanding how agriculture landscapes and associated services can
alter with hydropower projects and how mitigation proposals may be facilitated or hindered
by national/international regulations and conventions. A central focus in on the
reconfiguration of services and goods in the altered and new agricultural landscapes
associated with hydropower projects. Both national and transboundary case studies will be
presented.
Methods
Literature reviews, reports and interviews were used to study the issues mentioned. The
author has been personally involved in environmental and social assessments of hydropower
projects in Vietnam, Cambodia and Chile and, thus, draws on first hand knowledge (see
references).
Results
In Vietnam and Cambodia the hydropower projects and planning spans over two large rivers
(the Se San and Srepok) which are the lifeline of local communities. Land use is diverse and
traditional practices prevail. Here the hydropower projects alter significantly the agricultural
landscape and thus the ecosystem services. Hydropower operators, the governments of
Vietnam and Cambodia have come up with various mitigation and development options in the
past and at present. Legislation is not in place to link hydropower development with land use
and thus changes in agricultural landscapes. The poorest of the downstream inhabitants loose
traditional forms of river bank agriculture and divert to deforestation for agricultural
activities.
In similar lines in the case of Chile where cascades of hydropower projects are being
planned in low populated, grazing dominated agriculture and natural areas ecosystem services
90
Session A3: Integrated assessment of ecosystem services from agriculture
in relations to agriculture take a different form, with extensive agriculture is the only form.
The reconfiguration of services and goods render changes in livelihoods. In the case of the
Philippines the mitigation and reorganization of agriculture has had positive response from a
variety of sectors and a wide range of stakeholders. The responses are in part due to the
increase in the multi-functionality of land use, and increase in services.
Conclusions
In all the above cases policy and the lack of regulations are a deciding factor for the final
reconfiguration of the agriculture. Poverty alleviation or livelihood restoration may occur for
the key affected population but at large the long term consequences are turbid at best. How
ecosystems are maintained, altered and reconfigured in light of existing policy will be
discussed.
References
Integrated Impact Assessment of the Aysen Hydropower Project. Chile. Draft 2008. By
SWECO Norway
Rapid Environmental Impact Assessment on the Cambodian part of the Srepok River due to
Hydropower Development in Vietnam. 2007. Submitted by SWECO Grøner, ENVIRODEV, NIVA.
Rapid Environmental Impact Assessment on the Cambodian part of the Se San River due to
Hydropower Development in Vietnam. 2007. Submitted by SWECO Grøner, ENVIRODEV, NIVA.
Summary Environmental Impact Assessment (SEIA) – Main Report. July 2006. Song Bung 4
Hydropower Project, TA No. 4625-VIE, Vietnam. (Final GoV’s Report). www.adb.org
91
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
‘Green services’:
Bridging the gap between consumers and producers of landscape in
The Netherlands?
G.H.P. Dirkx
Wageningen-UR, PO-Box 47, 6700 AA Wageningen, The Netherlands
Contact: joep.dirkx@wur.nl
Introduction
The Dutch landscape is in essence an agricultural landscape. It has developed as the result of
agricultural land use systems that were in practice until the 19th century. These landscapes are
nowadays highly valued by the general public. With that, the meaning of the landscape has
undergone a shift from a mono-functional agricultural territory to a multi-functional
environment that is considered to provide a wide scope of ‘products’, varying from food to
attractive landscapes.
Agriculture has, however, changed. Attractive landscapes are not any longer the obvious
non-commodity output of agriculture. Modern agriculture may even be threatening valued
landscape features, such as field patterns or wooded banks, as they are impeding the
modernization of agriculture. Farmers are nevertheless still the producers of landscape
quality. The societal demand for attractive landscapes is not necessarily in their interest, as
there is no market for attractive landscapes. Farmers that are investing in attractive
landscapes, do not get a better price for milk or other farm products.
With the concept of green services the Dutch government aims to provide an alternative
for a market for attractive landscapes. Green services are services provided by farmers that go
beyond normal farm practice. Maintenance of landscape elements is an example. With the EU
approval, in February 2007, of a Catalogue describing services to be delivered and the height
of the allowances for these services, payment for green services has been brought in
accordance to the EU State Aid regulations. With this catalogue being approved, the concept
of green services can be brought into practice.
The paper analyses the first practical experiences with ‘green services’ in order to explore
the possibilities for involving agriculture in providing ecosystem services. Some policy
options will be given.
Methods
The first step in the analyses was to obtain an overview of the degree to which ‘green
services’ are actually being supplied. Information was provided by the Dutch counties. The
next step was to interview stakeholders involved in setting up green services. They have been
asked about the pro and cons of working with the Catalogue, about the interest in supplying
green services, the kind of green services being supplied and the height of the allowances.
Stakeholders have also been asked about the willingness to reserve public or private funds to
be able to pay farmers for the services provided.
Results and discussion
Although there is a lot of interest in the concept of green services, green services are hardly
being brought in practice. Only two of the 12 Dutch counties have decided on policy
instruments on green services. In one county policy instruments are being prepared. In these
three counties funds have been reserved for greens services. In the other nine counties there is
in general no readiness to reserve funds.
The counties that have made funds available, require 50% co-financing when contracts for
green services are being agreed upon. In most cases municipalities are the parties that are
providing the required co-financing. Therefore, within the counties with a green services
92
Session A3: Integrated assessment of ecosystem services from agriculture
policy instrument, only in municipalities that are prepared to co-finance green services, green
services can be supplied. In general there are no private parties willing to co-finance green
services. The possibilities to have investments in landscape quality financed by private
parties, turn out to be limited.
In the two counties where there are regulations, farmers are willing to supply green
services. The allowances are considered to be reasonable. The supply of green services turns
out to be exceeding the possibilities within the available funds. Making more funds available,
might help to increase the amount of ecosystems services supplied by agriculture.
Green services are meant to provide a supplement to the environmental stewardship
schemes that are only available in designated areas that are considered to be of national
importance. Green services enable farmers to supply ecosystem services in the other parts of
the country, the often called white areas. However, the reticence of counties and
municipalities to fund green services, causes the largest part of the Netherlands to remain a
‘white’ area in terms of ecosystem services.
There are fundamental differences in the way demand and supply are being brought
together. In one county, contracts are based on a landscape plan describing the kind of
services that are required in an area. Contracts are being composed out of fixed packets of
green services with fixed allowances. In the other county people in the area negotiate with
farmers about the kind of services to be delivered by the farmer and the allowance he will
receive for these services. In the latter county, demand and supply are indeed ruling the
agreements on green services.
93
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Quantification of the relationship between changes in farming and changes
in biological diversity: The Atlantic case
B.S. Elbersen, R.G.H. Bunce, A. Van Doorn
Alterra, Wageningen-UR,
P.O. Box, 6700 AA Wageningen, The Netherlands
Contact: Berien.elbersen@wur.nl
Introduction
The first stage of the paper is allocated to a providing the historical background to changes in
agricultural practices in Atlantic Europe and their links with biodiversity. It is widely
recognized that the intensification of agriculture due to support for increased production has
led to a decline in biodiversity and this trend has been quantified in the UK Countryside
Survey (CS). Relevant results from this survey are then presented. Elsewhere in Europe
disparate studies have shown the same pattern which is linked to globalization and technology
transfer (Petit et al., 2001). The driving forces and processes behind such changes are then
summarized.
Data from a stratified random series of sites from southern Britain are then used to
establish relationships between changes in agricultural practices and changes in biodiversity.
Whilst the major losses of biodiversity have been on agricultural land there have been parallel
losses because of declining management both on unfarmed features on farms but also in
woodlands and roadsides.
Finally the review is used to develop a model of future likely trends in biodiversity linked
to changes in farming based on the analyses of past patterns.
Methods
For this study we use the field survey data from a selection of agricultural CS plots (which are
allocated within the selected 1 km squares) collected in three survey years (1978, 1984,
1990). With this information the relationship between changes in farming and biodiversity
can then be further underpinned and it can be determined whether these relationships can be
quantified and made specific for certain types of farming. The identified relationships derived
from the CS results are then translated in general relationships which can be used for future
predictions in biodiversity change in response to changes in farming, modelled in the bioeconomic farm model FSSIM, part of SEAMLESS (Van Ittersum et al., 2008). For this study
these general relationships can only be applied to the Atlantic environmental zone as this is
well covered with the CS data.
Results and discussion
As widely reported in CS, biodiversity in intensively farmed lowland landscapes is now
largely restricted to linear features and small patches. The decline in biodiversity is greatest in
the farmed fields but is also taking place in the unfarmed categories partly because of the
indirect effect of eutrophication from agricultural practices but also because of independent
ecological processes such as canopy closure in woodlands. This statistical exercise shows that
there are real differences between the farmed/unfarmed classes and that they are changing in
biodiversity at different rates.
As to the future it is expected that whilst the intensification scenarios are all deleterious the
impact is much greater in infertile grassland because of the high initial capital (Robinson &
Sutherland, 2002). Extensification has some benefits although they are likely to be limited
because of the inherent resistance to change in highly fertile systems (Hopkins et al., 2000).
The most important influence is that the loss of species in infertile grassland may stabilize. By
contrast abandonment will have major benefits to cropland and fertile grassland but will be
94
Session A3: Integrated assessment of ecosystem services from agriculture
strongly negative in infertile grassland because in these systems the high resources of
biodiversity are dependent on traditional farm management.
References
Hopkins, A., et al., 2000. Journal of the Royal Agricultural Society of England 161: 210-223.
Petit, S., et al., 2001. Ambio 30: 81-88.
Robinson, R.A. & W. J. Sutherland, 2002. Journal of Applied Ecology 39: 157-176.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
95
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated assessment of multi-functionality of irrigated agriculture:
An application for the Mediterranean regions
G. Giannoccaro, M. Prosperi, J. Berbel
ETSIAM – University of Córdoba,
Dpt. Agricultural Economics, Sociology and y Policy,
Universidad de Córdoba, Campus de Rabanales Edificio C5, 14014 - Spain
Contact: es2gigig@uco.es
Introduction
Multi-functionality refers to the numerous benefits that agricultural systems may provide for a
region. Besides producing food and fiber, agriculture may also provide jobs for rural people
and contribute to the viability of the area, create a more stable food supply, and provide other
desired (and undesired) environmental and rural outputs. The multi-functionality refers to
comprehensive agricultural systems where irrigated areas cannot be excluded.
The vast majority of studies regarding agricultural multi-functionality have focused on the
supply (or producer) side (OECD, 2001a). They analyse issues such as the joint production of
agricultural outputs, market failures or options for ensuring the provision of public goods and
services from multi-functional agriculture. However, it is surprising to note that the multifunctionality debate has rarely stopped to consider the demand (or consumer) side by
analysing individual preferences for private and public goods and services, as well as
individuals’ opinions on the performance of agriculture within this multi-functional
framework.
Methods
Within this context, in the present research it is proposed an integrated assessment of multifunctionality of irrigated agriculture taking into account simultaneously the two points of
view. On the supply side, the role in terms of multi-functionality of irrigated agriculture will
be assessed. On the demand (or consumer) side, it will be considered the individual
preferences, as well as individual opinions on the performance of irrigated agriculture.
For these purposes, firstly the indicator-based assessment will be used in order to estimate
the functions provided by irrigated systems (see Table 1). Up to four different service classes
will be identified, namely: Food production (concern on quantity); Contribution to economy;
Employment generation; Environmental protection. Essentially, the definition and calculation
of the indicators will be redacted from OECD book (see OECD, 2001b).
Moreover, by comparing the rainfall to irrigated systems it will assess the relative role that
water in term of multi-functionality takes. It will implement the new course of the CAP based
on the decoupled payments and cross-compliance constrains.
Secondly, we will try to establish social utility function that should be maximized using
policy instruments. In this regards, it will be implemented the joint valuation of private and
public goods and services from irrigated agriculture.
Results
The empirical application will be referred to Spanish (Guadalquivir basins), and Italian
(Irrigated Board of Capitanata) regions where the irrigation is a very hard issue for the
achievement of adequate farm income. The first results confirm on the supply side the
important socio-economic role played by the irrigated systems in the Mediterranean areas.
However, the major environmental challenge of irrigated agriculture is to achieve a trade-off
between socioeconomic impacts and environmental sustainability. In this regards the new
policy regulation, such as uncoupled payment subsidies and the environmental restrictions
(e.g., pesticide’s prohibition), may be improve the environmental impact. From demand side
96
Session A3: Integrated assessment of ecosystem services from agriculture
the analysis points out the factors that affect individuals’ opinions on different services of
multi-functionality of irrigated systems. Basically the results reflect the role that the irrigated
systems play at local level. Consequently aspects such as place of residence and demographic
and socio-economic characteristics can significantly influence peoples’ opinions.
This research aims to contribute on the assessment of agricultural multi-functionality
through quantitative methods focusing on the irrigated systems. On the whole, the evaluation
of multi-functionality of irrigated agriculture both on the supply and demand side can help
policy-makers for the design and implementation of agricultural policies. In this sense, the
results will enable in policy decision making so as to optimize the multi-functional services of
irrigated systems according to the social preferences.
Table 1. Multi-functionality assessment of irrigated agriculture with respect to rain fed.
SERVICE
INDICATOR
increase
=
decrease
CLASS
Food production Food production
x
(quantity)
Contribution to
Gross Margin
x
economy
Pubblic support
x
GDP
x
Employment
Employment
x
generation
Soil cover
x
x
Biodiversity index
x
x
Environmental
Landscape
x
x
protection
maintenance
Water use
x
Nitrogen balance
x
Pesticides risk
x
Energy balance
x
x
References
OECD, 2001a. Multifunctionality: Towards an analytical framework. OECD, Paris.
OECD, 2001b. Environmental Indicators for Agriculture, Volume 3 - Methods and Results.
OECD, Paris.
97
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Land Use Functions:
An approach for integrated assessment of the goods and services provided
by agriculture in a multi-functional and sustainable context
M. Pérez-Soba, M.L.M. Jones, C. Pacini, M.L. Paracchini
Alterra, WUR, P.O. Box 47, 6700 AA Wageningen, The Netherlands
Contact: Marta.perezsoba@wur.nl
Introduction
The traditional role that agriculture plays in the European rural areas has evolved from the
traditional, mono-functional production of goods (food and fiber), to a multi-functional
provision of goods and services (landscape identity, cultural heritage, provision of habitat,
etc.). The impact that this new role of agriculture might have on sustainability of regions is
largely unquantified. There is, therefore, a need for approaches that help to identify the
multiple functions of agricultural systems and assess their interrelations with the other land
uses co-existing in the rural areas. These approaches should integrate the goods and services
provided by agriculture in a multi-functional context, through a true integration of the
economic, environmental and societal issues (three dimensions of sustainability) at a
meaningful spatial scale considering the territorial perspective.
The integration of economic, environmental and social issues demands complex systems of
thinking based on multi-scale integrated analysis and not a collection of independent analyses,
each based on a defined discipline. In addition, the assessment has to be performed at the
appropriate spatial scale. For example, the environmental services provided by the agricultural activities may vary among countries and regions depending on their agricultural
systems and social demand. Therefore, there is a need for flexible approaches that allow the
assessment to be performed at the appropriate regional scale. Finally, it is deemed vital to
consider stakeholder preferences when linking multi-functionality to the sustainability
concept.
In conclusion there is a need for an approach that (i) defines and measures the economic,
environmental and social goods and services – functions – provided by the multiple use of
agricultural land at territorial level, and (ii) helps to identify the sustainability
limits/thresholds/targets of these functions. An interdisciplinary team has addressed this need
by developing an innovative approach, the Land Use Functions (LUFs), which (i) link directly
the economic, environmental and social functions to the use of the land; (ii) provide a flexible
method to assess at the appropriate spatial resolution; and (iii) transparently address the
identification of the different functions that a specific land use (in this case agriculture) might
have, facilitating the explicit analysis of their trade-offs. This paper describes the conceptual
LUFs framework and highlights the advantages of its use.
Land Use Functions: the good and services attached to the land use
The LUF approach has its main roots in the concepts of multi-functionality in agriculture
(OECD, 2001), ecosystems goods and services (Costanza et al., 1997), and landscape
functions (Kienast et al., 2007). Land Use Functions are defined as the private and public
goods and services provided by the different land uses, that summarize the most relevant
economic, environmental and societal aspects of a region. For example, agricultural land use
might have several functions, such as provision of employment or landscape identity (social),
provision of food, timber or biofuels (economical), support for biodiversity and maintenance
of ecosystem processes (environmental). The LUFs allow for a cross-sectoral assessment
since they consider the main sectors involved in land use, i.e. agriculture, forestry, energy,
tourism and nature conservation. The LUFs aggregation method is built to be applied on a
Europe-wide indicator framework consisting of a large set of indicators, selected to represent
98
Session A3: Integrated assessment of ecosystem services from agriculture
key impact issues of sustainability of land use, as listed in the EC Impact Assessment
Guidelines (CEC, 2005). These impact indicators cover a wide variety of land uses across the
three dimensions of sustainability. At a higher level, there are Land Use Functions (LUFs)
(Pérez-Soba et al., 2008) which aggregate indicator types loosely into a multi-dimensional
space, which is linked to the three dimensions of sustainability. Each sustainability dimension
is represented by three LUFs: Economic (Residential and Industrial Services, Land-based
Production, Infrastructure), Environmental (Abiotic Resources, Provision of Habitat,
Ecosystem Processes) and Social (Work, Health and Recreation, Culture), giving nine Land
Use Functions in all. The LUF framework reduces the number of dimensions represented by
the set of impact indicators to make sustainability impact assessment interpretable. The
concept of sustainability is recognized at two principal scales – at the level of individual
indicators for which specific sustainability limits are defined, and at the level of Land Use
Functions where an optimum sustainability with respect to multi-functional land use is arrived
at by an end-user through evaluation of competing LUFs within the ‘trade-off evaluation
space’. The indicator sustainability limits are independent of the wider LUF evaluation
process, but may be subject to change given future knowledge, or when investigated within
the context of local-scale participatory approaches.
The advantages of using Land Use Functions
LUFs are a new tool for assessing the good and services provided by land use, such as
agriculture, in a multi-functional and sustainable context. They simplify complex assessments
based on a large set of indicators by grouping them into nine LUFs that cover the main
functions of the land. It, therefore, makes possible for policymakers, stakeholders and
scientists to identify at a glance those functions of the land that are hindered and those
functions that are enhanced by a change in the land use, e.g., policy option (Figure 1).
Figure 1. The impacts are
summarized in changes in
the nine LUFs, represented
in the spider diagram.
References
CEC, 2005. SEC 791.
Costanza, R., et al., 1997. Nature 387: 253-260.
Kienast, F., et al., 2007. Internal SENSOR deliverable. OECD, 2001. Paris.
OECD, 2001. Multifunctionality: Towards an analytical framework. OECD, Paris.
Pérez-Soba, M., et al., 2008. In: Eds. K. Helming, et al., Sustainability impact assessment of
land use changes. Springer Verlag, pp. 375-404.
99
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
How do policy options modify landscape amenities? An assessment
approach based on public expressed preferences
T. Pinto-Correia, S. Truchet, C. Machado, F. Barroso, P. Picchi, Y. Michelin,
N. Turpin, J.P. Bousset
ICAM-Group on Mediterranean Ecosystems and Landscapes, University of Évora,
Colégio da Mitra, 7000 Évora, Portugal
Contact: mtpc@uevora.pt
Introduction
‘Landscape’ means an area, as perceived by people, whose character is the result of the action
and interaction of natural and/or human factors (European Landscape Convention, 2000). The
definition is clear and nowadays well accepted. Trickier is to progress in the sense of
identifying landscape quality objectives, which should depend on the public demand of these
landscapes, i.e. their cultural and amenity values. New EU policies focusing on the
development of Europe’s rural areas, including its broad variation of landscapes, and specially
the New European Model of Agriculture, which stresses the territorial role of agriculture,
increase the need to better understand how these different landscapes are valuated by people,
and how farming may contribute to so-called landscape quality. Indeed, people do value
landscape patterns, elements, and their combinations (Dramstad et al., 2006). They are also
able to express their preferences for certain landscape patterns, but there are still challenges to
assess these preferences taking care of the diversity and complexity of the landscape (AlKodmany, 1999), and to combine these preferences with the increasing number of models
assessing how policy options can modify the chain “on-farm modifications –> agricultural
land use changes –> landscape pattern –> adequacy to people needs and expectations”. And
reversely, how human expectations can be considered in the design of specific policy options
can be coordinated with each other. This paper describes a landscape amenities evaluation
tool that aims at filling this gap. This tool is based on the identification of an optimum range
of landscape composition, defined through a survey assessment to user groups or expert
panels, and its relation, through selected indicators, to the landscape composition resulting
from different scenarios.
Methods
The approach focuses on human factors and follows the ‘subjective’ paradigm. Landscape
visual aesthetic quality is considered to be a product of the visible features of the landscape
interacting with personal cultural background of the observer. Landscape quality, for what
concerns its cultural and amenity functions (De Groot & Hein, 2007), is consequently “in the
eyes of the beholders”. Linking current preferences of various groups of people with future
changes in the landscape lead by policy options can be made assessing the preferences for
present land cover combination using photographs as visual stimuli (Fairweather & Swaffield,
2002; Lewis, 2008). We tested the index of function suitability (IFS) designed by PintoCorreia & Picchi (2008) associated with SEAMLESS outputs (Van Ittersum et al., 2008) for
various policy options in two case studies, in Portugal and in France. The approach of this IFS
is based on the identification of a range of optimum combinations in land cover, for the
support of the various amenity functions. This optimum is then compared with the land cover
combination that results from the scenarios of policy changes, in order to assess the function
suitability of that combination for each specific function.
We start from a landscape taxonomy jointly built by scientist and expert panels in each
region. This taxonomy is used to design landscape prototypes in each region. The prototypes
are key elements to link model outputs to preferences, through the use of snapshots and
photos.
100
Session A3: Integrated assessment of ecosystem services from agriculture
Then we capture the preferences on the capability of prototypes to contribute to landscape
functions (recreation, second housing, hunting, etc) from individual interviews, using
snapshots and IFS. We specify changes in the landscape prototypes as policy impacts by
using an interaction model between results of FSSIM simulations (pressure indicators) and
landscape (environmental function & state indicators).
Results and discussion
The use of the photographs as visual stimuli has proved to demand a very well planned
selection and manipulation. The photographs should show clearly what is aimed at being
assessed. Similarly, the organization of the inquiries and the sampling method are important.
Several tests have been developed. Results show it possible to identify the optimum range of
combinations. The relation to scenarios through the indicators selected is still under progress
but seems to lead to a possible assessment. Figure 1 show a test result of the IFS applied to
four different case studies in Portugal. Other results will be presented in the paper.
Figure 1. Landscape multi-functionality for four sample municipalities in the Alentejo region.
References
Al-Kodmany, K., 1999. Landscape and Urban Planning 45: 37-45.
De Groot, R. & L. Hein, 2007. In: Eds U. Mander, et al., Multifunctional Land Use, Meeting
Future Demands for Landscapes Goods and Services. Springer, pp. 15-36.
Dramstad, W.E., et al., 2006. Landscape and Urban Planning 78: 465-474.
Fairweather, J.R. & S.R. Swaffield, 2002. Intern. Journal of Tourism Research 4: 283-297.
Lewis, J.L., 2008. Landscape and Urban Planning 85: 45-59.
Pinto-Correia T. & P. Picchi, 2008. Landscape Amennities Handbook. Appendix to
SEAMLESS PD 224.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
101
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Regional sustainability contrasts in Brazil as indicated by the CompasSus:
Compass of Sustainability
Saulo Rodrigues-Filho, Vanessa M. De Castro, Fabio Bakker, Diego Lindoso,
Nathan Debortoli, Marcel Bursztyn
University of Brasília, FUB Center for Sustainable Development - CDS,
Av. L3 Norte, Bloco. C, Campus UnB, 70904-970, Brasília-DF, Brazil
Contact: srodrigues@unb.br
Introduction
Aiming at the current need for assessing sustainable processes and states, the objective of this
article is to present preliminary results obtained by applying the Compass of Sustainability
(CompasSus) to administrative regions of Brazil. The methodological approach is based on
both an original conceptual framework and existing assessment methodologies, such as the
Barometer of Sustainability and the Ecological Footprint.
Methods
CompasSus is elaborated under the perspective of the different operational concepts of
sustainable development, according to those methods. In spite of the advances on the
perspective of sustainable development perceived in international agreements for conservation
of nature, as well as in the implementation of national policies aiming at sustainability, there
still exists an operational ambiguity of the concept of sustainability. Thus, its fuzzy theoretical
consolidation for clear definition of objectives and lines of action still leads to different
assumptions for diverse authors.
Results
Looking at the Brazilian case within the LUPIS EU-Project, conservation strategies are
needed to protect the world’s largest tropical rainforest against a predatory business-as-usual
frontier expansion. On the other hand, socio-economic development is a legitimate demand of
Amazonian residents (12% of Brazil’s population) living in a region that occupies 58% of the
Brazilian territory and produces just 7% of the country’s GDP. The ecological-economic
literature on weak and strong sustainability has explored some important values and
Figure 1. Illustration of the conceptual
framework of CompasSus, a combination
of both weak and strong sustainability
approaches.
102
Session A3: Integrated assessment of ecosystem services from agriculture
interpretations affecting a desired operational concept of sustainability. The concept of critical
natural capital, for example, has a pivotal role for defining strong sustainability.
As for addressing trade-offs between environmental conservation and regional
development demands the Compass of Sustainability (CompasSus) introduces a Combined
Hemispheric Assessment of Sustainability (CHAS), as illustrated in Figure 1. Whereas the left
hemisphere reflects a weak sustainability approach, with a focus on local/regional impacts
caused by production systems, the right hemisphere mirrors a strong sustainability focused on
global impacts caused by consumption patterns, considering critical natural capital an
important component of sustainability. Preliminary results using the CompasSus approach
indicate its potential to combine strengths of both weak and strong sustainability approaches
so that a more comprehensive notion of sustainability for different Brazilian Regions is
achieved. According to the sustainability assessment performed the state of Santa Catarina in
the Southern Region of Brazil scored the best performance due to a combination of
environmental and socioeconomic indicators: high efficiency in energy consumption in
relation to wealth; low intensity of fertilizers and pesticides associated with high agricultural
yields; and an above average performance in all social and economic indicators. On the other
hand, the state of Maranhão in the Northeastern Region of Brazil scored the worst
sustainability performance as a consequence of its least developed social and economic
dimensions, including indicators of education, life expectancy and income.
103
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Economic modelling of multi-functionality at the farm level
E. Romstad, K. Louhichi, P. Zander
Dept. of Economics & Resource Management,
Norwegian University of Life Sciences,
P.O. Box 5033, 1432 Aas, Norway
Contact: eirik.romstad@umb.no
Introduction
Market failures constitute the primary reason for justifying policy intervention. Goods not
amicable for market exchange, also denoted non-market or public goods, are of particular
relevance in the case of agri-environmental policies and has received substantial attention
both in the academic literature and in real policy (see Lankoski & Ollikainen (2008) for an
overview). Multi-functionality basically implies that some non-market goods are produced in
conjunction with the ordinary production of market goods in agriculture.
This paper considers the jointness and policy implications of multi-functionality, and
provides insights to how multi-functionality can be modelled under various settings for data
availability.
Our starting point is that in an economically rational equilibrium there must be a negative
relationship between the market commodity and the non-market good or service per hectare of
land as long as the non-market attribute has a non-negative price. The reason for this is if this
was not the case, the farmer could earn larger expected profits by increasing the per hectare
supply of either of the goods. Figure 1 provides an illustration.
Figure1. The trade-off between the
extent of market (y) and non-market (z)
goods or services.
In Figure 1 only the thick segment on the production possibility frontier, that bounds the production possibility set, are technically efficient. To see this consider allocations on the frontier
left to F or below E, where it is possible for the farmer to increase the production of both y
and z without additional costs. For a given price ratio, –py /pz, the optimal allocation is in A
where the relative price line tangents the production possibility frontier.
Multiple authors claim that there is some strict technical jointness (again see Lankoski &
Ollikainen (2008) for an overview) pointing to empirical data. We claim that this perceived
jointness is due to two factors. First, if the price of the market commodity is increased, it
becomes profitable to increase the expenditures spent on commodity production, leading to an
expansion of the per hectare production possibility set. If this expansion more than offsets the
substitution of producing more of the market commodity due to the relative price change,
there will be an increase in the production of both the market and non-market good (Romstad,
2007).
104
Session A3: Integrated assessment of ecosystem services from agriculture
Second, the increased per hectare profits due to a price increase of the market commodity
entails that more acreage goes to this type of production. Consequently, also more of the nonmarket good may be produced on this acreage, even if the above substitution effect is larger
than the effect of the expansion of the production possibility set (Romstad, ibid.). These
analytical results have strong implications for how multi-functionality is to be modelled.
Ideally, one seeks to embed non-market goods or services included in the farm model
objective function so that one could fully study the impacts of relative price changes. Unfortunately, there has been little quantitative modelling done on this using real data. A notable
exception is Groot et al. (2007) who find the negative relationship illustrated in Figure 1 for
biological diversity and biomass production. Unfortunately, few of these multi-attribute
relationships have been analysed to capture the corresponding production possibility frontier.
Methods
FSSIM is a bio-economic farm model simulating farm level behaviour given a set of biophysical, socio-economic and region-specific policy constraints, allowing for technological
innovations and policy changes (Van Ittersum et al., 2008). It is a primal based approach (i.e.,
technology is explicitly represented) using nonlinear programming to account for selected
positive and the negative jointness in outputs (i.e., joint production) associated with the
production process. FSSIM is based on discrete production functions and a limited number of
externalities functions. These specifications enable FSSIM to directly explore the impacts of
some premium and policy changes not only on the relationship between market and nonmarket goods, but also on the production process (Louhichi et al., 2007).
Whenever the production possibility frontier is not well specified due to the before
mentioned lack of data, it is difficult to perform analyses using direct payments for the nonmarket goods or attributes. However, many multi-functionality attributes, like hedgerows or
solitary trees, are not joint products in a strict sense. In these cases there is a discrete land use
decision to be made by the land owner on adding such elements in the landscape if the
payments for these elements is sufficiently high to offset the reduction of the most profitable
land use in absence of targeted land use payments. Such elements represent a direct
competing land use that appear of particular relevance given the work done on the visual
characteristics of landscapes in Seamless, and enters the optimal solution if the corresponding
acreage or attribute payments are sufficiently high.
Results and discussion
The linkages between commodity production and acreage use on one hand, and landscape
attributes like scenic nature or biological diversity on the other hand, are not yet fully
implemented. Such results are foreseen before the end of 2008.
References
Groot, J.C.J., et al., 2007. Agriculture, Ecosystems and Environment 120: 58-69.
Lankoski, J. & M. Ollikainen (Eds.), 2008. Multifunctionality in agriculture. Evaluating the
degree of jointness. Policy implications. OECD, Paris, 252 pp.
Louhichi, K., et al., 2007. A generic template for FSSIM for all farming systems, PD 3.3.11,
Seamless Integrated project, IAMM, Montpellier, 82 pp.
Romstad, E., 2007. Multifunctionality in SEAMLESS-IF. PD 3.3.10, Seamless Integrated
project, Norwegian University of Life Sciences, Ås, 15 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
105
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Predicting combined impacts of land management and pollution on
biodiversity: A dynamic niche occupancy approach
E.C. Rowe, S. Smart, C.D. Evans, B.A. Emmett
Centre for Ecology and Hydrology,
Environment Centre Wales, Deiniol Road, Bangor, Gwynedd, LL57 2UW, UK
Contact: ecro@ceh.ac.uk
Introduction
Biodiversity has intrinsic value and may be essential to certain ecosystem functions. The
maintenance of biodiversity is thus a key ecosystem service (Millenium Ecosystem
Assessment Board, 2003), but the difficulty of predicting effects of changes in land use and
management on biodiversity means that this service is not easily incorporated into integrated
assessments of such changes. Niche envelope modelling has been widely applied to predict
impacts of climate change on species’ distributions. By incorporating other environmental
factors into niche models, in particular soil and vegetation characteristics, responses of
species to multiple drivers can be predicted (Coudun, 2006).
While sulphur pollution has declined greatly in Europe since the 1980s, pollution by
reactive nitrogen continues. Nitrogen pollution is causing a global loss of biodiversity
(Phoenix et al., 2006), due largely to effects on interspecific competition among plants. The
reduction of N limitation increases the growth of fast- and tall-growing plant species, thus
shading out smaller-growing, light-demanding plant species (Clark et al., 2007) and reducing
the availability of niches for many animal species (Wallisdevries et al., 2006). Semi-natural
European habitats such as grasslands and heathlands support large numbers of species
principally because the regular removal of biomass through grazing or fire reduces groundlevel shading.
We present a dynamic niche occupancy approach that can be used to assess the combined
impacts of management change and nitrogen pollution on plant species occurrence.
Methods
In the dynamic niche occupancy approach, changes in the likely occurrence of a species are
predicted by linking dynamic models of environmental change to regression models of the
niche for the species. The ‘realiszd niche’ of a plant species can be defined as the
hypervolume where the species occurs within a space defined by a set of environmental
factors and restricted by biotic interactions (Hutchinson 1957). Empirical realized niche
models have been derived for 822 higher plant and 315 bryophyte species in the UK, by
multiple logistic regression of their occurrence in relation to soil pH, total soil carbon and
nitrogen concentrations, soil moisture and canopy height. These factors are easily measured
and hence have been recorded together with plant species occurrence in several large datasets.
The set of species models is collectively called GBMOVE (Smart et al., 2009).
Predictions of changes in the environmental factors that define plant species niches were
obtained using the biogeochemical models MAGIC (Cosby et al., 2001) and SUMO
(Wamelink et al., 2009). Using the outputs from these models to solve the regression models
allows changes in the probability of occurrence of individual plant species to be predicted.
Plant species are valued differentially by biodiversity specialists, and lists of positive and
negative indicator species are available for particular habitats (JNCC, 2006). A habitatspecific index of biodiversity value was calculated by subtracting the mean probability of
occurrence of negative indicator species from the mean probability of occurrence of positive
indicator species.
106
Session A3: Integrated assessment of ecosystem services from agriculture
Results and discussion
The model chain was tested by comparison of predictions against observed changes in species
composition at a lightly grazed upland site in Cumbria, UK. Predicted rates of change in
probability of occurrence of species present at the site were correlated with observed rates of
changes in these species (Pearson correlation coefficient = 0.568, p=0.002). Historic and
projected changes to soil pH and C/N ratio, probability of occurrence of positive indicator
species and overall habitat quality are illustrated in Figure 1.
(b)
5.5
5.0
pH
C / N g g-1
30
20
4.5
10
4.0
0
1850
1900
1950
Year
2000
2050
C/N
pH
(c)
1.0
0.06
0.8
0.04
Habitat Quality
40
Probability of Occurrence / Pmax
(a)
0.6
0.4
0.2
0.02
0.00
-0.02
-0.04
-0.06
0.0
1850
1900
1950
Year
2000
2050
1850
1900
1950
2000
2050
Year
Figure 1. Simulated changes at Moor House long-term monitoring site, Cumbria, UK, under
the Gothenberg scenario for projected nitrogen and sulphur emissions: (a) soil pH and C/N
ratio simulated using the MAGIC soil chemistry model; (b) probabilities of occurrence of
positive Common Standards Monitoring indicator species rescaled to maximum probability
for the species; (c) overall habitat quality. Soil water content and canopy height were assumed
to be constant.
The model chain is currently being refined by further testing against long-term floristic
datasets, and by incorporation of a more sensitive indicator of soil N availability. The
approach is being used to assess the effects of changes in pollutant load under different air
pollution policy scenarios, and is being adapted as a tool to assist land managers in setting
appropriate levels of grazing and fertilization to maintain plant diversity. Forecasts of
biodiversity change will also be useful for broader assessments of ecosystem services.
References
Clark, C.M., et al., 2007. Ecology Letters 10: 596-607.
Cosby, B.J., et al., 2001. Hydrol. Earth Syst. Sci. 5: 499-517.
Coudun, C., et al., 2006. J. Biogeog. 33: 1760-1763.
Hutchinson, G.E., 1957. Cold Spring Harbor Symp. Quantitative Biol. 22: 415-427.
JNCC, 2006. http://www.jncc.gov.uk/page-2237
Millenium Ecosystem Assessment Board, 2003. http://www.millenniumassessment.org/
en/Framework.aspx
Phoenix, G.K., et al., 2006. Global Change Biol. 12: 470-476.
Smart, S.M., et al., 2009. Ecol. Model. (submitted)
Wallisdevries, M.F. & C.A.M. Van Swaay, 2006. Global Change Biol. 12: 1620-1626.
Wamelink, G.W.W., et al., 2009. For. Ecol. Mgmt. (submitted)
107
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Cost effectiveness of biodiversity in agricultural production systems
D.P. Rudrum, R.A.M. Schrijver, A. Corporaal, T.J. De Koeijer
LEI, Postbus 47, 6700 AA Wageningen, The Netherlands
Contact: dale.rudrum@wur.nl
Introduction
In The Netherlands, there is an ongoing discussion on how to reform the current Common
Agricultural Policy from payments based on historical production levels towards payments
based on delivered services according to societal demands. One of the demands being
considered is the conservation of biodiversity. To explore this option further, the question that
should be answered now is how biodiversity in the modern agricultural landscape could be
stimulated and which measures are the most cost effective. As measures in the agricultural
landscape should be taken by the farmer insight is needed in which measures could be
incorporated in the management of the agricultural production system and how these
measures affect the production system, biodiversity and income. The effect on income of the
farmer is important as it may be assumed that if a measure is profitable a farmer will be more
willing to adapt a measure than if the measure is not profitable. Assuming that farmers do not
have a willingness to pay for biodiversity, the loss of income provides a minimum cost price
for the concerning measure.
For the assessment of the cost effectiveness of biodiversity in the agricultural landscape an
indicator for biodiversity and a method to calculate cost price are needed. Both have to be
related to on farm measures aimed at influencing biodiversity. In this work the HEMO index,
a measure for biodiversity is incorporated into FIONA, a farm optimization model. Our
research is aimed at providing a first study to show the possibilities of the combination of
FIONA with the HEMO index. For this two cases are chosen. One area characterized by dairy
farms on sand soil situated in the east of The Netherlands, and one area characterized by
arable farms on clay soil situated in the south-west of The Netherlands. Possible measures for
the biodiversity in these areas are chosen with the help of local experts. Next based on data of
the farm structure in the area for each area two representative farms are identified. These
farms are modelled in FIONA to give both cost price and effects on biodiversity.
Each proposed measure is evaluated on its economic and ecological effects after which the
effects of combinations of measures will be analysed.
Methods
The biodiversity indicator should give a continuous measure of biodiversity with which
completely different options can be compared, and a tuning mechanism so that the order of a
list of possibilities is in accordance with policy priorities. In this work first results with a new
biodiversity indicator, the HEMO index, will be presented. In short, the index is calculated in
the following steps: the geographical location of a plot of land provides a list of possible and
or relevant species that may occur in the area. The specific environmental conditions are used
as a proxy for the occurrence of the species on that plot. Finally, the index is calculated based
on the sensitivities of the occurring species for environmental conditions. Habitats for
sensitive species score better than habitats only supporting common species.
Corporaal et al. (in prep.) introduced the niche number, an indicator based on the
environmental conditions like the fertility of the soil, the water table, the management of the
vegetation in which a species can live. If the species is indifferent to a condition it scores
100%, if however it is sensitive to the condition it scores a percentage based on the range in
habitat variation still acceptable for this species. The niche number of a species is the average
over all indicators. Species scoring 100% can live anywhere, and a decreasing niche number
is associated with an increase in sensitivity to habitat conditions. In our work the conditions
108
Session A3: Integrated assessment of ecosystem services from agriculture
are all weighted one, but distinctions can easily be made if required.
The niche number is used to calculate the HEMO index. This is the ratio of habitat
sensitive species to habitat indifferent species that occur (or in our case: are expected to
occur) in a certain area. The species considered are chosen in the context of the geographical
location. In this way, the populations of species supported by very different habitats can be
compared, essentially based on the likelihood of the ecosystem occurring elsewhere.
The next step is to link the conditions used for the niche number to management practices
on the farm. As the condition is used, and not the measure that provides the condition it is
made easier to compare the effect of separate measures and to explore the possibility to mix
measures. Obviously not all conditions can be influenced by farming practices. Soil type is a
given. Also some measurements remain directly linked to conditions, like grazing pressure
and fertilization. The possibility to mix measures is, therefore, considered the main advantage.
Finally the costs of the measures have to be assessed. This is done using FIONA. With this
model the economic optimal in and output can be identified given the restrictions of the farm
concerning the area and amount of labour. The model was developed by Berentsen for
analysis of economic and ecological effects of technical and institutional changes concerning
fertilization practices (Berentsen & Giessen, 1995). We adapted the model for the evaluation
of the participation level of farmers into on farm nature conservation schemes (Schrijver et
al., 2006). By the incorporation of a biodiversity index the model is also suitable for cost
effectiveness analyses.
Results and discussion
The results are not available yet but will be available by the end of the year 2008.
References
Berentsen, P.B.M. & G.W.J. Giessen, 1995. Agricultural Systems 49: 153-175.
Schrijver, R.A.M., et al., 2006. Agrarwirtschaft 55(5/6): 280-289.
109
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A novel site-specific methodology to assess the supply curve of
environmental services
J.J. Stoorvogel, A. Temme, J.M. Antle, L. Claessens, J.M. Schoorl
Land Dynamics Group, Wageningen University, P.O. Box 47,
6700 AA Wageningen, The Netherlands
Contact: Jetse.Stoorvogel@wur.nl
Introduction
Water use and watershed management have a long history of conflicts which continue to take
place as long as we deal with limited resources (Kashaigili et al., 2003). Recently a new
policy instrument has been introduced that is based on positive incentives through so-called
payments for environmental services (PES, Zbinden & Lee, 2005). However, there is a
structural lack of ex-ante screening tools that allow for a rapid assessment of the potential for
payments for environmental services. The issue has multiple dimensions including the
provision of environmental services, the economics of the adoption of alternative practices,
but also the proper implementation of PES schemes. A joint analysis should provide key
information to policy makers: What is the supply of environmental services at a particular
price? In this study we will demonstrate a site-specific methodology to assess this supply
curve of environmental services. The methodology will be illustrated with a case study to
secure sustainable land use of the Tugela watershed and the Woodstock dam in South Africa.
The 1150 km2 basin is an important provider of food crops but at the same time the Tugela
basin is an important provider of drinking water for Johannesburg and the hydro-electrical
plant in the Woodstock dam is a key player in the South African electrical network. As a
result water has a significant value and payments to the various land and water users in the
watershed are feasible if they result in a more efficient water use and an increase of water
availability.
Methods
Antle & Valdivia (2006) introduced a methodology to model the supply of ecosystem services
from agriculture. The methodology provides an efficient way to calculate the supply of
environmental services for a region. However, the methodology is not site-specific and
therefore not very suitable for environmental problems like soil erosion and water supply that
require the analysis of spatial interactions. We therefore linked the methodology to the
LAPSUS model (Schoorl et al., 2002) to be able to evaluate in a site-specific manner the
effect of the adoption of alternative practices in terms of water use and soil redistribution. The
methodology determines for each location in the area an expected opportunity cost to switch
to an alternative management practice on the water holding capacity at that location,
erosion/sedimentation, and expected net returns. The opportunity costs now allow us to
determine for each location and for various scenarios of the payments (prices for water and
sediment, or a fixed payment for adoption) whether farmers will accept the payments for
environmental services. This can be presented in a map, but the results can also be aggregated
for the region to calculate the supply curves for the environmental services.
Results for the Tugela river basin
Land use in the Tugela basin was mapped on the basis of a supervised classification of a
satellite image and classified in five major categories (Table 1). For each of these classes
possible changes in management were identified to conserve water use. Survey and literature
data provided insight in the variation in opportunity cost to switch practices and the effect of
the potential changes that were identified. The effects were evaluated with the LAPSUS
model to assess erosion and runoff.
110
Session A3: Integrated assessment of ecosystem services from agriculture
Table 1. Land use in the Tugela basin and proposed management changes to increase water
use efficiency and reduce erosion.
Land use
Proposed change
Large scale irrigated farming
Improved irrigation efficiency through drip irrigation
Large scale rangelands
Reduction of stocking density/ controlled burning
Subsistence farming
Simple erosion control measures
Communal lands
Fencing and rotational grazing
Forest areas
Improved trail management
The results (Figure 1) show that even without payments (corresponding to the negative prices
for ES) a certain percentage of the farmers may already adopt the conserving practices. With
an increase in the price for the ES the area under conservation will quadruple together with
the environmental services provided (right side of Figure 1). The key question that remains is
the assessment of proper transaction and adoption costs. These costs will shift the supply
curve leftward and may have a significant impact on the interpretation of the results.
1000
Price ES
(R/m 3)
750
500
250
0
12500
10000
7500
Participation (ha)
5000
2500
0
500
-250
1000
1500
2000
2500
Environmental service (000 m 3/yr)
-500
-750
-1000
Figure 1. The supply curve of environmental services (expressed as water preserved) at the
right side and the adoption of conserving practices in the watershed under different prices for
the environmental services on the left side of the graph.
Discussion
The case study illustrates the applicability of the modelling framework with limited data
availability. The linkage with site-specific models like the LAPSUS model provides new
opportunities to analyse processes that have a clear spatial interaction and for which the local
conditions play an important role in the effect of changes.
References
Antle, J.M. & R.O. Valdivia, 2006. Aust. J. Ag. & Res. Econ. 50: 1-15.
Kashaigili, J.J., et al., 2003. Phys. and Chem. of the Earth 28: 839-851.
Schoorl, J.M., et al., 2002. Soil Sci. Soc. Am. J. 66: 1610-1619.
Zbinden, S. & D.R. Lee, 2005. World Development 33: 255-272.
111
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Redefining landscape economics: A proposal to integrate economics
approaches and landscape disciplines
C. Tagliafierro, M. Antrop, V. Van Eetvelde, W.G. Hutchinson, A. Longo
Gibson Institute for Food Land and the Environment, School of Biological Sciences,
David Keir Building, Stranmillis Road, BT9 5AG Belfast, UK
Contact: ctagliafierro01@qub.ac.uk
Introduction
The need to implement effective incentive systems for landscape planning and management
requires policymakers, stakeholders and planners to understand the values that communities
attach to landscape. Placing monetary values on landscape and its characteristics has been
challenging economists for the last decades because of the complexity of components to take
into account when analysing landscape. In order to address this issue, the new discipline of
landscape economics needs interdisciplinary approach, integrating analytical methodologies
and involving researchers from the different landscape disciplines. Only in this way it is
possible to define a new valuation procedure that integrates the real landscape - as it exists in
the territory - with economic modelling approaches. Some attempts in this direction include
Geoghegan et al., 1997; Bell & Irwin, 2002; Campbell, 2007.
The methodology here presented aims at giving a contribution bridging knowledge and
methodologies of different landscape disciplines. It integrates landscape ecology, landscape
preference studies and environmental economics, through the analysis of biophysical and
cultural components of the landscape and the public’s preferences for the landscape using
discrete choice experiments. We argue that this approach represents a common framework
where landscape scientists and economists can collaborate on a common goal.
Methods and results
The case study area is the Peninsula of Sorrento, in the South of Italy. It represents a good
example of highly valuable Mediterranean landscape that embodies strong identity and
cultural values and provides a substantial contribution to the tourist activities in the area. The
growth of tourism activities and the decline in traditional farming practices due to economic
pressures are leading towards a loss of this unique landscape. The methodology steps are
schematically represented in Figure 1.
The first step of our approach is the identification of the landscape types in the study case
area, which form the bases for the selection of the ‘attributes’ to be used in the monetary
valuation of the landscape through the application of choice experiments. A parametric
landscape classification methodology using GIS-techniques is applied to the study area (Van
Eetvelde & Antrop, 2008), to identify landscape types and help describe them in terms of
their attributes – quantified with metrics (visual indicators). Digital maps have been analysed
and integrated with the GIS software to identify landscape structural components. The urban
density degree and the presence of scattered settlements is also analysed in detail.
Subsequently, principal components analysis is used to define homogeneous clusters for all
these components, which gives rise to classification of six main landscape types and ten sublandscape-types (distinct on the basis of altitude and settlement degree).
Based on the landscape classification, the second stage involves taking more than 300
pictures of the study area. Location of the pictures is geo-referenced; a viewshed analysis of
all the pictures is conducted and the attributes of the landscape covered by the view area are
quantified in visual indicators (McGarigal & Marks, 1995; Ode et al., 2008). An example of
such indicators is reported in Table 1. Finally, based on these attributes, a sequential
experimental design with a Bayesian information structure is used to increase the sampling
efficiency of a discrete choice experiment survey. The results from this are used to quantify
112
Session A3: Integrated assessment of ecosystem services from agriculture
the non-market benefits of landscapes and their attributes. These values may provide
policymakers, stakeholders and planners with a better understanding of the public interest for
landscape.
Landscape Ecology
Landscape
Classification
“Types”
Attributes
Visual
indicators
Integration
Economic model
Individuals’ preferences
WTP for landscape attributes
(Euros/household)
Environmental Economics
Table 1. Example of visual indicators.
Visual indicator
Meaning
N. of elements
Number of different landscape
patches
Built area
% of area covered by artificial
areas
Naturalism index
% of area covered by natural
systems
Shape index
Relation between perimeter and
area of patches, adjusted for a
constant related to the type of
data file used (raster or vector)
Total area of viewshed Hectares covered by the
picture/viewshed
Figure 1. Schematic
representation of the
methodology main steps.
Attribute represented
Degree of fragmentation
Presence and degree of
urbanization
Degree of natural character of
the area
Shape complexity related to
the geometry of patches
Depth of the view
References
Bell, K.P. & E.G. Irwin, 2002. Agricultural Economics 27: 217-232.
Campbell, D., 2007. Journal of Agricultural Economics 58: 497-483.
Geoghegan, J., et al., 1997. Ecol. Econ. 23: 251-264.
McGarigal, K. & B.J. Marks, 1995. FRAGSTATS: spatial pattern analysis program for
quantifying landscape structure. General Technical Report PNW-GTR-351. US Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, OR.
Ode, A., et al., 2008. Landscape Research 33(1): 89-117.
Van Eetvelde, V. & M. Antrop, 2008. A stepwise multi-scaled landscape typology and
characterization for trans-regional integration, applied on the federal state of Belgium.
Submitted to Landscape and Urban Planning (January, 2008).
113
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Towards indicators of multi-functionality: Assessing jointness between
commodity and non-commodity outputs
Nadine Turpin, Lee Stapleton, Guy Garrod, Eric Perret, Floor Brouwer
UMR 1273 Métafort, 24 avenue des Landais, B.P. 50085,
63172 Aubiere Cedex, France
Contact: nadine.turpin@cemagref.fr
Introduction
Multi-functionality is one way to reconcile agriculture with sustainable development:
agriculture, beyond the production of food and fiber, also provides important social,
environmental and economic functions to society. In general, much of the current literature on
agricultural multi-functionality is qualitative and narrative, and focuses on demonstrating the
existence of social and/or environmental functions. This paper goes beyond qualitative
analysis and seeks to measure a degree of multi-functionality. More precisely, the objective of
this paper is to provide a theoretical backdrop to the design of indicators that measure the
degree of multi-functionality involved in the co-production of commodity and noncommodity outputs by farms. It is important that decision makers can measure the sustainable
development implications of a given policy intervention in terms of how this affects the multifunctional attributes of a given area. Policy formulation that aims at supplying commodity and
non-commodity outputs separately will lead to higher implementation costs than when the
policy considers multi-functionality and encourages farmers to supply these outputs jointly
(OECD, 2001). More specifically, measuring the degree of jointness will provide insights to a
potential decrease of the use of public funds to support agriculture and its associated
amenities in the context of the European Union’s (EU) Common Agricultural Policy (CAP).
Methods
A brief review of the start-of-the-art in agricultural multi-functionality research emphasizes
that much of the current literature on the subject is qualitative rather than quantitative. This
literature review serves to highlight how there is a pressing need for operational indicators of
multi-functionality which can, for example, provide insights into the likely effects of a
decrease in the availability of public funds to support agriculture and its associated amenities
in the context of the EU’s CAP.
Therein lies the rationale and motivation for the work presented below whereby we present
a theoretical framework for joint supply based on the assumption that the degree of jointness
has consequences both in terms of commodity production costs and non-commodity
production. Based on the jointness definition, the assessment of indicators of multifunctionality relied on three sequential stages: identification of jointness, qualitative
assessment of jointness, quantitative assessment of jointness.
Results and discussion
Identification of jointness was carried out at the level of the farm gate for both an EU sample
and a regional case study of Auvergne, France. Full functionalities of SEAMLESS-IF have
been used, including various sets of indicators and their aggregated values (Van Ittersum et
al., 2008). Identification of jointness with a direct use of the integrated framework requires
the possibility of displaying dual values of active environmental constraints and crosselasticities of non-commodity outputs relative to commodity outputs.
Qualitative assessment of jointness examined the relationship between farm income and a
set of environmental and social indicators; both at farm and regional levels (Turpin et al.,
2006). It is worthwhile noting that for the Auvergne case-study it was possible to highlight,
for example, that the marginal effect of environmental subsidies on farm income is not
114
Session A3: Integrated assessment of ecosystem services from agriculture
constant over time (Figure 1). Moreover, the regional landscape is a patchwork of highly
multi-functional farms alongside less multi-functional farms. The impact of policies on
landscape multi-functionality would be in terms of both specific on-farm multi-functionality
and the spatial distribution of multi-functionality across farms.
marginal effect of env. subs. on income 2004
100
80
60
40
20
0
-10
0
-20
10
20
30
40
50
60
70
marginal effect of env. subs. on income 2000
Figure 1. Total marginal effect of environmental subsidies on farm income for FADN farm
groups in Auvergne (the size of the dots represent the share of the total agricultural area for
each group of farms).
Moving to assessing the final stage in our framework, quantitative assessment of jointness, we
extended the approach used to identify jointness at the farm gate level for an EU sample. The
results outlined in this paper serve to illustrate that multi-functionality of agriculture is far
from being negligible.
References
OECD, 2001. Multifunctionality: Towards an analytical framework. Paris.
Turpin, N., et al., 2006. Towards indicators of multifunctionality: Assessing jointness
between commodity and non-commodity outputs. PD2.2.3, SEAMLESS integrated project,
EU 6th Framework Programme (contract no. 010036-2), www.SEAMLESS-IP.org, 62 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
115
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Valuation of ecosystem services and decision making:
Combining a CBA with MCA
C.M. Van der Heide, Frans J. Sijtsma, Petra Van Egmond, Sonja Kruitwagen,
Arjen Van Hinsberg, Frank J. Dietz
Public Issues Division, LEI Wageningen UR,
P.O. Box 29703, 2502 LS The Hague, The Netherlands
Contact: Martijn.vanderheide@wur.nl
Introduction
To evaluate the effects of land use projects, the technique of cost-benefit analysis (CBA) is
widely used. CBA is a policy assessment method that quantifies in monetary terms the value
of all consequences of a programme, policy or other government intervention – usually
denoted by the term ‘project’ – to all members of society. That is, CBA has been defined in
terms of what the gains and losses are to society, and therefore, as cost-benefits analysts
claim, the method can provide an aid to decision-makers in evaluating projects with nonmarket environmental consequences. Nevertheless, although CBA is probably the most
widely practiced method of project appraisal, there are a number of difficulties posed by
applying it to land use projects in agricultural areas. One of the most prominent areas of
difficulty is the monetary valuation of the effects that these projects have on the state of agrobiodiversity in the area under consideration. As a result, adaptations of the method are
suggested in performing an evaluation of major decisions regarding land use projects with
non-market environmental consequences.
In this context, multi-criteria analysis (MCA) can be used for such an adaptation. MCA is a
non-monetary evaluation method, which compares attributes of different project alternatives
by assigning a scoring and weighting systems. It is a flexible methodological approach,
because it can deal with quantitative, qualitative or mixed data and does not impose any
limitation on the number and nature of criteria. The technique usually provides an explicit
relative weighting system for the different criteria The set of weights describe quantitatively
how important each criterion is with respect to the other criteria and reflects the preferences of
those who assign the weights, or can be based on expert judgment.
Methods
The combination of CBA with MCA is called ‘MCCBA’ and appears to be a helpful approach
to deal with the increasingly complex nature of land use projects in agricultural areas. That is,
by using MCCBA, the physical effects of land use projects on agro-biodiversity can be
determined and evaluated in combination with effects which can be fruitfully monetarized.
After quantifying the effects of a project physically, the outcome of a MCCBA is a
combination of aggregated monetary scores (i.e. monetary values of marketed goods) and
aggregated non-monetary scores (which are related to the effects of the project on agrobiodiversity). That is, the outcome of a MCCBA consists of the monetary net social benefits
(which equal the social benefits minus the social costs) complemented with aggregated nonmonetary scores.
This paper focuses on the possibilities of aggregating these physical non-monetary scores
into a single indicator. Developing such an indicator makes it possible to bypass the
controversial field of monetary valuation, which is generally based on eliciting respondents’
preferences for non-marketed goods. The indicator does not capture the final trade-off
between the costs and benefits to society from a given project alternative; neither does it
reveal the consequences of the alternative for societal welfare. Nevertheless, one single
indicator for measuring biodiversity effects improves the information about all the relevant
physical effects of a land use projects, and makes alternative projects better comparable.
116
Session A3: Integrated assessment of ecosystem services from agriculture
Note that, like the conventional CBA, a MCCBA for land use projects is expected to be
undertaken by specialists (and not by just some laymen) who are perfectly capable of
integrating the expert judgment of MCA into the combined framework of MCCBA.
Preliminary outline of the paper
1. The paper starts with a description of the difficulties related to the valuation of ecosystem
services, and how these difficulties impinge upon the performance of a CBA for a policy
design. We present a short review of the international literature, and provide an
introduction to three case studies of projects applying a CBA to evaluate land-use changes
in the Netherlands. Each of these three case studies uses a different approaches to capture
the impacts of land-use changes on ecosystem services.
2. Then we explore the two evaluation methods, CBA and MCA, and we describe their
advantages and disadvantages. This description forms the input for the next section.
3. The third section provides the theoretical basis for the combination of CBA and MCA. This
MCCBA-combination provides the best of both ‘worlds’ (i.e. methods).
4. We then discuss the development of a widely applicable indicator for (agro-)biodiversity
that proves to be useful within the MCCBA-combination. The indicator is developed on
the basis of information that is easily available from Environmental Impact Assessments
(EIAs).
5. In the fifth section we present the application of this (agro-)biodiversity indicator in the
three case studies, which are introduced in section 1.
6. Finally, we conclude our paper with a discussion of the pros and cons of our approach.
Here, we pay particular attention to: (i) the two different ‘worlds’ of CBA and MCA, and
the contribution of the MCCBA to bring both worlds closer together; and (ii) available data
and data demands for further international application of the MCCBA.
117
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Analysing, forecasting and valuating the effects of landscape change
on the ecosystem service of biological pest control
W. Van der Werf, D.A. Landis, M.M. Gardiner, A.C. Costamagna, J.M. Baveco,
F.J.J.A. Bianchi, N.C. Schellhorn, W. Zhang
Wageningen University, Centre for Crop Systems Analysis,
P.O. Box 430, 6700 AK Wageningen, The Netherlands
Contact: wopke.vanderwerf@wur.nl
Introduction
Landscapes are changing rapidly worldwide. Empirical evidence is mounting that these
changes may affect the suppression of pests in agricultural crops. A suite of modelling
techniques is available and in development to extend empirical results to realistic spatial
images of the distribution of the ecosystem service of biological control over agricultural
landscapes and highlight the economic impacts of landscape change on the value of
ecosystem services. These models can play a role in functional landscape design and policy
support.
Natural pest regulation is an important ecosystem service without which the productivity of
plants in natural and managed systems would be severely impaired, and agriculture might
well be impossible. Constanza et al. (1997) estimated the value of this ecosystem service at
more than 400 billion US$ per year at a world-wide scale (but see Pearce (1998) and
Bockstael et al. (2000) for a critique of Constanza’s methods). In agricultural landscapes in
the temperate zone, the natural pest regulation function is often positively related with the
presence of non-crop habitats (Bianchi et al., 2006) and several recent studies have shown
effects of landscape context on biological control (Tscharntke & Brandl, 2004; Tscharntke et
al., 2005). Non-crop habitats may stimulate natural enemy populations by the provision of
(alternative) food sources, hibernation habitat and prey or hosts (Landis et al., 2000). As a
consequence, non-crop habitats often serve as reservoirs of natural enemies, which can
colonize and suppress herbivore populations in arable fields. Interest is increasing in the
design of landscapes that maximize biological control as an ecosystem service, thus helping
make agriculture less dependent on technological inputs (Fiedler et al., 2008). Spatially
explicit simulation models for natural enemy movement and impact in artificial and real
landscapes can elucidate the economic returns to landscape manipulations aiming at higher
levels of the ecosystem service of biological control (Zhang, 2007).
Empirical studies
Convincing evidence that landscape composition affects biological control is provided by
empirical studies with ‘sentinels’. For instance, Bianchi et al. (2008) placed second and third
instar larvae of the diamond back moth, Plutella xylostella, on experimental Brussels sprout
plants in twenty two fields in different landscapes throughout the Netherlands in July 2006.
After two days of exposure, the P. xylostella larvae were recovered, dissected and checked for
the presence of parasitoid eggs. Parasitism rates were positively related with area of forests
within circles of 1, 2 and 10 km around the site, forest edges at a scale of 1 and 2 km and road
verges at a scale of 1 km.
Gardiner et al. (2008) found a significant (P<0.01) positive relationship between land use
diversity within a 1.5 km radius around a site and the biological control of soybean aphid in
soybeans. A strong negative relationship (P<0.001) was found between the corn acreage
around a site and biological aphid control. On average, the proportion aphid reduction on
plants that were exposed to predators was 77% in comparison to those plants that were
shielded from predators and where aphid population growth was essentially exponential.
These proportions reduction translate into small but significant savings in crop protection
118
Session A3: Integrated assessment of ecosystem services from agriculture
costs for growers that use pesticides and vast savings for growers that rely on biological
control alone. An increase in corn acreage for biofuels would carry quantifiable costs due to
loss of ecosystem service value (Landis et al., 2008).
Bio-economic modelling
In silico studies can be made in many different ways. Spatial probability distributions of
natural enemy impact can be estimated from sentinel data (Van der Werf et al., 2008). By
combining the estimated kernel functions with landscape maps, maps of impact across the
landscape can be made (Baveco et al., 2008). A second approach for in silico studies is based
on estimating an initial effect of predators by exclusion (Gardiner et al., 2008) and extending
this effect over a whole growing season using a validated model for pest population growth
(Costamagna et al., 2007). A third approach is based upon modelling the predator & prey
population processes from the bottom up, i.e. on the basis of detailed description of individual
processes (Bianchi & Van der Werf, 2003, 2004; Bianchi et al., 2007).
Process-based landscape simulations, such as those presented in Bianchi & Van der Werf
(2003, 2004) and in Bianchi et al. (2007), allow analysis of what would constitute optimal
landscapes for providing biological control. Such simulations elucidate basic design
characteristics of pest suppressive landscapes but are still subject to large uncertainties,
primarily due to fragmentary knowledge on the movement of pest natural enemies across
landscapes and habitat use by beneficial insects. Economic analyses of the simulations and
simulated or empirical maps offer insights on economically optimal landscape patterns
(Zhang, 2007) and may assist in land use planning and land use policy.
References
Baveco, J.M., et al., 2008. IOBC-WPRS Bull. 34: 21-24.
Bianchi, F.J.J.A. & W. Van der Werf, 2003. Env. Entomol. 32: 1290-1304.
Bianchi, F.J.J.A. & W. Van der Werf, 2004. Ecol. Mod. 171: 177-193.
Bianchi, F.J.J.A., et al., 2006. Proc. of the Royal Soc. B: Biological Sciences 273: 1715-1727.
Bianchi, F.J.J.A., et al., 2007. Landscape Ecol. 22: 1541-1554.
Bianchi, F.J.J.A., et al., 2008. Landscape Ecol. 93: 595-602.
Bockstael, N.E., et al., 2000. Env. Science & Techn. 34: 1384-1389.
Costamagna, A.C., et al., 2007. Agric. and Forest Entomol. 9: 297-305.
Costanza, R., et al., 1997. Nature 387: 253-260.
Fiedler, A.K., et al., 2008. Biological Control 45: 254-271.
Gardiner, M.M., et al., 2008. Ecol. Appl. (in press)
Landis, D.A., et al., 2000. Ann. Rev. of Entomol. 45: 175-201.
Landis, D.A., et al., 2008. Proc. Nat. Acad. Sci. 105: 20552-20557.
Pearce, D., 1998. Environment 40: 23-28.
Tscharntke, T. & R. Brandl, 2004. Ann. Rev. of Entomol. 49: 405-430.
Tscharntke, T., et al., 2005. Annal. Zool. Fennici 42: 421-432.
Van der Werf, W., et al., 2008. IOBC-WPRS Bull. 34: 113-116.
Zhang, W., 2007. Optimal Pest Management in the Presence of Natural Pest Control Service.
http://www.aec.msu.edu/agecon/theses/fulltext/zhang_phd.pdf
119
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Understanding agricultural ecosystem services with ARIES (ARtificial
Intelligence for Ecosystem Services):
Perspectives for assisted policy making
Ferdinando Villa, Marta Ceroni, Kenneth Bagstad, Gary W. Johnson Jr., Sergey Krivov
Ecoinformatics Collaboratory, Gund Institute for Ecological Economics,
Univ. of Vermont, 617 Main Street, Burlington, VT, 05405-0001, USA
Contact: Ferdinando.Villa@uvm.edu
Introduction
Agricultural practices depend on ecosystem services (ES; Farber et al., 2006) and at the same
time produce new ecosystem services. Because the dynamics, spatial and temporal scales, and
range of values of ecosystem services production and usage are not well understood, it is
difficult to locate agriculture within a broader context of ecosystem services-informed policy
making. As a result, agricultural policy making often lacks understanding of the short- and
long-term repercussions of decisions taken on the overall portfolio of valuable services
produced by nature and necessary for human well-being.
ARIES (Ecoinformatics Collaboratory, 2008) is a new methodology and web application
meant to assess ecosystem services and illuminate their values to humans in order to make
environmental decisions easier and more effective. By creating ad-hoc, probabilistic models
of both provision and usage of ES in a region of interest, ARIES helps discover, understand,
and quantify environmental assets, their likely beneficiaries, and what factors influence their
Figure 1. A screenshot from the ARIES toolkit calculating the potential of an area to provide
climate stability through carbon sequestration and storage.
120
Session A3: Integrated assessment of ecosystem services from agriculture
value according to specified needs and priorities. In this contribution, we discuss the use of
ARIES to understand how the consequences of agricultural decision-making may propagate
along the causal chain of the broader spectrum of Ecosystem Services and illustrate
perspectives for integration of ES thinking into agriculture.
Methods
Ecosystem services dynamics can be seen as a generalized source-sink problem, where
ecosystems are the source of benefits that meet the needs of specific human beneficiaries.
Modelling ES in a given spatial and temporal context requires: (1) determining the currencies
of these benefits, such as water, CO2 etc; (2) determining likely surfaces of both provision and
usage relative to the area and time of interest; (3) quantifying the rates of flow of the
correspondent benefits. It is the rate of flow (current or potential) that can be directly related
to the value of the ES, both in abstract and in economic terms.
Most of the many difficulties of modelling ES depend on the high heterogeneity of behaviour
exhibited by the benefits they produce. Among these:
1. Provision and usage happen at entirely independent scales in space and time. Therefore, a
scale-explicit approach needs to be taken, and theoretical instruments that can tackle multiscale systems are lacking.
2. The ‘currency’ of benefit provision is rarely an easily modelled biophysical quantity.
Easier cases include, e.g., CO2: quantification of its exchange from vegetation to
atmosphere may be all that’s needed to assess benefits of carbon sequestration. Things are
much more complex with currencies like sense of identity or avoided risk of flooding.
3. Little clarity exists in the literature about quantifiable definition of ES, their benefits, and
the modalities of their propagation from ecosystem to human beneficiary.
The ARIES methodology is based on explicit conceptualizations (ontologies: Villa et al.,
2009) that lay out first of all a novel vision of ES, based on the breakdown into individual
benefits, each of which is modelled independently, then linked to the others. Domain
ontologies in ARIES result from a large-scale expert consensus. Artificial intelligence
techniques (machine reasoning, pattern recognition) examine source data and extract from the
ontologies models that best represent the situation at hand. ARIES builds ad-hoc, probabilistic
Bayesian Network models (Cowell et al., 1999) that inform the users of the full probability
distribution of the outcomes of their decisions.
The result of an ARIES user session is an environmental asset portfolio that describes in
depth the spatial distribution of benefits produced the area, their potential and realized values,
and the causal relationships that link the values to each other, to their likely beneficiaries, and
to actual or potential policies. Users can enter a scenario explorer module to explore the likely
changes in Ecosystem Service (ES) provision and usage engendered by changed
environmental conditions, consequent to either natural change or their own actions.
We will discuss the ARIES methodology and demonstrate the software toolkit to highlight
its potential in informing ES-centric decision making in agriculture.
References
Cowell, R.G., et al., 1999. Probabilistic Networks and Expert Systems. New York, Springer.
Ecoinformatics Collaboratory, 2008. ARIES: Artificial Intelligence for Ecosystem Services.
White paper: http://ecoinformatics.uvm.edu/aries/uploads/media/ARIES.pdf
Farber, S., et al., 2006. Bioscience 56: 121-133.
Villa, F., et al., 2009. Modelling with knowledge: A review of emerging semantic approaches
to environmental modelling. Environmental Modelling & Software. (in press)
121
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Policy options for safeguarding the provision of public goods and services in
intensive and extensive farming systems
Henk Westhoek, Anne Gerdien Prins, Henk Van Zeijts
Netherlands Environmental Assessment Agency,
P.O. Box 303, 3720 AH Bilthoven, The Netherlands
Contact: Henk.Westhoek@pbl.nl
Introduction
The concept of agriculture providing valuable non-commodity outputs next to agricultural
commodities, has now been more or less generally accepted. Some of these non-commodity
outputs have the characteristics of (positive) externalities or public goods (OECD, 2001;
OECD, 2003). This can include aspects like biodiversity, landscape and water management.
However, as a result of intensification of agriculture in many regions, over the last fifty years,
the automatic delivery of these non-commodity outputs has come under stress. This has, for
example, resulted in strong downward trends for biodiversity in agricultural landscapes, such
as for farmland birds (EEA, 2005). Another potential driver for the decline of biodiversity lies
in the marginalization of agriculture in other regions. The discontinuation of traditional forms
of management, often in the form of extensive grazing systems, might lead to changes in
vegetation, landscape and the accompanying wildlife. Small-scale mosaics of land use and
landscape elements might be lost, resulting in a reduction of biodiversity. The policy
measures needed to safeguard the delivery of non-commodity outputs will probably differ
between those situations where intensification of agriculture is more likely, and those with
risks of marginalization. This paper explores the future of the delivery of non-commodity
output in relation to policies, with special attention to the situation in the EU.
Methods
The need for policies to ensure a continued, or even enhanced delivery of public goods and
services, will be assessed for the current situation and for the future, using the results of the
scenarios study Eururalis (Eickhout & Prins, 2008; Rienks, 2008; Verburg et al., 2008).
Currently, situations do exist where it is clear that policy targets (such as halting biodiversity
loss, by 2010) are unlikely to be met and where policy intervention might be needed. The
exploration of the expected future delivery of public goods by agriculture depends strongly on
the future of agriculture itself. The Eururalis scenarios show different developments in
intensification and marginalization, both drivers of biodiversity loss, due to other
developments in macroeconomic growth, population, technology and the role of the
government. The results of this study will be used to evaluate future changes in the delivery
of public goods to different regions of the EU. The costs and the effects of different policy
options will be assessed.
Results and discussion
The results for the different Eururalis-scenarios show that land abandonment will occur in all
scenarios, but the rate will differ from 2% to 7%, depending on the scenario (Table 1). In the
Continental Market scenario with a continuation of the existing farm support, the rate of land
abandonment because of marginalization is the lowest, whereas the rate is the highest in the
Global Co-operation scenario with less protection. Maps with projected land use show that
there are areas in Europe where land abandonment occurs in all scenarios (Figure 1). If it is
publicly accepted that agricultural practices should continue in these regions − because of the
role of agriculture in the provision of goods and services (such as maintaining traditional
landscapes and biodiversity) − policy action is needed. During the conference, maps will be
presented, showing where these regions are located. In addition, estimations will be given of
122
Session A3: Integrated assessment of ecosystem services from agriculture
farm-support programmes needed to safeguard the continued provision of public goods.
Furthermore, from the results of Eururalis and other studies on the future of European
agriculture it can be derived where intensification of agricultural practices is likely to take
place, with a consequent lower delivery of non-commodity outputs. If it is publicly accepted
that the delivery of these non-commodity outputs will continue at the present (or higher)
level, targeted policy will be needed to ensure this. This will be further elaborated at the
conference.
Table 1. Projected land use in the EU27 under four different Eururalis scenarios.
Agricultural area in 2000 (% of all land)
Agriculture land in 2030 (% of all land)
Agricultural area change (% of
agricultural land)
Agricultural area change (% of all land)
Abandoned areas in 2030 (% of all land)
Global Continental
Global
Economy
Market
Co-operation
47.9
47.9
47.9
43.6
47.0
41.9
–9
–1.9
–12.6
–4.3
4.4
–0.9
2.2
–6.0
6.7
Regional
Communities
47.9
42.8
–10.7
–5.1
5.9
Source: Eururalis 2.0 (Rienks, 2008; Eickhout & Prins, 2008)
Figure 1. Hotspot map of land
abandonment, showing where in
Europe and if so, in how many of
the four Eururalis scenarios. Source:
Rienks (2008).
References
EEA, 2005. Agriculture and environment in EU-15 - the IRENA indicator report 6/2005.
European Environment Agency, Copenhagen.
Eickhout, B. & A.G. Prins, 2008. Eururalis 2.0. Technical background and indicator
documentation. Wageningen University and Research and Netherlands Environmental
Assessment Agency, Bilthoven.
OECD, 2001. Multifunctionality. Towards an analytical framework. OECD, Paris.
OECD, 2003. Multifunctionality, the policy implications. OECD, Paris.
Rienks, W.A. (Ed), 2008. The future of rural Europe: An anthology based on the results of the
Eururalis 2.0 scenario study. Wageningen University and Research and Netherlands
Environmental Assessment Agency, Wageningen.
Verburg, P.H., et al., 2008. Annals of Regional Science 42: 57-77.
123
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Translating spatial policy into future landscape functions
L. Willemen, L. Hein, P.H. Verburg
Environmental Systems Analysis Group/Land Dynamics Group,
Wageningen University, P.O. Box 47,
6700 AA Wageningen, The Netherlands
Contact: Louise.Willemen@wur.nl
Introduction
Most rural regions contain a multitude of other landscape functions in addition to agricultural
production. The spatial variability of these functions depends on the spatial configuration of
the landscape. Spatial policies generally aim to influence the landscape in such a way that the
provision of one or more landscape services is improved. For example by creating zones
around natural areas to improve wildlife habitats, re-allotment of arable land to stimulate
agricultural production, or creating access to natural areas to boost recreational activities.
However, spatial policies will affect each single landscape functions in a different manner.
This paper presents an assessment of the effect of changes in spatial configuration of the
landscape on the provision of landscape services, using the rural Gelderse Vallei region in
The Netherlands as case study area. We analyse the change in supplied landscape services,
and we estimate the change in economic value of these services, as a consequence of growth
trend and a set of policies proposed for the Gelderse Vallei.
Methodology
The overall methodology consists of three steps:
1. Mapping landscape functions before policy implementation
2. Translating spatial policy and trends into future landscape function maps
3. Valuation of (future) landscape services in monetary units, using basic economic
indicators.
We base our study on set of integrated spatial policies which were designed for our study
region (the so-called Reconstruction Act). As baseline we use the situation in the year 2000,
as end point of our assessment we use the year 2015. We include eight different landscape
functions in our analysis, which all relate to one or more specific spatial policies: residential,
intensive livestock, cultural heritage, drinking water, tourism, plant habitat, arable production,
and leisure cycling function.
The first step of our methodology, the mapping of landscape functions, is based on linking
landscape indicators to landscape functions (Willemen et al., 2008). Hereafter spatial policies
and growth trends together with (empirically) quantified relations between landscape
indicators are used to quantify and map the future landscape functions. As a last step,
landscape services are valuated in monetary unites based on economic statistical data for the
region.
Results and discussion
For each landscape function the provided landscape services in 2000 and 2015 are mapped
and their economic value is estimated. As an example, the change in the leisure cycling
function is presented in Figure 1.
In the period 2000–2015, the gross revenues related to the leisure cycling function are
estimated to increase from 6.5 to 19.8 million € yr–1. These numbers are based on the trend on
the average spending per cyclist per day trip. The economic growth of this landscape function
indicates that although the spatial pattern is not expected to change a lot, the total value of the
leisure function in the region is likely do so. The analysis of the intensive livestock function
124
Session A3: Integrated assessment of ecosystem services from agriculture
shows opposite results, its spatial pattern is expected to change drastically in 15 year, but the
economic value for the study area will remain equal over time.
2000
2015
Figure 1. Change in leisure cycling function between 2000 and 2015.
By assessing change in landscape function both in terms of landscape services and in
monetary units, the overall contribution to society of future landscape functions a region can
be explored. Because of the high level of uncertainties in extrapolation methods and data, the
presented study should be seen as a contribution to methodological development rather than a
complete ex-ante evaluation to be used by policymakers.
Reference
Willemen, L., et al., 2008. Landsc. Urban Plann. 88: 34-43.
125
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme A
Methodology for integrated assessment
Session A4: The organization of tools for impact
assessment: Standards or diversity?
Session organizers:
Katharina Helming
ZALF, Müncheberg, Germany
Roger Moore
CEH, Wallingford, UK
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Impact Assessment practice to support
sustainable policy measures in Europe1
Peter De Smedt
EC - DG Research, Marsveldstraat 21, 1050 Brussels, Belgium
Contact: peter.de-smedt@ec.europa.eu
Introduction
Sustainable Development (SD) has become an overall policy objective in Europe. This is
confirmed by the recent issuing of a renewed European SD Strategy in 2006 and a variety of
national and regional SD strategies. Making the concept of SD operational for public policies
raises important challenges in terms of relevancy, accuracy and legitimacy. The growing
interferences between a wide range of global, regional and local developments are increasing
the necessity for new forms of knowledge in order to underpin policies in general, and
sustainable development strategies in particular. Without prospective approaches, including
foresight and a variety of assessment tools, sustainable policy measures have a risk to lack a
solid foundation.
Methods
The purpose of this paper is to analyse how, in Europe, policy instruments have been
developed in order to provide a solid foundation for sustainable policy measures. The analysis
involved the set-up of a scoping study during Spring 2008 to evaluate and compare different
current practices. The selected Impact Assessment (IA) exercises and research policy cases
have been analysed on the basis of a set of criteria developed for this scoping study. The
criteria for evaluating the use of IA tools incorporates: (i) the relevancy, i.e. ‘How closely
connected or appropriate IA of the EC and novel IA policy cases are to the renewed EU
SDS’; (ii) the accuracy, i.e. ‘The quality or state of being exact or precise and correct in all
detail, of being capable of, or successful in reaching the intended target’; and (iii) the
legitimacy, i.e. ‘The extent to which the IA conforms to a given standard (= EU SDS and EC
IA Guidelines)’.
Results and discussion
The empirical evidence of this scoping study confirms a broad variety of successfully
established IA-related initiatives in Europe and the interviewed policymakers and researchers
find the IA approaches legitimate on a conceptual basis. Formal activities and guidance for
IA, for example, are well established within the EC. Both communities however acknowledge
that the full potential of IA tools to support sustainable policy measures in practice is not yet
met. Researchers often find the scope of the current IA exercise too narrow and too sectoral to
support real change in order to anticipate the unsustainable developments. Yet, the
contribution of a formal IA exercise should be evaluated in its full context as being part of a
broader policy process. The framing of the policy question, for example, has most often been
established before the IA exercise was initiated. In addition, research projects often struggle
to bridge the gap between science and the formal policy process. The tools used in any such
process-based application must be simple, based as far as possible on rigorous analysis, while
recognizing explicitly where value judgements are included (Turnpenny, 2008). Moreover,
whilst being simplifications of reality, many scientific models remain so complex that they
are seen rather as black boxes instead of transparent analytical tools. Hence, some of what
modellers see to be the great strengths of modelling tools are felt by non-modellers to be
1
Keynote presentation
128
Session A4: The organization of tools for impact assessment: Standards or diversity?
serious weaknesses (Lotze-Campen, 2008). Consequently, research outcomes do not fully
reach the policymakers.
These findings support that, although IA can provide researchers and policymakers with a
relevant and legitimate common tool, in practice both communities only show a limited
collaboration. Still, the scoping study reveals some evidence of effective close collaboration
between researchers and policymakers. The study also confirms – and this is in contrast with
most scientific literature (such as Weaver & Jordan, 2008) – that these promising experiments
are not only limited to research projects, but can also be found in formal IA experiences
within the EC. This supports the importance of an intensive collaboration where researchers
and policymakers interact on equal basis to support a more integrated and explorative
approach. As Cash and colleagues (2003) also describe, an assessment process is often more
effective if the knowledge being produced and communicated at the interface between science
and policy is perceived by both sides to be credible, e.g., meets scientific standards,
legitimate, e.g., produced by a fair process that reflects the interests of the stakeholders – and
salient, e.g., answers questions seen to be relevant by potential users.
Conclusions
Decision-making can only proceed in a sustainable way if the effects of new policy measures
are explored and understood before they are introduced. Due to the nature and importance of
SD, science and policy have both an important responsibility in this matter. Most practice of
sustainability is based on a set of theories. However, the connection in the other direction, i.e.
between practice and theory, has traditionally been ignored (Gunderson et al., 2007).
Sustainable assessment provides a means where both communities of practice – researchers
and policymakers – can and should collaborate. To become more effective IA practice should
go beyond the traditional supply approach of science. This is needed because the dialogue
linking researchers and policymakers will not happen by itself (Liberatore, 2001). Further
research and policy initiatives should, therefore, include a joint collaboration between
researchers and policymakers to develop a shared understanding of what constitutes a
satisfactory IA (Lee, 2006). This will provide cross-fertilization and learning opportunities
among researchers and policymakers, providing a solid foundation for sustainable policy
measures.
References
Cash, D.W., et al., 2003. Proceedings of the National Academy of Sciences of the United
States of America 100: 8086-8091.
Gunderson, L., et al., 2007. Ecology and Society 12: 40.
Lee, N., 2006. Environmental Impact Assessment Review 26: 57-78.
Liberatore, A., 2001. From science/policy interface to science/policy/society dialogue. In:
Social Sciences for Knowledge and Decision Making. OECD Publishing, Paris.
Lotze-Campen, H., 2008. Int. J. Innovation and Sustainable Development 3: 70-92.
Turnpenny, J., 2008. Int. J. Innovation and Sustainable Development 3: 33-47.
Weaver, P.M. & A. Jordan, 2008. Int. J. Innovation and Sustainable Development 3: 9-32.
129
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A methodology for enhanced flexibility of integrated assessment of policy
impacts in agriculture
F. Ewert, M.K. Van Ittersum, I. Bezlepkina, O. Therond, E. Andersen, H. Belhouchette,
C. Bockstaller, F. Brouwer, T. Heckelei, S. Janssen, B. Jonsson, M.J.R. Knapen,
M. Kuiper, K. Louhichi, J. Alkan Olsson, L. Ruinelli, A.E. Rizzoli, N. Turpin,
J. Wery, J.J.F. Wien, J. Wolf
Institute of Crop Science and Resource Conservation, University of Bonn,
Katzenburgweg 5, D-53115 Bonn, Germany
(Former address: Plant Production Systems, Wageningen University, The Netherlands)
Contact: Frank.ewert@uni-bonn.de
Introduction
Agriculture interrelates with the socio-economic and natural environment and faces
increasingly the problem of managing its multiple functions in a sustainable way. Growing
emphasis is on adequate policies that can support both agriculture and sustainable
development. Integrated assessment and modelling (IAM) can provide insight into the
potential impacts of policy changes. An increasing number of IA models is being developed,
but these are mainly monolithic and are targeted to answer specific problems. Approaches that
allow flexible IA for a range of issues and functions are scarce. Recently, a methodology for
policy support in agriculture has been developed that attempts to overcome some of the
limitations of earlier IA models. The final project version of the proposed framework
(SEAMLESS-IF) will be released shortly and initial results from the testing of the framework
are available. The present paper provides a first evaluation of this methodology to improve
flexibility of IAM in agriculture.
Method
SEAMLESS-IF is a component-based framework for agricultural systems to assess, ex-ante,
agricultural and agri-environmental policies and technologies across a range of scales, from
field–farm to region and European Union, as well as some global interactions. The framework
is based on a software infrastructure that allows a flexible (re-)use and linkage of components.
The components considered include individual models, database and indicators that are linked
depending on the IA problem to be addressed. Usability of SEAMLESS-IF is supported by a
Graphical User Interface (GUI) specifically developed to support interactions with end-users
for all steps of the IA procedure. The methodology is described in more detail in Vvan
Ittersum et al. (2008) and Ewert et al. (2009). Two example applications are used to
demonstrate the flexible application of SEAMLESS-IF. These examples refer to (i) the
impacts on European agriculture of changes in world trade regulations and (ii) regional
impacts of the Nitrate Directive in combination with agro-management changes. The
improved flexibility of SEAMLESS-IF is assessed with respect to its individual framework
components (such as the indicator framework and library, database and models including their
linking) and the phases and steps of the IA procedure (such as system, problem and scenario
description, and the visualization of results).
Results
A summary of the results of the evaluation for the different framework components and the
IA steps is provided in Table 1, whereas detailed information can be obtained from Ewert et
al. (2009). A high level of flexibility has been achieved for most framework components. For
some components, e.g., the indicator framework the flexibility to change this or add new
frameworks is still limited, which may be subject to future development. Importantly, we
show that improving the flexibility of IAM requires flexibility in model linking but also a
130
Session A4: The organization of tools for impact assessment: Standards or diversity?
Table 1. Achieved degree of flexibility in SEAMLESS-IF for selected IA steps and
framework components.
IA step / framework
Characteristics
Degree of flexibility
component
System description
Problem and scenario
description
Indicator framework
Indicator library
Database
Model linking
Visualization of results
Spatial and temporal extent and resolution
Defines policies, farm characteristics,
changes in external conditions and indicators
Considers four classifiers such as level of
organization, environmental and economic
goals, etc.
Organizes indicators according to the
indicator framework characteristics
Database of all model inputs and outputs
including indicators and assessment results
Linking of models available in SEAMLESSIF and considered in the SEAMLESS-IF
ontology
Presentation and evaluation of results in form
of tables, graphs, maps.
Flexible
Flexible
Limited
Very flexible
Very flexible
Very flexible
Flexible
generic set up of all IA steps. This includes the problem and scenario definition, the selection
and specification of indicators and the indicator framework, the structuring of the database,
and the visualization of results. A very important aspect is the flexibility to integrate, select
and link data, models and indicators depending on the application. For instance, the linking of
cropping and farming system models allows consideration of a range of crop successions,
crop management options and their combinations which was not possible in earlier
frameworks. Technical coupling and reusability of model components are greatly improved
through adequate software architecture (with SEAMLESS-IF using OpenMI) and the use of
ontology strongly supports the conceptual consistency of data-model-indicator linkages.
Conclusions
We demonstrate that the proposed framework enhances flexibility in IAM and that it is a good
basis to further improve integrated modelling for policy impact assessment in agriculture. The
presented framework has also limitations which require further development, e.g., the
integration of new models (which requires specific programming expertise) or the
propagation of model uncertainties (which requires a close link to the end-users). Also, the
scientific basis for linking models across disciplines and scales is still weak and needs specific
attention in future research. Importantly, enhancing flexibility can have negative trade-offs
affecting model performance, quality of simulation outcomes and framework understanding
and transparency. Accordingly, finding the right balance between specific and generic model
solutions is crucially important when trying to improve flexibility in IAM.
References
Ewert, F., et al., 2008. Environmental Science & Policy. (in review)
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
131
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Choose complexity: How to bundle knowledge in
Integrated Assessment Tools?
Comparing implicit knowledge integration in SENSOR and PLUREL
K. Fricke, K. Helming, S. Sieber
Leibniz-Centre for Agricultural Landscape Research e.V., Eberswalder Str. 84,
15374 Müncheberg, Germany
Contact: katharina.fricke@zalf.de
Introduction
In the frame of sustainability impact assessment, tools for decision support are required that
facilitate the ex-ante assessment of land use and planning decisions on the multifunctional
performance of rural land use as well as of rural-urban linkages. Here, the FP6-founded
project SENSOR is developing Sustainability Impact Assessment Tools (SIAT) to analyse
European policy options related to rural land use, while the project PLUREL aims inter alia at
developing a Sustainability Impact Assessment Tool Kit for Rural-Urban Regions (SIATRUR) to display policy impacts on rural-urban linkages at European level. Both Integrated
Projects (IPs) face the challenge of integrating interdisciplinary knowledge such as land use,
socio-economics, environmental economics and landscape research.
Methods
This paper focuses on the knowledge integration in both tools while outlining end user
requirements towards its domain structure, design and architecture. Taking into account the
differences in the conceptual modelling approaches that reflect the altering prerequisites of
the two projects, both tools follow a comparable approach. SIAT is a scenario-driven metamodel based on response functions describing relations between (1) policy options and land
use changes and (2) land use changes and sustainability indicators. These response functions
are derived from existing economic and environmental models as well as expert-driven
knowledge rules at national and regional level. Building upon this analytical design of the
SIAT, the SIAT-RUR (1) anticipates consequences of selected global driving forces and
European policies on rural-urban land use types and (2) analyses how they affect social,
economic and environmental services and functionalities of the rural-urban regions. The latter
tool will support policymakers, scientists and interested lay persons in analysing urbanization
processes and trends in European regions, and will provide especially policymakers with
support in the development of strategies for a better guidance of these processes.
A systematic approach is developed to handle, focus and if necessary reduce the
complexity of the interdisciplinary knowledge integrated into these tools in order to improve
their performance and acceptance and thus their usability for discussion support (compare
Figure 1). Thus, the research question focuses on the level of explicit and implicit knowledge
of integrated Impact Assessment and its application in tools. This approach is derived from
the exemplary frameworks of both IPs and tested towards its general transferability
concerning a thematic shift in the initial question of these tools. Hereby, special regard is paid
to the concepts of handling complexity and their appliance in the conception phase of decision
support tools as well as the end user requirements towards these tools.
Here, the focus is to develop an approach that minimizes the complexity by both reducing
input variables and nevertheless handling complexity of the integration of interdisciplinary
knowledge fields. This has to be regarded already in the development process of meta-models
for the sustainability assessment of land use changes and relevant land use policies in
European regions. Merely then tools can facilitate end users to select the suitable tool for their
particular needs are deduced. First of all, the developed approach focuses on retrieving the
requirements on the content and the needed performance of the tools. Subsequently, the
132
Session A4: The organization of tools for impact assessment: Standards or diversity?
complexity of a broad variety of different input variables that are necessary to assess the
impacts consistent to different spatial scales is reduced. While maintaining the overall
transparency, the inherent knowledge of the decision making process is incorporated as
implicit knowledge. Meanwhile, the differentiating requirements of end users, stakeholders
and scientists accompany the development of these Impact Assessment Tools in participatory
processes.
Figure 1. Central thread of how to bundle complex information in SIA-Tools for end users.
Results and discussion
The paper analyses the conceptual approaches towards the SIAT and the SIAT-RUR taking
into account the different requirements concerning the content and the functionality of the
tools. A reliable and satisfying way of knowledge integration reduces the complexity of the
interdisciplinary cognition. Accordingly this ensures a distinguished user-friendliness of a
SIA-Tool while considering requirements and options for meta-information and structures.
This methodology to integrate interdisciplinary knowledge is regarded as a systemic approach
to handle the complexity in such tools and to enable the transferability of general tool
concepts. The implementation of an overall and transferable concept of knowledge integration
in tools such as SIAT and SIAT-RUR is seen as useful concept of handling the complexity for
stakeholder and end user discussions and therefore support these interest groups in choosing
the right tool.
References
Helming, K., et al., 2008. In: Eds K. Helming, et al., Sustainability impact assessment of land
use changes. Springer, pp. 77-105.
Sieber, S., et al., 2008. In: Eds K. Helming, et al., Sustainability impact assessment of land
use changes. Springer, pp. 107-128.
133
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Swiss Agricultural Life Cycle Assessment (SALCA):
An integrated environmental assessment concept for agriculture
G. Gaillard, T. Nemecek
Agroscope Reckenholz-Tänikon Research Station ART, CH-8046 Zurich, Switzerland
Contact: gerard.gaillard@art.admin.ch
Objectives and rationale
The need for an environmental assessment of agricultural activities has been dramatically
increased during last years. The most important stakeholders in agriculture (farmers,
authorities, food and energy industries, NGO’s) require different environmental information
and tools for their own purposes and according to their own context (farm management tool,
sector monitoring, food chain management, citizen and consumer information). Since often
the needs concern the same objects (e.g., a specific crop) and require a common background
(system analysis, data), it becomes clear for the practitioner that he has to look for synergies
between these different applications of agricultural environmental know-how. Instead of
developing a large number of tools in order to cover all types of applications (a tool for
national policy, a tool for food industry, a tool for farm management and so on), respectively
instead of separately developing modules of environmental assessment (here a database, there
a method for biodiversity, and other where an environmental communication tool), a general
concept is required integrating method, data and tools in order to deal with the increasing
demand in an efficient way. Considering this challenge, we developed the SALCA (Swiss
Agricultural Life Cycle Assessment) integrated concept with the objective of performing a
comprehensive environmental assessment for a large variety of agricultural systems.
SALCA
SALCA consists of the following components:
• Database for life cycle inventories for agriculture.
The database is developed in close co-operation with ecoinvent (especially regarding data
format and quality criteria), which makes the agricultural inventory data compatible with
all the other economic sectors. In this way, we developed a consistent set of Swiss,
European and American inventory data which is easily available for all interested people
(Nemecek & Kägi, 2007). Regarding agricultural production data, the close collaboration
with FADN (Farm Accounting Data Network) services supported by a network of 200
farms allows us to grant the required representativeness for Switzerland (Alig et al., 2008).
• Models for the calculation of direct emissions from field and farm. The non-linearity
inherent to agricultural processes requires for a credible environmental assessment the
consideration of specific models for direct emissions. Together with environmental
scientists in their respective discipline, we developed or adapted models for the most
current emissions encountered in agriculture and for some of them a separate
documentation is available online: Nitrate (Richner et al., 2006), phosphorus (Prasuhn,
2006), heavy metals (Freiermuth, 2006).
• A selection of impact assessment methods. The impact assessment methods developed by
the scientific community mostly are not focused on agriculture. A major support is to
assess and select a set of impact assessment methods appropriate for agricultural
applications. As an example a systematic comparison of 7 methods proposed in Europe for
assessing the eco- and human toxicological impact was performed with a focus on the
pesticide active ingredients (Kägi et al., 2008).
• Methods for the assessment of impacts on biodiversity and soil quality. There is enough
evidence that the restriction to the usual impact categories (like energy, greenhouse
warming potential or eutrophication) does not enable a correct consideration of all
134
Session A4: The organization of tools for impact assessment: Standards or diversity?
environmental impacts of agricultural activities. For these reasons and for a Mid-European
context of application, we developed two methods in order to cope with the environmental
impacts commonly summarized under the expression ‘land use’: SALCA-biodiversity
(Jeanneret et al., 2006) and SALCA soil quality (Oberholzer et al., 2006).
• Calculation tools for agricultural systems (farm, annual crop, perennial crop). The
repetition of similar cases together with a coherent system analysis required for consistent
comparisons (e.g., at crop rotation level, between farm types) the development of
calculation tools for some archetypical cases (such as farm, crop) based on a commercial
LCA (Life Cycle Assessment) software. Furthermore, an overall assessment requires the
integration of tools especially developed for direct emissions and impact assessment in a
consistent way allowing a precise calculation under consideration of complex interfaces
(Nemecek et al., 2009).
• Interpretation schemes for agricultural LCA. Especially for stakeholder not currently
dealing with environmental information like delivered by LCA, it is central to integrate
environmental results delivered by the tools in such a way that they can be used by the
stakeholder. An example of it is the interpretation and communication concept for
environmental farm management (Alig et al., 2008).
Analysis and conclusion
The close link of all these components proved to be efficient in many ways and is being tested
with success in different countries (Switzerland of course, but also France, Japan and several
European countries in the frame of European research projects). Major advantages are:
• Knowledge developed for one specific application and user is available in short delays for
all other types of application and users
• Consistency is granted over all applications of SALCA according to a coherent set of
quality criteria
• Development costs are minimized due to a modular re-use of the SALCA components.
Based on the SALCA experience, we are convinced that efficient environmental tools require
such an integrated approach to meet the challenges of satisfying the very different needs of
agricultural stakeholder – farmers, industry, state and consumers – with the required quality.
References
Alig, M., et al., 2008. LCM in agriculture: How to communicate? Proceedings of the
European SETAC conference.
Freiermuth, R., 2006. Heavy metals. http://www.art.admin.ch/themen/00617/00785/index.
html?lang=en
Jeanneret, P., et al., 2006. http://www.art.admin.ch/themen/00617/ 00785/index.html?lang=en
Kägi, T., et al., 2008. Multicriteria comparison of RA and LCA toxicity methods with focus
on pesticide application strategies, submitted to LCAFood08 Conference Proceedings.
Nemecek, T. & T. Kägi, 2007. Life Cycle Inventories of Agricultural Production Systems.
Final report ecoinvent V2.0 No. 15. Agroscope Reckenholz-Taenikon Research Station
ART, Swiss Centre for Life Cycle Inventories, Zurich and Dübendorf, CH, online at:
www.ecoinvent.ch.
Nemecek, T., et al., 2009. AgSAP Conference, these proceedings.
Oberholzer, H.R., et al., 2006. http://www.art.admin.ch/themen/00617/00785/index.html?
lang=en
Prasuhn,V., 2006. http://www.art.admin.ch/themen/00617/00785/index.html? lang=en
Richner, W., et al., 2006. http://www.art.admin.ch/themen/00617/00785/index.html? lang=en
135
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Linking agricultural pollution to receiving waters using OpenMI
L. Khanh Hoang, S. Van Griensven, J. Cools, L. Benedetti, A. Mynett
Department of Hydroinformatics and Knowledge Management, UNESCO-IHE,
P.O. Box 3015, 2601 DA Delft, The Netherlands
Contact: hnklinh@yahoo.com
Introduction
The EU Water Framework Directive (WFD) has introduced a new approach in water
resources protection in which there is a change from focusing on the control of point sources
of pollution (emission-based regulations) to integrated pollution prevention at river basin
level and setting water quality objectives for the receiving water (immission-based
regulations). This new policy requires the integration of all water quality issues, related to
both point and diffuse pollution sources, at river basin scale. With this new approach, it is
necessary to integrate catchment modelling (which aims at estimating the flows of water and
pollutants released from a draining catchment into the receiving water) with modelling of
water quality processes in the receiving water. Due to the past and present efforts in waste
water treatment for industries and households, agricultural pollution is becoming the major
concern, being often to the main cause of nutrification and eutrification of water bodies.
Methodology
This study aims at integrated catchment modelling using the Soil and Water Assessment Tool
(SWAT; Arnold et al., 2002) by linking it to several river model software (WEST®, SOBEK)
of different complexity, using the Open-MI model integration framework (Gregerson et al.,
2007). SWAT was made Open-MI compliant by converting the model engine into a
component (Getnet, 2008). A comparison of the different river modelling software is done
based on performance before and after calibration, on flexibility and applicability.
The chosen case study is the Grote Nete river basin (Belgium) in which there are pollution
problems from agriculture and from households and industries which directly discharge
wastewater to the river. Several scenarios with the aim at improving the water quality in the
Grote Nete river are evaluated with the models. A comparison is made by comparing the
curve of pollutant source reduction versus receiving water quality obtained for each river
model.
Discussion and results
There is a large difference in concepts and process description for river water quality
modelling (Table 1). A proper selection is therefore required.
Model integration through software such as Open-MI (Figure 1) is a good development
and allows for a case and problem dependent choice of the most appropriate software for each
subsystems. While Open-MI helps to tackle the technical aspects of water quality models, this
is not the most critical problem. Differences in concepts, description of variables, differences
in discretization between models are more difficult to overcome.
Table 1. comparison of the river water quality models.
SWAT
WEST
Hydraulic routing
Variable storage
Continuous stirred
Muskingum method
tank reactors in series
Water quality
Qual2E
RWQM1
OpenMI integration Water quantity and
Ongoing
quality done
136
SOBEK
1D Saint Venant
equation
DELWAQ
Water quantity done,
water quality ongoing
Session A4: The organization of tools for impact assessment: Standards or diversity?
Figure
1.
OpenMI
configuration editor link
property from SWAT to
SOBEK-RE model.
The integrated model between SWAT and WEST is built by linking the catchment modelling
SWAT and river model using WEST. SWAT provides result of flow and pollutant from
draining catchment as the input for river model in WEST which then uses continuous stirred
Tank Reactors (CSTRs) in series approach as routing method. Figure 2 shows the comparison
of flow simulation between integrated model SWAT_WEST and SWAT which uses
Muskingum routing method.
Comparison of flow simulation between SWAT and SWAT-WEST
14,000
WEST
12,000
SWAT
Flow (m3/s)
10,000
8,000
6,000
4,000
2,000
0,000
0
50
100
150
200
250
300
350
Day
Figure 2. Comparison of flow simulation between SWAT and SWAT-WEST.
References
Arnold, J.G., et al., 2002. Soil and water assessment tool, theoretical documentation,
Grassland, soil and water research laboratory - Agricultural Research Service, Texas, US.
Getnet, D.B., 2008. Integrated Sediment Transport Modelling using Open-MI (SWAT and
SOBEK-RE) for the Blue Nile river basin. MSc thesis Water Science and Technology, IHE
Institute for Water Education.
Gregerson, J.B., et al., 2007. Journal of Hydroinformatics 9: 175-191.
137
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A methodological framework for sustainability impact assessment of land
use policies in developing countries:
Re-using and complementing approaches
P. Reidsma, H. König, S. Feng, I. Bezlepkina, H. Van Keulen,
M.K. Van Ittersum, F. Brouwer
Plant Production Systems, Wageningen University, P.O. Box 430,
6700 AK Wageningen, The Netherlands
Contact: pytrik.reidsma@wur.nl
Introduction
Enhancement of sustainable development is an important issue in developing countries. Land
use patterns and land use changes are considered critical to sustainable development. To
improve systematic knowledge on the impact of land use policies on sustainable development
in developing countries, a methodological framework is developed in the LUPIS project
(www.lupis.eu). The framework builds on knowledge and tools developed within a European
context in two integrated projects SENSOR (www.sensor-ip.org; Helming et al., 2008) and
SEAMLESS (www.seamless-ip.org; Van Ittersum et al., 2008). The SENSOR methodology
develops ex-ante impact assessment tools at regional scale for EU policies related to land use,
with a focus on cross-sectoral trade-offs and sustainability side-effects. The SEAMLESS
methodology concentrates on the agricultural sector and targets at assessing agricultural and
environmental policies and technological innovations at multiple scales.
The goal of this paper is to present the LUPIS framework through an illustration of the
case study in China, where the land use problem considered is the water pollution in Taihu
Lake Basin. Re-using the methodologies developed in the SENSOR and SEAMLESS projects
allows realization of the dual objectives of the LUPIS project: (1) test the applicability of
European approaches in the context of developing countries and (2) analyse, for several case
study countries (including China, India, Indonesia, Mali, Kenya, Tunisia and Brazil), the
impacts of land use policies on a specific land use problem and sustainable development of
the selected region.
Methodological framework
The LUPIS methodological framework enables the complementary use of the methodologies
from SENSOR and SEAMLESS, and of additional tools, for assessment of land use policies
and sustainable development in developing countries (Reidsma et al., 2008). Since the interest
is on ex-ante assessment, modelling tools that can be used for simulations are core elements
of the LUPIS framework. Models can be complex mathematical models, with high data
requirements, simple models that are easy to parameterize, and/or knowledge rules as used in
SENSOR. To structure the analysis, the methodological framework for sustainability impact
assessment (SIA) considers three phases, like SEAMLESS (Figure 1). The first phase, premodelling, deals with problem analysis and the selection of indicators. In the second phase,
modelling, the impacts of policies on indicators are assessed. The selected set of assessment
tools can vary for the various case study sites, in dependence of data and model availability.
In the last phase, post-modelling, the impacts of policies on sustainable development are
evaluated, through an assessment and aggregation of indicators.
As part of the LUPIS framework, a common indicator framework is developed that guides
the selection of environmental, economic, social and institutional indicators, and specifies
procedures to aggregate single indicators. Following the SENSOR methodology, Land Use
Functions are used as a processed illustration of aggregation of single indicators that are
tailored to assess goods and services associated with land use (Pérez-Soba et al., 2008). Both,
138
Session A4: The organization of tools for impact assessment: Standards or diversity?
LUFs and single indicators are evaluated for sustainable development in the post-modelling
phase.
Pre-modelling: Problem definition
System
description
Case study
definition
Indicator
selection
Scenario
Description
Problem
Causal chains
SD targets
Current situation
Context
Land use sectors
Land Use Functions
Baseline scenario
Policies
Spatial scale
Indicators
Policy options
Adaptation and/or
development of
assessment tools
Application of
assessment tools
SENSOR
SEAMLESS
other tools
Models
Knowledge rules
Response functions
Define experiments
Parameterization
Simulation
Data
Stakeholders
Modelling: Assessing impact of policy on indicators
Review and selection
of assessment tools
Post-modelling: Assessing impact of policy on Sustainable Development
Indicator integration
and formation
Sustainable Development
evaluation
Single indicators
Multi-criteria analysis
Composite indicators
T rade-off analysis
Land Use Functions
Institutional assessment
Visualisation/
documentation/
communication
Tables, Graphs,
Spider-diagrams,
Maps, Text
Figure 1. Methodological framework for SIA in LUPIS case studies.
Discussion and conclusion
A first evaluation of the framework for the various LUPIS case study regions has indicated
that it is impossible to indiscriminately re-use the modelling frameworks that are being
developed in SEAMLESS and SENSOR within LUPIS. In China, the main focus is on the
development of bio-economic models at farm and regional scale to assess the impact of
policies and technological innovations in the agricultural sector on reducing water pollution
and sustainable development at large. The generic model developed in SEAMLESS (Louhichi
et al., 2007) is used as a basis, but needs adaptation for case study specific issues. These can
partly be based on other models developed for application in South-east Asia. Other drivers of
water pollution, such as industry and domestic sewage, and related indicators will be assessed
with knowledge rules to complement the SIA.
Hence, the SEAMLESS and SENSOR models and other tools are viewed as a set of
methodologies, from which can be selected for application in non-EU contexts, and which can
be complemented by other approaches. As the framework is generic and flexible, the main
procedures as depicted in Figure 1, are considered to match all LUPIS case studies.
References
Helming, K., et al. (Eds), 2008. Sustainability Impact Assessment of Land Use Changes.
Springer, Berlin.
Louhichi, K., et al., 2007. PD 3.3.11, SEAMLESS Integrated project.
Pérez-Soba, M., et al., 2008. In: Sustainability impact assessment of land use changes, Ed. K.
Helming, et al., Springer, Berlin, pp. 375-404.
Reidsma, P., et al., 2008. D4.1, LUPIS Integrated Project, Wageningen.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
139
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
From models to indicators: Ontology as a knowledge representation system
O. Therond, N. Turpin, S. Janssen, I.N. Athanasiadis, M.J.R. Knapen, C. Bockstaller,
J. Alkan Olsson, F. Ewert, I. Bezlepkina
INRA, UMR 1248 AGIR, F-31326 Castanet Tolosan, France
Contact: Olivier.Therond@toulouse.inra.fr
Introduction
Evaluation of policies requires targeted and systematic approaches to assess their economical,
environmental and social impacts at a wide range of scales. Integrated Assessment and
Modelling (IAM) (Parker et al., 2002) has been proposed as a method to ex-ante assess
indicators for such policy impact assessment. In IAM these indicators have to be linked to
model outputs and to user-relevant concepts, like scale and indicator framework and
dimension.
Among the various IAM tools currently developed, SEAMLESS-IF has been built as a
joint effort of thirty partners and their researchers (ca 150), each of them providing specific
knowledge in his own discipline (Van Ittersum et al., 2008). In the course of such tool
development crossbreeding between different scientific knowledge is necessary as each
discipline uses with its own way notions and concepts that sound similar, like resource
efficiency, profitability, productivity, environmental soundness or social viability. For IAM
projects, no explicit procedure was found in literature to homogenize meaning across
concepts, scales and disciplines. To solve this problem of multitude of meanings for
indicators and the related concepts, the SEAMLESS project developed an indicator ontology
i.e. a finite list of concepts and the relationships between these concepts. This indicator
ontology shared by all scientists from various disciplines and backgrounds working on an
integration task, serves as a knowledge-level specification of the joint conceptualization (for
more information on the ontology concept and objectives see Janssen et al., 2009) and
enabled implementing indicators in the IAM platform. The ontology supports and facilitates
the communication of complex concepts needed to define, present, compute and displays
social, economical and environmental indicators at the wide range of scales investigated by
SEAMLESS-IF. This paper describes the ontology for indicators as developed in the
SEAMLESS project and the process of ontology development.
Methods
Impact indicators in SEAMLESS are primarily based on modelling chain outputs. The
development of such indicators necessitated a strong iterative and structured interaction
between indicators, database, models and software developers as well as tool evaluators. The
indicator development and implementation work started from a literature study on
sustainability indicators and frameworks, evolved through the development of the indicator
ontology, the definition of indicators that can be computed by the SEAMLESS modelling
chains and the needed scaling procedures and other post-modelling processing. It resulted in
implementing the SEAMLESS indicator tools offering all the services necessary to
manipulate impact indicators for policy impact assessment focusing on the agricultural
systems. The developed work steps were accompanied by a cyclical evaluation-improvement
procedure involving most of the different type of developers.
Results and discussion
The specific SEAMLESS indicator ontology allows:
- To define ‘Indicator Group’ i.e. an indicator impact-oriented family grouping together a set
of indicators providing information on the same impact but at different scales (Bockstaller
et al., 2009). This indicator group allows highlighting links between indicators presenting
140
Session A4: The organization of tools for impact assessment: Standards or diversity?
-
-
-
different spatial and/or temporal extents but providing information on the same process
(e.g., the nitrate leaching group brings together the Nitrate leaching in kg N-NO3 ha–1 y–1 at
field and farm level and the share of the area with nitrate leaching over a given threshold
computed at landscape or regional level).
For each indicator group to have a specific link to an indicator group fact sheet describing
all the characteristics of the indicators (purpose, impact, described processes, scales,
detailed description of calculation, information needed for interpretation, possibilities of
up-scaling/aggregation and evaluation of the indicator).
To establish the location within the so called Goal Oriented Framework (GOF) of each
indicator (Alkan Olsson et al., 2009). The GOF aims at guiding the user in the selection of
indicators, preventing him from focusing on a single issue, and facilitating the
communication between researchers and policy experts related to different sustainability
dimensions.
To define thresholds as reference value to interpret the indicator value.
To highlight indicator tradeoff i.e. a relation of antagonism between different indicators.
To define the model output used to compute the indicator and by this way the functional
link between indicator and the modelling chains.
To identify the list of intermediate variable of the modelling chain necessary to interpret
and understand the causality chain lying behind the indicator values.
To describe the spatial and temporal resolution and extent of each indicator (i.e the
investigated process scales and the policy decision scales).
Figure 1. the concepts
IndicatorGroup, Indicator
and Model (large ellipses),
with some of the attributes
(small ellipses) and the
relationships between them
(arrows with name).
There are many indicators databases around the world1 where indicator ontologies include
metadata and semantic interoperability between the various indicators. The main specificity of
the SEAMLESS indicator ontology is to allow implementing sustainability indicators in an
IAM platform. This ontology structures all the knowledge and data in relation to indicators, so
that an indicator, in a transparent and explicit way, can be linked to the relevant modelling
chain outputs, be selected, assessed and displayed using the specific IAM tools and be
enriched with all the user-oriented information necessary to perform well qualified policy
impact assessment. The indicator ontology is a separate product of the SEAMLESS research
project and can now be reused in other research projects working with indicators and models.
Reference:
Alkan Olsson, J., et al., 2009. Environ. Sci. and Pol. (in review)
Bockstaller, C., et al., 2009. AgSAP Conference 2009, these proceedings.
Janssen, S., et al., 2009. Environmental Modelling and Software. (in review)
Parker, P., et al., 2002. Env. Mod. & Soft. 17(3): 209-217.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
1
For a recent survey and an operationalized ontology, see http://www.sdi.gov/
141
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme B
State-of-the-art of components for integrated
systems
Session B1: Large scale integrated policy assessment:
Tools and applications
Session organizer:
Thomas Heckelei
University of Bonn, Germany
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Impacts of EU biofuels directives on global markets and EU environmental
quality: An integrated PE, global CGE analysis1
Wolfgang Britz, Thomas Hertel
Nussallee 21, 53229 Bonn, Germany
Contact: Wolfgang.Britz@ilr.uni-bonn.de
Overview
Both the EU and the US have implemented biofuel mandates with the aim of reducing their
dependency on fossil fuels while simultaneously abating Green House Gas (GHG) emissions.
However, recent studies (Searchinger et al., 2008; Fargione et al., 2008) have questioned the
value of these mandates for reducing global warming. These studies have emphasized the
important role of indirect Land Use Change (iLUC) in increasing GHG emissions associated
with first generation biofuels. The basic idea is that the diversion of agricultural products into
the energy economy will induce crop land conversion, as additional land is brought into
production in order to meet the rather price-inelastic global demand for food. If these
converted lands are high in carbon content such as tropical forests or peat bogs, then the net
impact of the biofuels programme on GHG emissions may be adverse. Indeed, in the case of
corn ethanol produced in the US, Searchinger et al. (2008) suggest that GHG emissions could
even double, when compared to the continued use of petroleum products. While their
analytical framework is relatively simple, these papers make a compelling case for
considering iLUC in any assessment of the environmental impacts of biofuel mandates.
Accordingly, these mandates have now included provisions restricting the renewable fuel
standards to biofuels which meet minimum GHG reduction standards. For example, the 2007
US Energy Act requires that incremental corn ethanol contribute to at least a 20% reduction in
GHG emissions, relative to petroleum products. This, in turn, has generated a great demand
for studies of iLUC and biofuels. The California Air Resources Board has similar guidelines
for biofuels to qualify for the Low Carbon Fuel Standard.
The studies of iLUC have included both partial equilibrium (OECD, 2008; Tokgoz et al.,
2008) and general equilibrium analyses (Keeney & Hertel, 2008; Banse et al., 2008). Each of
these approaches has its strengths and limitations. The partial equilibrium studies typically
offer greater commodity detail, while the general equilibrium studies are better at capturing
the linkages between the farm and non-farm sectors – in particular the energy sectors in the
case of biofuels, and they also include explicit competition for land between agricultural and
other uses. In the present paper, we develop a methodology for linking partial with general
equilibrium models in order to capitalize on the strengths of each approach. We focus particularly on the impacts of EU biofuels programmes on global land use and GHG emissions.
Methods
Biofuels mandates provoke simultaneous adjustments in the markets for both fossil fuels and
agricultural raw products, and the ensuing effects will depend inter alia on the interplay with
policy instruments in these markets. Subsidizing biofuel processing or production of biofuel
feedstocks, to give an example, will reduce transport fuel prices and thus stimulate energy
demand, whereas obligatory blending could increase fuel prices and depress demand. The
effects on the overall economy thus depend on the method for implementation of the biofuel
policy, which is why CGE models have been used to analyse this issue. The CGE model
which we will build upon in this paper is GTAP (Hertel, 1997). In particular, we utilize the
biofuels version of that model (Birur et al., 2008), augmented with land use by AgroEcological Zones (Hertel et al., 2009) and by-products (Taheripour et al., 2008). This model
1
Keynote presentation
144
Session B1: Large scale integrated policy assessment: Tools and applications
has been widely used to establish the links between energy policies and global land use.
However, its capability to model detailed agricultural impacts in the EU is limited – both due
to commodity and regional aggregation (the GTAP model only includes national production
functions – albeit augmented by sub-national Agro-Ecological Zones).
In contrast to their general equilibrium counterparts, agricultural partial equilibrium
models profit from their specialized nature by offering more detail regarding dis-aggregation
in space and products, as well as improved treatment of domestic agricultural policies. Often,
their supply response is also judged more robust as their results had been validated over time
years in different policy relevant applications. We hence enrich the general equilibrium,
global analysis by integrating the CAPRI model of EU agriculture (Britz et al., 2007), into the
analysis. In order to achieve a mutually consistent supply behavior of GTAP and CAPRI; an
aggregate, agricultural supply response for CAPRI is derived from the detailed, regional
programming models of CAPRI. This multi-product crop revenue function provides the basis
for communication between the two models and it is explicitly incorporated into GTAP.
Specifically, we utilize a normalized quadratic, revenue function, calibrated to the
aggregated EU-wide Hessian obtained from the CAPRI model – aggregated from the 250
region-NUTS level to the level of the 18 GTAP AEZs associated with the EU-27. The
modified GTAP model is then used to analyse the impacts of the (recently revised) 2015 EU
biofuel mandates on global trade, production and global land use. Secondly, the changes in
equilibrium agricultural product prices simulated by GTAP are subsequently used as a shock
to CAPRI. CAPRI then simulates the resulting regional changes in farming practices driving
by the price changes which in turn allow calculation of different environmental indicators,
including changes to Global Warming emissions.
Anticipated results
Existing studies and analysis with CAPRI show that around 50% of the EU’s 10% biofuel
mandates would be covered by non-EU production, either directly by imports of biofuels or
by changes in EU net-trade of biofuel feedstocks or their substitutes. The EU’s changed nettrade status in these markets will provoke land-use changes in other regions of the world, and,
as global demand increases, this will raise food prices, worldwide. Inside the EU, the shifts
towards wheat replacing coarse grains, increasing rape seed shares and the increase in
agricultural raw product prices lead to a higher intensity in agricultural practices, e.g., to
higher mineral fertilizer applications resulting in a higher nutrient load. In addition, we expect
fallow land and extensive fodder production on arable land to be reduced as a result of these
mandates. Preliminary results from GTAP suggest that land cover change induced by EU
policies may be particularly important in Brazil, Eastern Europe, Canada and Africa. Land
cover changes will be disaggregated into net changes in cropland, grazing land and forestry
and associated carbon fluxes will be reported based on the figures reported in Searchinger et
al. (2008).
Selected references
Banse, M., et al., 2007. European Review of Agricultural Economics 35(2): 117-141.
Britz, W., et al., (Eds) 2007. Description of the CAPRI Modeling System, Final report of the
CAPRI-Dynaspat Project. Inst. for Food and Resource Economics, Univ. of Bonn.
Fargione, J., et al., 2008, Land clearing and the biofuel carbon debt. Science 319 (5867).
Hertel, T.W., et al., 2009. Chapter 6 in: Eds T.W. Hertel, et al., Economic Analysis of Land
Use in Global Climate Change Policy. Routledge. (forthcoming)
Keeney, R. & T.W. Hertel, 2008. GTAP Working Paper No. 52.
OECD, 2008. Biofuel support policies: An economic assessment. OECD, Paris.
Searchinger, T., et al., 2008. Science 319 (5867): 1238-1240.
Tokgoz, S., et al., 2007. http://www.card.iastate.edu/publications/synopsis.aspx?id=1050
145
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Modelling global N and P surface balances: Implementation of the
millennium ecosystem assessment scenarios
A.F. Bouwman, A.H.W. Beusen, G. Billen
Netherlands Environmental Assessment Agency,
P.O. Box 303, 3720 MA Bilthoven, The Netherlands
Contact: Lex.bouwman@pbl.nl
Introduction
Global population, food production, and energy consumption have increased approximately
2.5-, 3-, and 5-fold, respectively, during the past five decades (Grübler et al., 1995; FAO,
2008). Through activities such as fertilizer use, fossil fuel consumption, and the cultivation of
leguminous crops, humans have more than doubled the rate at which biologically available
nitrogen (N) enters the terrestrial biosphere compared to pre-industrial levels (Galloway et al.,
2004). The global phosphorus (P) cycle has also been greatly altered by human activity. P
production from rock phosphate and subsequent use as fertilizer, detergent, animal feed
supplement and other technical uses has more than doubled P inputs to the environment over
natural, background P from weathering (Mackenzie et al., 1998; United States Geological
Survey, 2008).
Method
This paper describes the implementation of the four Millennium Ecosystem Assessment (MA)
scenarios (Alcamo et al., 2006) for 2000–2050 to develop spatially explicit global N and P
surface balances for agriculture and natural ecosystems. We implemented the data from the
original MA work covering the period 2000–2050 in the IMAGE model (Bouwman et al.,
2006) to calculate the N and P surface balances and surpluses for nonpoint sources developed
for the MA scenarios. The surface inputs are specified for natural ecosystems (with inputs
from biological N2-fixation and atmospheric N deposition) and agricultural systems (N and P
fertilizers, animal manure, biological N2-fixation by leguminous crops, atmospheric N
deposition). We consider two output terms, i.e., crop nutrient export in harvested crop
products and grass and hay consumption by grazing animals, and ammonia volatilization. We
present the N and P surplus, which is an important indicator for the nutrient losses to the
environment.
Discussion
The MA scenario storylines allow for describing contrasting future developments in
agricultural land use, differences being related to the efficiency of nutrient use in agriculture.
In one scenario oriented towards closing N and P cycles (pro-active approach), the overall
global agricultural efficiency increases to values of 50% for N and 54% for P. In scenarios
with a reactive approach to environmental problems, the efficiencies are somewhat lower
(44% for N and 46–48% for P). The scenarios with a reactive approach to environmental
problems show increases in agricultural N and P surpluses in all developing countries (Figure
1). In the scenarios with a proactive attitude, N surpluses decrease and P surpluses show no
change or a slight increase, except for Africa where surpluses increase in all scenarios. In
Europe and North America the N surplus will decline in all scenarios, most strongly in the
environment-oriented scenarios; P surpluses decline (proactive) or increase slowly (reactive
approach to environmental problems). In North Asia (Russian Federation) with strongly
declining population in all scenarios, the N surplus shows slow or moderate changes, and P
surplus will increase in globalization scenarios with strong economic growth and decrease in
scenarios with regional orientation.
146
Session B1: Large scale integrated policy assessment: Tools and applications
Industrialized
Industrialized
150
20
Human
Crop-export
Fertilizer
NH3-volatilization
Manure
Other_losses
Fixation
Depos.
Human
Crop-export
P-fertilizer
Surplus
P-manure
15
2050_GO
1970
2030_GO
1970
2050_GO
2030_OS
P in Tg yr-1
N in Tg yr-1
100
2050_OS
2000
2030_TG
2050_TG
2030_AM
2050_AM
10
2030_GO
2030_OS
2050_OS
2030_TG
2000
2050_TG
2030_AM
2050_AM
50
5
0
0
Developing
Developing
500
75
450
400
2050_GO
50
2050_GO
300
2050_OS
2030_GO
250
2050_TG
2030_OS
2050_AM
2030_TG
200
2030_AM
2000
2050_TG
2030_GO
2030_OS
2030_TG
2050_AM
2030_AM
25
150
100
2050_OS
P in Tg yr-1
N in Tg yr-1
350
2000
1970
1970
50
0
0
Figure 1. Surface N balance (panels on the left) and surface P balance (panels on the right) for
industrialized (top) and developing countries (bottom).
References
Alcamo, J., et al., 2006. In: Eds S.R. Carpenter, et al., Ecosystems and human well-being:
scenarios. Island Press, Washington DC, pp. 279-354.
Bouwman, A.F., et al. (Eds), 2006. Integrated modelling of global environmental change. An
overview of IMAGE 2.4. Publication 500110002/2006, Netherlands Environmental
Assessment Agency, Bilthoven, 228 pp.
FAO, 2008. FAOSTAT database collections (www.apps.fao.org), Food and Agriculture
Organization of the United Nations, Rome.
Galloway, J.N., et al., 2004. Biogeochemistry 70: 153-226.
Grübler, A., et al., 1995. Global energy perspectives to 2050 and beyond, World Energy
Council /International Institute for Applied Systems Analysis, Laxenburg, Austria.
Mackenzie, F.T., et al., 1998. In: Eds J.N. Galloway & J.M. Melillo, Asian Change in the
Context of Global Climate Change. Cambridge University Press, New York.
United States Geological Survey, 2008. Mineral commodity summaries (http://minerals.usgs.
gov/minerals/pubs/mcs/). Digital Data Series DDS-37, US Geological Survey.
147
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Incorporating livestock in global integrated assessments
of land use and agro-ecosystems services
M. Herrero, P.K. Thornton, M. Rosegrant, P. Havlik, S. Msangi, S. Fritz, H. Lotze-Campen,
A. Notenbaert, A. Popp, B. Eickhout, C. Muller, J. Van de Steeg, D. Gerten, T. Sulser,
C. Ringler, S. Wood, H.A. Freeman
International Livestock Research Institute, P.O. Box 30709-00100, Nairobi, Kenya
Contact: m.herrero@cgiar.org
Introduction
The World is changing at unprecedented rates due to a range of drivers such as increased
human population, rural/urban migrations, income increases, dietary changes, climate change
and others. These changes exert significant pressures on the use of resources, and at least in
some places, they hamper the functioning of agro-ecosystems and affect several aspects of
human well-being (nutrition, mortality, incomes and others). In the process, the poor become
more vulnerable and the sustainability of ecosystems for future generations gets
compromised. The last decades have seen a crop of forward looking integrated assessments
that try to understand better these effects and to find solutions (policies, technologies,
investments) to satisfy the global requirements of future food production, ecosystem
functioning, poverty reduction and others. Notable examples of these assessments are the
Millennium Ecosystem Assessment (MA, 2005), the Comprehensive Assessment of Water in
Agriculture (CA, 2007), IPCC’s Fourth Assessment Report (IPCC, 2007), GEO4 (UNEP,
2007) and the International Assessment of Agriculture, Science and Technology Development
(IAASTD, 2008). Some of these have not considered livestock at all, or as explicitly as it is
required in order to fully elucidate the impacts and contribution of livestock on the use of land
and other resources, food security and other dimensions of human well being and ecosystems
functioning. This is somewhat surprising considering that livestock systems are the largest
land use system on Earth (Reid et al., 2008) and that they play a key role in the livelihoods of
many people around the World, especially the poor. This paper examines the key elements
and feedbacks of livestock systems that could be included to improve integrated assessments
of land use, agro-ecosystems services and human well-being.
Methods
We reviewed the main global assessments mentioned above and the tools and models they
used. We identified gaps of where the key linkages with livestock where missing in these
models and proposed the key aspects and ways of incorporating them for future assessments.
This led to the development of a framework for incorporating spatially differentiated livestock
systems, livestock numbers and productivities by species, management parameters and use of
resources for use in a range of global integrated assessment models.
Results and discussion
Some key features of livestock systems that need to be incorporated in global integrated
assessments are presented in Table 1. They all relate to dynamic aspects of livestock
production, competition and trade-offs for resources, sometimes between systems and others
with other sectors (i.e. water, grains), to the sustainability of ecosystems and to how they
support humans (food). Including these important dimensions will allow us to understand the
beneficial and negative aspects of global livestock production in the future, and to develop
policies to ensure that livestock keeps having an important role in global food security and in
the livelihoods of the poor.
148
Session B1: Large scale integrated policy assessment: Tools and applications
Table 1. Some aspects of livestock systems that need to be represented in global integrated
assessments.
Land use
Better estimates of global rangeland productivity.
Rangeland composition and dynamics for assessing future
change.
Better estimates of carrying capacity of rangelands.
Feed supply, both for monogastrics and ruminants.
Livestock productivity
Consumption of feeds by different species.
Changes in the productivity of different species as genetics
and feeding changes over time by production system.
Livestock numbers
Better understanding of what drives the spatial distributions
of animals (ruminants, pigs and poultry).
Feasibility of animal numbers in different systems due to
resource constraints.
Water use by livestock
Water embedded in the production of feeds for different
species and water intake by animals.
Competition for water between livestock and other sectors.
Livestock systems and their
Improved definitions of livestock systems (i.e. industrial,
changes
mixed, pastoralist, etc).
Systems transitions between pastoral and mixed systems due
to intensification, service and technology provision and
others.
Systems transitions between mixed and industrial systems.
Disaggregated food supply (milk, meat, others) from
different systems.
Intensification thresholds of livestock production.
Environmental impacts in different livestock systems
(excretions, etc).
Livestock and climate change Animal species changes due to changes in environmental and
production conditions.
Mitigation measures for greenhouse gases.
Feeding animals under different climate change scenarios.
References
CA, 2007. Water for Food, Water for Life: A Comprehensive Assessment of Water
Management in Agriculture. Ed. D. Molden, Earthscan, London, and IWMI, Colombo.
UNEP, 2007. GEO4: Global Environment Outlook 4, Environment for Development. United
Nations Environment Programme. Online at www.unep.org/GEO/geo4/
IAASTD, 2008. The International Assessment of Agricultural Science and Technology for
Development. Island Press, Washington DC.
IPCC (Intergovernmental Panel on Climate Change), 2007. Climate Change 2007: Impacts,
Adaptation and Vulnerability. Summary for policy makers. Online at
http://www.ipcc.cg/SPM13apr07.pdf
MA, 2005. The Millennium Ecosystem Assessment. Ecosystems and Human Well-being:
Scenarios, Volume 2. Island Press.
Reid, R.S., et al, 2008. In: Eds K.A. Galvin, et al., Fragmentation in Semi-Arid and Arid
Landscapes. Springer, Dordrecht, The Netherlands, pp. 1-24.
149
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
What if not all land is created equal? The role of heterogeneous land
when assessing the impact of trade liberalization on developing countries
Marijke Kuiper
International Trade and Development, Public Issues Division,
LEI, Wageningen UR, P.O. Box 29703, 2502 LS The Hague, The Netherlands
Contact: marijke.kuiper@wur.nl
Introduction
The impact of trade liberalization on developing countries has become a central topic in
discussions on international trade in the context of World Trade Organization (WTO)
negotiations. The GTAP database plays a prominent role in this debate by providing the basis
for the great majority of international trade models. In this paper, we focus on one aspect of
GTAP-based models: the modelling of land. Land is a key input in agriculture, which is the
most important sector in terms of employment and foreign exchange earnings in many
developing countries. In addition to the importance for developing countries, agriculture is
also the most contested area in the current WTO negotiations. One reason is the relative high
levels of current protection in agriculture compared to manufactured goods. This relatively
high current level of protection also implies that global gains from liberalization of agriculture
will be high (Hertel et al., 2007). The way in which land is treated in trade models, therefore,
appears crucial for understanding the impact of trade liberalization in developing countries.
The GTAP database distinguishes one type of land, next to two types of labour (skilled and
unskilled), capital and natural resources. A casual look at the map already suggests that
having a single type of land may not be well-suited for analysing the impact on developing
countries. Most rich countries are located in the temperate zones, whereas developing
countries are predominately found in the tropics. In this paper we explore whether the
expected impact of trade liberalization on developing countries is affected by differences in
land endowments. We hypothesize that developing countries have less productive land,
implying that benefits from trade liberalization are less than expected based on currently used
models with a single type of land.
Methodology
Our analysis relies upon a recent GTAP-compatible land use dataset containing data on land
endowments, harvested area and yields by crop, agro-ecological zone and country. These data
allow us to compute a yield by crop and agro-ecological zone. Ignoring for now differences
between countries we can establish for each crop the agro-ecological zone with the highest
yield. Normalizing the yields to range from 0 to 1 for the highest yield provides an overview
of potential of each AEZ for the eight crops distinguished in the GTAP database (Figure 1).
Figure 1 clearly indicates the variability in suitability for different crops with the tropical
AEZs (AEZ 1–6) being less productive than those in the temperate zones (AEZ 7–12),
supporting our hypothesis that developing countries may gain less from agricultural trade
liberalization than commonly expected.
Given the prominent role of agriculture in our analysis we start from the GTAP-AGR
model, a version of the GTAP model dedicated to analysing agricultural policy questions
(Keeny & Hertel, 2005). Instead of defining production by AEZ we maintain a single
production function by crop (as in GTAP-AGR) but redefine the single type of land as a land
aggregate composed of land of different AEZs. We then add a nest to the production function
to determine the composition of this land aggregate. By assuring that this land nest captures
the productivity differences of land across AEZs we incorporate the same amount of
information in the model as with defining production by AEZ (differences in yields across
AEZs) without having an explosion of the model’s dimensions.
150
Session B1: Large scale integrated policy assessment: Tools and applications
AEZ 18
AEZ 17
AEZ 1
1.00
Tropical
AEZ 2
AEZ 3
0.75
AEZ 16
AEZ 4
0.50
Temperate
0.25
AEZ 15
AEZ 5
0.00
AEZ 14
AEZ 6
AEZ 13
AEZ 7
AEZ 12
AEZ 8
AEZ 11
AEZ 9
Boreal
Arid
AE Z 1
Dry semi-arid
AE Z 2
Moist semi-arid
AE Z 3
Sub-humid
AE Z 4
Humid
AE Z 5
Very humid
AE Z 6
Arid
AE Z 7
Dry semi-arid
AE Z 8
Moist semi-arid
AE Z 9
Sub-humid
AE Z 10
Humid
AE Z 11
Very humid
AE Z 12
Arid
AE Z 13
Dry semi-arid
AE Z 14
Moist semi-arid
AE Z 15
Sub-humid
AE Z 16
Humid
AE Z 17
Very humid
AE Z 18
AEZ 10
Rice
Other grains
Oil seeds
Plant-based fibres
Wheat
Fruit and vegetables
Sugar cane and beet
Ot her crops
Figure 1. Normalized yields by GTAP crop and agro-ecological zone (AEZ).
First results
The aim of this study is to assess whether heterogeneity of land affects the expected impact of
trade liberalization on developing countries. We, therefore, compare the results of the model
including the productivity differences across AEZs (SEAMTAP) with the findings of GTAPAGR. We also compare the impact of changing the standard assumption of full employment
to presence of unemployment in all but the high income countries.
A first test run with the model reducing all tariffs by 25% results in a rather varied picture
although with limited differences in total welfare between the models with and without
productivity (between –0.9 and +1.6 percent point difference). Most country groups gain less
when accounting for productivity of AEZs, with the exception of LDCs with moist/sub-humid
land, middle income countries with arid/dry and moist/sub-humid land and high income
countries with dry/arid land. These exceptions show an increase in their land productivity
measure signalling a better match between crops and AEZs. Accounting for unemployment in
developing countries changes welfare effects considerably: all developing countries see a
welfare increase while all high income countries face a decline in welfare. Effects of
accounting for productivity are less pronounced when unemployment is introduced in the
model (between –0.2 and +0.8). The full results reported in the final paper will be based on a
baseline projection to 2014 and 2020 and assess the impact of a realistic representation of the
WTO negotiation results of July 2008.
References
Hertel, T.W., et al., 2007. Economic Policy 22(50): 289-337.
Keeney, R. & T. Hertel, 2005. GTAP-AGR: A Framework for Assessing the Implications of
Multilateral Changes in Agricultural Policies Technical Paper 24. Publisher, Purdue.
151
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated economic-environmental modelling for domestic policy
development and assessment: The Canadian experience
Hugues Morand, George Achuo, Robert J. MacGregor, Alexandre Lefebvre
930 Carling Ave., Floor 6, Room 689, Ottawa ON, Canada, K1A 0C5
Contact: morandh@agr.gc.ca
Introduction
Understanding how changes to policies and programmes will impact the agricultural sector’s
economic and environmental performance is critical for the policy development and
evaluation process (Junkins, 2005). Demand for this type of work continues to increase as
Governments are expected to be accountable for achieving outcomes and demonstrate
measured progress towards goals. Achieving this necessitates linking biophysical models to
economic models (Lefebvre et al., 2005). Under the National Agri-Environmental Health
Analysis and Reporting Program (NAHARP), Agriculture and Agri-Food Canada (AAFC)
has been developing an integrated economic – environmental modelling capacity to assess (expost) or predict (ex-ante) the combined economic and environmental impacts or effectiveness
of proposed programmes and measures, and identify trade-offs in policy formulation. The
presentation will: (1) provide an overview of this integrated system; (2) use ex-ante definition
of regional environmental targets as an illustrative application for policy implementation; and
(3) discuss challenges as well as more recent applications and developments.
Methods
The integrated modelling system uses an economic model to estimate changes in farm
resource allocation (crops and livestock) relative to a baseline level for selected scenarios, and
feeds this information into biophysical models to assess a suite of potential environmental
impacts (Figure 1).
The economic model used is the Canadian Regional Agriculture Model (CRAM) which is
a sector equilibrium, static, non-linear optimization model maximizing producer plus
consumer surplus minus transportation costs (Horner et al., 1992). CRAM is referred to as a
positive mathematical programming (PMP) model (Howitt, 1995). The basic commodity
coverage includes grains and oilseeds, forage, beef, hogs, dairy and poultry. Biofuels are a
Economic Parameters
Technology
Farm Management Practices
Physical Resource Base
Biophysical/Climate Models
Decision Models
Policy
Model
(CRA M)
Policy Scenarios
(relative to
baseline)
Resource Allocations
cropping patterns
livestock production
management practices
input use
Scientific Knowledge
Environmental Data
Governments
Industry
Academics
Biophysical Models
(AEIs, CEEMA,
C RAWUM)
FEEDBACK
Economic
Impacts
Environmental
Impacts
Policy and
Program
Development
Figure 1. Integrated Economic – Environmental Modelling System.
152
Session B1: Large scale integrated policy assessment: Tools and applications
recent inclusion in the model as a value added activity for grains and oilseeds. Spatially,
CRAM covers 55 cropping regions based on Statistics Canada Census of Agriculture while
provinces are the smallest spatial units for the livestock component of the model. CRAM runs
with GAMS© and uses MINOS as its solver.
Depending on purposes and issues to be analysed, CRAM has been linked over the years to
the following biophysical models: Environmental Policy Integrated Climate Model (EPIC);
Agri-Environmental Indicators (AEIs); Canadian Economic and Emissions Model for
Agriculture (CEEMA); and Canadian Regional Agriculture Water Use Model (CRAWUM).
EPIC is a plant growth model which has been used to forecast yields to be integrated into
CRAM to assess economic impacts of climate change. AEIs allow tracking progress of the
sector in terms of environmental performance, and CRAM results are fed into AEIs to look at
impacts of changes in production patterns on water, air, soil and biodiversity. Using CRAM
outputs as inputs, CEEMA calculates greenhouse gas emissions from the agricultural sector
using IPCC tier two coefficients. Finally, CRAWUM is aimed at assessing the total
agricultural demand for water by sub-sectors and regions using CRAM irrigated areas and
livestock activity levels.
Results and discussion
This modelling approach has been used in the analysis of possible strategies for mitigating
GHG emissions from agriculture (Kulshreshtha et al., 2002), developing environmental
outcome targets for the Agricultural Policy Framework (APF) implementation agreements
(Heigh & Junkins, 2004), estimating regional environmental impacts of agricultural trade
liberalization (OECD, 2004), and most recently to estimate regional environmental impacts of
a 5% biofuels mandate on all transportation fuel used in Canada.
Projects relying on this integrated modelling system which are currently underway include
an ex-post assessment of APF with respect to initial targets, environmental assessments of
business risk management programmes such as crop insurance and income stabilization
programmes as well as cellulosic-based biofuels strategy, integrated analysis of adaptation to
climate change, and assessment of the potential to enhance the provision of ecosystem
services from the agricultural sector. Finally, in order to facilitate the use of AEIs by policy
analysts and reduce the burden on scientists, a totally automated interface between CRAM
and AEIs is being developed.
References
Heigh, L. & B. Junkins, 2004. Quantitative Analysis of the Impact of Agricultural
Management Strategies on Environmental Indicators. Agriculture and Agri-Food Canada,
Ottawa, Ontario, Canada.
Horner, G.L., et al., 1992. The Canadian Regional Agriculture Model: Structure, Operation
and Development. Policy Branch, Agriculture Canada, Technical Report 1/92.
Howitt, R.E., 1995. American Journal of Agricultural Economics 77(2): 329-342.
Junkins, B., 2005. In: Eds A. Lefebvre, et al., Environmental Sustainability of Canadian
Agriculture: Agri-Environmental Indicator Report Series – Report no. 2. Agriculture and
Agri-Food Canada, Ottawa, Ontario, Canada, pp. 36-39.
Kulshreshtha, S., et al., 2002. Canadian Economic and Emissions Model for Agriculture
(CEEMA 2.0): Technical Documentation, CSALE Working Paper no. 12, University of
Saskatchewan, Saskatoon, Saskatchewan, Canada.
Lefebvre, A., et al. (Eds), 2005. Environmental Sustainability of Canadian Agriculture: AgriEnvironmental Indicator Report Series – Report no. 2. Agriculture and Agri-Food Canada,
Ottawa, Ontario, Canada.
OECD, 2004. Agriculture, trade and environment: Arable crops sector. Organisation for
Economic Co-operation and Development, Paris, France.
153
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Comparison of econometric techniques for deriving consolidated databases:
The AGROSAM case
M. Müller, I. Pérez Dominguez, W. Britz, T. Heckelei
Institute for Prospective Technological Studies of the European Commission,
Seville, Spain
Contact: Marc.Mueller@ec.europa.eu
Introduction
Integrated policy impact assessment at Pan-European or global scale requires large-scale
consolidated databases to feed economic or biophysical models or components. A key data set
for economic analysis are social accounting matrices (SAM) which represent the monetary
flows between productive sectors and institutions and thus may serve a large variety of
quantitative tools, especially Computable General Equilibrium (CGE) models. However, the
datasets underlying the SAMs, namely national Supply- and Use Tables (SUT) or symmetric
Input-Output Tables (IOT), are typically highly aggregated by sectors and commodities and
thus provide little detail for sub-sector specific analysis. The agricultural sector is e.g. often
represented as one row and column only in the national datasets.
This coarse representation is one reason for the limited application of CGEs for analysis of
the Common Agricultural Policy. The AgroSAM project (Müller & Pérez Dominguez, 2008)
hosted at the Institute for Prospective Technological Studies of the European Commission
(IPTS) addresses this issue by combining national SUTs for the EU Member States with the
highly disaggregated information on the agricultural sector provided by the CAPRI model
database (Britz, 2005). One of the main challenges for AgroSAM consists in overcoming
definitional and structural differences between the SUTs based on the European System of
National Accounts (ESA95) and the CAPRI database which is mainly structured according to
the Economic Accounts for Agriculture (EAA, Eurostat, 1997). As such, the AgroSAM
project is one example for constructing large-scale data bases for impact assessment where
different data sources are combined and consolidated.
Methods
As a result of the structural deviations between the databases to be combined, the totals of the
agricultural sub-sectors based on the EAA do not match the corresponding values in the
SUTs, and the obtained, disaggregated SAMs were not automatically balanced. So far a crossentropy (CE) framework balances the SAMs by staying as close as possible to the information
obtained from the CAPRI database while respecting the totals indicated by the SUTs. The
supports for the agricultural SAM entries are centered on priors derived from the original
CAPRI data. The spread of the supports around the priors steering the deviation between
final, balanced SAM entries and priors is so far based on a subjective evaluation of the
reliability such that, e.g., entries relating to cereals will deviate less from the priors compared
to entries relating to fodder crops.
A CE application requires for each estimate matching priors and weights, and in the case of
more then two supports, additional constraints. The resulting high number of variables and
constraints may cause computational difficulties for large-scale datasets during estimation and
will increase estimation time. Additionally, the implicit posterior density depends on the
interaction between the choice of supports, the a priori probabilities and the entropy criterion.
Both problems were addressed by Heckelei et al. (2008) and Witzke & Britz (2005) by
motivating a Highest Posterior Density (HPD) estimator which refrains from discrete support
points but still allows to express confidence by using informative priors on the variance of
each estimate.
154
Session B1: Large scale integrated policy assessment: Tools and applications
The aim of this paper is to compare the performance of CE and HPD estimators in the
context of the SAM balancing based on prior information. For this we investigate the
differences in the estimations and the performance gains in computation on the example of
selected SAMs for EU Member States. Further on, we test a less subjective way of defining
the second moment of the prior distributions by using the production values of the
commodities, value of sold quantities and subsidies paid to the sectors to define the variance.
The weights are motivated firstly by assuming that the economic importance of an activity
will steer to a certain extent the resources spend to generate statistical data for it. Secondly, a
high economic sectoral value implies aggregation over a large number of individual agents. If
the error in the aggregated value depends on individual reporting errors centered around zero,
the probability of an error in the aggregates reduces with the number of reporting individuals.
Both arguments support using production values, values of sold quantities and subsidies
received as a basis to define the expected variance of the priors.
Results and discussion
Currently, AgroSAMs with their highly dis-aggregated agricultural sector while keeping the
full sectoral breakdown from the national SUTS are available for 21 EU Member States in
2000, and estimation for the remaining six Member States is underway. The paper will firstly
present the large difference in agricultural detail between existing SAMs with an agricultural
breakdown as the GTAP data base and the new AgroSAMs. Secondly, it will discuss the
differences in results, implementation and computational viewpoints between the CGE
framework and a HDP one. The dataset is a key input in the FP VII project CAPRI-RD where
it will provide the basis to populate regional CGEs at NUTS II. In addition, we expect to
contribute to the current methodological work around SAM balancing and databasecombination, a common problem shared by large model chains like SEAMLESS, where
partial and general equilibrium models are combined.
References
Britz, W. (Ed.), 2005. CAPRI Modelling System Documentation. URL: http://www.ilr1.unibonn.de/agpo/rsrch/capri/capri-documentation.pdf, University of Bonn.
Eurostat, 1997. Manual on the Economic Accounts for Agriculture and Forestry EAA/EAF 97
(Rev. 1.1).
Heckelei, T., et al., 2008. A Bayesian Alternative to Generalized Cross Entropy Solutions for
Underdetermined Econometric Models. Agricultural and Resource Economics, Discussion
Paper 2008:2, Bonn, Germany.
Mueller, M. & I. Pérez Dominguez, 2008. Compilation of Social Accounting Matrices with a
Detailed Representation of the Agricultural Sector (AgroSAM). Presented at the 11th
Annual Conference on Global Economic Analysis, Helsinki, Finland.
Witzke, H.-P. & W. Britz, 2005. Consolidating trade flows and market balances globally
using a Highest Posteriori Density estimator. Presented at the 8th Annual Conference on
Global Economic Analysis, Lübeck, Germany.
155
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Agriculture and natural resources in a changing world:
The role of irrigation
T. Sauer, P. Havlík, U.A. Schneider, G. Kindermann, M. Obersteiner
Research Unit Sustainability and Global Change (FNU), Hamburg University,
Bundesstrasse 55 (Pavillon), 20146 Hamburg, Germany
Contact: timm.sauer@zmaw.de
Introduction
Fertile land and fresh water constitute two of the most fundamental resources for food
production. These resources are affected by environmental, political, economic, and technical
developments. Regional impacts may transmit to the world through increased trade. With a
global forest and agricultural sector model, we quantify the impacts of increased demand for
food due to population growth and economic development on potential land and water use. In
particular, we investigate producer adaptation regarding crop and irrigation choice,
agricultural market adjustments, and changes in the values of land and water. To our
knowledge this is the first large scale assessment of agricultural water use under explicit
consideration of alternative irrigation options in their particular biophysical, economic, and
technical context, accounting for international trade, motivation-based farming, and quantified
aggregated impacts on land scarcity, water scarcity, and food supply.
Methods
We apply a mathematical programming-based, price-endogenous sector model of the
agricultural and forestry sectors. The model depicts production, consumption, and
international trade in 11 world regions. It was programmed using GAMS software (General
Algebraic Modeling System). Market and trade equilibrium in global agricultural markets are
simulated to reveal commodity and factor prices, levels of domestic production, export and
import quantities, resource usage, and environmental impacts. The agricultural sector is
represented by more than 40 crops, and an aggregated livestock sector. For crop management,
the model can choose between different irrigation systems. Four types of irrigation are
portrayed: basin and furrow surface irrigation, localized drip, and sprinkler irrigation. For
each method, we evaluate biophysical and technical suitability to exclude inappropriate
irrigation system applications. Micro-economic data include production costs, resource
requirements, and expected yields. Production costs contain all expenses for management and
inputs required to reach the respective management-related yield. The interdisciplinary range
of factors that determine irrigation decisions in our model is shown in Table 1. Actual water
use is finally computed considering irrigation cost per spatial unit for all appropriate
combinations of regional geographic background, crop type, and irrigation system.
Results and discussion
This study integrates alternative irrigation methods into a global agricultural and forest model
to estimate regional adaptations in agricultural water use for different development scenarios.
The new model combines the heterogeneity of irrigation technologies and natural resources
with micro and macro-economic drivers. The integration of explicit irrigation systems into a
partial equilibrium model of the agricultural and forestry sectors improves global land use
change assessments and evaluation of interdependencies between policies, land use related
externalities and food supply.
Our simulations show that agricultural responses to population and economic growth
include considerable increases in irrigated area and agricultural water use, but reductions in
the average water use intensity. Furthermore, we show that irrigation is a complex decision
beyond the binary decision of adopting irrigation or not. Different irrigation systems are
156
Session B1: Large scale integrated policy assessment: Tools and applications
preferred under different exogenous conditions. To accurately estimate land and water
scarcity, the likely adaptation of farmers to different irrigation methods needs to be
quantified. Negligence of these adaptations would bias the burden of development on land
and water scarcity. Without technical progress in agriculture, a population and income level
as predicted under our scenario for 2030 would require substantial price adjustments for land,
water, and food to equilibrate supply and demand (compare Figure 1).
Table 1. Biophysical, technical, and economic determinants of irrigation choice.
Crop characteristics
Water application efficiency
Crop market prices
Soil infiltration rate
Operation time per event
Investment capital cost
Slope inclination
Level of pressurization
Energy prices
Length of growing period
(energy and labour requirement) Labour cost
Land and water prices
Water resource availability
Coverage per system unit
(resource economics)
Western Europe (WEU)
Central and East Europe (EEU)
South Asia (SAS)
Latin America and Caribbean (LAM)
Sub-Saharan Africa (AFR)
Figure 1. Results: Water price index by region.
Selected references (examples)
Alexandratos, N., et al. (Eds), 2006. World agriculture: Towards 2030/2050 - Prospects for
food, nutrition, agriculture and major commodity groups. FAO, Global Perspective Studies
Unit, Rome.
Brouwer, C., et al., 1988. Irrigation Water Management: Irrigation Methods. FAO Irrigation
Water Management Training manual no 5, Rome.
Buchanan, J.R. & T.L. Cross (Eds), 2002. Irrigation Cost Analysis Handbook. Agricultural
Extension Service, The University of Tennessee, PB1721.
Cowie, A., et al., 2007. Environmental Science & Policy 10: 335-352.
Foley, J.A., et al., 2005. Science 309: 570-574.
Wallace, J.S., 2000. Agriculture, Ecosystems and Environment 82: 105-119.
Wichelns, D. & J.D. Oster, 2006. Agricultural Water Management 86: 114-127.
157
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The fate of nitrogen in the Brazilian soybean chain
E.M.A. Smaling, R. Roscoe, J.P. Lesschen, A.F. Bouwman, E. Comunello
International Institute for Geo-Information Science and Earth Observation (ITC),
P.O. Box 6, 7500 AA Enschede, The Netherlands
Contact: smaling@itc.nl
Introduction
With economic growth, urbanization and changing diets, world demand for plant-derived oils
and their derivates is soaring. Oils are used inter alia in the food, feed and cosmetics industry,
and increasingly as a biofuel. Soybean is one of the major booming oil crops. In addition to
oil, it offers a very valuable by-product, i.e. protein-rich soy meal, a raw material for animal
feed. Increasing demand, mainly from China has been met with a supply response that is
particularly strong in Brazil and Argentina. Recognizing that soybean (i) is a highly important
commodity for the Brazilian economy, (ii) contributes to the conversion of forest and savanna
land, and (iii) travels long distances, ending up in many different foods and feeds worldwide,
there is a need to better understand the dynamics of the soybean sector, and how economic
and environmental targets can be realized simultaneously. In this study, nitrogen (N) is used
as a marker. N stocks and flows in and between the compartments of the Brazilian soybean
chain are quantified, to find out where in the process N remains inside or disappears from the
food chain. Compartments include forest and savanna conversion, soybean cultivation,
transport and processing, animal and human consumption, and waste disposal (Figure 1).
Methods
Spatial data sets on soybean area and yields, soils and rainfall were collected at municipality
level, for three time periods (1993–1995; 1998–2000; 2003–2005) and put together in a GIS.
Export to China and the European Union was included. Nitrogen dynamics by forest and
savanna conversion was taken from literature, whereas the nutrient balance model NUTMON
model (Lesschen et al., 2007) was used to calculate the cultivation part (Phase 2 in Figure 1).
For the calculation of the fate of nitrogen during Phase 3–5 (Figure 1), several assumptions
were made, based on literature and expert knowledge.
Results
Amazon forest and Cerrado savannah has most often been replaced by pastures. These are
increasingly replaced by soybean farms. Some parts of the Amazon are converted into
soybean fields directly. Soybean has also replaced other arable crops, and its growth can only
partially be linked to forest and savanna removal. Estimates (Phase 1) of N losses in forest
and savanna that can be attributed to soybean range between 2000 and 6000 million kg of N
yr–1. This comes on top of biodiversity loss, destruction of living areas, and the creation of
increasingly dry conditions due to reduced air circulation on the continent. Estimates (Phase
2) of N losses in soybean fields were 182 million kg of N yr–1 in 1993–1995, but in 2003–
2005, there is a gain of 62 million kg of N, due to generally improved practices of
conservation agriculture, and due to the N-fixing nature of (the leguminous species) soybean.
Other nutrients may be limiting, and/or having negative nutrient balances though (P, K, trace
elements), but this was not investigated. Also, the issue of genetically modified soybean
(currently estimated at 43%), and massive use of pesticides and associated health risks are not
dealt with here. Of the total soybean production, 75% is turned into soy meal, 20% into soy
oil, and about 5% is lost during processing and transport. As soy oil contains no N, it is no
longer relevant for the chain studied here. The soy meal is largely fed to animals, mainly to
pigs and chickens. During animal production and human consumption, a series of N losses
come to the fore, production of animal manure being the main (recyclable) loss. The N
158
Session B1: Large scale integrated policy assessment: Tools and applications
remaining in consumed meat for humans is about 20% of the N available in the harvested
soybeans. Finally, of the N excreted by humans into sewerage systems, 32% ends up in
surface waters.
Phase 1
Conversion
Forest
Pasture
Phase 2
Cultivation
Cerrado savanna
Agricultural land
Soybean-based
farming system
Fertilizer
Manure
Deposition
Leaching
Soybeans
Maize and
wheat
N-fixation
Phase 3
Transport and
processing
Crop residues
Brazil
Gaseous losses
Erosion
EU and China
Crush factory
Soy meal
Phase 4
Consumption
Soy oil
Feed
Livestock
Meat, eggs,
milk
Phase 5
Waste treatment
Waste disposal
Figure 1. Five phases of the soybean chain studied.
Reference
Lesschen, J.P., et al., 2007. Nutrient Cycling Agroecosyst. 78: 111-131.
159
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Spatial planning of livestock production and manure abatement
B. Van der Straeten, J. Buysse, S. Nolte, F.L. Marchand, L. Lauwers,
D. Claeys, G. Van Huylenbroeck
Ghent University, Department of Agricultural Economics,
Coupure links 653, 9000 Ghent, Belgium
Contact: Bart.VanderStraeten@ugent.be
Introduction
Some Western European countries (e.g., The Netherlands and Belgium) experienced a large
expansion in animal production. An excess of nutrients resulted from the feed compounds
trade balance and led to a high pressure on the environment (Feinerman & Komen, 2005;
Nesme et al., 2005). As a consequence of the Nitrate Directive (91/676/EEC), the Flemish
region has introduced a manure decree in 1991 which describes how manure should be
disposed. There are currently two options to dispose excess manure. Firstly, the manure can
be transported to farms with a manure deficit. Secondly, the manure could be processed.
Despite the fact that manure transport is running to its limits, the processing capacity has not
sufficiently developed yet to solve the manure excess problem.
One of the problems of the development of the processing capacity for private investors
and policymakers is the uncertainty of the development of the manure excess. The regional
concentration of animal production is very diverse and together with high transport costs this
creates a huge spatial differences in the demand for manure processing. Due to the interplay
between transport and processing, it is difficult to predict where demand for processing
capacity will arise. A simple, but incorrect, indicator of this demand is the comparison of
animal production density and available disposal capacity, which is currently used to make a
spatial differentiation of the policy interventions. However, this indicator ignores the
possibility to transport to neighbouring regions with a manure deficit and disregards the
fertilizing behaviour of the farms. Regions with high animal production density enclosed by
other regions with high animal production density have indeed a higher demand for
processing manure than the same regions surrounded by regions with a low animal production
density. The type of manure has to be taken into account as well, because both transport and
processing costs are very different.
Therefore, a spatial mathematical programming model is built based on a multi-agent
system (MP-MAS) that simulates the individual farmer’ behaviour confronted with all spatial
aspects of transport, spreading and processing the manure.
The paper shows how spatial mathematical programming can be used for environmental
and regional planning decisions (ex-ante and ex-post) and how the MP-MAS approach can be
extended to a very large dataset (38,000 farms) that contains the complete population. The
developed model and simulation results can provide valuable information that can reduce risk
to enhance the development of manure processing. The model results can also reduce costs
because transport costs of manure are very high and location of the processing plants is
therefore very important.
Methods
MP-MAS has two major advantages. First, it allows to deal with the regional heterogeneity of
manure production, processing and fertilization behaviour. Secondly, the model optimizes at
farm level and takes interactions between farms and their environment into account
(Boulanger & Brechet, 2005). The cost of manure allocation decision of each individual agent
is minimized subject to legal constraints of the manure policy, manure transport costs and
manure abatement limitations.
160
Session B1: Large scale integrated policy assessment: Tools and applications
The cost-minimized transports do not necessarily correspond to actual transports.
Therefore, the model is called a normative programming model, but the actual utilization of
available manure on the land is based on the empirical data.
Transport costs in the model are proportional to the distance between the farms based on
distances between municipalities. Farms within a municipality have the same distance to all
other farms, which allows to reduce the size of the between-farms transport matrix
considerably and making simulations possible with limited computer resources.
Results and discussion
The simulation results provide information for investors on the optimal location of processing
plants or additional animal production by showing the demand for additional manure
processing capacity. The spatial data on demand for processing also indicate that the current
developed processing capacity corresponds already quite good with the demand for
processing, but the total capacity is not sufficient.
Investments in manure processing not only depend on the needed capacity, but also on the
willingness to pay for manure disposal. The model calculates this willingness to pay and
shows that it increases with an increasing density of animal production and decreases as a
results of investments in additional capacity (Figure 1). Figure 1 shows the simulated
willingness to pay for each municipality in Flanders.
The simulation results are also useful for policymakers because they show the total manure
disposal cost of different incentive policies for manure processing. The results indicate that
the current manure processing obligation is not efficient because it results in an additional
manure disposal cost of 4.01%.
gemeenten
totaal080318.proxy_
0 - 0.06
0.06 - 0.17
0.17 - 0.31
0.31 - 0.59
0.59 - 0.95
0.95- 1.34
1.34 - 1.78
Figure 1. Willingness to pay for allocating one kg of nitrogen in each municipality in Flanders
(€ per kg N).
References
Boulanger, P.M.& T. Brechet, 2005. Ecological Economics 55(3): 337-350.
Feinerman, E. & M.H.C. Komen, 2005. Environmental & Resource Economics 32: 367-388.
Nesme, T., et al., 2005. Agricultural Systems 83(3): 297-314.
161
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The impact of modulation: Modelling first and second pillar CAP policies
H. Van Meijl, T. Jansson, M. Banse, G.B. Woltjer
LEI, Wageningen University and Research Centre,
P.O. Box 29703, 2502 LS The Hague, The Netherlands
Contact: hans.vanmeijl@wur.nl
Introduction
The aim of this paper is to assess the economic impact of the transfer of funds from direct
income support to farmers (Pillar One) to rural development money (Pillar Two) of the
Common Agricultural Policy (CAP) through the compulsory modulation mechanism, as
provided for under Article 10 of Council Regulation (EC) No 1782/2003. Modulation was
introduced, originally as a voluntary mechanism, in 2000 as a means of increasing support for
rural development within the CAP. This is achieved by transferring a proportion of the Pillar
One budget to the funding of rural development measures under Pillar Two. This requirement
became mandatory in 2003 as a result of the Mid Term Evaluation of the CAP. This currently
applies to the EU-15, however compulsory modulation will apply to the twelve new Member
States that acceded to the EU in 2004 and 2007 when their Pillar One payments reach the
same level as those of the EU-15. For the 2007–13 programming period, compulsory
modulation increases the financial support available to rural development measures by 8
billion euros to 88 billion euros.
Scope of the Study and Methodological Approach
This paper, therefore, aims to provide a quantitative and qualitative assessment of the impacts
of this transfer of funds from Pillar One to Pillar Two of the CAP through the use of the
compulsory modulation mechanisms on the social and economic performance of the
agriculture sector and rural areas. More specifically, it studies the impact on the environment,
the competitiveness of the agriculture sector, on rural communities and national rural
development budgets. The study also considers the re-distribution effects of modulation,
within and between Member States, between economic sectors and types of holdings. This
study is innovative as it is the first that models explicitly the various measures (all three axis)
of the second pillar of the CAP in a quantitative way.
Specifically, it considers the impacts of compulsory modulation under two distinct
scenarios, within the time horizon of 2013, and across the EU27. The first scenario consists of
the current rules under which compulsory modulation operates (5% modulation rate and
associated franchise and distribution rules). The second scenario comprises the changes
proposed under the CAP ‘Health Check’ in May 2008 (an additional 8% by 2013, with further
increases according to farm size).
Methods
In the Modulation project the commodity focus and regional / territorial focus have to be
connected. The global economy-wide dimension is covered by an economic general
equilibrium model (LEITAP, see Van Meijl et al., 2006). ESIM – an EU-wide partial
equilibrium model – is providing more agricultural detail for the EU-25 countries, CAPRI, an
EU25 regional partial equilibrium model, is distributing this impact to the regional (NUTS2)
level (see Britz, 2005). ESIM’s main contribution is the projection of developments in EU
agricultural markets into the future. CAPRI’s main contribution is changes in CAP policies
and the regional impact (NUTS2 level). To cover modulation impacts the CAPRI model is
extended with article 69 payments within the first pillar and with the second pillar measures.
LFA, N2000 and Agri Environmental payments are directly implemented in CAPRI, and the
remaining measures are captured by linking the costs and production technology of CAPRI to
162
Session B1: Large scale integrated policy assessment: Tools and applications
simulation results of LEITAP, where those other measures are explicitly implemented.
LEITAP is a global computable general equilibrium model that covers the whole economy
(Van Meijl & Tongeren, 2002; Van Meijl et al., 2006; Banse et al., 2008). A key feature of
modulation is that some measures like physical and human capital investment have dynamic
impacts. To include these dynamics the LEITAP model is extended to a recursive dynamic
version with endogenous technological change by specifying a relation between investments
and productivity change.
la
tio
n
od
u
M
1s
t
d
pi
pi
lla
r
la
r
l
2n
R
eg
io
na
En
v
gr
o
A
LF
A
Hu
m
an
C
ap
P
ita
hy
l
si
ca
lC
ap
ita
l
0.6
0.5
0.4
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
EU15
EU27
Figure1.
Production
volume of primary
agriculture (% change
progressive modulation
relative to baseline in
2013).
Results and discussion
The overall production effect due to progressive modulation is positive for primary
agriculture in the EU15 and EU27. The impact for EU15 is larger than for the EU27 as for the
NMS modulation hold only for the last years while it is in place for the EU15 for the whole
period. Next to the overall impact of progressive modulation this Figure shows the impact of
various groups of second pillar measures, the impact of the whole second pillar and the
impact of reducing the first pillar. The positive production effect is due to a positive effect of
redistributing the second pillar money. Within pillar two measures, especially impact of
physical capital investments is largest. A small positive impact have human capital
investments, LFA payments and agri-environmental payments. Reducing the first pillar has a
slightly negative impact on production due to that part of the payments are still coupled in
some countries in the baseline scenario and due to that decoupled payments have minor
production effects. Modulation has further a positive impact on competitiveness, environment
and quality of life.
References
Banse, M., et al., 2008. European Review of Agricultural Economics 35: 117-141.
Britz, W. (Ed.), 2005. CAPRI documentation. Available at www.ilr.uni-bonn.de
Van Meijl, H. & F.W. Van Tongeren, 2002. Eur. Rev. of Agric. Economics 29(4): 445-470.
Van Meijl, H., et al., 2006. Agriculture, Ecosystems and Environment 114: 21-38.
163
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Environmental impacts of the European livestock sector on different scales,
and prospects for reduction
Henk Westhoek, Bas Eickhout, Nico Hoogervorst, Trudy Rood,
Elke Stehfest, Geert B. Woltjer
Netherlands Environmental Assessment Agency (PBL),
P.O. Box 303, 3720 AH Bilthoven, The Netherlands
Contact: Henk.Westhoek@pbl.nl
Introduction
The European livestock sector is a large agricultural sector, employing many people,
producing good and safe meat, and dairy and other products for mainly European customers.
The (European) livestock sector has large impacts on rural areas, both inside and outside
Europe and both positive and negative. The positive impacts include the maintenance of
traditional agricultural landscapes through grazing animals, with accompanying biodiversity.
The contribution to rural economies and the production of traditional, regional products also
are considered to be positive.
However, livestock production has considerable negative impacts, such as the emission of
greenhouse gases and nutrients, and the conversion of nature areas − both inside and outside
Europe (Steinfeld et al., 2006). Especially, large impacts on nature areas outside Europe,
which are needed for the cultivation of feed products (notably soy beans), are often brought to
the fore. The imported animal feed contains large quantities of nutrients, which lead to
eutrophication in areas with intensive livestock production (Bouwman et al., 2005; Grote et
al., 2005).
In regions outside Europe, the consumption of animal products is rapidly rising, leading to
a projected doubling of the global consumption, over the next 50 years (FAO, 2003). Issues,
such as animal welfare, animal health and human health are also reasons for assessing
whether changes in production systems and/or consumption patterns might be needed, to
reduce the adverse societal impacts of livestock production (McMichael & Bambrick, 2005;
Walker et al., 2005). This paper aims to identify a number of plausible pathways to reduce the
environmental impacts of livestock production.
Methods
Starting from an analysis of issues accompanying livestock production, several options to
reduce negative impacts by the livestock sector will be defined. The discussion about the
effects of livestock production is more prominent in northwestern Europe. But, given the fact
that much of the relevant regulation (on environmental, animal welfare, veterinarian aspects,
market regulation) is on a European scale, most ‘improvement options’ will have to be
implemented on a European scale, too.
These options will include both technical options and policy options. For plausible
technical options we allow some time for them to become economically feasible, in the long
run. At the other end of the spectrum we examine policy-oriented options, like changing trade
regimes (import tariffs) for meat or feed stuffs produced outside the EU. The effects of dietary
changes also will be investigated, in terms of both of a reduction in meat and dairy
consumption, and a shift between meat types.
The effect of all options will be evaluated, qualitatively and quantitatively for a number of
environmental impacts (notably land-use changes and emission of greenhouse gases and
nutrients), to realize an integrated assessment of these options. This assessment will be done
on a global scale, since we expect many global displacement effects caused by the different
policy options. For this assessment, we will couple a CGE model (LEITAP, based on GTAP)
and an environmental impact model (IMAGE). Previous coupling of these two models proved
164
Session B1: Large scale integrated policy assessment: Tools and applications
to be very fruitful (Nowicki et al., 2006; Ten Brink et al., 2007; Eickhout et al., 2007;
Verburg et al., 2008). Chain analyses and LCAs for products of the livestock sector and their
substitutes are also used to estimate environmental impacts and option potentials. Recently,
we evaluated the effect of the EU biofuel policy proposal, using these three different kinds of
methodological approaches (Eickhout et al., 2007). Experiences gained in the biofuel case
will be evaluated and used in this case.
Results and discussion
At the conference, results will be presented followed by a discussion, since this issue concerns
a work-in-progress. From the modelling framework we expect results for a number of
indicators, which together make an integrated assessment of the effects of different policy
options possible. These indicators include human consumption of meat, dairy and other
sources of protein, the volume of livestock production, the quantity of land used needed for
this production (distinguished between cropland and pastures) and emissions of greenhouses
gases form livestock and agricultural land. Given the modelling framework, we expect that
we are better able to dynamically model the effects of lowering the consumption of animal
products in Europe on a global scale. This will probably lead to more realistic results than a
static approach would do. The dynamic modelling approach includes effects on food prices,
prices of agricultural commodities and factor prices.
References
Bouwman, A.F., et al., 2005. Pedosphere 15: 137-155.
Eickhout, B., et al., 2007. Land Use Policy 24(3): 562-575.
FAO, 2003. World agriculture: Towards 2015/2030 FAO, Rome.
Grote, U., et al., 2005. Environmental Science & Policy 8: 439-451.
McMichael, A.J. & H.J. Bambrick, 2005. Public Health Nutrition 8: 341-343.
Nowicki, P.L., et al., 2006. Scenar 2020: Scenario study on agriculture and the rural world EC
Directorate Gen. for Agriculture and Rural Development, Brussels.
Steinfeld, H., et al., 2006. Livestock’s long shadow: Environmental issues and options. FAO,
Rome.
Ten Brink, B., et al., 2007. Cross-roads of Life on Earth: Exploring means to meet the 2010
Biodiversity Target CBD Technical Series No. 31. Secretariat of the Convention on
Biological Diversity, Quebec.
Verburg, P.H., et al., 2008. Annals of Regional Science 42: 57-77.
Walker, P., et al., 2005. Public Health Nutrition 8: 348-356.
165
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Agricultural policy and land use
G.B. Woltjer
LEI, P.O. Box 29703, 2502 LS Den Haag, The Netherlands
Contact: Geert.Woltjer@wur.nl
Introduction
Agricultural policy has fundamental consequences for land use and therefore environmental
services like biodiversity. This paper will address these issues through a general equilibrium
model of the world economy in combination with a biophysical land use model where land
use issues can be analysed. The focus will be on methodological issues with respect to the use
of information from biophysical models of land use in a general equilibrium economic model.
The LEITAP model
LEITAP2 is developed at the Dutch agricultural research institute LEI, part of Wageningen
University and Research (WUR). The name is derived from the base model from which it is
derived, GTAP, and the name of the institute where it is developed. Compared with the
original version of the LEITAP model it is extended and stylized a lot.
The LEITAP2 model is based on the general equilibrium model GTAP (Hertel & Tsigas,
1997), developed at Purdue University, United States. It uses the carbon market and the rough
characteristics of the production structure of GTAP-E (Burniaux & Truong, 2001). It uses the
international capital flow accounting system of the dynamic GTAP model GTAP-DYN and
includes also some parts of the agricultural variant of GTAP, GTAP-AGR.
The LEITAP2 model includes a lot of extensions compared with the GTAP model. The
different extensions of the model can be switched on or off through a simple change in
coefficients form. First, an integrated production structure, with energy nesting (including
biofuels), feed and fertilizer nesting is included. The feed and fertilizer nest is also an
extension compared with LEITAP1. Second, there is a possibility to include dynamic
international investment in the model (was not available in LEITAP1). This will probably be
extended towards a model of investment between sectors in the near future. Third, production
quota can be implemented, in a more general way than in LEITAP1. Fourth, EU-policy,
including first and second pillar measures, can be switched on (also new compared with
LEITAP1). Fifth, land supply is modelled, based on biophysical model outcomes. Compared
with LEITAP1 the theoretical structure of the interactions between the models has been
improved a lot. Sixth, substitution between different types of land is modelled in a dynamic
way (in LEITAP1 this was static). Seventh, a dynamics of capital and labor mobility between
agricultural and non-agricultural sectors can be switched on (only static mobility was
available in LEITAP1). Eighth, the GTAP-E carbon market is implemented and will probably
be extended in the near future (New compared with LEITAP1).
Results and discussion
This is work in progress. Results will be presented at the conference. The focus of the paper
will be on methodological issues, where results are only presented to show the effect of the
model improvements. Special focus will be given on the relationship between the biophysical
models IMAGE (Bouwman et al., 2006; Eickhout et al., 2007) and DYNA-CLUE (Verburg et
al., 2002, 2006) and the LEITAP2 with respect to land supply, land productivity and land use
changes.
The map in Figure 1 gives an indication of the type of results that can be expected. The
map shows the second pillar effect of 13% modulation, i.e. a scenario where 13% of the first
pillar money is taken away and used for a combination of Less Favored Area Policy (15%),
Agri-Environmental Schemes (25%), Investment subsidies (25%) and regional measures
166
Session B1: Large scale integrated policy assessment: Tools and applications
(35%). These measures have different effects in different regions. In eastern Europe, the land
saving effects of technological improvements are most important and generate a decrease in
agricultural land use, while for example in France the land use increasing measures like LFA
and Agri-Environmental schemes are more important and generate an increase in land use.
Figure 1. Second pillar effect of 13% modulation on agricultural land use in the EU27.
References
Bouwman, A.F., et al., 2006. Integrated modelling of global environmental change. An
overview of IMAGE 2.4, Netherlands Environmental Assessment Agency (MNP),
Bilthoven, The Netherlands.
Burniaux, J.-M. & P. Truong, 2001. GTAP-E: An energy-environmental version of the GTAP
Model. GTAP Technical Paper no. 16, Center for Global Trade Analysis, Purdue
University, USA.
Eickhout, B., et al., 2007. Land Use Policy 24: 562-575.
Hertel, T.W. & M.E. Tsigas, 1997. Chapter 2 in Ed. T.W. Hertel, Global trade analysis:
Modeling and applications, Cambridge University Press, UK.
Verburg, P.H., et al., 2002. Environmental Management 30: 391-405.
Verburg, P.H., et al., 2006. Agriculture, Ecosystems & Environment 114(1): 39-56.
167
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Farm structural change in European regions
A. Zimmermann, T. Heckelei
Institute for Food and Resource Economics, Nussallee 21, 53115 Bonn, Germany
Contact: andrea.zimmermann@ilr.uni-bonn.de
Introduction
As the aggregate outcome of farm level structural adjustments in response to policy changes,
structural change links the farm with the regional level and plays a prominent role in the
integrated assessment of the agricultural sector (Van Ittersum et al., 2008). Our study thereby
focuses on the economic aspects of farm structural change. The purpose of the study is
twofold: (1) the direction of structural change defined as the change of the number of farms in
different farm types shall be identified for EU15 regions, and (2) the key exogenous factors
leading to and their impact on the structural developments shall be detected.
The analysis is conducted for a multi-dimensional farm typology combining ten
specialization classes (e.g., arable, dairy, mixed farming) and three economic size classes
based on the European standard grouping of farms (Andersen et al., 2006). The farm types
cover the whole farming sector and are mutually exclusive. In total 30 farm types plus an
entry/exit class are considered. The movement of farms between the farm types is represented
by transition probabilities which are derived using a non-stationary Markov chain approach.
The econometric specification makes use of FADN data on the transitions of sample farms
between the different farm types and combines it with data on the total number of farms per
farm type from 1990 to 2003. Exogenous factors assumed to influence structural change apart
from a trend variable are the unemployment rate and prices for various agricultural outputs.
Results for the transition probabilities as well as for the impact of the exogenous variables
will be shown for selected regions across Europe.
Methods
Methodologically, a two-step Markov chain estimation is applied, where in the first
estimation step non-stationary Markovian transition probabilities are derived:
n j (t ) = ∑ i =1 ni ( t −1) pijt ,
N
with n being the number of farms in farm type j at time t and depending on the number of
farms in all farm types i in the period before ( t − 1 ) multiplied by their respective transition
probabilities pij to move from farm type i to farm type j in one time period. The probability
constraints, non-negativity (pij ≥ 0) and summing-up to unity ( ∑ j =1 pij ) must hold. For
N
estimation of the first step a generalized cross-entropy estimator (GCE) similar to
Karantininis (2002) and Stokes (2006) is applied. The prior information on the transition
probabilities necessary for the GCE is derived from the actual movements of the FADN
sample farms (micro data), whereas the Markov equation is equipped with data on the total
number of farms per farm type and region (macro data). In the second estimation step, the
transition probabilities obtained with the GCE are used as left-hand side variable and
regressed against a set of explanatory variables using ordinary least squares (OLS). Since the
probabilities relate non-linearly to explanatory variables and coefficients (MacRae, 1977), the
equations are linearized by transformation of the transition probabilities into log-odd ratios
(Stavins & Stanton, 1980).
Results and discussion
Results will be shown for four European FADN regions: The Netherlands, Brandenburg
(Germany), Midi-Pyrénées (France), and Andalucia (Spain). The estimated transition
168
Session B1: Large scale integrated policy assessment: Tools and applications
probability matrices exhibit typical characteristics with high probability values for staying in
the same farm type as in the period before on the diagonal and lower values which tend to
concentrate around these, i.e. for transitions between size classes within the same
specialization classes. Transitions between specialization classes take mainly place between
various farm types and the mixed farming categories (mixed and mixed livestock). Also,
different mobility schemes between regions can be identified, e.g. with farms in Brandenburg
exhibiting nearly no structural change, whereas in Midi-Pyrénées significantly more and
larger non-zero probabilities can be found at the off-diagonals. In Brandenburg and Andalucia
the total number of farms seems to stay rather stable as the probabilities for both entry and
exit are very close to zero for all farm types. In The Netherlands and Midi-Pyrénées nearly no
market entries, but for most farm types significant large exit probabilities are reported.
In the second estimation step, transition probabilities being equal or greater than 0.01 are
regressed against a number of explanatory variables. The probabilities below 0.01 are
subsumed in a rest category. As explanatory variables apart from the constant and a trend, the
unemployment rate of a country as well as prices of different agricultural outputs are used.
The unemployment rate is assumed to impact especially the exit category (Garvey, 2006),
whereas price developments are mainly thought to have an effect on transitions between farm
types. Due to the non-linear character of the model, the impact of the exogenous variables on
structural change is given in the form of elasticities (Zepeda, 1995). As expected from the
results on the transition probabilities the trend only weakly impacts the transition probabilities
in Brandenburg and Andalucia, whereas significantly larger impacts can be found for The
Netherlands and Midi-Pyrénées. The influence of the unemployment rate on structural change
in the agricultural sector is very weak in Andalucia and weak in The Netherlands. In Germany
the unemployment rate is negatively correlated with exits from the sector which is in line with
the authors’ expectations. So far, the prices for milk, wheat, and pigs have been considered
and proven to be of significance to the analysis. Their impact on the transition probabilities
will be analysed in detail in the paper.
References
Andersen, E., et al., 2006. SEAMLESS Report No. 12, www.SEAMLESS-IP.org, 30 pp.
Garvey, E., 2006. SEAMLESS Report No. 15, www.SEAMLESS-IP.org, 35 pp.
Karantininis, K., 2002. J. Econometrics 107: 275-290.
MacRae, E.C., 1977. Econometrica 45: 183-198.
Stavins, R.N. & B.F. Stanton, 1980. Using Markov Models to Predict the Size Distribution of
Dairy Farms. Cornell University, Ithaca, New York.
Stokes, J.R., 2006. Agr. Resource Econ. Rev. 35: 357-373.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
Zepeda, L., 1995. Europ. Rev. Agr. Econ. 22: 41-60.
169
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme B
State-of-the-art of components for integrated
systems
Session B2: Indicators for integrated assessment
Session organizers:
Maria Luisa Paracchini
JRC, Ispra, Italy
Nadine Turpin
CEMAGREF, Clermont Ferrand, France
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Sustainability Solution Space: An indicator based tool for assessing the
sustainability of agricultural systems1
Claudia R. Binder, A. Schmid, J.K. Steinberger
Social and Industrial Ecology
Department of Geography
University of Zürich
Winterthurerstrasse 190
8057 Zürich, Switzerland
Contact: claudia.binder@geo.uzh.ch
Introduction
This paper introduces the method of Sustainability Solution Space (SSP) for assessing the
sustainability of agricultural systems. Current integrative and indicator-based assessment
approaches in agriculture usually have three main shortcomings: (i) there is an overall focus
on assessing the ecological aspects of agriculture neglecting to some extent economic and
social aspects; (ii) research has so far focused on filling important gaps in knowledge and
technology, but has missed to include the step towards utilization and implementation of this
knowledge; and (iii) the assessment results themselves are difficult to be implemented in
decision-making, as conflicting goals and the interaction between indicators has not been
sufficiently considered. We propose that for filling this gap an approach is needed which
fulfils systemic criteria, i.e., sufficient representation of the system including functional
interaction among indicators, which allows to depict goal conflicts; normative criteria, i.e.,
considering the different value perspectives of stakeholders by including them in the process
and designing sustainability ranges rather then threshold values; and procedural criteria, i.e.
pursuing the assessment in a true transdisciplinary process.
Methods
The core components of the SSP procedure (Wiek & Binder, 2005) are described in Table 1.
Preliminary to constructing an SSP the function the sustainability space has to fulfil has to be
defined (prerequisite phase). Who will use this tool and for what purposes? The
transdisciplinary approach in this prerequisite phase allows for including and balancing the
different views and objectives stakeholders might have.
The method itself consists of a systemic, a normative and an integrative module (Table 1).
The modules are interdependent; constructing an SSP is, thus, not a linear procedure but an
iterative process. The system module is the basis for the sustainability solution space. It (i)
describes and defines the system with its characteristics and its main problems, (ii) derives
indicators (environmental, economic and social), and (iii) determines the relationship among
the indicators. Note that the system module is already constructed in a transdisciplinary
process, i.e. with participation of stakeholders. The normative module sets the criteria for
defining sustainability ranges. It includes both the stakeholder as well as the scientific view.
For each indicator a sustainability range is defined, i.e., a minimum and maximum value is set
according to the selected criteria. The integrative module, finally, integrates the normative
module and the system module. With a computer tool (see below) the sustainability solution
space is calculated. It shows within which ranges the values of the indicators are sustainable
and allow for analysing trade-offs of measures.
Based on the SSP the system of the milk value added chain is described with a set of 17
indicators, 8 ecological (derived from LCA data) and 9 socio-economic. The sustainability
thresholds were obtained through literature research and stakeholder interviews. The
1
Keynote presentation
172
Session B2: Indicators for integrated assessment
relationship among the indicators was developed in a transdisciplinary workshop. The SSP
programme takes a geometric approach to determine the intersection space corresponding to
the satisfaction of the normative ranges while taking into account the functional interactions
of the indicators. We show some results of the sustainability solution space for the Swiss milk
value added chain and discuss the prerequisites, advantages and shortcomings of the method.
Table 1. Steps of SSP adapted to sustainability assessment of agriculture (after Wiek &
Binder, 2005; Schmid, 2008; Binder et al., 2008, 2009)
Step
Description
Prerequisite
Goals setting
Stakeholder involvement
Scale
Module I: Systemic Module
Step 1
Characterizing the region to be assessed
Step 2
Problem-oriented derivation of indicators (e.g., ecological, economic and social)
Step 3
Analysing the inter- and intra-linkages among the indicators as well as their
dynamics
Module II: Normative Module
Step 4
Specifying the sustainability ranges for the indicators
Module III: Integrative Module
Step 5
Defining the solution space for decision making
Step 6
Analysing trade-offs
References
Binder, C.R., et al., 2008. Nachhaltigkeit in der Schweizerischen Milchwertschöpfungskette.
Schlussbericht, Soziale und Industrielle Ökologie, Geographisches Institut. Universität
Zürich, Switzerland, 35 pp.
Binder, C.R., et al., 2009. Analyzing the normative, systemic and procedural dimension of
indicator-based sustainability assessments in agriculture: A review. Submitted.
Schmid, A., 2008. Nachhaltigkeit in der Schweizer Milchwertschöpfungskette:
Transdisziplinäre Untersuchung der sozialen und ökonomischen Dimension mit einer
indikatorenbasierten Systemanalyse und Nachhaltigkeitsbewertung mittels Solution Space
for Sustainability. Master thesis, Social and Industrial Ecology, Department of Geography,
University of Zürich, Switzerland.
Wiek, A. & C.R. Binder, 2005. Environmental Impact Assessment Review 25(6): 589-608.
173
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Model and indicators of sustainable development in rural areas from the
local perspective
M. Adamowicz, A. Smarzewska
Szkoła Główna Gospodarstwa Wiejskiego, Nowoursynowska 166, Warsaw, Poland
Contact: mieczyslaw_adamowicz@sggw.pl
Introduction
In the conditions of decreasing agriculture role in the rural economy and the development of
multifunctional agriculture it seems reasonable to discuss the concept of sustainable
development in relation to whole local arrangements involving both agriculture and its
surroundings. The concept of sustainable development in rural areas requires an application of
appropriate indicators and measurement systems (Borys, 2005). The paper attempts to create a
model of sustainable development in rural areas from the local perspective (NUTS 5) which is
empirically verified on selected rural districts from Lubelskie and Mazowieckie provinces.
The key role in shaping sustainable development can be assigned to the local government, due
to its decision-making rights and its potential to benefit from financial and administrative
instruments, strategic and operational instruments as well as educational instruments. Local
authorities held responsibility for the results of decision made, therefore, they should manage
their resources effectively. Thus, the attempt to conduct the analysis of agriculture and rural
societies should take into account a territorial approach that proves useful while determining a
country’s policy for rural areas. In addition to the implementation of its own tasks, the
economic policy of a district’s self-government should also address any developmental
problems concerning agriculture as rural areas. Local authorities should perceive agriculture
as one of the areas of the rural economy and create conditions for undertaking measures
facilitating non-agricultural economic functions and to create new jobs outside agriculture.
Methods
This work attempts at establishing the model of stable (Turner, 1993) and sustainable
development in local district (Meyer, 2004; Meyer & Elbe, 2005). The aim of creating this
model is to present the district self-government with the possibility of choosing the direction
which leads to achieving the state of a fully-balanced integrated order or partial orders. In
order to present the progress of local governments in implementing the concept of sustainable
development and to assess its level, 30 rural districts from Mazowieckie and Lubelskie
provinces were chosen on purpose to conduct relevant research. As a result of the research
conducted two groups of rural districts that applied the concept of stable and sustainable
development differently were distinguished. All researched districts undertook measures
concerning specific dimensions of sustainable development; however, these measures were
not implemented in a balanced manner. Detailed values of local indicators of sustainable
development created the so called general indicator for sustainable development (GISD) of
the researched rural districts. Based on it, ranks of individual districts were compared with the
entire research sample. General indicators for sustainable development of districts were
organized hierarchically, namely, from 1 to 30 – in reference to the research sample. Rank 1
was assigned to districts with the lowest value of GISD, whereas rank 30 to the district which
achieved the highest value of this indicator.
Results and discussion
The aforementioned discussion proved that for the new concept of sustainable development it
is worth to apply a definition which describes this development as a process of changes, in
which such qualities as stability and sustainability are put into practice and considered
positive, at least within the anthropocentric system of values. The measurement of sustainable
174
Session B2: Indicators for integrated assessment
development of rural districts on the basis of indicators suggested will allow determining the
level of this sustainability (the level of sustainability in limited substitution of capitals or the
level of sustainability in capitals complementarity) in relation to the proposed model of
sustainable development and will enable to assess the efficiency of local self-governments’
activities. The rating process made it possible to distinguish districts at different levels of
development, which were described as relatively below average, relatively average or
relatively above average. Districts with rank from 1 to 10 were defined as districts with
sustainable development level relatively below average, districts with rank from 11 to 20 were
assessed as districts with the average rank of sustainable development and districts with rank
between 21 and 30 were classified as districts with the development level relatively above the
average, in reference to the entire research sample. It should be assumed that sustainable
development measurement should be conducted at both the regional and supra-regional levels.
Too large differences in the values of local indicators of sustainable development of
researched districts concerning the social and institutional, economic as well as environmental
and spatial aspects may prove the level of sustainability between them is low and it may affect
the position of a given district in the GISD ranking. The greater the differences between the
specific aspects of development the more likely a drop in the GISD ranking of the general
indicator of sustainable development for a given district measured at the regional level i.e.
comparing the values of local indicators of researched districts of a given province with the
values of indicators of researched districts from the other province.
Slight differences in the value of local indicators of sustainable development in researched
districts between the aspects of sustainable development determine whether a district would
keep the GISD at the same level or not, both at the local and regional levels (the district
examined remains within the group of a certain level of sustainable development without
either an increase or fall out from a given group of districts). Therefore, it seems necessary to
conduct the measurement process of sustainable development in order to determine its level
and to assess its evolution. It seems reasonable to discuss the concept of sustainable
development in relation to whole local arrangements involving both agriculture and its entire
surroundings. Since it is not enough to conclude that the concept of sustainable development
is being implemented, but it is necessary to make an attempt to measure the level of this
development in a given district on the basis of sustainable development indicators at the local
level. Sustainable development measurement can be interpreted at the regional level i.e. in
reference to researched districts of a given province or at the supra-regional level i.e. in
comparison to researched districts from a different province. The choice of the perspective
may influence the hierarchy of districts in the GISD ranking of districts from researched
provinces. Therefore, it is possible to state that the sustainability of aspects of development
influences whether or not a certain district will keep its position in the GISD ranking and
indicates that the concept of sustainable development can be more effectively applied in these
rural districts as opposed to districts characterized by high value differences between the
aspects of sustainable development.
References
Borys, T. (Ed.), 2005. Wskaźniki zrównoważonego rozwoju (Sustainable Development
Indicators’), Wydawnictwo Ekonomia i Środowisko, Warszawa-Białystok.
Meyer, W., 2005. Evaluation, design and data collection. Part III: Investigating sustainability.
EASY-ECO Training, 27 september 2005, Bratysława.
Meyer, W. & S. Elbe, 2004. Local Network Governance. Perspectives and problems for the
German rural sector, XI World Congress of Rural Sociology – Globalisation, Risks and
Resistance in Rural Economics and Societies, 25-30 July 2004, Trondheim.
Turner, R.K. (Ed.), 1993. Sustainable Environmental Economics and Management: Principle
and Practice. Belhaven Press, New York, London, 389 pp.
175
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A structured set of indicators for integrated assessment of future
agri-environmental policies
C. Bockstaller, N. Turpin, L. Stapleton, M. Van der Heide, O. Therond, T. Pinto-Correia,
V. Voltr, M. Raley, I. Bezlepkina, J.P. Bousset, J. Alkan Olsson, F. Ewert
INRA, UMR Nancy-Université - INRA Agronomie et Environnement Nancy-Colmar,
BP 20507, 68021 Colmar Cedex, France
Contact: christian.bockstaller@colmar.inra.fr
Introduction
Ex-ante integrated impact assessment of new policies is a prerequisite for them to efficiently
support sustainable development (SD). Recently, SEAMLESS Integrated Framework has
been developed to assess ex-ante impacts of agricultural and agri-environmental policies and
technologies on agricultural systems across a range of scales, from field–farm to region and
the European Union (Van Ittersum et al., 2008). The aim of this paper is to present the set of
sustainability indicators developed within the SEAMLESS project.
Methods
An indicator list was developed within the SEAMLESS project which is structured and
presented through a new indicator framework, i.e. a goal-oriented indicator framework
(GOF). This framework covers a broad range of themes linked to the three main dimensions
(environmental, economic, social) of sustainability, and generic themes across the three
dimensions (Alkan Olsson et al., 2009), for two domains; the sustainability of agriculture
itself and the impact of agriculture on the rest of the world, i.e. on SD. Three objectives
underpinned the development of the SEAMLESS-IF indicator list across scales: (i) to provide
policy-makers and stakeholders with indicators which they usually use and/or which they
would like to use; (ii) to ensure scientific soundness of SEAMLESS-IF indicators, i.e. their
relevance to represent impacts at stake; (iii) to cover the various themes in each dimension of
the GOF (see Table 1). Within SEAMLESS-IF indicators are primarily assessed by models
(and model chains) and thus their development has been constrained by the nature of the
available model outputs. Outputs from three main models integrated in SEAMLESS-IF are
used for the indicator calculation: the agricultural sector model SEAMCAP; the farming
system model FSSIM; and the cropping system model APES. However, despite the range of
scales covered by the SEAMLESS-IF model chains some key indicators can currently not be
assessed directly from model outputs. However despite the high range of scales covered by
the SEAMLESS model chains some of key indicators cannot currently be assessed at certain
scales using model outputs. To address this problem generic upscaling procedures has been
developed and associated to each indicator that needs to be upscaled.
Results and discussion
Examples of indicators are shown in Table 1. Across scales a total of 80 environmental, 140
economic and only 11 social indicators are or are about to be integrated into SEAMLESS-IF.
This new structured set of indicators offered by SEAMLESS-IF enables a multi-scale
integrated assessment of SD from the farming systems to the agri-environmental zones and
the EU level. In comparison with many former initiatives the broad spectrum covered and the
type of the proposed indicators allows for a deeper analysis of environmental pressures and
impacts, economic costs and benefits and socio-demographic dynamics. For example, through
the integration of the APES model, indicators assessing emissions like nitrate leaching can be
calculated considering key processes, which is not the case for simple indicators describing
farmers’ practices like nitrogen use (Bockstaller et al., 2008). However, this requires a
detailed description of fertilization and pesticides management for a given area. Another
176
Session B2: Indicators for integrated assessment
example is the assessment of economic indicators at NUTS2 level with two related model
chains, that enables capturing complementary impacts of policy options, Social indicators in
this list were derived from economic data, on labour and income distribution since no social
model is, until now, integrated in SEAMLESS-IF.
Table 1. Example of environmental indicators within the goal-oriented indicator framework
(GOF) at different scales (farm, normal font; Nuts 2, italic; member state or EU level, bold).
Domain 1
Impacts on the agricultural sector
Dimension of sustainable development
Themes
Domain 2
Impacts on the rest of the world
Dimension of sustainable development
Environmental
Economic
Social
Environmental
Pesticide use
Net farm income
Equity
Nitrate leaching
Equity
Percent of
subsidies in farm
income
Equity
Pesticide leaching
Equity
Monetary
poverty rate
Crop diversity
Percent of
subsidies in farm
income
Ultimate goals
Agricultural
income
Economic
Social
Percent of area
with high leaching
Nitrate surplus
Processes for
achievement
Soil Org.Mat.
change
Direct
payments
Labour use
P balance
Direct
payments
N2O emissions
Productivity of
farm inputs
Volatization
First pillar CAP
expenditure
Total labour
use
NH3 emissions
export subsidy
outlays
Potential
employment
P balance
profit of the agr.
processing
industry
N2O emissions
Terms of trade
Value of farm
production
Soil erosion
Means
Water use
by irrigation
Energy use by
min. fertilizer
Use of mineral P
Share of animal
production
Share of animal
production
Share of animal
production
Total costs
Labour use
Labour use
Labour use
Soil erosion
Water use
by irrigation
Energy use by
min. fertilizer
Use of mineral P
Land shadow
prices
Land value
Fairness
Labour
use
Labour
use
Labour
use
Conclusion
The SEAMLESS-IF multi-scale approach with its explicit upscaling procedures, as well as
the integration of the indicators into a generic flexible software system linked to a large
database mark an important progress with respect to the creation of an efficient set of
indicators to assess the sustainability of future agri-environmental policies. However, some
methodological issues remain unclear, such as the determination of reference values and the
aggregation of indicators into composite indices. For the latter, methods have been explored
(Bockstaller et al., 2009). Furthermore, there are still themes not covered by the GOF, e.g.,
impacts on biodiversity, and only few indicators are available representing the social
dimension. However, as SEAMLESS-IF is a flexible system further extension of the indicator
list is possible through the integration of new models.
References
Alkan Olsson, J., et al., 2009. Environ. Sci. Pol. (in review)
Bockstaller, C., et al., 2008. Agron. Sustain. Dev. 28: 139-149.
Bockstaller, C., et al., 2009. AgSAP Conference 2009, these proceedings.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
177
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
From a disaggregated set of indicators to a synthetic, composite assessment
of sustainability: Paths and pitfalls
C. Bockstaller, L. Stapleton, M. Van der Heide, G. Geniaux, S. Bellon, E. Josien,
M. Raley, J. Alkan Olsson
INRA, UMR Nancy-Université - INRA Agronomie et Environnement Nancy-Colmar,
BP 20507, 68021 Colmar Cedex, France
Contact: christian.bockstalller@colmar.inra.fr
Introduction
SEAMLESS-IF endeavours to capture the complexity of sustainable development by
incorporating a wide range of indicators grouped along three dimensions (Van Ittersum et al.,
2008). Presenting outputs corresponding to disaggregated indicators in a table or a figure can
be considered as a preliminary step in the analysis of a given action or, rather, a policy in the
context of the SEAMLESS project. However, the question of aggregation arises when one
intends to conclude about the sustainability of a policy, or furthermore to compare two or
more policy options via a set of indicators. This contribution reviews different methods of
combining diverse information in an explicit, consistent and transparent way, and presenting
it in an easily intelligible form to facilitate ex-ante policy evaluation.
Aggregating a set of indicator values into a single, composite value
The most common and intuitive approach is to combine different source of information into a
single value, e.g., indicator scores into a global index or composite indicator. This is in many
cases calculated by means of a sum or a weighted mean (Rosnoblet et al., 2006). Some
assessment methods deliver a single score resulting from the sum of scores without an explicit
standardization of the single indicator values. This kind of approach presents several
methodological flaws like the risk of adding apples and pears. Several possible techniques for
normalization exist: linear scaling techniques, Gaussian normalization, distance to target,
ranking by experts, categorical scales, etc. (Geniaux et al., 2005). Another approach is to
convert all value into the same unit, monetary or physical (e.g., Ecological Footprint).
Aggregation methods based on a common monetary unit like in cost-benefit analysis raise the
issue of how to value non market goods and services like environmental assets, water quality,
biodiversity, etc. (Van der Heide et al., 2009).
Multi-criteria analysis
A serious drawback of using a single composite indicator is the loss of information through
the aggregation and hidden compensation. A possible solution to these problems is multicriteria analysis (MCA. MCA has its roots in management science and operational research
and is a methodology for selecting between or prioritizing different options described by a set
of criteria. Central to MCA is: (1) some form of criteria or scale for selecting between or
prioritizing different options; (2) a table to show the performance of the different options
relative to each other (Table 1). This is based on a weighting, rating or ranking procedure by a
group of experts or stakeholders (Josien et al., 2006); (3) statistical analysis or some
alternative methods for drawing conclusions or highlighting the key findings from the MCA.
Such analysis lays on a multitude of very different approaches, e.g., a utility function in the
case of a compensatory approach (allowing compensation between criteria) or pair-wise
comparisons like, for example, outranking approaches which are mostly non compensatory. In
this case, the number of criteria has to be limited.
Multi-criteria assessment based on a qualitative method (dashboards)
Qualitative approaches can also be considered as a way to aggregate. These types of
178
Session B2: Indicators for integrated assessment
approaches lead to a conclusion in the form of a quantitative value, or as classes of a given
criteria (e.g., sustainability). Technically, therefore, the dashboard approach can be considered
as a hybrid approach combining qualitative and quantitative elements. Such approaches are
based on decision rules expressed in “if then” language, presented either as decision trees
based on qualitative multi-attribute decision modelling or in the form of a dashboard (Figure
1). The number of criteria included in the analysis can be increased when those are structured
in a hierarchical tree.
Table 1. Hypothetical data on ranks and rates of economic themes by a group of experts.
Expert
Theme1
Rank (between 1 and 9)
Rate (total 100)
1
Viability
1
50
1
Performance
3
30
1
Capital
5
20
2
Viability
3
40
etc.
1
From the goal-oriented indicator framework (GOF) developed in the SEAMLESS project
(Alkan Olsson et al., 2009)
DASHBOARD
Tree diagram
Decision
rules
Pilot
gauge
Intermediate
gauge
Basic gauge
Warning
levels
Ind1, Ind 2, … Indi
Indicators
Figure 1. Overview of the ‘dashboard’ concept according to Girardin et al. (2005).
Conclusions
The MCA approach allowing a transparent weighting procedure as shown in Table 1 can
serve as a basis to a more qualitative approach using a dashboard presentation. This approach
provides a presentation of aggregation results in an easily intelligible form. However,
implementation in an assessment framework like SEAMLESS-IF remains open because it
requires additional software development work, which is beyond the means of the project.
References
Alkan Olsson, J., et al., 2009. Environ. Sci. Pol. (in review)
Geniaux, G., et al., 2005. PD 2.2.1 SEAMLESS project, pp. 148.
Girardin, P., et al., 2005. Indicateurs et tableaux de bord. Guide pratique pour l’évaluation.
Lavoisier, Londres, Paris, New York, 39 pp.
Josien, E., et al., 2006. PD 2.6.1 SEAMLESS project, 42 pp.
Rosnoblet, J., et al., 2006. In: Eds M. Fotyma & B. Kaminska, 9th ESA Congress, Warsaw,
Poland, pp. 707-708.
Van der Heide, M., et al., 2009. SEAMLESS book. (in review)
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
179
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Contextualization of on farm ex-ante evaluation of the sustainability of
innovative cropping systems in viticulture, using a multiple criteria
assessment tool (DEXi)
S. Delmotte, C. Gary, A. Ripoche, J.M. Barbier, J. Wery
INRA, UMR System, 2 place Pierre Viala,
34060 Montpellier Cedex 2, France
Contact: gary@supagro.inra.fr
Introduction
Grape producers face an economic and environmental crisis, the latter relating to the generally
high use of pesticides in vineyards. In relation to them, various stakeholders expect some
environmental improvements. Researchers are proposing novel cropping systems that should
be evaluated ex ante with respect to their contribution to sustainable development, meaning
both their environmental impacts and their economical and social adoptability by farmers.
Farms producing grape are very diverse in terms of size, soils, availability of staff and
equipment, and objectives of production (wine grade). In this context, we hypothesized that
the evaluation of a candidate cropping system should be carried out in the farm context,
taking into account information in relation with its adoptability.
Methods
To evaluate and compare cropping systems, Multiple Criteria Decision Aiding (MCDA)
methodologies seem to be relevant (Sadok et al., 2008). A decision support tool called DEXi
(Bohanec, 2008) was adopted; it enables to design decision trees based on a hierarchy of
criteria. The qualitative classes (such as ‘low’, ‘medium’ and ‘high’ for example) of criteria
are aggregated into a single note, which is in the present case the contribution of the cropping
system to sustainable development. It is necessary to define scales to convert values of
indicators into classes of criteria. The criteria are then aggregated using ‘if… then…’ decision
rules to obtain final classes for the overall criteria. Some criteria and indicators were derived
from a list proposed for field crops (MASC v.1.0, 2008) and adapted to viticulture by a group
of four experts. Two theoretical farms were defined, with contrasting production objectives
(low vs. high grade wine) and availability of labour and equipment. A process of aggregation
was then proposed to take into account these characteristics in the evaluation, and several
cropping systems were compared.
Results
The impact of cropping systems on environment was evaluated through five attributes (Figure
1): the pressure on biodiversity, the energy use and the impacts on the soil, water (both
surface and below ground) and air compartments. The economic adoptability of cropping
systems was evaluated through (i) the satisfaction of production objectives on average, and
the stability of production over the years, (ii) the total cost of implementation of the cropping
system. It indirectly reflected the efficiency of production and the productivity of labour. For
the social and human dimensions of sustainability, four attributes were considered: the
difficulties to implement the candidate system, the task overlap, the health risk, the social
recognition and the free time left to the farmer.
One parameterization of the decision tree was obtained for each farm context. It differed
on five points: (i) the calculation of the indicators of performance depended on the yield and
quality objectives of the farmer, (ii) some of the indicators were calculated in reference to the
actual system practiced by the farmer (iii) some qualitative criteria could take values that
reflected the opinion of the farmers (Figure 1 in grey), (iv) the scale to convert an indicator
value into a class of criteria could be fixed by them and reflect their objectives and
180
Session B2: Indicators for integrated assessment
constraints, (v) the aggregation of the criteria had to be set up by stakeholders for the
environmental criteria and by the farmers for the social and economics criteria, in order to
represent their priorities among the attributes (Figure 1 in dotted lines).
Figure 1. Splitting up of the sustainability of cropping systems into environmental,
economical and social/human criteria.
It resulted that the economical and social performances of some cropping systems differed
among the two farms (e.g., integrated farming with non-permanent cover cropping had a
lower social score and a higher economical score in the farm with low availability of labour
and equipment).
Discussion
In the present research, we aimed at developing a tool that could be used by extensionists,
farmers and stakeholders. The contextualization of the evaluation of cropping systems was
tested for the two theoretical farms by using role games with experts, mainly researchers and
extensionists. It proved to be powerful for discussing the decision tree and considering the
local and farming contexts. The evaluation outputs appeared sound to the experts, particularly
with respect to the adoptability of innovations, that is a crucial dimension of sustainability.
The method must now be tested in a real case and participatory methods must be chosen.
We propose to consult stakeholders of the area where the evaluation takes place to define the
utility functions for the environmental criteria. Independently, farmers would be consulted to
define scale and utility functions for social and economical criteria.
Acknowledgements
The authors thank A. Biarnès, L. Bouchet and C. Clipet for their expert contribution.
References
Bohanec, M., 2008. DEXi: A Program for Multi-Attribute Decision Making, version 2.0.
http://www-ai.ijs.si/MarkoBohanec/dexi.html.
MASCV1.0, 2008. IDDN.FR.001.040014.000.R.P.2008.000.30100.
Sadok, W., et al., 2008. Agronomy for Sustainable Development 28: 163-174.
181
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing the environmental sustainability of the CAP:
A DPSIR framework approach
G. Giannoccaro, J.M. Pistón, J. Berbel
ETSIAM – University of Cordoba, Dpt. of Agricultural Economics,
Sociology and Policy, Campus de Rabanales Edificio C5 14014 - Spain
Contact: es2gigig@uco.es
Agriculture plays only a small part in the economies of European Union (EU) member
countries, accounting for about 2% of GDP and 5% of EU employment. But in terms of its
impact on the environment and natural resources, the role of agriculture is more significant,
accounting for 45% of EU total land use and over 30% of total water use (OECD, 2001a). In
addition, agriculture in the rural areas is the major beneficiary of the Common Agricultural
Policy (CAP) in most EU countries; therefore, environmental concern about resource
depletion, and conservation of biodiversity, habitats and landscapes, inevitably involve
environmental sustainability issues (EC, 2001).
Assessing the CAP impact might help in re-addressing the CAP in the wider framework of
EU environmental objectives.
There are currently numerous scientific works on the assessment of environmental impacts
in relation to the agricultural activity, but most of them come from the researches published
by OECD (OECD, 2001b) and they deal with indicator definition and calculation. However,
the indicator set definition for sustainability assessment has become a high priority both in
scientific research and policy agendas.
This research will focus on the development of the environmental sustainability evaluation
framework, proposing a consistent and comprehensive methodology referred to as the DPSIR
framework (EEA, 2001), in order to study the interactions between CAP and environmental
issues in rural areas. The approach consists of taking into account all environmental themes
(soil, air, water and biodiversity) involving the agricultural system in rural areas. Each
environmental theme will be accounted according to the DPSIR model, forces, pressures,
state, impacts and responses (i.e. CAP supports, both the first and the second pillar). Stress
will be made on the trends in agricultural impacts in rural areas and the influence of CAP on
environmental sustainability (Table 1).
Further, by some deep interviews, we will attempt to highlight the changes in farmers'
behaviour in the event of CAP support modification. To achieve this, a preliminary
application case will be undertaken in Andalusia (Spain).
Table 1. DPSIR design for CAP environmental sustainability assessment.
Agricultural systems
Environmental
themes
Soil
Water
Air
Force
Pressure
State/Impact*
Response
From scientific
documents
highlighting the
relationship
between
environment
and agriculture.
Official reports
about pressures
(see OECD,
2008) focusing
on the main
problems within
rural areas, and
interview results.
Selected
indicator values
(time series and
threshold
values). Official
data.
CAP response
definition and
budget trend
amounts. Official
data.
Interview results.
Farmers’
behaviour
recognition.
Biodiversity
*
Because of the restricted framework on farming activities, the impact of the agricultural sector is ineffective.
182
Session B2: Indicators for integrated assessment
Spatial effects are the crucial aspect of environmental sustainability issues. Essentially,
there is a horizontal and a vertical component. The horizontal component depends on the
scale of application limited to the plot, farm or landscape (watershed, region or state),
respectively. On the contrary, the vertical component refers to the administrative issues.
We propose an approach that takes into account the main agricultural systems in rural areas
as the horizontal component, and aggregation at NUT 3 level as the vertical component,
focusing on household farms in rural areas.
The framework is restricted to on-farm activities (only cultivation practices) of the
production cycle. This means that impacts caused by off-stream activities such as transport,
food transformation and packaging are not accounted for.
References
EEA, 2001. Environmental Signals 2001. European Environment Agency, Copenhagen.
European Commission, 2001. Statistical Information needed for Indicators to monitor the
Integration of Environmental concerns into the Common Agricultural Policy. COM 144
final.
OECD, 2001a. Measuring the Environmental Impacts of the Common Agricultural Policy:
Challenges, Recent Trends and Outlook, and Future Directions. The Common Agriculture
Policy and the Environmental Challenge - New Tasks for the Public Administrations?
Maastricht, 14-15th May 2001.
OECD, 2001b. Environmental Indicators for Agriculture, Volume 3 - Methods and Results.
OECD, Paris.
OECD, 2008. Environmental Performance of Agriculture in OECD Countries since 1990.
OECD, Paris.
183
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Agri-environmental nitrogen indicators for EU27
A. Leip, F. Weiss, W. Britz
European Commission – DG Joint Research Centre,
Institute for Environment and Sustainability,
TP.050, 21014 Ispra (VA), Italy
Contact: adrian.leip@jrc.it
Introduction
Nitrogen is a key element to ensure modern agriculture’s output, sustaining the live and
lifestyle of billions of people. But nitrogen accounts also for key environmental problems that
challenge the well functioning of today’s societies. One molecule of nitrogen can contribute
to one or many environmental problems, including eutrophication, groundwater pollution,
climate change, and may affect human health via ozone formation or biodiversity via nitrogen
deposition on natural areas. This multiple impact of nitrogen is often referred to as the
‘nitrogen cascade’ (Galloway et al., 2003).
Accordingly, nitrogen plays an important role in several agri-environmental indicators
such as NH3 emissions, fertilizer N impact, gross N surplus, nitrates in water, GHG emissions
(EEA, 2005). Many of these indicators, however, are so far calculated independently and with
sometimes contradicting data sources (see, e.g., Grizzetti et al., 2007). This includes also the
first overview of the “European Nitrogen Case” that was presented by Van Egmond et al.
(2002) at the second International Nitrogen Conference held in Potomac (USA). Thus, a
system that calculates the detailed nitrogen balance and the related indicators for agriculture
in Europe on the basis of consistent data sets and advanced methodologies is highly desirable.
A closed balance of nitrogen is calculated in the CAPRI model, i.e., next to monetary
values and product balances, also the nutrient fluxes are in accordance with the law of massconservation (Britz et al., 2007). We present national nitrogen budget and related indicators
for the agricultural sector in EU27, but also for individual European countries, on the basis of
the CAPRI database for the year 2002.
Methods
In CAPRI, different parts of the agricultural sector are linked by the flow of (mass and)
nitrogen: the crop sector receives manure nitrogen from the livestock sector in the exchange
of animal feed; the animal sector receives feed and concentrates also from the agricultural
market and sells products for processing and consumption; the industry produces synthetic
fertilizer as major nitrogen input to agricultural soils that produce food and fiber for societal
use. Nitrogen losses occur both in livestock production system and from agricultural soils.
The nitrogen balance of the livestock sector is closed by estimating manure nitrogen excretion
as the difference from nitrogen intake with feed and nitrogen output (or retention) in products;
the soil nitrogen balance is closed by estimating soil nitrogen surplus from total nitrogen input
and quantified nitrogen output. Both manure excretion and nitrogen surplus are cross-checked
by independent data sources; soil surplus is split into nitrate leaching, NH3 volatilization,
denitrification (N2O and N2 emissions) using emission factors. The main nitrogen fluxes thus
determined are schematically shown in Figure 1.
Results and discussion
The largest N-input to EU27 agriculture (not considering Cyprus and Malta) is the application
of mineral fertilizer (11.3 Mt N), with almost equal amount coming from the livestock sector
(9.3 Mt N as manure). Only 14% of this nitrogen is available for human consumption. Net
import amounts to only 2% of the total N productivity (feed and food). The situation is very
different in the Netherlands, where a net N import of almost 200 kt N adds to an input of 290
184
Session B2: Indicators for integrated assessment
kt N in mineral fertilizers and 440 kt N in manure, reflecting the importance of the animal
sector in the Netherlands, which depends to a large degree on the import of feed concentrates.
This short selection of examples shows that the CAPRI modelling system is a powerful
tool for deriving nitrogen related agri-environmental indicators. These become even stronger
if downscaled (Leip et al., 2008; Britz & Leip, 2009) to the regional or watershed level.
Figure 1. Schematic representation of N-flows to and from the agricultural system.
References
Britz, W. & A. Leip, 2009. Development of marginal emission factors for N losses from
agricultural soils with the DNDC-CAPRI meta-model. Agriculture Ecosystems &
Environment. (submitted)
Britz, W., et al., 2007. Description of the CAPRI Modeling System. Final report of the
CAPRI-DynaSpat Projec. Institute for Food and Resource Economics, University of Bonn,
Bonn, Germany.
EEA, 2005. Agriculture and Environment in EU15 - the IRENA indicator report, EEA Report
No. 6/2005, Copenhagen: European Environment Agency, 128.
Galloway, J.N., et al., 2003. BioScience 53(4): 341-356.
Grizzetti, B., et al., 2007. Spatialised European Nutrient Balance, EUR 22692 EN,
Luxembourg: Office for Official Publications of the European Communities.
Leip, A., et al., 2008. Biogeosciences 5(1): 73-94.
Van Egmond, K., et al., 2002. Ambio 31(2): 72-78.
185
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Sustainability of Flemish farms: Advising farmers and policymakers
M. Meul, S. Van Passel
University College Ghent, Voskenslaan 270, B-9000 Ghent, Belgium
Contact: marijke.meul@hogent.be
Introduction
Today, ‘sustainability’ has rightfully gained its place in the vision, mission and strategy of
companies, organizations and governments, also in agriculture. However, putting the
theoretical concept into practice often proves to be very difficult. Hence, the objective of this
paper is to address the ‘sustainability-paradox’ between intention and action, by presenting
two complementary methods that are helpful to advise both farmers and policy makers
concerning sustainability of Flemish farms.
Methods
During the last decade, there has been an explosion of activity to develop sustainable
development indicators, in order to determine whether sustainable development is actually
being achieved. Hereby, two major approaches can be distinguished: (i) a set of indicators
listed or presented together within a single table or diagram (visual integration) and (ii) a
single, composite index of sustainability (numerical integration). Each approach has its pros
and cons and one has to choose a particular approach depending on the specific goal and
intended use. We present two approaches in this paper: a visual integration approach
(MOTIFS) and a numerical integration approach (the sustainable value approach). Both
approaches have already proven to be useful to assess the farm sustainability of Flemish dairy
farms.
MOTIFS
MOTIFS (Meul et al., 2008a) is an indicator-based sustainability monitoring tool for Flemish
dairy farms. It allows us to monitor farm progress towards integrated sustainability, i.e. taking
into account economic, ecological as well as social aspects, using a set of relevant indicators.
The tool offers a visual aggregation of indicator scores into an adapted radar graph,
considering ten sustainability themes related to ecological, economic and social aspects.
Sustainable Value approach
The sustainable value approach is developed by Figge & Hahn (2004) and it shows in
monetary terms the value that a company creates or destroys by the use of a set of different
resources. A positive value contribution indicates that a resource is used in a value-creating
way by a company. To determine how much value is created by the entire bundle of
resources, the sustainable value can be calculated by summing up all value contributions and
by dividing this value by the number of resources. More information of an application of the
sustainable value approach for Flemish farms can be found in Van Passel et al. (2007).
Case-studies
MOTIFS has been applied on 20 Flemish dairy farms participating in a Leader+ project
‘Strong with Milk, 2006–2008’ with the aim to monitor sustainability and stimulate
communication and exchange of knowledge between farmers. For a number of selected
sustainability themes, farmer discussion groups were organized, in which the MOTIFS-results
were discussed.
The sustainable value approach is currently applied in a European funded project (within
the 6th FP) called SVAPPAS, with a major aim to develop and adapt the sustainable value
methodology for the assessment of sustainability at farm, sector, national or cross-national
186
Session B2: Indicators for integrated assessment
level. Besides, the methodology is tested for different case study areas and the data needs and
data availability to arrive at a standard approach for sustainability analysis of farms and farm
policies are assessed.
Discussion and conclusions
Both methods can and should be used complementary to assess sustainability performance of
Flemish farms and to advise both farmers and policymakers on sustainability aspects. This is
shown in Figure 1.
Resources: labour, land, capital, nutrients, energy, water, biodiversity…
Sustainable value approach
arable
dairy
pig
sector x
Comparing sectors
farm 1
farm 1
farm 1
farm 1
Follow up evolutions within sectors
farm 2
.
.
.
farm 2
.
.
.
farm 2
.
.
.
farm 2
.
.
.
farm x
farm x
farm x
Advising policy makers
…
MOTIFS
Comparing farms within 1 sector
farm x
Follow up evolutions of farms
Advising farmers
Products and services: milk, meat, fruit, cereals, nature, care,…
Figure 1. Complementary use of the sustainable value approach and MOTIFS for
sustainability assessment and advice to farmers and policymakers within the Flemish farming
sector.
MOTIFS is designed to guide farmers towards a higher level of sustainability. The visual
integration of relevant themes of ecological, economic and social sustainability aspects and
sustainable entrepreneurship, allows an immediate and integrated interpretation of a farm’s
overall sustainability level and gives an overview of the farm’s strengths and weaknesses.
MOTIFS was found particularly interesting to be used in a discussion group of farmers to
mutually compare results and exchange knowledge and expertise (Meul et al., 2008b).
Moreover, by using the monitoring system to compare farm performances of an individual
farm over time, the farmer can follow-up whether management actions actually result in the
aimed effect. This makes MOTIFS a useful management tool for farmers.
The sustainable value integrates sustainability aspects in a numerical way. The approach is
extremely suitable to support decision makers in their selection of good resource users and to
follow up structural or sector evolutions. Policymakers can be informed on which are good
and bad performing sectors, so they may e.g. decide to help bad performers to improve their
sustainable resource use. The method can also be used to identify major characteristics within
a specific sector that influence the sustainability performance of the related farms. For
example, Van Passel et al. (2007) found that both structural and managerial characteristics
have an impact on the sustainable value of Flemish dairy farms.
References
Figge, F. & T. Hahn, 2004. Ecological Economics 48: 173-187.
Meul, M., et al., 2008a. Agronomy for Sustainable Development 28: 321-332.
Meul, M., et al., 2008b. Ecological indicators. (in press, available online)
Van Passel, S., et al., 2007. Ecological Economics 62: 149-161.
187
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Rural regions in Europe:
A new typology based on a multi-sectoral point of view
J. Scholz, S. Herrmann
Institute for Environmental Planning, Leibniz University Hannover,
Herrenhäuserstraße 2, 30419 Hannover, Germany
Contact: scholz@umwelt.uni-hannover.de
Introduction
The situation of rural areas in Europe is strongly discussed. Old recipes are no longer valid to
cope with current problems like demographic change, decreasing importance of agriculture or
impact of global change. Therefore, new approaches are needed to direct rural policies to
support the specific endogenous potential of rural areas.
One major objective of the recently started EU FP 7 project ‘RUFUS’ (Rural future
Networks, www.rufus-eu.de) is to generate a tool to specify the different dimensions of
potentials (social, economic, ecological) of rural regions. This characterization tool consists of
a typology of rural regions based on an indicator set developed by an interdisciplinary
methodology. By means of the typology, different types of rural regions with specific
characteristics concerning their development potential are derived.
This typology is framed by two additional project parts. First, a policy baseline assessment
delivers the basis for the typology concerning the policy sectors which have an impact on
rural areas (Nadin & Van Nes, 2008). On the other hand, the results of the typology serve as
basis for recommendations for integrated, multi-sectoral rural policy. Figure 1 shows the
relationships between these three investigation steps. The derived types will be reviewed by
regional stakeholders in case studies for each project country.
Figure 1. Workflow of RUFUS typology.
Extending existing rural typologies (e.g., Boscacci et al., 1999; OECD 2006) RUFUS is
targeted on a typology with a multi-sectoral (policies), multi-disciplinary approach for
indicator selection. The aim is to derive a set of indicators to draw a coherent picture of rural
areas in Europe. A set of qualitative and quantitative indicators is used to cluster European
regions regarding common characteristics of current state and development potentials. The
main objective is to highlight the diversity of rural regions among themselves, not the
comparison between rural and urban regions.
Working results of the first year of the RUFUS project will be presented, in particular the
first types of the RUFUS typology.
188
Session B2: Indicators for integrated assessment
Methods
The methodology consists of the following steps.
I. Basic settings
The typology is based on available data sets. No primary data selection was initiated. Most of
the data are taken from existing data bases like ESPON, EUROSTAT, CORINE Land Cover
and Landscan. Furthermore, additional indicators, especially social and ecological data sets,
were provided by each project country. Therefore, the project is able to extend general
indicator sets by regional information.
The reference units are the NUTS 3 regions of the project countries. The focus of the
typology is centered on the project countries: the Netherlands, United Kingdom, Sweden,
France, Hungary, Italy, Poland, Portugal and Germany.
II. Indicator selection
On the basis of the policy baseline assessment (Figure 1) selection criteria are derived for the
indicator selection. Two important selection criteria are the policy relevance and a balance of
social, economic and ecological indicators. Innovative indicators, especially for development
potentials and for social issues, have to be derived. Furthermore, the selection of indicators is
also strongly related to the analysed countries because the typology should reflect national
and regional differences. To reduce the number of indicators, statistical techniques are applied
(correlation value). The final indicator set is used by the second statistical process: the cluster
analysis.
III. Cluster-analysis
By using multi-variate statistical techniques (cluster analysis) the regions (NUTS 3) of
selected European countries are clustered by common indicator characteristics. By varying the
indicator set, different cluster sets can be developed. A hierarchical cluster analysis (Backhaus
et al., 2008) will be performed.
IV. Typology
Each cluster will be defined as a specific type by interpreting and describing their
characteristic aspects and their capability to reveal development potentials. Finally each type
should reflect the multi-sectoral approach to serve as basis for policy recommendations.
Results and discussion
The initial results of the RUFUS typology procedure so far are: the development of selection
criteria, the adaptation of the statistical methodology and the derivation of types, based on a
sustainable set of indicators.
Within the presentation different examples of the new types will be described. The
appropriateness of these types as well as pros and cons of the multi-sectoral approach will be
discussed. The question how typologies can serve as basis for policy recommendations will
also be answered.
References
Backhaus, K., et al., 2006. Multivariate Analysemethoden. Eine anwendungsorientierte
Einführung, Berlin.
Boscacci, F., et al., 1999. A typology of rural areas in Europe, Milan.
OECD, 2006. The new Rural Paradigm. Policies and Governance, Organisation for Economic
Co-Operation and Development, Paris.
Nadin, V. & A. Van Nes, 2008. Position Paper. Baseline assessment. (unpublished)
189
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme B
State-of-the-art of components for integrated
systems
Session B3: Cropping systems modelling
Session organizers:
Jim Jones
University of Florida, Gainesville, USA
Mike Robertson
CSIRO, Sustainable Ecosystems, Australia
Graham Russell
University of Edinburgh, UK
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Cropping systems modelling: Past performance and future prospects1
B.A. Keating, M.J. Robertson, P.S. Carberry, D.P. Holzworth, N.I. Huth
CSIRO Sustainable Ecosystems, 306 Carmody Rd., St Lucia 4067 QLD Australia
Contact: Brian.Keating@csiro.au
Introduction
Cropping systems models integrate climate, soil, and crop characteristics to simulate crop
growth and productivity, resource use, and changes in soil, nutrient and water characteristics.
In more elaborate modelling systems, pastures, animals and forests might feature as components of multi-paddock simulators. In this paper, we review developments in cropping
systems models and their evolution from simpler crop models over the last 25 years. We trace
the history of key modelling efforts in Europe, North America and Australia. In particular we
explore:
(a) Developing scale in cropping systems modelling efforts;
(b) The evolving focus from crop models to cropping systems simulators;
(c) Innovation in construction of cropping systems models;
(d) The evolving focus for model application.
Developing scale in cropping systems modelling efforts
The development and wider application of many crop models did not always extend far
beyond the original authors, reflecting both the incentives that existed for publishing new
models and oftentimes the difficulties in transferring models supposedly ‘validated’ on a
limited set of data to different environmental or management conditions. There were
exceptions to these isolated model development efforts and there are now a small number of
cropping systems simulators that dominate applications in the literature (Table 1). These
include the DSSAT-ICASA effort (based on the CERES and GRO models initially) coming
out of the USA, which did much to champion the cause of crop and cropping systems models
worldwide, the Wageningen models (e.g., SUCROS, MACROS, ORYZA, WOFOST) and
perhaps as importantly, the students from around the world that studied crop modelling and
systems analysis at Wageningen University in The Netherlands, CROPSYST from Italy but
with a broad international participation and the APSIM modelling system from Australia.
Evolution from crop models to cropping systems simulators
We have seen an evolution from crop models that focused on yield prediction of individual
crops in response to genetics and resource supply to cropping systems models that focused on
the systems level interactions between crops, pastures, soils, environment and management.
The early crop models on the 1980s (e.g., CERES, GRO models) invested in sufficient
physiological detail to predict crop yield response to resource supply (water, nutrients) and
weather, as modified by management decisions and genetic characteristics. There were
‘systems simulators’ available at this time (e.g., EPIC, CENTURY) but they lacked
physiological detail and were limited in their simulation of crop yield in response to
management and environmental factors. By placing the focus of cropping systems simulators
on the soil (not the crop) but retaining the crop physiological detain and management
responsiveness in crop modules to the systems simulator.
Innovation in construction of cropping systems simulators
The quality and robustness of the computer code in the early models was variable but
generally poor. As often as not, crop physiologists taught themselves FORTRAN and went
1
Keynote presentation
192
Session B3: Cropping systems modelling
Table 1. Frequency of web-based references for different modelling efforts using the search
terms, ‘crop model’ and ‘cropping systems model’ (Google search engine 1/1/2009).
Modelling Effort
CERES
DSSAT
APSIM
ORYZA
MACROS
WOFOST
CROPSYST
Crop model
3930
2340
1110
989
228
819
495
Cropping systems
model
207
117
507
47
9
21
209
Reference
Jones & Kiniry (1986)
Jones et al. (2003)
McCown et al. (1996)
Bouman et al. (2001)
Penning de Vries et al. (1989)
Van Diepen et al. (1989)
Stöckle et al. (2003)
off and ‘hacked’ code. This was a viable strategy for simple crop models but quickly became
untenable in the early 1990s as crop models were morphing into increasingly complex
‘cropping systems’ models. We can identify three forms of innovation relevant to cropping
systems simulator construction over the last 25 years. These are:
• Software engineering innovation - developments in terms of modularity and protocols for
inter-module communications, multi-language simulators, reusable code components,
version control and regression tests, etc.,
• Science innovation - developments in plant physiological and soil chemical/physical
concepts and their translation to robust model algorithms,
• Application innovation - Developments in model interfaces, visualization, scaling up
techniques (summary models, cross-scale model linkages) and model delivery systems
(such as web-based applications).
Evolving focus for model application
We have seen the forces stimulating model development and application as evolving from an
initial focus on investigation of crop physiological function through a period where decision
support for farm-level decision making dominated to the present day when models are
extensively used in policy development and program implementation concerning land use
planning, environmental protection, climate change impacts and adaptation assessment and
greenhouse gas mitigation. Over the last 20 years the use of simulation models in agronomic
and farming systems research has gone from being a ‘fringe’ activity to a tool accepted by the
mainstream agronomic research community in diverse applications.
At the same time as there has been an acceptance of cropping systems modelling by the
broader agronomic research community, we argue that the rate of innovation in scope, design
and predictive capacity has gradually declined. We present and discuss the evidence for this
contention and explore likely sources for future innovation in cropping systems models
including novel delivery systems (e.g., via the internet), linking models at different scales to
deal with farm business and environmental issues, the ability to simulate livestock in cropping
systems, and the derivation of summary models to aid agricultural extension and education.
References
Bouman, B.A.M., et al., 2001. ORYZA2000: Modelling lowland rice. IRRI and WU, 235 pp.
Jones, J.W., et al., 2003. European Journal of Agronomy 18: 235-265.
Jones, C.A. & J.R. Kiniry, 1986. CERES-Maize: A simulation model of maize growth and
development. Texas A&M University Press, College Station, TX, USA.
McCown, R.L., et al., 1996. Agricultural Systems 50: 255-271.
Penning de Vries, F.W.T., et al., 1989. Simulation of ecophysiological processes of growth in
several annual crops. Pudoc, Wageningen, The Netherlands.
Stöckle, C.O., et al., 2003. European Journal of Agronomy 18: 289-307.
Van Diepen, C.A., et al., 1989. Soil Use and Management 5: 16-24.
193
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
APES, an agricultural production and externalities simulator evaluated for
two main crops in Midi-Pyrenees
M. Adam, H. Belhouchette, E. Casellas, O. Therond, J. Wery
Plant Production Systems Group, Wageningen UR, Haarweg 333, P.O. Box 430,
6700 AK Wageningen, The Netherlands
Contact: myriam.adam@wur.nl
Introduction
An agricultural production and externalities simulator (APES) has been developed within the
SEAMLESS project (Van Ittersum et al., 2008), as part of a modelling chain enabling ex-ante
impact assessment of agricultural and environmental policies and technological innovations.
The reason to include such model was to predict the impact of different land-bound activities
on regional agriculture from a bottom-up perspective. A first step towards this objective is to
evaluate the performance of APES at field level for the main crops of the investigated regions.
We conducted this analysis for Midi-Pyrenees to subsequently include this model for the
assessment of the impact of the Nitrate Directive on the agricultural systems of South of
France (Belhouchette et al., 2009). This study uses the APES Modelling Solution (Casellas et
al., 2009) developed to simulate arable crop activity with a focus on two contrasting crops.
Methods
APES is a modular system of biophysical components that simulates biophysical impacts of
crop management, soil and climate on cropping systems. APES integrates various existing
modelling approaches representing the soil-water budget, soil-plant nitrogen budget, crop
phenology, crop canopy and root growth, biomass production and partitioning, crop yield,
residue production and decomposition, soil erosion. Data were collected in INRA-Toulouse
(station Auzeville) from 1996–2002, for two major crops in different fields and management
practices in Midi-Pyrenees (Table 1).
The key phenological stages have been estimated using data for emergence, flowering and
harvesting time. The performance at field level of the model with default parameter values
Table 1. Main soil properties and management practices of the experimental data sets.
Plot name
Main soil properties
Main management practices
and year
Maize
T3-1996
T4-1996
X3-1997
X4-1997
Y3-1997
Y4-1996
Z1-1999
Z6-1998
Durum wheat
T1-2000
T4-1998
X4-1996
Y5-1999
Y7-1997
Soil type
Organic matter
content (%)
Bulk density
(t m–3)
Sowing
date (DOY)
Irrigation
(mm)
Clay loam
Clay loam
Clay loam
Clay loam
Loam
Clay loam
Clay loam
Clay loam
0.97
1.00
0.83
0.80
0.71
0.69
0.59
0.79
1.32
1.30
1.34
1.29
1.37
1.34
1.30
1.29
102
102
102
102
101
102
99
125
186
144
158
134
125
176
311
286
183
183
171
171
171
204
201
206
Clay loam
Clay loam
Clay loam
Loam
Clay loam
0.70
1.00
0.82
0.63
0.66
1.33
1.30
1.29
1.36
1.33
312
316
309
336
341
0
0
0
0
43
106
185
138
77
169
194
Fertilization
(kg ha–1)
Session B3: Cropping systems modelling
were evaluated comparing observed data with simulated data (grain yield, nitrogen uptake and
leached) using statistical criteria such as mean, standard deviation (SDV), the root mean
square error (RMSE) and the Pearson correlation coefficient (r).
Results and discussion
Figure 1 summarizes the main simulation outputs compared to the experimental data. Overall,
we have a slight underestimation in yield (–10% for maize to –5% for wheat) and a significant
overestimation of nitrogen uptake (+30% for maize and +20% for wheat) which might explain
the low level of nitrate leaching simulated. The overestimation of nitrogen uptake is likely to
be due to a too high nitrogen concentration in plant tissue. Another potential cause of the low
nitrogen leached can be a too low water drainage.
Maize yield (t.ha-1)
Durum wheat yield (t.ha-1)
13
8
7
11
6
predicted
predicted
1:1
12
10
9
1:1
5
4
3
8
8
9
10
11
12
2
13
2
3
4
observed
Maize N uptake (kg.ha-1)
7
8
300
1:1
350
250
predicted
predi cted
6
Durum wheat N uptake (kg.ha-1)
400
300
250
1:1
200
150
200
200
5
observed
250
300
350
400
100
100
observed
150
200
250
300
observed
Figure 1. Observed and simulated grain yield and crop nitrogen uptake for maize and winter
durum wheat in Auzeville, Midi Pyrenees (France) from 1996–2002.
However, as the use of the model targets the assessment of the impact of management
practices induced by the Nitrate Directive (low N fertilization), the most important criteria for
the accuracy of the model lies on its ability to capture differences in management practices.
Model performance is evaluated through the relationship between observed data and the
simulated outputs for different years and field managements. The correlation coefficient (r)
was 0.62 and 0.66 for yield and 0.84 and 0.32 for nitrogen uptake for maize and durum wheat,
respectively. This evaluation is a first step to demonstrate the good behaviour of the model in
various conditions for two contrasting crops. It shows the need for improvement in some parts
of the model related to plant nitrogen dynamics. Further testing on other crops will be carried
out according to Casellas et al. (2009). After proper calibration at field scale, APES
performance at regional level will be evaluated following the methodology of Wallach et al.
(2009). Finally, the performances of APES in a modelling chain will be tested to assess the
impact of crop management on environmental externalities in several European regions.
References
Belhouchette, H., et al., 2009. AgSAP Conference 2009, these proceedings.
Casellas, E., et al., 2009. AgSAP Conference 2009, these proceedings.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
Wallach, D., et al., 2009. AgSAP Conference 2009, these proceedings.
195
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Performances of two crop models in various conditions:
The importance of underlying assumptions
H. Belhouchette, M. Adam, E. Casellas, F. Celette, M. Corbeels, J. Wery
UMR-SYTEM, 2, Place Pierre Viala,
34060 Montpellier Cedex 1, France
Contact: Hatem.Belhouchette@supagro.inra.fr
Introduction
APES, a modular model was developed in the SEAMLESS project (Van Ittersum et al., 2008)
to assess, in different EU regions, the impact of farm management on crop production and
environmental externalities. Such purpose assumes the capability of APES to simulate
cropping systems in a wide range of climate, soil and crop management. The main topic of
this study is to evaluate APES, under different stresses and management practices
(temperature, water and nitrogen), to simulate crop production and nitrogen and water
dynamics. For this evaluation, APES was compared to the CropSyst (Stöckle et al., 2003)
model (widely evaluated and used under different biophysical conditions and management
practices). An explicit description of the limitations and specificities of each modelling
approach is related to the behaviour of each model in different situations.
Methods
Both models were calibrated independently, using the same experimental data sets collected
for rainfed durum wheat in the Midi-Pyrenees region (Mahmood, 2008; Adam et al., 2009).
Soil is mainly clay-loamy and weather data were collected on site, including rainfall,
temperature and radiation. The performance of each model has been tested, under different
scenarios, to define how they reacted to different gradients of temperature (scenario 1: Taverage
to Taverage + 4°C + no water stress+ no N stresses), irrigation regime (scenario 2: Taverage + no
N stress + irrigation doses from 0 to 700 mm) and nitrogen fertilization (scenario 3: Taverage +
no water stress + N fertilization from 0 to 200 kg N ha–1). The behavior of the two models
under different scenarios was defined analysing model outputs such as leaf area index, above
ground biomass and grain yield, water uptake and nitrogen uptake and leaching enabled.
Results and discussion
Temperature impact on leaf area development and biomass accumulation
The dynamics of the total leaf area development and biomass accumulation generated by the
two models are different (data not shown). When the average temperature increases, the LAI
decreases in CropSyst, while in APES it remains constant (data not shown). The approaches
used in each model explain these differences. LAI development in CropSyst at potential
production is directly dependent on biomass production, following a logistic curve with a
LAImax. In APES, LAI development is simulated following two phases, the first one being
the juvenile phase with an exponential growth, followed by a linear growth directly dependent
on biomass production. Another important (indirect) effect occurs through simulated biomass
production: in CropSyst there is a temperature effect on the radiation use efficiency during
early growth, while in APES the RUE is considered constant through the crop cycle. This
means that in CropSyst simulated biomass at the beginning of the cycle decreases with
increased average temperature, while in APES, it remains constant. Concerning leaf
senescence, the decrease of the LAI in APES is the same as in CropSyst, even though
different approaches are used. In APES, senescence responds to shading, temperature and
ageing (leaf duration), while in CropSyst only the ageing factor is directly considered by
simulating the leaf area duration.
196
Session B3: Cropping systems modelling
Biomass (t/ha)
Biomass (t/ha)
Water-limited conditions
16
16
Figure 1 illustrates the total biomass
8
8
trend when irrigation increases
gradually from 0 to 700 mm.
0
0
0.0
0.4
0.8
CropSyst is more sensitive than APES
0
200
400
600
waterstress
(0-1)
to the low amounts of water, but more
irrigation (mm)
APES
CropSyst
CropSyst
APES
biomass is simulated with the larger
irrigation scenario. In APES, water
1. Simulated total biomass for
Figure 2. Total biomass simulated for
stress (actual transpiration /potential Figure
different amounts of irrigation
different amounts of irrigation v.s. water
transpiration) is only effective above a
t
given threshold level. On the other hand, in CropSyst, water stress affects linearly the biomass
accumulation from emergence to flowering. After the growing period, the harvest index is
adjusted to account for sensitivity to water stress during flowering and/or grain filling. The
‘threshold value’ in APES causes a lower water stress sensitivity (also linked to the function
continuity) than CropSyst for equivalent levels of Actual/Potential transpiration (Figure 2).
Nitrogen-limited conditions biomass and nitrogen balance
APES simulates more biomass than CropSyst when N fertilization increases gradually from 0
to 150 kg ha–1 (Table 1). In both models, the crop experiences N stress when its N concentration drops below a critical value for unrestricted growth (NNI approach, Lemaire, 1989).
The two models simulate N transformation in a similar way using first-order kinetics. Both
models include routines for the simulation of soil temperature, and its effect on N
transformations. The main difference between the two models is that in CropSyst the
microbial community is considered as not limiting to the nitrogen transformation process,
which is driven only by water and temperature. While, APES, based on the formalism of
G’DAY model represents the role of soil micro-organisms in a mechanistic way through the
mineralization-immobilization turnover processes during organic matter decomposition.
N
(kg ha–1)
0
50
100
150
Biomass
N leaching
Total mineralization
N uptake
(t ha–1)
(kg ha–1)
(kg ha–1)
(kg ha–1)
CropSyst APES CropSyst APES CropSyst APES CropSyst APES
6.5
7.2
1.4
7.2
27.0
31.7
80.0
157.0
9.9
13.3
1.0
13.3
26.7
22.8
130.0
194.0
12.7
14.9
0.9
14.8
26.4
24.2
180.0
213.0
14.7
14.9
0.9
14.8
26.3
22.2
230.0
215.0
Table 1. Biomass
and nitrogen
balance: percentage
difference between
CropSyst and APES.
Conclusions
From our model comparison it appears that even if models use different approaches to
simulate growth, water and N dynamics, the final results in term of crop production and externalities are very similar. However, under specific conditions of water, N or/and temperature
stress, some modules are more suitable to be used than others. The following main
conclusions can be drawn. CropSyst seems more sensitive to heat stress than APES,
especially to predict phenology and biomass accumulation. For water stress and N stress, the
results seem more ambiguous due to the complexity of the approaches used that relate both to
crop and soil processes.
References
Adam, M., et al., 2009. AgSAP Conference 2009, these proceedings.
Lemaire, G. (Ed.), 1997. Diagnosis of the nitrogen status in crops. Berlin, Springer, 239 pp.
Mahmood, F., 2008. Mémoire de Master, SupAgro-Montpellier, 39 pp.
Stöckle, C.O., et al., 2003. European J. of Agronomy 18(3-4): 289-307.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
197
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A methodology for the evaluation and improvement of a generic biophysical
Soil-Plant-Atmosphere crop model based on ‘Mini-Application’
E. Casellas, J.E. Bergez, H. Belhouchette, O. Therond, M. Adam, A. Metay, J. Wery
INRA, UMR SYSTEM 2, Place Viala, 34060 Montpellier Cedex 1, France
Contact: casellas@supagro.inra.fr
Introduction
The modelling platform for integrated assessment of scenarios, SEAMLESS-IF
(http://www.seamless-ip.org/) has been developed for impact assessment of agricultural
systems from field to global scales. The field function is modelled with APES
(http://www.apesimulator.org/), a dynamic soil-plant-atmosphere model (SPA) simulating the
behaviour of a large range of crops (arable crops, vineyards, agro-forestry, grassland…).
APES has been developed as a shared effort among several research groups and provides an
extensible and modular modelling framework. The development of the different components
was then performed independently, i.e. component by component. Biophysical “composite”
models usually show a high level of interactions between the different composing submodels. Those may be either direct and/or indirect. Stand-alone model components present
only part of the reality and different sources of errors can be compensated by a proper
parameter calibration for a given range of conditions. As each component is independent with
its own underlying assumptions (explicit and/or non-explicit), some incompatibilities may
appear. The overall consistency and robustness of the resulting model has therefore to be
tested. Furthermore, some investigation and/or adaptation may then be required on some
components and their combination to simulate a given type of crop. A specific methodology
was developed to perform this iterative evaluation-improvement procedure allowing linking
of components in a ‘Modelling Solution’ of APES (APES-MS, i.e. a given list of components,
the links among them and the model options selected for each component).
‘Classical’ model evaluation is usually only based on a quantitative comparison with
observed data, in a more or less broad range of conditions and crop management, and usually
for a few simulated outputs of the model. We defined a more extensive and broad-range
evaluation procedure based not only on this type of quantitative evaluation but also on
conceptual and qualitative evaluation for the whole range of conditions for which the model
will be used and with the key state and flow variables. The objective of this paper is to present
this procedure, based on relevant Mini-Applications (MA), and its main outcomes.
Methods
In this approach, we assume that all
individual components have already
been calibrated and evaluated as
independent stand-alone components.
We then focus on the composite
model as a whole and on the
interactions among the different
components. Our evaluation-improvement procedure is based on three
steps: (i) a conceptual evaluation that
aims to evaluate the relevancy and
consistency of the concepts of the
different components put together in Figure 1. Schema of methodology for conceptual,
an APES instance; (ii) a qualitative qualitative and quantitative evaluations steps for an
evaluation of APES-MS based on SPA modular model.
198
Session B3: Cropping systems modelling
trends of key variables in comparison with expert knowledge; (iii) a quantitative evaluation
using specific MA test cases for which we have observed data and/or expert knowledge.
These three steps allow identification of the main problems to solve and the type of
improvement (model calibration, change of model structure, Figure 1). For the qualitative and
quantitative evaluations, MA are first described as a crop/soil/climate/agro-management
combination representative of the future use of the model and for which we have sufficient
expert knowledge to describe a priori the evolution of the major state and flow variables. The
model is then challenged for some specific processes by varying the input conditions and
investigating how it reacts compared to expected behaviour in such conditions as defined by
the conceptual evaluation. To properly run a MA, agronomists, modellers and computer
scientists are required.
Each of these evaluations is managed using a 3-step process. In a first step, PlantAtmosphere functions are analysed without the influence of the soil resources. This
corresponds to theoretical ‘potential’ plant growth (no effect of water or nitrogen on plant
production). It allows the evaluation of temperature and light effects decoupled from soil
effects. In a second step, the effect of water constraints both on soil water dynamics and
interactions with plants are evaluated. We first make sure that fully-irrigated plants react in
the same way as in potential production. We can then use simulations with observed
irrigations, which allow investigating soil water / plant interactions (e.g., water extraction by
plants or plant water stress impact). The third step is simulating water- and nitrogen-limited
conditions, enabling the investigation of soil nitrogen dynamics and interactions with plant
and soil water. Again we make sure that with full irrigation and full fertilization we can reach
‘potential’ plant growth. We finally use observed irrigation and fertilization to investigate soil
nitrogen dynamics and its interactions with soil water and plants (e.g., plant nitrogen
extraction or nitrate leaching). For nitrogen issues, some bare soil simulations with controlled
temperature and soil water content have also been used to investigate internal soil nitrogen
dynamics without any plant and soil water interactions.
Results and discussion
The three types of evaluations using the three different steps have proved to be necessary to
ensure the proper development of the APES-MS. This procedure was used during dedicated
workshops, each one combining an MA and an APES-MS, and covering the set of crop types
supposed to be simulated by APES in the SEAMLESS project. We only give below some
examples of the outputs of these workshops and how it helped to improve the model.
During the conceptual evaluation of the plant and soil water interactions, we noticed that
there was a process for plant water demand reduction that was handled both in plant and water
extraction components, leading to an overestimation of water stress and the impossibility to
reach potential production.
An example of qualitative evaluation can be illustrated by the expected sigmoid plant
response to water effect of progressive irrigation doses (this water being the only one coming
in the system as we removed rain to the weather file).
An example of quantitative evaluation can be taken from the MA held in the Midi-Pyrenees
with durum wheat that has been compared to various observed field data (Adam et al., 2009)
resulting in a new parameter calibration.
MA described in detail can be run again with improved APES-MS, this is part of the
quality of model construction. This methodology is still to be investigated (for dispersed work
of several team members in particular). But it was still very efficient in improving APES-MS
and created a collective dynamic between agronomists, modellers and informatics.
References
Adam M., et al., 2009. AgSAP Conference 2009, these proceedings.
199
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Using a cropping system model for large scale impact assessment in Europe
E. Casellas, O. Therond, H. Belhouchette, D. Wallach, R.J. Oomen, G. Russell,
F. Ewert, H. Hengsdijk, M.K. Van Ittersum, M. Adam, J. Wery
INRA, UMR-SYTEM, 2, Place Pierre Viala,
34060 Montpellier Cedex 1, France
Contact: casellas@supagro.inra.fr
Introduction
Agricultural policy of the European Commission aims at improving agricultural sustainability
at field, farm, regional and EU scales. Ex-ante assessment of the possible economic,
production and environmental consequences of policies may support decision making. The
SEAMLESS project (Van Ittersum et al., 2008) provides an integrated assessment and
modelling platform to support such policy assessments. This platform uses the cropping
system model APES (http://www.apesimulator.org/) to predict yields and externalities such as
nitrogen leaching of agricultural activities. Like other cropping system models APES operates
at the field level using daily time steps. To predict crop yields and externalities at the scale of
the European Union APES need to be used for a much larger geographic area. How to do so,
and how to evaluate the model at this scale, are the problems addressed in this paper.
Methods
The APES cropping system model includes components that take into account water and
nitrogen stress but not yield reductions due to diseases or weeds. Model inputs are daily
weather, soil characteristics, initial soil conditions and management practices. A large number
of annual crops (cotton, maize, oats, peas, potatoes, rape, rice, rye, sorghum, soya, spring
barley, spring soft wheat, sugar beet, sunflower, triticale, winter barley, winter soft wheat,
winter durum wheat) can be simulated in rotations.
The first problem in using a field level model for a large area is the spatial soil-climate
heterogeneity. Therefore, Europe is divided into zones with relatively homogeneous
properties from an agronomic perspective. In SEAMLESS, Europe is divided into
AgroEnvironmental Zones (AEnZ), which are combinations of 13 environmental zones
(primarily based on statistical analysis of climate and geomorphological variables), a soil
classification (based on 6 topsoil organic carbon classes) and NUTS2 administrative regions
(Baruth et al., 2006). A total of 195 AEnZ in 16 sample regions (from Andalucia-Spain to
Etela-Suomi-Finland and from Southern and Eastland-Ireland to Thessalia-Greece) have been
chosen to represent the biophysical conditions of Europe. For each AEnZ, historical weather
(from the MARS database) and soil characteristics (from the European soil database) are
available.
Classical model evaluations for APES have been carried out at the field level (Adam et al.,
2009; Casellas et al., 2009), but for large areas it is important to add an additional evaluation
allowing to compare model results with corresponding observed values in each AEnZ, for a
range of regions throughout Europe. Therefore, current management practices for the major
crops in each AEnZ need to be specified. Such data do not exist at a pan-European level; the
Farm Accountancy Data Network (FADN) contains data about input use but at the farm level
rather than the crop level, and does not specify timing of input use. We therefore developed a
new data base, taking advantage of the fact that SEAMLESS has a large number of partners
throughout Europe. These partners identified local experts who in turn specified the major
regional crop rotations and associated yield levels. In addition, the experts estimated average
(over fields and years) management activities for these crops, i.e. sowing and harvesting date,
nitrogen input, total amount of irrigation water and number of irrigations. Then, based on that
information, experts and agronomists developed generic decision rules specifying detailed
200
Session B3: Cropping systems modelling
management practices for each crop (Oomen et al., 2009). These management rules are
important not just for quality evaluation of model outputs, but also for simulation of current
practices.
The APES model has simulated 21 crops in each AEnZ of the 16 sample regions, for a
period of 25 years. The first five simulation years are used for initialization of the soil
conditions. Average date of physiological maturity (assumed to be the same as harvest date in
expert data) and average yields are calculated using results from the remaining 20 years.
Three criteria are used to assess the agreement between model and observed data. First, we
calculate the average number of degree days up to the ‘observed’ (i.e. expert value) harvest
date for each AEnZ. This value can be compared to the model parameter value that represents
the required number of degree days to maturity (determined by modellers according literature
information and field level calibrations). The ratio of the two is the first error factor, kpheno.
Secondly, simulated yields in each AEnZ are compared to the yields as specified by local
experts. The ratio is the second error factor, kyield1. Finally, one can first calibrate the
phenology parameters of the model (by multiplying all these parameters – that represent
degree days to different phenological stages – by kpheno) and then calculate a new yield error
factor kyield2. The difference between kyield1 and kyield2 indicates to what extent the error in
yield can be reduced by correcting for the errors in phenology. Remaining errors can still refer
to many issues, i.e. misspecification and parameterization of the model, flaws in soil, climate
and agro-management data, and to the fact that pests and diseases are not modelled in APES.
Runs of APES and calculation of the error factors are currently underway and will be
presented at the Conference.
Results and discussion
The feasibility of using a cropping system model at the scale of Europe is closely related to
the availability of input data and evaluation data. In general, weather and soil data are
available in existing European data bases resulting in the AEnZs as described. The model can
be run for given management to predict yields and environmental effects at European scale.
For model evaluation, we have created a new data base that contains current crop
management and evaluation data for AEnZ throughout Europe, based on expert opinions.
The assessment criteria as discussed here serve two functions. Firstly, they indicate the
degree of agreement between model and observed data. It will be of particular interest to
analyse systematic patterns in the error factors, e.g. across regions. Secondly, the error factors
can be used as correction factor to modify model outputs. The effectiveness of such a factor in
improving prediction of externalities and estimation of alternative agro-management options
still needs to be assessed.
References
Adam, M., et al., 2009. AgSAP Conference 2009, these proceedings.
Baruth, B., et al., 2006. PD4.3.2. A spatial framework for environmental data, SEAMLESS
integrated project, EU 6th Framework Program, contract no. 010036-2, www.SEAMLESSIP.org. 35 pp.
Casellas, E., et al., 2009. AgSAP Conference 2009, these proceedings.
Oomen, R.J., et al., 2009. AgSAP Conference 2009, these proceedings.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
201
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Spatializing a bio-decisional crop-model to assess the sensitivity of a region
to modifications in agricultural practices
L. Clavel, M.-H. Charron, D. Baudet, J. Soudais, D. Leenhardt
INRA, UMR1248 - AGIR (Agrosystèmes et développement territorial),
B.P. 52627, 31326 Castanet Tolosan Cedex, France
Contact: Lucie.clavel@toulouse.inra.fr
Introduction
Recent societal expectations, e.g., food security and environmental resource quality and
availability, need to consider agriculture not only at its traditional spatial scale, the field plot,
but at higher spatial scales and complexity levels. They also suggest considering new ways of
producing and managing crops. Crop models that simulate the crop growth and yield for a
given soil and climate combination can be used to measure the impact of new management
practices at the plot scale. To account for more complexity, it is useful to use cropping
systems models that integrate constraints and behaviours of farmers. To account for larger
spatial extents that include multiple fields and a great diversity of cropping systems, a solution
is to distribute or spatialize these models over the considered area (Faivre et al., 2004).
We propose here a way to distribute a bio-decisional crop model over a large agricultural
area in order to assess its sensitivity to modifications of crop management practices.
Methods
The study area is the Neste System, a catchment in south-western France. This 800,000 ha
area gathers the catchment areas of 18 rivers artificially supplied with water by a single canal
(the Neste canal). The land is mainly dedicated to agriculture: 500,000 ha are cultivated from
which 50,000 ha are irrigated (Hurand, 2000). Thanks to the canal and European common
agricultural policy, the irrigated area has been multiplied by around three since the 1970s. As
a consequence, eight years out of ten, the system faces a lack of water that threatens the
environmental equilibrium of the rivers and the satisfaction of agricultural water needs.
Since irrigation is a key issue of this area, we decided to select a model that could represent
modifications of irrigation practices. The MOuSTICS model is a bio-decisional crop model. It
integrates (i) a crop model (STICS, Brisson et al., 1998) that simulates the growth of different
crops in the Neste system and (ii) a decisional model that, instead of simulating the water
needs of the considered crop, explicitly simulates irrigation applications through farmers’
decision rules that depend on crop, soil, climate and water resource constraints (Bergez et al.,
2001).
Cropping system, soil and weather information is required as input data for running
MOuSTICS. To account for the spatial variability of these input data, we divided the Neste
system into 67 support units (SUs) resulting from the intersection between water management
units (sections of the 18 river catchments) and Small Agricultural Regions (delineated
according to agriculture-related criteria, mainly soil and climate conditions). We associated
with each SU one daily weather series interpolated by Météo-France (Prats & Pérarnaud,
2001). We classified the soils of the study area into 13 main classes. For each SU, we
determined the proportion of each soil class by GIS. Experts from local extension services
helped us to identify ten farm types for the whole area according to economic, technical
orientation and irrigation criteria. We used agricultural census data (Agreste, 2000) to
calculate for each SU, the average Usable Agricultural Area and crop distribution for each
farm type (distinguishing between irrigated and non-irrigated). Finally, we redistributed the
farm types over the SUs by using expert information describing the relationship between farm
types and soil types. This process allowed us to identify, for each SU, the area of each
crop×soil combination. We stored this information into matrices (one per SU) that made
202
Session B3: Cropping systems modelling
explicit the farm type, the soil and the crop (Figure 1)
(Clavel & Leenhardt, 2008). Management practices
associated with each combination were determined
by expertise.
MOuSTICS can be run for each crop×soil×farm
type combination of the matrix with different crop
management practices scenarios (Table 1). In order to
assess the irrigation demand and the crop production
at SU level, we can multiply the MOuSTICS outputs
with the areas stored in the matrix corresponding to
the SU. As the matrices present the same
organization for each SU, outputs can be aggregated
at levels greater than the SU: Small Agricultural
regions, water management units, and the whole
system.
Figure 1. Each cell of the matrix
contains the area A of a given
crop×soil×farm type combination.
Such matrix is built for all SUs.
Table 1. Simple scenarios can be tested in a first step.
Scenario
Purpose
Alternatives
Scenario 1
Reducing number of water At sowing; at last application; increase of
applications
the period between 2 application
Scenario 2
Reducing water volume applied 5 mm; 10 mm; 15 mm etc.
Scenario 3
Change maize earliness
Early; 1/2 early; 1/2 late; late
Results and discussion
The work is in progress, we will quantify the effect of given practices on the whole area. This
effect will be assessed by two main variables: irrigation water demand and crop production.
By comparing various scenarios we intend to estimate to which practices the system is most
sensitive to, and this at the different scales considered (farm, Small Agricultural Region,
water management unit, whole system).
Spatializing a bio-decisional crop model can show the effective impact of cropping
systems innovations at broader scales than the plot. This type of tool can be used in
prospective research for assessing which cropping systems should be modified over a region
for achieving dual aims in term of environment and production. In the frame of the present
study, it is planned to use the tool to evaluate more complex cropping systems scenarios built
with stakeholders.
References
Agreste, 2000. La Statistique Agricole, Recencement Agricole 2000. Le recensement agricole
en bref. http://www.agreste.agriculture.gouv.fr/
Bergez, J.E., et al., 2001. Ecological Modelling 137: 43-60.
Brisson, N., et al., 1998. Agronomie 18: 311-346.
Clavel, L. & D. Leenhardt, 2008. International Environmental Modelling and Software
Society. International Congress on Environmental Modelling and Software.
Faivre, R., et al., 2004. Agronomie 24: 205-217.
Hurand, P., 2000. La gestion opérationnelle d’un système complèxe: ‘Le système Neste’.
CACG, 23 pp.
Prats, S. & V. Pérarnaud, 2001. Projet ADEAUMIS. Contribution de Météo-France, 19 pp.
203
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The SoilC&N module of APES: Simulating the immobilizationmineralization turnover for predicting short- and long-term C and N
dynamics in the soil
M. Corbeels, E. Casellas, F. Celette, A. Mérot
UMR-SYTEM, Centre de coopération International en Recherche Agronomique pour le
Développement, 2, Place Pierre Viala, 34060 Montpellier Cedex 1, France
Contact: marc.corbeels@cirad.fr
Introduction
Carbon (C) and nitrogen (N) dynamics in the soil can be described by a number of
approaches, ranging from simple empirical regression equations to complex and detailed
process-based models. The disadvantage of empirical models is that they are site and time
specific and they need to be parameterized for each new situation. Simple two-compartment
models comprising a labile and stable organic matter pool can be analytically solved and
parameter estimation for a given situation is relatively simple (e.g., ICBM, Kätterer &
Andrén, 2001). However, these types of models do not incorporate important feedbacks of
soil C and N to changing environment. More comprehensive models, such as CENTURY
(Parton et al., 1987), have been developed for this purpose. Most of these models do not
consider explicitly microbial physiology as the driving factor of N immobilizationmineralization turnover, while this is fundamental for an adequate description of
decomposition (e.g., Van Veen et al., 1984).
Model description: novel features for simulating soil C and N dynamics
SoilC&N, a decomposition model for C and N cycling in the soil is briefly presented here
(Figure 1). More details of the model are given in Corbeels et al. (2005). It was adapted for
use as a module in the APES crop growth simulator (www.apesimulator.org). The distinctive
features of this model are: (1) growth of microbial biomass is the process that drives N
mineralization-immobilization, and microbial succession is simulated; (2) decomposition of
plant residues may be N-limited, depending on soil inorganic N availability relative to N
requirements for microbial growth; (3) N:C ratio of microbial biomass active in decomposing
plant residues is a function of residue quality and N availability; (4) ‘quality’ of leaf and fine
Foliar and fine root litter
MP
CO2
LCP
L
HCP
CO2
Foliar and fine root litter
Woody litter
MP
WP
L
Woody litter
LCP
L
HCP
WP
L
CO2
CO2
Inorganic
N
Microbial
biomass
CO2
CO2
CO2
Microbial
biomass
Inorganic flows
CO2
Mineralisation
Sm
CO2
Sy
Old
SOM
Immobilisation
Young
SOM
Young
SOM
Organic flow
Stabilisation/
humification
(a)
Old
SOM
(b)
Figure 1. Pools and fluxes of (a) C and (b) N in the SoilC&N model. MP: metabolic pool;
HCP: holocellulosic pool; LCP: ligno-cellulosic pool; L: lignin; SOM: soil organic matter;
Sm: stabilization coefficient for microbial biomass; Sy: stabilization coefficient for Young
SOM (from Corbeels et al., 2005).
204
Session B3: Cropping systems modelling
root residues is expressed in terms of measurable biochemical fractions; and (5) N:C ratios of
soil organic matter (SOM) pools are not prescribed but are instead simulated output variables
determined by plant residue characteristics and soil inorganic N availability.
The model includes above- and below-ground plant residue pools and three SOM pools
(microbial biomass, Young and Old SOM) with different turnover times (Figure 1). Rates of
decomposition are modified by temperature, moisture, lignin content of the residues and N
availability. Stabilization of SOM is simulated by transferring fractions of decomposed
microbial biomass and Young SOM into more recalcitrant forms (respectively into Young and
Old SOM). Nitrogen is mineralized to, or immobilized from, the soil inorganic N pool to
maintain the N:C ratio of decomposing microbial biomass within a specified range. Balancing
potential microbial N demand against inorganic N availability determines whether the activity
of decomposers is limited by N. If so, then simulated microbial use efficiency and
decomposition fluxes are reduced. The maximum rate of microbial N uptake is proportional to
soil inorganic N content. Lignin transformation to Young SOM promotes additional N
immobilization into the Young SOM pool, which simulates the process of chemical N
immobilization.
Application examples
Short term dynamics of N supply
SoilC&N can be used as a stand alone model or coupled to a plant production model to
simulate within-season soil N mineralization dynamics from SOM and added organic sources.
The model responds to quality of added organic matter and predicts N immobilization or
mineralization rates in time. The N immobilization peak depends on the biochemical quality
of the plant residues and the available inorganic N. When soil inorganic N becomes severely
limiting, decomposition of residues is slowed down. With a proper parameterization of plant
residue ‘quality’, the model can acceptably predict N dynamics from crop residues ranging
from green leguminous leaves to woody residues.
Global change modelling
Coupled to a plant production model, SoilC&N is particularly suited to simulating the impacts
of land-use change and consequences of climate change on soil carbon storage and N
availability for plants. For example, the model is able to predict long-term storage of soil C
following a change in land-use from forest to cropland, as a result of simulated changes in
microbial activity, soil N availability and SOM C:N ratios to changes in plant residue quantity
and quality. The incorporation of the feedbacks in the model between plant residue quality, N
availability and microbial activity increases the mechanistic integrity of the model, compared
to other models such as CENTURY or RothC (Coleman et al., 1997). Climate change can
strongly affect simulated microbial activity via the water and temperature factors that modify
the rates of decomposition.
In conclusion, the ability of SoilC&N to adequately describe both short-term events such as
soil N supply during one growing season, and long-term dynamics, e.g., soil C storage over
several decades, is an important asset when coupling to a plant production model.
References
Coleman, K., et al., 1997. Geoderma 81: 29-44.
Corbeels, M., et al., 2005. Ecol. Model. 187: 426-448.
Kätterer, T. & O. Andrén, 2001. Ecol. Model. 136: 191-207.
Parton, W.J., et al., 1987. Soil. Sci. Soc. Am. J. 51: 1173-1179.
Van Veen, J.A., et al., 1984. Plant and Soil 76: 257-274.
205
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
APES: The Agricultural Production and Externalities Simulator
M. Donatelli1, G. Russell, A.E. Rizzoli
Agriculture Research Council,
Bologna, Italy
Contact: marcello.donatelli@jrc.it
Introduction
Many simulation tools allow the impact of agricultural management on production activities
in specific environments to be studied (e.g., Brisson et al., 2003; Keating et al., 2003; Jones et
al., 2003; Stöckle et al., 2003; Van Ittersum et al., 2003). Such tools are usually targeted at
one or more specific production activities: arable crops or cropping systems, grassland,
orchards, agro-forestry, livestock etc. Some include an estimate of system externalities which
may have a negative environmental impact, for example, erosion, nitrogen leaching or the fate
of pesticides. Very often, the structure of such systems neither allows an easy addition or
modification of models for new agricultural production activities, nor the use of different
approaches for the simulation of processes via alternative formulations. Furthermore,
documentation of such tools is often not up-to-date, and may not follow a single standard,
which makes it difficult to access information. Finally, when such systems are proprietary
systems of either research groups or projects, it may not be possible for third parties to
effectively develop the system further. There was, thus, a need within project SEAMLESS
(Van Ittersum et al., 2008) for a flexible and open modelling platform that could be extended
and adapted to achieve the SEAMLESS modelling goals. APES, the Agricultural Production
and Externalities Simulator, was thus developed as a simulation model system for estimating
the biophysical behaviour of agricultural production systems at the field scale in response to
the interaction of weather, soil and agro-technical management options.
Methods
APES is based on a set of models (components in software terms) that are used to simulate
the behaviour of different domains within the weather-soil-plant system (Figure 1). It was
designed to allow the subsequent incorporation of additional models to simulate additional
processes, such as plant pests, and the replacement of models with alternative versions.
Biophysical processes are simulated using deterministic approaches based on mechanistic
representations. APES includes utilities for managing appropriate weather and soil data
including cases where there are incomplete data. The AgroManagement component was
designed to represent farmers’ decisions realistically, taking into account the state of the crop
and the environment. The system permits the analysis of multiple years to take account of
rotations and varying weather patterns. Associated with APES are procedures for modelling
new crops by adjusting parameters, for adjusting parameters to take account of differences
between regions and to calibrate the models. It exists as both a stand-alone version and an
integrated version within SEAMLESS-IF. Finally, there are internal checks on the validity of
combining particular sets of inputs and outputs.
Results and discussion
APES has now been run to provide inputs for farm system models both for regions where
there is a complete set of input data and for those where the available information is
incomplete, which is often the case when predictions have to be made at regional level.
Parameter sets for 20 crops are already included together with several types of grassland and
1
Currently seconded to the Joint Research Centre, Institute for the Protection and Security of the Citizen, MARS
Agri4cast, Ispra, Italy.
206
Session B3: Cropping systems modelling
vineyards. APES has already been used successfully to model vineyards intercropped with
grass and for a series of mini-applications involving contrasting crop-environment
combinations for locations where there are good test data.
Users of APES can choose a modelling solution to meet their particular requirements, where a
modelling solution is a combination of components and options within components. The
criteria for selecting the currently available modelling solutions were based on the need to: (1)
account for specific processes to simulate soil-land use interactions, (2) obtain input data to
run simulations, (3) simulate agricultural production activities of interest (e.g., crops, grasses,
vineyards, agro-forestry), and (4) simulate agro-management decisions and their impact on
the system. The Graphical User Interface (GUI) allows users to run simulations and explore
the outputs of APES in response to changing inputs. The GUI is also made with components
and tools which can be re-used in different systems.
Part of the rationale of APES was to ensure wide availability of relevant information.
Descriptions of APES, including help files and video tutorials, are thus available on a website
(http://www.apesimulator.org or through http://www.seamless-ip.org). There is an APES
newsletter (news-subscribe@apesimulator.it). Subscribers can receive news of updates.
Figure 1. The main typologies of APES models and outputs.
References
Brisson, N.C., et al., 2003. Eur. J. Agron. 18: 309-332.
Jones, J.W., et al., 2003. Eur. J. Agron. 18: 235-265.
Keating, B.A., et al., 2003. Eur. J. Agron. 18: 267-288.
Stöckle, C.O., et al., 2003. Eur. J. Agron. 18: 289-307.
Van Ittersum, M.K., et al., 2003. Eur. J. Agron. 18: 87-393.
Van Ittersum, M.K., et al., 2008. Eur. J. Agron. 96: 150-165.
207
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The effects of climate change on the phenology of cereal crops in Hungary
Éva Erdélyi
Corvinus University of Budapest, Department of Mathematics and Informatics,
1048 Budapest, Villányi út 29-43, Hungary
Contact: eva.erdelyi@uni-corvinus.hu
Introduction
Using geographical analogues Gaál & Horváth (2006) showed that the possible future climate
in Hungary would be similar to the present climate of South-Southeast Europe. Increased
mean annual temperatures in our region, if limited to two or three degrees, could generally be
expected to extend the growing season. In the case of crops where phenological phases
depend on accumulated heat units (Varga-Haszonits, 1987), phenophases could become
shorter. In this paper we investigated the effects of climate change on the growing periods of
maize and winter wheat, which are the two most widely cultivated plants in Hungary.
Materials and methods
Weather data. Climate scenarios can be defined as relevant and adequate pictures of how the
climate may look in the future. During our research, we applied the principles defined by
IPCC (1996) and used the most commonly accepted scenarios presented in international
reports, which give predictions for the middle of the century. We have used climate models of
the United Kingdom Meteorological Office and the Geophysical Fluid Dynamics Laboratory
(USA). In this work five climate scenarios from these General Circulation Models were
downscaled to Debrecen, an important centre of agricultural production in Hungary to provide
31 years of daily values for temperature, precipitation and solar radiation. These were
compared with baseline data representing the current climate.
Crop model. We used the 4M crop model (Fodor, 2002), which has been developed by the
Hungarian Agricultural Model Designer Group. It contains several sub-models to describe the
physiological interactions of soil-plant systems and offers the possibility of developing
different systems models for the specific purposes of the users. The CERES model was
chosen as the starting point and was adapted for Hungarian conditions. The simulations were
run using, as weather inputs, the data from the climate scenarios and the historical data of the
reference period.
Results and discussion
The phenological phases of maize shortened (Table 1) and began earlier as a result of
temperature increase in all scenarios. In case of winter wheat we compared only the historical
data and the UKTR scenario. Like maize, the starting dates of the phenological phases were
earlier, especially in the first growing period (Figure 1). Harvesting is predicted to occur on
average eight days earlier in the future.
Table 1. The length in days of the phenological phases of maize for six climate scenarios.
Phenophase
2
3
4
5
6
7
BASE
16
25
7
43
14
56
GFDL25
16
24
7
40
12
42
GFDL55
13
22
6
40
12
42
208
UKHI
7
16
6
37
11
32
UKLO
6
15
6
35
11
33
UKTR
14
24
6
40
13
52
Session B3: Cropping systems modelling
Phenological phases of winter wheat
8
phenologic al phase
7
6
5
4
3
2
1
0
0
30
60
90
120
150
180
210
240
270
days in the vegetation period
UKTR
1960-90
Figure 1. Comparison of the starting dates of the phenological phases of winter wheat for the
reference period 1960–90 with the UKTR climate scenario for Debrecen.
Climate change has already had significant impacts in Hungary on the environment, human
health and society. Considering the possible changes we have to answer many questions in
order to prepare for the future (Erdélyi, 2006). Living under changing climate conditions, one
of our most urgent tasks is to create well-designed descriptive-forecasting systems (Erdélyi et
al., 2006), as well as to define the optimal preparation, adaptation and response strategies to
the changing conditions. Modelling is a good tool for investigating future circumstances
without the need for carrying out expensive and long experiments. It helps identify strategies
for preparing for the future and can take account of the many direct and indirect ways in
which climate change can affect agriculture.
References
Erdélyi, É., 2006. Climate change and temperature needs of winter wheat. Ecological
problems of our days – from global to local scale. Keszthely, pp. 274-279.
Erdélyi, É., et al., 2006. How climate change influences the field crop production I. and II.–
Use of Spatial Analogy and Yield variability of maize. 4th International Eco-conference –
Safe Food, Novi Sad, pp. 1-12.
Fodor, N., et al., 2002. European J. of Agron. 18: 389-393.
Gaál, M. & L. Horváth, 2006. Geographical analogies in climate change research. HAICTA,
Greece, pp. 840-846.
IPCC, Climate Change, 1996. In: Eds. J.T. Houghton, et al., The Science of Climate Change.
Cambridge University Press, Cambridge.
Varga-Haszonits, Z., 1987. Agrometeorológiai információk és hasznosításuk, Mezőgazdasági
Kiadó, Budapest.
209
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Estimating external costs from the application of plant protection products
in Europe
P. Fantke, A. Kuhn, S. Wagner, R. Friedrich
Institute of Energy Economics and the Rational Use of Energy (IER),
Hessbruehlstr. 49a, 70565 Stuttgart, Germany
Contact: Peter.Fantke@ier.uni-stuttgart.de
Introduction
Widespread public concern exists about the hazards posed by plant protection products (PPPs)
and other organic compounds that are released into the atmosphere, water and soil by
agricultural activities. Humans are exposed to these contaminants via multiple pathways, i.e.,
inhalation, dermal exposure and ingestion of different food items and drinking water.
However, for most currently used PPPs we do not yet fully understand their behaviour in and
between the atmosphere, water and soil and their effects on human health and the
environment at different concentrations. As an example, there exist numerous PPPs that are
subject to transformation processes leading to an altered mode of action in humans (Hodgson
& Rose, 2008). The present paper aims to give an insight into the challenges of conducting a
full chain assessment of PPPs for estimating external costs due to human health impacts at the
European scale as a basis for national and international policy making.
Methods
When assessing and valuing welfare losses
due to the application of PPPs, most
questions
are
interdisciplinary
and
complex. Linking concentrations of PPPs
in the food chain, which is the most
important exposure route for humans
(Juraske, 2007), to specific health effects
demonstrates the complexity of such
questions (Figure 1). To provide a comprehensive answer to the even more specific
question “How much and in which way do
changes in the application of PPPs affect
human health throughout Europe?”, an
integrated assessment modelling approach
is therefore required. A typical integrated
assessment model (IAM) is RAINS
(Amann, 2004), which addresses health and
ecosystem impacts of particulate pollution,
acidification, and eutrophication from air
pollution sources. Furthermore, questions
that are generally tackled by IAMs range
from sustainable development (MalkinaPykh, 2002) to policy assessment, e.g.,
impacts of agricultural policies at different
scales on the whole agricultural sector (Van
Ittersum et al., 2008).
For a full chain approach from the
application of PPPs to the estimation of
health effects and related external costs it is
Figure 1. Conceptual structure of a full chain
assessment of PPPs. Arrows denote a PPP’s
environmental pathway.
210
Session B3: Cropping systems modelling
essential not only to consider the atmosphere as the receiving compartment as only 30–50%
of the applied chemical is normally lost to air (Van den Berg et al., 1999). It is, thus,
necessary to consider all compartments, i.e., agricultural soils, surface water and groundwater,
and the surface of plants, where it is important to distinguish between the chemical fraction
that resides at the surface and the fraction that penetrates the plant tissues (Leistra, 2006). The
crop component of the IAM needs to be able to take account of differences between crops and
management operations as the impact of a PPP can be significantly increased by incorrect
application. In addition, all relevant emission routes have to be considered for the exposure
assessment, i.e., spray drift and runoff to surface water, leaching to groundwater, and
accumulation in topsoils and field crops (Peeters et al., 2008). The long-range transport
potential of a PPP must also be considered. Trade of different crops is included as an
extension of the natural fate. The most challenging part of the effect assessment is the
consideration of mixture toxicity as in reality chemical contaminants rarely, if ever, occur in
isolation, and interactions of components in a mixture can cause complex and substantial
changes in the apparent properties of its constituents, resulting in synergistic or antagonistic
effects (Knauert et al., 2008).
Results and discussion
An integrated assessment model for estimating human health impacts and damage from the
application of PPPs in Europe is being developed within two EU projects, EXIOPOL1 and
HEIMTSA2. The most limiting information is that related to epidemiologically-derived
effects, and spatially resolved emission/application data. Several emission data sets have been
reviewed for most European countries. These, however, mainly refer to consumption of PPPs,
grouped according to their intended purpose, e.g., insecticides, at national levels. Application
statistics on single crops are really needed at sub-national level for all EU countries, but are –
except for the UK – not available at present. The existing consumption and application data
will now be processed and combined with all relevant fate and exposure pathways in order to
derive human health impacts and damages. First results, i.e., external cost estimates due to
application of current PPPs for most European countries, are expected to be available in late
2009.
References
Amann, M., 2004. The RAINS model: Documentation of the model approach prepared for the
RAINS peer review 2004. IIASA, Laxenburg, Austria.
Hodgson, E. & R.L. Rose, 2008. Pest Management Science 64: 617-621.
Juraske, R., 2007. Advances in Life Cycle Impact Assessment of Pesticides: Methodological
Improvements and Experimental Studies. Universitat Rovira i Virgili, Tarragona.
Knauert, S., et al., 2008. In: Proceedings of SETAC Europe 18th Annual Meeting, Warsaw,
pp. 247-248.
Leistra, M., 2006. Estimating input data for computations on the volatilisation of pesticides
from plant canopies and competing processes. Alterra, Wageningen, 80 pp.
Malkina-Pykh, I.G., 2002. Ecological Indicators 2: 93-108.
Peeters, F.M., et al., 2008. PRIMET version 2.0. A Decision Support System for assessing
Pesticide RIsks in the tropics to Man, Environment and Trade. Alterra, Wageningen.
Van den Berg, F., et al., 1999. Water, Air & Soil Pollution 115: 195-218.
Van Ittersum, M.K., et al., 2008. Agric. Syst. 96: 150-165.
1
2
http://www.feem-project.net/exiopol/
http://www.heimtsa.eu
211
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Modelling water and nitrogen interactions in durum wheat:
A first step towards the conception of joint management strategies
for irrigation and nitrogen fertilization
S. Guillaume, J.E. Bergez, E. Justes
INRA, UMR 1248-AGIR, B.P. 52627, 31326 Castanet Tolosan Cedex, France
Contact: Solenne.Guillaume@toulouse.inra.fr
Introduction
In durum wheat production, grain protein concentration is of great importance because it
determines the quality of the grain and thus the selling price to the producer. Grain protein
concentration is strongly influenced by nitrogen nutrition status, which is a function of the
rates and dates of N-fertilizer application, but also by soil water status, which influences soil
nitrogen availability.
Due to climate change, farmers will have to deal with the reduction of water supply from
rainfall during wheat growth. Irrigation may be needed to ensure an adequate yield and
quality. This must be considered together with nitrogen fertilization to maximize N
availability for the crop and to minimize N losses. Moreover, the response may depend on the
cultivar of durum wheat that is grown.
Biophysical models may be useful tools to evaluate the impact of irrigation and N
fertilization on production levels. However existing models do not always accurately simulate
yield and grain protein concentration and only a few of them are parameterized for durum
wheat. Moreover, they do not take into account the constraints met by farmers that lead (or
not) to a decision to adopt a particular irrigation strategy.
The aim of this study is to improve a generic biophysical crop model to obtain a specific
crop model for durum wheat which simulates yield and grain protein concentration under
different water and nitrogen availabilities.
Methods
The proposed method is based on six main steps:
1) A soil-crop model has to be chosen based on its ability: (i) to represent the main
processes of the crop growth such as biomass accumulation and grain N accumulation,
(ii) to take into account water and nitrogen dynamics and (iii) to provide outputs at the
crop management scale for irrigation and N fertilization.
2) A database of field observations is needed to calibrate and evaluate the model. It has to
encompass observations from a range of contrasting irrigation and N fertilization
strategies made in pedoclimatic situations typical of durum wheat growing areas.
3) The model has to be parameterized using the information in the database while limiting
as much as possible compensation effects between parameters due to mathematical
optimization.
4) An in-depth model evaluation has to be conducted to assess the accuracy and the
robustness of the model and to identify processes that need to be improved.
5) “Faulty” processes have to be changed by new model equations (options) identified from
the literature or coming from other models.
6) New model evaluations have to be conducted before and after parameterization and after
equation implementation in order to compare the accuracy and the robustness of the
different options.
Results and discussion
The STICS model was chosen as the base model for improvement. STICS is a generic model
that simulates crop growth, grain yield and grain nitrogen concentration affected by nitrogen
212
Session B3: Cropping systems modelling
and water stress at field scale with a daily time step (Brisson et al., 2003). A total of 763
year/site/management combinations distributed over 13 years (1995–2007) and 10 sites and
representing 19 durum wheat cultivars were introduced into the database. Parameter
estimation was conducted on a subset of the total database following several independent
steps and using a simplex algorithm. Evaluation of the effectiveness of the modified model
was based on the comparison between observed and simulated values using Root Mean
Square Error and model efficiency as indicators. Equations from SIRIUS (Jamieson &
Semenov, 2000), AZODYN (Jeuffroy, 1999; Jeuffroy et al., 2000) and CERES-WHEAT
(Ritchie et al., 1998) were analysed to identify new formalisms (Figure 1). We introduced
new equations into STICS in order to provide a new version specific for durum wheat that can
simulate more accurately grain yield and grain protein concentration for various cultivars
under different simulation options.
The next step of the project will be to develop a decision support model that represents
farmers’ decision rules and to connect this decision model to the biophysical one in order to
evaluate strategies that jointly manage irrigation and nitrogen fertilization.
Figure 1. Variables, parameters and stress indices used in the models STICS and AZODYN to
simulate Grain nitrogen concentration. Boxes are state variables, squares are model
parameters and hexagons are related to water and nitrogen stress. Solid arrows indicate
information flows. Dotted arrows indicate that the state variable with the minimum value is
considered to determine the value of N accumulation.
References
Brisson, N., et al., 2003. Eur. J. Agron. 18: 309-332.
Jamieson, P.D. & M.A. Semenov, 2000. Field Crops Res. 68: 21-29.
Jeuffroy, M.H. & S. Recous, 1999. Eur. J. Agron. 10: 129-144.
Jeuffroy, M.H., et al., 2000. In: Eds P. Maillard & R. Bonhomme, Fonctionnement des
peuplements végétaux sous contraintes environnementales. INRA, Paris, pp. 289-304.
Ritchie, J., et al., 1998. In: Eds G.Y. Tsuji, et al., Understanding options for agricultural
production. Kluwer Academic Publishers, Dordrecht, pp. 79-98.
213
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Optimization of upland rice breeding programme based on a crop
modelling approach
A.B. Heinemann
Embrapa Rice & Beans – Rod. GO-462, km 12,
75375-000 Santo Antônio de Goiás, GO, Brasil
Contact: alexbh@cnpaf.embrapa.br
Introduction
Sub-tropical flooded rice cultivated in the South of Brasil accounts for 65% of Brazilian rice
production. However, in the South the area available for increasing rice production is limited.
Also for food safety and logistical reasons, it is not appropriate to concentrate rice production
in one region. Hence, there is increasing interest in improving the upland rice systems of
Goiás State, in the Brazilian savannas. In the early 1970s, at the start of the development of
savanna agriculture in Brasil, upland rice was grown in Goiás State. However, from 1985
onwards, these efforts were primarily targeted on the more favourable climatic zones
(Pinheiro et al., 2006) in the northwest of Brasil as a consequence of direct selection for grain
yield and quality, and blast resistance. The average upland rice yields in Goiás State are only
1800 kg ha–1 due to high spatial and temporal rainfall variability. An environmental
characterization of Goiás State, based on stress patterns could improve the local upland rice
breeding programme, allowing the identification and development of superior genotypes for
particular regions that could be combined into elite varieties. This study explores how to
adjust the local breeding systems to optimally fit the range of environments found in Goiás
State. The objectives were to (a) determine the drought patterns for upland rice in Goiás State;
and (b) develop modelling strategies that can improve the efficiency of the upland rice
breeding programme there.
Methods
In this study, the target population environment (TPE) corresponds to Goiás State, located in
the central part of Brasil. To characterize TPE, 12 locations with soil and climatic data were
selected. The most common soil types, covering 46% of the region, are Oxisols and Ultisoils.
Soil water holding capacity is about 100 mm m–1. Rooting depth is generally limited by soil
acidity which increases with depth, and may be as shallow as 0.3 m under low input management. This was taken into account in the modelling study by setting the maximum rooting
depth to 0.4 m (low input scenario) and 0.8 m (well managed cropping system scenario). One
reference upland rice varietal type (short cycle), based on the characteristics of the most
commonly planted genotypes in the region, was parameterized using the crop model RICE06,
implemented on the ECOTROP modelling platform of CIRAD (Kouressy et al., 2008).
The relative water stress impact on yield (RWSI) was evaluated by expressing attainable
yield as a fraction of potential yield. Simulations were undertaken for a range of sowing dates,
sites and years, using recommended management practices. Sowing dates were defined at 15day intervals during the main planting season defined as 01 November–31 December.
To develop a typology of drought patterns for rice, a matrix consisting of location, sowing
date, year and growth phase (100 °Cd periods) was established for the simulated mean daily
water stress index. Following Chapman et al. (2000), a hierarchical agglomerative clustering
method was used to classify the drought stresses into three main groups based on the
similarities in the phenological patterns of water stress index.
Results and discussion
Three predominant stress patterns were observed, called low (L), mid-season (M) and
terminal (T) stress (Figure 1). The most severe but less frequent stress pattern, T (Table 1)
214
Session B3: Cropping systems modelling
began about 600 °Cd after emergence (38 DAE), at the beginning of reproductive phase and
was most intense between 800 (52 DAE) and 900 °Cd (59 Days After Emergence). The T
stress pattern also affected grain filling, particularly under conditions of shallow soil, and
reduced potential yield by 50% in both soil types. The L pattern was most frequent for deep
soils (46%) and the M pattern for shallow soils (48%) with smaller relative effects on yield.
The overall, weighted yield reduction confounding all stress patterns was 18% for deep soil
and 36% for shallow soil (Table 1).
The results of the present study on rice TPE indicated that yield reduction caused by
drought (RWSI) is less than 50% overall for both soil depths (Table 1). On deep soils, the L
(for both short and medium duration TPEs), plus the M (short-duration) stress patterns
represent 84% of all the stress pattern frequencies, i.e., the drought environment that causes
yield reduction of < 50% occurs with a frequency of 20% in deep soils. Consequently, for the
deep soil, water deficit is not the main constraint to be addressed by the upland rice breeding
programme, especially if it is largely avoided by planting around November 15.
The frequency of severe drought is likely to be much higher for shallow soils, which in
Goiás State are typical of small, family based holdings associated with low inputs. An
additional drought screen (i.e., specific site, planting date combination) may therefore be
necessary to ensure that germplasm can be selected that is suitably adapted to the types of
stress experienced in shallow unimproved soils.
Table 1. Potential yield (GYpot, without water limitation), attainable yield (GYatt, with water
limitation), relative water stress impact (RWSI) on yield and stress frequency (S.F).
GYpot
GYatt
RWSI
S.F. GYpot
GYatt
RWSI
S.F.
Stress
–1
–1
–1
–1
type
(kg ha ) (kg ha )
(%)
(%) (kg ha ) (kg ha )
(%)
(%)
0.8 m Soil Depth
0.4 m Soil Depth
L
3937
3718
6
46
3927
3516
10
23
M
4120
3378
18
38
4027
2742
32
48
T
4217
1982
53
16
4270
1634
62
29
Average
4051
3311
18
4074
2599
36
L/0,4 m
M/0,4 m
T/0,4 m
L/0,8 m
M/0,8 m
T/0,8 m
1.0
Stress index
0.9
0.8
0.7
0.6
REPRODUCTIVE
GRAIN FILLING
1300
1200
1100
1000
900
800
700
600
500
400
300
0.5
Figure 1. Stress patterns from the cluster
analysis. The legends indicate the
different stress types (L - low; M - midseason; T - terminal) and rooting depth
limitations (0.4 and 0.8 m).
Crop Grow th (C.day)
References
Chapman, S.C., et al., 2000. Aust. J. Agric. Sci. 51: 209-221.
Kouressy, M., 2008. Agric. and Forest Meteorol. 148: 357-371.
Pinheiro, B. da S., et al., 2006. Field Crops Res. 97: 34-42.
215
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Extracting useful information from daily GCM rainfall
for cropping system modelling
A.V.M. Ines, J.W. Hansen
International Research Institute for Climate and Society
The Earth Institute at Columbia University, NY, USA
Contact: ines@iri.columbia.edu
Introduction
There is a growing interest in using daily GCM outputs for crop yield predictions because of
GCM capability to simulate seasonal climate in advance and the daily weather requirements
of cropping system models (Ines & Hansen, 2006; Baigorria et al., 2007, 2008). However,
given that GCM simulate climate at very coarse spatial resolution, the simulated weather are
usually biased relative to a point location limiting their direct applications to crop simulation
models (Cohen, 1990). GCM can predict better inter-annual climate variability than the
absolute meteorological values (Baigorria et al., 2008). GCM rainfall for example is
characterized by having high frequencies and low intensities resulting into too many rainfall
events with too little amounts during rainy seasons (Ines & Hansen, 2006). The biases in
rainfall frequency and amounts will result into under-prediction of crop yields if directly used
in crop simulation models (Mavromatis & Jones, 1998). For this reason, Ines & Hansen
(2006) developed a simultaneous bias correction approach to correct the biases in GCM
rainfall frequency and intensity. While they observed improvements in simulated yields with
bias corrected daily GCM rainfall, the under-prediction of crop yield was still evident which
they attributed to the inability of the bias correction method to correct the temporal structure
of daily GCM rainfall mismatching the lengths and distributions of dry spells. In this paper,
we explore a strategy to extract useful information from the bias corrected GCM rainfall to
improve the simulation of dry spell lengths and their distributions with the aim of predicting
better crop yields using cropping system models.
Methods
We used the bias correction method of Ines & Hansen (2006) to correct rainfall frequencies
and intensities of ECHAM4.5 grid daily rainfall (1970–1995, 24 ensemble members)
encompassing our study location in the Machakos district of eastern Kenya (1°35’ S, 37°14’
E). After bias correction, we derived the monthly rainfall frequencies for all years, for each
ensemble member and then used them to condition a weather generator (Hansen & Ines,
2005). The stochastic disaggregation was applied in two ways; (i) for each ensemble member
using their respective rainfall frequencies to condition the weather generator, and (ii) using
the mean of monthly rainfall frequencies from all ensemble members to condition the weather
generator. We used 24 realizations in our stochastic disaggregation matching the number of
ensemble members of the GCM, and then we replicated ten times. We tested the performance
of our bias corrected-disaggregated GCM based-rainfall by linking them to CERES-Maize to
simulate yield. A local maize variety ‘Katumani B’ was used in the simulations. The cropping
season is from October-February. For comparison, we ran the crop model using observed,
uncorrected and bias-corrected GCM rainfall. For all simulations, we used long-term average
minimum, maximum temperature and solar radiation values conditioned on dry and wet days.
Results and discussion
Table 1 shows the performance of the bias corrected-disaggregated GCM-based rainfall on
crop yield prediction. The combined bias correction (BC)-stochastic disaggregation (DisAg)
using mean of monthly rainfall frequencies of all ensemble members to condition the weather
generator (BC-DisAg2) showed the best performance in which all indicators have improved
216
Session B3: Cropping systems modelling
Table 1. Performance of the proposed combined BC-DisAg on prediction of yield. Note: R correlation; MBE - mean bias error; d - index of agreement (Willmot, 1982), MSE - mean
squared errors; subscripts R and S indicate random and systematic components.
Method
R
MBE
d
MSE
MSER
MSES
(-)
(Mg ha–1)
(-)
(Mg ha–1)2 (Mg ha–1)2 (Mg ha–1)2
Uncorrected
0.61
–2.35
–1.14
6.61
1.06
5.55
BC only
0.70
–1.04
0.50
1.95
0.86
1.09
BC-DisAg1
0.63
–0.41
0.63
1.22
1.01
0.21
BC-DisAg2
0.73
–0.20
0.74
0.91
0.79
0.12
considerably compared to bias correction only. The yield mean bias is reduced and the
prediction skill and d-statistics have increased suggesting that the BC-DisAg2 method
attempts to correct random and systematic error components of the simulated yield. Using
each ensemble member rainfall frequency to condition the weather generator (BC-DisAg1)
only performed fairly compared to using BC only; most of the improvements are observed on
yield mean bias. This performance of BC-DisAg2 can be attributed to its better skill in
simulating dry spell length distributions (Figure 1c) as the weather generator we used applied
a hybrid first- and second-order Markov process in simulating rainfall occurrence. BC only
slightly improved the dry spell length distributions (Figure 1b) compared to Uncorrected
GCM runs (Figure 1a) and since GCM rainfall is highly auto-correlated, the longer dry spells
have not been corrected by deterministic bias correction.
0.5
Observed
Probability
0.4
0.3
Uncorrected GCM runs
Observed
Observed
Bias Corrected GCM runs
BC ‐ Disaggregated GCM runs
4
Mean observed
Mean GCM
0.2
0.1
a
b
c
0
0.0
0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50
Dry spell length
Figure 1. PDF of dry spell lengths during anthesis period (Nov. 15–Dec. 31) from (a)
uncorrected, (b) BC only and (c) BC-DisAg2 (best trial).
References
Baigorria, G., et al., 2007. Clim. Res. 34: 211-222.
Baigorria, G., et al., 2008. Agric. For. Meteor. 148: 1353-1361.
Cohen, S.J., 1990. Bull. Am. Meteorol. Soc. 71: 520-526.
Hansen, J.W. & A.V.M. Ines, 2005. Agric. For. Meteorol. 131: 233-246.
Ines, A.V.M. & J.W. Hansen, 2006. Agric. For. Meteorol. 138: 44-53.
Mavromatis, T. & P.D. Jones, 1998. Clim. Change 41: 583-614.
Willmot, C.J., 1982. Bull. Am. Meteorol. Soc. 63: 1309-1313.
217
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Uncertainties in simulating crop performance in degraded soils and low
input production systems
J.W. Jones, J. Naab, D. Fatondji, K. Dzotsi, S. Adiku, J. He
P.O. Box 110570, University of Florida, Gainesville, FL 32611, USA
Contact: jimj@ufl.edu
Introduction
Models are increasingly being used as research tools to predict outcomes of cropping systems
under different climate, soil, and management conditions in both developed and developing
countries. Many papers have been published on research that demonstrated that cropping
system models perform adequately for the intended purposes and then used the models to
study impacts of different cultivars, irrigation, fertility, and cultural management practices on
yield and other predicted outputs. Many of these studies have emphasized the importance of
incorporating climate uncertainty to adequately consider risks to production and profitability
(e.g., Hammer & Muchow, 1991; Thornton & Wilkens, 1998). Impacts of, and adaptation to,
climate change have made extensive use of crop models, and now these models are being
used for simulating years of crop rotation for projecting long term changes in soil carbon and
other properties that affect sustainability of production in different environments.
Typically in these studies, researchers are interested in only a few factors that may limit
growth and yield, such as water and nitrogen in addition to climate. There may be a number
of factors that limit production in farmers’ fields that present challenges to model users. This
is particularly true in developing countries where (i) soils are low in fertility and hold very
little water, (ii) where farmers typically do not apply fertilizer or irrigate, (iii) there is
considerable annual variability in climate, and (iv) subsistence farming practices are used.
Matthews & Stephens (2002) pointed out the difficulties of obtaining inputs to operate
cropping system models in developing countries, and this presents one of the challenges in
reliable use of cropping system models in those countries. However, there is another major
challenge that has been ignored in most previous studies, even if inputs for model studies
were collected – uncertainty in environmental parameters and inputs. Model developers
routinely emphasize the importance of obtaining accurate genetic coefficients in order to
apply cropping system models in local studies, which suggests that without reliable values for
these parameters, the models will not adequately simulate the responses to climate and
management that users are studying. Little attention has been given to uncertainty analysis of
different soil and management inputs relative to prediction of cropping system performance.
Based on studies conducted in West Africa, we hypothesize that uncertainty in soil
parameters, initial conditions, and nutrient inputs contribute more to prediction uncertainty
than genetic parameters in low input, rainfed cropping systems on low fertility and water
holding soils.
The objectives of this study were (1) to quantify the uncertainty of simulated crop
production as affected by estimated uncertainties in important soil parameters, soil initial
conditions, genetic coefficients, and nutrient inputs; and (2) to determine which uncertain
inputs dominate model output uncertainty under different environments and management
conditions.
Methods
We conducted an uncertainty analysis on three cropping systems with contrasting climate,
soil, and management inputs, two in West Africa and one in the USA. We also quantified the
contributions of uncertainties in different soil and genetic parameters on that uncertainty using
global sensitivity analysis. The first crop was irrigated maize grown in Gainesville, Florida
with high nitrogen input to represent a typical high input production system. We used the
218
Session B3: Cropping systems modelling
fully irrigated, high nitrogen fertilizer treatment from a 1982 experiment (Bennett et al.,
1989). The data from this experiment are distributed in DSSAT (Jones et al., 2003) and have
been used in training workshops for and testing different maize model versions. The second
crop was low input maize represented by an experiment conducted in 2004 by Naab (2005) in
Wa, Ghana, using the treatment that had no nitrogen added but adequate P and other inputs.
The third crop was millet with the data taken from a millet experiment conducted in 1999 in
Damari, Niger by Fatondji et al. (2006) in which a rainfall harvesting planting technique was
compared with traditional flat planting. We used their flat planted, 3000 kg ha–1 manure
application treatment as the basis for the uncertainty and sensitivity analysis. Soils in all three
experiments were sandy with low water holding capacities and low organic matter contents.
The two experiments in West Africa were rainfed. Soil parameters, initial conditions,
management details, and genetic coefficients were based on prior simulation studies by the
authors of the studies.
A Monte Carlo approach was used to generate 5000 sample sets of inputs for each of the
three sites from estimated distributions of soil inputs (water holding limits, rainfall runoff
parameter, initial values of soil water, carbon, nitrate, and ammonium, and manure input N
concentration) and genetic coefficients. Soil water limits were sampled taking into account
correlations between the inputs, based on the approach used by He (2008). The DSSATMaize and Millet models were simulated using each sample of inputs to create 5000 outputs
for each site. Distributions of simulated biomass and grain yield were generated and
compared with observed responses. Also, first order and total sensitivities of these outputs to
each uncertain input were computed using the Sobol decomposition of variances.
Results and discussion
Results are presented showing empirical model output distributions of biomass and grain
yield for each of the three cases. First order global sensitivities of these outputs to each of the
uncertain inputs are compared and ranked for each case study. The implications of this
research will be discussed relative to use of cropping system models in harsh environments,
the importance of characterizing and communicating uncertainties in model outputs when
analysing cropping system performances, suggestions for reducing uncertainties when
evaluating models for specific experimental conditions, and practical implications relative to
communicating results to decision makers.
References
Bennett, J.M., et al., 1989. Field Crops Res. 19: 297-311.
Fatondji, D., et al., 2006. Nutr. Cycling in Agroecosyst. 76: 203-217.
Hammer, G.L. & R.C. Muchow, 1991. In: Climatic Risk in Crop Production: Models and
Management for the Semi-Arid Tropics and Subtropics. CAB International, Wallingford,
UK, pp. 205-232.
He, J., 2008. Best Management Practice Development with the CERES-Maize Model for
Sweet Corn Production in North Florida, PhD Dissertation, University of Florida, 329 pp.
Jones, J.W., et al., 2003. Eur. J. Agron. 18: 235-265.
Matthews, R.B. & W. Stephens, 2002. Crop-soil Simulation Models: Applications in
Developing Countries. CABI Publishing, Wallingford, UK, 280 pp.
Naab, J.B., 2005. Measuring and Assessing Soil Carbon Sequestration by Agricultural
Systems in Developing Countries, 2004 Annual Report. Savanna Agricultural Research
Institute, Wa, Ghana.
Thornton, P.K. & P.W. Wilkens, 1998. In: Understanding Options for Agricultural
Production: Systems Approaches for Sustainable Agricultural Development. Kluwer
Academic Publishers, Dordrecht, pp. 329-345.
219
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Crop systems evaluation for strategic investment decisions
J. Koo, S. Wood, Z. Guo
International Food Policy Research Institute, 2033 K St., NW, Washington, DC 20006, USA
Contact: j.koo@cgiar.org
Introduction
Investors in development and policymakers have a growing need for data and tools to help
them target and prioritize interventions so as to achieve the greatest possible food security
impacts in cost-effective ways. Although crop systems modelling has been used successfully
as a cross-disciplinary decision support tool, many crop models have been developed for use
on comparatively small unit areas, which are typically assumed to be homogeneous in crop
growth, environmental conditions, and management regimes. However, typical investment or
policy-making decision covers large areas with significant heterogeneity in crop growth
conditions and thus crop responses to potential interventions. Potential benefits and pitfalls of
increasing the spatial resolution in crop systems modelling studies of large areas have been
discussed and tested, but the choices of models and scenarios in these studies have been
relatively simple and limited in scope, due to incomplete spatially-explicit knowledge of the
crop systems attributes, a lack of sufficiently high-resolution data at regional or global scales
required to evaluate such models and results, and the computational cost of running these
models over a large area at a high resolution under multiple scenarios. Despite many
challenges, a Sub-Saharan Africa (SSA) and South Asia (SA) region-wide decision-support
platform employing crop systems models at its core is being developed by a network of crop
modellers facilitated by the HarvestChoice project (http://HarvestChoice.org) to help meet the
demand for a large scale analytical research platform with accompanying data. The platform
is being developed to assess the spatially-explicit likely-changes in crop yield and biomass
production and regional resource and market implications of technology and management
innovations, under different scenarios of changes in crop management practices and
agricultural policy at a grid-based regional scale. This paper discusses the platform
development process, and presents a series of use-cases describing potential analytical uses of
the platform for the high-spatial-resolution regional estimation of: (1) crop calendar, (2)
baseline crop productivity level, and (3) impacts of biotic and abiotic stresses on crop
production, under a range of management and climate change scenarios.
Methods
The analytical crop systems modelling platform is being developed by focusing on following
issues: (1) characterizing cropping systems based on a blend of macro and micro data, (2)
compiling/developing regional-scale crop model input data layers in a standard format,
including climate/weather, soil properties, pest and disease prevalence, and model evaluation
datasets, (3) developing a set of regional-scale scenarios of potential changes in R&D
investment, technology and market access, and farm-scale adoption, (4) developing a database
of spatially-explicit pre-run crop systems model results that allow users to assess the potential
impacts of scenarios of changes based on the user-defined baseline, and (5) developing a suite
of tools to provide an easy interface for running crop growth models in a grid-based regional
context with pre-loaded yet modifiable datasets and user-defined scenarios.
Based on the preliminary version of the datasets and tools being compiled and developed, a
spatially-explicit database of model results based on following use-cases was developed:
1. Crop- and site-specific length of growing period (LGP) and cropping calendar and their
comparison with existing regional databases developed using rule-based systems and
observations;
220
Session B3: Cropping systems modelling
2. Baseline production under a range of supplementary nutrient and water management
practices for two representative varieties for each crop, under current and future climate,
with:
a. Impacts of drought on growth stage-specific water stress and yield,
b. Pest infestation based on the surveyed and simulated pest prevalence data and estimated
damage,
c. Drought and pest infestation occurring simultaneously.
Each case used DSSAT v4.02 (Jones et al., 2003) model for maize and groundnut at a 5 arcminute (approximately 10 km) grid in SSA. For current and future climate, 30 years of daily
weather for each grid cell was simulated using WeatherMan (Hoogenboom et al., 2006),
MarkSim (Jones & Thornton, 2000), and WorldClim (Hijmans et al., 2005). Representative
soil profiles for major soils in each grid cell were compiled using the Harmonized World Soil
Database 1.0 (FAO, IIASA, ISRIC, ISSCAS, JRC, 2008) and the ISRIC-WISE 1.1 (Batjes,
2002). Crop yield in each grid cell was estimated as a weighted average based on the relative
area of each soil type.
Results and discussion
Overall outputs of the use-cases showed that
location-specific information on the potential
changes of crop production under a wide range
of scenarios can be useful to help set
investment priorities and indentify efficient
technology development and adaptation
strategies. These outputs will be integrated with
kg/ha
501 - 1,000
an economic evaluation to provide insights into
1,001 - 1,500
the potential spatial and socio-economic
1,501 - 2,000
2,001 - 2,500
patterns of impact under simulated investment,
2,501 - 3,000
policy, and environmental change scenarios at
3,001 - 3,500
3,501 - 4,000
regional and global scales.
> 4,000
The performance of crop systems models
can be compromised if the environmental or Figure 1. Simulated rainfed/low-input
production conditions of interest greatly differ maize yield potential in Sub-Saharan
from those used during model development and Africa and CIMMYT maize field trial
evaluation, and the risk can be relatively higher locations (dots).
in the regional-scale application. As part of the
analytical platform development, a set of field measurement data are being compiled from
multiple sources (e.g., CGIAR field trial database, national/sub-national agricultural census
data) for use as modelling control points. Rigorous efforts on model evaluation will be critical
in gaining the confidence of scientists, analysts, investors, and policymakers.
References
Batjes, N.H., 2002. A Homogenized Soil Profile Data Set for Global and Regional
Environmental Research (WISE, version 1.1) 2002/01. International Soil Reference and
Information Centre, Wageningen, The Netherlands.
FAO, IIASA, ISRIC, ISSCAS, and JRC, 2008. Harmonized World Soil Database [Online].
Available by FAO (Rome, Italy) and IIASA, Laxenburg, Austria.
Hijmans, R.J., et al., 2005. Int. J. Climatol. 25: 1965-1978.
Jones, J.W., et al., 2003. Eur. J. Agron. 18: 235-265.
Jones, P.G. & P.K. Thornton, 2000. Agron. J. 92: 445-453.
Hoogeboom, G., et al., 2006. Weatherman. Release 4.0.2.0. Univ. of Hawaii, Honolulu, HI.
221
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A modelling approach to optimize the management of pea-barley
intercropping in Europe
M. Launay, N. Brisson, S. Satger, H. Hauggaard-Nielsen, G. Corre-Hellou, E. Kasynova,
E.S. Jensen, M.J. Gooding, C. Dahlmann, P. Von Fragstein, A. Pristeri, M. Romeo, M. Monti
Unité Agroclim, INRA, Site Agroparc, 84914 Avignon Cedex 9, France
Contact: Marie.Launay@avignon.inra.fr
Introduction
Intercropping, defined as two or more species grown mixed together in the same field during
the same cropping season, may be a good strategy for low input agriculture. Indeed, available
resources, such as light, water and nutrients can be more efficiently utilized by the intercrop
as a result of complementary acquisition strategies. Grain legume-cereal intercrops, and
especially pea-barley intercrops, have been shown to maximize the biomass produced per unit
of nitrogen available (Hauggard-Nielsen & Jensen, 2001). Given the complexity of
intercropping systems modelling can be a helpful tool to improve understanding of
interspecific interactions (Brisson et al., 2004) and to test various agronomic strategies. We
have used a modelling approach to test pea-barley systems in three sites in three northern
European countries involved in the INTERCROP FP5 EU project, in order to take into
account soil and weather variability. The strategies being tested, which were selected
according to their feasibility for farmers, were compared by answering the following
questions: (i) how do pea-barley intercrops (IC) compare with sole crops (SC) as regards yield
and quality and their stability? (ii) in order to increase the crops’ access to light, would it be
worth delaying the sowing date of the quick-growing species? (iii) how can the use of
nitrogen resources be optimized by choosing the most suitable preceding crop and/or the most
appropriate soil? To answer those questions, a virtual experiment was carried out using 10
years’ weather data from Angers (France), Reading (UK) and Roskilde (Denmark) sites.
Materials and methods
The STICS growth model (Brisson et al., 2003), extended to intercropping (Brisson et al.,
2004), has already been evaluated on pea-barley intercrops (Corre-Hellou & Brisson, 2007).
Weather records for about 10 years from Angers (France), Reading (UK) and Roskilde
(Denmark) were used for the simulations. The simulation protocol depended on the question
being addressed:
• Pea-barley IC/SC comparison in terms of quantity, quality and stability: the simulation
design mimicked the experimental design from the INTERCROP project (pea SC, barley
SC, additive pea-barley IC and replacement pea-barley IC);
• Delay in sowing dates: 2 weeks’ delay between pea and barley sowing for both density
treatments (additive and replacement designs) were simulated.
• Test of soil nitrogen availability: the agronomic design mimicked the additive and
replacement pea-barley IC of the INTERCROP experimental design, using two hypotheses
of soil nitrogen availability factorially combined to give four combinations:
- 20% increase in the soil organic nitrogen content compared to the actual one of the field
sites.
- two initial mineral nitrogen profiles at sowing accounting for two types of preceding
crop (oilseed rape and maize, for example).
Results and discussion
(i) The model showed the value of intercropping for making the best use of environmental
resources as regards yield and quality, but indicated a site effect as regards yield stability.
Total IC yield was more stable over the years than any of the SC yields. It identified factors
222
Session B3: Cropping systems modelling
explaining the competitiveness of the two species: pea growth appears to be strongly linked to
soil moisture through nodulation activity, and barley yield was determined by nitrogen uptake
through rooting, and by light interception due to its height relative to pea.
(ii) Sowing barley before pea involves a reduction of only 5% of the grain yield Land
Equivalent Ratio (LER) averaged over all three sites. There was no effect in Angers and only
a small one in Roskilde. Although the effect was not statistically significant even in Reading,
a reduction of 10% would be agronomically important. In Angers, where barley is more
competitive than peas, it appears that sowing peas before barley is preferable while in
Roskilde and Reading, simultaneous sowing is equally good (Table 1). This shows that the
sowing strategy must be adapted to the location, being dependent on temperature and thus
latitude.
Table 1. Simulated Grain yield LER for additive and replacement designs, when sowing
barley before pea, barley and pea at the same date, and pea before barley, at Danish, English,
and French locations.
Grain Yield LER
Additive
Barley before Same sowing
Pea before barley
pea
date
1.17 a
1.21 a
1.20 a
Location Roskilde (Denmark)
Reading (UK)
1.12 a
1.24 a
1.22 a
Angers (France)
1.11 a
1.11 a
1.28 b
All sites
1.13 a
1.19 a
1.24 a
Values are the mean over years. In each row, grain yield LER values followed by the same
letter are not significantly different at p < 0.01.
(iii) Grain yield and plant N content responses to soil initial mineral nitrogen content are
higher than to soil organic nitrogen. Increasing nitrogen availability at the beginning of
growth reduces the proportion of pea in the IC yield, as already shown experimentally for a
wheat-pea intercrop by Ghaley et al. (2005). Barley responded more to the organic nitrogen
increase in Reading (UK) and Angers (France) than in Roskilde (Denmark). In Roskilde, low
temperatures slow down mineralization during winter (barley grain yield is increased by 9%
in Reading whereas it does not increase in Roskilde when organic nitrogen content was 20%
more than the nominal content of the soil). Similarly, the bigger barley response to initial
mineral nitrogen content in Roskilde and Angers was due to wetter conditions in Reading
resulting in increased leaching of nitrogen (barley grain yield was increased by 27% in
Roskilde and Angers whereas it was increased by only 13% in Reading with a 12 g N m–2
increase in initial mineral N content). These results underline the importance of a long term
strategy, including mineralization management through organic residue supply and rotation
management, in order to increase soil nitrogen availability, and favour total grain yield and N
accumulation of the intercrop.
References
Brisson, N., et al., 2003. Eur. J. Agron. 18: 309-332.
Brisson, N., et al., 2004. Agronomie 24: 409-421.
Corre-Hellou, G. & N. Brisson, 2007. Field Crops Res. 103: 76-85.
Ghaley, B.B., et al., 2005. Nutrient Cycling in Agroecosystems 73: 201-212.
Hauggaard-Nielsen, H. & E.S. Jensen, 2001. Field Crops Res. 72: 185-196.
223
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Establishment of quinoa (Chenopodium quinoa Willd.) and simulation of its
cultivation with the STICS crop model
Sophie Lebonvallet, Nadine Brisson, Jean-Pierre Raffaillac, Marie Launay
Agroclim, INRA Domaine St Paul Site Agroparc, 84914 Avignon Cedex 9, France
Contact: foflebonval@yahoo.fr
Introduction
Quinoa is a species native to South America, grown mainly on the high Andean plateau
(Mujica et al., 2001). It appeared on the international market about twenty years ago, thanks
partly to its high nutritive value and partly to its production under an organic farming label.
Bolivia is today the main world exporter (Laguna, 2002).
This work aims to explain the issues involved in quinoa establishment and the methods of
cultivation on the Bolivian Altiplano. The climatic risks for agriculture there are very high
(low and very variable rainfall, frequent frosts, especially during the final stages of
development of the crop) which keep yields fairly low, from 0.6 to 1 t ha–1 on average. Crop
establishment, therefore, should not be too late, to avoid the crop experiencing frost after
flowering, which would destroy the yield (Bois et al., 2005). However it should not be too
early, so that the young plants are not subjected to too severe drought. The sowing period is in
fact just at the start of the rainy season; the soil water content at this time depends on the
water accumulated during the previous rainy season, its movement during the winter and on
the first rains, which are irregular, of the new rainy season (Vacher et al., 1994). The cultural
practices used, the success of establishment and good growth of the young plants are,
therefore, determinant phases for the rest of the growing period and the yield.
Methods
Due to the small number of studies and small amount of data available on quinoa, we decided
to study all these aspects by modelling and
numerical experimentation, by adapting the crop
model STICS (Brisson et al., 2003) to quinoa and to
its growing conditions in the Bolivian Altiplano.
This allowed us to rapidly generate a large amount
of data.
Formalisms had to be added to or modified in
STICS in order to better represent the conditions of
the Bolivian Altiplano and the cultivation techniques
used for quinoa. They are concerned in particular
with taking account of altitude in the climatic data,
the effect of drought, cold and soil capping on seed
germination and establishment of the crop, and also
the traditional method of sowing in poquets (Figure
1). The model was parameterized for quinoa using
Figure 1. Sowing in row/poquet.
data from a set of experimental plots laid down by IRD
between 2001 and 2006 (Raffaillac et al., 2007).
Results and discussion
STICS proved to be able to simulate the growth of quinoa with an efficiency of 60-80%,
which is relatively satisfactory.
Firstly, the model was used just to study the effect of cultural practices on the
establishment of quinoa. These practices differ from the north to the south of the Altiplano in
terms of date, depth and type of sowing (row or poquet), ploughing date, and the
224
Session B3: Cropping systems modelling
susceptibility of the soil to crusting. All these effects were factorially combined for sites,
years and soil types, leading to the simulation of more than 11,000 cases. From this largescale numerical experimentation there emerged various results underlining the adaptation of
techniques to their environment. First of all, these results illustrate the variability in optimal
sowing depths and dates, which explains the difficulties encountered by the farmers and their
low yields. They also enable us to better understand the role of traditional sowing in poquets
(higher soil moisture, reduced time for germination and emergence). They underline the value
of traditional manual sowing at several depths (a method of control against drought and frost
due to the successive flushes of emergence) and explain the early ploughing, in the middle of
the previous rainy season more than six months before sowing in the driest regions, to provide
the maximum amount of stored soil water at the time of sowing.
Secondly, the model was used to study the effect of cultural practices over the whole of the
growing period, and thus on the yield. The results obtained, particularly for the effect of
sowing date, illustrate perfectly the difficulty of managing the climatic risks in the Altiplano,
which is very much a compromise between a yield which is moderate but reliable and one
which is high but uncertain. Furthermore, paradoxically, this shows how a variable sowing at
several depths reduces the yield variability and leads to much more uniform yields than in the
case of sowing at a single depth.
Thirdly and finally, different rotations were studied to see the effect of the duration of
fallow. In fact, traditionally the quinoa crop was preceded by 10-50 years of fallow, which
allowed regeneration of the soil fertility thanks to the natural vegetation which could grow
and then be incorporated into the soil (Hervé et al., 2003). However, since the quinoa
« boom » and its entry into world commerce, the necessary increase in production has led to
these fallow periods being reduced, sometimes to only two years, which could explain, among
other reasons, the observed fall in yields. Thus, the simulations carried out can show that the
reduction in the length of the fallow period results in a fall in the soil water and nitrogen
reserve, which results in practice in lower yields, especially in the southern region of the
Altiplano, where the conditions are much more difficult.
Lastly, the use of the model, guided by these questions, has allowed us to better understand
the system, the issues involved in crop establishment, and the sustainability problems in this
environment which is restrictive and subject to great variability. This work has some
limitations in terms of processes being simulated too simplistically compared with reality, due
to a shortage of data because of difficult experimental conditions or processes which are not
taken into account; but in a situation where quinoa is beginning to be studied more and more,
it nevertheless opens the way to more advanced research, both to improve the tool created
here, but also to use this tool in other work on sustainability, effects of climate change,
cartography, decision aids etc.
References
Bois, J.F., et al., 2006. European Journal of Agronomy 25: 299-308.
Brisson, N., et al., 2003. European Journal of Agronomy 18: 309-332.
Hervé, D., et al., 2003. Nature Sciences Societes 11: 243-254.
Laguna, P., 2002. Debate Agrario, Analysis y Alternativas 34: 95-169.
Mujica, A., et al., 2001. In: Quinua, ancestral cultivo andino, alimento del presente y futuro,
Eds Mujica et al., CIP, UNAP. FAO, CD Cultivos Andinos, Chili.
Raffaillac, J.P., et al., 2007. Una red agronómica quinua en el altiplano boliviano. Congreso
Internacional de la Quinua, CIHDE & Univ. Ed. A. Prat, 23-26 octubre de 2007, Chili.
Vacher, J.J., et al., 1994. In: Eds. D. Hervé, et al., Dinamicas del descanso de la tierra en los
Andes. IBTA-ORSTOM, La Paz, pp. 127-139.
225
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
An integrative modelling approach to simulate the agricultural system
through a combination of a decisional,
technical and biophysical sub-system
A. Mérot, J.E. Bergez, J.C. Mailhol, J. Wery
UMR System, SupAgro, 2 place Pierre Viala, 34090 Montpellier, France
Contact: anne.merot@supagro.inra.fr
Introduction
Because farmers allocate resources, such as equipment or water at farm level, innovative
cropping systems should be designed at this level. However it is not easy to perform because
modellers have to cope with a diversity of processes and scales. Integrative modelling is
therefore required to evaluate cropping systems ex-ante.
Due to environmental impacts, conventional water management practices in southern
France (Crau Plain) must be modified in order to ensure higher water productivity and lower
environmental impacts while maintaining high hay production. In this context, farmers, water
managers and policymakers expressed the need for information to face these challenges. In
this work, we illustrate the use of a conceptual framework (Le Gal et al., 2007), based on
three sub-systems (biophysical, technical, decisional), to guide a process of integrative
modelling. The example of the dynamic bio-decisional model IRRIGATE, representing the
functioning of the cropping system at farm scale for water management, will be used as an
example.
Methods
The conceptual framework is based on three sub-systems (Le Gal et al., 2007). The decision
sub-system can be defined as the farmer’s combination of objectives, rules, indicators and
constraints to adopt and implement technical practices. The technical sub-system determined
by the decision sub-system is the ‘combination of techniques applied by farmers to the
biophysical system in order to fulfil production objectives’ (Le Gal et al., 2007). The
biophysical sub-system under the control of the technical sub-system is defined by the
interactions between physical and biological components, such as water, soil, climate and
pests, and the plant growth/development processes. Each sub-system can be described as a
combination of components in interaction, status variables with associated dynamic, temporal
and spatial scales, a set of inputs and a set of outputs.
General Soil and Plant
characteristics
Farm design (Water access, channels,
fields, plots) and soil specificities
Decisional system
(cropping system)
Climate
series
Water access
context
Biophysical system
(Field, plot)
Water supply, irrigation duration
Hydraulic context
Irrigation model
Technical system
Decision model
θs, LAI
Irrigation and Mowing plan
Water supply,
Crop-soil model
Agronomic context
∑T, θs
Irrigation and mowing calendars
Labour
Water consumption
Yields
Water loss
Figure 1. The conceptual model of the studied agricultural system based on the framework.
226
Session B3: Cropping systems modelling
Integrative modelling process can be summarized by four scientific questions: Which
processes of each sub-system to model? Which formalisms to represent these processes? How
to integrate these formalisms in a temporal and spatial multi-scale context? How to evaluate
the integrative model? Based on the conceptual framework and following a six-step
methodology, we built the integrative bio-decisional model composed of modules each corresponding to a sub-system: (i) identification of the limits of the system; (ii) identification for
each sub-system of the status/flow variables, their interactions in terms of information
exchange and fluxes, the main dynamics of the variables; (iii) identification of the temporal
and spatial scales associated with these processes; (iv) qualification/quantification of these
interactions by in-field experiments and surveys; (v) simplification of the modelling of each
sub-system by choosing the most important processes and associated formalisms; (vi)
integration of the modules to simulate the functioning of the cropping system and evaluation.
This last stage implies also the transition from the conceptual model to the numerical model.
Results and discussion
Description of the three sub-systems (Figure 1): The biophysical sub-system is a soil-plantwater system. The water component corresponds to the dynamics of run-off and infiltration.
The plant component has three status variables (actual evapotranspiration, Leaf Area Index
and biomass) and represents plant growth and water uptake in soil. The soil component
connects the water and plant components. It is linked to the soil water dynamics. The
technical sub-system combines two components, one for each practice simulated: mowing and
irrigation (Mérot et al., 2008a). The decisional sub-system is divided into nine components,
each corresponding to a decision rule. The decision process addresses the question of ‘do I
mow or do I irrigate’, with an interactions between the decisions (Mérot et al., 2008a). The
links with the status of the biophysical sub-system is given by three indicators: the soil water
deficit, the leaf area index, and the crop development stage. The decisional sub-system
generates daily irrigation and mowing schedules for the entire season, which composes the
technical sub-system.
The bio-decisional model IRRIGATE (Figure 1): The bio-decisional model IRRIGATE was
built from the conceptual framework. It includes: (i) a dynamic decision model at farm scale
(Mérot et al., 2008a) simulating irrigation and mowing management, (ii) a dynamic crop
model at field scale (Mérot et al., 2008b) simulating plant and soil functioning in relation to
water supply, and (iii) a hydraulic model at border scale (Mailhol & Mérot, 2008) simulating
an irrigation event according to both the biophysical and the water distribution context. They
are written as three independent and modular modules.
The conceptual framework helps us to structure cropping system models as a combination
of a biophysical, technical and decisional system. It leads to a better delimitation of the three
sub-systems, to a better description of their major interactions and it facilitates up-scaling and
down-scaling. The same scientific investment is done for the analysis and the modelling of
each sub-system. Compared to classical crop models, the representation of the biophysical
system can be different when linked with the decision system. However, this
conceptualization does not avoid three important difficulties of an integrative approach: (i) a
higher complexity in the modelling process, (ii) generality and robustness of the integrative
model, (iii) evaluation of the model.
References
Le Gall, P.Y., et al., 2007. Farming Systems design 2007. Sept 10-12, 2007, Catania.
Mailhol, J.C. & A. Mérot, 2008. Irrigation Science 26(4): 289-302.
Mérot, A., et al., 2008a. Agricultural Systems 98: 108-118.
Mérot, A., et al., 2008b. European Journal of Agronomy 29: 163-174.
227
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Modelling cover-cropped vineyards with APES:
A modular simulation platform
A. Metay, C. Gary, E. Casellas, A. Ripoche, K. Kansou, J. Wery
Supagro UMR SYSTEM (Supagro, INRA, CIRAD), 2 Place Viala,
34060 Montpellier Cedex 2, France
Contact: metay@supagro.inra.fr
Introduction
Introducing a cover crop into vineyards is a recently introduced practice in wine production
areas (Celette et al., 2008) that provides a variety of environmental and agronomic services.
For example, it reduces runoff and, therefore, soil erosion (Battany & Grismer, 2000; Le
Bissonais et al., 2004). In fertile zones, it can also inhibit the excessive vegetative
development of grapevine that would be unfavourable to high quality wine production (Smart
et al., 1991) and favourable to fungal diseases. However, the risk of excessive competition for
soil water between the intercrop and the grapevine has markedly limited the adoption of this
practice in Mediterranean regions. As it is not easy to quantify the trade-off over long-term
experiments, we have taken advantage of the APES simulation platform which was designed
within the EU FP-6 SEAMLESS project (Van Ittersum et al., 2008) to assess the agronomic
performance and environmental impacts of these cropping systems. Specific components
were developed for the simulation of cover cropped vineyards and evaluated with a field
experiment database in south-eastern France.
Methods
Modelling APES (Agricultural Production and Externalities Simulator) is a field scale
modular simulation system describing the soil-plant-atmosphere behaviour under specific
climate and technical management conditions (http://www.apesimulator.org/). The concepts
used in the simulation of biophysical processes are based on existing models and published
studies. The main novelty is in the modular programming approach used in the model implementation. The simulation of cover cropped vineyards requires the use of both the Crop and
Tree components, and of climate, soil and resource arbitration components.
In the Crop component, crop development depends on thermal time. Crop growth is based
on radiation interception by green leaf area and its conversion into dry matter; the model
simulates both potential growth and attainable growth, as affected by water and nitrogen
limitations. Dry matter and nitrogen are partitioned between the growing organs on the basis
of partitioning tables with fixed parameters for each phase of the crop cycle. Parameters and
modelling approaches can differ according to the crop simulated; in the present study, the
‘Winter barley’ parameter set was used.
The Tree component is a generic perennial woody crop model that has been calibrated for
grapevine. It simulates the vineyard growth, production and product quality. It is similar to
the Crop component, with some additional features that are specific to woody crops. As vines
are generally planted in rows, their canopy is heterogeneous, and so the Tree component
converts the daily carbon increment into an increase in crown dimensions which then affect
the light interception component (Pronk et al., 2003). As woody crops are perennial, carbon
and nitrogen can be stored during a crop cycle and used during the next one, particularly at
budbreak (Castelan-Estrada, 2001). The Tree component calculates quality variables for fruit
crops. For grapevine, the dynamics of fruit water and sugar contents depends on the thermal
time during the second phase of berry growth (Ollat et al., 2002).
Intercropping can be simulated in APES. The light interception component enables light to
be partitioned between two crops (grapevine and barley in the present case). The water and
nitrogen uptake components compute water and nitrogen demand in each soil layer in
228
Session B3: Cropping systems modelling
proportion to the fraction of root from each crop that is present in the layer.
Field experiment The experiment (Celette et al., 2008) was carried out from 2003 to 2006 on
a vineyard (cv. Aranel/Fercal) near Montpellier (43°31' N-3°51' E). The climate is Mediterranean with a water deficit occurring from April to September i.e. from budburst to harvest.
Soil was a deep and homogeneous clay loam with low organic matter and nitrogen content.
Two treatments were studied: a vineyard on bare soil obtained with chemical weed control
and a vineyard with a temporary intercrop of barley (Hordeum vulgare L.) sown between the
rows in autumn, covering about 60% of the area and destroyed at grapevine flowering.
Results and discussion
The APES model adequately simulated grapevine phenology (not shown) and inter-annual
variations of yield for vineyard on bare soil (Figure 1). Canopy development of grapevine was
simulated with and without competition with barley (Figure 2). Cover cropping generated a
depression of grapevine LAI, particularly during the dry summers of 2005 and 2006. The
model also simulated a decreasing LAI for barley year after year, which was not observed in
the field.
LAI vineyard (Vineyard + Barley system)
LAI barley (Vineyard + Barley system)
LAI vineyard (Vineyard only)
4
6
observed
simulated
kg/ plant
5
3
4
3
2
2
1
0
2003
2004
2005
2006
Figure 1. Fruit fresh yield (kg/plant) in a
vineyard on bare soil, from 2003 to 2006.
1
0
2004
2005
2006
Figure 2. Simulated LAI of vineyard and barley
for two treatments: vineyard only, vineyard and
barley with water and nitrogen stresses.
These first results show that the setting of various components in the APES modular platform
was fully functional for adaptation to a perennial plant intercropped with an annual crop and
provided crop and soil outputs within realistic ranges of values. In the present research, the
consequences for grapevine yield of inter-annual climate variability were correctly simulated.
Competition for soil resources was simulated and generated trends of vegetative development
that were qualitatively correct. Nevertheless, the effectiveness of the resource arbitration
simulation should be explored in more detail particularly with respect to intercrop behaviour.
References
Battany, M.C. & Grismer, M.E., 2000. Hydrol. Process. 14: 1289-1304.
Castelan-Estrada, M., 2001. PhD Thesis INA Paris-Grignon, France, 121 pp.
Celette, F., et al., 2008. Eur. J. Agron. 29: 153-162.
Le Bissonais, Y. et al., 2004. Agronomie 24: 129-136.
Ollat, N., et al., 2002. Journal International des Sciences de la Vigne et du Vin 36: 109-131.
Pronk, A., et al., 2003. Neth. J. Agric. Sci. 51: 279-295.
Smart, R.E., et al., 1991. South African Journal of Enology and Viticulture 11: 3-17.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
229
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A systems approach to optimizing agronomic and economic performance
of legume-based organic maize production
S.B. Mirsky, J.R. Teasdale, R.A. Mangum, M.A. Cavigelli, J. Maul, J.T. Spargo
10300 Baltimore Avenue, Bldg 001, Rm 117,
BARC-West Beltsville, Maryland, 20705, USA
Contact: steven.mirsky@ars.usda.gov
Introduction
Incentive-driven agricultural management practices (i.e., cover crop subsidy programs) that
enhance nonmarket agro-ecosystem services could have long-term impacts on environmental
stewardship as well as crop productivity (Boody & DeVore, 2006; REAP, 2007). Cover crops
are a multi-functional management tool that can improve soil fertility, reduce soil erosion,
sequester soil carbon, increase soil water infiltration and storage and suppress weeds. This
multi-functional management practice can improve soil conservation, reduce nontarget
pollutants, and decrease a cropping system’s C footprint. There is interest in developing cover
crop-based organic no-tillage (no-till) field maize (Zea mays L.) production systems which
combine two different soil building strategies, organic farming and no-till farming. However,
this management system has had limited success in achieving adequate crop yields, primarily
due to the challenges in achieving adequate cover crop-based fertility and weed management.
Weed suppression from natural levels (3,000-10,000 kg ha–1) of cover crop surface residues
has been incomplete (Mirsky et al., 2007). When utilizing cover crops for weed suppression
in reduced tillage and herbicide cropping systems, integrating multiple management tactics is
necessary (Williams et al., 1998). Coupling a new approach to cover crop surface residue
management for organic no-till systems, the roller/crimper, with additional cultural and
mechanical weed management tactics may increase the potential success of organic no-till
field crop production. More information is also needed to determine how nitrogen (N)
availability to maize in congruence with these weed management tactics would be affected by
rolling versus ploughing and delaying maize planting dates. Therefore, the objective of this
experiment was to evaluate how cover crop and weed management tactics in tillage and
reduced tillage organic field maize production optimizes nitrogen availability, weed
suppression, and maize performance.
Methods
This experiment was established on a certified organic field site in Beltsville, Maryland
(United States). Vicia villosa Roth. was established in the fall and terminated in the spring
with either a roller/crimper or by disking. Since V. villosa can only be killed adequately by
rolling after flowering begins (Teasdale & Rosencrance, 2003), the cover crop was terminated
on three dates by ploughing: early May (vegetative); mid-May (75% flowering); and late-May
(early pod stage); and on two dates by rolling, mid- and late-May. Maize was planted 7 to 10
days after rolling or disking. A no management control and cultivation treatment (disked
twice; rotary hoed twice; and sweep cultivated three times for tilled plots; and high residue
cultivated no-till plots) was included for both tilled and no-till systems. Weed-free plots were
included for each cover crop kill method by termination date treatment to assess the influence
of treatment effects on maize in the absence of weed competition. Nitrogen availability and
associated maize stress were determined by measuring: (1) N content of cover crop and
maize; (2) soil mineral N content to a depth of 30 cm prior to significant soil mineralization in
late March and when maize was at the V5 stage (i.e., pre-sidedress nitrate test); (3) maize ear
leaf N; and (4) chlorophyll content of ear leaf. Weed and maize mergence patterns, weed and
maize biomass, and maize population and grain yield were also measured. In addition, relative
230
Session B3: Cropping systems modelling
water content and soil water content to a depth of 30 cm were assessed to estimate drought
stress.
Results and discussion
Delay in cover crop termination increased maize yields due to higher levels of N availability
compared to earlier termination dates. Delayed maize planting permitted more cover crop
biomass production and N content, and greater available N for a succeeding maize crop. This
response in V. villosa and the resulting influence on maize has also been demonstrated by
other researchers (Clark et al., 1994; Wagger, 1989). Despite no seed treatment or pesticide
application, maize in the tilled system had a higher population density than in the no-till
system. This is in contrast to the no-till system which had significantly decreased maize
populations due to a seedling pest (i.e., Delia platura Mg.). Delay to the latest cover crop
termination date reduced the effects of insect damage to the maize population. This was
attributed, in part, to faster emergence and growth of maize due to warmer soil temperature.
However, a delay in cover crop termination can also decrease crop available soil water due to
increased evapotranspiration from the cover crop and increased heat units in the summer.
As expected, the ploughed maize had superior weed control. However, delays in cover
crop termination resulted in greater weed suppression for both tilled and no-tilled systems.
The timing of cover crop termination also influenced the weed community structure. The
effect of shifts in weed community structure on weed competitiveness and crop performance
will be discussed. Structural equation models will be employed to determine the relationship
between cover crop and weed management on maize agronomic and economic performance.
Optimization of organic no-till maize agronomic and economic performance will require
cropping system models that can integrate both the challenges of legume-based fertility,
drought stress, and pest management.
References
Boody, G. & B. DeVore, 2006. Bioscience 56: 839-845.
Clark, A.J., et al., 1994. Agon. J. 86: 1065-1070.
Mirsky, S.B., et al., 2007. In: 7th European Weed Research Society Workshop on Physical
and Cultural Weed Control. Salem, Germany.
REAP, 2007. Resource Enhancement and Protection Act of Pennsylvania.
http://www.cbf.org/REAP. (accessed 8 Nov. 2007).
Teasdale, J.R. & R.C. Rosencrance, 2003. Amer. J. Altern. Agric. 18: 95-102.
Wagger, M.G., 1989. Agron. J. 81, 533-538.
Williams, M.M., et al., 1998. Weed Sci. 46: 595-603.
231
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Development of generic management rules for
crop growth simulation models
R.J. Oomen, H. Hengsdijk, O. Therond, J.E. Bergez, G. Russell, E. Casellas,
S. Janssen, M.K. Van Ittersum
Wageningen University, Haarweg 333, 6709 RZ Wageningen, The Netherlands
Contact: Roelof.Oomen@wur.nl
Introduction
Crop growth simulation models are powerful tools for assessing the effects of management
practices or policy changes on the environment and agricultural production (Van Ittersum &
Donatelli, 2003). Assessments can be made at a very fine scale, e.g., for one farm, or at a
much coarser scale, e.g. assessing the impacts of the implementation of the Water Framework
Directive on European agricultural systems. These models simulate the behaviour of crops or
cropping systems taking into account soil, climate, agro-management and crop characteristics.
An example of such a model is the Agricultural Production and Externalities Simulator
(APES; Donatelli et al., 2009), which allows entire crop rotations to be simulated taking into
account the specific characteristics of such systems. Whatever the scale of the assessment, the
usually large data demands of these cropping systems models have to be met. Obtaining the
required input data becomes a challenge when cropping systems models are applied to
assessments for large and heterogeneous areas such as Europe. Soil and daily climate data are
available from the European databases assembled by the Joint Research Centre of the
European Union. However, little detailed data is available on the management of crops, i.e.
the timing, amount and type of input used and the application method. This agro-management
information refers to all operations relevant for crop production, for example tillage, sowing,
fertilization and irrigation (crop protection is not taken into account). Surveys aimed at
collecting detailed management information are cumbersome as they require the involvement
of many experts and they only provide average information on agro-management that is in
fact flexible and variable in practice as it depends on factors associated with soil, climate and
crop conditions.
In this paper, we describe an approach that has been applied to generate agro-management
information for 21 crops and 19 regions in the EU on the basis of ‘easily-obtainable’ survey
data and generic expert rules. We compare this agro-management information with detailed
agromanagement information from a survey carried out in four EU regions. This research has
been conducted within the SEAMLESS project (Van Ittersum et al., 2008).
Methodology
Two surveys were carried out, one for collecting detailed agro-management information for
crops grown in four EU regions, and one for collecting more general and easy-to-obtain
information on agro-management in 19 regions (Borkowski et al., 2007). The detailed survey
contains data on crop sequence (i.e. rotations) and inputs, outputs and timings of all
management operations. The simple survey has crop sequence data combined with aggregate
agro-management data and inputs and outputs, and provides only the average timing of
sowing.
Both in the detailed survey and in APES, management events are defined as the
combination of a timing rule, governing when the event takes place, and an ‘impact’,
describing the agro-management operation (Donatelli et al., 2006). These operations are
grouped into tillage, irrigation, fertilization, crop protection and sowing/planting and
harvesting. Timing rules can be based on actual observed dates or on (a combination of)
biophysical conditions like rainfall, plant available water (PAW) in the soil, crop phenological
stage, etc.
232
Session B3: Cropping systems modelling
The crop experts amongst the authors of this contribution developed a set of generic agromanagement rules and operations for management events that allow simple survey data to be
transformed into APES compatible inputs.
Results and discussion
The simple agro-management data to be extended by these so-called ‘simple management
rules’ consists of an average sowing week, fertilizer amount, and an irrigation amount and
average number of irrigation events. When converted, the result is a set of two tillage events,
a sowing and a harvesting event, zero to four fertilization events and an irrigation rule
triggering the specified number of irrigation events.
Figure 1. Example of the application of simple management rules to wheat agro-management.
The generated detailed agro-management has been compared to the surveyed detailed agromanagement from a number of sample regions and the simple management rule parameters
were tuned such that they result in the best matching agro-management in most cases. For
further tuning of the rule parameters we compare simulations of both types of detailed agromanagement, i.e. surveyed and generated, in terms of both yields and externalities.
Conclusions
Using crop growth simulation models with little information on crop management poses a big
challenge for coarse scale applications. However, given easily obtainable management data
we developed a consistent and generic set of rules summarizing agricultural knowledge.
These rules can then be used to generate detailed event-based agro-management for regions
where such information is lacking (see Casellas et al., 2009). The generated agromanagement data are evaluated as to their suitability for using in a coarse scale application for
answering research questions at, e.g., country or continental level.
References
Borkowski, N., et al., 2007. In: Eds M. Donatelli et al., Farming Systems Design 2007. Int.
Symposium on Integrated Analysis of Farm Production Systems. Catania, Italy.
Casellas, E., et al., 2009. AgSAP Conference 2009, these proceedings.
Donatelli. M., et al., 2006. Proceedings of the iEMSs 2006 Conference, Vermont.
Donatelli, M., et al., 2009. In: Eds F. Brouwer & M.K. Van Ittersum, Springer Academic
Publishing, Dordrecht. (forthcoming)
Van Ittersum, M.K. & M. Donatelli, 2003. Eur. J. Agron. 18: 187-394.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
233
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Crop rotation modelling for integrated environmental assessment:
Case study results and empirical evidence
Martin Schönhart, Erwin Schmid, Uwe A. Schneider, Peter Liebhard
Department of Economics and Social Sciences, BOKU Vienna,
Feistmantelstrasse 4, 1180 Vienna, Austria
Contact: martin.schoenhart@boku.ac.at
Introduction
Crop rotations are commonly regarded as an important factor for the sustainability of
agricultural systems (Ball et al., 2005). They influence environmental quality (e.g., soil,
water), the utilization of natural resources, and the aesthetics of agricultural landscapes.
Furthermore, they are a common way of managing risks on the farm and are the consequence
of farmers’ decision making. Integrated agricultural land use models increasingly
acknowledge the role of crop rotations in assessing the economic and environmental impacts
of agricultural production systems. However, since empirical data on crop rotations are
usually not, or hardly, available, different types of crop rotation model have been developed.
The software tool ROTAT (Dogliotti et al., 2003) for example has been developed to provide
all possible combinations of crop rotations from a given set of crops and according to
agronomic criteria. Similarly the rule-based model ROTOR (Bachinger & Zander, 2007)
generates agronomically sustainable crop rotations taking into account plant nutrition, weed
infestation and phytosanitary effects. Detlefsen & Jensen (2007) developed a network flow
model to find optimal rotations by maximizing the gross margins for each sequence of crops.
Constraints are the shares of crops to be modelled in the year of interest. El-Nazer & McCarl
(1986) developed a procedure for the identification of optimal long-run crop rotations to be
integrated into linear programming (LP) models. In their study, empirical data on the
economic effect of crop sequences was derived from regression analysis. While the first
example considers (only) agronomic criteria in delineating crop rotations, the latter two
generate optimized rotations based on gross margins derived from empirical data.
Here, we present the crop rotation optimization model CropRota that integrates agronomic
criteria and historical crop mixes at farm or regional scales. CropRota is validated against
seven years of field survey data from an Austrian farm.
Methods and data
CropRota is a linear optimization model that derives optimal crop rotations from observed
crop mix data. Crop mix data are relative shares of crops on a farm, region, or any other
spatial unit in a time period. A crop rotation cross (‘Fruchtfolgekreuz’) serves as a value point
matrix ranging between 0 and 10 points. It provides agronomic judgments about all available
pre-crop – main crop sequences. Judgments are ranging from ‘agronomically impossible’ (0
points) to ‘highly recommendable’ sequences (10 points). This value point matrix is
maximized in CropRota using six decision variables that contain the optimal sequence of
crops indexed by one to up to six crop sequences per rotation. The decision variables also
provide the relative shares of the crop rotations based on the observed or statistical crop
mixes, which are used as constraints. Therefore, this procedure implicitly integrates economic
relationships. Consequently, one can construct the crop mix by using crop rotations and their
relative shares. CropRota currently considers more than 30 crops. Further constraints can be
implemented to restrict prohibitive crop combinations and sequences in a rotation, e.g., a
limitation on the frequency of sugar beet, potatoes or rape seed to one year in four in a
rotation.
In this preliminary analysis, we apply CropRota to a crop farm in Lower Austria. The
model was validated by comparing the model output with observed crop sequences in the
234
Session B3: Cropping systems modelling
fields of the farm between the years 2001 and 2007. In this period, 10 different crops were
grown in 28 fields occupying 42.5 ha in total. For the final analysis we will validate CropRota
using the same type of data from several hundred farms, which have been made available
recently.
Results and discussion
Two approaches A and B have been assessed. In approach A, crop rotations for each single
year were modelled based on annual crop mixes. In B, one average crop mix was calculated
for the years 2001 to 2007. In approach A, CropRota derived 26 different crop rotations with
two to six crop sequences. In approach B, 10 crop rotations were identified. To validate the
model we compare the frequencies of observed and modelled two-crop sequences.
In Figure 1 such a comparison between observed and modelled outcomes is shown for
likely crop sequences with sugar beet (SBEET) and winter wheat (WWHEAT). For example,
corn silage (SILCORN) follows sugar beet (SBEET) on about 2% of the total farm land. For
most cases, annually modelled crop rotations (approach A) are closer to the observed crop
sequences than in approach B. Furthermore, the assumption of constant annual crop mixes as
well as the restrictions on the frequency of certain crops (e.g., sugar beet) in a rotation have
proven to be acceptable for this farm.
Figure 1. Comparison between model and empirical results on crops following SBEET (left)
and WWHEAT (right).
The pilot case study application of CropRota to a single farm and some simple model
validation exercises show the potential of CropRota as a valuable tool for identifying likely
crop rotations. These crop rotations provide important information for biophysical process
models and economic land use optimization models, which are increasingly used to jointly
assess economic and environmental impacts of alternative agricultural systems.
References
Bachinger, J. & P. Zander, 2007. Eur. J. Agron. 26: 130-143.
Ball, B.C., et al., 2005. Can. J. Soil Sci. 85, 557-577.
Detlefsen, N.K. & A.L. Jensen, 2007. Agric. Syst. 94: 566-572.
Dogliotti, S., et al., 2003. Eur. J. Agron. 19: 239-250.
El-Nazer, T. & B.A. McCarl, 1986. Amer. J. Agric. Econ. 68: 127-136.
235
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Implications of heterogeneous urinary-N return for modelling pasture
production and N losses from grazed pastures under climate change
V.O. Snow, M. Lieffering, B.R. Cullen, I.R. Johnson
AgResearch – Grasslands Research Centre, PB 11-008, Palmerston North, New Zealand
Contact: Val.Snow@agresearch.co.nz
Introduction
Urine excretions in grazed systems cover only a few percents of the paddock area during any
one grazing resulting in very high N concentrations, 500–1000 kg N ha–1, in urine-affected
areas of the paddock. These areas are the primary source of leaching and are also important in
denitrification processes. Previous work (Snow et al., 2008) concluded that the non-linearities
in the processes that contributed to leaching necessitated an explicit representation of urine
patches in the model, but that simpler uniform urine return models could adequately simulate
pasture production and N fixation. Here, we reconsider that conclusion in the context of
climate change simulations and examine the effect of climate change on pasture species
composition and on N2O emissions.
Methods
EcoMod is a dynamic biophysical simulation model (Johnson et al., 2008) with a simulation
option that allows heterogeneous urine return (Snow et al., 2008). Two soil types were
simulated; a clay loam and a sand with plant available soil water of 210 and 100 mm
respectively to 1 m deep. Under current climatic conditions, Manawatu, New Zealand is on
the ecotone of C4 occurrence in pasture. A pasture species mix of a C3 grass (perennial
ryegrass), a legume (white clover) and a C4 grass (Paspalum) was set up in the model. The
mix of the three species responded dynamically to the growing conditions, changing within
and between simulations. Every 28 days the pasture was grazed. At each grazing 15% of the
ingested N was removed from the simulation to mimic export in animal production. The
remaining nutrients were partitioned between dung and urine according to the algorithm in
Johnson et al. (2008). Dung was returned uniformly to the paddock. For each soil type and
climate change scenario, simulations were run with the urine returned heterogeneously to 2%
of the paddock at each grazing (Het) or returned uniformly over the paddock (Uni).
Three climate scenarios were simulated.
The base ‘2000’ scenario was obtained
for the period 1972 to 2007 for the
Manawatu climate (New Zealand, 40.35
ºS 175.61 ºE). Average air temperature
was 13 °C (ranging between monthly
averages of 4.3 °C in July and 22.3 °C in
January). Annual rainfall was 951 mm/yr
distributed evenly through the year and
had low variability with 740 and 1201
mm yr–1 in the driest and wettest years,
respectively. Two climate change
scenarios were simulated based on the
‘A1B’ projection (MfE 2008). The ‘2030’
scenario was for a 0.7 °C increase in
temperature and a 10% increase in
rainfall. The ‘2080’ scenario was a 2.1 °C
increase in temperature and a 20%
2.5
2.0
1.5
1.0
2000
2030
2080
Climate change scenario
Figure 1. Box-whisker plots showing the ratio
between modelled N2O denitrification by the Het
model and the Uni model for the clay loam soil.
The dashed line indicates the ideal 1:1 ratio.
236
Session B3: Cropping systems modelling
increase in rainfall. The values quoted are
the annual averages and actual changes
varied by season as described in MfE
(2008) and were supplied to the model by
modifying the weather file.
Clay Loam
0.0
0.1
nW
lu m
pa
0.5
rt io
as
0.6
2000 Uni
2030 Uni
2080 Uni
po
0.3
0.7
P ro
0.2
2000 Het
2030 Het
2080 Het
0.8
lo v
nP
rt io
po
eC
er
P ro
h it
Results and discussion
0.4
0.4
The differences between N leaching
0.5
modelled by the Het and Uni assumptions
0.3
(data not shown) were so large that it was
0.6
0.2
not possible to correct a Uni model to
0.7
0.1
appropriately simulate leaching in a grazed
0.8
system. The ratio between N2O emission
0.0
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
modelled by the Het and Uni assumptions
Proportion Ryegrass
for the scenarios on the clay loam soil are
shown in Figure 1. The moderate variation Figure 2. Ternary diagram showing the effect
and skew in the ratio for the ‘2000’ of heterogeneity assumption and climate
scenario, compared to values for leaching change scenario on pasture species mix. Only
shown by Snow et al. (2008), suggests that the clay loam data are shown.
provided year-to-year variation is not very
important the effect of urine patches on modelled denitrification can probably be corrected by
a rate parameter change. However different corrections would be needed for different soil
types (data not shown here) and for different climate change scenarios (Figure 1). For
example, if a Uni model was calibrated to give appropriate results it would overestimate N2O
emission for the ‘2080’ scenario by some 25%. A model with uniform urine return might also
be inappropriate when simulating mitigation measures directed at N2O emissions in grazed
systems.
There is a clear separation caused by climate change scenario with greater C4 dominance in
the ‘2030’ and ‘2080’ scenarios than the base ‘2000’ scenario (Figure 2). However spreading
the urine uniformly over the entire paddock suppressed the increase in the contribution of
white clover in the ‘2030’ and ‘2080’ scenarios to the species mix. There was a similar effect
on pasture protein concentration (data not shown). Other results showed that there were
relatively minor differences between the Het and Uni models for drainage and intake but
larger differences for N fixation.
These results suggest that, in grazed systems, a model used to assess the impacts of climate
change should include an explicit representation of urine patches. With the later climate
change scenarios pasture N concentration declined and so did the concentration of N in the
urine patches. However, the concentrations were still too high to allow reasonable simulation
of leaching, denitrification or pasture species mix with a model assumption of spatially
uniform return of urine.
References
Johnson, I.R., et al., 2008. Aust. J. Exp. Agric. 48: 621-631.
MfE, 2008. Climate change effects and impacts assessment. Ministry for the Environment,
Wellington. xviii + 149 p. (www.mfe.govt.nz/publications/climate/climate-change-effectimpacts-assessments-may08)
Snow, V.O., et al., 2008. Aust. J. Agr. Res. (in review)
237
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The interaction between sub-optimally available resources for crop
production as implemented in the simplified model FIELD
P. Tittonell, M. Corbeels, K.E. Giller
PPS, Wageningen University, P.O. Box 430, 6700 AK Wageningen, The Netherlands
Contact: pablo.tittonell@cirad.fr
Introduction
Crop production in the tropics, and particularly in sub-Saharan Africa (SSA), takes place
under sub-optimum conditions in terms of resource availability. FIELD (Field-scale resource
Interactions, use Efficiency and Long-term soil fertility Development) is a simplified model
of the cropping system developed with the primary aim of simulating interactions between
sub-optimally available resources (light, water, nutrients, labour) determining short term
responses and long term trends in crop productivity within smallholder farms (Tittonell et al.,
2007, 2008). The model simulates crop responses to applied and/or available nutrients (N, P,
K), considering the agro-ecological potential and climatic variability of a certain site, the
heterogeneity in soil fertility within the farm, and keeps track of changes in soil carbon and
nutrient stocks in time. Central to the calculation of crop yields is the simulation of the
interaction between water, N, P and K. Such interactions are represented following the
concepts used in the model QUEFTS (Janssen et al., 1990) but expressed in a generic rather
than empirical way to be able to perform dynamic, long term simulations. This allows an
implementation of a ‘Liebscher’ approach – where the effect of one limiting resource acts by
reducing the efficiency with which another resource is used – rather than a ‘von Liebig’ type
of function. Here, we illustrate this by showing how the interaction between nutrients taken
up by a crop are simulated in FIELD, with a few examples from SSA.
Resource availability, interactions and crop response
The approach taken in FIELD is represented with the 3-quadrant diagram of De Wit & Van
Keulen (1987) but including a fourth quadrant: resource availability (Figure 1). Nutrient
inputs to the soil (e.g., applied fertilizers or manure) may become more or less available
depending on their form, timing and placement, and on the characteristics of the
soil/landscape where the crop grows (the Av0 – AvI range); e.g., applied/available nutrients
may be temporarily or permanently ‘lost’ in P-fixing red soils or by N leaching in sandy soils.
Crop biomass
The ability of the crop to intercept and
RT3
CoI
take up the available nutrients depends
RT2
on root architecture and activity (the
In0 – InI range), and together with the
Agronomic
Ability of the
efficiency &
crop to
previous quadrant they determine
economic result
‘convert’
nutrient capture efficiency. Nutrients
RT1
Co0
taken up are ‘converted’ into crop
biomass with an efficiency that is the
Resource captured
inverse of the average nutrient Resource input
(e.g., applied)
(e.g., taken up)
concentration in different plant parts
(Co0 and CoI represent their maximum
Av0
Ability of the
concentration and dilution, respectively).
crop to
‘intercept’
Effect
of
The conjunction of these effects leads to
‘Soilscape’,
input & method
different patterns in the observed
In0
InI
response of crops to nutrient inputs (RT1,
AvI
RT2 and RT3), which determine the
Resource available
agronomic result or input use efficiency.
Figure 1. Four-quadrant diagram (see text).
The ability of a crop to convert resource
238
Session B3: Cropping systems modelling
A into biomass depends on the availability and uptake of resources B, C, etc. (Figure 1 upperright quadrant). For example, the N conversion efficiency (NCE) of tropical maize varies
between 60 and 170 kg DM kg N–1 taken up (based on maximum and minimum N
concentrations in plant); even when N is in ample supply, poor P availability or limited water
supply will lead to a relative ‘concentration’ of N in the plant (i.e., a poorer NCE). In the
model FIELD, the maximum NCE (or maximum dilution of N in plant) is corrected for the
availability of P and K; similarly, the maximum P conversion efficiency is corrected for N
and K; etc. These correction factors (CF) are calculated by relating the availability of the
balancing nutrient to its target value, which is derived from the product of the water-limited
production level and its average concentration in the plant.
Crop biomass (t ha-1)
N and P use efficiencies within heterogeneous smallholder farms – an example
The response to nutrient applications by crops growing on smallholder farms is highly
variable. On-farm research trials are conducted to analyse the origin and magnitude of such
variability and evaluate nutrient management strategies for improved crop productivity.
Figure 2 corresponds with the upper-right quadrant in Figure 1 and was built with data from
multi-locational on-farm trials in Kenya (Ky) and Zimbabwe (Zb) (Vanlauwe et al., 2006;
Zingore et al., 2007), and with simulated data using the simple FIELD approach as described
above.
n
tio
Kenya - simulated
lu
20
i
d
The prediction of crop yield across
Kenya - measured
N
um
Zimbabwe - simulated
fields with different soil types and history
m
i
ax
Zimbabwe - measured
M
16
of management, receiving varying doses
of N, P and K as organic and/or mineral
12
fertilizers was satisfactory (RMSE 2.2 and
–1
1.2 t DM ha , Ky and Zb), as was the
on
trati
8
prediction of N uptake (RMSE 22 and 30
ncen
o
c
mN
i mu
Max
kg N ha–1, Ky and Zb) (data not shown).
4
The prediction of NCE was within the
range of measurements at both locations
0
(Figure 2), but in the case of Kenya the
0
30
60
90
120
150
180
model tended to underestimate NCE. This
N uptake (kg ha-1)
could be due to an overestimation of water
Figure 2. Simulated and measured N uptake and crop
stress in FIELD, which would require
biomass (see text for explanation).
simple calibration against data from
different seasons and/or locations.
In conclusion, a generic approach to nutrient interactions allowed us to capture the
variability in crop responses observed in the field, as a result of spatial soil heterogeneity
(both inherent and due to past management), rainfall variability and/or poor agronomy. These
factors affect nutrient use efficiency operating mostly on the efficiency of nutrient capture
(Tittonell et al., 2007). A simple model with few parameters, adapted to the data-scarce
environments of SSA, is a useful tool for ex-ante assessment of the impact of soil fertility
management interventions. This is of particular importance in light of the major investments
in soil fertility planned for SSA (e.g., www.agra-alliance.org).
References
De Wit, C.T. & H. Van Keulen, 1987. Geoderma 40: 253-265.
Janssen, B.H., et al., 1990. Geoderma 46: 299-318.
Tittonell, P., et al., 2007. Field Crops Res. 100: 348-368.
Tittonell, P., et al., 2008. Agron. J. 100: 1511-1526.
Vanlauwe, B., et al., 2006. Nutr. Cycl. Agroecosyst. 76: 171-182.
Zingore, S., et al., 2007. Field Crops Res. 101: 296-305.
239
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The effect of conservation agriculture practices in Europe
M. Van der Velde, F. Bouraoui, P. Prosperi, J.M. Terres, B.S. Elbersen
IES-Joint Research Centre of the European Commission,
Via Fermi 2749, 21027 Ispra (VA), Italy
Contact: marijn.van-der-velde@jrc.it
Introduction
The Common Agricultural Policy (CAP) is increasingly aiming at enhancing the
sustainability of agro-ecosystems while responding to the sustained demand for quality and
reliable food production. Environmental requirements such as cross compliance are part of the
policy instruments available to address the integration of environmental concerns into the
CAP. We wish to assess the potential environmental benefits of adopting conservation
agriculture practices and adopting standards of good agricultural and environmental
condition. Ideally we wish to identify those regions in Europe where the adoption of
conservation farming and/or soil conservation practices would have the biggest potential
benefits for the environment. At the same time we aim to study if trade-offs exist between
conservation and production through the modification of plant-soil-water interactions in the
agricultural system.
Methods
We used a pan-European spatialized version of the EPIC model (Williams, 1995) that runs on
a 10 by 10 km grid with relevant meteorological, land use, terrain, soil and management
information (Figure 1). Daily meteorological data were obtained from JRC’s MARS climatic
database given on a 50 by 50 km grid, land use information was obtained combining satellite
land cover data (CORINE 2000) and farm structure survey statistics on crop areas (Grizzetti
et al., 2007), digital terrain information was derived from SRTM (Shuttle Radar Topographic
Mission), soil data were obtained from the European Soil Bureau Database (ESBD 2.0) and
sowing dates were determined using a potential heat units programme (for more details see
Williams, 1995; Bouraoui & Aloe, 2007; Van der Velde et al., 2008). This approach allows us
to include relevant soil functions such as water storage capacity in evaluating environmental
responses to changes in crop management practices.
Figure 1. Conceptual design of the
EPIC-EAGLE geospatial modelling
framework.
240
Session B3: Cropping systems modelling
We consider the environmental effects of no-till and the introduction of a cover crop (clover)
in respectively barley and maize. The no-till scenario is compared to conventional tillage
under barley. The cover crop scenario is compared to irrigated and non-irrigated maize
without the growth of the cover crop during winter months. Currently, the EPIC-EAGLE is
set-up so that plant nutrient requirements are satisfied by chemical NO3-N application. This
may be interpretable as a form of optimal precision fertilizer management.
Results and discussion
We emphasize that the simulation results should solely be used as indications, and should be
used along with other information to come to informed policy decisions. Our modelling
results indicate that no-till practices will be effectively reducing erosion across Europe
compared to conventional tillage practices. Modelling does not reveal any significant
differences in yield between tilled and non-tilled barley. Also, crop water stress is not very
different. Largest reductions in soil loss are obtained in areas with steeper slopes and/or high
annual precipitation rates such as the north west of Spain. In absolute terms low-lying areas,
such as the Netherlands, benefit much less from no-till practices. The reduction of erosion
also leads to lower N transported with runoff. Nitrate leaching and fertilizer use are fairly
similar between the no-till scenario and the reference.
Preliminary results indicate that the introduction of a cover crop leads to a reduction of
erosion under non-irrigated maize over most of Europe, although certain parts experience a
slight increase in erosion (~20%). Modelled yields are lower when a clover cover crop is
grown in the winter months. The explanation is given by the increase in water stress that
generally occurs at the start of the maize growing season associated with soil moisture
depletion by cover crop water use in early spring. Soil moisture depletion by the cover crop
potentially is a problem if rainfall in the month previous to maize sowing is not sufficient to
replenish soil moisture in the root zone. Similarly, reduced soil moisture leads to smaller
maize cover and thus a larger exposure of bare soil and subsequent increases in erosion that
may not be off-set by the reduced erosion in the winter months. The model results suggest
that a trade-off may exist between lowering erosion and increasing water stress (decreasing
yields) when using a cover crop in water-limited environments. Modelling further suggest that
if a slight increase in water stress is allowed, for example determined by a minimal reduction
in yields <0.25 ton ha–1, erosion will be substantially reduced across Europe when using a
cover crop in non-irrigated maize.
References
Bouraoui, F. & A. Aloe, 2007. European Agrochemicals Geospatial Loss Estimator: Model
development and applications. EUR – Scientific and technical research series, 22690,
Office for Official Publications of the European Communities, Luxembourg, 118 pp.
Grizzetti, B., et al., 2007. Spatialised European nutrient balance. EUR – Scientifical and
Technical Research series, 22692, Office for Official Publications of the European
Communities, Luxembourg, 98 pp.
Van der Velde, M., et al., 2008. Global Change Biology 14: 1-14.
Williams, J.R., 1995. In: Ed. V.P. Singh, Computer models of watershed hydrology. Water
Resources Publications, Littleton CO, USA, pp. 909-1000.
Disclaimer: The statements expressed here are purely those of the authors and may not in any
circumstance be regarded as stating an official position of the European Commission
241
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Toward gene-based crop simulation models for use
in climate change studies
Stephen M. Welch, Amity Wilczek, Judith L. Roe, Brooke Moyer, Martha Cooper,
Johanna Schmitt, Sanjoy Das, Praveen Koduru, Xinye Cai
Kansas State University, Department of Agronomy, Manhattan KS 66506, USA
Contact: welchsm@ksu.edu
Introduction
Several lines of reasoning point toward the urgent need to develop crop simulation models
that are more firmly grounded in genetics that those that are currently available. For example,
current general climate models are including increasingly realistic plant physiology in their
descriptions of land surface processes (e.g., Foley et al., 1996). These sub-models do allow
the distribution of plant types to change under the impact of competition, but they make no
provision for changes in life history or physiology due to natural selection. However, Ward et
al. (2000) has shown that CO2 itself can act as a selection agent leading, in as few as five
generations, to earlier flowering plants with reduced final biomass. Even without the onus of
climate change, both commercial and private breeders are under increasing pressure both to
shorten product development cycles and/or to develop genotypes with prescribed
characteristics. Molecular genetics, in principle, has the potential to meet this need. A major
difficulty, however, has been interrelating features at the gene level with desired,
environmentally-influenced endpoint phenotypes – the genotype-to-phenotype (G2P) problem
(Cooper et al., 2002). These two issues converge when one considers that, on the scale of the
North American Prairie, changing crop physiology has the power to alter climate at the same
time that changing climate is altering the target population of environments for plant breeders.
Methods
We believe that the G2P problem is best solved using a two-pronged approach that seeks to
build from molecular levels toward physiology, while simultaneously exploiting physiological
information to augment research at the genetic level. Tardieu (2003) and Tardieu et al. (2005)
identified a class of models they termed genetic meta-mechanisms which, to us, constitute the
proper common meeting ground. These models capture the underlying behaviour of the
genetically-controlled processes without necessarily incorporating all of the detail.
Specifically, they said these models should meet two criteria:
1. Models should reproduce phenotypes in a wide range of environments;
2. Model constants that were asserted to be genotypic properties should associate with
defined genomic regions in a manner independent of the environments in which
measurements were made.
The first criterion corresponds to standard goodness-of-fit tests long applied to crop
models. The second, however, mandates that to be considered as ‘gene-based’, one should be
able to show, either by reference to specific genes or at the resolution of association mapping
studies, that (i) there are or are likely to be gene(s) that control the properties that a given
constant parameterizes and (ii) that the values that the constant takes in different lines do not
result from detectable G×E effects. Stated in the usual vernacular of crop modelling, ‘genetic
coefficients’ that take on different values in different environments constitute prima facie
evidence of some invalidity, which may be in the model itself or in the measurement of
environmental data used to drive it (Welch et al., 2002).
To meet the needs alluded to in the introduction, there is a third criterion which should be
explicitly added:
242
Session B3: Cropping systems modelling
3. Models should be able to reproduce the results of selection.
For plant breeders, this equates to models that predict the phenotypes of offspring from the
genetics of the parents. The model of Reymond et al. (2003) meets this test. In the broader
context, it means that one should be able to predict the evolution of phenotype from a
knowledge of the genotypes present (natural variation), the progression of environments, and
the relationship of phenotype to fitness, and the selection gradients. Classical quantitative
genetics has been seen to fail this test by overestimating the response to selection (Coors,
1999).
Results and discussion
We present two studies that move in the directions just described. In the first, a classical
photothermal model of flowering time was linked to the known underlying genetic network of
Arabidopsis thaliana (Wilczek et al., 2009) The model was parameterized with literature data
and phenotypic measurements from several mutant lines planted in common gardens from
Valencia, Spain, to Oulu Finland. The model was independently verified from a combination
of growth chamber studies and repeated-planting field data. The model reveals a narrow
widow in late summer when flowering times become exquisitely sensitive to germination date
and rapid cycling life histories transition into winter annuals. This has significant
consequences for patterns of selection. The model also has been used to characterize 206
natural accessions and suggests a possible genetic trade-off wherein slowed short day
development rates limit premature flowering at high latitudes in lines with lower chilling
requirements. Because of its exceptionally low computational requirements this model would
be directly usable within general climate models.
None of the three criteria above constrain the mathematical forms that might be used for
gene meta-mechanisms, so they might, in fact, be networks. The second, in silico study uses
genetic programming and particle swarm optimization methods to extract plausible network
structures from synthetic phenotype and gene expression data generated by a known network
(Cai et al., 2009). The best networks found had structural features that were reminiscent of the
true network. Although the actual network structure was not recovered, all three of the above
criteria were met, including the critical third one. This shows that partial or only partially
correct network structures can serve useful roles in crop simulation.
Between them, these two studies illustrate the dual G2P & P2G approach that we feel is the
way forward to truly gene-based plant simulation.
References
Cai, X., et al., 2009. Int. J. Bioinformatics Res. and Appl. (in press)
Cooper, M., et al., 2002. In Silico Biol. 2: 151-164.
Coors, J.G., 1999. In: Eds J.G. Coors & S. Pandey, The genetics and exploitation of heterosis
in crops. ASA-CSSA-SSSA, Madison, Wisconsin, USA, pp. 225-245.
Foley, J.A., et al., 1996. Global Biogeochemical Cycles 10: 603-628.
Reymond, M., et al., 2003. Plant Physiology 131: 664-675.
Tardieu, F., 2003. Trends in Plant Science 8: 9-14.
Tardieu, F., et al. 2005. Australian Journal of Agricultural Research 56: 937-946.
Ward, J.K., et al., 2000. Oecologia 123: 330-341.
Welch, S.M., et al., 2002. Trans. ASAE 45: 1163-1175.
Wilczek, A., et al., 2009. Science. (under review)
243
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme B
State-of-the-art of components for integrated
systems
Session B4: Bio-economic farm modelling
Session organizers:
Guillermo Flichman
IAMM, Montpellier, France
Pytrik Reidsma
Wageningen University, The Netherlands
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Some conceptual issues for modelling the interaction between agriculture
and environment through the bio-economic mathematical programming
models1
G. Flichman, K. Louhichi
CIHEAM-IAMM, 3191 Route de Mende, 34093 Montpellier, France
Contact: flichman@iamm.fr
In the last years, there has been a significant development of bio-economic models, especially
those integrating biophysical models (called also agronomic or cropping system models) and
economic mathematical programming models. This development was enhanced by the
conjunction of several factors such as the multiplicity of objectives in new agricultural
policies, the increase of demand for multi-disciplinary approaches for integrated assessment,
the call for more dialogue and cooperation between scientists from diverse disciplines. An
important number of bio-economic models was developed and tested on different farming
systems and under various agro-ecological conditions (Janssen & Van Ittersum, 2007). This
rich body of literature on empirical applications of bio-economic models was not followed by
a conceptual development related to this type of models. Indeed, there is not enough literature
regarding the interaction between this models and the economic theory, their main interest
compared to conventional economic approaches, their specifications and contributions in
strengthening collaboration and improving integration between different disciplines.
The aim of this paper is to develop some conceptual and theoretical issues related to bioeconomic models (principally the farm models) and to present the suitable way to use this
type of approach for modelling the relations between agriculture and environment and more
largely for the integrated assessment.
A bio-economic model is known generally as a linkage between models from different
disciplines to provide multi-disciplinary and multi-scales answers to a given problem. In
reality, the philosophy behind this approach is more complicated. A bio-economic model
should not be a simple link between models through an exchange of information but a real
integration in both conceptual and technical terms. This has twofold implications: first, we are
facing a new approach which should have a clear position in the economic and agronomic
theory corpus, and second, the construction of each model should take into account the
specificity and the conceptual basis of the other. We try to better understand the economic
theoretical issues behind the used mathematical programming models and also to present the
main specifications that these economic models should have to ensure a consistent integration
with the agronomic ones.
For modelling the relation between agriculture and the environment, economic theory has
summoned up several approaches: the application of the standard micro-economic analysis
(i.e. ‘Environmental Economics’), the integration of original methods and tools based on the
agent’s revealed preferences into the conventional theoretical corpus (i.e. the so-called
‘London School’), or the exploration of new methodologies and knowledge stemming from
other disciplines in particular from Natural Sciences (i.e., Constanza & Daly, 1987). From this
classification it appears at first sight that the bio-economic modelling method is a part of the
Ecological Economics approach. However, as the bio-economic models are often based on
optimization models it could also be possible to situate this method under the conventional
economic theory (i.e., ‘Environmental Economics’). The arguments behind each of both
theses will be developed in detail in the final version of this paper.
Regarding integration, these models should have a set of specifications ensuring a
consistent integration with the agronomic models. The first specification is the primal based
1
Keynote presentation
246
Session B4: Bio-economic farm modelling
approach: technology should be explicitly represented to deal simultaneously with biophysical
and economic aspects and to quantify all the inputs and outputs associated to agricultural
activities (i.e. production processes) in physical and economic terms. A dual based approach,
based only on costs cannot fulfil these purposes.
The second specification is that the models should be activity based, what means that one
product can be produced by several different activities1 (Koopmans, 1951), and each activity
produces several products. Each activity is defined by the technical coefficients that represent
the use of inputs needed to produce one unit of output. What is important to stress for our
purpose is the fact that all the basic information concerning production is related with
activities and not with products. This characteristic allows taking into account the positive and
the negative jointness (i.e., joint production, see Baumgärtner et al., 2001) associated to the
production process, and making a suitable integrated assessment of new policies which are
mainly linked to activities and not to products.
The last specification is that the inputs and the outputs (including externalities) should be
represented in discrete forms and expressed in transparent way (i.e., yield and cost functions
per product are expressed as discrete functions) in order to make easily the integration with
biophysical models and also to ensure that the impact of each input can be assed separately
with respect to the others. Indeed, the biophysical model provides a set of multi-inputs and
multi-outputs production functions, which are unsuitable to be properly represented through
continuous forms. Moreover, the complex problems related to non-linearity in the relation
between agricultural production and environment can not be expressed easily or at all
applying ‘well-behaved’ functional forms.
All these specifications are illustrated in the Farm System Simulator (FSSIM) developed
within the EU FP6 SEAMLESS project (Van Ittersum et al., 2008). FSSIM is a bio-economic
model simulating farm level behaviour given a set of biophysical, socio-economic and regionspecific policy constraints, and its responses to technological innovations and policy changes.
It is a primal based approach to account for selected positive and negative jointness in outputs
associated with the production process. FSSIM is based on discrete production functions and
a limited number of externalities functions which are expressed in physical terms (Louhichi et
al., 2007). These specifications enable FSSIM to directly explore the impacts of some policy
changes not only on the relationship between market and non-market goods, but also on the
production process itself.
References
Baumgärtner, S., et al., 2001. Ecological Economics 36: 365-372.
Costanza, R. & E. Daly, 1987. Ecological Modelling 38: 1-7.
Janssen, S. & M.K. Van Ittersum, 2007. Agricultural Systems 94: 622-636.
Koopmans, T., 1951. Activity Analysis of Production and Allocation. Proc. of a Conference,
Cowles Commission, New York. http://cowles.econ.yale.edu/P/cm/m13/index.htm
Louhichi, K., et al., 2007. A generic template for FSSIM for all farming systems,
SEAMLESS PD3.3.11, 82 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
1
This method, which precludes the separate measurement of alternative processes to produce the same
commodity, or the recognition of joint production, can be and is being supplemented by the study of engineering
information (Koopmans, 1951).
247
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated home garden model for food security,
nutrition and poverty alleviation
M. Akkas Ali, M. Robiul Alam, M.S.H. Molla, M.A. Momin
On-Farm Research Division, Agricultural Research Station, Panba 6600, Bangladesh
Contact: makkas62@yahoo.com
Introduction
In Bangladesh, more than 30,000 children are suffering from blindness each year and
majorities of its population are in lack of the required amount of vitamins (81%), protein
(60%) and minerals (Mahmud, 1985). There are about 17.5 millions of homesteads in the
country which can help producing sufficient vegetables and fruits for the concerned families.
Farmers practice different patterns of vegetables and fruits in the vicinity of the household,
but almost all are unplanned, poor yielded and non-scientific. Thus, it was felt that a complete
model is needed for homestead production. A model was applied to nine possible production
units to avoid the shortfalls stated above with the following objectives (i) Maximum
utilization of homestead spaces and time round the year with fruit and vegetables, (ii) Ensure
food security round the year and build up food for family consumption with nutritional
quality, (iii) Create employment opportunity for family members, cash generation and
develop women members for decision making and gender equity.
Materials and methods
The study was carried out at the Farming Systems Research and Development (FSRD) site,
Pabna during 2001–2005 with 15 female-participating farm families. Nine cropping patterns
were used for 9 production niches of each homestead (viz. open land, fence, trellis, non-fruit
tree, partial shady area, roofs of cottage, marshy land, home boundary and backyard) and the
selection of crop varieties were finalized with the active participation of the co-operators in
accordance with their need for assessment, preference and resources in decision-making
process. There was a flexibility of plot and/or space sizes of each production niche to avoid
complexity of the study. Recommended crop production technologies were used for the study.
To attain environmental benefit some technologies like composting with kitchen and house
waste, bio-pesticides, use of BARI cooking oven and use of bi-gas plant were also tested. The
collected data were checked, processed and analysed for interpretation.
Results and discussion
Production of vegetable and fruits The production of vegetables increased remarkably in the
integrated model (Table 1). Average production was 746 kg family–1 during 2001-05, which
was above 4 times compared to a previous (178 kg) model (Islam et al., 1996). Highest
production was obtained from the creeper group due to better management with improved
skill in production practices. The total production from newly included spaces was 2.38 times
higher than the open space. The fruit yield from existing trees was 810 kg family–1 y–1 where
main contribution was from mango. Mango yield was increased by about 3.37 times due to
better pest and agronomic management. The new fruit trees introduced in the model and
management of existing fruit trees are expected to increase production remarkably.
Food security and family nutrition Adequate amount of nutritious fruits and vegetables were
supplied round the year averaging 710 g d–1 person–1 (Table 1). The average consumption of
fruits was 920 g d–1 family–1, much higher than daily requirement and 3.3 times higher (280 g)
than non-project areas (Akhtar et al., 2000). The average production per day of both
vegetables and fruits were 2.04 and 2.22 kg which all together, was 2.84 times higher than the
need for family consumption and 5 times higher than the national average (396 g) family–1.
248
Session B4: Bio-economic farm modelling
As production of food items lead to its added consumption and also increase distribution
(22%) to relatives and sale (35%) to the buyers (Table 1). The percent consumption of
vegetables and fruits were 45 and 42% of production only. The supply of nutrients from fruits
and vegetables of the tested model surplus the need for most of the wanting essential nutrients
like Vitamin A, C, calcium and iron-previously deficient in the diet. An ample amount of
Vitamin B1, B2, protein and energy were also obtained from the supplied food of the model
(Table 2).
Table 1. Average yield of vegetable, fruits, gross return, disposal pattern and income of the
model. Currency: US$ 1 = Tk (Taka) 68.
Yield
Crops
family–1
(kg)
Vegetables
746
Mean family–1 d–1
2.04
Fruits
810
Mean family–1 d–1
2.22
Grand total (Veg. + Fruits)
1,556
G. Mean family–1 d–1
4.26 (710 g)
Total cost for the model (Tk.)
Benefit cost ratio
-
Return
family-1
(kg)
2,832
8
8,664
24
11,496
31.5
350
32.85
Disposal of the produce family–1 (kg)
Consumption Distribution
Sale
Cash income
(Tk.)
337 (45%)
105 (15%) 304 (41%)
1,143
0.92
0.29
0.83
3.13
334 (42%)
238 (29%) 238 (29%)
1,979
0.92
0.80
0.65
5.42
671 (43%)
343 (22%) 542 (35%)
3,122
1.84
0.94
1.48
8.55
-
Table 2. Average yield of vegetables, fruits and nutrient contents of integrated model.
Crops
Vegetables
Mean family–1 d–1
Fruits
Mean family–1 d–1
Total both resources
Daily needs family–1 *
% requirement supplied
*
Yield
(kg)
746
2.04
810
2.22
4,026
1.5
284
Content of nutrients
Carotene Vit.C Vit.B-1 Vit.B-2 Calcium
Iron
Energy Protein
(kcal)
(gm)
(μgm)
(μgm) (μgm) (μgm)
(μgm)
(μgm)
14,959 320,298 1306
655 1409,927 113,322 767,412 16,861
41
878
4
2
3,863
310
2,102
46
29,597 187,588 404.9
318
82,117 21,864 397,511 6,164
81
514
1.11
0.87
225
60
1,089
16.89
122
1,392
5.11
2.87
4,088
370
3,191
6.89
10
260
7.0
6.3
3,000
145
14,100
284
1,220
535
73
46
136
255
23
22
Estimated from data provided by Haque (1985) for six members per family.
Income generation, poverty reduction, empowerment of women and gender equity The
average gross return was obtained Tk. 11,496 per family from the model with a very little
cash investment (Tk. 350 y–1, 1 US$ = Taka 68). The average BCR on cash cost basis was
over 32.85. This cash is generating round the year enabling the poor co-operators to meet up
immediate family needs like purchase of edible oil, lighting fuel (kerosene), pulse, salt and
spices. Sufficient amount of nutrients supplying from the food of the model, which is helped
in crossing poverty level (23% energy, Table 2). The women members were actively
participated in the program and involved in majority of the gardening activities, earned 24%
of family income, and participated in different group activities, trainings and field days. Even
40% women alone made decision in different activities. This empowerment enabled women
in attaining gender equity and increase prestige in the family as well as in the society.
References
Akhtar, M.S. et al., 2000. Annual report 2000. OFRD, BARI, ARS, Pabna, pp. 7-18.
Islam, S., et al., 1996. In: 14th International Symposium on Sustainable Farming Systems,
Colombo, Sri Lanka, pp. 81-89.
Mahmud, S.A., 1985. Introductory: Cultivation and nutritional identity of fruits and
vegetables.1985. Dept. of Agricultural Extension, Khamarbari, Dhaka.
249
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Economic impact of voluntary agri-environmental measures:
The case study of a ‘Test-Action’ in a basin of south-western France1
M. Ben El Ghali, A. Ridier, C. Kephaliacos, G. Nguyen
Toulouse University, LEREPS, ENFA, France
Contact: elghali_ma@yahoo.fr
Introduction
The impact of pesticides and nitrogen use on water quality is one of the principal
environmental issues today. To handle this problem, the European Union has adopted, since
2006, compulsory legal measures adressed to farmer within the framework of the cross
compliance of the CAP 1st pillar subsidies (environmental regulations and Good
Environmental and Agricultural Practices - GAECS). But voluntary measures are also applied
for a long time inside the Rural Development Policy inside the CAP. Some of these measures
are local initiatives (Territorial Action Plans organized in small regions, Test-Actions initiated
in France by Water Basin Agencies of the Ministry of Ecology). The research presented here
concerns such a local program in a river basin in the Midi Pyrénées region (region of GersAmont). For this program concerning new farm management practices, we have observed that
the rate of contracting measures concerning the change of farming practices is low (compared
to measures subsidizing investment projects). The issue is to analyse and explain the reasons
for this low adoption rate of voluntary measures concerning rotation practices and soil
management.
Methodology
This research is based on mathematical programming methods, developed at the farm-level, in
order to analyse the impact of the new practices on the farm income and on labour
organization. Thus, labour is explicitly accounted for in the model because it seems to be an
essential component in the adoption decision of new agricultural practices.
After including agronomic data (yields defined as a combination of crop, soil type,
previous crop, agro-management type) as well as economic data (costs, premiums, prices,
resources requirement…), the mathematical programming model seeks to represent the
farmer’s observed behaviour and projects responses under new agricultural practices. The
main outputs generated by this model are the land use, the input use according to different
cropping techniques (soil type, previous crop, etc.), the farm income and the needs in labour.
Labour data is explicitly introduced according to the type and number of farm operations and
to the time needs for each operation.
The general structure of the economic farm model is formulated as follows:
Maximize: U = p ' y ' x − c' x + s ' x
Subject to: Ax ≤ B ; x ≥ 0
U represents the utility function of the farmer (here the expected gross margin), p is the vector
of producer prices, y is the vector of yields for each cropping activity, c is the vector of costs
per unit of cropping activity, s is the vector of subsidies per unit of cropping activity
(depending on the regulation of the CAP), x are the quantities produced for each crop, A is the
matrix of technical coefficients, B is the vector of available resource levels.
The model is calibrated for two typical crop farms corresponding to two main areas in the
river basin: the first farm type is specialized in maize and located in alluvial corridors, the
second farm type is located in hillsides and its cropping plan is specialized in ‘dry cereals’.
After validation of the model according to these typical farms, a set of two measures
corresponding to the program ‘Action-test’ is simulated: (1) the ‘long rotation’ consisting in
1
Research financed within the framework of the European project Life ‘Concert’eau’.
250
Session B4: Bio-economic farm modelling
setting up a higher number of crops; (2) the ‘minimum level of soil maintenance’ by using
techniques without ploughing. This set of measures is performed for the year 2005.
Results and discussion
The results of the simulated scenarios are compared to a reference scenario, which represents
the farms situations before setting up voluntary test-actions.
The technical results show that the long rotation requirement has been respected for farms
situated in the area predominated by irrigated maize (Figure 2). In this area, colza and
sunflower are introduced. However in the dryness cereals area (Figure 1) the changes are not
very important, farmers in this region already and traditionally cultivate many crops in order
to face the risk of market prices and yields variability.
The implementation of the measure of ‘minimum level of soil maintenance’ can explain
the changes in crop pattern: this technique reduces the time required by 6% for ‘dry cereals’
farm and by 7% for the maize farm. Thus, the farmer can optimize his time availability and
reduce the labour peak periods in order to practice new crops that are more profitable (Table
1).
Area2(pre dom inance of ir rigate d
maize )far m type2_120 ha
Are a1(Drynes s cere als ) Farm type1_150 ha
Before test action
After test action
Figure 1. Crop pattern for farm type 1
(150 ha) in cereal areas (area 1).
Fa
ll o
w
ol
za
C
a
flo
we
r
Su
n
So
y
ai
z
M
W
he
at
e
w
Fa
llo
C
ol
za
100
80
60
40
20
0
Su
n
flo
we
r
ze
M
ai
at
W
he
So
ya
Before test action
After test action
50
40
30
20
10
0
Figure 2. Crop pattern for farm type 2 (120 ha)
in irrigated maize area (area 2).
The economic results: the changes of economic results before and after test-action show that,
even if the crop pattern is unchanged, farmers’ net income on farm type 1 increases with
3.5%, mainly because of the premium allocated in compensation of the set of measures. In
farm type 2, the farm income increases with 8.3%, mainly because of the additional income
due to more profitable new crops and additional premiums received in response to the respect
of the long rotation and no ploughing technique (Table 1).
Table 1. Economic results of scenarios simulations.
SC1: Before test-action
Area 1: dryness cereals
Gross Margin in €
63,320
Time labour in hour
1330
Area2: irrigated maize
Gross Margin in €
71,560
Time labour in hour
1178
SC2: after test-action Variation %
65,530
1247
3.49
-6.24
77,482
1085
8.28
-7.89
Conclusion
The results show that, according to different agronomic contexts, the implementation of new
farming practices can be an opportunity for farmers to re-allocate their available time towards
more profitable crops or projects. This results also confirm that the quantity and the quality of
labour spent in the change of practices is probably the main obstacle to the adoption of
innovative farming practices.
251
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
An appraisal of risk on perennial energy crops in France
G. Bocquého, F. Jacquet, L. Bamière
INRA, Avenue Lucien Brétignières, 78850 Thiverval-Grignon, France
Contact: gbocqueho@grignon.inra.fr
Introduction
Ligno-cellulosic crops are said to be a credible agricultural biomass in the long term to
produce biofuels. They have already been widely spread in some of Northern Europe
countries, for instance Salix in Sweden or miscanthus in Great-Britain. As for France, lignocellulosic crops for energy have just been introduced into farming systems for three years, but
at a very quick rate. Ligno-cellulosic crops are mainly miscanthus, switchgrass and short
rotation coppices of several forest species. As new and perennial crops, sold on an emerging
energy market, they show a novel combination of features that make them original in
comparison with French traditional crops, for food as well for energy purposes. In particular
they can require an extensive investment cost for planting, first incomes being received a few
years later, while the land they are planted on is tied up. Such crops could then be qualified as
risky ones. Nevertheless French farmers keep on converting farmland to these crops in
increasing numbers, and despite weak incentives for the Common Agricultural Policy. Indeed
they can be grown on set aside lands, like any other non-food crop, while the 2003 energy
crops scheme applies to both annual and permanent energy crops. The aid was worth less than
€ 32 ha–1 in 2007 (to be compared with the € 89 ha–1 subsidy on cereals on average in France
in 2007). French government offers no additional aid.
We were interested in investigating the outward paradox between the raising number of
farmers willing to adopt perennial energy crops and the risky nature of these new crops, given
their lengthy establishment period and the uncertainty on market outlets for them. Policy
makers and other stakeholders also need to assess the economic benefits for farmers and the
extent to which they would be willing to convert their farmland to ligno-cellulosic production.
Methods
Firstly we built up two simple micro-economic models of farmers’ behaviour, basing us upon
a literature review of miscanthus and switchgrass characteristics. We simulated farmers’
decisions when they face the opportunity to invest in miscanthus, assuming they only had to
make one decision, which is to choose the number of hectares to convert from wheat to
miscanthus in order to maximize their total utility of profit. Optimal solution was constrained
by the total arable area and profits were subject to uncertain selling prices, uncertainty being
higher on wheat prices. The first model only took into account static variability of incomes,
whereas the second one also included their intertemporal variability (Hardaker et al., 1997;
Weil, 2002).
Secondly, we lead field interviews1 with French producers of miscanthus or switchgrass,
following a semi-directive questionnaire. A sample of eleven farmers with different structural
characteristics (age, production system) has been selected in the main producing regions
(Eure-et-Loir, Indre-et-Loire and Calvados) We were seeking to assess their perception of the
various types of risks they bear when they are cultivating perennial energy grass, and to
understand their motives for planting them.
1
This part of the work was financed by the European Union through the Network of Excellence Bioenergy, and
was done inside a partnership with IIIEE, Lünd University, Sweden.
252
Session B4: Bio-economic farm modelling
Results and discussion
According to our calculations, in 2008 in France area cultivated with perennial energy crops
increased by 30% in relation to 2007, to get to a total of 1565 ha in more than 300 farms
(AUP, 2008). Western and Northern France is especially concerned, with plantations mainly
dating back to 2007 or 2008. Three quarters of the total area is miscanthus, while 11% is
willow, 7% is switchgrass, and 8% is false acacia or poplar tree. However, permanent species
are still a very small share of the total of energy crops cultivated in France (0.25%). Fodder
dehydration plants buy directly to the farmers 33% of the miscanthus grown and 26% of the
short rotation coppices
The results of our simulations are given in Table 1. We found that diversifying a farmer’s
portfolio of activities with miscanthus is optimal if we take into account the variability of
incomes due to selling prices, but it is no more optimal if we consider also the expected cash
flows throughout time.
Table 1. Simulation results.
Gross annualized margin (€/ha/year)
Optimal areas of crops (ha)
Static risk
only
Static risk and
intertemporal
variability
Correlated prices
Independent prices
Correlated prices
Independent prices
Correlated prices
Independent prices
Correlated prices
Independent prices
Expected annualized gross income
(€ ha–1 yr–1)
Optimal areas of crops (ha)
Expected annualized gross income
(€ ha–1 yr–1)
Wheat
502.6
74.1
63.2
Miscanthus
409.9
25.9
36.8
478.6
468.5
99.6
0.4
98.9
1.1
502.3
501.6
Our field surveys showed that French farmers see perennial energy grass as globally less risky
than cereals, apart from a market outlet risk which can be strong because of an emerging
market. However this is usually shared with buyers thanks to contractual arrangements of
different kinds and the farmers seldom grow more than 10% of their arable land. The
following motives for investing in perennial energy grass have been set out: positioning
themselves on a new market, using marginal lands, saving time, and getting involved in an
environment-friendly activity.
Further research will deepen the analysis with the aim to conduct to an integrated
assessment of growing perennial energy crops. On the one hand we plan to include
environmental outputs (water, soil fertility) in the assessment by collaborating with
researchers from others disciplines. On the other hand, we will strengthen the analysis on
farmers’ behaviour, focusing on risk and intertemporal substitution attitudes (Epstein & Zin,
1989), and real options theory (Dixit & Pindyck, 1994).
References
Agence Unique de Paiement, 2008. Statistiques sur les cultures ligno-cellulosiques
bénéficiant des aides PAC. French Ministry for Agriculture and Fisheries, Paris.
Dixit, A.K. & R.S. Pindyck, 1994. Investment under uncertainty. Princeton University Press,
Princeton, 476 pp.
Epstein, L.G. & S.E. Zin, 1989. Econometrica 57: 937-969.
Hardaker, J.B., et al., 1997. Coping with risk in agriculture. CAB International, Wallingford,
274 pp.
Weil, P., 2002. In: Risques, Les cahiers de l’assurance. Vol. 49, SCEPRA, Paris, pp. 70-76.
253
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Economic analysis of cover crop (white mustard) adoption
to reduce take-all disease and N-leaching
G. Goulevant, S. De Cara, F. Jacquet, A. Reynaud, M.H. Jeuffroy, P. Lucas, F. Montfort
INRA-Economie Publique,
BP01-78850 Thiverval Grignon, France
Contact: fjacquet@grignon.inra.fr
Introduction
Winter wheat is a major crop in the world with more than 220 millions ha out of 1500
millions ha of arable crops. In the European Union, it represents one quarter of arable crops,
leading to frequent occurrence of wheat in main field crop rotations. However, due to its low
N uptake during winter, N-leaching risks can be high during this winter crop, in particular, in
conditions of warm climate in autumn and rainy conditions during winter.
There are several ways to reduce nitrate leaching in a winter crop rotation. Reduction of Nleaching can be achieved by straw incorporation of the preceding crop, leading to a decrease
in net mineralization between harvest and late autumn. Another solution is to use cover crop
(or catch crop) such as white mustard. The efficiency of cover crop or catch crop to reduce
nitrate leaching has been widely demonstrated (Meisinger et al., 1991), especially before
spring crops. The rational is that the use of an over-winter cover crop decreases nitrate
leaching (by retaining nitrogen via crop N uptake). As a result, autumn cover crop in order to
reduce nitrate leaching before spring crops is currently strongly encouraged by public
policies.
Despite important public subsidies, we do not observe a strong development of catch
crops. The main explanations are a low private benefit for farmers, as the sowing of a catch
crop is costly for farmers, and a supplement work to do just after harvest, in a period with
already a heavy work calendar. As a result, the economic private incentives for introducing
winter cover crop in wheat rotations are low. Adoption requires then high subsidies by public
authorities.
In this paper, we introduce another kind of benefit for the farmer of using cover crops
which has not yet been analysed in the literature. It is well known that producing the same
crop in the same field year after year is unsustainable because of the decline in yields. In the
specific case of wheat, the main problem is related to the development of the take-all disease.
The take-all disease is a problem related to soil-borne pathogens, Gaeumannomyces graminis
tritici (Ggt), that infect the roots of that crop, but that die out while the field is planted by a
different non-host crop. Agronomic studies have shown that some cover crops could have an
impact on the take-all disease. According to Ennaïfar et al. (2007), the take-all disease caused
by Ggt, and the incidence on the following winter wheat, can be decreased by a white mustard
when residues are ploughed in soil. Kirkegaard et al. (1994) have also shown that Brassica
crops may have positive effects on growth and yield of wheat by reducing Ggt inoculum
releasing.
In this paper, we argue that the benefit of the reduction in the take-all disease risk should
be taken into account by a farmer when choosing to adopt cover crops. This has strong policy
implications since in such a case it may reduce the amount of public subsidies required to
induce adoption. Evaluating these effects requires however a coupling of an agronomic crop
growth model describing the link between climate variability and wheat yield, with a model
simulating take all incidence and severity, and with an optimization-driven micro-economic
model describing the adoption by a risk-averse farmer of the cover crop.
254
Session B4: Bio-economic farm modelling
Methods
To model the process of cover-crop adoption by a farmer producing wheat taking into account
N-leaching and take-all disease, we coupled three types of models. First, the wheat growth
process and the N-leaching are simulated by STICS, which is a generic soil-crop model
simulating the crop growth from sowing to harvest at a daily time scale (Brisson et al., 1998).
From the characteristics of the climate, the soil, the grown variety and the crop management
applied, STICS calculates output variables relative to the production (quantity and quality of
harvested grains), to the environment (drainage and nitrate leaching) and to the evolution of
the soil characteristics under the influence of the crop. Second, the take-all disease model is
the one developed by Ennaïfar et al. (2007) allowing to simulate take-all incidence and
severity according to climate characteristics. Third, the economic model is a standard
optimization-driven micro-economic model describing the behaviour of a risk-averse farmer
producing wheat. This micro-economic model has been written in GAMS. To control for
local conditions, we have calibrated the three models for two very different regions: Le Rheu
in Brittany (48°07' N; 1°47' W) and Grignon close to Paris (48°50' N; 1°57' E).
Our approach has then been the following. First we have studied the behaviour of the
farmer without risk of take-all disease. The decision variables of the farmer included four
ways for managing the winter crop: a bare soil from harvest of preceding crop to sowing (the
wheat sowing being in early October), wheat volunteers as cover crop, a white mustard as
cover crop with an early destruction and white mustard with a late destruction. The microeconomic model has been simulated in various contexts including various levels for the taxes
on N-leaching or for cover crop adoption subsidies. In each of these contexts, we analysed the
optimal adoption decisions. A similar analysis has been conducted including the risk of takeall disease and a comparison of the farmer decisions without take-all risk has been realized.
Results and discussion
Preliminary results suggest that the cost for inducing cover crop adoption is significantly
lower when the farmer internalizes the take-all risk.
References
Brisson, N., et al., 1998. Agronomie 18: 311-346.
Ennaïfar, S., et al., 2007. European Journal of Plant Pathology 118: 127-143.
Kirkegaard, J.A., et al., 1994. Australian Journal of Agricultural Research 45: 529-545.
Macdonald, A.J., et al., 2005. Plant and Soil 273: 355-373.
Meisinger, J.J., et al., 1991. In: Ed. W.L. Hargrove, Cover crops for clean water. Soil and
Water Conservation Society, Jackson, Tennessee, USA, pp. 57-68.
255
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Farming Systems SIMulator: First generic bio-economic farm model
S. Janssen, K. Louhichi, A. Kanellopoulos, P. Zander, G. Flichman, H. Hengsdijk,
Hongtao Li, T. Heckelei, G. Stokstad, N. Borkowski, M. Hecker, A. Oude Lansink,
M. Blanco, E. Meuter, M.K. Van Ittersum
Plant Production Systems Group, Wageningen University,
P.O. Box 430, 6700 AK Wageningen, The Netherlands
Contact: sander.janssen@wur.nl
Introduction
Policymakers and farmers have an interest in making ex-ante assessments of the outcomes of
their choices in terms of policy and farm plan (cf. Zander & Kächele, 1999; EC, 2005). A
Bio-Economic Farm Model (BEFM) is defined as a model that links formulations describing
farmers’ resource management decisions to formulations that describe current and alternative
production possibilities in terms of required inputs to achieve certain outputs and associated
externalities. Janssen & Van Ittersum (2007) recently reviewed the usefulness of BEFMs for
predicting the impact of policy changes and identified a lack of re-use of BEFMs, e.g., models
were only used for one purpose and location for which they were initially developed and not
for other purposes or locations afterwards. Instead in cropping systems models, the re-use for
diverse purposes and locations is wide-spread. This lack of re-use for other purposes and
locations might hinder the usefulness of BEFM for policy assessment. In SEAMLESS
(System for Environmental and Agricultural Modelling: Linking European Science and
Society), we choose to develop the Farming Systems SIMulator (FSSIM) as a generic bioeconomic farm model. Here we define generic as being useful for a range of agrienvironmental zones, different farm types, different innovations or policy questions and
applications that require different level of detail in input or output data. The objective of this
contribution is to introduce FSSIM as a generic BEFM model, to present its structure and how
it can be applied for different purposes.
Generic features of FSSIM
FSSIM has several features that make it generic. First, it has not been built as a monolithic
model, but it is the result of a combination of several components. Second, the structure of
FSSIM can be adapted to the purpose for which it is being used and third, it has been coupled
to different user interfaces. Each of these features will now be discussed in some more detail.
FSSIM exists out of two main modules, FSSIM-Mathematical Programming (MP) and
FSSIM Agricultural Management (AM) (Figure 1). FSSIM-MP captures resources, socioeconomic and policy constraints and the farmer’s major objectives (Louhichi et al., 2007).
The aim of FSSIM-AM is to describe, generate and quantify production techniques of current
and alternative production enterprises which can be simulated by a cropping system model
such as the Agricultural Production and Externality Simulator (APES; Donatelli et al., 2009)
in terms of production and environmental effects. The fully quantified activities, i.e. the
complete sets of agricultural inputs and outputs, are assessed in FSSIM-MP on their
contribution to the farmer’s and policy goals considered. As FSSIM-AM and MP are quite
large entities, these have been further sub-divided into components or sub-modules that have
a more specific role and a stand-alone value (Figure 1). Components exist for collecting data
on current activities (either with detailed or aggregated information on agricultural
management), for specifying livestock activities or livestock related constraints, for
specifying policies and for calibration (Janssen et al., 2008).
Conceptually, FSSIM fulfils two main purposes. The first purpose is to provide supplyresponse functions for so-called NUTS2-regions (corresponding to provinces in many cases)
that can be upscaled to EU level, while the second purpose is to allow detailed regional
256
Session B4: Bio-economic farm modelling
impact assessment of agricultural and environmental policies and technological innovations
on farming practices and sustainability of the different farming systems. The dual purpose of
FSSIM resulted in applications that were more data intensive and applications that were less
data intensive.
F S S IM -A g ricu ltu r al
M a n a g em e n t (A M )
S urv e ys o n
d e taile d / c urre nt
a ctiv itie s
F S S IM -M a th em a tical P ro g ra m m in g
(M P)
Co m mo n
M o du le
(o bje ct ive ,
risk)
Liv esto ck F ee d
C o m po ne nt
Pe ren nial
Ac tiv ity
Co m po n ent
T e ch nica l
C oe ffic ien t
G en era to r
Ex p er ime nt
M o d ule=
M o d el
De cla ration
De ta iled /
S im ple C ur rent
Ac tiv itie s
C o m p on en t
P ro du ctio n
E nte rp rise
G en era to r
Cro p s
M o d ule
L ive sto c k
m o d ule
P ere nnia l
M o d ule
P ro du ction
T e ch niq ue
G en era to r
Re sults
Po lic y
m o d ule
PM PC alib ratio n
mo d ule
Figure1. FSSIM-MP and AM with their individual components or modules.
FSSIM has been coupled to two distinct graphical user interfaces, e.g., SEAMLESS-IF and
FSSIM-GUI. In the SEAMLESS-IF, FSSIM is integrated with other models and is run as part
of a model chain. The FSSIM-GUI (Meuter et al., 2009) allows modellers and integrative
modellers to make model runs with one or more components of FSSIM. The FSSIM-GUI
should, thus, help modellers and integrative modellers to evaluate components one by one and
work with FSSIM across data-sets.
Concluding remarks
FSSIM has been applied to 11 regions to assess supply responses for EU, to 4 regions for a
more comprehensive regional analysis, to both arable and livestock systems and to one region
(Mali) outside the EU. The distinction between two different purposes, the subdivision of
FSSIM in different modules and the coupling to the different user interfaces has proven to be
useful for achieving an appropriate configuration of the model with respect to data
availability, research question and location.
References
Donatelli, M., et al., 2009. In: Eds. F. Brouwer, M.K. Van Ittersum, et al., APES: The
Agricultural Production and Externalities Simulator. Springer. (forthcoming)
EC, 2005. Impact Assessment Guidelines. European Commission, Brussels, 99 pp.
Janssen, S. & M.K. Van Ittersum, 2007. Agricultural Systems 94: 622-636.
Janssen, S., et al., 2008. Integration of all FSSIM components within SEAMLESS-IF and a
stand alone Graphical User Interface for FSSIM. D3.3.12, 53 pp.
Louhichi, K., et al., 2007. A generic template for FSSIM for all farming systems. PD3.3.11,
SEAMLESS integrated project, EU 6th Framework program, contract no. 010036-2
www.SEAMLESS-IP.org pp. 82.
Meuter, E., et al., 2009. AgSAP Conference 2009, these proceedings.
Zander, P. & H. Kächele, 1999. Agricultural Systems 59: 311-325.
257
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Using time driven activity-based costing data for modelling complex
innovations: The case of batch farrowing
L. Lauwers, P. Everaert, J. Van Meensel, K. Bryon
ILVO – Social Sciences Unit,
Burg. Van Gansberghelaan 115,
9820 Merelbeke, Belgium
Contact: Ludwig.lauwers@ilvo.vlaanderen.be
Introduction
Time-driven activity-based costing, TD-ABC (Kaplan & Anderson, 2004) has been
successfully adopted in complex decision-making processes (Everaert et al., 2007).
Originated as a tool for the assessment of activity-specific labour use in a multi-activity
processing environment, the method is able to link activities to various input factors.
Moreover, recent work of Bryon et al. (2008), applying TD-ABC for evaluating conversion to
batch farrowing in farm farming, show that the method can be adapted for ex-ante evaluation
in a multi-criteria decision-making situation. TD-ABC is rather simple to apply, provided a
thorough system analysis and decomposition of activities into basic tasks is possible. As such,
the method can be considered as a well-grounded budgeting. The question arises whether this
mass of information can also be used, without much extra efforts, in optimization models.
This paper examines whether TD-ABC and linear programming, LP, can be combined in
one simple spreadsheet model, in order to take profit of both methods. The model uses data of
a former study on batch farrowing, which has been found as a valuable innovation from
corporate social responsibility (CSR) viewpoint (Bryon et al., 2008) and of an ongoing study
on the economic-ecological efficiency in pig production (Van Meensel & Lauwers, 2008).
The model is first used for a case farm (433 sows) and then generalized to a sample of 120
farrow-to-finishing farms.
Methods
The pig production process is decomposed in various main activities. In total, 20 activities are
analysed, of which the major labour-consuming are given in Table 1. TD-ABC models
highlight the time required for performing an activity, based on all possible variants of the
activity. The TD-ABC model considers drivers and time equations. The characteristics that
drive the variants of the activity are called time-drivers, because they ‘drive’ the time spent
for a particular case. Time equations model how time drivers drive the time spent for an
activity. In complex environments where the time needed to perform an activity is driven by
many drivers, TD-ABC can include multiple drivers for each activity.
Results and discussion
Table 1 gives an overview of estimated work load. In total 385 hours of labour is saved while
moving from the 1-week to the 4-week system, or from 13.92 to 13.03 hours per sow per
year. Figures are expressed per sow as a production unit, also including the piglet and
finishing pigs. The method is not only used for its original objective, but extended to CSR
criteria, like labour income, nitrogen emission and animal welfare (Table 2).
This information is brought in a spreadsheet based LP, constructed for evaluating choice
options in pig farms, in this case, to compare the batch farrowing systems on a 2 to 5 weekly
basis. Starting points are the different economic, ecological and sociological outcomes of the
time-drivers assessed with the adapted TD-ABC method. The LP model then chooses the
farrowing system, characterized by a binary variable, with the highest utility to the farmer.
258
Session B4: Bio-economic farm modelling
Table 1. Labour need for the different systems (in minutes/year).
Activity
1-week system
4-week system
Feeding and controlling pigs
109099
113706
Controlling sows
35259
35259
Feeding piglets at sow
32850
24635
Feeding sows
28991
28861
Farrowing
24960
15600
Moving sows
16744
14404
Difference
+4607
0
–8215
–130
–9360
–2340
Table 2. TD-ABC based estimation of change in CSR criteria for the two systems.
Activity
1-week system
4-week system
Difference
Labour income (Euro/sow.year)
21
36
+15
Nitrogen emission (kg N/sow.year)
137
138
+1
Animal welfare
Determined quantitatively
++
Feasibility of batch farrowing is finally simulated for a sample of 120 farrow-to-finishing
farms. With respect to the innovation, we conclude that batch farrowing is an eco-efficient
production process, which satisfies CSR objectives. The added value of the method is:
- comprehensible, farm-specific and transparent decomposition of production process;
- easy template to fill in specific farm features, usable for optimization modelling,
which gives the opportunity costs of a priori assigned decisions options.
- sufficient opportunities for a handsome sensitivity analysis
As such, the frame work allow for Monte Carlo simulations of uncertain parameters,
providing extra information for farmers reluctant for new farming systems with uncertainty.
References
Bryon, K., et al., 2008. Contribution of batch farrowing to more sustainable pig production.
Master thesis, Ghent University, Belgium.
Everaert, P., et al., 2008. Intern. J. of Phys. Distr. and Logistics Management 38(3): 172-191.
Kaplan, R.S. & S.R. Anderson, 2004. Harvard Business Review 82(11): 131.
Kaplan, R.S. & S.R. Anderson, 2007. Time-driven activity-based costing, a simpler and more
powerful path to higher profits. Harvard Business School Press, Massachusetts.
Van Meensel, J. & L. Lauwers, 2008. An Excel work sheet for economic-ecological
efficiency diagnosis and linear programming planning in pig production (work in
progress).
259
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Higher prices, less nature management?
Modelling farmers’ decisions on dairy farms
Hans Leneman, Raymond Schrijver
LEI, Wageningen UR,
P.O. Box 29703, 2502 LS The Hague, The Netherlands
Contact: hans.leneman@wur.nl
Introduction
The prices of grain, compound feed and milk in particular have risen strongly recently. This
paper comprises an analysis of the possible consequences of the changing market conditions
within the agricultural sector for the willingness to participate in agricultural nature
management. We focus on decisions by dairy farmers in the Netherlands to participate in
nature management schemes.
Methods
The basis for the analysis is the profile and the motives of the agricultural nature managers
and the way in which nature management can fit in with farm economics. We deducted the
profile and the motives of dairy farmers from literature. We combined this with results of
FIONA, a Linear Programming model at farm level (Groeneveld & Schrijver, 2006),
originating from the model presented by Berentsen (1999) (Figure 1). FIONA is especially
designed to fit in nature management in the operational management on farms.
Figure 1. FIONA (Farm based Optimization model for Nature and Agriculture).
Results of FIONA are presented using cost curves, which illustrate the relation between costs
and the optimum area of nature management. These costs curves are determined for an
average and a more intensified farm planning. Nature management consists of a meadow bird
protection scheme, with a good actual participation. The curves were estimated by entering
different area of the scheme (1, 3, 5… 29 ha). The compensation payment was not taken into
account.
260
Session B4: Bio-economic farm modelling
Results and discussion
Nature management on dairy farms stands for more than 60% of the total area under nature
management and this management is mainly aimed at meadow birds (Voskuilen & De
Koeijer, 2006). Dutch farmers’ participation has an economical motive; they are satisfied with
the compensation they receive. They also start their nature management activities because of
general interest in nature and landscape. Data show that both motives are equally important
(Leneman & Graveland, 2004). Another factor that explains farmers‘ participation is the
presence of environmental cooperatives. These cooperatives lower the administrative burden
for individual farmers, related to an application with the government.
Results of FIONA point at two effects of higher prices of milk and compound feed. First, it
will lead to a decline of the area of nature management. We estimated that participation will
decrease with some 20–40%, depending on the farm intensity, using a milk price of 45 € per
100 kg. If the compensation is increased by € 100 per ha, this decrease could be prevented.
The second effect is a higher threshold for new participants. The margins of profit on the first
hectares nature management will be lowered by a third.
Based on literature and on farm simulation results, we expect participation of dairy farmers
in nature management schemes to decrease, as prices for milk, grain and compound feed rise.
References
Berentsen, P.B.M., 1999. Economic-environmental modelling of Dutch dairy farms
incorporating technical and institutional change. PhD thesis, Wageningen Agricultural
University, Wageningen, The Netherlands.
Groeneveld, R.A. & R.A.M. Schrijver, 2006. FIONA1.0 – Technical description
Werkdocument 36 Wettelijke Onderzoekstaken Natuur & Milieu, Wageningen, 30 pp. (in
Dutch)
Leneman, H. & C. Graveland, 2004. Deelnamebereidheid en continuïteit van het Agrarisch
Natuurbeheer. Rapport 7.04.06. LEI, Wageningen UR, Den Haag, 2004. (in Dutch)
Voskuilen, M.J. & T.J. de Koeijer, 2006. Profiel deelnemers agrarisch natuurbeheer. WOtrapport 27. Wettelijke Onderzoekstaken Natuur & Milieu, Wageningen. (in Dutch)
261
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
EU dairy farming in the face of change:
An exploration using a bio-economic farm model
K. Louhichi, H. Hengsdijk, S. Janssen, G. Bigot, E. Perret, M.K. Van Ittersum
CIHEAM-IAMM; 3191 Route de Mende, 34093 Montpellier, France
Contact: Louhichi@iamm.fr
Introduction
Dairy farmers in the European Union (EU) face important challenges associated with changes
in policy and market conditions. First, the European Commission has proposed to abolish the
milk quota system in 2015 as part of the reform of the Common Agricultural Policy (CAP).
The milk quota system was introduced in 1984 because of the surplus of milk production in
the EU and the heavy burden on budget costs of intervention in the market for dairy products.
Anticipating the complete abolishment of the quota system in 2015, the EU has decided to
increase the milk quota with 2% per year in the beginning of 2008. Second, dairy farmers
have been confronted recently with a sharp increase of the price of feed concentrates due to a
number of reasons among others the increased demand for biofuels. Feed concentrates are one
of the major inputs in dairy farming that have contributed to the productivity increase of the
EU dairy sector during the last decades.
Both developments imply that EU dairy farmers need to cope with more volatile market
conditions with respect to both milk prices and prices of concentrates. An ex-ante assessment
of the likely consequences of the market instability supports the dairy sector and policy
makers to adapt to these changing conditions. The aim of this paper is to explore the effects of
changes in milk quota regime and in feed concentrate prices on farm income, resource use and
milk production using a bio-economic farm model. This model has been applied in two EU
regions with contrasting dairy farming systems, i.e. an intensive system (about 1.5 dairycow
per ha) in Flevoland (Netherlands) and a more extensive system (about 0.5 dairycow per ha)
in Auvergne (France).
Methods
The used method is based on the Farm System Simulator (FSSIM) developed within the EU
FP6 SEAMLESS project (Van Ittersum et al., 2008). FSSIM is a bio-economic farm model
which simulates the behaviour of farmers given a set of biophysical, socio-economic and
region-specific policy constraints, and their responses under technological innovations and
policy changes. It consist of a static comparative programming model, which maximizes a
non-linear utility function defined as a combination of expected income and risk, according to
the Mean-Standard deviation method (Louhichi et al., 2007). FSSIM integrates a large
number of crop and animal activities to facilitate the endogenously matching between feed
availability and feed requirements in mixed farming system. The model is calibrated on
average farm data applying a variant of the Positive Mathematical Programming (PMP)
approach and information from the Farm Accountancy Data Network (FADN). In this case
study, FSSIM is applied for two dairy farm types as identified in the EU-wide farm typology
developed within the SEAMLESS project (Andersen et al., 2007).
Two different policy scenarios are simulated and compared to a baseline scenario
incorporating policy changes formulated in the 2003 CAP reform. The first scenario,
simulates the reform of the milk quota regime based on an annual increase of milk quota by
2% during the period 2008–2013 and the second scenario, adds to the previous one a
progressive increase of feed concentrates prices from 0 to 100%.
262
Session B4: Bio-economic farm modelling
Results and discussion
Compared to the baseline scenario, the implementation of the milk quota reform leads to a
slight increase of both milk production and farm income in the two regions (Point of
intersection of the lines with the Y-axis in Figure 1): milk production and farm income
increase with 2% and 9%, respectively, in Flevoland and with 1% and 4%, respectively, in
Auvergne. The increase in milk production is associated with the intensification of milk
production in both regions, i.e. higher milk yields per cow. The economic effects in Flevoland
are associated with a strong increase of the farm gate nitrogen surplus mainly due to an
increase in the use of mineral fertilizers.
Results of the gradual increase of the concentrate prices show the strong dependency of the
intensive farm type in Flevoland on feed concentrates as farm income is much more affected
than the extensive farm type in Auvergne. The response of farmers to the gradual increase of
concentrate prices differs between both farm types, i.e. a strong a reduction in the number of
animals (i.e. selling of animals) in Flevoland and an extensification of production system (i.e.
lower supplement feed import) in Auvergne. The slightly higher farm income as a result of
the relaxation of milk quota is completely lost in both farm types if current concentrate prices
increase by more than 40%. The increase of concentrate prices has a positive impact on the
farm gate nitrogen surplus due to the lower import of nitrogen in concentrates.
Concentrate price
Concentrate price
30
10
15
0
0%
20%
40%
60%
80%
100%
0
-10
(%)
0%
-15
-30
-30
-40
-45
40%
60%
80%
100%
(%)
-20
20%
Farm type: Auvergne
Farm type: Flevoland
-60
-50
Farm income
Milk production
Nitrogen farm surplus
Livestock unit
Farm income
Milk production
Nitrogen farm surplus
Livestock Unit
Figure 1. Change of farm income, milk production, livestock units and nitrogen surplus due to
the increase of the milk quota and of the concentrate price (% change to baseline).
This case study shows that future policy and market changes may affect EU dairy farming
differently depending on the characteristics of the prevailing farming systems. The application
of this bio-economic modelling approach illustrates its potentials to integrate technical,
economic and environmental knowledge and to make the policy debate on the CAP reforms
more transparent thus contributing to well-informed decision-making.
References
Andersen, E., et al., 2007. J. Environ. Manage. 82: 353-362.
Louhichi, K., et al., 2007. A generic template for FSSIM for all farming systems,
SEAMLESS PD3.3.11, 82 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
263
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Dynamics of suckler cow farms under stochastic crop yields:
A recursive discrete stochastic programming approach
C. Mosnier, A. Reynaud, J. Agabriel, M. Lherm
INRA, UR1213 Herbivore, 63122 Clermont-Ferrand, France
Contact: cmosnier@clermont.inra.fr
Introduction
Farming is inherently a risky business. Seasonal weather conditions and market prices
account for important sources of variability, a variability enhanced by climate change. This
changing environment challenges French suckler cow farmers1 who continuously have to
adapt their production management. These farms indeed rely on forage production which is
very sensitive to weather conditions According to Boyer (2008), first recipients of the French
fund for agricultural calamities are herbivorous farms, mainly because of drought on forage
crops.
Different on-farm strategies can be used to manage weather risks. A previous work
undertaken by Mosnier et al. (2008) focused on simulating production adjustments to crop
yield shocks in those farms. However, this study did include neither farmers expectation
regarding beef prices and crop yield variability nor their risk aversion. Two kinds of risks can
be anticipated: embedded risks which occur when farmers plan to adjust their decisions
following the realization of some uncertain events (such as for instance the possibility to use
feed stocks or buy some feed when crop and pasture production are too low) and non
embedded ones when farmers do not consider the possibility of mitigating risks impacts once
the shocks have occurred. This study aims at better assessing how seasonal weather
conditions affect farm outcomes, on both short and mid terms, by taking into account both
risk anticipation and adjustment capacity.
Methods
Two well known modelling approaches can take into account embedded risks: Discrete
Stochastic Programming (DSP) and Stochastic Dynamic Programming (SDP). Previous bioeconomic livestock farm models using DSP approach are though limited by the number of
decisions stages introduced and by their short time span (one year in the case of Lambert,
1989; Kingswell et al., 1993; Jacquet & Pluvinage, 1997; Lien & Hardaker, 2001). Model
size increases with the number of decisions stages considered. Livestock farm models using
these SDP approach have to reduce the number of activities considered (Moxnes et al.,
Kobayashi et al., 2007): model size explodes with the number of dynamic variable. To
overcome the modelling limitations, we propose to use a sequence of recursive DSP model as
proposed by Blanco & Flichman (2001).
Our model is formulated to represent average French suckler cow farms which consist of
beef cattle production based on a suckler cow herd and of grain and forage crop production.
This production system must be managed monthly by a farmer in order to maximize their
expected utility of profit over a 5-year planning horizon, under beef price risks and two
annual weather-related embedded risks. We assume both inter-temporal preferences which
favour present and regular income and a constant relative risk aversion. The recursive process
is introduced so as that only decisions made during the current period are binding. Farmers
are supposed to revise partly their plan each month, if new information becomes available.
The bio-economic model is coupled with sub-models estimating pasture growth (Jouven et
al., 2006) and animal requirement according to their theoretical live weight (INRA, 2007).
Our model is free to adjust diet composition and diet energy content for each kind of animals
1
Suckler cows system consists of raising cows with their mother milks.
264
Session B4: Bio-economic farm modelling
(the herd is divided into 12 annual animal classes) as long as intake capacity constraints are
not exceeded and weight gains lie within an interval of plus or minus 10% of the theoretical
ones. Herd size and herd composition are optimized too, thanks to the monthly sales of
animals. Eventually, land allocation to forage and cereal crop production, haymaking, and
feed stock management (purchase and sale) are chosen endogenously. Structural constraints
such as land, labour and building, limit production possibilities. The model is parameterized
in order to represent average suckler cow farms located in the north of Massif Central (an area
famous for its Charolais beef production) that finish most animals, grow mainly pasture crops
and also, to a lesser extent, cereal crops.
Results and discussion
We simulate a sequence of 60 years: the 10 first years correspond to average weather
conditions (average yield over the period 1990–2006) and the subsequent 50 years to
stochastic ones. Each year, values for stochastic variables are randomly drawn from a sample
containing six states (defined upon observations over the period 1990–2006) that have the
same probability. We suppose that states of crop prices, cereal and forage crop yields are
linked. Farmers anticipate two states of weather condition and beef prices: a bad and a good
ones equal to average observation minus or plus one standard deviation.
Preliminary results reveal firstly that, at equilibrium, production under risk is at the same
level than without: the stocking rate is not lowered in order to constitute hay stock or to
decrease farm exposure to production risk. Farmers indeed plan to sell less concentrate and
buy some additional straw if some unfavourable weather conditions would happen.
Secondly, it appears that weather shocks of the previous year affect the production system
the following years. When two bad years occur in a row, stocking rate drops even more
during the second year. Taking into account explicitly the value of feed and animal stock
variation into the calculation of profit enables to consider most of weather effects the year of
the shock: when a good year occurs, feed stocks are built and during the bad one this stock is
depleted. Good correlations between this calculation of profit and shock intensity are found.
Eventually, very bad years damage more farm economic results than good years benefit them.
Consequently, over the last 50 years time span, average profit is much lesser that the average
one estimated for an average year.
References
Blanco, M. & G. Flichman, 2001. Working paper, Institut Agronomique Méditerranéen,
Montpellier.
Boyer, P., 2008. Notes et Etudes Econ.
INRA, 2007. Alimentation des bovins, ovins, caprins. QUAE, 307pp
Jacquet, F. & J. Pluvinage, 1997. Agric. Syst. 53: 387-407.
Jouven, M., et al., 2006. Grass and For. Sc. 61: 112-124.
Kingswell, R.S., et al., 1993. Agr. Econ. 8: 211-226.
Kobayashi, M., et al., 2007. Amer. J. Agr. Econ. 89(3): 205-817.
Lambert, D.K., 1989. West. J. of Agr. Econ. 14(1): 9.
Lien, G. & J.B. Hardaker, 2001. Eur. Rev. of Agr. Econ. 28(1): 17-36.
Mosnier, C., et al., 2008. In: 12th Congress of the European Association of Agricultural
Economics.
Moxnes, E., et al., 2001. Ecological Modelling 145(2-3): 225-241.
265
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A dynamic bio-economic model to simulate optimal adjustments of suckler
cow farm management to production and markets shocks
Claire Mosnier, Jacques Agabriel, Michel Lherm, Arnaud Reynaud
URH, INRA Clermont-Ferrand-Theix,
63122 Saint Genes Champanelle, France
Contact: claire.mosnier@clermont.inra.fr
Introduction
Tactical adjustments to seasonal weather conditions and beef price may generate additional
incomes or avoid losses in French beef cattle farms. Due to the length of the suckler cow
production cycle, adjustment decisions may impact not only on current production and profit
but also on future farm outcomes. To better understand the consequences of shocks and
subsequent production adjustments on the evolution of farm earnings and production over
time, we built an original dynamic recursive bio-economic farm model. Some bio-economic
models have already tested some tactical decisions on production but none has yet introduced
simultaneously the possibility of adjusting herd size and herd composition, diet composition
and diet energy content, as well as crop rotation, haymaking and feed stocks, taking into
account both their short and long term consequences. This is the objective of the model
developed here.
Methods
The model is formulated to represent average French suckler cow farms. Such farms consist
of beef cattle production based on a suckler cow herd and of grain and forage crop production.
This production system must be managed by a farmer over a finite horizon of T years. Each
year starts in April at the beginning of the grazing season and is divided into monthly
intervals.
The cattle herd is characterized by twelve annual animal classes which differ in age, sex
and fattening or storage objective. Classes are described by two dynamic variables: the
number of animals and their average live weight. These dynamics can be controlled by
monthly decisions related to animal sales, diet, reproduction and rate of fattening. Concerning
crop production, five feeds (grazed grass, hay, maize silage, grains and straw) can be
produced from four different crop productions (permanent and temporary pasture, maize, two
kinds of cereals). They are characterized by their qualities and their quantities. Feed quantities
are dynamic variables as their values depend on balance between previous stock and decisions
of produce consumed by the herd, produce purchased and sold. Grass production is calculated
on a monthly basis thanks to an herbage growth model developed by Jouven et al. (2006) and
filled with weather data from Nevers meteorological station (Météo France). Cattle
production is parameterized for the Charolais.
Our model assumes that farmers make decisions to maximize their utility of net profit over
a 5-year planning horizon. Introducing the utility function takes into account farmer intertemporal preferences which may favour current or regular income. Net profit is defined as the
difference between annual receipts (sales and Common Agricultural Policy - CAP- payments)
and costs (variable and fixed). Moreover, it is assumed that farmers adjust their decisions
when new information becomes available. This is modelled thanks to a recursive sequence of
multi-periodic optimizations. Each optimization is reinitialized by incorporating dynamic
variable values of the previous optimization and updated market, weather or CAP conditions.
Originalities of our model lie first in our specification of the production system. It indeed
introduced a higher degree of flexibility than other bio-economic livestock models found in
the literature. On the one hand, most dynamic bio-economic models allow animal diet
composition to vary but dynamics of animal live weights remain exogenously given. To date,
266
Session B4: Bio-economic farm modelling
these characteristics are fully endogenous only in Lambert (1989) and Kobayashi et al.
(2007). However, live weight management is not individualized for the different kinds of
animals whereas it might be of interest for farmers to favour for instance animals for sales. On
the other hand, some models explicitly incorporate feed stock dynamics (Barbier & Bergeron,
1999; Louhichi et al., 2004 etc.), but they concern dynamics of conserved forage. None of
them have explicitly integrated standing grass biomass stock dynamics. This confers on our
model possibilities of modulating forage use (grazing or haymaking) according to seasonal
conditions, given that current use will affect future availability of hay and standing biomass.
Secondly, like most of these models, decision variables are made to maximize a sum of
discounted utility of annual gross margin. However, time preferences and farmer anticipations
are explicitly taken into account. The recursive dynamic framework adopted gives us the
possibility of simulating impacts of non-expected shocks on farmer decisions over both the
short and the long term.
Results and discussion
Comparison of model outputs with a panel dataset of 25 farms over the period 2000-2006
demonstrates that it realistically predicts evolution of net profit, CAP subsidies, animal sales
and adjustments of concentrate feed in the diet. Adjustments of pasture cut, crop product
receipt, animal live weight and percentage of finished females point out some divergences due
in part to better control of biological process in the model, to the absence of feed security
stocks and to an unrealistically easy switch between several animal productions. The
application provided here helps to understand how the different sources of adjustments –
animal sales, crop product sales and purchases, animal diet, and haymaking – can be
combined to face temporary crop yield and beef price shocks, and above all how the optimal
mix is modified according to shock intensity. It can contribute as well to assessing the farm’s
capacity to handle shocks. However, ‘ability to manage shocks or hazards is a complex
function of existing behaviour that themselves represent long term or structural adaptation to
predictable shocks, crisis behaviour and by external responses (policy) to a predicted and
actual crisis. The next step will be to introduce risk anticipation in the decision sub model in
order to study jointly shock anticipation decisions and shock adjustment decisions.
References
Barbier, B. & G. Bergeron, 1999. Agric. Syst. 60: 1-16.
Jouven, M., et al., 2006. Grass and For. Science 61: 112-124.
Kobayashi, M., et al., 2007. Amer. J. Agr. Econ. 89(3): 205-217.
Lambert, D.K., 1989. West. J. of Agr. Econ. 14(1): 9.
Louhichi, K., et al., 2004. Anim. Res. 53: 363-382.
267
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Modelling Nitrogen balance for a regional scale livestock-pasture system as
a discussion support tool
U.D. Nidumolu1, M. Lubbers, V. Alary, P. Lecomte, H. Van Keulen
CIRAD, Département Elevage et Médecine Vétérinaire, Ile de la Réunion
Contact: uday.nidumolu@csiro.au
Introduction
Since the end of the 1980s, Réunion Island (located about 800 km east of Madagascar in the
Indian Ocean) has developed intensive livestock farming in order to increase its selfsufficiency in food and to preserve agricultural employment. Although these objectives have
been reached, considerable amounts of livestock effluent are produced. Because of the
shortage of land suitable for spreading manure and the mismatch between the types of the
manure produced and the needs of existing crops, livestock enterprises generate increasing
risks of pollution. These include emerging conflicts with other activities, such as tourism, due
to bad odour. Mastering the management of livestock wastes is, therefore, deemed necessary
by local authorities (Aubry et al., 2006). As argued by Flamant et al. (1999), the sustainability
of livestock farming should be assessed primarily in relation to local conditions, as
representations of visible or potential crises originating from conflicting interests of animal
husbandry and other local activities with respect to land use. According to Thornton &
Herrero (2001) and quoted by Aubry et al. (2006), the likely trends of smallholder croplivestock systems development within the next 20–30 years will require models to enable
analysis of these complex systems, assess their impacts, and help farmers improve their
performances. Currently, there is a major concern regarding agri-environmental issues.
Farmers are viewed not only as food suppliers but also as the custodians of the countryside.
This role of farmers has been officially acknowledged in the EU Common Agricultural Policy
(CAP) through a number of regulations that enforce agri-environment schemes and crosscompliance (Pacini et al., 2004). A detailed study on the Nitrogen issues in livestock is given
in a recent FAO publication (Steinfield et al., 2006). In keeping tune with this trend, in the
development of the regional model the objective was more to calculate N as an environmental
indicator which is introduced by two sources (i) by way of fertilizer for fodder requirements
and (ii) by the manure excreted by the dairy animals.
Method
The model was developed in a linear programming with multi-purpose method. It was
implemented under GAMS ‘General Algebraic modelling system’ with objective function
which will be optimized by taking account of the different constraints existing on the farm
and in the region where the model performed. These constraints include: land, labour, etc.
Ksheera is a mechanistic model, using a normative approach (Janssen & Van Ittersum, 2007).
It performs dynamically with a step of 6 months to calculate the objective function. The main
objectives of the model are to optimize: (i) income, (ii) nitrogen excess, and (iii) labour hours.
The classical one-dimensional approaches are less effective because of multi-parameters
which should be taken into account before making decision. As the model uses a normative
approach, the bench marking with the reality is difficult (Hazell & Norton, 1986). The
comparison is based on a reference year and the validation was attempted for the first six
months of this year. The error rate is less than 10%. Environmental policy is included in terms
of N management and the indicator chosen is N contributed from the dairy farms per ha. For
details of the model please refer to Nidumolu et al. (2008) and Nidumolu (2007).
1
Current affiliation: CSIRO Sustainable Ecosystems/SARDI, Bldg 11a, Hartley Grove St, Waite Campus,
Urrbrae 5064, South Australia, Australia.
268
Session B4: Bio-economic farm modelling
Results and discussion
Based on the calculation for N (by way of fertilizer and manure) the following outputs are
derived for the four sub-regions for Year 1 – Year 6 (for base scenario) and are given in the
Table 1 below.
Table 1. Nitrogen (kg per ha) calculated from N in fertilizer and N in manure.
Year
Cafre
Joseph
Ouest
Palm
1
320
236
141
347
2
336
240
149
364
3
330
234
207
361
4
341
242
215
375
5
339
240
214
374
6
343
242
215
376
The sustainability of dairy farms will depend increasingly not only on profitable milk
production but also on farmers' ability to comply with nutrient management regulations
(Powell, 2003). The model calculates the N per ha by both fertilizer and manure. The
innovation is the dynamism of the model (both land and animals) over a time-step of six
months. By a logical extension of the dynamism of land and animals, the fertilizer use (and N
use) and manure N excreted by animals is dynamic over the same time step. N in the context
of Réunion is a management option than a modelling problem. Therefore, the idea is to use
the results of the model in terms of N calculations to discuss N management options. The
driving forces for discussing these options are EU Nitrate Directive, fertilizer costs, manure
transportation issues, transformation options (compost for ease of transport, efficiency etc)
and manure as a source of energy. The management options are discussed based on the
empirical values calculated by the model.
References
Aubry, C., et al., 2006. Agricultural Systems 88: 294-315.
Flamant, J.C., et al., 1999. Livestock Production Science 61: 275-286.
Hazell, P. & R.D. Norton, 1986. Mathematical Programming for economic analysis in
agriculture. MacMillan, London, 400 pp.
Janssen, S. & M.K. Van Ittersum, 2007. Agricultural Systems 94: 622-636.
Nidumolu, U.B., 2007. Report on the EU Marie Curie ‘MODLAIT’ project at CIRAD EMVT
- Pole Elevage, Saint Pierre, La Réunion 2005-2007,120 pp.
Nidumolu, U.B., et al., 2008. In: Proceedings of Australian Agronomy Conference, 21-25
September 2008, Adelaide.
Pacini, C., et al., 2004. Agriculture, Ecosystems and Environment 102: 349-364.
Powell, M., 2003. U.S. Dairy Forage Research Ctr., Madison, WI.
Steinfield, H., et al., 2006. Livestock’s Long Shadow: Environmental Issues and Options.
FAO Publication, Rome. P 408.
Thornton, P.K. & M. Herrero, 2001. Agricultural Systems 70: 581-602.
269
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Linking cropping systems models with farm bio-economic models to
examine the relationship between profit and natural resource condition
Michael J. Robertson, Andrew Moore, Andrew Bathgate
CSIRO Sustainable Ecosystems, PMB 5, PO Wembley, WA 6913, Australia
Contact: Michael.Robertson@csiro.au
Introduction
Natural resource indicators are used by catchment management organizations (CMOs) in
Australia as targets for land use management. However, the nature of the trade-off function
between natural resource management (NRM) outcomes and whole-farm profit is ill-defined.
Defining the function will assist CMOs and farmers to evaluate the achievability of particular
targets, and help determine the size of economic incentives required to offset any expected
loss in farm profit associated with meeting targets. More generally, it will also allow for
quantification of the NRM impact of broadacre farming and the scope for more positive NRM
outcomes through changes in land use and management.
This paper brings together two modelling approaches to address the challenge of defining
NRM-profit trade-offs. Dynamic biophysically-based simulation models of agricultural
systems, such as APSIM (Keating et al., 2003) and GRAZPLAN (Donnelly et al., 2002), can
produce estimates of production, soil states and processes in relation to weather and
management and are commonly applied at point (i.e. field) scale. Whole-farm linear
programming models, such as MIDAS (Kingwell & Pannell, 1987), represent the biological,
physical, technical and managerial relationships of a mixed farm that is representative of
production systems within a defined region. The MIDAS model allocates available resources
in order to maximize the objective function of whole-farm profit, subject to resource,
environmental and managerial constraints.
In this paper, we use the biophysical models to derive values for NRM indicators (leakage,
bare ground) under a range of land uses and use these in MIDAS to explore the trade-offs
between profit and the indicators when subject to a range of constraints. The analysis was
conducted in two contrasting mixed farming regions in the cereal-livestock zone of Australia
to explore potential regional differences in the trade-offs.
Methods
Versions of the whole-farm economic model, MIDAS have been developed for each of the
southern mainland states of Australia. Versions representative of two contrasting regions were
used in this analysis. In one, from the Mediterranean systems of Western Australia, farming
systems are dominated by cereals, oilseeds, grain legumes and sheep production from annual
pastures. In the other, from the uniform rainfall zone in southern New South Wales, farming
systems have similar proportions of cropping and pastures, including the perennial pasture
species, lucerne.
APSIM and GRAZPLAN biophysical sub-models were linked into a soil-crop-pastureanimal model and simulations were conducted for all possible rotations of crops and pastures
on all soil types in each MIDAS model. The large set of simulations required was generated
automatically from a much smaller set of templates and executed on several hundred
computers that were managed with the Condor CPU cycle-harvesting software. Rotations
were simulated over 100 years of historical daily climate at both locations using realistic
management settings extracted from MIDAS (e.g., N fertilizer rates, stocking rates, flock
structure, grazing management). Simulated yields and pasture utilizations were checked for
consistency against those in MIDAS and were generally realistic. For each rotation x soil type
combination, long-term mean values of the NRM indicators were calculated for: water
leakage (mm yr–1) and nitrate leaching (kg N yr–1) below the root zone, ground cover (live
270
Session B4: Bio-economic farm modelling
and dead plant cover), and rate of change in soil organic carbon in the 0-10 cm layer. Ground
cover was expressed as the proportion of days where cover was less than certain thresholds.
Values of the NRM indicators for each rotation and soil type were entered into MIDAS. The
following sensitivity analyses were then run:
1. Holding the value of the indicator constant at a range of levels and optimizing for whole
farm profit. Plotting the value of optimum profit against each NRM indicator level enabled
the profit-NRM indicator response to be described.
2. Holding the percentage of the farm under cropping constant at a range of levels and
deriving the NRM indicator obtained when whole-farm profit was optimized. By plotting the
resulting NRM indicators against percentage crop, the influence of enterprise mix (crop vs.
livestock) on NRM indicators was examined.
Results and discussion
This analysis is the first of its type to combine biophysical data on NRM indicators within a
whole-farm economic model. It revealed a number of points:
• Win-wins are possible in the system where gains in the indicator (e.g., less leakage) are
matched by improvements in whole-farm profit.
• Trade-off situations are also apparent and in the case of leakage the ability to improve
the NRM indicator is dependent upon the availability and profitability of perennial
pastures, such as lucerne. In some cases substantial profits will have to be foregone to
reach NRM targets.
• NRM indicators vary in terms of their responsiveness to enterprise mix, with
groundcover less responsive than leakage.
• NRM indicators can respond negatively, positively or neutrally to changes in enterprise
mix, highlighting that NRM is a multi-dimensional concept.
• Methodological challenges include ensuring that simulated plant production from the
biophysical models is similar to that in the whole-farm model; the computational
challenge of simulating a large number of rotation × soil type combinations; and
maintaining the rank order of soil type ‘performance’ in both types of model.
These analyses will be useful in the quantification of the NRM impact of broadacre farming
and the scope for more positive NRM outcomes through changes in land use and
management. More complex analyses will be required to consider optimization where more
than one NRM indicator is being considered
References
Donnelly, J.R., et al., 2002. Agricultural Systems 74: 115-139.
Keating, B.A., et al., 2003. Eur. J. of Agronomy 18: 267-288.
Kingwell, R.S. & D.J. Pannell (Eds), 1987. MIDAS, a bioeconomic model of a dryland farm
system. Pudoc, Wageningen.
271
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Adoption of new forage technology: Impact on the socio-economic
sustainability of milk production in Moc Chau, Vietnam (DAIVIE model)
P. Salgado, M. Lubbers, R. Schipper, H. Van Keulen, V. Alary, P. Lecomte
CIRAD, UPR Systèmes d’élevage, Montpellier, 34398 France
Contact: paulo.salgado@cirad.fr
Introduction
In Vietnam, livestock make up about 30% of the total agricultural products and is a strategic
policy of the Vietnamese government due to its potential contribution to employment creation
and income generation. As many countries in Asia, Vietnam does not have a dairy tradition.
However, with the changes of food habits, the consumption of dairy products significantly
increased. The 100,000 dairy herds produced approximately 235,000 tons of milk in 2007
which satisfies 30% of the domestic consumption. Even if there are real opportunities for
dairy development in Vietnam, there are still major constraints including low management
skills of farmers, lack of suitable forage resources for cattle feeding and high cost of
concentrated feeds. New policies and agricultural innovations are being proposed in order to
solve the main issues related to the sustainability of the dairy sector. Agronomist researchers
provided alternative solutions using new forage technologies based on temperate forage
species. The first questions are: which are the good decisions to be implemented? How to
assess and evaluate the impacts of agricultural innovations? For this type of exercise,
modelling appears as an interesting decision tool because it helps to assess the complex
interactions found in the farm system as well as to analyse the configuration of alternative
technical innovations (Louhichi et al., 2004).
Methods
Participatory methods were employed to analyse the farming systems and establish
appropriate research and development activities. The first priority was to test and select
alternative forage plants, namely temperate species, which are resistant to cold temperatures
and able to produce enough quantity and good quality grass during winter. The forage
experiments were carried out under normal farming conditions and monitored by a team
composed of farmers, extension workers and researchers. Periodic surveys and meetings were
conducted to measure and observe species adaptation and evolution and, further, served as a
platform for discussion and evaluation process.
An exhaustive assessment survey with 120 representative dairy farms of Moc Chau district
was carried out in 2006 in order to establish a farm typology (Warter, 2006). Three main
criterions were selected: soil quality, herd dimension, and experience of farmers in dairy
activity. The adoption of temperate forage species by farmers started in 2004. In order to
assess and evaluate the impact of the new forage technology adoption, the modelling
approach was used. A descriptive farm-level linear programming (LP) model was created
under GAMS (General Algebraic Modelling System) using a bio-economic and integrative
approach. The stake of the model was to develop the essential interrelationships between
animal nutrient requirements and farm feed (nutrients) supplies related to profit maximization.
Two indicators (profit and labour time) were selected to evaluate the impact of the forage
Table 1. Socio-economic impact of oat adoption in average farm types.
Typology
No adoption
Oat adoption
Difference (%)
criterion
profit (€) labour
profit (€)
labour
profit
labour
Soil fertility
17,118
20
18,994
27
+11
+35
Experience
21,627
27
25,321
37
+17
+37
Herd dimension
14,859
19
17,009
25
+14
+32
272
Session B4: Bio-economic farm modelling
innovation on the socio-economic sustainability of dairy farms. Using this mathematical and
data-processing tool, it will be possible to identify and confirm the factors constraining the
adoption of the new forage technology and tackle the question of the socio-economic
sustainability of farms.
Results and discussion
The Avena species (oats) proved to be the best solution in terms of agro-ecological adaptation,
high production yields, excellent nutritive value and low production costs. In addition, the use
of oats allows regular production of milk throughout the lactation period. In the first year of
the scaling-up experiments (2004), 2.0 ha of temperate forage species were planted in Moc
Chau dairy basin by 36 farmers. Two years latter, the area for oat production was about 60 ha
and more than 140 farmers (28% of total farmers) were involved. However, the opinion of
few farmers related to the potentialities of temperate forage evidenced some dissatisfaction
judgments which possibly influenced further adoption. The reasons used by farmers to justify
their no-adoption includes: low growth capacity of oat in poor soils; insufficient number of
harvest periods; high input costs; labour-intensive. In order to realize the importance of these
adoption factors we create a multi-period farm-scale model (DAIVIE). The DAIVIE model
evaluate the impact of oat forage adoption, over the profit (total income – total costs), labour
time and more generally on sustainability of farms in Moc Chau dairy basin. The model takes
into account the interactions between the different components of the farm system (forage,
feeding, and livestock) and the market as well as between the past and present decisions in
which effects will have an impact in the future (Alary, 2000).
In all the tested scenarios the model selected the oat production as the optimal solution to
maximize farmer profits. This selection confirms the observations in the field concerning the
economic interest of oat use during the winter period. The results obtained for the total profit
(8 years) and labour time (days/month) in average farm types are presented in Table 1.
The model showed that the adoption of oat forage will lead to an increase in the farmer’s
income. This is directly related to an increase in the selling of milk and young animals, as
well as the reduction in the feeding costs. However, the amount of increase depended on the
typology and farm type used. The impact of oat forage adoption for the farmers is an increase
in labour time during the winter period. According to the model results, the experience of
farmers in dairy activity and the dimension of their herd did not seem to be crucial factors for
the adoption of the oat technology. On the other hand, the forage yields and the additional
labour needed are likely the main issues constraining the successful adoption of oat by dairy
farmers. These results are in total agreement with previous field evaluation process (Salgado,
2008). Almost all farmers declared that the technical protocol for oat production is rather
simple and do not pose difficulties even for farmers with less experience in dairy breeding.
For labour requirements, it is important to note that during the winter period farmers are
engaged in other activities in addition to milk production. Farmers’ decisions are governed
not only by productivity considerations but also by other factors, such as traditional practices
and cultural preferences. The introduction of labour-intensive technologies will result in
higher costs for hiring agricultural workers to perform additional tasks and in some cases
farmers are not interested to hire labour.
Finally, DAIVIE model conceptualization illustrates the importance of the interdisciplinary
work between modellers, agronomist and socio-economist researchers in order to obtain an
accurate mathematical approach and a fine representation of a dairy farm system.
References
Alary, V., 2000. L’harmattan, Paris, 236 pp.
Louhichi, K., et al., 2004. Animal Research 53: 1-19.
Salgado, P., 2008. Rapport scientifique, CIRAD, Montpellier, 95 pp.
Warter, J.L., 2006. Rapport de stage ISTOM, 57 pp.
273
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integration of biophysical and economic models for spatially explicit land
use and landscape analysis on farm level
Martin Schönhart, Thomas Schauppenlehner, Erwin Schmid
Department of Economics and Social Sciences, BOKU Vienna, 1180 Vienna, Austria
Contact: martin.schoenhart@boku.ac.at
Introduction
According to Keating & McCown (2001), two key components constitute agricultural
systems: the ‘biophysical production system’ and the ‘management system’. Integrated land
use modelling aims at the acknowledgement of both by increasingly combining biophysical
and economic models that are frequently connected via GIS-tools to make them spatially
explicit. Such model system is expected to deliver more accurate results while at the same
time permit interdisciplinary research questions to be solved (Antle et al., 2001).
Comprehensive ex-ante assessments of land use changes and induced effects to the natural
and social environment become feasible. Nevertheless, exact definition of interfaces between
data and models as well as consistency criteria for data and model integration are one of the
biggest challenges faced by modellers.
In this article we present such integration procedure, its challenges and trade-offs for a
modelling system containing the land use model FAMOS[space] and the biophysical process
model EPIC (Williams, 1995). Spatially-explicit data on field scales are available and build
the fundament for both models. This system is developed to analyse land use and landscape
changes by developing a set of alternative policy scenarios as well as to assess production and
environmental impacts of agricultural systems in the Austrian ‘Mostviertel Region’.
Methods and data
FAMOS [space] is an integer programming model at farm scale with spatial contexts. Land
use and production decisions at field-scale are modelled at the farm level by maximizing total
gross margins. Important decision variables encompass choices in crop and livestock
production, crop rotation, tillage systems, animal housing systems, input intensities, farming
systems (i.e. organic or conventional farming), and environmentally friendly management
measures (e.g., winter cover crops). A key feature of the model is the acknowledgement of
historical farm-specific production patterns in animal and crop production, e.g., crop
rotations. Crop yields and environmental impacts of agricultural systems are simulated with
EPIC for each site (e.g., field) and are integrated into FAMOS[space]. A site represents
homogeneous soil and topographical conditions. Furthermore, landscape structures are
assessed via landscape metrics.
The site data represents field properties which emerge through integration (e.g., zonal
statistics) of various data sources, e.g., digital soil map, land use map, and elevation map,
weather data, crop management data. Like in other land use models, soil is regarded as
‘integrated landscape element’ (Frede et al., 2002). Land use data are also describing the
geography of the fields and are taken from the INVEKOS-GIS data base of the Austrian
ministry of agriculture (BMLFUW).
Results and discussion
In a first model run, we applied the model system (EPIC and FAMOS[space]) to a typical
farm (arable land: 14 ha, grass land: 12 ha) in our project region ‘Mostviertel’. We used EPIC
to simulate crop and straw yields, soil sediment loads, and soil organic carbon (SOC) for
different sites and alternative productions systems and over a period of 10 years. Figure 1a
shows average SOC for the topsoil (0.3 m). As expected, for all CR removing of straw from
the field leads to lower average SOC contents on average. Although sites are quite similar on
this farm, variations in crop yields can be observed (compare to Figure 1b).
274
Session B4: Bio-economic farm modelling
Figures 1. Variation in average SOC for different crop rotations (CR) and straw management
systems (a) and variation in crops yields (t dry matter ha–1) for different arable crops (b).
The EPIC output is available in flat file structure and needs further processing to input in
FAMOS[space]. Therefore, a data interface was developed to transform EPIC output data to a
relational database. Table 1 presents various figures which are outputs of FAMOS[space].
Under current production conditions, the farm can generate 29,800 € of total gross margins
from animal and plant production as well as agricultural subsidies. Straw from grain
production can be sold or used in animal production systems. Without application of organic
fertilizers, this may cause SOC depletion over the long run. As counter measure, we
introduced an environmental payment for arable land subsidizing straw conservation.
Depending on the subsidy level, increasing farm incomes are accompanied by potentially
higher SOC-contents and shifts in the CR-system.
Table 1. Farm management strategies and average SOC for different SOC-subsidy levels.
SOC-Subsidy Farm income
Strawweighted
CR1
CR5
CR6
(€/ha)
(€)
Management
SOC (t/ha)
(ha)
(ha)
(ha)
0
29,800
straw removed
56.61
9.7
0.6
3.8
100
29,800
straw removed
56.61
9.7
0.6
3.8
150
30,500
straw conserved
58.74
10.4
3.8
200
31,300
straw conserved
58.74
10.4
3.8
This straightforward application of the model system (EPIC and FAMOS[space]) gives
insights into model features and challenges of model development. Foremost, the trade-off
between variation in model input data and model complexity has to be solved. The
combination of five soil classes, two management strategies, and seven CR already results in
70 EPIC runs. Still, more detailed variations in soil slope classes as well as management
strategies are necessary. Next steps in model development will be the specification of
environmental payment programs, the linkage of GIS and economic data and the integration
of further environmental indicators. In the final paper, the application of the model system
will cover the analysis on the landscape level in the ‘Mostviertel Region’.
References
Antle, J.M., et al., 2001. Ecosystems 4: 723-735.
Frede, H.-G., et al., 2002. Journal of Plant Nutrition and Soil Science 165: 460-467.
Keating, B.A. & R.L. McCown, 2001. Agricultural Systems 70: 555-579.
Williams, J.R., 1995. In: Ed. V.P. Singh, Computer Models of Watershed Hydrology. Water
Resources Publications, Colorado, pp. 909-1000.
275
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A bio-economic model to analyse the performance of the cotton sector
in Mali
B. Traoré, K. Louhichi, B. Rapidel
Institut d’Economie Rurale (IER), Mali
Contact: boubasiditraore@yahoo.fr
Introduction
Mali’s cotton production was doubled and nears record levels in the last decade attributed to
the increased planted area, as well as favourable weather and few pest problems. However,
this record was not followed by an improvement of cotton productivity (i.e. yield’s level)
which practically stagnated since several years (IER/CMDT/OHVN, 1998). This condition,
overlapped with the increase of world cotton price (Nubukpo & Keita, 2006) and the volatile
of the input price due to a number of reasons among others the rise of petrol prices, has leaded
in the last years to a drop of farm income in most of the cotton farms. To enhance farms
performance and support farmers and policymakers to design the future strategies to adopt
face this new context, an ex-ante assessment could be very useful. The objective of this paper
is to contribute on this issue by assessing the impact of this conjunction of factors on farm
income, land use and cotton production and to anticipate the effects of the likely management
strategies to adopt in this context. This impact assessment is performed through a generic bioeconomic farm model developed within the EU FP6 SEAMLESS project. This model is
applied in a set of farm types representative of the cotton farm in the Sikasso Malian region.
Methodology
The used method is based on the bio-economic farm model ‘FSSIM’ (Farm System
Simulator) developed within the EU FP6 SEAMLESS project to assess the economic and
environmental impact of agricultural and environmental policies and technological
innovations (Louhichi et al., 2007). The principal specifications of this farm model are: (i) a
static comparative model with a limited number of variants depending on the farm types and
conditions to be simulated; (ii) a risk programming model with a basic specification relating
to the Mean-Standard deviation method in which expected utility is defined under two
arguments: expected income and risk; and (iii) a positive model in the sense that its empirical
applications exploit the observed behavior of economic agents and where the main objective
is to reproduce the observed production situation as precisely as possible. For our case study,
FSSIM model was calibrated using average farm data collected through a farmer survey
carried out in a sample of cotton farms. FSSIM was applied on a set of farm types in order to
explore farmer’s response to change on the market conditions, particularly the increase of
input and output prices, and to simulate the impact of some agro-management strategies
developed recently to handle the problem of market instability and yield stagnation. In this
paper we show the results of only one policy scenario combining a 15% decrease of the
producer cotton price and a 23% increase of the nitrogen fertilizer price. This policy scenario
was compared to a ‘status quo’ scenario representing the current situation.
Results and discussion
The model was applied to the three identified farm types, however, for several reasons we
have decided to show in this paper the results of only one farm type called ‘large farm’. The
main characteristics of this farm type are an extensive agro-sylvo-pastoral system based on
cotton crop grown on biennial and triennial rotations and a farm size around the 12 ha. The
results of FSSIM calibration in the selected farm type was compared to the reference run. It
shows that the model reproduce approximately the real decision-making process of farmers,
for both the bio-technical management and the economic results. Indeed, the percent deviation
276
Session B4: Bio-economic farm modelling
between the observed and the simulated area of the principal crops such as cotton, sorghum,
millet and mani does not exceed 2%. The only difference was represented by the substitution
of groundnut by maize which is overestimated. However, it is necessary to recall that only the
current cropping techniques were taken into account in the calibration phase. Farmers and
researchers have approved the results of the calibration phase and have judged positively the
model quality.
After model calibration and validation, we started simulation by dropping 15% the cotton
producer price and increasing the nitrogen fertilizer price by 23%. The results of this
simulated scenario are shown in Table 1, in comparison to reference run.
As expected, the conjunction of the increase of nitrogen fertilizer price and the decrease of
cotton producer price have systematically involves the reduction of allocation area reserved
for the cotton crop and maze-bean (mani) association. The cumulated effect of these two
scenarios gives similar results to the effect of each scenario tested separately. Any time, the
reduction of cotton area is more significant in the combined effectThe income seems to be
less affected that is certainly explained by the increasing of allocation area reserved for maze
and mile which made it possible to deaden the negative effect. In all cases, the sorghum
allocation varied very little.
Table 1. Impact of the change on cotton and urea prices on land use and farm income in the
large farm type.
Farm
Crop allocation
Mani Cotton Maze Mile Sorghum income (€)
Reference run (%)
9.10 28.93 12.38 32.20
17.39
2313
Dropping of cotton price of 15%
–0.72 –2.82 3.30 1.20
–0.95
–4.8
Raising of N fertilizer price of 23%
–2.26 –0.30 –0.31 1.65
1.22
–1.5
Cumulated effect of two variables
–1.26 –3.50 4.16 1.92
–1.32
–3.01
In famer’s opinion, these results correspond to their rotation management strategy which
according to external pressures gives priority to the cereals which make it possible to ensure
family consumption. Farmers prefer to decrease the allocation reserved for the cotton crop
which is consuming much nitrate fertilizer.
Reference
IER/CMDT/OHVN, 1998. Mémoire de réunion de concertation sur la baisse de rendement de
la variété NTA 88-6 au cours de la campagne 97/98. Institut d’Economie Rurale, Tarla,
Mali.
Louhichi, K., et al., 2007. A Generic Mathematical Programming Model (FSSIM-MP) for
farming systems analysis. Farming System Design 2007, Catania, Italy.
Nubukpo, K. & M.S. Keita, 2006. Cahiers Agricultures 1(15), janvier-février 2006.
277
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Exploring production-theoretical insights for economic-environmental
trade-off analysis
J. Van Meensel, L. Lauwers, G. Van Huylenbroeck, S. Van Passel
ILVO – Social Sciences Unit
Burg. Van Gansberghelaan 115, 9820 Merelbeke, Belgium
Contact: jef.vanmeensel@ilvo.vlaanderen.be
Introduction
Decision making becomes more and more complex for farms. On the one hand, farms have to
keep competitive under strengthening competition. On the other hand, the internalization of
environmental effects forces farms to take into account environmental performance.
In order to provide better guidance for improving economic and/or environmental
performance, the trade-off between both performances has to be explored. Trade-offs can be
positive or negative. A positive trade-off implies that economic and environmental performance improve simultaneously. A negative trade-off implies that economic performance
improves, while environmental performance diminishes or vice versa.
The objective of this paper is to provide a framework for economic-environmental tradeoff analysis. The framework provides a reference basis for evaluating in a consistent manner
the effect of different types of measures influencing economic and/or environmental
performance.
Methods
Based on a literature review, various economic-environmental trade-off paradigms and
theories are combined and different trade-off measures are put forward. The trade-off between
economic and environmental performance is typically presented by the marginal abatement
cost (MAC) curve that links a firm’s emission level to the cost of additional units of emission
reduction (McKitrick, 1998). Conventional theory bases the MAC curve on the assumptions
that production and pollution abatement are separable and actual production is efficient. These
assumptions imply that a firm can control its emission by either investing in pollution control
equipment or by reducing output.
Numerous authors argue that conventional assumptions underlying abatement cost analysis
provide too little flexibility to describe accurately economic-environmental trade-off
possibilities. Hill et al. (1999), for example, distinguish three main stages in the process of
farm level transition to environmentally sound production practices: (1) efficiency improvement, (2) substitution of inputs or production processes and (3) redesign, that is, output
reduction or the use of new or additive technology for environmental purposes.
For our research, we distinguish between integrated and additive trade-off measures1.
While integrated measures address input-output transformation and the relation with negative
externalities and profit generation, additive measures aim at pollution abatement after
externalities have already been generated. Since additive measures are costly, they always
imply a negative trade-off, improving environmental performance and diminishing economic
performance. Some integrated measures, however, imply a positive trade-off.
In order to analyse besides additive also integrated trade-off measures, productiontheoretical foundations of input-output transformation and the link with economic and
environmental performance have to be elaborated. Production, profit and emission functions
that are theoretically consistent are constructed, taking into account the materials balance
principle for pollution. Numerous textbooks (e.g., Varian, 2003) describe the relationship
1
Rennings (2000) also distinguishes between integrated and additive measures but only focuses on measures that
improve environmental performance, without considering economic performance.
278
Session B4: Bio-economic farm modelling
between production, (marginal) costs and (marginal) revenues and economic performance.
We construct a similar linkage between production and environmental performance (Figure
1).
Marginal incoming
quantity in input
Pollution
Output
Yfarm
Marginal outgoing
quantity in output
Xfarm
Yfarm
Input
Output
Figure 1. Production and environmental performance of a farm.
Results and discussion
The relation between production, economic and environmental performance constitutes a
framework for analysing besides additive also integrated trade-off measures. Consider for
example efficiency improvement. It is shown that a technical efficiency improvement always
implies a positive economic-environmental trade-off. Input re-arrangement implies a positive
or negative trade-off, depending on the input-output technology of the farm, the input
combination the farm is currently using and the proportions of respectively input prices and
environmental coefficients of inputs. A similar reasoning applies to output re-arrangement.
Farms exhibiting economies of scale can increase their economic performance through
scaling up. This increase in output, however, results in more pollution (cf. Figure 1), so there
will be a negative economic-environmental trade-off. Vice versa, farms exhibiting
diseconomies of scale can increase their economic and environmental performance
simultaneously through scaling down. Consequently, a positive trade-off is established.
Certain measures, like improving output quality or buying the same inputs at lower prices,
improve economic performance, without affecting environmental performance. Contrarily,
using the same inputs, but with less harmful impact on the environment, results in a better
environmental performance without influencing economic performance.
References
Hill, S.B., et al., 1999. Agriculture, Ecosystems and Environment 73: 107-110.
McKitrick, R., 2002. Journal of Environmental Economics and Management 37: 306-314.
Rennings, K., 2000. Ecological Economics 32: 319-332.
Varian, H.R., 2003. Intermediate microeconomics: A modern approach. W.W. Norton &
Company, London, 688 pp.
279
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A whole farm simulation model to improve multi-criteria system efficiency
J. Vayssières, M. Vigne, V. Slegten, P. Lecomte
CIRAD Département ES, Station de Ligne Paradis,
7 Ch. de l’IRAT, 97410 La Réunion, France
jonathan.vayssieres@cirad.fr
Introduction
World population is expected to increase by about 40% in just over 40 years. An obvious
question is “How are we going to satisfy food requirements while preserving the environment”? The answer to this question can at least partly be found in the concept and practice of
ecological intensification. It seeks to increase the efficiency of resources use in agricultural
systems. Taking the example of dairy farming systems in Réunion tropical island we examine
how a whole farm model can be used to design more sustainable and efficient farms.
Methods
A wide variety of separate crop and livestock models exists, but integrated crop–livestock
simulation models are comparatively under-developed (Thornton & Herrero, 2001). GAMEDE
is a simulation model that represents dynamically the whole dairy farm functioning. It is
composed of two sub-systems: (i) a decision system that simulates management actions
performed by the farmer on a daily basis, and (ii) a biophysical system that simulates consequences of these actions on daily biomass flows and main ecological processes (crop growth,
livestock production, manure conditioning). The model calculates numerous technicoeconomic, social, and environmental indicators. Environmental indicators are based on (i)
‘farm gate balance’ methods: the apparent nitrogen surplus and efficiency (Nevens et al.,
2006), and (ii) ‘life cycle analysis’ methods: the energy balance and efficiency (Bochu, 2007).
The model’s detailed description and evaluation is proposed in a separate paper (Vayssières et
al., 2009).
Inspired by participative modelling approaches around multi-agent systems (Antona et al.,
2005), GAMEDE has been designed and used with six farmers chosen to represent dairy system
diversity in Réunion. The model is not used to assess the impact of policies but rather to
describe practices, to analyse differences in efficiency and to test ex-ante improvement
solutions, directly with farmers (Carberry, 2002). Value of computer-generated quantitative
information in farmers’ decision-making was assessed by qualitative interviews conducted by
an external observer (Kerdoncuff, 2007).
Results and discussion
One of the six farms is used here to illustrate GAMEDE’s relevancy to explore ‘hypothetical’
options. Farm 2 can be considered as an average farm in terms of technical and environmental
results. Starting from a project of the farmer, ‘the building of cow cubicles’, we considered a
serial of hypothetical scenarios. The three most relevant are presented here (Table 1).
- Scenario 1. With cubicles, cows’ pastures would be converted into silage grasslands because
animals could stay in the barn all the day. The surplus and more concentrated organic
fertilizer available would allow the farmer to stop mineral fertilizers without reducing land
productivity and with still an overall silage surplus.
- Scenario 2. This surplus of silage would be used to replace a part of imported forages.
Moreover we realized, from the model dynamic representations, that some pastures were
under-exploited with the actual rationing practices. A reduction of concentrate quantities
distributed to heifers and dry cows would allow a better valorization of these pastures without
affecting weight of replacement animals and then milk production of descendent cows.
- Scenario 3. Reduction of calving interval is also an alternative to improve farm efficiency.
280
Session B4: Bio-economic farm modelling
Table 1. Effects of different practice changes on sustainability indicators of farm 2.
Scenario
0. actual practices
1. better
valorization of
on-farm produced
organic fertilizers
2. better valorization of on-farm
produced forages
3. improvement
of herd reproduction performances
by 5%
Total effect
Nitrogen
efficiency
(Dmnl)
Energy
efficiency
(Dmnl)
Work
time
(h/ w)
Gross
Margin
(€/ yr)
13.8
Concentrate
efficiency
(milk L/ kg
FM of conc.)
1.16
0.26
0.35
85.1
61 300
+ 10
– 8.5
0
+ 24
0
+8
0
+1
+ 14
+8
+9
+6
–2
+ 13
–2
+7
+1
+7
+3
0
+6
+9
+ 12.5
+9
+ 40
+9
+6
+ 19
Land
eff.
(UFL/ha
/ yr)
4 600
Work
efficiency
(€/ h)
Values of Table 1 are calculated with GAMEDE and are means for the 2004-2006 period.
Except for the first line (scenario 0.), which is in absolute value, all values are percentages of
variation of farm results with reference to values of scenario 0.
As described for farm 2, interactive simulations have shown that farmers can improve the
efficiency of their systems just by changing their day to day management practices. Contrary
to preconceived ideas, environmental impact of dairy farming systems can be reduced while
maintaining or improving farmers’ revenue. In fact, the principal limitation to the adoption of
these “environmentally friendly practices” is not economical but social: the work time surplus
that is required.
Three of the six farmers involved in the project have adopted more efficient practices
found during interactive simulations. The others have stopped dairy activity, mainly for
economical reasons, while no solution was found to improve the efficiency of their farm in
the respect to their workload capacity. In both cases, farmers have attached importance in
their decision making to representations produced during GAMEDE design and simulations.
The interactive simulations were certainly limited to six farms but diffusion of participative
methods and systemic knowledge has been quick and wide. Today, one year after the
participative modelling experience, extension services plan farmers meetings similar to the
ones organized around the model and they now consider the family objectives and labour
ability as important aspects to be taken account in reasoning advice to farmers. Moreover the
dairy sector now defends farming system diversity, whereas till two years ago it had a Dairy
cooperative prescribed unique model, because indicators regarded by advisers are not only the
production but also the global efficiency of the system.
Using the type of tool proposed in this study directly with producers could be an
alternative to environmental policy directives (often badly accepted by farmers) to identify
and promote environmentally friendly practices adapted to each farm specificity. This
approach is especially relevant in developing countries, where agricultural policy and control
are less operational than in Europe and where many smallholder farming systems rely on
crop-livestock interactions.
References
Antona, M., et al., 2005. Nature Sciences et Sociétés 13: 165-168.
Bochu, J.L., 2007. Rapport scientifique. ADEME, 248 p.
Carberry, P.S., et al., 2002. Agricultural Systems 74: 141-177.
Kerdoncuff, M., 2007. SUPAGRO. Montpellier, 60 pp.
Nevens, F., et al., 2006. Agricultural Systems 88: 142-155.
Thornton, P. & M. Herrero, 2001. Agricultural Systems 70: 581-602.
Vayssières, J., et al., 2009. Agricultural Systems. (submitted)
281
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme B
State-of-the-art of components for integrated
systems
Session B5: Integrated assessment methods for regional
and landscape levels
Session organizers:
John Antle
Montana State University, USA
Daniel Auclair
INRA, Montpellier, France
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated assessment modelling: How to assess regional impacts?1
J.J. Stoorvogel
Land Dynamics Group, Wageningen University,
P.O. Box 47, 6700 AA Wageningen, The Netherlands
Contact: jetse.stoorvogel@wur.nl
Introduction
The Kenyan government identified in its strategy to revitalize Kenyan Agriculture numerous
possible interventions to combat soil fertility decline (Government of Kenya, 2004).
However, the document does not include an analysis of the potential impact of these
interventions. The 2003 reform of the European Common Agricultural Policy (CAP)
fundamentally changes the support of agriculture in Europe (Osterburg et al., 2007). Although
the changes aim to support landscape functions, the impact is still unknown. In the
Ecuadorian Andes, various NGO’s and farmer organizations have been concerned with the
intensive use of pesticides in the potato production system (Crissman et al., 1998). Their
concern was the start of a political debate towards more sustainable management practices
and an improvement of environmental quality and human health. A debate that started without
information on the actual impacts of pesticide use and the consequences of alternative
management practices.
The three cases above differ significantly from each other but, at the same time, they all
require some type of integrated assessment to properly explore the potential impacts of
interventions and to take informed decisions. Environmental problems cross the borders of
academic disciplines and need an integrated approach in which knowledge from various
disciplines is integrated.
Tools for integrated assessment
The broad range of tools that are currently available for integrated assessment (e.g., Bouma et
al., 2007) is the direct consequence of the plethora of questions that are being asked. Where
some of the questions have a more explorative character and aim at the identification of the
window of opportunities, others focus on the prediction of changes after technological
changes or policy interventions. But at the same time unique agro-ecological conditions may
require specific approaches. It simply matters whether we deal with, for example, a rather
uniform commercial potato-pasture rotation in the Ecuadorian Andes or with complex
subsistence, mixed cropping systems in Machakos district, Kenya. Different research groups
are working on these problems and developed different approaches to the problem. Where
some focus on multi-agent techniques, others focus more on econometric approaches, fuzzy
logic and optimization models. It seems logical that due to the variety of questions,
conditions, and approaches, we have a large number of tools available. The question that
remains unanswered is why the actual use of these tools for integrated assessment is still so
limited?
Case studies
In this paper, we look at three very distinct case studies: (1) the subsistence, small scale
production system in Machakos district (Kenya), (2) the National Landscape ArkemheenEemland in The Netherlands with its commercial agriculture in combination with its unique
environmental characteristics, and (3) the commercial potato-pasture system in the Carchi
area in the Ecuadorian Andes (Figure 1). The three cases deal with different regional issues
like poverty, the conservation of an historical peat landscape, and environmental quality. At
1
Keynote presentation
284
Session B5: Integrated assessment methods for regional and landscape levels
Machakos
Arkemheen-Eemland
Carchi
Figure 1. Case study areas (source: Google Earth).
the same time the three cases differ in terms of agro-ecological conditions, inherent variation
of the systems, and their dynamics.
As a result it is logical that the three case studies require different approaches to integrated
assessment modelling. The specific conditions in the three case studies are linked to a review
of different modeling approaches to illustrate how case specific the different approaches are.
Discussion
Despite the wide variation of available modelling approaches, the actual application is still
rather limited. Data availability and model complexity are certainly two of the most important
causes behind this limited use. As a result we can only conclude that there is a call for models
with reduced complexity and reduced data requirements. This can be accomplished by
modelling aggregate effects rather than modelling all the inherent variation in cases where the
variation is relatively limited or aggregate results are required. In other cases, one can wonder
whether it is necessary to explicitly model all the processes of the system. If the system is
driven by a few key processes one can search for a reduced form that focuses on those
processes. Finally, one has to evaluate seriously whether statistical or more mechanistic
approaches are required.
Integrated assessment modelling is the result of a serious call from society to evaluate
environmental systems in an integrated manner. Although the modelling tools provide a solid
basis, further development is necessary to make them applicable in practice and answer the
policy questions.
References
Bouma, J., et al., 2007. Adv. Agron. 93: 257-311.
Crissman, C.C., et al., 1998. Economic, environmental, and health tradeoffs in agriculture.
Kluwer Academic Publishers, Dordrecht, The Netherlands.
Government of Kenya, 2004. Strategy for Revitalizing Agriculture 2004-2014. Nairobi,
Kenya.
Osterburg, B., et al., 2007. Landscape issues of the EU common agricultural policy reform:
Implications for modelling. In: Eds. Ülo Mander et al., Multifunctional land use: Meeting
future demands for landscape goods and services. Springer, Berlin, pp. 269-279.
285
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated Assessment of crop and water allocation under conditions of
uncertain water supply in the Khorezm Region of Uzbekistan, Central Asia
I. Bobojonov, E. Berg, C. Martius, J.P.A. Lamers, J. Franz
Centre for Development Research, University of Bonn,
Walter-Flex-Str. 3, 53113 Bonn, Germany
Contact: ihtiyor@uni-bonn.de
Introduction
Low water productivity, poor irrigation system efficiency, and low field application
efficiencies have resulted in shallow, saline groundwater tables and secondary soil
salinization endangering land productivity. In the Khorezm region of Uzbekistan, this is
exacerbated by increasing water variability in the Amu Darya River, the main supplier of
irrigation water. This reduces the probability of sufficient water supply in particular to the
tail-end water users, which, in turn, increases income risks of farmers (Bucknall et al., 2003).
Farmers in this region are not free in their decision-making on land and water allocation,
and more than 70% of the area is allocated to cotton and winter wheat due to the state
procurement system which leaves little room for crop diversification. This study analyses the
potential economic and ecological gains of different policy recommendations affecting water
availability and use. Mathematical programming models have been used in a systems
perspective where socio-economic and environmental aspects of land and water use are
simultaneously considered.
Methods
Primary and secondary data sources were used, such as literature, statistical data and
information collected in the Khorezm region by the ZEF/UNESCO project within the Center
for Development Research, University of Bonn, as well as via farm, household, and market
surveys to obtain information on crops, cultivated area, production factors, and prices.
The Expected Value-Variance (EV) approach, combined with chance constrained
programming (Charnes & Cooper, 1959; Berg, 2003), was selected for its suitability in taking
into account multi-dimensional sources of risk associated with irrigated agriculture. Risk
associated with price fluctuations and yield variability is considered in the objective function
of the model, while risk associated with unreliable water supply is accounted for in the
constraints part of the model. Optimization of water and land allocation is carried out for 300
fields, belonging to 99 farmers in one Water Users Association (WUA) in Khorezm. The
spatial data (e.g., water distribution canals, distances) and agro-ecological properties (e.g., soil
fertility, soil texture) of these fields were imported from GIS data into the mathematical
programming model.
Optimal spatial cropping pattern and water distribution is found by allocating crops
according to the agro-ecological comparative advantages (e.g., soil fertility, distance from the
water source) and risk associated with each type of crop growing activity. Several farm and
WUA level constrains are taken into account in the optimization process. Cotton, winter
wheat, rice, maize (for grain), fodder crops, potatoes, vegetables and melons are the main
crops in the model. Each of these crops could be cultivated with different levels of water use.
The model was calibrated using different Constant Relative Risk Aversion (CRRA) levels.
Following the model calibration with the observed situation in the WUA, the findings of the
base run were compared to those of various scenarios. The analyses and comparisons
improved the understanding of the influence of different policy changes on the ecological and
economic situation. The scenarios included the introduction of a direct water price, waterwise technologies (such as drip irrigation), a change in the state procurement policy, different
water availability levels in the region, and various combinations of these factors. The effects
286
Session B5: Integrated assessment methods for regional and landscape levels
of changes in these variables on economic, social-welfare and ecological indicators were
assessed.
Results and discussion
Under the current system (status quo), the income of agricultural producers is relatively low
due to the cropping of winter wheat and cotton enforced on up to 70% of total arable area, and
owing to the low, state-fixed prices of these commodities at concurrently rising input prices.
The model simulations demonstrated the scope of direct water pricing for increasing water
use efficiency (WUE). However, the introduction of water pricing sharply decreased the
expected income on farm as well as on WUA level under the existing state order regime.
Unless the existing state procurement system is changed, this bears the danger of bringing
more poverty to the region and increasing the income risk of agricultural producers. Water
pricing seems nevertheless to be a promising solution for WUE increase under a liberalized
market economy. Expected incomes of farmers were still several times higher when water
pricing was introduced under a liberalized market economy condition.
Simulation findings also indicated that farms located at the tail end of the irrigation system
have to cope with the highest income losses during water scarce years, which could be
buffered by measures such as insurance policies for the ‘drought’ years. The introduction of
water-wise innovations will not automatically bring about widespread environmental
improvements owing to the high initial investments for such technologies. Under the present
regime, it would be economical beneficial to employ technologies such as drip irrigation and
laser-guided land levelling on a very limited area because farmers have very limited farm
capital for such investments. Thus, this model successfully allows exploring different options
for technology and policy improvement under a farm risk perspective, while concurrently
addressing economic and ecological effects.
References
Berg, E., 2003. Modeling the impacts of uncertainty and attitudes towards risk on production
decisions in arable farming. Poster Paper submitted for the 25th International Conference of
IAAE, Durban, South Africa.
Bucknall, L., et al., 2003. Irrigation in Central Asia. Social, Economic and Environment
Considerations. World Bank.
Charnes, A.W. & W. Cooper, 1959. Management Science 6(1): 73-79.
287
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
An agent-based model for assessing the impacts of decoupled agricultural
support on landscape values
Mark Brady, Konrad Kellermann
SLI, Box 730,
22007 Lund, Sweden
Contact: mark.brady@sli.lu.se
Introduction
In this paper, we present a spatial, dynamic agent-based simulation model of an agricultural
landscape for assessing the impact of agricultural policy reform on landscape mosaic and
biodiversity value. To make the empirical assessment feasible it was carried out using a
regional case-study approach to capture some of the diversity of agricultural and socioeconomic conditions in the EU-25, these being; high-cost or marginal regions (Jönköping and
Västerbotten Counties in Sweden), EU-10 (Vysočina in the Czech Republic) and high-value
Mediterranean agriculture (Marche and Calabria in Italy). Three policy scenarios were
evaluated and simulations extended from 2001 to 2013 (the final year of the current
programme period). The reference scenario represents continuation of the Agenda 2000
framework. The principle ingredients of this ‘coupled’ policy are payments based on crop
area and livestock numbers. The second scenario or Reform reflects the decoupled policies
implemented in each country in 2005. These payments are conditioned on land being
maintained in “Good Environmental and Agricultural Condition” (GAEC). The third scenario
is a hypothetical pure income transfer or Bond in which farmers have no obligations in return
for support. It should illustrate the effects of eliminating or reducing decoupled payments,
which will be natural policy considerations in the scheduled reviews of the CAP in 2008 and
2013.
Methods
A fundamental insight from the literature is that a standard static or marginal economic
analysis is likely to provide misleading results because decoupling has the capacity to affect
farmers’ strategic decisions – whether to enter or exit farming, make new investments,
renegotiate land contracts, etc. As such the analysis requires a dynamic framework where
changes in the farmers’ opportunity set can be considered. The consideration of space is
important to obtain a realistic model of structural development and its impacts on the
landscape. The more fragmented and dispersed fields are in a region the lower the potential
for profitable farm expansion. The size distribution of fields is also a defining characteristic of
an agricultural landscape and its mosaic. Finally the availability of agricultural and seminatural habitat is important for conservation of biodiversity in Europe.
To take these aspects into account we chose to extend the AgriPoliS model which is fully
described in Happe et al. (2006). The principle advantage of AgriPoliS compared with other
agricultural-economic models is that it can simulate agricultural development in a region over
time and in space. It is also based on individual farms (or agents) which makes it possible to
model the policy framework in detail and analyse impacts at both the farm and aggregate
levels. Farm decision making is modelled using mixed-integer programming. Fundamental to
AgriPoliS is that it models the competition for land between farms via a land market. The
dynamic nature of the model makes it possible to consider farmers entry and exit decisions as
well as investment and land rental activities. We extended the landscape modelling in
AgriPoliS in several ways to make landscape analysis possible, the details of which are
documented in Kellermann et al. (2007). Indicators for measuring biodiversity value (based
on species-area relationships) and mosaic (Shannon Index) were also introduced (Brady et al.,
2007).
288
Session B5: Integrated assessment methods for regional and landscape levels
Results and discussion
The decoupling Reform caused a significant reduction in the cultivated area in high-cost
regions, yet the GAEC condition ensured that this land was not abandoned. However,
converting land to GAEC resulted in an increase in the area of the dominating land-use in
high-cost regions, i.e., grass, and hence reductions in mosaic value (Figure 1a). In regions
where cultivation of commodities remained profitable after decoupling (EU-10 and
Mediterranean regions) it produced a small negative to positive impact depending on the
regional crop mix in both decoupling scenarios. The positive impact can be explained by the
relative increase in profitability and hence area of previously unsupported crops.
(a)
(b)
(a) Change in landscape mosaic
(b) Change in biodiversity value
15%
10%
5%
10%
0%
-5%
Jönköping
Västerbotten
Vysocina
Marche
Calabria
5%
-10%
0%
-15%
-5%
-20%
-10%
Jönköping
Västerbotten
Vysocina
Marche
Calabria
-25%
-15%
-30%
-20%
-35%
-40%
-25%
AGENDA
REFORM
BOND
AGENDA
REFORM
BOND
Figure 1. Impacts of decoupling on landscape values (a) mosaic and (b) biodiversity.
The type of decoupling (with or without GAEC) was found to be potentially very important
for maintaining biodiversity value in high-cost regions (Figure 1b). For example a large
portion of biodiversity value in Jönköping is associated with semi-natural grassland. GAEC
for this land requires that it be grazed annually which is almost identical to the condition
associated with Pillar II agri-environmental payments. Without the GAEC condition
biodiversity value would have fallen by almost 15% or 26 red-listed species. A similar result
did not emerge for Västerbotten because the area of semi-natural habitat is minimal.
Similarly, the case of Vysocina illustrated that small changes in land use can have significant
implications for biodiversity value in regions where marginal biodiversity value is high which
is the case for pasture in this region.
Overall our results show that decoupling has potentially quite negative consequences for
the landscape but these will be regionally specific and might be offset by interaction with
existing environmental and national support schemes. A minimum land management
requirement, as embodied in the GAEC condition, was not necessarily sufficient to preserve
landscape and biodiversity values.
References
Brady, M., et al., 2007. Environmental Impacts of Decoupled Agricultural Support: A
Regional Assessment, IDEMA Deliverable 24/SLI- Working Paper 2007: 1. www.sli.lu.se/
IDEMA.
Happe, K., et al., 2006. Ecology and Society 11: 49. http://www.ecologyandsociety.org/
vol11/iss1/art49/.
Kellermann, K., et al., 2007. AgriPoliS 2.0. Documentation of the Extended Model, IDEMA
Deliverable 20, IAMO www.sli.lu.se/IDEMA. Halle, Germany.
289
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Generic bio-economic modelling of plant diseases: An exploration
A. Breukers, W. Van der Werf, M. Mourits, A. Oude Lansink
P.O. Box 8120, 6700 EW Wageningen, The Netherlands
Contact: Annemarie.Breukers@wur.nl
Introduction
Cost-effective management of pests and diseases in plant production chains requires a
thorough understanding of the system’s dynamics and the effects of control measures.
Existing analytical models of plant disease epidemics often do not match with managers’
perception of the system (McRoberts et al., 2003; Madden, 2006) while simulation models
generally have a case-specificity that limites the domain of application and does not enable
generic insights (e.g., Thackray et al., 2004; Willocquet & Savary, 2004; Breukers et al.,
2006). Consequently, there is a need among plant health risk managers for generic tools that
fit with their perception of the system and yet provide plausible results. To meet this need, we
developed a generic bio-economic conceptual framework for plant disease epidemics and
disease management. In this framework dynamical processes in the plant production chain are
modelled from the perspective of a plant disease manager operating at the national level, i.e. a
plant health authority such as the Dutch Phytosanitary Service.
The conceptual framework
The framework considers the true world as a collection of objects, structured in
compartments. Objects represent (aggregations of) entities that can become infected or
infested with the disease of interest. All objects of the same type are grouped in one
compartment. Examples of compartments and corresponding objects are a production chain
containing lot objects of a particular crop, arable land containing arable field objects, and
woodland containing forest objects. Objects can multiply, die, and transmit disease within and
between compartments via one or more pathways. Pathways may be partly or completely
blocked by control measures. The likelihoods of birth, death, infection, and control differ
periodically over time due to seasonality in crop production activities. This periodic variation
is represented in the framework by splitting one time step (normally a year) in a series of
periods, called phases. Figure 1 provides a diagrammatic representation of the system.
Time (t)
t=0
System (compartments)
t = m-2
Production
chain A
Production
chain B
t = 2m
t=m
t = m-1
t=m
X
X
t =2m-2
t = 2m-1
t = 2m
X
X
Agricultural
fields
Environmental
compartment
Figure 1. Schematic representation of modelled system according to the generic framework.
Figures represent objects; shaded ones are infested. Arrows indicate transmission between
objects; a bar indicates that transmission is blocked by a control measure. Objects with a cross
are removed from the compartment.
290
Session B5: Integrated assessment methods for regional and landscape levels
The framework is structured according to the theory of periodic matrix modelling (Skellam,
1966), where each phase has its own transition matrix. Adjustments and extensions to this
structure were made to overcome some inherent limitations of matrix modelling. First,
nonlinearity is introduced to account for the probability of multiple infection, which is a
function of the fraction infested within a compartment. Second, memory is introduced to
represent history effects of crop rotation on disease dynamics. Control measures are included
as parameters that affect susceptibility of healthy objects (preventive measures), or
transmission (reactive measures) and survival (eradication measures) of infested objects. An
economic module quantifies the losses from crop damage and the direct costs of control
measures. Indirect costs, such as impacts on trade and product prices, are not accounted for.
Results and discussion
The conceptual framework as described above was translated into a functional prototype
model for three pathogens: Ralstonia solanacearum, a bacterium that causes brown rot of
potatoes, Phytophthora infestans, an oomycete causing potato late blight, and Meloidogyne
chitwoodi, a nematode causing damage to many root crops. Table 1 shows the compartments
and types of objects defined for each of the three cases.
The three model applications comprise pathogens with very different biology, dispersal
pathways, host sites, and potential control measures. Yet, simulation results for each pathogen
were plausible based on expert judgment and agreement with empirical evidence, confirming
that the framework is generic as well as valid for evaluating options for disease management
at the production chain level. Important strengths of the modelling framework are thus
generiticy, plausibility in three case studies, and congruence with the level of decision making
of plant health authorities that have a responsibility for mitigating economic impacts of plant
disease at the national level.
The generic framework presented and tested in this paper provides a new analytical
method for obtaining insight into dynamics and control of plant diseases. While its basic
structure resembles a periodic matrix model, it deals with more complex systems for which
the assumptions of matrix modelling (e.g., linearity in transition parameters, unrestricted
population growth) do not hold. Also, it contains several unique features not captured by
existing epidemiological models, and thus provides a valuable contribution to fundamental
plant disease modelling.
Table 1. Overview of types of objects per case study. Each object type represents one
compartment.
Compartment types
R. solanacearum
P. infestans
production chains
•
•
Seed potato lots
Ware potato lots
•
•
Seed potato lots
Ware potato lots
arable land
•
•
•
Fields, not detected
Fields, detected
Surface water
•
Fields
•
Waste piles
environmental
M. chitwoodi
•
•
•
•
•
•
•
Seed potato lots
Ware potato lots
Flower bulb lots
Other root crop lots
Non-root crop lots
Fields, not detected
Fields, detected
References
Breukers, A., et al., 2006. Agricultural Economics 38: 137-149.
Madden, L.V., 2006. European Journal of Plant Pathology 115(1): 3-23.
McRoberts, N., et al., 2003. Australasian Plant Pathology 32(2): 167-180.
Skellam, J.G., 1966. Seasonal periodicity in theoretical population ecology. Proceedings of
the 5th Berkeley Symposium on Mathematical Statistics and probability, Berkeley CA.
Thackray, D.J., et al., 2004. Virus Research 100(1): 67-82.
Willocquet, L. & S. Savary, 2004. Phytopathology 94(8): 883-891.
291
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integration of agriculture, groundwater and economic models using the
Open Modeling Interface (OpenMI): Application to the Ogallala Aquifer
Tom Bulatewicz, Xiaoying Yang, Jeffrey M. Peterson, Scott Staggenborg,
David R. Steward, Stephen M. Welch
Kansas State University,
Dept. of Computing and Information Sciences,
Manhattan KS 66506, USA
Contact: tombz@ksu.edu
Introduction
Developing comprehensive understanding of agricultural systems requires knowledge that
crosses traditionally isolated disciplinary procedures, approaches, methodologies, and
ontologies. This study describes current efforts to address this challenge by an
interdisciplinary team of engineers and scientists at Kansas State University. This builds upon
ongoing projects funded by the USDA/ARS Ogallala Aquifer Initiative, which seeks to
understand the economic and environmental risks associated with policy change. It also builds
upon the Kansas NSF funded Ecoforecasting Initiative, which seeks to forecast ecological and
environmental change.
Methods
This study brings together three core models for agriculture, groundwater and economics. The
agriculture simulation is provided by the Environmental Policy Integrated Climate (EPIC)
model that uses the inputs of crop type, weather, soils, and production practices to produce
field-scale outputs of yield, irrigated water use, biomass, percolation to groundwater, and
runoff. The groundwater model is based upon the Analytic Element Method and uses inputs
of well pumping rates, geologic aquifer parameters, and groundwater/surface water
interactions to produce outputs of groundwater elevation and flow rates across wide
geographic areas. The economic model is based upon the polychotomous choice selectivity
model and uses inputs of market input and output prices and factors that affect economic
decisions such as type of irrigation, climate, soils, and historical land use to produce outputs
of crop selection and irrigated water use at the field-scale.
Building on earlier work (Steward et al., 2008; Bulatewicz et al., 2008), we have
integrated these three models using the Open Modeling Interface (OpenMI), which describes
a standard way for models to interact. A wrapper program was written for each model that
conformed the model to the standard interface. As part of each wrapper, we chose a selection
of model inputs and outputs that can be exchanged with other models through the interface. In
this way, each model has no knowledge of the other models, mitigating the challenging task
of integrating the three models in different compositions as well as allowing for further
integration with other OpenMI-compliant models. The input data to the integrated model was
provided by a geo-database (Yang et al., 2008).
The models have been integrated such that each one exchanges data with the other two on
an annual basis as illustrated in Figure 1. The economic model informs the crop model about
what crops are likely to be grown in a particular year. This decision is based in part on the
saturated thickness of the aquifer as estimated by the groundwater model. The saturated
thickness of the aquifer is in turn influenced by the amount of irrigation water that is pumped
to grow the crops simulated by the crop model. Although the models use different time steps
internally, data is exchanged once per year.
292
Session B5: Integrated assessment methods for regional and landscape levels
Results and discussion
The Open Modeling Interface provides an effective integration methodology to bring together
traditionally isolated modelling approaches. Results will be shown that illustrate the impacts
of policy change on agricultural production as well as groundwater resources and economic
activity. These findings provide a scientific support to guide policy analysis and management
decisions.
Figure 1. The three models and the pattern of data exchange between them.
References
Bulatewicz, T., et al., 2008. Journal of Hydrology. (in review)
Steward, D.R., et al., 2008. Water Resources Research. (accepted)
Yang, X., et al., 2008. International Journal of GIS. (in review)
293
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
How to design technical and organizational innovations to promote
sustainable development in catchments with intensive use of pesticides
C. Gary, J.M. Barbier, P. Rio, P. Andrieux, J.M. Blazy, X. Louchart, M. Bonin,
J.L. Diman, F. Causeret, H. Ozier-Lafontaine
INRA, 2 place Pierre Viala, 34060 Montpellier Cedex 2, France
Contact: gary@supagro.inra.fr
Introduction
The sustainable development of agriculture in regions where perennial crops such as
grapevine in south France and banana in the French West Indies dominate is questioned in
relation to their high use of pesticides. The resulting degradation of the environment generates
damages for various activities including agriculture. The cost of adopting alternative crop
protection strategies and/or restoring water quality is high. The consequences on human
health and environment generates conflicts with other stakeholders with feedback
consequences on agriculture in terms of policy (regulations), market and social recognition.
The observed low diffusion of low-input cropping systems results from technical, economic
and organizational limitations at several scales, from field and farm to catchment and region.
Then any proposal of alternative technologies should be embodied in sets of consistent
innovations of different natures and at different scales.
In terms of research methodology, the challenge is to design novel agricultural systems and
carry out ex-ante their assessment in a way that connects various scales and balances all
dimensions of sustainability (Van Ittersum et al., 2008). Various methods of integrated
assessment have been proposed; they are all based on systems analysis, they mobilize in a
concerted way several disciplines and use models as a mean to explore the effectiveness of
various scenarios (Parker et al., 2002).
In the present project, skills in human (economy, geography, sociology) and biophysical
(agronomy, hydrology, engineering) sciences were gathered to (i) design innovative farming
systems that would reduce the use and diffusion of pesticides, (ii) evaluate their ecological
effectiveness and likelihood of adoption by farmers and (iii) identify the organizations and
regulations that would favour sustainable development in the studied catchments.
Methods
A generic framework was adopted to organize the various scientific disciplines and
approaches (Figure 1). The focus was more on the integration and consistency of these
approaches than on the formal connection of a set of models differing in various ways:
static/dynamic, mechanistic/empirical, biophysical/decisional, field/farm/catchment.
The influence of the institutional context was examined in two ways. First, a typology of
mechanisms of incitation or repression was built and their potential impact on farming
systems assessed with linear programming. Second, the role of networks of information
among farmers in relation with the diffusion of innovations was studied and modelled with
Multi-Agent Systems.
The design of innovative farming systems was made according to two approaches. In the
grapevine catchment, surveys were carried out to analyse the diversity of farmers’ strategies
of weeding, soil management and crop protection. Some of these existing strategies were
identified as innovative. In the banana catchment, it was considered that an input of novel
techniques had to be introduced. To this end, a process of prototyping was engaged with
experts (agronomists, geneticists, nematologists) after the typology of farming systems. It
produced innovative cropping systems and bio-economic modelling was used to select those
potentially fitting with the various types of farming systems.
The economic and environmental performances of the innovative farming systems were
294
Session B5: Integrated assessment methods for regional and landscape levels
assessed with biophysical models and/or indicators. This assessment focused on farm scale in
the case of banana (the innovation resulting from the adoption of novel cropping systems by
farmers) and on catchment scale in the case of grapevine (the innovation resulting from new
distributions of performing types of farming systems within the population of farms).
At last the adoptability of the most effective farming systems was evaluated with new surveys
and the conditions of adoption were identified with an econometric model.
context (regulations,
institutions, networks, market)
innovations from
research
(laboratory models)
farm
constraints
+
choice of
candidate innovations
candidate
innovations
+
prototyping
action
model
regional diversity of farms
(structure, strategy,
functioning, effectiveness)
likelihood of adoption
econometric
model
evaluation of
conditions of
adoption
innovative
farming systems
biophysical
models
and indicators
+
productive and
environmental
effectiveness
+
evaluation of innovative
systems
soils and climate
Figure 1. Scientific framework for the integrated assessment of innovative farming systems
(grey rectangles are models, white rectangles, the real world, and ovals outputs).
Results and discussion
Government and farmers’ institutions recently introduced new instruments and, in some
cases, their theoretical effectiveness could be assessed. The fluxes of information within
farmers’ networks appeared to limit the diffusion of innovation. Yet the process of design of
innovative farming systems differed among grapevine and banana catchments, this social
context was considered as a forcing variable in both cases. The economic and ecological crisis
was more severe in banana catchments, which justified the interest for radical technical
innovations and for their thorough assessment at farm scale.
A set of assessment tools were produced, from field to farm and catchment, including
breakthroughs such as the coupled simulation of the dynamics of crop and nematode
development in banana fields, or the coupled simulation of cultivation techniques and
resulting surface transfer of water and pesticide distributed within a grapevine catchment.
Attention was paid to the likelihood of adoption of innovative systems by farmers, in relation
to the economy and organization of their farm and to the innovation and policy attributes.
Combined with the evaluation of crop production and externalities, it provided a framework
for an appraisal of the contribution to sustainable development of existing and alternative
farming systems. The coupling of various scales and criteria of evaluation should facilitate the
analysis of the consequences of specific policies designed to promote novel farming systems.
In this perspective, the interaction with stakeholders, including policymakers, will have to be
more formalized.
References
Parker, P., et al., 2002. Environmental Modelling & Software 17: 209-217.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
295
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Visualizing changes in agricultural landscapes
S. Griffon, D. Auclair
CIRAD, UMR AMAP, TA A-51/PS2,
34398 Montpellier Cedex 5, France
Contact: sgriffon@cirad.fr
Introduction
Rural land managers, foresters and farmers, but also local decision makers, local authorities
and members of local governments, are increasingly aware of the necessity to take into
account the perception of the landscape by the general public, and to predict the evolution of
landscapes according to management decisions (Bergen et al., 1995). Different management
choices can lead to similar, or to very different landscapes. The positioning of woodlots, of
fields, and of agro-forestry areas, the type of silvi-cultural management (selective or
systematic thinning, artificial pruning, clear-cut or shelterwood systems, reforestation, choice
of species, etc…) or agricultural system (rotation, land attribution, crop allocation, etc…) and
the balance between forest and agriculture, are susceptible of drastically modifying the visual
aspect of the landscape. Although land managers generally have a good experience of what
result can be expected from their decisions, they are often faced with difficulty when trying to
communicate the visual impact of a future management option to all the stakeholders (local
and regional decision makers, land managers, landscape planners, and various communities
involved in outdoor activities). Three-dimensional visualization of the landscape provides
means that are better understood than maps, especially for the general public. With such
methods, visual changes of the landscape can be shown very impressively, which can allow
for an intuitive assessment of the visual landscape quality.
The present work describes a landscape visualization tool which has been developed within
the SEAMLESS project (Van Ittersum et al., 2008). This landscape visualization component
should be launched at the end of a policy simulation to allow for exploration of landscape
changes, as a post-model analysis, to be used as a basis for discussion and negotiation within
the community.
Methods
The users (modeller or policy-maker, or any other stakeholder) who are interested in a
specific landscape should initially start with a description of the area they wish to visualize.
This requires a description, preferably mapped in GIS format, of environmental data such as
land cover and land use. Pressures causing changes in landscape can be simulated by a bioeconomic farm model (such as the FSSIM module of the SEAMLESS project). This is then
translated into changes in the spatial configuration of the landscape. A 3D data conversion
plug-in has been developed and integrated into the open-source GIS software QGIS
(http://www.qgis.org/). It works as an external module that allocates each agricultural parcel
to a specific land-use by importing the proportion of each land-use class computed by the bioeconomical farm model and distributing it on the field pattern according to specific rules (in
the present version, randomly). Then the plug-in can export the extent selected by users (most
reasonably around 10 km by 10 km). Data are cropped, fused and formatted to be visualized.
The generation of the 3D model and its rendering is done by a specific software component,
that we have called here ‘Seamless Landscape Explorer’ (SLE).
SLE is then launched and the user can edit the geo-typical configuration (land-use textures
and vegetation) for each land-use class. A dynamically optimized elevation mesh from the
digital terrain elevation raster is first computed using the Geographic MipMaps technique (De
Boer, 2000). Then the mesh is textured with Texture Splatting technique and with satellite
imageries or thematic maps and vegetation can be spatialized according to user parameters.
296
Session B5: Integrated assessment methods for regional and landscape levels
The different layers of vegetation consist in trees, shrubs, small plants, and other objects, such
as rocks, are added to complete the impression of natural complexity. The rendering is done
in real time.
The example of application presented here concerns the Mediterranean territory of Pic
Saint Loup (near Montpellier, Hérault, France), where four scenarios were studied, as part of
a participatory process for planning the regional peri-urban and agri-environmental policy
(Nespoulous, 2004). The scenarios concern high or low biodiversity and high or low urban
pressure. Land-use maps were produced for each scenario and the resulting scenes were
processed with SLE.
Results and discussion
Figure 1 shows an example of land-use change according to two scenarios: biodiversity with
agriculture (left) and peri-urban pressure with environmental concern (right).
Figure 1. Example of Pic Saint Loup landscape change.
These images, as well as fly-over videos, were discussed with stakeholders (both professional
and general public). These computer simulations were used as a support for understanding
and characterizing the landscape.
Decisions support systems are increasingly being applied in spatial planning, and virtual
landscapes become an important part of decision making. Planners recognize realism as an
important factor in this type of visualization (Appleton & Lovett, 2005). It is important to
define an appropriate level of realism, because the ‘photorealism’ can have potential negative
effects if it is not linked to real-world data. The tool presented here provides the opportunity
to control several aspects, in particular the viewpoint. In the example shown here, the Pic
Saint Loup mountain in the distance is important for local stakeholders to recognize clearly
their familiar landscape.
References
Appleton, K. & A. Lovett, 2005. Computers, Environment and Urban Systems 29: 321-339.
Bergen, S.D., et al., 1995. Journal of Forestry 93: 33-37.
De Boer, W.H., 2000. http://www.flipcode.com/articles/article_geomipmaps.shtml
Nespoulous, A., 2004. MA Thesis, Univ. Montpellier, 89 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
297
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated analysis of the effects of agricultural management and spatial
planning on environmental quality at landscape scale
J. Kros, T.J.A. Gies, J.C. Voogd, R. Smidt, W. De Vries
Alterra, Wageningen University and Research Centre,
P.O. Box 47, 6700 AA Wageningen, The Netherlands
Contact: Hans.Kros@wur.nl
Introduction
An important threat of terrestrial Natura 2000 sites in The Netherlands is atmospheric
nitrogen (N) deposition. High N deposition may lead to adverse effect on terrestrial
ecosystems, such as the loss of plant species diversity. For the setting of emission standards
and deposition targets, critical nitrogen deposition loads are used. In The Netherlands most
(almost 70%) of the N deposition on the Natura 2000 sites is coming from ammonia emission
from agricultural sources. Consequently, most emphasis is lead on the reduction of ammonia
emission from agriculture. Because most of the emitted ammonia is deposited at a distance of
less than 30 km, the landscape scale is the typical scale for looking for solutions.
Models that simultaneously assess atmospheric emissions from complex agricultural
landscapes and deposition to nearby areas have been limited so-far. An example is the model
INITIATOR2 (Integrated Nutrient ImpacT Assessment Tool On a Regional scale) that can be
used for an integrated assessment of the effectiveness of policies aiming at the simultaneous
reduction of all relevant element fluxes (nutrient and contaminants) to atmosphere, ground
water and surface water (De Vries et al., 2005). INITIATOR2 predicts atmospheric emissions
of ammonia and greenhouse gases (CO2, CH4 and N2O) from housing and manure storage
systems and from soils and also the accumulation, leaching and runoff of nitrogen (N),
phosphate (P) and metals from soils. This paper is aiming at the effectiveness of emission
control measures in agriculture on the deposition on Natura 2000 sites, using the
INITIATOR2 and focusing on the NH3 emission and resulting N deposition. This integrated
modelling system is used for predicting current and future deposition, to identity which
measures are most promising for one province in The Netherlands and to identify the side
effects on other emissions such as N2O and CH4 and leaching of N.
Modelling approach and measures
To gain insight in the environmental impacts of management measures on the environment,
INITIATOR2 was applied to predict atmospheric emissions of ammonia from housing and
manure storage systems and from soils. A database (GIAB) with spatially explicit data on
animal numbers in many animal categories, agricultural practices and land management, such
as manure application techniques available data for each farm in The Netherlands was used
for the year 2004, being a reference year for the calculations (Anonymous, 2004).
Furthermore, spatially explicit input data related to soil parameters (soil map 1: 50.000) and
hydrology was used to make the model predictions. The NH3 emissions in the field and those
from housing systems form the input of an atmospheric transport model (OPS, Van Jaarsveld,
1995) to assess the N deposition on agricultural and non-agricultural systems.
Evaluated measures were amongst others (i) emission reduction from poultry and pig
housing system by using air washers, (ii) low protein feeding, (iii) reducing fertilizer amount,
(iv) conversion towards biological dairy farms, and (v) moving farms to a larger distance
from the Natura 2000 sites.
Results and discussion
The critical N deposition of the Natura 2000 sites occurring in Overijssel (province located in
the North-eastern part of the country) is exceeded in almost 60% of the area (Figure 1).
298
Session B5: Integrated assessment methods for regional and landscape levels
Exceed ance of critical N
deposition ( %)
No exceedances
Figure 1. Relative exceedances of the critical N deposition for nature target types in the
province of Overijssel, The Netherlands.
Shifting from traditional agricultural management towards biological dairy farming within
zones of 3 km surrounding the Natura 2000 sites resulted is one of the most promising
measures this resulted in almost a reduction of 50% of the emissions from that zone. The N
deposition on the Natura 2000, however, resulted in a reduction of only 7%. This is due to a
large part of the N deposition (more than 70%) that is originated from other sources than
ammonia or from sources outside this province. At the AgSAP conference also other emission
(N2O, CH4) and the leaching of N will be presented as well as the relation between generic
measures and local measures. It is concluded that measures in a zone surrounding a nature
areas are especially beneficial for local situations with close-by high emissions, but for an
overall protection more stringent generic measures are necessary.
References
Anonymous, 2004. Statistics for agriculture and horticulture, 2004 (in Dutch). Landbouw
Economisch Instituut (LEI) en Centraal Bureau voor de Statistiek (CBS), Den Haag.
De Vries, W., et al., 2005 In: Eds. Z. Zhu, et al., 3rd International Nitrogen Conference Contributed papers. Science Press USA Inc., Monmouth Junction, pp. 859-870.
Van Jaarsveld, H.J.A., 1995. Modelling the long-term atmospheric behaviour of pollutants on
various spatial scales. PhD Thesis, Universiteit Utrecht, Utrecht, 235 pp.
299
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessment of nitrogen balances at farm and regional level
Bernhard Osterburg, Thomas G. Schmidt
Institute of Rural Studies, Johann Heinrich von Thünen-Institut (vTI),
Bundesallee 50, 38116 Braunschweig, Germany
Contact: bernhard.osterburg@vti.bund.de
Introduction
Nitrate leaching into ground water caused by agricultural nitrogen (N) surplus is a main
challenge of environmental policies in many European areas. The implementation of the EU
Water Framework Directive (WFD) calls for new efforts to reduce N surplus and groundwater
pollution in order to achieve a good status of water bodies by 2015, or at least reach trends for
the better (EC, 2004). Regions with high animal stocking densities are more likely to face
problems of achieving these targets, due to high organic N inputs and less efficient use of N
from animal excretions.
N balances are the point of departure for analysing nitrate pollution of the groundwater,
and for identifying mitigation options. Established approaches (e.g., PARCOM guidelines
(PARCOM, 1995), OECD N balance database (OECD, 2001)) deal with figures from
agricultural statistics at national level. Core problem of assessing N balances at the landscape
level is the lack of regional data for mineral fertilizer input. Therefore, in most approaches,
normative estimates for mineral inputs are applied in order to disaggregate national data. In
these approaches, livestock density and fixed factors for N efficiency (N output/input relation)
are used without broader knowledge about real distribution of N efficiency in the farming
sector. The assessment of regional N balances presented in this contribution seeks for
improvements regarding the depiction of mineral N input.
Methods
The main data input for the calculation is agricultural census data containing numbers of
crops, animals and farm types for the municipality level (LAU-2, formerly NUTS 5 of the EU
classification), as well as several additional pieces of information like regional yields and
farm import or export of manure. Parameters for manure production by animal category as
well as N contents of agricultural products were provided by the Agricultural Chamber of
Lower Saxony. In addition, accounts of 6,600 farms (2-year average from 1999/2000 and
2000/2001), containing information on physical mineral fertilizer input, were used to model
mineral N input. Organic manure from livestock is calculated as fixed output coefficients per
livestock species multiplied by herd size according to the agricultural statistics. N excretion of
dairy cows is computed in dependence of milk yield per cow and year. As endogenous
variables, N output, legume N-fixation, N from animal excretion, soil quality index and the
regional density of pig production (as a proxy for manure imports not stated in the accounts)
were used. The model derived from farm data is applied to the regional farm survey data,
using the sales of mineral fertilizer recorded in agricultural statistics on federal the state level
as consistency frame.
The N uptake is calculated as the nitrogen in cash and forage crops. Ammonia losses into
the atmosphere are assessed separately, as they are part of the N deposition, too. The uptake
of cash crops is calculated on the basis of statistical regional average yield and acreage per
crop type. In order to cope with the uncertainties related to roughage production, an estimation based on animal input-output assures consistency between roughage as well as fodder
concentrate input, and manure and animal product outputs. A multiple regression equation for
concentrate input is derived from farm accounting data, allowing for an improved estimation
of roughage production. Endogenous variables explaining the N intake from concentrates are
livestock category, milk yield, and livestock stocking density per hectare of forage area.
300
Session B5: Integrated assessment methods for regional and landscape levels
Real mineral N input (kg per hectare)
Results and discussion
Mineral nitrogen input per hectare is simulated for each farm using the new model, and
compared with real data from farm accounts. The R-square (coefficient of determination) of
the regression model is about 0.3, thus the model performance is rather weak. This is due to
high variance of mineral N input per hectare even in farm groups with similar structures.
However, an analysis of residual values between real and estimated mineral N input values
(both for structural or regional characteristics) shows that the model estimates do not contain
systematic deviations for specific subgroups, so that positive and negative deviations cancel
out each other (see Figure 1). Therefore, the results are regarded to be sufficiently exact to
depict the average situation of different regions, farm types and structures.
400
300
200
100
0
0
100
200
300
Estimated mineral N input (kg per hectare)
Figure 1. Estimated and real mineral fertilizer input per hectare for 6,600 farms.
An assessment of regional N balances using normative assumptions helps to identify regional
hot spots of N surplus, e.g. based on statistics of livestock stocking densities. However, for
using N balances as a basis for policy recommendations, more knowledge on the productive
use of nitrogen at farm level is needed. The high variance of mineral N inputs indicates that
there are considerable ‘hot spots’ due to different individual farm performance. Obviously,
management and advice are important factors for N efficiency and resulting N surplus at farm
level. The data analysis shows that average N efficiency of animal manure is higher in farms
with high livestock density. Thus, recommending a lower livestock density, allowing for
spatial reallocation of the livestock herd, does not automatically tackle the underlying
problem of low N efficiency. The results show that micro data analysis is needed to better
identify sources of N emissions and to reveal scope for further improvements of N efficiency.
References
EC, 2004. Monitoring under the Water Framework Directive, Guidance Document No 7.
Produced by Working Group 2.7 – Monitoring, Common Implementation Strategy for the
Water Framework Directive (2000/60/EC). European Commission, Brussels.
OECD, 2001. Nitrogen balance data-base. Organisation for Economic Co-operation and
Development, Paris.
PARCOM - Paris Commission for the Environmental Protection of the North East Atlantic
(1995) PARCOM Guidelines for Calculating Mineral Balances; Oslo and Paris Conventions for the Prevention of Marine Pollution Programmes and Measures Committee
(PRAM), Oviedo: 20-24 February 1995, Annex 15 (Ref: § 7.19).
301
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing the institutional compatibility of new policies
Christian Schleyer, Insa Theesfeld, Olivier Aznar
Humboldt-Universität zu Berlin, Department of Agricultural Economics and Social Sciences,
Division of Resource Economics, Philippstr. 13, Building 12, 10099 Berlin, Germany
Contact: christian_schley@hotmail.com
Introduction
Policy assessment from an institutional perspective follows the concept of institutions for
sustainability, which is defined as the necessary institutional structure capable of delivering
economic, social, and environmental sustainability objectives. Thus, the effectiveness of a
policy and the cost-effectiveness of its implementation depend to a large extent on the degree
of compatibility between this policy option and the respective institutional context. However,
not least because institutions usually relate to a great diversity of situations, the state-of-theart in institutional economics offers hardly any standardized procedures for institutional
analysis that can easily be combined with environmental and agricultural models widely used
for policy impact assessment (Theesfeld et al., 2008). To assess the compatibility between
policy options and various institutional contexts a standardized methodology has been
developed within the SEAMLESS-Integrated Project (Van Ittersum et al., 2008) that provides
for an institutional dimension in modelling: the ‘Procedure for Institutional Compatibility
Assessment’ (PICA) (Schleyer et al., 2007).
Following an overview about the state-of-the-art in approaches for policy assessment, this
contribution will focus on the PICA methodology. Here, the four distinct steps are elaborated
on using the EU Nitrate Directive as a concrete policy example to illustrate the procedure (see
Figure 1). Finally, some ideas will be presented how PICA serves as a tool for pre- and
postmodel analysis of environmental and agricultural models for policy impact assessment in
a complementary way.
Method
PICA comprises four distinct steps:
(1) Policy options are clustered according to the type of intervention (regulatory, economic,
and advisory), the area of intervention (hierarchy/bureaucracy, market, and self-organized
network), possibly involved property rights changes, and the attributes of the natural
resource addressed. This classification allows identifying the generic structure of a policy
option.
(2) Each policy cluster is linked to specific sets of crucial institutional aspects (CIA) that
may constrain or foster policy implementation.
(3) Institutional indicators are used to evaluate the potential of a respective CIA.
(4) The information provided by the institutional indicators is used for a qualitative
assessment of each identified CIA. Subsequently, the CIA and the related assessments are
arranged in thematic categories of institutional compatibility leading to qualitative
statements about the probable effectiveness of a policy option.
Results and discussion
PICA allows for a systematic institutional ex-ante assessment of (agricultural, environmental,
and rural development) policies. This enables policymakers and decision makers in charge of
implementing policies to identify at an early stage (potential) institutional incompatibilities
between policy options and the various institutional contexts in different countries and
302
Session B5: Integrated assessment methods for regional and landscape levels
regions. In addition, PICA provides hints for a better policy design in terms of effectiveness
and cost-efficiency. This may include redesigning or adapting the policy options and the
design of complementary policy measures.
Society
Policy Option
e.g., EU Nitrate
Directive
Type of
Intervention
Agricultural
Subsidies
Estimate
Area of
Intervention
Number of farms
Number of farmers
that are member
Number of people
in a farmers’ association
employed in the farming sector
Property
Rights Change
Step 3: Select indicators
(e.g., Membership in farmers’
associations)
Step1: Identify policy type
(e.g., regulatory / market)
Step 2: Extract crucial institutional aspects
(e.g., Bargaining power of farmers’ associations)
Use existing institutional indicators
or elaborate new proxies that indicate
the extent of the crucial institutional aspect
Step 4: Conclude on e.g., Communication capacity
Figure 1. PICA applied to the policy option ‘EU Nitrate Directive’.
References
Schleyer, C., et al., 2007. Approach towards an operational tool to apply institutional analysis
for the assessment of policy feasibility within SEAMLESS-IF, SEAMLESS Report No.29,
SEAMLESS integrated project, EU 6th Framework Programme, contract no. 010036-2,
www.SEAMLESS-IP.org, 169 pp.
Theesfeld, I., et al., 2008. Ex-ante Policy Assessment from an Institutional Perspective: A
Procedure for Institutional Compatibility Assessment. Eds V. Beckmann & K. Hagedorn,
Institutional Change in Agriculture and Natural Resources (ICAR) Discussion Paper
20/2008.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
303
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing sustainable agriculture development using the MicroLEIS DSS in
Souma area, Iran
F. Shahbazi, A.A. Jafarzadeh, M.R. Shahbazi
Soil Science Department, Faculty of Agriculture, University of Tabriz,
51664-16471 Tabriz, Iran
Contact: shahbazi@tabrizu.ac.ir
Introduction
This study was performed in the Souma area with about 4100 ha extension which has located
between 44°35' to 44°40'east longitude and 37°50' to 37°55' north latitude in the North-West
of Iran (west Azarbaijan). Site, soil and climate data of study area were stored in SDBm plus
(De la Rosa et al., 2002) and CDBm, major components of MicroLEIS DSS. The agricultural
land uses considered for evaluation were wheat (Triticum aestivum), alfalfa (Medicago
sativa), sugar beet (Beta vulgaris), potato (Solanum tubersum), and maize (Zea mays) crops
which are relevant in the region. Sustainable agriculture development for these crops is
important and also necessary. A specific agricultural use and management system on land that
is most suitable according to agro-ecological potentialities and limitations is the best way to
achieve sustainability (FAO, 1978). Since the early 1990s, and in this conceptual framework
was developed the land evaluation decision support system MicroLEIS DSS (De la Rosa et
al., 2004). The MicroLEIS DSS system has been widely used over the last 20 years for many
different purposes which the highest number is for Europe mainly Spain and other
Mediterranean and Semi-arid countries, additionally, the South America countries also have
many users. This system has been applied for the first time in west Asia (Iran), for land use
planning in Ahar region (Shahbazi et al., 2008).
Methods
MicroLEIS DSS system, through its 12 land evaluation models, analyses the influence of
selected soil indicators on critical soil functions referred to: (1) land productivity (agricultural
and forest soil suitability, crop growth, and natural fertility), and (2) land degradation (runoff
and leaching potential, erosion resistance, subsoil compaction, workability, and pollutant
absorption and mobility). In this research work, bioclimatic deficiency evaluation, segregation
of agricultural lands, restoration of marginal areas, and diversification of crop rotation to
achievement the sustainable agriculture is more importantly attended (Table 1).
Table 1. Micro LEIS land evaluation models according to the soil function simulated and the
specific strategy supported.
Constituent
Soil function
Specific strategy
model
(Modelling type)
Land use planning-related
Terraza
Cervatana
Sierra
Almagra
Bioclimatic deficiency
(Parametric)
General land capability
(Qualitative)
Forestry land suitability
(Qualitative/Neural network)
Agricultural soil suitability
(Qualitative)
Quantification of crop water supply and frost
risk limitation
Segregation of best agricultural and marginal
agricultural lands
Restoration of semi-natural habitats in marginal
agricultural lands: selection of forest species (61
species)
Diversification of crop rotation in best
agricultural lands: for traditional crops (12
crops)
304
Session B5: Integrated assessment methods for regional and landscape levels
Input variables are physical/chemical soil parameters (e.g., useful depth, stoniness, texture,
water retention, reaction, carbonate content, salinity, or cation exchange capacity) collected in
standard soil surveys, monthly agro-climatic parameters for long-term period (from the 36
consecutive years of Urmieh meteorological station), and agricultural crop and management.
Validation tests as far as possible have been done in the study area.
Results and discussion
The predicted results of applying the four agro-ecological land evaluation models (Terraza,
Cervatana, Sierra and Almagra) constituents of MicroLEIS DSS are presented and discussed
for nine applicable soil series. Results showed that the annual yield reduction of maize is the
highest amounts (74%) between the selected crops (Table 2).
Table 2. Evaluating results: Bioclimatic classes and annual reduction of crop production.
Selected Land uses
Rys%
Bioclimatic classes
Wheat
0
H1
Alfalfa
39
H2
Potato
55
H3
Sugar beet
60
H4
Maize
74
H4
Soil and erosion risk factors in total of 810 ha, caused to be urgently reforested with Swamp
Pine species. The rest of study area (3300 ha), are considered as agricultural lands. Also
results obtained from the model application and integrating with GIS are presented and
discussed in this research work (e.g., Figure 1). Cultivation of irrigated maize can be
recommended in the study area with attention to Terraza and Almagra model results.
Figure 1. Maize suitability map of study area.
References
De la Rosa, D., et al., 2002. Environmental Modelling and Software 17: 721-731
De la Rosa, D., et al., 2004. Environmental Modelling and Software 19: 929-942.
FAO, 1978. Report on the agro-ecological zones project. World Soil Resources Report 48,
Rome.
Shahbazi, F., et al., 2008. Int. Journal of Agrophysics 22: 277-286.
305
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing and predicting agro-structural change in the Alpine Region
T. Streifeneder, C. Hoffmann, F.V. Ruffini
Eurac Research, Drususalle 1, 39100 Bozen, Italy
Contact: thomas.streifeneder@eurac.edu
Introduction
In the past decades, mountain agriculture in the Alps has been affected by a profound
structural change, with distinct variations noted between individual regions (MacDonald et
al., 2000; Streifeneder et al., 2007). Regions with moderate farm abandonment rates are located mostly in the northern Germanic parts of the Alps, in contrast to the high abandonment
rates of those regions in the southern Romanic areas of the Alpine bow (Figure 1).
Based on the data from the AGRALP-project (www.eurac.edu/agralp), the present
approach assesses the future of agro-structural change until 2020. Therefore, a multiple
regression model has been created that utilizes agro-structural, socio-economic and geostatistical data. The model’s focus, thus, concentrates on the following research questions:
a) What factors most influence the future development of the agricultural sector?
b) What future farm abandonment rates can be expected?
c) What consequences for economic, social and landscape structures can be expected as a
result of the predicted agro-structural development?
Methods
The analysis was based on a comprehensive harmonized statistical database for integrated
assessment, with key indicators on the regional level (NUTS level 2) and the municipal level
(LAU 2 level). The prediction of the agro-structural change was adapted to the approach of
Weiss (2007). The forecasting process was conducted in two steps:
a) Clustering: By means of relevant driving factors only available on NUTS 2 (e.g., GDP,
unemployment rate, accommodations, etc.), six homogenous clusters were built from the
27 Alpine NUTS 2 areas.
b) Model building: Based on municipal data, a multiple regression model was devised for
each of these clusters aiming at:
• assigning the significance of the independent variables, and
• estimating the future development of the agro-structural change until 2020 following
three different scenarios that are essentially adapted to the Scenar 2020 approach (EC,
2006) and considers the results of the Seamless project.
1) Trend Extrapolation Scenario (Baseline):
Based on the long time-series of the Institute’s database, the trends of the observed
developments are extrapolated to estimate future changes.
2) Alternative Scenario 1 (Liberalization):
The liberalization approach is adapted to the predictions of the WTO. Assumed in this
approach are significant changes to the CAP in Europe regarding direct payments and
subsidies for agro-environmental measures or less-favoured areas.
3) Alternative Scenario 2 (Regionalization):
Characteristics of this scenario include slight modification of the current CAP measures (e.g.,
stronger modulation), a moderate decrease in direct payments and subsidies, special
compensation measures for mountain areas as well as small-structured (e.g., dairy cow
premium) farms, and higher product prices due to diversification strategies (e.g., high quality
products, agro-tourism).
306
Session B5: Integrated assessment methods for regional and landscape levels
Results and discussion
In the Alps, variances between areas with and without agriculture will be more pronounced in
the future. In all scenarios, stronger declines are detected in the southern regions (Italy) than
in the northern regions (Austria, Switzerland). This can be attributed to the regional
prevalence of small-structured farms and over-aged farmers, as well as to the lack of farm
successors. Negative impact from agro-structural change is felt mainly in peripheral regions
that lack tourist appeal, have limited infrastructure, and have poor accessibility. The Baseline
Scenario leads to an annual farm-abandonment of around 2–3%. Depending on the level of
progress of agro-structural change and the regional framework conditions, full-time and parttime farming industries develop quite heterogeneously. Overall, part-time farming increases.
The Liberalization Scenario records an annual farm-abandonment of 3–5%. Areas that until
now have only been moderately affected become more strongly affected under agro-structural
change. Full-time farms face a greater impetus to capitulate, as they can no longer expand or
specialize. Part-time farms are generally endangered with abandonment.
Within the Regionalization Scenario the farm-abandonment rate does not exceed 1–2%.
Full-time farms are less affected; the abandonment rate of part-time farms is low.
Figure 1. The farm abandonment rate depicts large regional differences in the Alpine
Convention area (5,954 municipalities).
References
European Commission, 2006. SCENAR 2020 - Scenario study on agriculture and the rural
world. EC, Directorate-General Agriculture and Rural Development.
MacDonald, D., et al., 2000. Journal of Environmental Management 59: 47-69.
Streifeneder, T., et al., 2007. Journal of Alpine Research 95(3): 27-52.
Weiss, F., 2007. Modellierung landwirtschaftlichen Strukturwandels in Österreich: Vergleich
einer Modellprognose mit den Ergebnissen der Strukturerhebungen (1999-2005). DP-332007. Universität für Bodenkultur, Vienna.
307
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Sustainability impact assessment of EU policies at regional level:
An analysis of land use dynamics and environmental effects
with the MEA-Scope tool
Sandra Uthes, Claudia Sattler, Amanda Osuch,
Kathrin Happe, Annette Piorr, Peter Zander
Leibniz-Centre for Agricultural Landscape Research (ZALF),
Eberswalder Strasse 84, D-14374 Müncheberg, Germany
Contact: sandra.uthes@zalf.de
Introduction
Sustainability impact assessment of EU agricultural policies is an important issue addressed
by a number of EU impact assessment tools. Yet social and environmental impacts are poorly
covered in many tools particularly in those that act at comparatively aggregate scales, e.g.,
with administrative regions as smallest scale. The EU-STREP MEA-Scope (2004-2007) has
developed a spatial bio-economic modelling tool for ex ante policy impact assessment that
attempts to describe economic and social events by modelling the behavior of individual
agents, while the geographic information attached to every agent also allows for a regional or
local approach to policy analysis. The MEA-Scope tool has proved useful in evaluating
economic and ecological impacts of policy changes at the micro level (Happe et al., 2006,
Piorr et al., 2008). This paper analyses the spatial distribution of a number of environmental
indicators subject to different scenarios of the Common Agricultural Policy (CAP), i.a.
Agenda 2000, single farm payment, phasing out of direct payments with and without agrienvironmental measures, in four EU case study areas: Ostprignitz-Ruppin, Germany (DE);
River Gudenå, Denmark (DK); Mugello, Italy (IT); Piešťany, Slovakia (SK).
Methodology
The MEA-Scope tool is based on three farm-level models, in which the development of
spatially localized single farms in reaction to different CAP scenarios is tracked over a time
horizon of 10 to 15 years. The models are loosely coupled in a hierarchical order. The agentbased model AgriPoliS simulates the interactions among the farms on the land market and
their investment decision, the bio-economic farm model MODAM simulates the cropping and
livestock patterns of the farms being the basis for a fuzzy-logic-based environmental impact
assessment, and the process-oriented dynamic simulation model FASSET simulates on-farm
matter flows. Typical farms of the Farm Accountancy Data Network (FADN) and various
GIS data sources are used to reproduce farm structure and farm localization in the regions
(Kjeldsen et al., 2006). The resulting single farms own or rent particular plots of land with
different soil, climate and elevation characteristics (grid cell size 1 ha). Farm activities
encompass land use and production decisions, rental activities, labour allocation decisions and
investments. During the simulation, a farm can change its characteristics such as farm size,
labour endowment, specialization and production activities. The fuzzy-logic based impact
assessment procedure (Sattler et al., 2006) delivers an Index of Goal Achievement (IGA) for a
number of region-specific abiotic and biotic indicators (listed in Table 1) not all of which are
relevant to all regions as indicated by the respective column of each region. The term ‘goal
achievement’ refers to indicator-related goal definitions, such as ‘prevention of nitrate
leaching’ or ‘enhancement of habitat quality for skylarks’. The closer the index value is to one
the higher is the assessed suitability of a particular agricultural management option to
contribute to goal achievement. Area-weighted aggregation of the IGAs allows for comparing
environmental impacts of the policy scenarios among land capability classes, farms or
regions.
308
Session B5: Integrated assessment methods for regional and landscape levels
Table 1. Environmental indicators implemented in the MEA-Scope tool.
Abiotic
Biotic
Abbrev.
NO3
NP
Pest
GWR
WaEro
Amph
Sky
Hare
Hover
Flora
Indicator
Nitrate entry into groundwater
Nutrient (N/P) entry into surface water
Pesticide entry into ground and surface water
Groundwater recharge/proliferation
Water erosion
Red belly toad (amphibians)
Skylarks (field breeding birds)
Field hares (mammals)
Hover flies (beneficial insects)
Wild flora species (winter annuals)
DE
x
x
x
x
x
x
x
x
x
x
DK
x
IT
x
x
x
x
x
x
x
x
x
x
x
SK
x
x
x
x
x
x
Results and discussion
The same scenarios applied to different regions can lead to opposite or conflicting results with
regard to the environmental and economic performance of the farms in the regions. Although
decoupling direct payments, for example, led to a clear intensification on arable land in all
regions, this overall intensification was not necessarily accompanied by an increasing mean
farm income. In the no direct payments scenario, for example, 2nd pillar programs were of
particular relevance for biotic indicators in all regions, but participation in these programs
could only partly compensate occurring income losses from the 1st pillar of the CAP.
Spatial distribution analysis allows identifying winners and losers, where environmental hot
spots are likely to occur, and how the environmental performance of different farms (e.g., the
index of goal achievement for amphibians, Figure 1) changes over time under the different
CAP options.
The modelling results demonstrate how much the diversity of European regions matters for
practical policy implementation. An explicitly spatial focus as adopted in the MEA-Scope
tool is, therefore, argued to be crucial for future policy analysis.
18
16
14
12
10
8
6
4
2
0
-2
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
Index of Goal Achievement [IGA] for Amphibians
base year (w ith direct payments)
nine periods later (w ith direct payments)
nine periods later (w ithout direct payments)
Figure 1. Kernel density distribution of the Index of Goal Achievement for Amphibians
[0 to 1] among the farms of the German region Ostprignitz-Ruppin (n=585) with 0 indicating
lowest and 1 indicating maximum possible goal achievement.
References
Happe, K., et al., 2006. Agrarwirtschaft 55(5-6): 268-279.
Kjeldsen, C., et al., 2006. Danish Journal of Geography 106(2): 35-43.
Piorr, A., et al., 2008. Environ. Sci. Policy. (accepted)
Sattler, C., et al., 2006. IARGE 5(2-3): 309-325.
309
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessment of mitigation measures for air pollutants and greenhouse gases
from agricultural systems in Germany at high spatial resolution
S. Wagner, O. Beletskaya, U. Kummer, J. Theloke, E. Angenendt, R. Friedrich, J. Zeddies
Institute of Energy Economics and the Rational Use of Energy (IER),
Universität Stuttgart, 70565 Stuttgart, Germany
Contact: susanne.wagner@ier.uni-stuttgart.de
Introduction
Agriculture is one of the main sources of air pollutants (NH3, PM) and greenhouse gases
(CH4, N2O, CO2). As these emissions have adverse impacts on the environment and human
health, respective policies demand their reduction. A detailed analysis of emissions impacts,
mitigation measures and their costs at farm level is important, because mitigation measures
influence directly farm management and consequently its revenues. Due to the variability of
emissions, mitigation potentials and costs across regions, an application of different
modelling approaches to the regional level is also necessary. To further develop emissions
mitigation strategies, spatially resolved emission data help to identify highly affected areas
and regional key drivers. This paper presents a method for integrated assessment of emission
reduction measures in agricultural systems in Germany with response to the environment,
farm economics and agricultural management at high spatial resolution. Provided that data
requirements, i.e., statistical, technical and farm accountancy data, are met, this method is
applicable to other countries as well.
Methods
In a project devoted to mitigation of emissions
from German agriculture, an economicecological model (Economic Farm Emission
Model (EFEM)) and a model for high spatial
resolution of emissions (Emission Calculation
Model (ECM)) are combined for the
development of regionalized cost-efficient multipollutant mitigation strategies and the assessment
of agriculture’s contribution to emissions
compared to other sectors (see Figure 1). This
approach allows scaling up emissions from farm
to regional scale, and furthermore, to the national
level.
EFEM is targeted to optimize farm
managements of four farm types (arable, mixed,
fodder- and livestock-production farms) and to
assess the emission of air pollutants and
greenhouse gases from different agricultural
activities at different scales. Thereafter, an
assessment of political and technical strategies
will be carried out, and different simulations and
scenarios will be compared.
For EFEM three German regions are
considered: Brandenburg, Lower Saxony and
Baden-Württemberg. Activities such as animal
husbandry, crop production, fodder preparation,
manure management, land preparation and
310
Figure 1: Combination of EFEM
(Triebe, 2007) and ECM (Pregger et al.,
2007, adapted).
Session B5: Integrated assessment methods for regional and landscape levels
harvesting operations are built into EFEM as real or potential sources of different emissions.
Emission intensities, farm and regional capacities are compiled within the production module
with the help of linear programming. Emission factors have been either determined or taken
from studies about European agricultural particulate emission and adopted for EFEM.
Emissions data at county level resulting from EFEM will be integrated into ECM, which
allows the calculation of spatially resolved emission information, complemented by proxy
data such as animal numbers and agricultural area (Pregger et al., 2007). Emission data are
spatially resolved by intersection with the CORINE land use data which are available for
western Europe with a resolution of 250 m × 250 m (EEA, 1997), comprising land use
categories such as arable land and pastures. Emissions in the model areas are resolved in a
5 × 5 km grid. By parameterization, the results are extrapolated to the whole of Germany at a
10 × 10 km grid.
Results and discussion
An implementation of the EFEM-modelling approach allows selection of the most efficient
mitigation options, regarding technical and non-technical measures in particular under
consideration of the economic impacts to the recipients as well as other regional effects. The
effects of mitigation options for air pollution and greenhouse gas emissions from agricultural
activities in Germany will be assessed on the base of comparison of the reference year 2003
and scenario 2013. ECM provides high resolution emission data for the whole of Germany
for the reference situation and scenarios that can be depicted in maps. They allow identifying
regions with higher emission concentrations and therefore developing regionalized mitigation
strategies. ECM also calculates and visualizes contribution of different agricultural activities
to total emissions. The quality of the high resolution emission data is determined by the
accuracy of input data. Proxy data for spatial allocation can be obtained from official
statistics; CORINE land use data are the best available for the time being. A literature review
on emission factors evidenced, however, that, especially for particulate matter, more
measured data are required. The combination of EFEM and ECM results in an improved
German emission inventory. A validation of the emission inventory is considered necessary;
therefore, validation experiments, i.e. comparisons of model results with measurements, are
essential.
References
EEA, 1997. ETC Landcover Data. European Environment Agency, European Topic Centre on
Land Cover, Kiruna.
Pregger, T., et al., 2007. Documentation of the anthropogenic GHG emission data for Europe
provided in the frame of CarboEurope GHG and CarboEurope IP. Institute of Energy
Economics and the Rational Use of Energy (IER), Universität Stuttgart.
Triebe, S., 2007. Möglichkeiten zur Verminderung von Treibhausgasen aus der
Landwirtschaft in der Bundesländern Brandenburg und Niedersachsen. EUL Verlag, Köln.
311
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Regionally optimized animal farm manure transports in an area with high
intensity animal farming systems
Sylvia Warnecke, Markus Biberacher, Gabriele Broll
ISPA, Hochschule Vechta, P.O. Box 1553, 49364 Vechta, Germany
Contact: swarnecke@ispa.uni-vechta.de
Introduction
Within the past decades, the north-western part of the county Lower Saxony, Germany, has
emerged as a European centre of high intensity animal agriculture. In several districts of
Lower Saxony, which comprises 2.9 million hectare of agricultural land, nutrients from
animal farm manure are in surplus to what can be sustainably used as fertilizer on the existing
farm land. Long-term over-application of animal farm manure in those districts resulted in
nutrient enrichments in soils and waters (e.g., Leinweber et al., 1993). The German Fertilizer
Ordinance (DüV, 2006), implementing the European Nitrates Directive, regulates fertilizer
use on the farm level and defines a threshold limit for the application of animal farm manures.
This results in an export of the excess manure into districts with lower animal densities
(Warnecke et al., 2008). Logistics and transportation are costly, especially for manures with
high water contents. Thus, this study aims at optimizing the flows of animal farm manure
between the districts in Lower Saxony to avoid over-application in the surplus districts and at
the same time make use of the fertilizer value in districts with lower animal densities. A linear
optimization model is used for the integrated assessment of (a) the impact of national policies
on the extend of animal farm manure transports in Lower Saxony and (b) the impact of farm
level decisions and techniques on these transports (e.g., systems altering the composition of
the manure).
Methods
On basis of Biberacher (2007) and Biberacher et al. (2008) a linear optimization model is
established to find an optimal regional (potentially supra-district) usage of the existing animal
farm manures on the available agricultural acreage (model input) while complying with legal
and environmental standards and considering transport distances and market prices for
nutrients (model constraints). With special focus on type (dung or slurry) and composition
(content of N, P, K and water) of the manures (determined as described in Warnecke et al.,
2008), the model output is the overall minimal transport effort for manures at the lowest
transportation costs (Figure 1). The modelling language is GAMS. The model is linked to an
Excel template serving as an interface for data input and result visualization.
District 1
Input
• Number of
farm animals
• Factors determining
manure composition
• Agricultural acreage
Constraints
• legal
• economic
• ecologic
District 2
District 3
Figure 1. Model framework and workflow.
312
Output
Minimal manure
transport between
districts in an
optimized case
Session B5: Integrated assessment methods for regional and landscape levels
Special emphasis is dedicated to “if ... then” analyses to show the impacts of changing
constraints on the overall optimal solution. Hence, the model allows the definition of a
comprehensive and integrative pool for farm manure management options that are available
on the different scales, serving as information to the individual decision makers.
Results and discussion
Initial calculations have been done for the 45 districts of Lower Saxony for the year 2008.
The basic scenario’s constraint assumption is the application limit of animal farm manures
(170 kg N ha–1 yr–1) as defined by the German Fertilizer Ordinance’s regulations (DüV,
2006). If a maximal amount of animal farm manure equivalent to 170 kg N ha–1 yr–1 is applied
to the available agricultural acreage of the individual districts, then almost 340,000 t of
poultry dung have to be exported from two districts to five adjacent districts (Table 1). Hence,
the nutrients in water rich slurries can be completely used as fertilizers in the districts of their
origin.
Table 1. Animal farm manure transports between districts in an optimized case.
From district
To district
Manure category
[tons yr–1]
Vechta
Diepholz
poultry dung
165,673
Cloppenburg
Oldenburg district
poultry dung
69,526
Cloppenburg
Emsland
poultry dung
56,493
Cloppenburg
Ammerland
poultry dung
44,156
Cloppenburg
Oldenburg city
poultry dung
3,640
These initial results are subject to change with the following additional input parameters and
constraints which are being incorporated into the model: As biogas production plays an
important role in Lower Saxony data on the amount of nutrients that enter the agricultural
material flows with the renewable raw materials are added to the input data. The digestates
from the fermentation process are, just like the slurries, costly to transport. They compete for
agricultural acreage close to their location of production. For this reason, further on-farm
transformation processes are included into the model, e.g., slurry or digestate treatment
techniques that alter the nutrient:water ratio. Additionally, further legal, economic and
environmental constraints will be considered.
In the end, specific scenarios are supposed to show the most relevant impacting parameters
and hence options for regulating regional animal farm manure and digestate flows most
efficiently. As spatial units – as opposed to individual farms – are analysed in this study, the
results will be of particular relevance for policymakers and extension work.
References
Biberacher, M., 2007. Modelling and optimisation of future energy systems. Using spatial and
temporal methods. VDM Verlag Dr. Müller. ISBN-13: 9783836424356
Biberacher, M., et al., 2008. In: Eds M. Sànchez-Marrè, et al., Proceedings of the iEMSs
Fourth Biennal Meeting. S. 1306-1313.
DüV, 2006. Verordnung über die Anwendung von Düngemitteln, Bodenhilfsstoffen,
Kultursubstraten und Pflanzenhilfsmitteln nach den Grundsätzen der guten fachlichen
Praxis beim Düngen (Düngeverordnung - DüV).
Leinweber, P., et al., 1993. Phosphorversorgung der Böden im agrarischen Intensivgebiet
Südoldenburg. Vechtaer Druckerei und Verlag GmbH, 62 pp.
Warnecke, S., et al., 2008. In: IAMO Forum 2008 “Agri-Food Business: Global Challenges –
Innovative Solutions”, Proceedings CD. Online: http://www.iamo.de/uploads/
media/5b.1_Warnecke_CD.pdf
313
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme B
State-of-the-art of components for integrated
systems
Session B6: Software infrastructures and tools for
integrated assessment
Session organizers:
Robert Argent
Bureau of Meteorology, Canberra, Australia
Jan Erik Wien
Alterra, Wageningen UR, The Netherlands
Andrea E. Rizzoli
IDSIA, Lugano, Switzerland
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated modelling tools, mass collaboration and the opportunities ahead1
R.V. Moore, J.J.F. Wien
NERC Natural Environment Research Council – Centre for Ecology & Hydrology,
Wallingford, Oxfordshire, OX10 8BB, United Kingdom
Contact: rvm@ceh.ac.uk
Introduction
Some modellers have long dreamed of a time when linking models would be as simple as
drag and drop. When it happened, they would be able to explore and predict the wider
consequences of events and may be foresee, for example, that the move to biofuels could lead
to food shortages. The barriers to progress have been technical and cultural. ICT advances
have now swept away enough of the technical problems to make linking operationally
feasible, though many challenging issues remain to be addressed. However, it is the new
internet and the cultural changes following in its wake that may well open up opportunities
beyond the original dream. This talk will summarize the driving forces behind integrated
modelling and then speculate on where they may take us.
Driving forces behind the development of integrated modelling tools
At least three forces are driving integrated modelling: events, policy and industry. Events,
such as climate change, affect almost every aspect of life. To understand those effects and
plan our response requires not just the ability to model the individual environmental, social
and economic processes but also their interactions. With respect to policy, the impact
assessment (IA) guidelines used by European Commission officials emphasize the importance
of achieving an integrated form of assessment, covering social, environmental and economic
impacts jointly (EC, 2005). This requires consideration of the wider implications of measures
before they are implemented, leading to a need for integrated modelling and hence integrated
modelling tools. In this context, industry covers the model and software developers, the consultants, who use the modelling tools, and their clients, often governments or their agencies.
Amongst the developers, the driving forces are the need to make model linkage simpler and
hence cheaper, combined with their need to grasp the opportunities for consultancy created by
the events and policies. For the end users the driver is the need to implement integrated
management. Integrated management requires that the tools should not only predict the
response of an individual process such as a river during flood conditions, but be able to
predict how many interacting processes will respond to different management policies.
Integrated modelling at present
The introduction of standard model linking interfaces, such as the OpenMI (Gijsbers et al.,
2002), and frameworks for component based model integration, like SEAMLESS-IF (Van
Ittersum et al., 2008) has opened up enormous opportunities for exploring how processes
interact. Now models from any discipline can, if complying with the standards, be linked with
increased efficiency, reduced costs and risks, leading to more easily re-used tools. It is,
therefore, easier to consider questions such as ‘What will be the impact of climate change on
ecosystem services?’ Already developers are creating tools that put the whole model
integration and run process with in the range of a much wider user community, the first
example of such a toolset being OpenWeb2.
One of the practical barriers to collaboration on integrated modelling between European
competent authorities relates to the licence issues surrounding data and models. When it
1
Keynote presentation
2
http://www.hrwallingford.co.uk/index.aspx?activities=company&topic=Research&facets=company&subtopic=OpenWEB
316
Session B6: Software infrastructures and tools for integrated assessment
becomes practical to link models across networks and the models to be linked no longer have
to be located on one machine, then many of the real and imaginary barriers to
interdepartmental collaboration caused by data and model ownership issues come down.
Some of the present difficulties should be removed by version 2.0 of the OpenMI in 2009/10.
An important point to note about the driving forces is that they emerge from problems whose
resolution requires knowledge from several disciplines. Often, more knowledge is needed
than any individual or organisation can possess. The creation of standards and frameworks
like the OpenMI and SEAMLESS-IF is also beyond the intellectual and financial resources of
most individual organisations. Collaboration is, therefore, not merely desirable but essential.
Although, the OpenMI has had substantial funding to date, this situation is unlikely to
continue. A community of practice therefore needs to be built around the OpenMI, which has
sufficient interest in the future of the OpenMI to sustain it with ideas and so maintain its
relevance. There is now evidence to believe this is happening across Europe and in the US.
Individuals are working on ways of extending the OpenMI’s usefulness, and feeding the
results back to the OpenMI Technical Committee. It is the task for the OpenMI to learn how
to sustain this virtuous circle.
Challenges and opportunities ahead
The new internet will make global communication very much easier for both people and
computers. Access to knowledge will become far more widespread. It is therefore possible to
consider IA tools based, for example, on Twitter services (http://twitter.com/), a free
networking and micro-blogging service that allows its users to send and read other users'
information. The combined and shared knowledge of large groups of experts and
stakeholders, in this way, could be used for massive collaborative assessments and solutions.
Wikis and other collaborative platforms, have proved to be successful when based on the four
principles openness, peering, sharing, and acting globally (Tapscott & Williams, 2006), as are
interactive virtual worlds (e.g., Second Life, http://secondlife.com/) where users, for example
residents in an area, can interact with each other through avatars. Residents can explore, meet
other residents, socialize, participate and cooperate in individual and group activities, possibly
solving environmental problems. Combined with integrated models these environments could
enrich IA’s with visual impact assessments that show impacts of policies or events in virtual
but realistic future worlds. Stakeholders could participate by testing strategies and building a
better understanding of the aspects of the real world which the virtual world depicts (Wien &
Van der Wal, 2004).
There are still many challenges ahead, e.g. on how a viable collaborating community is to
be created that will enable the opportunities opened up by integrated modelling to be
exploited for the benefit of society. Will it happen spontaneously or does its creation require
positive action by those with an interest in it happening? Will mass collaboration enabled by
Web 2.0 technologies be the critical success factor for the future of integrated modelling?
References
European Commission, 2005. Impact Assessment Guidelines. Brussels.
Gijsbers, P.J.A., et al., 2002. Towards OMI, an Open Modeling Interface and Environment to
harmonise European developments in water related simulation software. Proceedings of
Hydroinformatics, Cardiff. IWA Publishing, London, pp. 1268-1275.
Tapscott, D. & A.D. Williams, 2006. Wikinomics. How mass collaboration changes
everything. Atlantic Books, London.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
Wien, J.J.F. & T. Van der Wal, 2004. Real learning in virtual worlds; the essence of playing
games, proceedings of the 8th international workshop on experimental learning in chains
and networks, Wageningen, The Netherlands.
317
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The Trade-off Analysis software:
Modelling tools to support informed policy decision making
J.M. Antle, J.J. Stoorvogel, R.O. Valdivia
Montana State U., Bozeman, MT 59717 USA
Contact: Jantle@montana.edu
Introduction
Trade-off Analysis (TOA) is a participatory approach to integrated assessment of agricultural
systems designed to bring together policy decision makers, stakeholders, and a research team
to provide quantitative assessment of policy options (Stoorvogel et al., 2004). The approach is
based on the identification of quantifiable sustainability indicators and alternative policy and
technology scenarios. Trade-offs are defined as the inter-relationships between the
sustainability indicators observed as key parameters, such as the prices farmers face, are
varied.
The Trade-off Analysis software was developed as a tool to implement analysis of the
trade-offs among key environmental and economic indicators, and how those trade-offs
change with technology and policy scenarios identified by stakeholders. The TOA software
integrates spatially referenced data and disciplinary models to simulate agricultural systems
on a site-specific basis and then aggregate results to the level of spatial units (such as
watersheds, political units) relevant for policy analysis. The TOA approach has been used to
study various issues including technology transfer, poverty, food security, climate change,
and ecosystem services with applications in Ecuador, Peru, Ghana, Kenya, Nepal, The
Netherlands, Panama, Senegal, Uganda and United States.
Methodology
The TOA software (see below) has a modular structure designed to ‘loosely couple’
disciplinary models to estimate and simulate an agricultural system under alternative
technology and policy scenarios. In the estimation step, the software implements the DSSAT
crop models to simulate ‘inherent productivity’ of crops on a site-specific basis using
spatially-referenced soil and climate data. The inherent productivities are used with field-level
and farm-level data to estimate systems of output supply and input demand equations for each
activity in the system. These equations are used to construct an econometric-process
simulation model that determines land management as a function of economic variables and
inherent productivities. In the simulation step, the user samples a set of sites from the region
to be simulated, specifies the technology or policy scenario, and defines the ‘trade-off
variable’ that the model uses to generate changes in behavior that give rise to trade-offs in the
system (e.g., crop or input prices). The econometric-process model is then simulated, and
results are passed to environmental process models. Finally, the economic and environmental
outcomes at each sample point can then be aggregated and displayed graphically, or saved for
analysis and display using spatial analysis and other data analysis tools.
Recent extensions of the TOA software have been made to address extensions of the basic
framework. First, feedbacks from environmental changes (e.g., soil erosion) to crop
productivity and management have been incorporated. Second, a market equilibrium model
has been linked to the TOA system, so that users can solve for the market equilibrium prices
that would result from introduction of technology and policy scenarios.
The analysis of agricultural systems using spatially referenced data and highly detailed
disciplinary models requires a large amount of high-quality data. Often such models are more
detailed than needed to provide policy guidance. An alternative approach, referred to as a
minimum-data (MD) approach by Antle & Valdivia (2006), is to use data that characterize the
318
Session B6: Software infrastructures and tools for integrated assessment
spatial distributions of relevant economic and environmental variables within an ecozone,
together with relatively simple economic decision models. The MD approach has been
utilized in a number of recent studies of ecosystem service supply, and has been adopted for
new projects on climate change and desertification in North Africa and East Africa.
Using the TOA modelling tools
The TOA software, documentation, and several example applications are publicly available at
www.tradeoffs.nl. However, users should be aware that its use requires knowledge of the
disciplinary models and data – a team with appropriate training is recommended. The TOAMD model has been implemented with software and documentation available on the internet
at www.tradeoffs.montana.edu. The TOA team provides training periodically to interested
users through courses and web-based training.
Figure 1. Trade-off Analysis and Minimum Data software.
References
Antle, J.M. & R.O. Valdivia, 2006. Australian Journal of Agricultural and Resource
Economics 50(1): 1-15.
Stoorvogel, J.J., et al., 2004. Agricultural Systems 80(1): 43-66.
319
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Granular ontologies for integrated assessment of agricultural systems
Ioannis N. Athanasiadis, S. Janssen, E. Andersen, A.E. Rizzoli, J.J.F. Wien
Dalle Molle Institute for Artificial Intelligence, Lugano, Switzerland
Contact: ioannis@idsia.ch
Introduction
The key features of integrated assessment studies are interdisciplinary research, participatory
processes involving stakeholders, and knowledge integration. Integrated assessment studies of
agricultural systems need to meet such objectives. This paper introduces a remedy to tackle
the complexity of agricultural modelling and data management, by utilizing a set of
ontologies for domain modelling and management of agricultural knowledge.
An ontology consists of a finite list of concepts and the relationships among these concepts
(Antoniou & Van Harmelen, 2004). Typically, an ontology is specified in a formal language,
which is understandable by computers, as the Web Ontology Language (OWL) (McGuinnes
& Van Harmelen, 2004). An ontology captures the concepts involved in a certain domain and
their relationships. A shared ontology is a collaborative product by a group of researchers,
which captures the common understanding of the domain, as it perceived by a community.
A shared ontology in the case of an integrated assessment study is required to take under
account different scales, dimensions, databases, methodologies, models, indicators, typologies
and stakeholders involved. Such an effort requires an explicit procedure for building the
shared ontology, while avoiding getting confused by the project complexity. In this work, we
describe the content and structure of a set of shared ontologies that we developed as part of
the SEAMLESS-IP project (http://www.seamless-ip.org).
Case study
The SEAMLESS project (System for Environmental and Agricultural Modelling: Linking
European Science and Society) (Van Ittersum et al., 2008) develops a computerized and
integrated framework (SEAMLESS-IF) to assess the impacts on environmental and economic
sustainability of a wide range of policies and technological improvements across a number of
scales. In SEAMLESS-IF, different types of models and indicators are linked into model
chains, where each model uses the outputs of another model as its inputs and ultimately
indicators are calculated.
Methods
To arrive at a shared ontology that specifies the domain of models, indicators, scales,
dimensions and assessment questions in SEAMLESS-IF, a collaborative group (task-force)
was set up. The task force was formed by a core of three scientists with different
backgrounds, which were responsible to stimulate discussions, capture the domain
specifications, and translate them into an ontology written in OWL. In the task-force were
included for achieving certain goals, a community of dozens of domain-experts. Domain
experts (agronomists, economists, social scientists, and data experts) were involved in
different parts of the work, on activities related to their expertise, together with the core
group. The task-force adopted a collaborative approach (Holsapple & Joshi, 2002), in which
researchers formed work-groups to reach to a common understanding of a particular domain
and ultimately produced a shared ontology. An iterative prototyping approach was used to
develop versions of the shared ontology, which were subsequently reviewed by panels of
experts and further improved.
320
Session B6: Software infrastructures and tools for integrated assessment
Results and discussion
The collaboration of scientists involved in the task-force resulted to a shared ontology,
covering scales, models, indicators and dimensions relevant to the SEAMLESS project.
Instead of making one large ontology, spanning across different sub-domains of the project,
we developed eleven small ontologies, each one of which refers to a distinct aspect of the
project. Common concepts and relationships are shared across granular ontologies.
The content of the eleven small ontologies is:
1. Crop.owl fomalizes the domain related to crops, crop products and grouping of crops.
Several models refer to such concepts at different scales, and this ontology clarifies these
concepts.
2. Farm.owl describes concepts about farms and geographical regions, in which these farms
are found, along with soil and climate information.
3. Prodent.owl contains concepts related to rotations and choices in production made by
farmers (production enterprise).
4. Activity.owl gives an overview of the different types of activities (e.g., arable, livestock,
and perennial) farmers.
5. Livestock.owl specifies in more detail livestock activities.
6. Agrirule.owl specifies in more detail arable activities.
7. Capri.owl provides concepts related to the CAPRI model (Heckelei & Britz, 2001).
8. Farmopt.owl provides concepts related to farm economics and optimization of farm
income.
9. Indi.owl specifies concepts relevant to indicators and the use of indicators in integrated
assessment (Turpin et al., 2009).
10. Seamproj.owl describes SEAMLESS-specific scenarios and integrated assessment
problems.
11. Pica.owl specifies the analysis of institutional compatibility assessments.
There are two main advantages in defining smaller ontologies as part of the larger shared
ontology. One is that interactions between different domains become explicit. It becomes
apparent which are the concepts shared across sub-domains, which typically are the issues of
conflict in interdisciplinary research, and the critical points for software development and
interoperability of model components. Second, ontologies of finer granularity can be re-used
independently of the rest. They could also be extended or substituted by new (finer)
conceptualizations.
The SEAMLESS-IP ontologies
http://ontologies.seamless-ip.org.
are
available
for
download
and
re-use
on
References
Antoniou, G. & F. Van Harmelen, 2004. A Semantic Web Primer. The MIT Press, pp. 238.
Heckelei, T. & W. Britz, 2001. Concept and explorative application of an EU-wide regional
agricultural sector model (CAPRI-Project). 65th EAAE Seminar, pp. 281-290.
Holsapple, C.W. & K.D. Joshi, 2002. Communications of the ACM 45(2): 42-47.
McGuinness, D. & F. Van Harmelen, 2004. OWL Web Ontology Language Overview.
www.w3.org/TR/owl-features/.
Turpin, N., et al., 2009. AgSAP Conference 2009, these proceedings.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
321
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A software tool to support multi-criteria-based assessment of animal
welfare1
R. Botreau, I. Veissier, P. Champciaux, J.-P. Brun, A. Lamadon, P. Perny
INRA, UR1213 Herbivores, Site de Theix, 63122 Saint-Genès-Champanelle, France
Contact: rbotrea@clermont.inra.fr
Introduction
During the last decades, numerous trade groups (producers, processors, retailers, restaurant
chains…) have put in place certification schemes. Most of these schemes are meant to ensure
product quality and an increasing number of them takes into account other concerns like the
protection of the environment or animal welfare (e.g., IKB by the Dutch meat industry,
Swedish Broiler Control, or McDonald’s Europe). However, there is no common standard for
assessing animal welfare. One objective of the European project Welfare Quality® (2004–
2009) is to design an overall assessment of animal welfare of cattle, pigs and poultry on the
farm or at the slaughterhouse, and to turn this assessment into product information. This
information could be delivered as such to consumers or, alternatively, be used by stakeholders
(producers, retailers…) to certify products.
To assess all dimensions of welfare on an animal unit (health, behaviour, stress…),
numerous measures are to be collected on that unit. Then, the data need to be interpreted in
terms of welfare and integrated into an overall assessment. We designed a mathematical
model for such an assessment following a sequential multi-criteria process: four welfare
principles (good feeding, good housing, good health and appropriate behaviour), subdivided
into 12 criteria (e.g., absence of hunger and thirst for good feeding) are defined and assessed
on a 0-100 scale, and then aggregated by comparison to predefined reference profiles which
delimit four welfare categories (Excellent, Good, Basic, Not classified) at unit-level (Botreau
et al., 2009). This assessment model was tuned according to expert opinion and this resulted
in calculation of scores based on various functions. To facilitate the implementation of the
assessment system, we propose a software tool for data acquisition and processing.
Development of data acquisition and web-based software, and database
The objectives of the tool are to ease the acquisition of data and their storage, to automate the
scores and overall assessment calculations, to present a synthesis of results to users
(producers, retailers, consumers…), and to simulate potential improvements. The tool is
organized in four interconnected modules (Figure 1):
Module 1 facilitates the acquisition of data directly on farms or at slaughter. This software
is to be used by the assessor on a lap-top or a Tablet PC. In addition to data collection, this
module makes some basic calculations at unit level (e.g., % animals affected by a given
problem) and prepares files containing only necessary data at unit-level to be sent to
Module 2. We call such data ‘raw data’. These files (XML format) are exported to the
database by the assessor.
Module 2 is a database storing the data collected on farms and slaughterplants across years
(from Module 1) and their related scores calculated in Module 3. The database can be
consulted trough Module 4.
Module 3 is a calculation application importing raw data from Module 2, calculating
criterion- and principle-scores and overall welfare assessment, and exporting these to
Module 2. Module 3 may also be used by Module 4.
1
The present study is part of the Welfare Quality® research project which has been co-financed by the
European Commission, within the 6th Framework Programme, contract No. FOOD-CT-2004-506508. The text
represents the authors’ views and does not necessarily represent a position of the Commission who will not be
liable for the use made of such information.
322
Session B6: Software infrastructures and tools for integrated assessment
Module 4 is an interactive platform to enable dialog with stakeholders. This module
provides information on measures and the calculation of scores and overall assessment.
Figures are produced on the distribution of results among a given population (all animal
units, only those visited on one year, in one country etc.) while producers are given access
to their own data. Simulations are also possible, e.g., producers can simulate an
improvement or a decrease in one or several measures and see the consequences on their
scores and overall assessment. To execute a simulation, Module 4 may import data from
Module 2 or create a new set of measures to work on, and calculations are made by
Module 3.
Module 1
Acquisition of data
on farms / slaughterplants
Module 2
Storage of data
from farms / slaughterplants
Module 3
Calculation of scores
for farms / slaughterplants
Module 4
Interaction
on measures and scores
Data acquisition tool
Web-based tool
Figure 1. Software modules to ease the implementation of Welfare Quality® assessment
system.
Module 1 is a local application for data acquisition on farms or slaughterplants, whereas
Modules 2 to 4 correspond to a web customer-server application based on a MySQL database.
The support used for each application is presented in Table 1.
Table 1. Modules’ development.
Module
Function
Module 1
Data acquisition
Module 2
Database management
Module 3
Calculation of scores
Module 4
Interactions on results
*
PC SOFT products
Development tool
WinDev 12*
Database
HyperFile*
Support
PC
WAMP5
(PHP MySQL)
MySQL
Web server
Since different user-types may have access to the web-based software (producers, assessors,
certification bodies, researchers or mere visitors), several user-profiles are defined and have
restricted access.
Discussion
A prototype of the tool for assessing dairy cow welfare on farms is developed. It will be
extended to other animal types studied in Welfare Quality® (veal calves, fattening pigs…).
Overall assessment of animal welfare is by definition a complex exercise and support tools
are necessary for calculations to be used by stakeholders. We hope the present tool will help
stakeholders figure out the kind of results that can be obtained with the assessment system
designed in Welfare Quality®. It can show them how improvements on farms or at slaughter
are likely to affect the results of the assessment of these units. The tool shall also help the
taking up of the assessment system by easing the collection of data and making the
assessment more transparent (since all intermediate scores are accessible).
Reference
Botreau, R., et al., 2009. Foundations of Computing and Decision Sciences.
323
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The role of the Object Modelling System (OMS) for integrated assessments
of conservation on agricultural land in the United States
J. Carlson, O. David, J. Ascough, F. Geter, L. Ahuja
Agricultural Systems Research Unit, Agricultural Research Service,
U.S. Department of Agriculture, 2150 Centre Avenue, Building D,
Fort Collins, Colorado 80526, USA
Contact: Jack.Carlson@ars.usda.gov
Introduction
Operators of approximately 2.1 million farms, ranches, and small woodlands make daily
decisions to manage a combined 1.1 billion acres of private land, about 50% of the land base
of the United States. Federal and state conservation programs deliver technical and financial
assistance to help operators to sustain the health of the land to maintain its productivity over
the long term. Federal programs deliver about $3 billion annually to help defray the cost of
installing and maintaining practices contained in the operator’s conservation plan. These
practices reduce erosion, control nutrient and pesticide leaching and runoff, increase soil tilth,
prevent overgrazing, improve wildlife habitat, and provide other ecosystem services.
U.S. Department of Agriculture (USDA) agencies deliver conservation program services
through a network of approximately 3,000 offices at the state, area, and county levels. Field
conservationists work with farm operators to develop conservation plans, and the national
database system contains 1.4 million active plans covering 300 million acres. These plans
provide the basis for financial assistance through 200,000 active program contracts, also
maintained in the system.
Up to 35 thousand conservation plans are serviced each work day. About 5,000 field
conservationists use the system each week, and during peak periods, the system
accommodates as high as 2,000 concurrent sessions. Each night data is summarized and
processed to a data mart for daily reporting to agency managers on the progress with
delivering program services. Although the system can measure this progress, it cannot yet
adequately evaluate its impact on improving the health and sustainability of the land.
The 2002 Farm Bill established the Conservation Effects Assessment Project (CEAP),
which uses existing stand alone agro-environmental models for evaluating program
effectiveness. However, current efforts do not yet meet the CEAP vision of a consistent
national integrated assessment process.
Methods
CEAP analysts at multiple locations must be able to run calibrated and certified agroenvironmental models deployed to a centralized platform. Achieving an integrated assessment
infrastructure for CEAP involves (1) adopting a modelling framework; (2) establishing a high
performance model run-time platform; (3) creating an integrated assessment model base; and
(4) provisioning data for all areas in the national assessment. The analyses performed
examined recent developments in cloud computing, multi-threading support to on-line
commercial applications, lightweight non-invasive programming framework technology,
semantic web technology, service oriented architecture (SOA), and opportunities to leverage
existing resources and capabilities within participating agencies. The findings were applied to
define an improved integrated assessment infrastructure.
Results and discussion
In February 2008, USDA agencies adopted the Object Modelling System (OMS) framework
(Figure 1) to support integrated assessment, through formal agreement to manage it as part of
an approved technical architecture. OMS (Ahuja et al., 2005; David & Ahuja, 2006)
324
Session B6: Software infrastructures and tools for integrated assessment
developed in Java on the NetBeans platform contains a component builder, model builder,
repository, core framework services, data access and visualization tools. A Java Annotation
Specification is being added to remove model component dependency on the framework API.
Model components will be developed using Plain Old Java Objects (POJOs) and bound to the
framework at run time. Therefore, model components can be more easily adapted to other
frameworks, and their longevity expected to increase.
The OMS team has defined a run-time platform that enables multi-threading and deploying
models and components as services in an elastic computing cloud. The platform supports the
situation when many analysts across the country run a model concurrently and the situation of
the modeller spooling many model runs concurrently for sensitivity analysis.
The CEAP strategy has established the concept of a model base, containing related models
and components. The integrated assessment model base contains core components from J2Ks,
a combination of the J2000 and SWAT models, with components added from other models as
requirements evolve. The architecture supports the CEAP analyst running models tailored to
particular scenarios and region. A knowledge base will help ensure consistency in
assessments across regions.
The OMS team is defining a data provisioning architecture, featuring a data mart with web
services designed to support the requirements of the model base. Data stewardship
responsibility will be organized by integrated assessment regions.
Figure 1. High-level OMS architecture.
References
Ahuja, L., et al., 2005. Advances in Geosciences 4: 29-36.
David, O. & L. Ahuja, 2006. The Object Modeling System – A Modeling Platform. Proc. 3rd
Federal Interagency Modeling Conference, Reno, Nevada, 33 pp.
325
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
OptimiSTICS, a software for STICS crop model parameter estimation and
evaluation which was build up to function with other dynamical models
E. Justes, P. Lecharpentier, S. Buis, M. Launay, J. Bourges, J. Soudais, P. Clastre,
J.E. Bergez, P. Debaeke, D. Wallach, M. Guérif
INRA, UMR INRA/INP-ENSAT 1248 AGIR,
31326 Castanet-Tolosan Cedex, France
Contact: Eric.Justes@toulouse.inra.fr
Introduction
Because of the inherent complexity of crop models, it is important to have a range of software
tools for automating the runs of such models, analysing their behaviour, evaluating results and
estimating parameters. The calibration of soil-crop models in particular is a difficult task
which requires both the development of procedures adapted to the specificities of models and
the development of software tools to implement those procedures. The purpose of this paper
is to describe a package for model calibration and evaluation with a focus on OptimiSTICS,
which has been implemented to work with the STICS soil-crop model (Brisson et al., 1998),
but was develop to function with other dynamic models in the OptiLib software version.
Methods
Optimistics is a software or a data-processing utility which makes it possible to automate the
optimization of the parameters of the STICS model. This tool makes part of a library of
functions developed for model STICS but which are adaptable to other models. The software
package is written in the MATLAB© mathematical programming language. The package
further contains a series of tools such as: (i) full factorial design where the factors can be
parameters or input variables (MultiSimLib), (ii) global sensitivity analysis using the EFAST
method (SensiLib), (iii) evaluation tools to compare observed versus simulated data (OptiLib
and OptimiSTICS tools). Thus, given a set of site-year-management combinations and a set of
observed data, OptimiSTICS runs the model and, based on the comparison of observed and
calculated results, calculates mean difference, mean squared error, the decomposition of mean
squared error into systematic and unsystematic contributions, modelling efficiency, linear
regression between calculated and observed. It also automatically creates various types of
figures: (i) measured vs simulated data for all SU and for each SU in dynamics, (ii) residuals
according to observed data, (iii) evolution of convergence criteria during optimization, …
OptimiSTICS allows one to estimate parameter values in stages or steps, with different
parameters and different observed data used at each stage (Figure 1).
In each stage, the software
allows one to designate groups
of SU (corresponding to
genotypes,
or
soils,
or
managements …) and groups of
parameters that will have some
specific
values
differing
between groups of SU and
others as applying to the crop
species in general. A stage may
involve only a single type of
output or it may involve several
types of output. If only a single
type of output is involved,
Figure 1. Algorithm diagram of iterative process.
ordinary least squares is used.
326
Session B6: Software infrastructures and tools for integrated assessment
This procedure is automated and based on a description of the stages by the user. The
advantage of this approach is that each set of parameters is adjusted using only data that it
affects strongly. Furthermore, one can order the stages so that one first adjusts parameters to
state variables that strongly affect other state variables. The goal is to reduce the risk of
obtaining parameter values not ‘actual processes based’ due to compensation of errors.
There is an option to first do a log transformation of the data, which is appropriate when
model error is likely to be approximately proportional to the size of the response. When
several data types are used, the error for each data type should be weighted by the inverse of
the standard deviation of model error for that data type. To face the problem of possible
correlations between errors, the software provides two options. In the first, all errors are
treated as independent. This tends to give more importance to data types and plots with many
measurement dates. In the second option, an average squared measurement error is calculated
for each measurement type in each plot. That average error is treated as though it were the
single error for that data type and plot. In this case, each data type in each plot contributes just
one term to the overall squared error, regardless of the number of measurement dates. For
each option, the weight for each data type can be calculated automatically, using the method
of concentrated likelihood, or provided by the user. Once again, there is an option of log
transformation for each data type. The simplex algorithm is used for searching for the
parameter values that minimize the ordinary or weighted least squares criteria. The
‘fminsearchbnd’ function of Matlab© (D’Errico, 2005) was implemented. One can
automatically request several runs with different starting points, to improve the chances of
converging to a global minimum.
One further option, which has been programmed but not yet thoroughly tested, is a
Bayesian parameter estimation approach. An importance sampling algorithm is used here
(Guérif et al., 2006). The user specifies the prior distribution and the form of the likelihood
and the algorithm then calculates a weighted sample which approximates the posterior
distribution.
Results and discussion
The parameter estimation tools of the software package are quite recent, but we can already
draw some conclusions from our initial studies. OptimiSTICS allows one to implement
complex parameter estimation procedures that would be extremely time-consuming to carry
out otherwise. This has allowed us to use our data set to estimate parameter values for STICS
(last 6.2 version) for several varieties of maize, winter wheat, durum wheat, sunflower,
sorghum, spring pea and soybean.
The parameter estimation package allows one to fully automate the stage-by-stage
estimation of parameters. However, we have found that initially it is important to work with
one stage at a time, in order to evaluate the results at each stage.
The parameter estimation software allows one to estimate model parameters in several
different ways. One can choose from two different ways of weighting different measurements,
one can input weights or calculate them automatically, one can log transform the data or not,
one can define the stages of parameter estimation in different ways, etc. Then more than a
software package to estimate models’ parameters and evaluate models’ efficiencies,
OptimiSTICS tool will also have a methodological role in testing and comparing different
approaches to parameter estimation.
References
Brisson, N., et al., 1998. Agronomie 18: 311-346.
D’Errico, J., 2005. http://www.mathworks.com/matlabcentral/fileexchange/8277.
Guérif, M., et al., 2006. In: Eds D. Wallach et al., Working with Dynamic Crop Models.
Elsevier.
327
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
ProST: A process support tool for supporting water stress mitigation
through a participatory approach
Ayalew Kassahun
Applied Information Technology Group, Wageningen University,
Hollandseweg 1, 6706 KN, Wageningen, The Netherlands
Contact: ayalew.kassahun@wur.nl
Introduction
Participatory approach to solving water management problems is now widely recognized as
essential because the solutions affect many stakeholders and a variety of policy areas. In such
a process achieving common understanding, trust and confidence is challenging. It is essential
to make the process explicit so that all participants have clear and realistic expectations and
leave audit trail to ensure transparency. Pahl-Wostl & Hare (2004) found that when
stakeholders don’t know what is going to happen next or lose track of the goals of the whole
project, trust between them and the researchers is lost. To achieve a broadly accepted water
management plan they suggest that the project goals, roles of participants and the activities of
the process should be discussed and made explicit. ProST, a Process Support Tool, built in the
context of the AquaStress Project addresses this problem. ProST is part of a suite of loosely
integrated tools made available to the user community by the AquaStress project (Kassahun et
al., 2008). ProST along with a knowledge base system and a web portal constitute the main
part of I3S.
Methods
A participatory process for water management can be viewed as a collection of interrelated
tasks done by a team of researchers and stakeholders to solve a particular problem. Currently
available tools to define and manage processes can be roughly classified in two groups:
project management and workflow management tools. Project management tools are used to
manage a one-off and mostly long-lived processes and workflow management tools are used
to manage repetitive and mainly short-lived processes. However, in the field of water
management such tools are rarely used in practice.
A process support tool for supporting collaboration and participatory problem solving
should incorporate the necessary features from both project and workflow management
systems. Workflow management follows a life cycle roughly comprising of three steps: (i)
design & model, (ii) operational implementation and (iii) analyse & improve (Hollingsworth,
2004). In conventional workflow management, a project manager or a business analyst
designs a process. Thereafter the process is implemented, meaning the process is executed
many number of times. Tasks are dispatched to individual workers, while managers monitor
the progress of work and make the necessary decisions. While following the well tested
workflow life cycle can be beneficial for ProST the emphasis for support is mainly on
enabling stakeholders to participate actively in the planning and the execution of work.
Like a project management system ProST should support long-lived processes like water
stress mitigation projects. Project Management Institute considers projects as a one-off
process (PMI, 2000). Unlike conventional project management, participatory processes for
water stress mitigation projects are not purely one-off processes but one-of-a-kind processes
with recurring elements, which means project teams can reuse parts of water stress mitigation
project plans that previously were successfully applied elsewhere. In AquaStress, a default
and reusable participatory process for water stress mitigation is being defined based on
experiences from AquaStress test cases.
328
Session B6: Software infrastructures and tools for integrated assessment
Results
A process support tool ProST was developed that provides process support in three phases:
process definition, project set-up and project execution as shown in Figure 1. In the process
definition phase members of project team collaboratively plan a water stress mitigation
project. The plan describes the tasks to be done and any relevant information that can be of
use in executing the tasks. During project set-up the project leader(s) imports the process
definition for their site or case study from the knowledge base into ProST and initialize their
project by enlisting users, selecting tasks and setting-up authorization settings.
The main function of ProST is in the project execution phase. Once the project is set up,
individual members of the team use ProST to guide them in carrying out their tasks based on
their authorization. An important aspect of ProST is that it not only allows users to look
forward and know what is going to happen next, but also to look backward and see what has
been done by themselves and other team members. A flexible reporting tool enables users and
project managers to generate reports and analyse the progress of the project.
process definition
Knowledge base (KB) of participatory processes
Team
Teamdefines
definesaa
process
Project
team
process
define
process
process definition
transfer process definition from KB to ProST
project set-up
Process Support Tool (ProST)
Project leader(s)
set-up &
launch
project journal
transfer project journal to a shared project server
project execution
Project Server
shared project
journal
ProST
ProST
change project
setup, monitor
Project leader(s)
ProST
get guidance, store &
share results
Project member
get guidance, store &
share results
Project member
Figure 1. The steps of supporting participatory water stress mitigation project.
References
Hollingsworth, D., 2004. Workflow Management Coalition: 295-312.
Kassahun, A., et al., 2008. Integrated solution support systems for water management. iEMSs
2008, International Congress on Environmental Modelling and Software Barcelona.
Pahl-Wostl, C. & M. Hare, 2004. Journal of Community & Applied Social Psychology 14(3):
193-206.
PMI, 2000. Guide to the Project Management Body of Knowledge (PMBoK® guide)
Pennsylvania, USA, Project Management Institute, Inc.
329
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Use of OpenMI in SEAMLESS
M.J. Rob Knapen, Ioannis N. Athanasiadis, Benny Jonsson, David Huber, J.J.F. Wien,
Andrea E. Rizzoli, Sander Janssen
Group Systems, Alterra, Wageningen UR, P.O. Box 47, 6700 AA Wageningen
Contact: Rob.knapen@wur.nl
Introduction
Integrated Assessment (IA) is one option for ex-ante analysis of the impacts of policy changes
and technological innovations on agriculture. Integrated Assessment and Modelling (IAM;
Parker et al., 2002) is based on quantitative analysis involving the use of different modelling
tools. This paper focuses on an IAM approach, in which comprehensive models are linked
into model chains. When models are executed in a model chain at run time, they need to
exchange data. The Open Modelling Interface and Environment (OpenMI; Moore & Tindall,
2005) has been proposed as a standard to facilitate the exchange of data at run time between
models.
SEAMLESS, System for Environmental and Agricultural Modelling: Linking Science and
Society (SEAMLESS; Van Ittersum et al., 2008) is an IAM research project, in which a
linking is achieved between a set of models from biophysical and economic disciplines. In
SEAMLESS the OpenMI 1.4 standard was adopted for linking models at run time through
exchange items. As OpenMI is originally targeted at the hydrological domain, focusing on
models that only exchange time series of floating point values, it needed some extension to
support richer data types and other types of simulation engines before it could be applied to
the biophysical and economic models in SEAMLESS. This paper describes the adapted
version of OpenMI we developed for SEAMLESS and discusses the use of OpenMI for
model coupling in this context.
Methods
In OpenMI 1.4 (OpenMI, 2009) each model should have an interface and be wrapped as a
linkable component. Each linkable component has zero or more input exchange items and
zero or more output exchange items, which describe the data the model requires to run and the
data it produces. The input exchange items can be derived from other models or from data
sources, while the output exchange items give what the model produces. Each linkable
component has a getValues-method (Moore & Tindall, 2005) which is called by another
linkable component as a request to get input data for a specified simulation time and
geographic region. This results in a pull-based approach in which one linkable component
pulls data from another linkable component.
Within SEAMLESS, existing and new models are wrapped to make them OpenMI
compliant (so that they provide the standard OpenMI API for model components) and then
linked into a model chain. Special components – the triggers – are also added to the chain and
are eventually used to start the pull-based calculation. Following the links in the chain each
model is initiated to perform its run, requesting other models for data first when this is
needed. SEAMLESS uses the Java-version of OpenMI as the SEAMLESS framework is fully
build in Java, however the wrapped models are written not only in Java but also in GAMS
(General Algebraic Modelling System, 2009) and C#.
Results and discussion
For SEAMLESS an adaptation software layer had to be developed to make the original
OpenMI 1.4 suitable for use in the project. In its core it is build around the exchange of time
series of floating point typed quantitative data, which is enough to work with hydrological
models. The models used in SEAMLESS are from the economic and biophysical domain and
330
Session B6: Software infrastructures and tools for integrated assessment
are not operating on time series of floating point numbers but use complex data structures as
their input and output exchange items (e.g., a Crop, a Farm, or a EU Member State). An
ontology is used to define concepts for these data structures, and models exchange instances
of those concepts. Accordingly most of the adaptations in the SEAMLESS OpenMI Layer are
used to facilitate model components that exchange concepts. Within this layer the most
important classes are (see Figure 1):
•
•
•
•
SeamChain: A convenience class for building SEAMLESS model chains and
executing them. It is based on standard OpenMI classes.
SeamLinkableComponent: An extended version of the LinkableComponent of
OpenMI. It defines input and output exchange items that support the use of the
concepts defined in the ontology.
SeamTrigger: An specific version of the OpenMI Trigger class. It requires as input a
Problem concept, which describes the research problem and a IIndicatorValue
concept, which describe the results of a model run.
SeamExchangeItem: Base class for SEAMLESS exchange items that support the use
of concepts defined in the ontology. Subclasses for input (concept needed by the
model) and output (concept produced by the model) exchange items are implemented.
Figure 1. Simplified UML class diagram, which illustrates the SeamChain, SeamTrigger and
SeamLinkableComponent and their relation with the OpenMI Standard.
The described OpenMI extensions already are contributed to the OpenMI Association (as
open source) and are under discussion for inclusion in the OpenMI 2.0 standard that is under
development. This will increase the applicability of OpenMI for domains outside of the
hydrological domain.
References
General Algebraic Modeling System, 2009. www.gams.com.
Moore, R.V. & C.I. Tindall, 2005. Env. Science & Policy 8(3): 279-286.
OpenMI, 2009. www.openmi.org.
Parker, P., et al., 2002. Env. Mod. & Soft. 17(3): 209.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
331
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A graphical user interface for a generic bio-economic farm model (FSSIM)
Eelco Meuter, Sander Janssen, Martin K. Van Ittersum
Plant Production Systems Group, WU, P.O. Box 430,
6700 AK Wageningen, The Netherlands
Contact: eelco.meuter@wur.nl
Introduction
The Farm Systems SIMulator (FSSIM; Van Ittersum et al., 2008; Janssen et al., 2009) is a
Bio-Economic Farm Model (BEFM) that is designed as a component based model. It is to be
used for the range of agri-environmental conditions in the European Union, the farm types
that occur in these conditions and a broad range of innovation and policy questions that may
be relevant to those farm types. A bio-economic farm model links formulations describing
farmers’ resource management decisions to formulations that describe current and alternative
production possibilities in terms of required inputs to achieve certain outputs and associated
externalities. Janssen & Van Ittersum (2007) found that implementations of BEFMs are
hardly ever re-used by the research community for new problems or applications. In addition
to a modular structure, one of the features that could stimulate the use of these BEFMs by a
larger community is an easy to use and accessible graphical user interface (GUI). We
developed such user interface for FSSIM as the FSSIM-GUI.
This contribution introduces the functionality, the software architecture and technical
implementation choices of the FSSIM-GUI.
Functionality
The aim of the stand-alone FSSIM-GUI is to have an intuitive, easy to use user interface that
allows users to initialize, run and modify data for the simulation of farm systems which are
modelled by the FSSIM components. The application is web-based, which makes the
application easy accessible for the research and farmer’s community. The functionality is
primarily targeted at users with no or little experience in the use of BEFMs in general. The
user interface has a public and restricted part. Within the public part of the website, we inform
the user about FSSIM and its components. The user must register in order to gain access to
the restricted part. By registering users, the simulations executed by each user can be stored,
retrieved and repeated at a later stage by that user. Also, modifications of input data or outputs
created through a simulation are stored on a central database and accessible for the user that
created the simulation.
In order to run a simulation a user must create a new model experiment. This experiment
contains meta-data about the purpose and assumptions of the simulation. Then, the user
selects and parameterizes the FSSIM components with default or user-defined variables, such
as components to generate and quantify alternative cropping or livestock activities, to assess
current activities or to simulate farm behaviour in response to defined policies (i.e. the
mathematical programming model). After a successful simulation, the output is stored in the
database and the user can view some of the key aspects of the output or export it for further
analysis.
Architecture
The application has a service-oriented architecture, which is a software design that consists of
two parts; a set of services and the service infrastructure. A service is a declaration that clearly
defines a specific functionality of the application. This functionality can be used by other
services within the application or any other client. A service infrastructure is a framework that
deals with the information flows between services and tiers. The actual implementation of the
service declaration registers the service at the infrastructure and can use the infrastructure to
332
Session B6: Software infrastructures and tools for integrated assessment
perform its task in a simple and efficient manner. The tiers are loosely coupled via the service
declarations and service infrastructure.
The application has three tiers; a data tier, a service tier and a user interface tier. The first
tier, the data tier, is an object-oriented data model representing the variables for the different
model components and user management. In order to persist the data, the data model is
mapped to a relational database using an object relational mapping tool that is part of the
service infrastructure. The second tier is the service tier, which contains the service
declarations and its implementation. These service declarations are the gateway to the
functionality of the application and hides all technicalities, e.g., the implementation and
linkages between tiers, for a client (which can be a user interface, another service or another
application). It clearly specifies what the application can do. The third tier, the user interface,
is actually a client of the application. It only uses the registered service declarations and the
service infrastructure to manage the simulation runs. The user interface implements an event
model that handles the different actions from a user in order to successfully use FSSIM.
Technical Implementation
We opted to exploit the leading technologies and industry standards for complex web-based
application and to use only open source technology. After considering the service
infrastructures Spring and Java Enterprise 5, more precisely Enterprise Java Beans 3 (EJB,
JSR 220), we choose the standard EJB, knowing that EJB can be easily incorporated into the
Spring framework. The choice of EJB also allows to choose different object relational tools
like Hibernate or Toplink depending on the choice of database and different application
servers that implement Java Enterprise 5. We choose JBoss as service infrastructure. JBoss is
a popular application server with native support for Hibernate. Based upon these choices and
the actual implementation of the application services we limit ourselves to comply to Java 1.5.
The graphical user interface is built in Adobe Flex 3, a leading technology for building rich
internet applications; internet sites that have a look and feel of a desktop application. As Flex
is based on a different technology, called Actionscript, than Java, we choose Granite Data
Services as an infrastructure to connect the service tier to the GUI. Granite is an open source
tool that uses the same protocol and a similar implementation to connect Flex to Java as the
popular Adobe Flex Data Services or BlazeDS. Also, Granite is not limited to one CPU. The
Java data model is automatically mapped to an Actionscript data model by Granite. The
Actionscript data model allows the developer of the GUI to have access to the same data types
as defined in the data tier of the application.
Conclusions
With this contribution we aim to demonstrate the functionality of the FSSIM-GUI as well as
informing the research community about the advantages and challenges of using a service
oriented architecture in combination with leading state-of-the-art technologies and industry
standards. The FSSIM-GUI is available at http://fssim.seamless-ip.org and published under
the GNU public license, respecting the separate licenses and copyrights for the included thirdparty open source libraries.
References
Janssen, S. & M.K. Van Ittersum, 2007. Agricultural Systems 94: 622-636.
Janssen, S., et al., 2009. AgSAP Conference 2009, these proceedings.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
http://java.sun.com/javaee/technologies/javaee5.jsp http://java.sun.com/products/ejb/
http://www.springframework.org/
http://www.jboss.org/
http://www.hibernate.org/
https://glassfish.dev.java.net/
http://www.graniteds.org/confluence/pages/viewpage.action?pageId=229378
333
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
What are the key factors for successful integration of life cycle assessment
into FADN?
T. Nemecek, S. Blaser, D. Dux, G. Gaillard
Agroscope Reckenholz-Tänikon Research Station ART, 8046 Zurich, Switzerland
Contact: thomas.nemecek@art.admin.ch
Objectives and rationale
Farmers, authorities and stakeholders have an increasing need to assess the sustainability of
individual farms as well of the whole agricultural sector. Since several decades, the economic
situation of the Swiss agricultural sector is regularly monitored by a farm accountancy data
network (FADN). On the contrary, no such monitoring was carried out for the environmental
impacts of the farms. A feasibility study on 50 farms was performed to check the possibility
to implement a combined economic-environmental analysis by coupling FADN to
environmental life cycle assessment (LCA). The study concluded that a valuable assessment
of farms is feasible, provided that data collection and analysis could be automated (Rossier &
Gaillard, 2001). Therefore, a monitoring project was initiated in 2004, with data collection on
60, 170 and 170 farms in the years 2006, 2007 and 2008. The farms were selected by a
stratified sampling procedure using the criteria of farm type, region (lowlands, hills and
mountains) and farming system (integrated or organic).
The two main goals of the project are an analysis of environmental performance of the
farms with feedback to the farmer as a basis for environmental management and an
environmental monitoring of the agricultural sector. Furthermore, the results will serve for
various research projects.
Software infrastructure for environmental and economic farm assessment
The software infrastructure was based on the three existing software tools, namely the farm
management software AGRO-TECH, the LCA method and tools SALCA (Swiss Agricultural
Life Cycle Assessment) and the FADN database. These tools however had to be adapted (e.g.,
allocation between farm branches) and extended (e.g., data collection) in order to cover all
needs of the integrated assessment. Furthermore, interfaces were built for data extraction from
the farm management software for calculating an LCA (SALCAprep) and for validation of
the LCA results and export to the FADN database (SALCAcheck). A modular structure with
flexible interface programming was chosen to implement the software environment. The
workflow is presented in Figure 1. An LCA centre with specifically trained staff was created
for the purpose of centralized data collection and LCA calculation. Several feedback loops are
included in the workflow, e.g., data correction by the farmer after the plausibility check by
SALCAprep or validation of the LCA results by SALCAcheck. The latter component
compares the environmental performance of each individual farm to average reference farms
(benchmarking) and the farmer gets a feedback concerning the relative environmental impacts
of his farm, together with an analysis of possible improvement measures. The FADN
evaluation centre collects in economic and environmental data in the FADN database and
uses them for feedbacks to farmers (for environmental management), sectoral monitoring and
research projects.
Due to the high data requirements, ensuring data quality is a key issue. The main measures
are clear user guidelines, a hotline for the farmers, checks during data entry in the farm
management software (mandatory fields and range checks), plausibility checks during data
extraction in SALCAprep (mandatory fields, range checks and conditional checks)
complemented by visual control by staff of the LCA centre, validation of the LCA results in
SALCAcheck (comparison to benchmarks) as well as plausibility checks with general farm
data and economic results before importing into the FADN database.
334
Session B6: Software infrastructures and tools for integrated assessment
Accountancy
data
Technical
data
AccountancySoftware
(AGRO-TWIN)
SALCAprep
data extraction
Accountancy
Data
Plausibility tests
FADN evaluation centre
Farm management
software
(AGRO-TECH)
AccountancySoftware
LCA centre
Trust and
accounting office
Farm
Key factors for successful integrated assessment
After three years of data collection and the second year of data analysis, we can preliminarily
conclude about the key factors for an integrated assessment of economic-environmental
performance of farms:
• Automation is crucial to process a large number of datasets in an acceptable time.
However, it is a challenge, since farms are complex enterprises in a large variety of pedoclimatic and socio-economic conditions.
• Keeping a strict modular structure of individual components enables to manage the
complexity.
• Ensuring the data quality is essential, checks are needed at several levels.
• Finally a feedback to farmers is a key to motivate the farmers to participate and to provide
high quality datasets; the economic remuneration is only one element for motivation.
SALCAcalc
LCA calculation
SALCAcheck
LCA validation and
benchmarking
LCA-Data
Export ÖBStelle
FADN
database
Environmental
management
Monitoring
Research
Feedback to
farmers
Farm types
Regions
Sectors
Modelling
Planning
Figure 1. Flow chart for the data from the farm to the centralized evaluation.
Reference
Rossier, D. & G. Gaillard, 2001. Bilan écologique de l’exploitation agricole: Méthode et
application à 50 entreprises. Forschungsanstalt für Agrarökologie und Landbau (FAL),
Zürich, 176 pp.
335
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
An integrated framework for management of spatial and temporal data on
a drained watershed
O. Salazar, A. Joel, I. Wesström
Swedish University of Agricultural Sciences, Department of Soil and Environment,
P.O. Box 7014, Uppsala 750 07, Sweden
Contact: Osvaldo.Salazar@mark.slu.se
Introduction
The mechanisms determining the hydrology of drained watershed are complicated and depend
on many factors. The development of watershed scale models has provided methods to
describe the hydrologic components in these areas. Distributed models treat the watershed as
a spatially variable physical system with various functioning hydrological units, which is
more realistic and has significant theoretical advantages that make it more useful than other
types of models. Although spatial and temporal variations in such models are necessary, these
may be stated with the integration of geographical information system (GIS) with models.
This study describes the development and evaluation of a watershed integrated framework
that operates with Arc Hydro schematic features (Maidment, 2002) in the ArcGIS 9
ModelBuilder environment (ESRI, 2004). This framework is based on the field hydrology
DRAINMOD model (Skaggs, 1978) and uses a distributed approach to route water from the
field edge to the watershed outlet. This modelling approach was tested and compared based
on the ability to predict discharge on a 750 ha artificially drained watershed in a coastal area
in south-east Sweden where coarse-textured is the dominant soil type.
Methods
The Kleva watershed is typical of agricultural landscapes in the coastal area of Öland Island
in the Baltic Sea. It is a 750 ha artificially drained watershed located in Mörbylånga in
Kalmar County in south-east Sweden (56°31' N, 16°23' E) (Figure 1). The watershed was
divided into 95 fields. Each field was assumed homogeneous about soils, land use, and water
management practices. Hydrology data collected during six periods between 2003 and 2008
were used to test the framework.
1 km
Kleva watershed
Legend
: field
■ : measurement station
─ : Kleva river
─ : watershed boundary
336
Figure 1. Diagram of the study
area in Kleva watershed.
Session B6: Software infrastructures and tools for integrated assessment
In this study, a framework proposed by Whiteaker et al. (2006) for integrating Arc Hydro
features with a schematic network to simulate hydrologic watershed behaviour was used. The
DRAINMOD model version 5.1 was run to simulate a water balance on each field with input
based on the individual characteristics of each field. Daily drainage outflow and runoff
predicted by the model was summed on each field and stored in a time series of outflow. A
distributed approach was used to route the time series of daily DRAINMOD-simulated
outflow from each field drainage point to the watershed outlet, where the outflow from each
field were summed through the stream network to predict the discharge at the outletwatershed. The performance of the framework was evaluated considering the uncertainties of
the model inputs using the Generalized Likelihood Uncertainty Estimation (GLUE)
methodology (Beven & Binley, 1992).
Results and discussion
The GLUE estimates (uncertainty bands 5% and 95%) and measured monthly discharges
were in a satisfactory agreement. Overall the relative discharge volume deviation (Ed)
indicated that neither over prediction nor under prediction was found during the calibration
and validation periods (Table 1). However, compared with observed values, the relative
discharge volume deviation at 5% (Ed,5%) showed a positive value during Period 5. This
positive value was caused by an over prediction of the GLUE estimates occurred during
January-March in Period 5, when exceptionally mild weather conditions were registered.
During this period DRAINMOD had difficulties to divide rain and snow, and consequently to
predict flow events correctly.
Table 1. Relative discharge volume deviation between observed and 5% (Ed,5%) and 95%
(Ed,95%) prediction intervals for calibration (Period 1 to Period 3) and validation (Period 4 to
Period 6).
Period
na
Ed,5%b
Ed,95%c
9
–0.5
0.2
2
12
–0.3
0.3
3
3
–0.5
1.9
1-3
24
–0.4
0.3
4
6
–0.4
0.1
5
6
0.2
1.0
6
8
–0.5
0.7
4-6
20
–0.2
0.5
a
c
n is the number of months; b a positive value indicates that the discharges are outside the prediction interval
a negative value indicates that the discharges are outside the prediction interval.
It was concluded that major discrepancies in discharge predictions were due to errors in snow
accumulation and melt in mild winter periods. The results indicated that the framework can be
an effective tool for describing the discharge from artificially drained watersheds under cold
conditions in south-east Sweden.
References
Beven, K. & A. Binley, 1992. Hydrol. Process. 6: 279-298.
ESRI, 2004. ArcGIS 9, What is ArcGIS? Environmental System Research Institute (ESRI),
Redlands, CA, USA.
Maidment, D.R., 2002. Arc Hydro: GIS for water resources. Redlands, CA, ESRI Press.
Skaggs, R.W., 1978. A water management model for shallow water table soils. Technical
Report 134. Water Resource Research Institute, NCSU, Raleigh, NC, 178 pp.
Whiteaker, T.L., et al., 2006. Trans. GIS 10: 219-237.
337
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Groundwater concepts and agriculture: A GIScience data model study
David R. Steward, Xiaoying Yang, Willem J. De Lange
Kansas State University, Civil Engineering, 2118 Fiedler Hall, Manhattan, KS 66506, USA
Contact: steward@ksu.edu
Introduction
Groundwater is important for agriculture, providing a secure water source through drought. In
semi-arid regions across the world, modern society is wrestling with how to balance shortterm and long-term water needs for irrigated agriculture. It is important to develop scientific
support to guide societal planning and policy.
Methods
Understanding the impacts and interactions between agriculture and groundwater resources is
supported by a large number of models. Groundwater models utilize a variety of numerical
techniques based upon the Analytic Element Method, the Finite Difference Method and the
Finite Element Method. These computational tools currently lack data interoperability
between groundwater models that employ different mathematics, data types, and legacy
input/output file formats. Even so, existing groundwater modeling methods are all rooted in
the same fundamental hydrologic properties of mass, flux, pathways, and residence time.
AquiferLayer
Type: •Aquifer
•Aquitard
1
*
1
*
*
1
AquiferPoint
1
AquiferLine
1
*
AquiferPolygonPoint
1
*
AquiferPolygonLine
1
*
AquiferLinePoint
*
1
1
*
*
1
*
1
AquiferPolygon
*
*
*
1
*
AquiferPolygonLinePoint
1
AquiferProperty
Type: •Base elevation
•Conductance
•Dispersivity
•Hydraulic conductivity
•Porosity
•Resistance
•Specific storage
•Specific yield
•Thickness
•Top elevation
•Transmissivity
Units
Value: [x,y,z]
*
*
AquiferRaster
1
*
WaterPoint
1
*
*
WaterLine
1
*
1
21°30’0"E
1
*
1
*
WaterPolygonPoint
1
*
WaterPolygonLine
1
*
WaterLinePoint
WaterPolygon
*
1
1
*
*
1
*
1
*
WaterPolygonLinePoint
1
WaterProperty
Type: •Discharge
•Head
•Velocity
•Water boundary conductance
•Water boundary thickness
•Water boundary porosity
•Water boundary resistance
Units
Value: [x,y,z]
Time
22°0’0"E
22°30’0"E
23°0’0"E
23°30’0"E
24°0’0"E
24°30’0"E
18°30’0"S
18°30’0"S
19°0’0"S
19°0’0"S
19°30’0"S
19°30’0"S
20°0’0"S
20°0’0"S
*
20°30’0"S
20°30’0"S
*
WaterRaster
1
1
q
21°30’0"E
0
20
40
22°0’0"E
80
22°30’0"E
120
23°0’0"E
23°30’0"E
24°0’0"E
24°30’0"E
Kilometers
160
Figure 1. The GIS concepts datamodel is used to organize datasets across the world and
supports a wide range of groundwater modelling tools, which in turn are related to data for
other processes such as agriculture and economics (Steward & Bernard, 2006; Steward et al.,
2008, 2009; Steward & Ahring, 2009; Yang et al., 2009).
338
Session B6: Software infrastructures and tools for integrated assessment
These fundamental groundwater system concepts serve as the basis of in designing a new
object-oriented conceptual groundwater data model. Geographic Information Science
(GIScience)-based tools and methodology provide a framework for quantitative analysis
where a groundwater system as a series of aquifer layers with defined aquifer properties and
water boundary conditions. This data model provides a mechanism to structure and organize
groundwater datasets pertinent to agriculture, and to store a conceptual view of groundwater
that is common across groundwater models.
This data model is not intended to compete with existing Graphical User Interfaces (GUIs)
for existing tools. Instead, the data model provides a central storage mechanism. With built-in
database management capabilities, well-defined domain, relationships, and topological rules,
spatial data models improve data storage efficiency and enforce data integrity.
Results and discussion
The groundwater concepts data model addresses current lack of data interoperability across
models, provides flexibility in management of data and naming conventions, and provides
enhanced data integrity. Case studies (in the references Steward & Bernard, 2006; Steward et
al., 2008, 2009; Steward & Ahring, 2009, in review) illustrate the new conceptual
groundwater data model’s utility as an effective mechanism for structuring and organizing
groundwater datasets and storing a conceptual model of groundwater, and its flexibility to
support groundwater models across existing modeling techniques.
References
Steward, D.R. & E.A. Bernard, 2006. Ground Water 44(1): 56-61.
Steward, D.R., et al., 2008. Proceedings of The Royal Society of London, Series A, 464: 223248.
Steward, D.R. & T. Ahring, 2009. Journal of Engineering Mathematics. (in press)
Steward, D.R., et al., 2009. Water Resources Research. (accepted with minor revisions)
Yang, X., et al., 2009. International Journal of GIS. (accepted with minor revisions)
339
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Tools for transparent spatial integrated assessments
P.J.F.M. Verweij, M. Van Eupen, J. Roos-Klein Lankorst, W. Nieuwenhuizen
Alterra, Wageningen-UR, P.O. Box 47,
6700 AA Wageningen, The Netherlands
Contact: Peter.Verweij@wur.nl
Introduction
Model-based sustainability impact assessments (SIA) have a strong integrated character,
analysing environmental, social and economic impacts and side effects of an intended policy
option.
Societies see the emergence of new governance concepts, based on the assumption that
processes of planning and decision taking are no longer hierarchical but the product of
complex interactions between governmental and nongovernmental organizations, and the
general public (the model of ‘co-production of knowledge’; Callon, 1999). All involved are
seeking to influence the collectively binding decisions that have consequences for their
interests. To account for this changing governance and the increased number of stakeholders
involved, decisions need to be assessed in an integrated context.
This paper discusses the OSIRIS modelling environment (Verweij, 2004) which is a
generic framework for qualitative knowledge rule based reasoning. OSIRIS was originally
developed from Dutch spatial landscape planning perspective, but has spread to a wider
community within and outside of Europe, such as Brasil (Jongman, 2004), West Africa and
Asia (Van Eupen et al., 2007).
Method
Alterra, a research institute of Wageningen University, develops solutions for applied and
basic research across a number of environmental domains, including ecology, hydrology,
agriculture, landscape and spatial planning. Around 2001 many of the disciplines were
assessing (integrated) impacts of spatial plans. At the same time, the in-house software
engineering group was developing software tools to facilitate singular discipline impact
assessments. Similarities and differences in system requirements were analysed in order to
facilitate the integration of knowledge from different disciplines and reduce time and effort
building software tools. Analysis was done around: (i) existing systems (e.g., LEDESS,
Harms et al., 2000) by interviews and studying available tools and literature and; (ii) outline
of future projects, e.g., MENES (Boogaard, 2003), KELK (Roos-KleinLankhorst et al., 2004).
The analysis resulted in a common conceptual model framework consisting of: (a) spatial
explicit scenarios in terms of spatial strategy (e.g., clustering, or diffuse) and area based
quantities; (b) the need to transparently explicate knowledge by rules-of-thumb, like
knowledge matrices and decision trees; and (c) be able to drill down in modelled results by
looking up the rules-of-thumb.
Results and discussion
As an example of an application, Osiris was used to build a model, which determines how
vegetation and fauna are affected when, abiotic parameters in the Yellow River Delta are
changed (such as flooding frequency and groundwater availability) (Van Eupen et al., 2007).
Strong advantage of the used OSIRIS modelling environment was found both during
development of the model and final application of the scenarios:
− The tool was used in several workshops in which typologies were determined and
relationships between them have been defined by means of rules of thumb. Participants
needed to think in a structured way following the conceptual framework of spatial
340
Session B6: Software infrastructures and tools for integrated assessment
strategies and typologies linked with available spatial data and the available expert
knowledge.
− In the application of scenarios drilling down allowed exploring causal relationships of
impacts of spatial plans by looking up the rules of thumb as defined by participants during
the workshop itself. This knowledge transparency makes it possible to have several
iterations of fine-tuning the model during a single stakeholder workshop.
The OSIRIS modelling framework is a useful tool to interactively implement expert
knowledge and determine impacts of spatial scenarios in all kind of domains. This improves
communication and is helpful in the decision making process.
References
Boogaard, H.L., et al., 2003. Agro Informatica 16 1: 5-8. (in Dutch)
Callon, M., 1999. Science, Technology, and Society 4(1): 81-94.
Harms, W.B., et al., 2000. In: Eds Ü. Mander & R. Jongman, Landscape Perspectives of Land
Use Changes, WIT, Southampton, UK. Advances in Ecological Sciences 6, 1-17.
Jongman, R.G., et al., 2004. The Pantanal Taquari project: Decision Support System for
catchment based river management. Final Report, Programme Partners for Water. Alterra,
Wageningen.
Roos-Klein Lankhorst, J., et al., 2004. VBK Editorial Management 12(4): 38-39.
Van Eupen, M., et al., 2007. Impact modelling of scenarios on vegetation and fauna in the
Yellow River Delta. Yellow River Delta Environmental Flow Study Sino-Dutch Study
Programme Final Report 2007 YRCC, Yellow River Water Resources Protection Bureau,
Alterra, Delft Hydraulics, Arcadis Euroconsult, RUN.
Verweij, P.J.F.M., 2004, Osiris manual W!SL, Wageningen.
341
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Parameter estimation software for crop models
D. Wallach, F. Van Evert, Myriam Adam
INRA UMR AGIR,
BP 52627, 31326 Castanet Tolosan Cedex, France
Contact: wallach@toulouse.inra.fr
Introduction
Parameter estimation is a major aspect of crop modelling. Together with the functional forms
of the equations, it is a major determinant of prediction quality. On the other hand, it is a
difficult and time-consuming exercise, and requires expertise in regression analysis not
always available in a modelling project. This suggests that it is of major interest to provide
software to automate model parameter estimation. That is the purpose of the work reported
here. There are two separate problems that must be addressed. The first is to provide
algorithms that are appropriate for crop models. The second is to make these algorithms
available to a large number of models.
Parameter estimation for crop models is generally based on a least squares approach, but in
the case of crop models the problem is complicated by the complexity of the data and the
models. Different algorithms may be appropriate in different cases, and for complex data
there is no standardized approach. Therefore the software includes several different estimation
algorithms, and others can easily be added.
The software consists of functions in the R statistical computing language (R Development
Core Team, 2007), which is freely available. Coupling with modelling software is separated
into specific functions. We have coupled the software with two different platforms, namely
the MODCOM modelling platform (Hillyer et al., 2003), which is used in the SEAMLESS
project (Van Ittersum et al., 2008) and the RECORD modelling platform (Chabrier et al.,
2007). Both are built around a simulation motor based on the DEVS paradigm (Zeigler &
Praehofer, 2000). A major attraction of linking the software to such platforms is that in this
way it is not specific to a particular model. Rather, it is available to any model or any model
components that are implemented on those platforms.
Methods
The initial version of the software has five different algorithms for parameter estimation. The
different parameter estimation methods correspond to different hypotheses about model error
(Makowski et al., 2006). The first is ordinary least squares, which is appropriate for the case
of a single measured variable such as yield. The second algorithm minimizes the likelihood
for the case where several different variables are measured, but the same measurements are
available for each field. This leads to a determinant criterion (Bates & Watts, 1988). The third
and fourth algorithms both involve minimizing a weighted least squares criterion. They are
adapted to the case where several different variables are measured and different fields may
have different measured variables (or variables measured at different times). The third
algorithm assumes independence between errors; the fourth uses an empirical criterion to
avoid giving extra weight to fields with many measurements (Wallach et al., 2001). In
algorithms three and four the user can input error variances for each variable or they can be
estimated together with the parameter values. Also, the user can either treat the original data
or first do a log transformation, which is indicated in the case where model error is expected
to increase as the measured value increases (Bates & Watts, 1988). The fifth algorithm is a
Bayesian algorithm, based on importance sampling. This algorithm calculates a probability
distribution for each estimated parameter.
The MODCOM software is written in the C sharp language. Information is passed from the
R functions to MODCOM using Microsoft COM technology. The R software requires the
342
Session B6: Software infrastructures and tools for integrated assessment
following information, which must be stored in a data base: the observed data, the name of the
model to be run, the paths to the input files for the contexts of the data (for example climate,
soil and management files for each context), the list of model parameters and indicators as to
which are to be estimated and finally information related to the correspondence between the
data and the model output. At each iteration the R routine sets the parameter values in
MODCOM to the current values, executes the model for each context, retrieves the results,
calculates the criterion to be minimized and determines the parameter values for the next
iteration. The search algorithms are built into available R functions.
Results and discussion
It is envisioned that the software be used in several different ways. First, it can be used to
estimate parameters for individual model components, using data that concern a single
component decoupled from the rest of the model. Second, it can be used to estimate
parameters using the full model and multiple types of measurements. Finally, it can be used as
a methodological tool to compare different estimation algorithms.
When a model does not agree well with data, the problem no doubt arises from both the
way the equations describe the processes and from the values of the parameters. It is hoped
that by providing a convenient and flexible set of parameter estimation algorithms, the
software we propose will allow modellers to deal more simply and more efficiently with the
parameter value aspect of the problem. It should then be possible to better evaluate the role of
the process equations in model error.
References
Bates, D.M. & D.G. Watts, 1988. Non linear regression analysis and its applications. Wiley,
New York.
Chabrier, P., et al., 2007. In: Farm Systems Design 2007, an international symposium on
Methodologies for Integrated Analysis of Farm Production Systems, Catania, Sicily, Italy.
Hillyer, C., et al., 2003. European Journal of Agronomy 18: 333-343.
Makowski, D., et al., 2006. In: Eds D. Wallach et al., Working with dynamic crop models.
Elsevier, Amsterdam, pp. 101-150.
R Development Core Team. 2007. A language and Environment for Statistical Computing. R
Foundation for Statistical Computing, Vienna, Austria.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
Wallach, D., et al., 2001. Agronomy Journal 93: 757-766.
Zeigler, B.P. & H. Praehofer, 2000. Theory of modeling and simulation. Academic Press.
343
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Architecture of SEAMLESS-IF as framework application for integrated
modelling
J.J.F. Wien, A.E. Rizzoli, M.J.R. Knapen, I.N. Athanasiadis, S. Janssen, L. Ruinelli,
F. Villa, M. Svensson, P. Wallman, B. Jonsson
Alterra, Wageningen UR, P.O. Box 47, 6700 AA Wageningen
Contact: jan-erik.wien@wur.nl
Introduction
The SEAMLESS (System for Environmental and Agricultural Modelling; Linking European
Science and Society) consortium develops a computerized and integrated framework
(SEAMLESS-IF) to assess the impacts on environmental and economic sustainability of a
wide range of policies and technological improvements across a number of scales (Van
Ittersum et al., 2008). In the SEAMLESS-IF, different type of models and indicators are
linked into model chains, where each model uses the outputs of another model as its inputs
and ultimately indicators are calculated. This integrated modelling requires interoperability,
which is the ability of two or more systems or components to exchange information and to use
the information that has been exchanged (IEE, 1990). This paper describes the software
architecture and design of SEAMLESS-IF with emphasis on semantic interoperability.
Semantic interoperability
In the SEAMLESS project, a large number of scientists co-operate working on a wide range
of issues, like different type of models, data and databases, indicators, assessment problems
and user interaction, and they need support for specifying data, models, projects and their
relationships. Semantic interoperability is the ability of systems/components to share and
understand information at the level of formally defined and mutually accepted domain
concepts. In SEAMLESS, we employ semantic modelling and we have developed an
ontology to establish a set of shared domain concepts. All the commonly shared data types in
SEAMLESS are declared in the ontology (starting from projects, up to the model exchange
items). This is an important shift in everyday practice of modellers: modellers specify the data
requirements of their models on a higher level, i.e. that of an ontology, This ontology is
automatically transformed into a relational database model, to which ‘data collecting’
activities need to comply with.
Client server architecture
As seen in Figure 1, SEAMLESS-IF is based on a layered, client-server architecture. The end
user interacts with the system by means of two web-based Graphical User Interfaces (GUI)
which run as clients. The client-server architecture of SEAMLESS-IF allows for future clients
to be developed and linked to the server, in order to cater for specific needs of different user
groups. The currently available clients are:
- PE GUI, the policy expert GUI: through this interface an expert can evaluate the impact of
alternative agricultural policies from the different aspects of sustainability;
- IM GUI, the integrative modeller GUI: the module that guides the end-user to manage
projects and request the execution of model chains, in order to produce results to be later
used by policy experts;
The above client applications interact with the software services provided by the SeamFrame
Server, which include:
- The Modelling Environment: a programming framework that offers a series of facilities to
encapsulate and wrap existing models for execution by the processing environment. It
allows to deliver model components wrapped by a SeamFrame specific interface,
344
Session B6: Software infrastructures and tools for integrated assessment
compliant with the Open Modelling Interface (OpenMI) standard (www.openmi.org)
(Gijsbers et al., 2006), so that it can be executed by the Processing Environment.
- The Processing Environment: both a programming framework and a software application
that receives user requests for the execution of chains of model-components. It enables
model composition and execution. The actual exchange of data among the models is based
on the OpenMI that provides a standardized interface to define, describe and transfer data
between software components that run sequentially.
Figure 1. SEAMLESS-IF
architecture.
Technical Implementation
The current software stack for SeamServer consists of an application server (the market leader
Tomcat), data storage (PostgreSQL ) and a data access layer (Hibernate). The business logic,
where the preparation and management of experiments to be run by the processing
environment happens, is based on the Spring Framework, a Java based solution delivering the
a full-stack Java/JEE application framework. Finally, the remoting technology we adopted is
BlazeDS, a server-based Java remoting and web messaging technology that allows the
connection to Adobe® Flex® and Adobe AIR™ applications for delivering rich Internet
application (RIA).
Results and discussion
SEAMLESS has achieved a very important shift in integrated modelling practice, i.e.
modellers specify the data requirements of their models in a higher level, i.e. that of an
ontology., This ontology is than automatically transformed into a relational database model,
to which ‘data collecting’ activities need to comply with. This ensures the match between data
collection and data requirement for running the model-chain. It also provides the required
semantic interoperability. The client and server technologies that were adopted in
SEAMLESS-IF have advantages in the field of:
- Maintainability: Less custom source code to maintain, instead leveraging of popular open
source frameworks and source code generated from a higher level ontology.
- Stability and security: Replacing custom coding with well-known and tested open source
frameworks.
- Performance: Replacing XML with the binary AMF format. Tests shows that this is at
least 4 times faster.
- Flexibility: Use of well-known design patterns, program by interfaces and Spring
dependency injection.
References
IEE, 1990. IEEE Standard Computer Dictionary: A Compilation of IEEE Standard Computer
Glossaries. New York, NY.
Gijsbers, P.J.A., et al., 2006. In: Proceedings of HIC 2006. Nice, France, pp. 72-81.
Van Ittersum, M.K. et al., 2008. Agricultural Systems 96: 150-165.
345
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme B
State-of-the-art of components for integrated
systems
Session B7: Databases for integrated assessment of
agriculture and rural development
Session organizers:
Erling Andersen
University of Copenhagen, Denmark
Haluk Kasnakoglu
Middle East Technical University, Ankara, Turkey
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Usefulness, feasibility, limitations and sustainability of an integrated
international database on agricultural statistics1
Haluk Kasnakoglu
Mesa Koru, Sardunya E/21, 06810 Ankara, Turkey
Contact: haluk.kasnakoglu@gmail.com
Introduction
An international database on agricultural statistics has a minimum of four dimensions, namely
subject (domain), item, space and time. The subjects include quantities and values of
resources, production, trade and consumption. The items vary for subjects but are in general
factors of production, products and commodities for consumption. The space consists of the
sub-regions, regions, regions and groupings. Finally the time dimension may range from days
to months, years and decades. A statistical database contains data points on different
combinations of elements from these dimensions. An integrated database is one where data
within and between different domains and dimensions are comparable, consistent and
compatible and hence can be used together to harvest the added value from combining data
points both within and between different domains and dimensions (Figure 1).
Country A
Item
Country B
Subject
Figure 1. Integrated database
structure.
Time
Usefulness
There is no doubt that being able to use data points from different domains over different
dimensions adds value. Statistics on yields for a specific commodity is of little use if it cannot
be compared over time and space and with other commodities. The information on yield
changes gains an additional value if the change in production quantity can be valued by
introducing product prices and even more so by introducing the changes in costs of
production to compute the value added implications of yield increases. Apart from the added
value of new information generated, an integrated database contributes to the consistency,
comparability and compatibility of data in different domains and over different dimensions as
well as the analysis conducted with such databases.
Few examples of non-integrated data
Unfortunately there does not exist an integrated statistical database on food and agriculture at
the international level. Furthermore there are very few cases at the national level where
agricultural statistics are fully integrated within themselves and with the rest of the statistical
system. As a result the users must live/struggle with inconsistencies such as the following:
•
52 Billion US$ total world cereal imports is 13% larger than 46 Billion US$ total world
cereal exports for 2005.
•
Kenya imports 141,000 tons of rice from Pakistan paying 26 million US$ in while
Pakistan exports 334,000 tons receiving 56 million US$ from Kenya in 2004.
1
Keynote presentation
348
Session B7: Databases for integrated assessment of agriculture and rural development
•
2.0 million tons of cereal production reported in the world for 2006 in one database while
10% higher production (2.2 million tons) reported in another database for the same year
by the same organization.
Feasibility
The pre-requisites for integrating databases in different domains are common definitions,
scalable and mapped classifications and a conceptual integration framework. Candidates for
classifications are ISIC, HS and CPC with fairly good mapping to each other. Good
integrating frameworks are economic accounts (to integrate real data with monetary data and
resource data with output data) and supply utilization accounts (to integrate production, trade
and consumption data). Integration involves aggregations as well as disaggregations to match
the differences in the sizes and dimensions of different domains. The process of integration
brings to light the inconsistencies and flows in the constituting datasets. This implies a serious
process of reconciliation where it is unavoidable to modify some first-hand data. This requires
on the one hand, the use of well accepted state-of-art estimation methodologies supported by
good meta data but at the same time very good communication techniques and political
strength as what is at stake are the revision of official statistics reported by countries. In
summary, the ingredients for an integrated database are available but putting them together
and even more so selling it is challenging.
FAOSTAT experience
FAO with its well known database FAOSTAT is the international organization responsible
from compilation and dissemination of food and agriculture statistics. FAOSTAT has long
been producing time series data on production, trade, selected resources and prices. FAO has
also been estimating the numbers of undernourished to monitor progress toward the FAO
Food Summit and UN MDG goals of reducing hunger in the world. This involves
constructing supply utilization accounts and food balance sheets which are already major
steps toward integrating food and agriculture statistics. FAO has recently made an effort to
construct a fully integrated database within the framework of modernization and update of
FAOSTAT and disseminated it on June 2006. The integrated component of new FAOSTAT
referred to as the “Core” was short lived and has been turned-off at the end of 2007 hence
going back to old FAOSTAT era with a new face. The main reasons for the unsuccessful
attempt were more on the socio-political side than on the technical side. Among the factors
that have contributed to the failure of the attempt one can list:
•
Resentment by staff to new technology and methodology which implied more reliance on
machinery and models and less on human labour and judgment and hence need for retraining and re-positioning.
•
More work for existing users to understand and adapt to the new user interface and
classifications.
•
Revisions in the past series with implications on the analysis conducted with them,
including revisions in the FAO’s politically sensitive estimates and projections of number
and prevalence of undernourishment (which proved to be supported with the recent
developments in world markets and FAO’s new estimates of hunger published).
•
Lack of resources for a more user friendly user interface and communication.
Concluding remarks
It must be understood that an integrated database is a non-unique analytical database and is
not necessarily one which will satisfy all needs equally well. It will necessarily mean loss of
some detail and estimation in the process of standardization of data dimensions and some
deviation from official data in the process of data integration and reconciliation. On the other
hand it should not be considered as a substitute but complementary to the existing datasets.
349
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrating farming system, environmental and socio-economic data:
The SEAMLESS database
E. Andersen, B. Baruth, B.S. Elbersen, G.W. Hazeu, S. Janssen, I. Staritsky,
I.J. Terluin, C.A. Van Diepen, D. Verhoog
Forest & Landscape, UoC, Rolighedsvej 23,
1958 Frederiksberg C, Denmark
Contact: eran@life.ku.dk
Introduction
The SEAMLESS integrated project has developed a computerized, integrated and working
framework (SEAMLESS-IF) to assess environmental, economic and social effects of
alternative agricultural and environmental policy options at multiple scales (Van Ittersum et
al., 2008). This includes an integrated database combining farming systems, environmental
and socio-economic data covering the European Union (EU25).
In the integrated database all data used in the project are stored including model input and
output data, contextual data and spatial information for assessment and visualization of
indicators. The database is implemented and managed in the open source object-relational
database management system Postgres with an extension to handle geographical data using
PostGIS and Geoserver. It is expected that a Web Feature Service (WFS) will be used in the
final version of the system to visualize model results.
To reach an integrated database all the data that are included in the SEAMLESS database
are either processed from the original datasets and adapted to the use in the SEAMLESS
project or, in a few cases only, gathered specifically for the project. This paper describes how
data was processed to fit the integrated structure of the database and how the different data on
farming, the environment and on socio-economic issues are linked.
Methods
Three different modes of processing data for integration in the SEAMLESS database are
described relating to (a) farming system data, (b) biophysical data, and (c) data on allocation
of farm types.
In the SEAMLESS database data on farming systems stemming from the EU wide dataset
Farm Accountancy Data Network (FADN) have been aggregated to farm types. This is based
on a farm typology elaborated in earlier projects and adapted to SEAMLESS. The typology
includes 189 different farm types based on a combination of three different dimensions: size
(3 classes), combined specialization and land use (21 classes) and intensity (3 classes)
(Andersen et al., 2006).
The biophysical data in the SEAMLESS database are processed to a spatial framework
based on an agri-environmental zonation aiming to stratify Europe according to the main
biophysical factors that determine the agronomic production capacity (Hazeu et al., 2006).
However, to ensure linkages to farming system modelling, also administrative regions
(NUTS2 regions) are used to delineate the agri-environmental zones. The spatial framework
and the included biophysical data can thus be described as follows: (1) Each administrative
region is divided into one or more climate-zones, described by one time series of climate data.
(2) Each climate-zone is divided into more agri-environmental zones, described by one set of
soil characteristics. To reach this the original climate data has been processed from a 50 km
grid and the soil data has been selected as the dominating soil unit within the specific agrienvironmental zone.
In order to use both the farm type information and the biophysical data as input data for the
integrated modelling in SEAMLESS additional information on the location of the farm types
was added to all farm types. This is achieved by applying statistical procedures to reach an
350
Session B7: Databases for integrated assessment of agriculture and rural development
optimal distribution of crops taking into account information on altitude and Less Favoured
Areas (Elbersen et al., 2006). The result of this is that the database includes information on
the area within an agri-environmental zone managed by a certain farm type, whereas all other
characteristics of the farm types are given for administrative regions.
Results and discussion
The SEAMLESS database consists of 379 tables including 2,379 different fields and the
number of records in the database now exceeds 12 million. The different tables are linked by
a total of 487 relations. The database, thus, integrates data on farming systems with
environmental and socio-economic data covering the territory of the European Union (EU25).
The most important feature of the integrated database is that it ensures consistency
between the data coming from very different domains and sources and across spatial and
temporal scales. Other important aspects are that the integrated database facilitates easy
access compared to storing the data in several separate databases and that transparency in
relation to modelling is facilitated by storing data and model outputs together. Finally, it
should also be mentioned that the processing needed for the integration of the data enhances
the options for respecting the rights of the holders of the original data.
One of the few shortcomings of an integrated database is the loss of detail compared to
some of the original datasets due to the processing for integration. This seems to be inevitable
for integrated databases and probably also for integrated assessments as such. However, the
loss of detail is mediated by the gain of knowledge of the complex relations between the
different domains included.
References
Andersen, E., et al., 2006. SEAMLESS Report No.12, SEAMLESS integrated project,
SEAMLESS-IP.org, 30 pp.
Elbersen, B.S., et al., 2006. SEAMLESS Report No.19, SEAMLESS integrated project,
www.SEAMLESS-IP.org, 107 pp.
Hazeu, G.W., et al., 2006. SEAMLESS Report No.14, SEAMLESS integrated project,
www.SEAMLESS-IP.org, 55 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
351
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
European environmental stratifications: an overview
G.W. Hazeu, E. Andersen, B. Baruth, M.J. Metzger, C.A. Mücher, M. Pérez-Soba,
Ch. Renetzeder, C.A. Van Diepen
Alterra, Wageningen UR, P.O. Box 47, 6700 AA Wageningen, The Netherlands
Contact: gerard.hazeu@wur.nl
Introduction
Different environmental stratifications of Europe are developed within different European
projects. All stratifications aim to divide environmental gradients into convenient units and to
use these units as areas in which objects and variables might have relatively consistent
characteristics (Jongman et al., 2006). The stratifications can be used as basis for up-scaling
and as a sampling framework. However, as the objectives of the projects are different also the
characteristics of the stratifications will be distinct from each other.
This paper presents an overview of the Environmental Stratification (EnS/EnZ), the AgriEnvironmental Zonation (AEnZ), the Spatial Regional Reference Framework (SRRF), the
European Landscape Classification (LANMAP) and the FARO typology. The objectives,
similarities and differences between these typologies will be described.
Methods
The different environmental stratifications are reviewed regarding their objectives and
applications. Robustness of the classifications, spatial extent and the different thematic
domains are considered. Some stratifications are not purely environmental classifications as
they incorporate also administrative or socio-economic data to suit the user’s needs. The
stratifications are described and compared regarding (i) the themes and number of dimensions
(represented by the so-called basic datasets); (ii) the spatial and temporal scales of their
dimensions; and (iii) the modes of processing the basic data (e.g., statistical methods and
expert judgment).
Results and discussion
The Environmental Stratification of Europe (EnS) consists of 84 strata, which have been
aggregated into 13 Environmental Zones (EnZ) (Metzger et al., 2005). The stratification has a
1km2 resolution. The EnS has been constructed using tried and tested statistical procedures.
Climate data, data on geomorphology, oceanity and northing (latitude) are data used in the
statistical approach. It forms an appropriate stratification for stratified random sampling of
ecological resources, the selection of sites for representative studies across the continent and
for the provision of strata for modelling exercises and reporting at European scale.
The Agri-Environmental Zonation (AEnZ) is meant to stratify Europe on main biophysical
factors that determine the agronomic production capacity in Europe. It is an agrienvironmental framework used for modelling within the SEAMLESS project (Van Ittersum et
al., 2008). The environmental zones (EnZ) are combined with organic carbon topsoil data
(OCTOP) to cover the wide range of agri-environmental diversity of Europe. Furthermore the
EnZs/OCTOP land units are combined with 270 administrative NUTS2 regions into 3513 socalled Seamzones (Hazeu et al., 2006).
The aim of the European Landscape Map (LANMAP2) is to distinguish different
landscape types, their geographical distribution and their key characteristics. It is used as a
tool to overarch fragmented and yet integrate relevant regional and national approaches.
LANMAP2 was produced on basis of state of the art technology and four core data layers
with a high spatial resolution: (i) climate, (ii) altitude, (iii) parent material and (iv) land use.
The European Landscape Classification is a hierarchical classification. Level one is based on
climate only and has eight classes. Level two is based on climate and altitude and has 31
352
Session B7: Databases for integrated assessment of agriculture and rural development
classes. Level three is based on climate, altitude and parent material and has 76 classes. Level
four is based on all four data layers and is the most detailed level and has 350 landscape types
(Mücher et al., 2006).
The Spatial Regional Reference Framework (SRRF) was developed in the FP6 IP
SENSOR project (Renetzeder et al., 2008). The framework stratifies European territory into
27 relatively homogeneous clusters, integrating biophysical (climate, elevation and parent
material), socio-economic (land cover and ESPON/EUROSTAT data) and regional
administrative (NUTS-X1) aspects. It supports the regional assessment of the impact of
European policies affecting the sustainability of the land.
The FP6 FARO-EU project has developed a new typology to describe the different
ruralities of EU27 (Van Eupen et al., in prep.). It has three dimensions, i.e. one biophysical
(13 EnZ described above) and two socio-economic (economic density and accessibility). It is
the first rural typology with a high spatial resolution (1 km2) that can be aggregated to any
administrative level. The FARO-EU typology provides European rural policies with a flexible
framework to analyse current trends as well as future projections, and to support flex policy
rural development.
The classifications have the climate dimension in common. Furthermore, data on
geomorphology (altitude, slope), soil data or parent material and sometimes land cover data
are important to refine the stratifications. Spatial and thematic detail of the stratifications is
determined by the datasets. The stratifications serve the need to have a spatial framework that
describes systematically the variation in environment and socio-economic issues. Such spatial
frameworks are needed to assess impacts of policies and monitor changes. The different
stratifications serve different objectives and the usefulness of a stratification depends on the
user needs.
References
Hazeu, G.W., et al., 2006. Regional typologies of ecological and biophysical context,
SEAMLESS Report No.14, SEAMLESS integrated project, www.seamless-ip.org, 55 pp.
Jongman, R., et al., 2006. Landscape Ecology 21: 409-419.
Metzger, M.J., et al., 2005. Global Ecology and Biogeography 14: 549-563.
Mücher, C.A., et al., 2006. In: Eds. R.G.H. Bunce & R.H.G. Jongman, Landscape Ecology in
the Mediterranean: Inside and outside approaches. Proceedings of the European IALE
Conference 29 March - 2 April 2005 Faro, Portugal. IALE Publication Series 3, pp. 233243.
Renetzeder, Ch., et al., 2008. In: Eds K. Helming, et al., Sustainability impact assessment of
land use changes. Springer-Verlag, pp. 249-268.
Van Eupen, M., et al., (in prep.). A high spatial resolution typology of rurality in Europe.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
1
NUTS (Nomenclature of Units for Territorial Statistics) is a hierarchical classification of areas that provides a
breakdown of the EU’s economic territory. NUTS-X is a combination of NUTS-2 and NUTS-3. The NUTS-X
coverage has been prepared following an initial approach by EEA within the IRENA project framework, and
extended to EU-25 + Bulgaria and Rumania + Norway and Switzerland by the M3 Alterra team. The result is a
NUTS-X map with 475 units.
353
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Detailed spatial distribution of farming systems in Europe:
A basic starting point for the further environmental assessment of changes
in farming and CAP reform
M. Kempen, B.S. Elbersen, I. Staritsky, E. Andersen
Institute for Food and Resource Economics, University Bonn, Nussallee 21, Bonn, Germany
Contact: markus.kempen@ilr.uni-bonn.de
Introduction
Because of the regional variation in climate, natural resources (soils, vegetation, etc.) and
social structures, the increasing move towards de-centralization of policy implementation and
reorientation of the Common Agricultural Policy (CAP) to deliver more environmental
benefits, there is an increasing need to appraise the multi-functional effects of agriculture at a
range of scales. This integrated and multi-scale approach requires the use of farm information
that is as spatially explicit as possible as this enables to relate market response behaviour to
environmental performance of farms. In this context, a methodology was developed to
spatially allocate farm information to a specific environmental endowment. At this moment
the EU farm information is only available at the administrative level of the regions (NUTS 1
or NUTS 2 regions). The spatial allocation approach adds a spatial dimension to all individual
farms contained in the Farm Accountancy Data Network (FADN) making it possible to
aggregate these both to natural and to administrative regions.
This spatial dimension is a reference to biophysical units with relatively homogenous
conditions for farming, either a Homogenous Spatial Mapping Unit (HSMU) or a Farm
Mapping Unit (FMU) (a cluster of HSMUs). Since HSMUs can be clustered to any
administrative or biophysical entity the farms can also be grouped to these different spatial
entities. For the presentation of the allocation results we have chosen to group the farms to
Agri-environmental zones (Hazeu et al., 2006).
Methods
The allocation approach of FADN farm types builds on the methodology developed in the
CAPRI-Dynaspat project (Kempen et al., 2006) and uses the detailed land use maps resulting
from this project as a main input. The Dynaspat approach disaggregates the crop information
in the Farm Structural Survey (FSS) from the NUTS 1/2 regions to the much smaller
Homogeneous Spatial Mapping Units (HSMUs) by developing allocation algorithms in a
statistical procedure. First a locally weighted logit model (Anselin et al., 2004) is applied to
analyse crop observations from the LUCAS survey (EC, 2003). Then a Bayesian highest
posterior density estimator is applied to achieve consistency between the prior probabilities
resulting from the previous step and regional statistical data. The HSMUs are defined by
relatively homogeneous production conditions for agriculture, e.g., soil conditions (FAO,
1998; Driessen et al., 2001) or prevailing land use (Gallego, 2002); rather than administrative
boundaries. For the spatial allocation of the FADN farm information the land use information
and other attributes assigned to the HSMUs in the Dynaspat project are taken as the main
input basis. The methodology for the farm allocation (e.g., Elbersen et al., 2006) is very
similar to that used for producing the land use allocation in Dynaspat. The main difference is
however, that instead of using the HSMUs as the basic spatial entities to which farms are
allocated a clustering of HSMUs, so-called Farm Mapping Units, are used. This clustering is
necessary to reduce the complexity of the allocation procedure. The final allocated results are
still linked back to the original HSMUs of which the FMUs are composed. For the
presentation of the results in this paper the FADN-farms who have obtained a spatial
dimension through a link to an FMU and an HSMU are aggregated to farm types (see
Andersen et al., 2006) and then to larger agri-environmental zones.
354
Session B7: Databases for integrated assessment of agriculture and rural development
FADN records report for each sample farm whether it is located in a specific altitude zones
and in a Less Favored non-Less Favored Area (LFA). Following HSMUs are aggregated into
FMUs, which are homogeneous regarding soil type, Nuts2/3 boundaries, altitude zone and
Less Favored Area classification. Finally each sample farm is allocated to a FMU achieving
the best possible fit on:
• Information on LFA and altitude zones;
• Cropping patterns of FMUs and farms belonging to it;
• Yield levels of FMUs and farms belonging to it;
• Farm structure information at Nuts2/3 level
Applying a Bayesian Highest Posterior Density method allows to find the optimal allocation
of farms based on information available at different spatial levels.
Results and discussion
The results of the disaggregation approach for FADN farm information are delivering good
results in terms of validation at least for the old Member States. However, for the allocation
results of the FADN farms for the new MS the validation delivered lower probabilities. This
lower accuracy is not necessarily related to the quality of the allocation procedure, but more
likely to the quality of the FADN data used in the allocation. The FADN sample size in the
new Member States is significantly smaller than in the old Member States and leaves out
many small farms.
WHEA 1
M AIZ 1
3.0
3.0
2.5
2.0
1.5
2.0
1.5
2.5
1.0
0.5
FMU1
3t wheat
7t maiz
FMU2
7.5t wheat
2.5t maiz
1.0
0.5
0.0
0.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
MAIZ2
0.6
0.8
1.0
0.6
0.8
1.0
WHEA 2
2.5
2.5
2.0
2.0
1.5
1.5
1.0
1.0
0.5
0.5
Figure 1. Using
Probability Density
Functions
for
matching the yields
on farms with FMU
yields.
0.0
0.0
0.0
0.2
0.4
0.6
0.8
1.5
0.0
1.0
*
0.2
0.4
1.33 = 2.00
FARM 1
FARM2
FARM3
FARM4
Wheat
6
6.5
3
4.5
“Probability density
Yield
Maiz
4
3.5
7
5.5
π PDF1
0,23
0,07
5,30
2,42
PY f , fmu = ∏ pdf fmu ,c ( sc )
π PDF2
2,00
2,59
0,05
0,50
c
References
Andersen, E., et al., 2006. SEAMLESS report no. 12, May 2006.
Anselin, L., et al., 2004. Advances in Spatial Econometrics. Springer Verlag, Berlin.
Driessen, P., et al., 2001. World Soil Resources Reports 94. FAO, Rome.
EC, 2003. The Lucas survey. European statisticians monitor territory. Theme 5: Agriculture
and fisheries, Series Office for Official Publications of the EC, Luxembourg, pp. 24-40.
Elbersen, B., et al., 2006. SEAMLESS report no. 19, November 2006.
FAO, 1998. World Reference Base for Soil Resources. World Soil Resources Reports 84.
FAO, ISRIC and ISSS.
Gallego, J., 2002. Fine scale profile of CORINE Land Cover classes with LUCAS data, in
European Commission (ed.): Building Agro Environmental Indicators. Focussing on the
European area frame survey LUCAS, Vol. EUR Report 20521 EN.
Hazeu, G.W., et al., 2006. SEAMLESS report no. 14, June 2006
Kempen, M., et al., 2006. A statistical approach for spatial disaggregation of crop production
in the EU. University of Bonn and JRC-Ispra.
355
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Need for a theoretical framework in building an integrated database
P. Lavelle, X. Arnauld de Sartre, S. Dolédec, B. Hubert, I. Veiga
Université Paris VI, UMR Biodiversité et Fonctionnement du Sol,
Centre IRD Ile de France, 32 rue Henri Varagnat,
93143 Bondy Cedex, France
Contact: Patrick.Lavelle@bondy.ird.fr
Introduction
Our goal in this communication is to discuss our experience in building a database that aims
to integrate socio-economic, environmental and agronomical dimensions of family farming in
the Amazon.
In two Amazonian regions that represent most of the diversity of man made landscapes in
the ecoregion, we aimed to analyse and model the relationships that link (1) socio-economic
parameters; (2) composition and structure of landscapes; (3) biodiversity (of plants, birds,
Drosophilidae, Lepidoptera Saturnidae, ants, termites, earthworms), and (4) crop production
and other ecosystem goods and services (EGS) used by family farmers.
After presenting the method used to build a common database, we will stress the need to
clearly define, at the same time as building the database, a theoretical framework that can help
to analyse the data collected.
Methods
In two Amazonian countries (Brazil and Colombia), an original sampling protocol allowed us
to collect fully compatible socio-economic, landscape, agronomic and ecological datasets in
order to measure and understand the variations of EGS among family farms. Those datasets
are supposed to allow rigorous statistical analyses, in order to formulate new hypotheses on
the mechanisms that link these different sets of variables.
Work has been divided in five groups:
- WP1 - SOCEC: assessment of relevant socio-economic parameters;
- WP2 - LANDSCAPE: a quantitative analysis of the composition and structure of the 6
landscape «windows» in the two countries;
- WP3 - BIODIVERSITY: an assessment of the diversity of seven different groups of
ecosystem engineers that matter for EGS production (plants, earthworms, termites, ants)
and/or species that indicate specific trends and responses in patterns of biodiversity (birds,
Saturnidae, Drosophilidae);
- WP4 - ECOSYSTEM SERVICES: a quantification of a few basic EGS: soil services (C
storage, chemical fertility, water infiltration and storage), production of food, fiber and fuel
and other products, landscape control of disease (Chagas);
- WP5 - MODELLING: identification of the statistical relationships that link (1) socioeconomic context to (2) composition and structure of landscape, (3) biodiversity for the
seven groups and (4) provision of EGS. These relationships will be used to simulate the
consequences of different socio-economic scenarios using a multi-agent modelling
approach.
Groups 1 to 4 have collected their own datasets, which can be linked to one another, as
indicated in Figure 1.
356
Session B7: Databases for integrated assessment of agriculture and rural development
ENVIRONMENTAL
drivers
SOCIAL
drivers
ECONOMIC
drivers
3
2
Land Use Practices
1
LANDSCAPE
6
7
4
BIODIVERSITY
5
ECOSYSTEM GOODS
AND SERVICES
Figure 1. A simple empirical model of the impact of socio-economic changes on ecosystems
and the production of ecosystem goods and services
Data collection was organized in two phases:
- 150 farms in three windows with different land use histories and practices have been
characterized on the basis of socio-economic and landscape variables.
- A sample of 27 farms in each country, representative of the diversity of farms in the three
windows of the country, was selected. Biodiversity and EGS production have been
measured in each farm in these samples.
Due to this sampling method, which allows us to work in different scales, and to the pluridisciplinary structure of our team, we needed to define a common theoretical framework to
build and analyse the database containing all the datasets collected.
Results and discussion
The database, currently under construction, will be presented at this session.
Although statistical analyses enabled us to identify links among the different sets of
variables, to separate variables with significant co-variations and threshold effects in their
relationships, we found that it was necessary to develop a conceptual framework to guide the
analysis. To illustrate the need for a framework, we will present an example of an analysis of
a part of the data.
This discussion is based on the Panarchy Theory (Gunderson & Holling, 2002), which
served as a basis for our modelling. During our seminars, we suggested that different team
members define their own vision of the database organization based on this theoretical
framework.
Reference
Gunderson, L.H. & C.S. Holling (Eds), 2002. Panarchy. Understanding Transformations in
Human and Natural Systems. Island Press, Washington-Covelo-London, 438 pp.
357
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Some reflections on the design and use of typologies of EU regions
in socio-economic analyses
Ida J. Terluin, David Verhoog, Frans E. Godeschalk
LEI, Agricultural Economics Research Institute,
P.O. Box 29703, 2502 LS The Hague,
The Netherlands
Contact: ida.terluin@wur.nl
Introduction
Typologies of EU regions can help to identify regions with similar characteristics. As such,
they may facilitate the study and interpretation of types of regions. In literature, it is suggested
that a number of questions should be taken into account in the design of a typology (Berry &
Smith (1972) cited in Guffens & Latten, 1979; Hauer & Van der Knaap, 1973). These
questions include among others:
(a)
What is the purpose of the typology?
(b)
What is the individual unit in the classification?
(c)
What are the differentiating characteristics on which the typology is to be based?
(d)
Which method will be used for the classification?
(e)
How many types will be distinguished in the typology?
In these questions, no attention is paid to problems which may arise due to lack of data. Such
problems might play in particular a role in the definition of the individual unit in the typology
and the selection of differentiating characteristics. This may result in ‘second best’ typologies,
using less appropriate individual units or differentiating characteristics. In rural development
analyses these problems often arise, especially when data from different data sources, like
Eurostat REGIO, EU Farm Structure Survey and the Farm Accountancy Data Network, are
integrated.
The objective of this paper is twofold. First, the design and use of typologies in recent
studies on socio-economic analyses in EU regions is discussed. Second, it is explored to
which extent these typologies can be used for impact assessment of agricultural and rural
policies in the EU (Van Ittersum et al., 2008).
Methods
For the discussion of regional typologies, a framework with questions is developed. This
framework includes the questions (a)–(e) above, questions on used data, questions on the
integration of different data sources and questions whether the typology has been used for
impact assessment of agricultural and rural policies in the EU.
The set of regional typologies, that will be discussed, covers among others the OECD rural
typology (OECD, 1996), a typology based on GDP/capita, a typology derived from the
integration of regions in the national/global economy (EC, 1997), a typology of leading and
lagging regions (Terluin, 2003; Bollman et al., 2005), an LFA typology (ESPON 2.1.3
project, 2004; Terluin et al., 1994), a typology based on population change (Johansson and
Rauhut, 2005) and a typology of agricultural regions (ESPON 2.1.3 project, 2004).
Results and discussion
Based on the discussion of the different regional typologies in this paper, we hope to identify
a number of regional typologies that are useful for impact assessment of agricultural and rural
policies in the EU. On the whole, such typologies have in common that they are simple,
highly transparent and easy to understand.
358
Session B7: Databases for integrated assessment of agriculture and rural development
Table 1. Useful typologies of EU regions for impact assessment of agricultural and rural
policies.
Name of typology
Regional level
Policy
Typology x
NUTS ..
LFA
Typology y
NUTS ….
Modulation
etc
etc
etc
References
Berry, B.J.L. & K.B. Smith, 1972. City classification handbook: Methods and applications.
John Wiley Interscience, New York.
Bollman, R., et al., 2005. Paper for the Congress of the European Regional Science
Association ‘Land Use and Water Management in a Sustainable Network Society’, Vrije
Universiteit Amsterdam, 23-27 August 2005.
ESPON 2.1.3 project, 2004. The territorial impact of the CAP and Rural Development Policy.
www.espon.lu
European Commission (EC), 1997. Rural developments: Situation and outlook. CAP 2000
Working Document, Brussels.
Hauer, J. & G.A. Van der Knaap, 1973. Sociale geografie en ruimtelijk onderzoek,
kwantitatieve methoden (Human geography and spatial analysis; statistical methods).
Universitaire Pers, Rotterdam.
Johansson, M. & D. Rauhut (Eds), 2005. The spatial effects of demographic trends and
migration. Final report ESPON project 1.1.4, Stockholm.
OECD, 1996. Territorial indicators of employment: Focusing on rural development. Paris.
Terluin, I.J., 2003. Journal of Rural Studies 19(3): 327-344.
Terluin, I.J., et al., 1994. The agricultural income situation in Less Favoured Areas of the EC.
Luxembourg/Brussels.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
359
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme C
Case studies and application of tools and
empirical methods
Session C1: Integrated assessment of biomass for biofuel
or renewables: Trade-offs with feed and food
Session organizers:
Martin Banse
LEI, Wageningen UR, The Netherlands
Ken Moore
Iowa State University, USA
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Challenges in modelling environmental bio-energy potential(s)1
Jan-Erik Petersen
European Environment Agency,
Kongens Nytorv 6,
1050 Copenhagen K, Denmark
Contact: Jan-Erik.Petersen@eea.europa.eu
Introduction
The presentation summarizes experience from EEA projects and information gathered
through a number of expert workshops to describe the main challenges when aiming to model
the bio-energy potential of a given geographic region. Particular attention is given to the
range of environmental factors to be considered, the analysis of scale effects and the impact of
indirect land use change on total environmental potential. The discussion of modelling frames
includes questions relating to governance approaches to improving the environmental profile
of bio-energy production.
Methodological considerations in modelling bio-energy production
Bio-energy production has potentially significant effects on energy and agricultural markets,
it implies considerable changes in the nature and intensity of global land use and leads to
often strong environmental impacts that can be negative or positive. Analysing and modelling
the overall effects of bio-energy policies is therefore a very complex exercise that ideally
involves a combination of (bio-)energy, agro-economic, global land use and biophysical
modelling approaches. Such an integration has until now not really been achieved and there is
still a considerable need to analyse and develop approaches for model integration.
Given the focus of the conference the current review is mainly limited to (agro-)economic and
land use models. The multiple pathways and technologies for utilizing biomass in energy
production are a complex modelling field on its own but not covered in detail in this review.
Within the modelling approaches discussed an attempt is made to cover the following issues:
•
•
•
•
1
Key characteristics are the regional disaggregation (grid cells, NUTSx, countries,
country aggregates), regional coverage (region, country, Europe, world), time horizon
in applications, and the drivers (economic, political, technical) of a model.
Relevant environmental impacts include total greenhouse gas balance (before and after
taking indirect effects into account) and a number of environmental issues relating to
type and intensity of energy cropping and associated land use change. These relate to
soil erosion, quality and quantity of water in affected aquifers and surface waters,
landscapes and biodiversity.
Biomass production details provide crucial information for an environmental
assessment. Important questions are, for example, types of land use changes (within
arable land, between permanent grassland or forestry and arable land) farming system
properties (farm type, rotation, land cover in winter, fertilizer use, irrigation) and the
area of set aside and fallow land.
Regarding energy crops a rich coverage of biomass uses is of interest and hence the
alternative uses for food, feed, energy and manufacturing, possibly differentiated
(transport fuels, heating, electricity). For each processing alternative technical details
may be crucial for the economic and environmental balance (by-products, economies
of scale in processing, logistic constraints render location issues crucial).
Keynote presentation
362
Session C1: Integrated assessment of biomass for biofuel or renewables
As biofuels in particular are under close environmental scrutiny several policy measures have
been proposed for improving their greenhouse gas balance and minimizing other
environmental impacts. The most prominent approach is the introduction of so-called
‘sustainability standards’ but other options also exist, e.g., differentiated incentives in
electricity feed-in schemes, producer training and advice etc. Current agro-economic and land
use modelling approaches do not yet take such policy measures into account. This will be
necessary, however, if one wants to determine the likely environmental impact of a given bioenergy production approach and/or policy. On the other hand, (agro-)economic and other
models can also help determine the likelihood of success for different bio-energy policy
instruments in limiting environmental impact. The final section discusses, therefore, options
for integrating certain policy approaches into different modelling frameworks and proposes
some key questions to be addressed.
Documents used:
Petersen, J.-E., 2008. Informal consultation meeting between OECD, EEA and invited
technical experts: Linking economic and biophysical modelling in relation to bio-energy.
Summary document published on the web: http://bioenergy.ew.eea.europa.eu/
Britz, W., et al., 2008. Review of European spatial modelling approaches for assessing the
implications of bioenergy production in an environmental perspective. Unpublished report
to the European Environment Agency, Copenhagen.
363
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Life cycle assessment of the sugarcane production system in Brasil
Paola J. Bocardo, Kazumi Kondoh, Juliana C.S. Cruz, Dorothea S. Kimura
Tokyo University of Agriculture and Technology,
3-5-8 Saiwai-cho, Fuchu, Tokyo 183-8509, Japan
Contact: pjbocardo@yahoo.com.br
Introduction
The area allocated to sugarcane in Brasil for the crop-year 2008/09 is around 9 million of
hectares and approximately 559 million tonnes of cane will be crushed to produce alcohol and
sugar (SIAMIG, 2008). Part of this amount is harvested by hands after burning leaves,
releasing to the environment a high quantity of ash, potentially causing serious disease
problems for the population, resulting in social and environmental problems (Campos, 2003).
The harvest by hands has been substitute by machines, due to economic, legal and
environmental issues (Galdo, 2007). Thus, life cycle assessment (LCA) can be utilized to
analyse the whole life cycle of the Brazilian sugarcane system. The objectives of this study
were: (1) to analyse the dioxide carbon (CO2) emission during sugarcane production in Brasil,
comparing burning and green harvest without burning, and (2) to compare the socio-economic
and environmental aspects of these harvesting systems. Based on these analyses, the
cultivation method with the least CO2 emission and the best socio-economic aspects will be
discussed.
Material and methods
System boundaries and functional unit
The sugarcane system flowchart involved cultivation (land preparation and growth
management), harvest and transport. The processes considered for the cultivation were: (1)
land preparation (lime, gypsum, fertilizer and defensive applications, planting and covering);
(2) growth management (cover fertilizing, defensive, irrigation and mill mud application); (3)
harvest (burning and green cane); and (4) transportation (inputs, cane and workers) (Figure 1).
The functional unit was 1 t of raw cane.
Burning
cane
Land
preparation
Input
Planting
Cover
fertilizing
Harvest
Green
cane
Figure 1. Flowchart for the life cycle of sugarcane production system.
Inventory analysis
The data collection included the input and output of the sugarcane cultivation, harvest and
transport. Model, fuel consumption, wagon/passenger’s capacity, performance and distance
were considered for the data of machines/implements, trucks and buses for labour transport.
For lime, gypsum, fertilizers and pesticide application, the kind, amount/ha and times of
application/year were considered. The climate conditions were related to the air and soil
characteristics, as humidity, type of soil, nutrient leaching, gas emissions, temperature and
precipitation. For the social data, the kind of work, cost/time, numbers of workers required in
each stage, performance and level of education were collected.
364
Session C1: Integrated assessment of biomass for biofuel or renewables
Data sources and methods
Information about production was obtained from private sugarcane companies, institutes and
specialist’s opinions in universities. For the carbon (C) balance calculations each component
of this system was considered separately. The emissions from machines, trucks and burning
were converted into CO2 equivalents by calculating the consumed fuel for each management
practice. In case no data were available, it was obtained from the literature. The worker’s
condition was analysed descriptively and the total of workers required to produce 1 t of raw
cane related to social aspect. For the economic aspect, the cost to produce 1 t of raw cane was
calculated.
Selection and definition of impact categories
Considering the C stock compartments (plants, trash and soil) , the C storage in the plants by
photosynthesis was 5.7 t C ha–1, trash 0.5 Mg C ha–1 yr–1 and soil 1.0 Mg C ha–1 yr–1.
Generally, the green cane system can mitigate 5 Mg C-CO2 ha–1 yr–1 in a period of 3 years
(Campos, 2003). This mitigation potential can be changed, if the fossil fuel used by the
harvesting machine and the vehicles are considered. The social conditions are different in
both systems. It has been reported that one machine substitutes 100 workers (Ustilin &
Severo, 2001). Field work under hot conditions can cause severe health problems. The
labour’s salary depends on the amount of cane harvested in tones per day, inducing
exhaustive work. Thus, the working conditions for the burning system should be changed
where the burning harvest is inevitable.
Conclusion
About 45% (4 million of hectares) of the total sugarcane area in Brasil requires burning at
harvest due to topography. This can aggravate the environment and socio-economic aspects in
sugarcane production system, such as carbon sequestration and working conditions. A holistic
analysis considering carbon flows as well as socio-economic aspects is needed to create a
sustainable system.
References
Campos, D.C., 2003. Potencialidade do Sistema de Colheita sem Queima da Cana-de-acucar
para o Sequestro de Carbono. Doctor Thesis, Universidade de Sao Paulo, Brasil, 103 pp.
Galdo, M.V., 2007. Dinamica do carbono no solo no agroecossitema cana-de-acucar. Doctor
Thesis, Universidade de Sao Paulo, Brasil, 101 pp.
SIAMIG, 2008. Safra da cana-de-acucar 2008/09. Boletim eletronico numero 288.
www.siamig1.com.br/index.php?option=com_content&task=view&id=275&Itemid=107.
Accessed Sept. 09, 2008.
Ustilin, E.J. & J.R. Severo, 2008. Proteger o ambiente e continuar gerando empregos.
www.cna.org.br/Gleba99N/Set01/cana01.htm. Accessed Sept. 13, 2008.
365
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Application of Life Cycle Assessment methodology to compare sunflower
oil and diesel for fulfilling farm energy needs
L. Coletto, A. Brigi, A. Calgaro, S.G. Sandrini, S. Bona
Dipartimento di Agronomia Ambientale e Produzioni Vegetali,
AGRIPOLIS - Facoltà di Agraria - Università di Padova,
Viale dell’Università n. 16 - 35020 Legnaro (PD), Italy
Contact: lucia.coletto@unipd.it
Introduction
The recent European policy is aimed to develop energy production systems that minimize
environmental impact and to promote the production of biomass for energy purposes (oil
crops included) (CEC, 2001). Among these, of particular relevance in Italy is sunflower oil.
Sunflower is particularly worth to be studied as it is well adapted to Italian conditions and
requires very low inputs. Furthermore, the chemical/physical properties of sunflower oil
affect its performances if used as fuel in endothermic engines (Schlick et al., 1988).
Vegetable oils are interesting alternatives to fossil fuel owing to their relatively low
production costs, easiness to produce and, therefore, suitable in situations where homeproduced energy at low cost is required (Al-Hasan, 2002; Scrosta, 2005; Riello & Bona,
2005).
Life Cycle Assessment (LCA) is a methodology to assess the environmental impact
associated with a product, an activity or a process. Nowadays LCA is mainly applied to
industrial products or processes, but its use is increasingly widespread in agriculture.
In this study, the suitability of LCA methodology to analyse the environmental impact of
the use of sunflower oil on the farms to meet their internal energy requirements is
investigated. LCA is used to analyse the effective possibility of using sunflower oil instead of
diesel and to estimate the consequent changes in farm organization.
Methods
The case study consisted of 19 simulated farms of 10 ha differing in crops (maize, soybeans,
wheat) and in hectares for each cultivated species: 3 farms had monospecific system while the
other 16 differed for percentage of land occupied by the over cited crops (e.g., 7.5 ha of maize
and 2.5 ha of soybeans). In order to obtain the necessary oil supply, during the study the crop
distribution was rearranged assigning land to grow sunflower (e.g., 7.5 ha of maize and 2.5 ha
of soybeans become 4.7 ha of maize, 1.6 ha of soybeans and 3.7 ha of sunflower). The crops
were transformed into energy data using the parameters defined by Riello & Bona (2005).
The sunflower area was defined on the farm’s energy requirement basis, whereas the land
devoted to the other crops reflected, where possible, the percentage distribution before the
introduction of sunflower. We also hypothesized that all the farm tractors were converted for
the use of sunflower oil as fuel. The two managements (sunflower oil vs. diesel) were
compared with LCA methodology using SimaPro 7.18 based on Eco-indicator 99 methods.
The LCIs (Life Cycle Inventory) included soil cultivation, sowing, weed control,
fertilization, pest and pathogen control, harvest and drying of the grains, processes. Machine
infrastructure and shed were included. Inputs of fertilizers, pesticides and seed as well as their
transports to the farm were also considered. The direct emissions from the field were also
included in the inventories. The system boundary was at the farm gate. Sunflower oil
production process was included in machine inventory. The Project was financed by Progetto
FISR-SIMBIO-VEG.
366
Session C1: Integrated assessment of biomass for biofuel or renewables
Results and discussion
The resulting index, eco-indicator 99 value, evaluated the impact associated with different
energy source (sunflower oil vs. diesel) on the following environmental effects: production of
carcinogens, respiratory organics and inorganics, effect on climate change, radiation, ozone
layer, acidification/eutrophication, minerals and fossil fuels consumption. Those effects could
be grouped in three categories: Human Health, Ecosystem Quality, Resource. Figures 1a, 1b,
1c and 1d represent the relative advantage/disadvantage of the conversion of the considered
farms from fossil fuels to sunflower oil.
Human Health
Human Health
7E-02
W100
W100+SF
5E-04
6E-02
M100
M100+SF
4E-04
5E-02
S100
S100+SF
4E-02
3E-02
(DALY)
(DALY)
(a)
3E-04
(b)
W100
W100+SF
M100
M100+SF
S100
S100+SF
2E-04
2E-02
1E-04
1E-02
0E+00
0E+00
Carcinogens
(c)
Respiratory
inorganics
Respiratory
organics
Climate change
Ozone layer
(d)
Resource
1600
W100
W100+SF
35000
W100
W100+SF
1400
M100
M100+SF
30000
M100
M100+SF
1200
S100
S100+SF
25000
S100
S100+SF
(MJ Surplus)
(PDF)
Ecosystem Quality
Radiation
1000
800
600
20000
15000
400
10000
200
5000
0
0
Ecotoxicity
Acidification/ Eutrophication
Minerals
Fossil fuels
Figure 1. Impacts of maize (M), soybeans (S) and wheat (W) cultivated with diesel machinery
compared to the same crops cultivated with sunflower (SF) oil machinery. Figure a and b
represents human health’s impact (data expressed as years of life lost by humans); figure c the
ecosystem quality (data expressed as biodiversity lost) while figure d represents the use of
resources (MJ per hectare).
The higher carcinogens impact observed in crops cultivated with sunflower oil machinery was
due to the introduction in the farm of a sunflower mill produced by a factory with emissions
in all the compartments. In general, allocating a portion of farm land to cultivate sunflower
for producing oil for energy was environmentally positive, in particular if the crop substitutes
maize. The main environmental disadvantages were found for wheat while soybean had an
intermediate impact.
References
Al-Hasan, M., 2002. Biomass and Bioenergy 23: 381-386.
CEC, 2001. Commission of the European Communities, COM (2001) 68 final.
Riello, L. & S. Bona, 2005. Ed. Veneto Agricoltura, Padova, 80 pp.
Schlick, M.L., et al., 1988. Trans. ASAE, 31, 1345-1349.
Scrosta, V., 2005. PhD thesis, University of Marche, Italy, 106 pp.
367
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Family production systems with dende (Elaeis guineensis) and the
alternative of biodiesel:
Analysis in the Cajaíba community, municipality of Valença (Bahia-Brasil)
Alynson Dos Santos Rocha, Maciene Mendes da Silva
Universidade Federal da Bahia, Rua Jose Seabra s/n, antigo Colégio Padre Vieira,
Centro Barreiras, Bahia - Brasil
Contact: alynson@bol.com.br
Introduction
In this paper, family production systems with dende (Elaeis guineensis) in the Cajaíba
community, municipality of Valença (2006), state of Bahia (Brasil), are compared with the
alternative of biodiesel. The analysis focused on the dynamics of production considering the
non-agricultural activities that have become alternatives for the family farmers.
Biodiesel – biofuel produced from vegetal raw-materials – is presented as an energy source
that has few environmental impacts when compared to non-renewable fuels (oil), due to the
possibility of a reduction in pollutants emitted from automobiles equipped with combustion
engines. Concomitantly, economic and social advantages are posited when the diversity of
crops is valued with benefit potential, especially for family farming. Regarding the production
of vegetable oil crops (for example: soy – Glycine max; mamona – Ricinus communis; babaçu
– Orrbignya speciosa; and dende), the gradual insertion of biodiesel might mean an important
step to the economic development of Brazilian regions such as the semi-arid (north-eastern)
and other areas of family farming in the state of Bahia, that produces some of main raw
materials for biodiesel (Garcia, 2008).
Specifically for Bahia, dende farming appears to have significant potential for biodiesel
production due to the climate and soil conditions conducive to crop development, particularly
in the South. It is added to the fact of a new economic alternative to local family farming,
considering the productive dende chain (still) narrowly linked to oil production for the food
industry (Bahia, 2005). In this context, the National Program of Production and Use of
Biodiesel (PNPB) – initiative of the Federal Government of Brasil, created in 2004, combines
an attempt to diversify the Brazilian energy matrix with tax, credit and marketing incentives,
whose objective is to implement the production and use of biodiesel, at the same time
promoting socioeconomic development – must consider the organization of the productive
agents widely, the traditional techniques already used by the primary producers, the
importance of the dende farming to the productive units and the local prices received and the
dynamics of the market faced by the family farmers (Chiaranda et al., 2005).
Methods
The Agrarian Systems methodology is applied on the case study in the Cajaíba community.
This methodology analyses not only the economic relations concerning to the production
systems; the social and ecological aspects are analysed as well, and mainly the relations
between them. This systematic approach is used in Brasil by the National Institute of
Colonization and Agrarian Reform (INCRA) in its reports about rural settlements, but its
applicability is broader, and may also be used on the evaluation of diverse types of production
systems such as capitalist and family production units, as well as others. The method can be
summarized as follows: (a) analysis of secondary data on the agrarian systems; (b) landscape
characteristics; (c) characterization of farmers, unit and systems of production (crop, creation
and raw-materials transformation subsystems); (d) elaboration of typology of producers and
production systems; (e) crop subsystems characterization; (f) creation subsystems
characterization; (g) transformation subsystems characterization; (h) combination of the
368
Session C1: Integrated assessment of biomass for biofuel or renewables
subsystems; and (i) economic evaluation of the production systems in order to identify to the
family farmers income composition (Brasil, 2008).
Results and discussions
The results achieved indicate that in all types of family farming production systems studied
the dende subsystem not always is the most important crop in the farmer’s income
composition. Also, the production analyses have revealed that the dende might potentially be
raw-material to the biodiesel industry. One of positives aspects that confirm this observation
is the high degree of oil per hectare when dende is compared to other vegetable oil crops.
However, some impediments had been observed: not addition of values, due to the frequent
traditional aspect of the production; useful life of the dendezeiros (dende trees) exceeded,
causing low productivity; small scale producers without the possibility of expansion;
existence of more lucrative alternatives, such as the dende oil to the food industry.
Regarding to the viability of dende as raw-material for biodiesel some questionable points
emerge. The oil price practised, for example, on the first auction to the PNPB, R$ 102.50
(Brazilian currency) per ton of fresh product (in natura, entire), results in positive incomes to
the producers. However, the market value of these same products when destined for others
ends, approximately R$ 130.00, is higher than biodiesel. Also, to devote the production of
dende only for the PNPB means to give up the local subsystem of transformation (Rodão or
big wheel). This subsystem increases strongly the farmers’ incomes.
Considering oil prices for biodiesel compared to other products and observing the
dynamics of production systems, the PNPB goal of the socioeconomic inclusion of family
farmers would not be feasible because most family farmers are oriented to oil production for
the food industry with traditional techniques and equipment. The unfeasibility is based on the
prices received by family farmers who adhere to PNPB, which are below those realized in the
local-regional market. Additionally the resistance to allocating areas for only one crop, would
generate disequilibrium in the entire productive unit, causing economic losses to the family
farmers.
References
Bahia, 2005. Secretaria de Agricultura, Irrigação e Reforma Agrária (SEAGRI). Culturadendê. Available on: www.bahia.ba.gov.br/seagri/Dende.htm. Accessed may 07, 2005.
Brasil, 2008. Instituto Nacional de Colonização e Reforma Agrária (INCRA). Guia
Metodológico: Análise-Diagnóstico de Sistemas Agrários, 65 pp.
Chiaranda, M., et al., 2005. A produção de biodiesel no Brasil e aspectos do PNPB. Grupo de
Estudos e Extensão em Desenvolvimento Econômico e Social. Departamento de Economia
Administração e Sociologia. ESALQ-USP. Piracicaba (SP), 32 pp.
Garcia, J.R., 2008. O Programa Nacional de Produção e Uso de Biodiesel e a Agricultura
Familiar do Nordeste: Panorama, Perspectivas e Desafios. Workshop FEA/Unicamp –
Produção de Biodiesel Através de Agricultura Familiar. IE/UNICAMP, 15 pp.
Valença, 2006. Memorial da Câmara Municipal de Valença. Available on:
www.cmvalenca.ba.gov.br. Accessed June 30, 2006.
369
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
The availability of crop residues for energy, assessing global potentials
T. Guille, A. Fallot, F. Jacquet, P. Havlík
INRA-Économie publique, Avenue Lucien Brétignières,
78850 Thiverval Grignon, France
Contact: thomas.guille@grignon.inra.fr
Introduction
The research presented aims at globally assessing the potential availability of crop residues
for energy through an updated series of data and hypotheses on the determinants of the
potential volumes. Such systematic work is necessary in a context where second generation
biofuels need to ensure large biomass potentials at low cost. Crop residues are therefore
considered as a privileged resource, without production cost of the feedstock or impact on
competition for land, contrary to dedicated energy crops or plantations. The few potential
assessments that have been conducted at the global level show widely varying results with
diverse definitions of potential availability, use of fixed recuperability factor only
differentiated according to the kind of crop residue (primary or secondary) and little critical
review on the significance of results (Hall et al., 1993; Berndes et al., 2003; Hoogwijk et al.,
2003; Kim & Dale, 2004; Smeets et al., 2007). In this context, our work aims to estimate the
global availability of crop residues for energy and focuses on residues that could be collected
without competing with preservation of soil fertility and animal feeding.
Methods
The updated assessment proposed proceeds in various steps, progressively accounting for
crop characteristics, current agricultural practices and uses of the crop residues. Hypotheses
are made on the priority residue uses for soil conservation and animal feeding. Calculations
are based on most recent FAO statistics available in July 2008 on crop production, residue to
product ratios and estimations of current residue use, through recuperability factors. The
recuperability factors are differentiated according to crop species and residue type using
available knowledge in soil science and others disciplines. An emphasis is put on the level of
extraction of residues, this latter playing a crucial role in the evaluation of the effective
availability of crop residues for a sustainable energy production. Afterwards, a critical
overview of other assessments enables us to compare approaches (methodologies,
hypotheses) and results.
Results and discussion
Results at a world level indicate that agricultural activities generate more than 3.36 billions
tonnes of agricultural residues, representing 54.6 EJ yr–1. Given the hypotheses we detail,
some 21.4% of these residues are potentially available for energy, i.e. 719 million dry tonnes
representing 11.9 EJ yr–1 (approximately 3% of the current final energy demand). The largest
potentials are from wheat and rice straw, bagasse and rice husk. The richest countries in crop
residues are China, India, the US and Brasil (cf. Figure 1). Compared to other assessments,
our estimation stands at lower levels. Differences illustrate the importance of hypotheses on
the priority uses of crop residues. At last, even thought our estimation is low, it shows that the
potential of crop residues for energy clearly exist for a sustainable energy production.
This work is a first step in the evaluation of the availability of crop residue for energy,
based on physical availability. It doesn’t take into account harvesting and transporting costs
and opportunity costs associated with alternative residue uses (soil fertility, soil protection
against erosion, animal feeding, etc.). This work will be done by introducing crop residues in
a global land use model – GLOBIOM (Global Biomass Optimization Model) – developed at
IIASA (Havlík et al., 2008). It will allow us to take into account the competition with animal
370
Session C1: Integrated assessment of biomass for biofuel or renewables
feeding and to provide an analysis on the global supply of crop residues for energy and its
impacts according to different scenarios and policy objectives.
2 000,00
1 800,00
1 600,00
1 400,00
Secondary residue
1 200,00
Primary residue
1 000,00
800,00
600,00
400,00
200,00
ce
Fran
Mala
ysi a
Pakis
ta n
Viet
Nam
Russ
ia
Arge
ntina
T hai
land
Indo
nesia
Braz
il
USA
India
Chin
a
0,00
Figure 1. Countries with the largest potential of crop residues for energy (in PJ).
References
Berndes, G., et al., 2003. Biomass and Bioenergy 25(1): 1-28.
Hall, D.O., et al., 1993. In: Eds T.B. Johansson, et al., Renewable energy – sources for fuels
and electricity. Island Press, Washington, pp. 593-651.
Havlík, P., et al., 2008. Avoided Deforestation and Biofuel Production: A Global Analysis.
Yale Student Chapter of ISTF 2008 Conference, March 27-29 2008, New Haven, USA.
Hoogwijk, M., et al., 2003. Biomass and Bioenergy 25(2): 119-133.
Kim, S. & B.E. Dale, 2004. Biomass and Bioenergy 26(4): 361-375.
Smeets, E.M.W., et al., 2007. Progress in Energy and Combustion Science 33: 56-106.
371
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Soil optimized suitability analysis for allocation of energy grasses
L. Kukk, E. Suuster, A. Astover, M. Noormets, H. Roostalu
Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences,
1 Fr. R. Kreutzwaldi St., 51014 Tartu, Estonia
Contact: liia.kukk@emu.ee
Introduction
In recent years interest in producing bio-energy has increased. Several studies have estimated
agricultural bio-energy potential at global, EU and national scales (Voivontas et al., 2001;
Hoogwijk et al., 2005; Edwards et al., 2005), but final decisions and recommendations about
land use conversion to energy crops must be made at a more detailed spatial scale. Allocation
of land to energy crops should take account of local pedo-climatic conditions. Thus soil-crop
specific models are required for sustainable bio-energy planning. In Estonia, a large-scale
digital soil map is available but is not often used in the decision-making process. The value of
soil information can be realized only when it is actually used but the application of collected
soil information has still been modest because of poor availability, and limited knowledge of
the decision makers. The aim of the present study was to develop and use a model for
allocation of energy grasses based on soil data using the example of one municipality in
Estonia.
Methods
Municipality Aseri is located in north-eastern Estonia. A map layer of agricultural field
parcels (3,029 ha) from the Agricultural Registers and Information Board was used for
analysis. The most important land reserves for bio-energy production in Estonia are
abandoned agricultural areas, which are classed as agricultural land. Therefore, our study
provided suitability maps for total agricultural land and for abandoned fields separately. We
considered that field parcels without any applications for CAP single area payments to be
abandoned. The digital soil map at scale 1:10,000 was used in the current study but the
qualitative nature of information from the soil map makes it difficult for non-soil scientists to
understand and soil suitability is not a characteristic that can be directly read from a soil map.
To overcome this shortcoming a supplementary database of soil mapping units (based on soil
type and texture) was created with crop-specific suitability indexes. These were previously
developed on the basis of perennial biomass yields from a network of field experiments and
from agricultural enterprises (Kask, 1994; Kõlli, 1994). We evaluated areas suitable for reed
canarygrass (Phalaris arundinacea L.), Caucasian goat’s-rue (Galega orientalis Lam.) and
alfalfa (Medicago sativa L.) as these energy crops have been investigated in Nordic
conditions and can grow in a soil conditions that range from Leptosols (dry) to Histosols
(wet). Soils were classified into three categories: not suitable, moderately suitable and
suitable. Thereafter the field boundaries were superimposed on the soil classes and weighted
average suitability scores were calculated for each field parcel. Fields with a weighted
average suitability score less than 1 were considered as unsuitable for the selected energy
crops. Priority was given to the crop with the highest suitability score.
Results and discussion
Of the total agricultural land in Aseri municipality abandoned areas form 416 ha (14%)
(Figure 1). Alfalfa, reed canarygrass and Caucasian goat’s-rue could be grown successfully on
49%, 28% and 2% of the total agricultural land respectively. Alfalfa could be grown on 34%
of the abandoned fields. Similarly, 37% and 5% of the abandoned areas were suitable for
growing reed canarygrass and Caucasian goat’s-rue. Approximately a quarter of both total
372
Session C1: Integrated assessment of biomass for biofuel or renewables
agricultural land and abandoned areas is not suitable for the studied energy grasses. Further
bio-energy analysis should be supplemented with economic and environmental criteria.
Figure 1. Soil optimized suitability maps for energy crops on total agricultural land (A) and
on abandoned agricultural land (B).
Supplementing the existing database of the soil map with suitability indices has provided
knowledge-based recommendations about suitability for bio-energy production for each
agricultural field. This approach makes soil information more self-explanatory and accessible
for stakeholders. This case study serves as a pilot study for further development of fieldspecific resource management for bio-energy planning and can be used as a tool in
knowledge-based decision-making processes throughout Estonia.
References
Edwards, R., et al., 2005. In: Eds L. Sjunnesson, et al., 14th European Biomass Conference,
17-21 October 2005, Paris, France.
Hoogwijk, M., et al., 2005. Biomass and Bioenergy 29: 225-257.
Kask, R., 1994. J. Agric. Sci. 5: 405-423. (in Estonian)
Kõlli, R., 1994. Soil suitability and agrogroups. Estonian Agricultural University, Tartu, 85
pp. (in Estonian).
Voivontas, D., et al., 2001. Biomass and Bioenergy 20: 101-112.
373
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Environmental impacts of agricultural land allocation between
bio-energy crops and food/feed crops
J. Lankoski, M. Ollikainen
2, rue Andre-Pascal, 75775 Paris Cedex 16, France
Contact: jussi.lankoski@oecd.org
Introduction
This paper analyses the multiple environmental effects of land allocation between bio-energy
crops and food/feed crops. It adopts an integrated economic and natural science modelling
approach: an economic model of farmers’ decision making is combined with a biophysical
model predicting the effects of farming practices on crop yields and multiple environmental
effects. The analysed environmental effects include GHG emissions over the life cycle,
nitrogen and phosphorus runoff, herbicide runoff and the quality of wildlife habitats.
Methods
This paper focuses on three environmental issues: surface water quality, climate, and
biodiversity. Both nitrogen and phosphorus runoff from cultivated fields to watercourses is
estimated. As regards pesticide runoff, the focus is on herbicide runoff (MCPA as an active
ingredient)1. Greenhouse gas emissions are modelled on the basis of life cycle assessment
(LCA) estimates provided by Mäkinen et al. (2006). In this paper, the following elements are
included: (i) CO2-eq emissions related to the transportation of crops, (ii) CO2-eq emissions
related to the manufacturing, transportation and application of fertilizers, herbicide, and lime
(iii) CO2 emissions from soil and (iv) CO2-eq emissions from tillage practices, such as
plowing, harrowing and planting as well as CO2-eq emissions from harvesting and grain
drying. The effects of land allocation on biodiversity are quantified by a wildlife habitat
indicator - a habitat quality index, developed by Lehtonen et al. (2008). This index measures
the impacts of land use on the quality of wildlife habitats. The monetary valuation of
environmental effects is used to aggregate the environmental effects. These valuation
estimates are based on published Finnish valuation studies quantifying the consumers’
willingness to pay for reducing nutrient and herbicide runoff or to promote biodiversity. The
price of emission allowances is used as a proxy for the climate damage (CO2-eq emissions).
The empirical application is on crop production in south-western Finland, which represents
20% of cultivated land in Finland. Reed canary grass (RCG, a perennial grass with 14 years
rotation period) represents second generation biodiesel, while rape represents first generation
biodiesel, barley is used for ethanol, oats is used for feed, and wheat is the food crop.
Results and discussion
Table 1 shows that reed canary grass (RCG) is cultivated in the 8 lowest productivity parcels
with low nitrogen use intensity. The low nitrogen application rate is due to the high unit
transportation costs and thus a low effective output price for RCG. However, support
payments and low production costs make it profitable to cultivate RCG in the lowest
productivity parcels.
Table 2 presents total environmental effects. It illustrates that reed canary grass performs
well as regards environmental impacts. Its good environmental performance is mainly driven
by low CO2-eq emissions. This is explained by the fact that RCG sequesters carbon and thus
soil CO2 emissions for it are negative, whereas for other crops, which are cultivated with
conventional tillage, soil CO2 emissions are significant.
1
For details of nutrient and herbicide runoff modelling see Lankoski et al. (2006).
374
Session C1: Integrated assessment of biomass for biofuel or renewables
Table 1. Land allocation, input use intensity, production and farmers’ profits.
Crop
Land area N-use Herbicide Production
Total
Profits
use
production kg
ha
kg ha–1
kg ha–1
kg ha–1
€ ha–1
RCG
8
8.9
2969
23 753
139
Barley
8
84.8
0.83
3743
29 940
168
Oat
18
58.1
0.72
3103
55 850
150
Rape
8
69.1
0.84
1669
13 352
188
Total
42
122 895
-
Total
profits
€
1109
1344
2694
1502
6649
Table 2. Total nitrogen runoff, total phosphorus runoff, total herbicide runoff, total CO2-eq
emissions.
N-runoff
P-runoff
Herbicide
CO2-eq
Habitat index
Crop
kg
kg
runoff, kg
emissions, t
value
RCG
26
6
0.5
Barley
53
10
0.08
23
Oat
99
22
0.14
46
Rape
48
10
0.08
21
Total
227
47
0.30
90
130.5
Figure 1. Expost
social
welfare
for
alternative land
uses (€ ha–1).
Figure 1 shows that the land use type that delivers the best environmental performance (reed
canary grass) is the least profitable for farmers. This social welfare ranking illustrates that expost social welfare of alternative land use types is mainly driven by profitability of land use
rather than the social valuation of environmental effects.
References
Lankoski, J., et al., 2006. European Review of Agricultural Economics 33: 193-221.
Lehtonen, H., et al., 2008. Unpublished manuscript.
Mäkinen, T., et al., 2006. Greenhouse gas balances and new business opportunities for
biomass-based transportation fuels and agri-biomass in Finland. VTT Tiedotteita 2357.
Espoo, Finland. (in Finnish with English abstract)
375
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Biofuels: The way forward
A. Leip1, Robert W. Howarth, Stefan Bringezu
European Commission – DG Joint Research Centre,
Institute for Environment and Sustainability, TP.050,
21014 Ispra (VA), Italy
Contact: adrian.leip@jrc.it
Introduction
In response to calls for energy security, minimizing climate change, and rural development,
several nations (U.S., European Union, Brasil) have or will shortly establish mandatory
targets for the incorporation of biofuels into their liquid fuel portfolio. Global production of
liquid biofuels tripled between 2000 and 2007, yet still contributed only 1.8% of the world’s
liquid transportation fuels in 2007, while requiring 5–6% of the global harvest of all grains
(corn, wheat, rice, and others), 8% of vegetable oil, and 28% of the sugar cane harvest.
A number of recent papers have pointed out that there are hard biophysical constraints on
production – the amount of carbon fixed by all crops globally is already exceeded by the
carbon burned by fossil fuels, and producing biofuels on all currently abandoned land might
yield only ~7% of our current energy use (Campbell et al., 2008; Field et al., 2008).
Legitimate concerns exist about the relative climate benefit of various biofuels (Crutzen et al.,
2008), and competition for arable land between food, fiber, fuel and other ecosystem services
(Robertson et al., 2008; Searchinger et al., 2008). Biodiversity might be affected by the
increased production of liquid biofuels (Raghu et al., 2006) and freshwater scarcity might be
aggravated (de Fraitur et al., 2008). The environmental consequences of liquid biofuels
depend on what crops or materials are used, where and how these feedstocks are grown, how
the biofuel is produced and used, and how much is produced and consumed. Both positive
and negative effects on the environment can occur.
Methods
This paper reviews the key environmental problems and future perspectives for a sustainable
production of liquid biofuels on the basis of the SCOPE rapid assessment workshop and its
products, a policy brief and discussion chapters on the topics of (i) mitigation strategies (ii)
quantitative, integrated biofuels assessment; (iii) biofuels and developing countries; and (iv)
final land limits.
Results and discussion
The magnitude of the environmental consequences depends on how, where, and how much
biofuels are produced and consumed, as illustrated by the following examples (SCOPE
biofuel project, 2009).
How?
• The corn grown in the Mississippi River basin to produce ethanol is often heavily
fertilized, contributing to the Dead Zone downstream in the Gulf of Mexico.
• Producing ethanol from switchgrass and other perennial plants requires far less fertilizer,
resulting in much less nutrient pollution of aquatic ecosystems.
1
This presentation summarizes results of the SCOPE rapid assessment, hold in Gummbersbach 22-25 September
2008, See http://www.eeb.cornell.edu/howarth/SCOPEBiofuels_home.html
376
Session C1: Integrated assessment of biomass for biofuel or renewables
Where?
• Oil palm plantations established on cleared tropical forests result in loss of biodiversity
from one of the world’s hot spots of diversity and huge emissions of greenhouse gases
when peat soils are drained.
• Oil from jatropha grown on degraded lands in Mali powers generators for electricity for
cell phone microwave towers and provides local jobs with no adverse environmental
impact.
How much?
• Massive land-use conversions, especially in the tropics, would be required to produce
enough biofuels to meet just 10% of the current global use of gasoline and diesel. Many
biofuel technologies can be pursued in an environmentally benign manner when land
requirements are small and biofuel crops are not competing with food crops.
A balanced and comprehensive assessment of liquid biofuels production poses a challenge for
both scientists and decision makers. Recommendations of the workshop include:
• Current mandates and targets for liquid biofuels should be reconsidered in light of the
potential adverse environmental consequences, potential displacement or competition with
food crops, and difficulty of meeting these goals without large-scale land conversion. Nonfood biomass should be preferentially used for material purposes.
• Policy instruments are needed to help adjust the overall demand for (non-food) biomass at
levels which can be supplied by sustainable production; (i) increase efficient use of
biomass and mineral resources; (ii) reduce fuel consumption for transportation.
• Biofuels based on low input cultivation of non-food crops offers promise in developing
countries as a source of energy, in part because energy use is often very low at present.
Biofuel markets can serve as an opportunity to trigger additional investments that could
lead to increased production of food as well as biofuel crops by small-scale farmers.
Further research on the use of indigenous non-food crops should be encouraged.
References
Campbell, J.E., et al.., 2008. Environ. Sci. Technol. 42(15): 5791-5794.
Crutzen, P.J., et al., 2008. Atmos. Chem. Phys. 8: 389-395.
de Fraitur, C., et al., 2008. Water Policy 10(S1): 67-81.
Field, C.B., et al., 2008. Trends in Ecology & Evolution 23(2): 65-72.
Raghu, S., et al., 2006. Science 313(5794): 1742-10.1126/science.1129313
Robertson, G.P., et al., 2008. Science 322(5898): 49-50.
SCOPE biofuel project. 2009. Biofuels: Environmental Consequences and Interactions with
Changing Land Use.
Searchinger, T., et al., 2008. Science 319(5867): 1238-1240.
377
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Exploring the land, water and energy linkages in agriculture:
An expanded partial-equilibrium approach to modelling
environmental impacts
Siwa Msangi, Rüdiger Schaldach, Christina Kölking, Jennifer Koch, Joseph Alcamo
c/o International Food Policy Research Institute,
2033 K Street NW, Washington, DC 20006, USA
Contact: s.msangi@cgiar.org
Introduction
The increasing concern over global food prices and their implications for human well-being
has encouraged policy analysts and researchers to look more closely at the constraints facing
agriculture, in terms of both policy restrictions and distortions, as well as physical limits of
growth. The realization that energy and agricultural markets are now becoming more closely
tied together, has also highlighted the need to better understand the economic as well as
physical linkages that cause the increase in the demand for energy products to translate into
increased demand pressure (and prices) within agricultural markets. The ongoing increases in
the price of both energy and food products have direct consequences for human well-being,
and also have implications for the environment and future trade-offs that will inevitably occur
between the need to provide food, feed, fiber and fuel for a growing world population, while
maintaining the quality of ecosystems and the ability to provision needed services from them.
In this paper, we explore the implications of socio-economic and environmental drivers of
change for agricultural food production systems, global markets, and the impacts of land and
water use change over time, within a linked modelling framework. The linkage that is created
is between a global, partial-equilibrium, agricultural market model and a global spatiallyexplicit land-use change model. The linkage is both dynamic in nature and explicit in the
physical quantities of agricultural production (decomposed into yield and area), projected
consumption of crop and animal products, and the geographical area needed for future crop
production, in relation to other important land uses that are required for future growth needs
such as settlement areas. By creating a strong linkage between the various components of the
agro-ecosystem that are of key relevance to future agricultural growth – and to the main
drivers of change that affect their evolution over time – we are better able to model and
ultimately to understand the impacts of important factors, such as biofuels production growth.
Methods
The methodology that is used to model the trade-offs between the increasing demands of
food, feed and fuel and the pressures on both agricultural and non-agricultural land uses is to
establish a linkage between a global, partial-equilibrium model of agricultural production,
consumption and trade, IMPACT (Rosegrant et al., 2001), and a global land-use change
model, LandSHIFT (Schaldach et al., 2008) that operates on a geographic raster with a
resolution of five arc-minutes (Schaldach et al., 2008). While there are now a class of
computable general equilibrium modelling tools that have been developed to handle questions
of land-use change – their construction is often too coarse in resolution and not as explicit in
the actual quantities of either consumption, production or land-use change that are occurring
to be useful for environmental assessments. The linked model system seeks to overcome these
limitations by applying of a pair of models that are more disaggregated over the key food
commodities, key land-use change processes (e.g., settlement and agriculture) and spatial
regions that are of greatest concern to the policy research community. Moreover it takes into
consideration biophysical constraints such as water availability and climate change effects on
crop growth. It is precisely in those regions (like Sub-Saharan Africa) and for those staple
commodities (maize, cassava or sorghum) that GTAP-based general equilibrium models tend
378
Session C1: Integrated assessment of biomass for biofuel or renewables
to aggregate over into classifications that no longer allow for a detailed examination of the
impacts upon them. The fact that these models also focus on quantities, rather than production
and consumption values, allows us to derive results that are much more relevant to natural
resource managers, and environmental scientists and analysts who are interested in the
absolute quantities of change as a foremost indicator.
Results and discussion
The first result of the ongoing research is a prototype version of the coupled IMPACTLandShift model and an exemplary application for a simulation study of the impacts of
biofuel development in India. By taking into account the key drivers of change that are both
socio-economic and environmental in nature, and how they affect the way in which the
demand and supply sides of agricultural markets behave (Figure 1), we are able to gain a
better understanding of how future trends might result in positive or negative impacts in
human and environmental well-being in various parts of the world.
Figure 1. Key drivers of change that will be modelled.
References
Rosegrant, M.W., et al., 2001. Global food projections to 2020: Emerging trends and
alternative futures. Washington DC, International Food Policy Research Institute.
Schaldach, R., et al., 2008. Simulating the impact of biofuel development on country-wide
land-use change in India. Biomass and Bioenergy. (under review)
379
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Energy from agricultural residues and the trade-off to the food system
S. Nonhebel
Center for Energy and Environmental Studies, University of Groningen,
Nijenborgh 4, 9747 AG Groningen, The Netherlands
Contact: s.nonhebel@rug.nl
Introduction
Using biomass as an energy source is often mentioned as an option to mitigate the increasing
greenhouse effect. Biomass for energy purposes can be obtained from dedicated energy crops
and/or from agricultural residues. The available amount of residues is large and suggests a
significant energy potential. It is estimated that in The Netherlands potentially 190 PJ can be
obtained from these residues; on a global scale values of over 12 EJ are mentioned. However,
most of these residues are currently used as livestock feed, which forms the basis for
important proteins in the human diet. Use of residues for energy generation is likely to affect
the supply of proteins in human diets, and therefore adaptations in the food system are
required to compensate for this loss.
This paper focuses on the question: what are the consequences of using residues for energy
generation instead of using them for livestock feed. It studies the adaptations required in the
food system to compensate for the loss of residues. Three different systems are recognized:
the present one where residues are used for livestock feed (Figure 1a) and energy is obtained
from energy crops (wood), and two systems where residues are used for energy generation.
The loss of livestock feed is compensated for by growing extra protein crops (in combination
with a change to a vegetarian lifestyle; Figure 1b) or by the growing of extra feed crops
(Figure 1c).
Figure 1
(a)
Figure
(b) 2
Energy system
Forests/
wood
Food system
Energy system
Figure
(c) 3
Food system
Energy system
Primary
Protein
production crops
Primary
production
Residues
Residues
Primary
production
Food
industry
Livestock
Livestock
Energy
Consumption
Feed crops
Residues
Food
industry
Food
industry
Food system
Energy
Consumption
Energy
Consumption
Figure 1. Schematic presentation of the three systems studied: Energy Crops system (a),
Vegetarian System (b), Feed Crops System (c).
Method
For all three systems the magnitude of the various flows is quantified through determining:
1. The magnitude of available residues in kg/person.
2. The amount of meat that can be produced on basis of these residues
3. The amount of beans/pulses needed to compensate for this amount of meat in the
menu
4. Quantity of wheat required as livestock feed
5. The value of the residues as energy source
6. The amount of wood from biomass plantations to compensate for energy in residues.
7. Finally the acreage needed for producing the beans, the wheat and the wood in the
various systems.
380
Session C1: Integrated assessment of biomass for biofuel or renewables
Results
Table 1 shows the acreage required for the production of proteins and energy in the three
systems studied. No acreage is attributed to the agricultural residues since it is assumed that
they are ‘unwanted’ by-products of food production. This implies that in the Vegetarian
System (Figure 1b) and the Feed Crops System (Figure 1c) no land is attributed to energy and
that in the Energy Crop System (Figure 1a) no land is attributed to the meat production. The
production of 36 kg beans in the Vegetarian System requires 120 m2. In the Energy Crops
System 80 m2 is needed for the production of 2.2 GJ energy (121 kg wood), and the
production of 120 kg wheat in the Fodder Crops System requires 170 m2.
Table 1. Comparison of the acreage required for producing proteins (33 kg pork (on residues
or 120 kg wheat as fodder) or 36 kg beans) and 2.2 GJ energy (on residues or 121 kg wood) in
the three different food-energy production systems.
Energy crops (Figure 1a)
Vegetarian (Figure 1b)
Feed crops (Figure 1c)
m2
m2
m2
33 kg pork
0
36 kg beans
120
120 kg wheat
170
121 kg wood
80
2.2 GJ energy
0
2.2 GJ energy
0
Total
80
Total
120
Total
170
The large differences that occur between the systems are striking. The Energy Crops System
and the Fodder Crops System produce the same commodities (energy and pork) but the
Fodder Crops System requires nearly 100 m2 more to do so. The Vegetarian system also
requires a larger acreage than the Energy Crops System (120 m2). It should be noted that
values mentioned concern values per person per year. 33 kg pork is over 70% of the annual
meat consumption per person. And 120 m2 seems a small amount of land but multiplying it
with the number of inhabitants results is vast amounts of land needed.
Conclusion
The analysis above allows some general conclusions on use of agricultural residues for energy
generation. When residues have a value as livestock feed, use of these residues as an energy
source results in tremendous trade-offs to the food system. These trade-offs are due to the fact
that the loss of livestock feed needs to be compensated for to maintain a healthy diet for the
human population. The loss in the food system is far larger (120–170 m2) than the gain in the
energy system (80 m2).
381
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing the environmental effects of an increasing biogas production on
agricultural farms
J. Schuler, G. Uckert, P. Zander
Leibniz-Centre for Agricultural Landscape Research (ZALF e.V.),
Institute for Socioeconomics,
Eberswalder Str. 84, 15374 Müncheberg, Germany
Contact: jschuler@zalf.de
Introduction
Biogas production has been developed and fostered to mitigate negative externalities of fossil
energy use. But with the large scale adoption of this technology, negative side effects of the
increased biomass production became apparent. Therefore, this study aimed to identify the
trade off with nature conservation objectives, to explore efficiency and conditions of biogas
production and to contribute to the development of novel production systems taking into
account environmental objectives. On the basis of bio-economic farm modelling, the study
investigated the economic and environmental performance of novel cropping systems, their
impact on habitat conditions and conditions and possibilities to maintain or improve the
profitability of biogas plants.
Methodology
The assessment of the environmental and economic effects was based on an adaptation of the
modelling system MODAM (Zander & Kächele, 1999; Schuler & Kächele, 2003). For the
purpose of this study, a biogas module was introduced into the farm model. The module
implements linkages to other parts of the farming system such as crop production and
livestock production through competition for area, fodder and slurry usage. The substrate
usage of biogas plants is optimized on the basis of biogas potential of substrates and the
requirements of the fermenter with respect to structure, energy content and dry matter share of
the substrate.
For the environmental assessment, available indicators from the bio-economic system
MODAM were used (Happe et al., 2006; Sattler et al., 2006; Sattler, 2008; Schuler & Sattler,
2008). The indicators are based on expert knowledge and allow the comparison of different
scenarios through a dimensionless index system.
The model was applied to both existing farms from a case study as well as a set of
prototype farms representing typical farms of North-Eastern Germany. Different novel
production systems were introduced and a number of scenarios (prices, different methane
yield levels, installed fermenter and energy units) were examined with respect to
environmental and economic impacts.
Results and discussion
The modelling results show that the cultivation of energy crops is not harming the
environment per se. Depending on the crops used and the intensity of the cultivation, the
overall effects can differ. Furthermore, the efficiency of the biogas fermenter has a large
effect on the environment. Full capacity utilization of the biogas plant and less efficiency of
the biogas process is leading to a higher biogas substrate demand and accordingly a higher
area consumption for the energy crops. The increase of competition at the expense of
primarily set-aside areas showed the most harmful effects.
Figure 1 shows the overall effects of different farm types on a set of environmental
indicators. The farm types differ in the crops used and the biogas yield that is achieved in
each fermenter (representing different skills levels), K0 is a reference farm with no biogas
plant installed.
382
Session C1: Integrated assessment of biomass for biofuel or renewables
Amph
WaEro
Sky
1
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
Amph
Flora
GWR
Hare
Hover
NO3
N/P
Pest
Sky
WaEro
Flora
= Amphibians
= Wild flora species
= Groundwater recharge
= Field hare
= Hover fly
= Nitrate leaching
= N/P-entries surface water
= Pesticide entries water
= Skylark
= Water erosion
GWR
Hare
Pest
K0 (no Biogas-Plant)
K1 biog_basis
K1 biog_kons
N/P
Hover
K1 epfl_basis
K1 epfl_kons
NO3
Figure 1. Environmental effects of different resulting land use patterns for a selection of
regionally relevant indicators. (Farms K0= no biogas plant, K1= biogas plant installed, biog =
standard cropping practices used for energy production, epfl = adapted cropping practices for
energy production, basis = high biogas yields, kons = low biogas yields; 1 = positive, 0 =
negative effects).
In summary, the introduction of biogas plants can improve the environmental impact of arable
farms, if they make use of the possibilities to reduce input intensity and through a more
diversified crop rotation. In livestock farms these effects are less pronounced as
extensification options are less probable to occur and energy and fodder crops are largely
identical, thus narrowing the crop rotation.
References
Happe, K., et al., 2006. Agrarwirtschaft 55( 5/6): 268-279.
Sattler, C., 2008. Ökologische Bewertung und Akzeptanzanalyse pflanzenbaulicher
Produktionsverfahren. Dissertation, Humboldt-Universität zu Berlin. dissertation.de Verlag im Internet GmbH, 236 pp.
Sattler, C., et al., 2006. Determination of trade-off functions to analyse the provision of
agricultural non-commodities. International Journal of Agricultural resources, Governance
& Ecology 5, No. 2/3, pp. 309-325.
Schuler, J. & H. Kächele, 2003. Environmental Science and Policy 6(1): 51-55.
Schuler, J. & C. Sattler, 2008. Land Use Policy, doi:10.1016/j.landusepol.2008.05.001
Zander, P. & H. Kächele, 1999. Agricultural Systems 59(3): 311-325.
383
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
First and second generation biofuels up to 2030:
Possible scenarios and their environmental impacts
Elke Stehfest, Geert B. Woltjer, Anne Gerdien Prins, Bas Eickhout, Martin Banse
Netherlands Environmental Assessment Agency (PBL),
Postbus 303, 3720 AH Bilthoven, The Netherlands
Contact: Elke.Stehfest@pbl.nl
Introduction
Until recently biofuels seemed to be an attractive alternative to fossil fuels in order to reduce
greenhouse gas emissions and climate change, and to increase energy security. However,
concerns about the induced land-use change, their small greenhouse gas savings, and the
impact on global food prices has dampened this enthusiasm substantially. Initially ambitious
targets of bio-energy use for example in the European Union have been reduced, and options
to avoid the negative impacts have been introduced, e.g., via sustainability criteria. The
interaction with the agricultural sector via competition of land, and indirect land use effects
are highly complex and therefore hard to control, and the discussion on appropriate biofuel
policies continues.
In this paper, we explore possible scenarios of biofuel use and production under different
European and international biofuel targets using a coupled agro-economic and integrated
assessment model.
Methods
We used the LEITAP model, a modified GTAP model and database, (Van Meijl et al., 2006),
coupled to the integrated assessment model IMAGE (MNP 2006). The LEITAP model is used
to calculate agricultural production, trade and consumption of food crops and bio-energy
crops over the time period 2000-2030. Its projections of agricultural production and
intensification are passed to the IMAGE model, which allocates land use on a 5 minutes grid,
and calculates resulting environmental impacts, greenhouse gas balances and climate change
under the respective scenario. The coupling between the two models has been initiated in the
EUruralis framework (Verburg et al., 2008), is described in detail elsewhere (Eickhout et al.,
2008). Also the work presented here as been carried out as a part of the EUruralis project.
As a reference case we developed a ‘business as usual’ scenario, with continuous high
economic growth, moderate population growth, and no new policies. Following this approach,
biofuel policies of the European Union are set to the current 6% target in 2020. On top of this
reference case, we explore possible developments and policy options relevant for European
and global bio-energy. These options include variations of the European target, variations of
the bio-energy policies in countries outside the European Union, possible contributions of
second generation biofuels, technological development in both the food crop and the bioenergy sector, and different scenarios of nature conservation excluding bio-energy crop
production on designated areas like forests or high value grasslands.
Results and discussion
For the reference scenario and the variants listed in Table 1 the LEITAP–IMAGE model suite
calculates European and global land use effect, the full greenhouse gas balance of biofuel and
thereby their expected climate mitigation potential. While food prices are not the focus of the
study, they are also briefly reported.
(Results will be produced during the next months, and can, therefore, not be reported yet).
384
Session C1: Integrated assessment of biomass for biofuel or renewables
Table 1. Different options for the bio-energy scenarios explored in this study.
Category
Options
Target EU
No bio-energy policy, 6% EU baseline target, 10% high EU
target
Worldwide bio-energy
Baseline, and high implementation of biofuels in other
policies
countries
Other sources of bio-energy Different contributions from forest products and agricultural
residues
Technological development Different development of technological improvement for
bio-energy due to R&D investments and external
bio-energy
assumptions
Agricultural intensification
Different assumptions on the increase in crop productivity
Restricted areas
Free bio-energy production, or restricting bio-energy
production to non-forested lands, degraded soils, or
excluding high biodiversity areas.
References
Eickhout, B., et al., 2008. In: Eds T.W. Hertel, et al., GTAP Working Paper No. 47 Chapter 9
of the forthcoming book Economic Analysis of Land Use in Global Climate Change
Policy.
MNP, 2006. Integrated modelling of global environmental change. An overview of IMAGE
2.4. Netherlands Environmental Assessment Agency (MNP), The Netherlands.
Van Meijl, H., et al., 2006. Agriculture, Ecosystems and Environment 114: 21-38.
Verburg, P.H., et al., 2008. Annals of Regional Science 42: 57-77.
385
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Spatial assessment of abandoned agricultural land
for bio-energy production
E. Suuster, L. Kukk, A. Astover, M. Noormets
Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences,
1 Fr. R. Kreutzwaldi St., 51014 Tartu, Estonia
Contact: elsa.suuster@emu.ee
Introduction
Increasing dependency on fossil energy has forced researchers to think more about alternative
energy carriers, e.g., bio-energy. In Estonia there is great potential for cultivating energy
crops due to the availability of a lot of abandoned agricultural land. The decline in arable land
use in Estonia was regionally variable and especially high in marginal districts with low soil
quality (Peterson & Aunap, 1998; Astover et al., 2006). Therefore, the planning of bio-energy
production on abandoned areas requires precise location-specific analysis. The aim of present
study was to investigate the amount and location of abandoned agricultural land and to
analyse its suitability for producing energy crops in Saare County. Saare County is one of the
15 counties in Estonia situated in the country’s western part. The county’s total area covers
6.7% (292,219 ha) from Estonia’s surface.
Methods
The study identified abandoned field parcels in Saare County, using the Estonian Basic Map
(1:10,000) and field layer of Agricultural Registers and Information Board (ARIB) and
databases of Common Agriculture Policy (CAP) payments in 2007. We considered field
parcels without any applications for single area payments as entirely abandoned and parcels
where area payments covered 50–99% of total area as partially abandoned. Agricultural areas
outside of ARIBs fields (not valid for CAP subsidy schemes) were estimated also as entirely
abandoned and determined from the Estonian Basic Map. For this, field layer of ARIB fields
was cut out from basic map. Remaining agricultural areas from basic map were thereafter
cleaned topologically: (i) areas less than 0.3 ha were eliminated, (ii) areas with perimeter-area
ratio over 5 were eliminated, and (iii) visual assessment and manual correction of area
boundaries based on orthophotos. These approaches enabled eliminate small unsuitable areas
for bio-energy crops for further analysis. A GIS environment, MapInfo Professional 7.5,
enabled us to perform topology analysis of the field layers and the soil map polygons. We
identified the soils of abandoned land using the Estonian Land Board’s digital soil map (scale
1:10,000) and assessed the suitability of these areas, depending on the soil type and texture,
for short rotation energy forestry and energy grasses (Kõlli, 1994; Laas, 2004). The areas
suitable for potential bio-energy production were evaluated for willow (Salix sp), grey alder
(Alnus incana), hybrid aspen (Populus × wettsteinii Hämet – Ahti), reed canary grass
(Phalaris arundinacea L.), Caucasian goat’s-rue (Galega orientalis Lam.) and alfalfa
(Medicago sativa).
Results and discussion
The total agricultural land in Saare County included for analysis in the current study is 76,454
ha (68,654 ha ARIB fields and 7,800 ha from basic map) which represents 26% of the
county’s total land area. Total abandoned area formed 27,046 ha of which entirely and
partially abandoned accordingly 22,648 and 4,398 ha. Most of the entirely abandoned
agricultural land is situated on the coastal line of the county (Figure 1). Abandoned fields
mean size is 5 ha and entirely abandoned fields, 2.7 ha. The field’s size could significantly
limit re-using agricultural land again in the future.
386
Session C1: Integrated assessment of biomass for biofuel or renewables
Entirely abandoned
Partially abandoned
0
12.5
km
25
Figure 1. Location of entirely and partially abandoned agricultural land in Saare County.
On the entirely abandoned agricultural land Calcaric Cambisols, Leptosols and Fluvisols
dominate. The textures of these soils are mainly sands, loam, sandy loam and the same
materials with more stones in it. From total abandoned agricultural land Caucasian goat’s-rue
could be cultivated on 15,267 ha, alfalfa 14,439 ha and grey alder 14,389 ha. The potential
bio-energy production from these fields would form accordingly 493, 466 and 697 GWh.
Reed canary grass could be grown in 10,704 ha producing 395 GWh of bio-energy. Hybrid
aspen and willows are suitable for 1/5 of total abandoned agricultural land. Their potential
bio-energy production for each is 147 GWh. The energy consumption of Saare County in
2004 totalled 614 GWh (Hallemaa, 2006). Growing Caucasian goat’s-rue, alfalfa or grey
alder in suitable abandoned areas in Saare County, it is possible to cover the entire islands
energy consumption without reducing current food production.
References
Astover, A., et al., 2006. Archives of Agronomy and Soil Science 52(2): 223-231.
Hallemaa, H., 2006. Sustainability of energy use. Counties in Figures 2001–2005. Tallinn, pp.
18-38. (in Estonian)
Kõlli, R., 1994. Soil suitability and agrogroups. Estonian Agricultural University, Tartu, 85
pp. (in Estonian)
Laas, E., 2004. In: Ed. E. Asi, Afforestation of arable land: A methodical guide for
afforestation of arable land. Centre of Forest Protection and Silviculture, Tartu, pp. 31-57.
(in Estonian)
Peterson, U. & R. Aunap, 1998. Landscape and Urban Planning 41(3-4): 193-201.
387
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Enhancing the assessment of biofuels feedstock production through more
realistic representation of farming systems
P.J. Thorburn, D. O’Connell, T. Grant
CSIRO Sustainable Ecosystems, 306 Carmody Rd,
St. Lucia Qld 4068, Australia
Contact: Peter.Thorburn@csiro.au
Introduction
In common with global trends, there is growing interest in biofuels in Australia to help reduce
greenhouse gas (GHG) emissions and provide a more sustainable energy supply. The
Australian biofuels industry currently only supplies less than 0.5% of the total transport fuels
(O’Connell et al., 2007) using waste starch and C-Molasses for ethanol, and used vegetable
oil and tallow for biodiesel. Expansion of biofuel supplies requires different feedstock sources
because the supply of traditional feedstocks is reaching its limit. The sustainability of these
new feedstocks needs to be assessed. GHG emissions of biofuels have long been of critical
interest to policy makers as a key aspect of sustainability, and have been fundamental to the
argument for government support.
While the GHG emission profiles for a range of standard first generation biofuels have
been well studied (O’Connell et al., 2007), many of these assessments paid limited attention
to the details of the farming systems producing the feedstocks. For example, average values
for management inputs and practices were used and inter-regional variation ignored. Further,
soil carbon (C) is generally assumed to be constant, an assumption not met in regions recently
cleared of native vegetation (Fargione et al., 2008). Additionally, nitrogen (N) fertilizer is an
important input to many farming systems. Apart from the energy inputs associated with
fertilizer manufacture, it is increasingly recognized that emissions of nitrous oxide, a potent
GHG, have been underestimated in assessments of GHG emissions from biofuels production
(Crutzen et al., 2008). This paper describes our work linking agricultural simulation models to
life cycle assessments to enhance the assessment of GHG emissions in the feedstock production end of the biofuels value chain, using sugarcane production in Australia as a case study.
Methods
We compared GHG emission profiles of three contrasting Australian sugarcane production
systems: (1) The Tully region, in the super-humid tropics where weed control is a major issue
and crop residues are conserved; (2) The Burdekin region, in the dry tropics where sugarcane
production fully irrigated and crop residues are burnt at harvest; and (3) The Maryborough
region, in the subtropics where sugarcane production uses supplementary irrigation and crop
residues are also conserved. These production systems were simulated over a ~40 year period
with the APSIM-Sugarcane model. Model outputs for annual production, soil C and N losses
to the environment were used in the life cycle assessment (LCA), together with details of the
farm management practices such as tillage, pesticide applications, etc. employed in the
regions. In the Tully region, the site modelled was adjacent to a native forest remnant, and
soil C in the forest was used to represent soil C at the commencement of cropping (44 years
previous). This allowed us to also examine the impact of forest clearing and soil C run-down
in the LCA. Finally, uncertainty in nitrous oxide (N2O) emissions was investigated by
undertaking LCA’s with either standard, constant emission factors (used in Australian GHG
inventories) or N2O emissions modelled explicitly in APSIM (Thorburn et al., 2008).
Results and discussion
GHG emission varied by ~50% between the different sugarcane production systems (Figure
1a). Tully has the highest carbon dioxide emissions because of the greatest application of
388
Session C1: Integrated assessment of biomass for biofuel or renewables
Carbon dioxide
Methane
Nitrous oxide (EF)
Land transformation
70
(a)
60
50
40
30
20
10
0
Burdekin
Maryborough
Tully
Carbon dioxide
Greenhouse emissions kg CO2e / t cane
Greenhouse emissions kg CO2e / t cane
herbicides to control weeds in this wet environment. However, the Burdekin has high
methane emissions from burning of the crop residues, resulting in the highest total GHG
emissions. Ceasing burning would be an option for increasing sustainability of sugarcane
production in this region.
Methane
Nitrous oxide (mod)
Land transformation
70
(b)
60
50
40
30
20
10
0
Burdekin
Maryborough
Tully
Figure 1. Greenhouse gas emission profiles for sugarcane production in three contrasting
regions in Australia with N2O emissions (a) estimates with constant emission factors, or (b)
mechanistically modelled for each production system.
Soil C was predicted to be declining marginally (~80 kg ha–1 yr–1) in each region under
current production systems. At Tully however, including the initial rapid run-down of soil C
following clearing of the native forest increased average soil C decline to > 500 kg ha–1 yr–1.
Including emissions due to this change in soil C increased total GHG emission for Tully by
180% (data not shown). GHG emissions are further increased by an order of magnitude if
emissions from forest clearing are included. It would take ~59 years of producing ethanol
from C-Molasses before carbon neutrality from petrol offsets was reached if forests were
cleared for biofuel production in this region, illustrating the small net GHG benefits for
ethanol production from sugarcane in newly cleared land.
N2O emissions using constant emission factors reflect the different N fertilizer application
rates in these regions (Figure 1a). However, when N2O emissions were explicitly modelled,
they increase markedly at all sites, especially Maryborough (Figure 1b), illustrating the
conservative nature of standard N2O emission factors. The large impact at Maryborough is
due to the relative over-fertilization of sugarcane in that region compared with yields,
suggesting that GHG emissions from sugarcane production in this region could be reduced.
This study illustrates some of the complexities of assessing emission profiles for biofuels.
Agricultural production systems vary, responding to differing environments, infrastructure
constraints/opportunities and local customs. It will be important to capture the specifics of the
farming system when assessing GHG emission from biofuel feedstock production. As well
there is uncertainty in representation of N2O emissions. Standard emission factors for N2O are
used because of the complexity of the processes resulting in N2O emissions. However,
agricultural simulation models are developing and becoming an increasingly useful tool in
assessing sustainability of biofuels feedstock production.
References
Crutzen, P.J., et al., 2008. Atmos. Chem. Phys. 8: 389-395.
Fargione, J., et al., 2008. Science 319, 1235.
O’Connell D., et al., 2007. Biofuels in Australia – issues and prospects. Rural Industries
Research and Development Corporation, Canberra, Australia.
Thorburn, P.J., et al., 2008. Agric. Ecosys. Env. (submitted)
389
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Greenhouse gas emissions from grassland swards:
A Scottish perspective
C.F.E. Topp, R.M. Rees
SAC, West Mains Road, Edinburgh, UK, EH9 3JG
Contact: Kairsty.Topp@sac.ac.uk
Introduction
Emissions of nitrous oxide from grasslands in Scotland have been shown to be high by
comparison to other European countries (Flechard et al., 2007). Agriculture was responsible
for 79% of total N2O emissions in Scotland in 2006 (Jackson et al., 2007). Despite this,
grasslands tend to operate as net greenhouse gas sinks largely as a consequence of their high
CO2 sink strength (Soussana et al., 2007). Hence, it is important, that the opposing effects of
C uptake and N2O release are taken into account when assessing future climate and
management scenarios in order to calculate a net greenhouse gas balance. Improved
grasslands are an important part of the Scottish landscape covering about 22% of the
agricultural land area in 2005. The manure and fertilizer management of these grasslands is
known to have a significant impact on their N2O emissions and hence greenhouse gas
balance. Modifications to their management based upon an assessment of current and future
climatic conditions could, therefore, play a valuable role in reducing greenhouse gas
emissions. The DNDC (DenitrificationDecomposition) model (Saggar et al., 2004; Li et al.,
1992; Li et al., 2006), which has been extensively applied around the world, simulates the
daily fluxes and pool sizes of carbon and nitrogen in agroecosystems. The objective of this
study was to use the model to explore the impact of different fertilizer and manuring
strategies on greenhouses gas emissions at two cut grassland sites in Scotland.
Methods
Two sites, located near Edinburgh (Cowpark) and Dumfries (Crichton), were used to assess
the impact of fertilizer and manure applications on the global warming potential (GWP) of
grasslands. The DNDC model was used to simulate management for 30 years, with the results
presented for the final 20 years as this gives time for the soil pools to stabilize. The fertilizer
treatment was three applications each of 100 kg N in mid March, mid May and mid July.
Slurry applications were made in mid April and mid June with 150 kg of available N being
applied at each application. Three cuts of silage were taken per year with cuts in mid June,
late August and late October. The weather generator Earwig (Kilsby et al., 2007) was used to
create 30 years of baseline (1961–1990) and 2020 low and medium-high UKCIP scenarios
(http://www.ukcip.org.uk/).
Results and discussion
At both Crichton and Cowpark there was a net uptake of GHGs (negative GWP) driven by the
large carbon sink strength and all management and climate combinations. Nevertheless, there
was a reduction in the GHG sink strength (less negative GWP) when baseline conditions were
compared with either of the 2020 climate scenarios (Figure 1). At Crichton this was due to the
combination of a small increase in N2O emission and a small decline in the C sink, while at
Cowpark the change was mostly due to a reduction in the C sink. Both the fertilizer and slurry
treatments were net GHG sinks under all climates (Figure 1), but the slurry treatment at both
sites resulted in a greater net carbon uptake than the fertilizer treatment. However, the
environmental benefits of the slurry were counterbalanced by the increase in nitrate leaching,
which was particularly noticeable at the wetter Crichton site. There was also a significant
increase in the soil organic matter pool size for the slurry treatment relative to the fertilizer
treatment.
390
Session C1: Integrated assessment of biomass for biofuel or renewables
The results suggest that the net sink strength for greenhouse gases in the grasslands that
have been studied will decline by up to 40% over the next 20 years. This is driven largely by
the predicted changes in climate, and the effects highlight the potential for strong regional
differences in ecosystem responses. The warmer and wetter conditions in the west of Scotland
are predicted to lead to increased losses of N2O. In contrast, drier summers and overall
increases in temperature in the east would reduce the carbon sink strength, but have little
impact on N2O emissions. However, there are clearly uncertainties associated with these
predictions. This is partly a consequence of model uncertainties, although predicted
greenhouse gas fluxes are broadly consistent with those from measurements at Scottish sites
(Jones et al., 2006; Jones et al., 2007). In addition, there are also uncertainties about future
climates. However, as a consequence of our changing climate, this study indicates that over a
relatively short period of time the GHG sink strength may show a significant decline in
Scottish grasslands. These feedback effects will make targets for significant reductions in
greenhouse gas emissions even more challenging.
3
5
(a)
2
0
-1
CO2-C
N20
GWP
-2
-3
t CO2-e / ha
1
t CO2-e / ha
(b)
0
-5
CO2-C
N20
GWP
-10
-15
-4
-20
-5
-6
-25
Cr Base
Cr Low
Cr MH
Cow Base Cow Low Cow MH
Cr Base
Cr Low
Cr MH Cow Base Cow Low Cow MH
Figure 1. Carbon sequestration, N2O and global warming potential for the (a) fertilizer
treatment and (b) slurry treatment for Crichton (Cr) and Cowpark (Cow) weather conditions
for current (Base), and the low (Low) and medium-high (MH) UKCIP02 emission scenarios.
Acknowlegments
RERAD and EU project NitroEurope IP (Contract no: 017841) funded this work.
References
Flechard, C.R., et al., 2007. Agriculture, Ecosystems & Environment 121: 135-152.
Jackson, J., et al., 2007. Greenhouse gas inventories for England, Scotland, Wales and
Northern Ireland:1990-2005. AEA Technology PLC, Oxon, UK, 40 pp.
Jones, S.K., et al., 2006. Soil Use and Management 22: 132-142.
Jones, S.K., et al., 2007. Agriculture, Ecosystems & Environment 121: 74-83.
Kilsby, C.G., et al., 2007. Environmental Modelling and Software 22: 1705-1719.
Li, C., et al., 1992. J. Geophys. Res. 97: 9759-9776.
Li, C.S., et al., 2006. Ecological Modelling 196: 116-130.
Saggar, S., et al., 2004. Nutrient Cycling in Agroecosystems 68: 243-255.
Soussana, J.F., et al., 2007. Agriculture, Ecosystems & Environment 121: 121-134.
391
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Water implications of biofuel crop cultivation and set-aside suspension in
the European Union
M. Van der Velde, G. Wriedt, F. Bouraoui
IES-Joint Research Centre of the European Commission,
Via Fermi 2749, 21027 Ispra (VA), Italy
Contact: marijn.van-der-velde@jrc.it
Introduction
The Common Agricultural Policy (CAP) of the European Union (EU) is in a process of
review. On the 20th of May 2008, European Commissioner Mariann Fischer Boel presented
the so-called CAP ‘Health Check’ proposals (CEU, 2008). The proposals were adopted on the
20th of November 2008 (EC, 2008) and will be published in December 2008. One of the
consequences of the Health Check is the abolishment of set-aside in 2009. Set-aside,
withdrawing land from agricultural production, became an instrument of the CAP in 1988.
The primary objective of set-aside was reducing surplus production and an accompanying
objective was the protection of the environment and natural resources. It has been applied
under various conditions, voluntary or obligatory, for different percentages of arable land.
The crops that are likely to be grown on previously set-aside land include wheat and
oilseed rape because of expected market demand. Oilseed rape cultivation is stimulated by the
role agriculture has been given in mitigating climate change. The abolishment of set-aside
combined with the increase in rapeseed cultivation for biodiesel may lead to additional
pressures on European waters. The EU’s Water Framework Directive (EP & CEU, 2000)
aims to protect European waters by reducing polluting discharges in ground- and surface
water. The link between water and agriculture has become increasingly important over the last
several years. One of the challenges of the current CAP reform is to meet the environmental
objectives of the Water Framework Directive (WFD) while ensuring food supply and finding
the right balance for the role of agriculture in climate mitigation.
Methods
We assume that the arable land that previously had been set-aside has been completely
converted to a mix of oilseed rape, common wheat and durum wheat. We use a pan-European
spatialized version of the EPIC model (Williams, 1995) that runs on a 10 by 10 km grid with
relevant meteorological, land use, terrain, soil and management information. The model
previously has been tuned for oilseed rape, common wheat and durum wheat yields using
regional yields from across Europe as described by Van der Velde et al. (2008) for oilseed
rape. To ensure that the proposed changes to the CAP minimally impact the targets set by the
WFD it is pertinent to have relevant agricultural information available at the river basin level
and therefore we summarized our modelling results at that scale.
Results and discussion
We identified river basins with a high share of utilized agricultural area as well as those river
basins with a high percentage of set aside compared to the total utilized agricultural area and
identified those basins where pressures on water resources resulting from set-aside conversion
and related increased usage of fertilizers are likely to be most immediate.
The release of set aside’s production capacity requires safeguards to continue the
ecological and sustainable benefits of this instrument. Proposed policy changes specifically
related to water quality and quantity issues include additional GAEC standards on water
management and landscape features.
Our results suggest that a more continuous river-basin approach coupled with farm-scale
measures is needed to enable off-setting the additional agrichemical loading and soil particle
392
Session C1: Integrated assessment of biomass for biofuel or renewables
transport by increasing the buffering capacity of the agricultural-ecological landscape
continuum. Increasing the buffer capacity of the agricultural-ecological continuum will be
beneficial to retain soil, water and nutrients in the landscape. This will allow us to reap more
resilient environmental services and will help agriculture in adapting to increasing water
scarcity and droughts that are expected across Europe under further climate change.
We argue that careful consideration should be given to the objectives of the CAP as
evaluated by the Health Check to ensure that they do not impinge on the environmental
targets set by the Water Framework Directive.
References
CEU, 2008. Proposal for a Council regulation. Establishing common rules for direct support
schemes for farmers under the common agricultural policy and establishing certain support
schemes for farmers (etc.). Commission of the European Communities, COM(2008) 306
final, 161 pp, Brussels, Belgium.
EC, 2008. Agriculture: CAP Health Check will help farmers meet new challenges. European
Commission, Press Communication, IP/08/1749, Brussels, Belgium, 2 pp.
EP & CEU, 2000. Directive 2000/60/EC establishing a framework for Community action in
the field of water policy. European Parliament and the Council of the European Union,
Official Journal of the European Union, Brussels, Belgium, L 327/1-73.
EP & CEU, 2003. Directive 2003/30/EC on the promotion of the use of biofuels or other
renewable fuels for transport. European Parliament and the Council of the European
Union, Official Journal of the European Union, Brussels, Belgium, L 123/42-46.
Van der Velde, M., et al., 2008. Global Change Biology 14: 1-14.
Williams, J.R., 1995. In: Ed. V.P. Singh, Computer models of watershed hydrology. Water
Resources Publications, Highland Ranch, CO, USA, pp. 909-1000.
Disclaimer: The statements expressed here are purely those of the authors and may not in any
circumstance be regarded as stating an official position of the European Commission
393
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Bio-ethanol from sugar beet in The Netherlands: Sustainability assessment
G.W.J. Van de Ven, C.L.M. De Visser, H. Langeveld, S.C. De Vries, L. Van den Brink
Plant Production Systems, Wageningen University,
P.O. Box 430, 6700 AK Wageningen, The Netherlands
Contact: Gerrie.vandeVen@wur.nl
Introduction
The European Union imposed a minimum percentage of biofuels to be blended with fossil
transport fuels. A crop offering perspectives for bio-ethanol production in Europe is sugar
beet due to the high sugar yields per ha. The aim of this study is to (partly) apply the Dutch
Framework for sustainable biomass production (Cramer et al., 2007) and to assess the
sustainability of ethanol production from sugar beet in The Netherlands from real farm data.
Methodology
The Dutch Framework for sustainable biomass production aims at certification of biomass
production. The indicators cover a wide range of subjects related to the people-planet-profit
principles. Minimum/maximum values have to be quantified or relevant information has to be
reported. We have selected the indicators relevant for the Dutch production situation and
added the indicator net energy production.
We analysed the real data of two farms of Wageningen-UR over three years (2005–2007).
Sugar beet cultivation was according to normal commercial practices. Both farms are located
in regions with sugar beet in the standard rotation, on old marine clay (Westmaas) and on
former peat soil (Valthermond). Both farms are well managed with yields above average.
The bio-ethanol production chain is divided into four phases: (i) cultivation of sugar beet;
(ii) transport to the processing plant; (iii) conversion of sugar beet into ethanol and byproducts; (iv) distribution of bio-ethanol to users. The energy requirements and greenhouse
gas (GHG) emissions have been quantified for each phase separately. Energy use during
cultivation is the sum of direct and indirect energy required for the use of machinery, diesel,
seeds, nutrients (inorganic and organic) and crop protection agents, based on farm registration
and standard energy contents (Mortimer et al., 2004; Bos et al., 2007). The transport and
distribution distance are set at 100 km. Conversion of sugar beet into ethanol is according to
Mortimer et al. (model 4a; 2004), as no real processing data were available. The energy yield
is calculated from product yield and their lower heating values.
GHG emissions (CO2, N2O, CH4, expressed in kg CO2 equivalents) for manufacturing
inputs, transport and conversion are based on Mortimer et al. (2004). N2O emission during
cultivation, resulting from manure application, crop remains and decomposition of organic
matter, is calculated according to IPCC (2006). We assume that frequency of sugar beet
cultivation is increased from 1:8 (Westmaas) and 1:6 (Valthermond) to 1:4 on both farms, i.e.
limit to good agricultural practice. The changes in soil organic matter (SOM) have been
quantified according to Yang & Janssen (2000). Qualitative indicators are effects of changing
land use and no abuse of laws and regulations with respect to biodiversity and environment.
At both farms the latter indicator is met.
Results and discussion
Total energy production in ethanol and by-products (vinasse and pulp) exceeds the total
energy input by a factor 2.25 and net energy production is 140–195 GJ ha–1. If only the
energy output with ethanol is considered net energy yield is 35–50 GJ ha–1. However, if part
of the energy input is allocated to the by-products, based on the lower heating values this is
42%, the result is a net energy production in ethanol of 75–110 GJ ha–1. Hence, net energy
production per ha is positive.
394
Session C1: Integrated assessment of biomass for biofuel or renewables
Cultivation requires 10–18% of the energy input and conversion, 75–82%. Total GHG
emission varies between 8.3 and 11.6 t CO2 equivalents per ha, cultivation being responsible
for 35–50%. This implies that measures to reduce energy input should be focused at
improving conversion technologies, and measures to decrease GHG-emission need attention
during cultivation of sugar beet.
Figure 1 shows that direct N2O emission, from crop residues, decomposition of SOM and
nutrient application, is the major contributor to GHG emission during cultivation. In
Valtermond, this is substantially higher than in Westmaas, as the mineralization rate is 50 kg
N/ha more. One would expect lower nutrient application rates, but these are similar on both
locations. In Valthermond, animal manure is applied, which has a lower uptake efficiency and
we assumed that manure application leads to similar GHG emissions as fertilizer per kg of
nutrient applied. If the GHG related to production of manure is not taken into account, the
GHG emission in Valtermond will decrease by 0.8–1.3 t CO2-equivalents.
6.0
GHG emission (t CO2 eq./ha)
5.0
4.0
machines
diesel use
N2O
cpa
manure
fertiliser
3.0
2.0
1.0
0.0
2005
2006
Westmaas
2007
2005
2006
2007
Figure 1. GHG emission from sugar
beet cultivation during three years at
two locations.
Valthermond
Compared to gasoline the total GHG emission is reduced by 22–39%, if allocation is not
taken into account and 50-65% if allocation to by-products is taken into account. Hence, it
depends on assumptions and calculation methods, if the criteria of 30% reduction is met.
SOM content is hardly affected by growing sugar beet more frequently. It replaces corn in
Westmaas and barley in Valterhmond and under current practices the effects of differences in
effective organic matter applied with these crops are too small to assess.
In The Netherlands, the area under sugar beet was 83,000 ha in 2006, 10.6% of the total
arable area. Good agricultural practice limits the maximum area to 180,000 ha. Changing land
use will consist of replacing the least profitable crops first, e.g., grass seed, maize and some
small crops. Grass seed and maize are feed crops and replacing them on a large scale may
ultimately increase milk/meat prices. Sugar beet will only be grown for ethanol, if it is more
profitable than sugar production, depending on the EU market regulations for sugar. These are
being reformed towards world market prices, improving the scope for bio-ethanol production.
However, average cultivation costs were €1400 ha–1. At yields of 6.1–6.4 t ha–1, ethanol has to
receive about €225 t–1 to compensate variable costs.
References
Bos, J. et al., 2007. Energy use, GHG emission and C-sequestration, report 140, WUR, 79 pp.
Cramer, J., et al., 2007. Testing framework for sustainable biomass. Final report from the
project group ‘Sustainable production of biomass’. Commissioned by the Energy
Transition’s Interdepartmental Programme Management (IPM), 62 pp.
IPCC, 2006. IPCC guidelines for national GHG inventories. Volume 4, Hayama, Japan.
Mortimer, N.D., et al., 2004. Energy and GHG emissions from bio-ethanol production from
wheat grain and sugar beet production. Final report for British Sugar plc., UK, 180 pp.
Yang, H. & B.H. Janssen, 2000. Eur. J. Soil Science 51: 517-529.
395
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Energy maize for phytoremediation of metal-enriched soils and production
of energy: The Campine region, Belgium
V. Van Slycken, N. Witters, E. Meers, K. Adriaensen, T. Thewys,
J. Vangronsveld, F.M.G. Tack
Laboratory of Analytical Chemistry and Applied Ecochemistry,
Ghent University, Coupure Links 653, 9000 Gent, Belgium
Contact: stijn.vanslycken@ugent.be
Introduction
Atmospheric deposition of trace metals (e.g., Cd, Zn and Pb) from metal refinery activities
over the last century has caused elevated concentrations in agricultural soils in the Campine
region. Regional policy therefore prescribes that the soils should be remediated, while at the
same time it is desirable to keep the income of the farmers constant. Both goals can be
achieved using phytoremediation in combination with the growth of energy crops and brings
us to the concept of a multi-functional biomass system (Berndes et al., 2008).
Phytoremediation involves the use of plants for the removal of pollutants from the
environment or to render them harmless (Garbisu & Alkorta, 2001).
Methods
For the study of phytoremediation potentials with energy maize samples were collected from
a field (6ha) situated on a moderately contaminated soil in Lommel, Belgium. The planting
and management of this site is part of a demonstration project in the context of the
INTERREG-project BENEKEMPEN.
The concentrations of Cd in this study area vary between 0.5 and 12 mg Cd kg–1 soil. Other
trace metals (e.g., Cu, Pb and Zn) together with the pH and soil conductivity were more
homogeneously distributed throughout the field. As the concentrations of Cd in the sampled
plots are exceeding the threshold values for remediation for agricultural land (2 mg kg–1) in
Flanders, this study will mainly focus on phytoremedation purposes of Cd. In this field 1 ha
was reserved for investigations with energy maize. Regional policy prescribes that the soils
should be remediated, while at the same time it is desirable to keep the income of the farmers
at a constant level. Therefore, the impact on the revenue of the farmer originating from the
phytoremediation activities is calculated. To take into account the uncertainty involved,
sensitivity analyses are performed for several variables. The final aspect included is the
energy component of the project: an input-output ratio is calculated and avoided CO2
emissions are given.
Results and discussion
In the site for screening of energy maize, the total concentration measured in an aqua regia
destruction is 5 ± 1 mg Cd kg–1 soil. In 2007, a yield of 20 ± 3 ton dry biomass/ha was
obtained. No significant difference in Cd concentration between the cultivars could be
measured for each plant compartment (stem, leaves, bract, rachis, grain). Also no great
significant difference was found in yearly removal of trace element between different the
investigated cultivars, so that the yearly removal of Cd can be estimated at 18 ± 6 g Cd ha–1.
For total removal of 5 mg Cd kg–1 soil to 1.2 mg Cd kg–1 soil (acceptable value) more than
800 years will be needed when energy maize is used.
Nevertheless, the Cd concentration in the biomass is exceeding legal threshold values for
fodder crops (1.1 mg Cd kg–1 dry matter) which implies that the biomass produced cannot be
used as fodder but must be applied for other industrial purposes such as energy generation.
Batch-tests for anaerobic digestion, performed at OWS (Organic Waste Systems, Belgium)
showed no difference in biogas potential of the silage of the contaminated maize in
396
Session C1: Integrated assessment of biomass for biofuel or renewables
comparison with a reference material. This points out that energy maize does not have as its
main goal remediation of the contaminated soil, but as a more valuable alternative for the
farmers. Energy maize as an alternative crop offers great potential, but further research on
metal balance in this process and the disposal of the digestate is still ongoing.
Economic measurements need to be included to fully evaluate phytoremediation purposes.
We start from the assumption that basic activities remain (dairy cattle rearing). Therefore,
energy maize is grown and fodder maize has to be bought outside the contaminated area.
When the biomass from energy maize is used for energy production by digestion, the basic
income of the farmer can be supported by using and selling renewable energy (heat and
electricity) out of this polluted biomass. When farmers digest the maize themselves and
former activities and revenues thereof remain, their average yearly net income will grow with
€ 227 ha–1, to be added to the basic income of € 1.047 ha–1 (2005). However, Risks, legal
norms and extra efforts needed in such a project cannot be underestimated. A farmer can
therefore decide to sell the polluted energy maize by contracts with an energy partner that
converts the polluted biomass into energy. In the latter case, the average yearly income of the
farmer grows with € 115 ha–1.
As already mentioned, the project is conceived as a multi-functional biomass system.
Another environmental benefit, besides remediation, of this project is the production of
‘green’ energy. Taking into account all energy input (going from transport of fodder to
processing the waste product), digestion of energy maize delivers for each part of fossil
energy input, 8–12 parts of renewable energy. One hectare of energy maize delivers a net
yield of 80 GJ electricity and 30 GJ heat, to be sold locally or to be used on the farm. In a
traditional installation (coal), this amount of electricity would cause an emission of 9.6 t CO2.
To study the possibilities for phytoremediation of trace elements in the Campine region,
using energy crops, an integrated study is needed. The agronomical study of energy maize
shows that the produced biomass is comparable with energy maize grown on noncontaminated sites. Because of the low concentration of trace elements in the different plant
parts, a long remediation period is needed. Nevertheless, the high biomass and low
concentration are very promising for its use in the non-food industry, including the
provisional results for usage in energy generation by anaerobic digestion. Moreover, the
economic outlook for the farmers in the contaminated region is positive, as simulations have
shown. In this paper, energy maize is, therefore, presented as a sustainable alternative for
traditional farming activities. Research is also ongoing for other crops like short rotation
coppice and rapeseed. These crops can show even higher extraction rates, but have
disadvantages concerning economic, energetic or social aspects. All this information
combined with metal balances in the different energy flows will allow full comprehension of
the feasibility of the various phytoremediation approaches for a safe management of metal
enriched agricultural soils.
References
Berndes, G., et al., 2008. Biofuels, Bioprod. Bioref. 2: 16-25.
Garbisu, C. & I. Alkorta, 2001. Bioresource Technology 77: 229-236.
397
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme C
Case studies and application of tools and
empirical methods
Session C2: Modelling adaptation strategies to cope
with climate change
Session organizers:
Jim Hansen
Int. Research Institute for Climate Prediction, New York, USA
Lennart Olsson
Lund University, Sweden
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Will global mitigation policy enhance or undermine local adaptation?1
H. Meinkea, R. Nelsonb
CCSA, Wageningen University, The Netherlands, bCCAF, CSIRO, Australia
Contact: holger.meinke@wur.nl
a
Introduction
Adapting to climate change combined with providing the policy frameworks that facilitate
sound adaptation is essential for the survival of our agricultural sectors. Yet decision makers’
research needs on both sides – practice and policy – are often neglected as their interests cross
disciplinary and institutional divides. This can lead to maladaptation, as shown by some of the
recent expansion of biofuel production in the wake of policies with unintended consequences.
Bridging the practice – science – policy divide requires all three to adapt. Proactively
designed and sustainable adaptation action will only occur if and when climate-related risks
are treated holistically in conjunction with other drivers of risk (e.g., market, environment or
social risks), supported by policies that take multiple domains and outcomes (e.g., sustainable
development) into account. We call for adaptation science to provide integrated vulnerability
assessments that are policy relevant and trigger regionally appropriate adaptation responses.
Results and discussion
Proactively designed adaptation does not come easily to a sector that values tradition and
whose decision needs are rarely met by the climate change science community. Adaptation
requires changed attitudes and practice by all participants, including the science and policy
communities (Nelson et al., 2009a, b) and the recognition that science will only ever provide
partial answers to societal problems (Jasanoff, 2007). The insidious nature of ongoing climate
variability and future change poses a particular challenge: climate is a widely acknowledged
risk factor for most agricultural activities, but without being the sole or even dominant driver
for most of them. Yet without due consideration of climatic impacts, the dual goals of
agricultural production – profitability and sustainability – cannot be achieved. Further, the
considerable opportunities that are created by good climatic conditions and new, climaterelated policy measures often fail to translate into real benefits.
Resource
management
Climate Warning
Adapt
now
farm
Mitigate
TEMPORAL
SPATIAL
catchment region
Improved
livelihoods
future
state
country
Integrated
adaptation science
SECTORAL
enterprise business industry sector economy
Well-informed policy
development
Figure 1. Temporal, spatial and sectoral dimensions of adaptation and mitigation (left) and the
role of integrated adaptation science to inform policy as well as practice (right).
Garnaut (2008) states that ‘Contemplating the adaptation challenges … helps to focus our
minds on the more difficult dimensions of mitigation choices’. This is a call for the proactive
design of adaptation options backed by well-informed policies, an essential requirement for
vibrant rural sectors in charge of their own destinies (Figure 1). Adaptation science (AS), a
special form of sustainability science at the boundary between science and society, can build
social networks that institutionally connect agricultural and climate science with decision
1
Keynote presentation
400
Session C2: Modelling adaptation strategies to cope with climate change
makers, thus generating ‘social capital’ needed to create adaptive capacity. Through new
‘boundary-spanning organizations’ (Guston, 2001), AS provides novel applications that
explicitly recognize science’s ability to reduce, but not to eliminate uncertainty.
Australia, which only recently committed itself to the Kyoto targets, has a rich history of
applied climate risk management (due mainly to its highly variable, semi-arid climate) and
plays a key role in agricultural climate adaptation research (Howden et al., 2007; Meinke et
al., 2007). Garnaut (2008) identified seasonal climate forecasts as a key technology in
Australia’s adaptation challenge. Yet their impact has been disappointing, a direct
consequence of their low compatibility with decision making under uncertainty (Hayman et
al., 2007). The insidious nature of climate results in highly variable co-limitations that cannot
be overcome via single technological fixes. By defining forecast quality as the characteristics
of a forecast product and/or forecast service that enables action and satisfies identified and
agreed needs of the user community, the issue of co-limitations could be addressed. It is
therefore paramount not to focus on single scientific measures of a forecast (e.g., skill or lead
time). Instead, science is required to acknowledge co-limiting factors such as knowledge
barriers, sound governance and the ability to compare choices, chances and consequences.
Based on a review of supply and demand for integrated vulnerability assessments, we
conclude that our conceptual understanding of the issues has progressed to the point where it
is no longer acceptable to substitute impact modelling for integrated vulnerability assessments
when providing policy advice. For instance, confining an analysis to biophysical impacts
suggests that inland Australia is most vulnerable due to high exposure to a variable climate.
When farm incomes are used as a more integrative measure of exposure, the spatial
vulnerability of agricultural communities becomes considerably more complex (Figure 2).
Figure 2. Vulnerability of rural
Australia to climate variability:
Pasture growth variability vs
adaptive capacity (left) and farm
income variability vs adaptive
capacity (right).
no data
least vulnerable
moderately vulnerable
most vulnerable
We further conclude that policy relevant vulnerability assessments that support adaptation
action flow from collaboration between scientists from diverse disciplines and agencies. Our
work is an overt attempt to create policy relevant measures of vulnerability that trigger
appropriate action by or on behalf of specific individuals, communities and governments to
reduce it. We show how interdisciplinary collaboration can overcome methodological
challenges to providing policy relevant vulnerability assessments while impact modelling can
lead to entirely erroneous conclusions about the vulnerability of agricultural communities.
Rural communities that are vulnerable to climate variability and change tend to be vulnerable
for a complex set of interacting environmental, economic and social reasons.
References
Garnaut, R., 2008. Cambridge University Press, Port Melbourne, Australia, 634 pp.
Guston, D., 2001. Cambridge University Press, Cambridge.
Hayman, P., et al., 2007. Australian Journal of Agricultural Research 58: 975-984.
Howden, S.M., et al., 2007. PNAS 104(5): 19691-19696
Jasanoff, S., 2007. Nature 450: 33.
Meinke, H., et al., 2007. Aust. J. Agric. Res. 58: 935-938.
Nelson, R., et al.., 2009a, b. Part I & II - Environmental Science and Policy. (submitted)
401
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Exploring options to improve adaptation to climate change in crop
production of south-eastern South America
W.E. Baethgen, J.W. Hansen, A.V.M. Ines, A.M. Greene, L. Goddard
IRI, 61 Route 9W, Monell Building, Palisades, NY 10964-8000, USA
Contact: baethgen@iri.columbia.edu
Introduction
Even under the most optimistic scenarios of globally coordinated actions to drastically reduce
the net emissions of greenhouse gases (GHG) during the next decades, climate science
confirms that warming is already unavoidable due to past emissions (Meehl et al., 2007), and
therefore adaptation will be necessary to address the impacts resulting from such global
warming. Much of the research on expected impacts of climate change and options to
improve adaptation to possible climate change scenarios (IPCC WG2, 2007) does not include
any measure of the uncertainties that are intrinsic to the climate projections, and assumes that
technology remains unchanged throughout the decades that are covered in the study period.
These two limitations can drastically reduce the value of climate change research for
informing policy and decision making. Decisions that are based on deterministic climate
scenarios (i.e., scenarios lacking any measure of uncertainty or probabilistic ranges) are likely
to be inadequate and may even be maladaptive. Furthermore, policies based on adaptation
studies on adaptation to climate change that consider possible changes in technology can be
directed to stimulate the use of technologies that reduce vulnerability to climate.
We describe methods that are intended to overcome the limitations mentioned above. We
propose methods for establishing a suite of possible climate change scenarios based on GCM
projections and on the changes observed in the climate over the last several decades. We then
use crop simulation models to explore different levels of adaptive measures including
technological changes.
Methods
Our study region is Southeastern South America (SESA: southern Brasil, Uruguay, central
Argentina), one of the main world’s food baskets. We use three methods for establishing
climate change scenarios. First, we modify the observed weather data (last 30 years) with
anomalies (i.e., GCM climate change scenarios minus GCM climatology) from GCMs
available from IPCC to create a climate dataset with possible expected anomalies in the
means and no change in the variability for each one of the GCMs available from IPCC. We
may also explore ensembles having different relative weights for the component GCMs (e.g.,
Greene et al., 2006.). Second, we statistically characterize the changes observed in
temperature and precipitation in SESA throughout the last 70 to 100 years, and use the
observed changes in precipitation and temperatures (means, variability, dry spell length, storm
intensity, etc.), to project the observed trends into the near future. Third, we identify
coincidences in the GCM projections for the region (e.g., changes in temperatures and rainfall
in different seasons), and use stochastic weather generators (Semenov & Barrow, 1997;
Semenov, 2007) to create ‘synthetic’ daily weather datasets that incorporate the changes for
which most of the GCM agree (e.g., summer precipitation increase, minimum temperature
increase, etc.). The method is also used to create new datasets that increase the variability of
rainfall (both amounts per rainfall event and number of rainy days) and temperature, and
hence increase frequency of extreme events.
Results and discussion
This process results in a large number of possible future climate scenarios that that reflect
different sources of uncertainty (e.g., uncertainty related to socioeconomic scenarios used in
402
Session C2: Modelling adaptation strategies to cope with climate change
the GCMs, uncertainties on the projections of changes observed in the last 70-100 years, etc.).
The suite of possible climate change scenarios are thereafter used as input for crop simulation
models (in our study we use one for wheat and one for maize) to study the impact of those
possible climate scenarios on the expected yields in the study region. The models that we use
(DSSAT) allow for considering different technologies (different cultivar characteristics,
sowing dates, fertilizer strategies, irrigation, etc.) and therefore, we explore adaptive measures
that work best under the various expected climate scenarios. We explore different degrees of
technology changes: from selecting cultivars that already exist, to improved crop breeding
and pest management strategies that are under development and/or can be the result of future
investments in adaptation.
The results are expressed in formats (graphs, maps) that include measures of the
uncertainty (e.g., median, 10th, 90th percentiles; probability of exceedance of given values,
etc.), as opposed to only mean expected values. The research is helpful for the climate science
community to emphasize the need of connecting the climate data with tools that ‘translate’
climate scenarios into results with socio-economic relevance. Furthermore, it is useful for the
agricultural community to show the need and benefits of using several methods to establish
climate change scenarios which will allow communicating the uncertainties of the results.
Furthermore, it demonstrates the importance of including technology changes in climate
change impact studies.
References
Greene, A.M., et al., 2006. J. Climate (19)17: 4326-4343.
IPCC WG2, 2007. In: Eds M.L. Parry, et al., Climate Change 2007: Impacts, Adaptation and
Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge,
UK, 976 pp.
Meehl, G.A., et al., 2007. In: Eds S. Solomon, et al., Climate Change 2007: The Physical
Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge,
UK.
Semenov, M.A., 2007. Agric. Forest Meteorology 144: 127-138.
Semenov, M.A. & E.M. Barrow, 1997. Climatic Change 35: 397-414.
403
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessment and predictions of agricultural drought for the
Upper Blue Nile Basin, Ethiopia
Y.G. Bezabij, R. Sintanyeho, A. Van Griensven
Department of Hydroinformatics and Knowledge Management,
UNESCO-IHE Institute for Water Education, P. O. Box 3015,
2601 DA Delft, The Netherlands
Contact: bezab1@unesco-ihe.org
Introduction
Recent scientific studies indicate that drought is driven by natural variation of climate forced
by the internal interactions of the atmosphere and feedbacks from the oceans and land surface.
This makes drought phenomenon dependent not only on local but also on global weather
circulation and interactions. ENSO is attributed to the occurrence of drought in most parts of
the globe where its impact causes economic losses for developed and loss of life for
developing nations.
Methods
To successfully mitigate these impacts of drought necessitates reliable forecasting tools with
longer lead time. Integration of physically based models with data driven models can give an
answer by compensating each other’s drawbacks. Here, the physically based hydrologic
model ‘Soil and Water Assessment tool’ (Arnold et al., 2002) is used in combination with
recursive feed forward Artificial Neural Network (ANN) data driven provide predictions in
the form of short term (couple of days) forecasts of soil moisture conditions, mid-term early
warning of ENSO events (seasonal) and long-term predictions (several years) of climate
change impacts on soil water and the cropping season.
The models are developed by making
use of open source remotely sensed data
and GCM outputs downscaled using
statistical downscaling method. In this
study an attempt is made to knit the
oceanic sea surface temperature (SST)
with the land hydrological variables. The
SWAT model, being calibrated and
validated by ground stations data, and
the ANN are used for the assessment and
prediction of the soil moisture stress.
Two indicators are used: the Soil Water
Lake Tana
Deficit Index and the evapotranspiration
deficit indices (ETDI) as proposed by
-3.99- -3
Narasimhan & Srinivasan (2005).
-2.99 - -2
-1.99- -1
Results and discussion
As seen in Figure 1, the north, central
and western part of the Blue Nile region
is most affected by drought (negative
SMDI).
-0.99 - -0.50
-0.5 - 0.5
0.5 - 1
1-2
Figure 1. Soil Moisture Deficit Index based on
SWAT simulations for the Upper Blue Nile (January
2005) (Tessema, 2007).
404
2-3
Session C2: Modelling adaptation strategies to cope with climate change
It is known that drought left a scar and that this will be worsened by the impacts of climate
change and by desertification. The need to mitigate the impacts of drought asks for an
integration of our presented forecasts and predictions into communication tools for
stakeholders and decision makers. The short-term forecasts are useful for day to day crop
agricultural and water management, the medium term provides an early warning to activate
planning for disaster prevention and the long term predictions play a role in the assessment of
drought trend under the continuing climate change caused by either natural or anthropogenic
factors. These predictions are to be incorporated in strategic planning and policy considering
the effects of climate change.
Acknowledgement
The authors wish to thank the Nile Basin Capacity Building Network for supporting the
research of Yisehak Gebregziabher Bezabij.
References
Arnold, J.G., et al., 2002. Soil and water assessment tool, theoretical documentation,
Grassland, soil and water research laboratory – Agricultural Research Service, Texas, US.
Narasimhan, B. & R. Srinivasan, 2005. Agricultural and Forest Meteorology 133: 69-88.
Tessema, R., 2007. Agricultural Drought Assessment for Upper Blue Nile Basin, Ethiopia
using SWAT. MSc Thesis, UNESCO-IHE, Delft, The Netherlands.
405
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Pre-requisites for crop models used to test strategies for adapting to, or
mitigating effects of climate change
K.J. Boote, J.W. Jones, G. Hoogenboom, C. Rosenzweig
Agronomy Dept, Univ of Florida, Gainesville, FL 32611-0500, USA
Contact: kjboote@ufl.edu
Introduction
Climate change for a given region can be characterized by increased day and/or night
temperature, increased carbon dioxide (CO2), altered rainfall frequency and intensity, as well
as extended growing season. Simulation models for different crops can be used as strategic
tools to evaluate the consequences of climate change on production, and to evaluate shifts in
crops, sowing dates, cultivars, irrigation, and fertility management practices to adapt/mitigate
the effects of climate change. Genetic improvement in heat and drought tolerance can be
tested in a hypothetical manner. Mitigation strategies to enhance C sequestration can be
tested. There are several pre-conditions to successful use of crop models for such strategic
tests. First, the crop models must be accurately parameterized as to temperature and CO2
effects on growth and development. This is not easily accomplished, as the modelling
community has not focused sufficiently on full testing and improvement of crop models
relative to the latest scientific literature, and some models are too simple in their ability to
fully reproduce interactions of these climate change factors. Secondly, accurate representation
of the effects of rainfall variation requires two pre-conditions: good models of soil water
balance and evapotranspiration as well as accurate parameterization of soils for the specific
regions. Insufficiently detailed mechanisms in the model can be a limitation, particularly
relative to simulated responses to CO2, increased temperature, increased vapour pressure
deficit, and their interactions. The third pre-condition, is that the crop models must have good
soil temperature prediction, good prediction of emergence and early growth, as well as code
to mimic frost or freeze susceptibility in the seedling phase and grain-filling phase. Models
lacking frost or freeze damage are poorly suited to evaluate the concepts of varying sowing
dates to escape effects of hot temperatures or drought in summer.
Methods
Example simulations are presented for climate change impact assessment and adaptation
options for soybean and maize production in Iowa, Florida, and India, with and without
irrigation, where variables of sowing date, cultivar, and N fertilization were modified. These
three cases represent contrasting environments of continental climate with a good soil in
Iowa, subtropical climate with a sandy soil in Florida, and tropical arid monsoon climate in
India with a moderately good soil. These cases have either rainfall or frost hazards that limit
growing seasons. The soybean, maize, and sorghum models in DSSAT V4.0 (Jones et al.,
2003) were used to predict growth and yield for 30 years of historic weather, compared to
weather from climate change projections for those regions by 2100. Consequences of very
early to very late sowing dates were illustrated, to highlight differences in the three models
relative to temperature or drought termination of growth. CROPGRO-Soybean has explicitly
defined Tmin values, which reduce leaf area index and/or terminate crop development, from
frost damage and freeze kill, respectively (Boote et al., 1998). CERES-Maize lacks actual
freeze kill, but does have a reasonably robust simulation of crop maturation due to slowed
grain filling (less than 10% of ‘normal’ for 5 consecutive days) or ‘crop failure’ caused by
near-zero assimilate supply attributed to low temperature effects on radiation use efficiency,
limited water supply, or N stress. CERES-Sorghum lacks such signalling and may simulate
through freezing winter into the next year. Hypothetical genetic improvement in life cycle,
daylength sensitivity, heat-stress tolerance of grain-set, and drought tolerance were evaluated.
406
Session C2: Modelling adaptation strategies to cope with climate change
Maize hybrids with longer life cycles or soybean maturity groups with greater photoperiod
sensitivity may be needed to fully utilize longer growing season in good regions. Alternately,
shorter cycle cultivars sown early may be needed to avoid rainfall deficits in mid-summer, as
practiced for soybean production in some regions such as Texas, Arkansas, and Mississippi.
Costs of fuel, N fertilizer, irrigation, and drying versus crop value can be evaluated to decide
on optimum economic crop or biofuel choice for a given farmer, using the pricing structure in
the seasonal analysis of the DSSAT software. To adequately consider crop choice, one should
have models that consider the disease-pest benefits of rotation as well as the carry-over of N
and water from one season to the next.
Results and discussion
For the Ames, Iowa location under current weather, early sowing (May 1) was optimum for
maize and soybean yield, while yield was progressively reduced for later sowing because the
crop used less of the available season. As sowing was progressively delayed from June 1 into
July under current weather, the simulated maize and soybean crop yields were reduced
because of shorter season and increased occurrence of freeze in fall before maturity. Full
season maize and MG 3 soybean were optimum cultivar types. Sorghum simulations did not
perform well at Ames, Iowa, especially if sowing was delayed.
For the Gainesville, Florida location under current weather, the optimum sowing date for
maize was March 1 if irrigated, but was May 1 and later under rainfed conditions.
Interestingly, late sowing in Florida has not been a common production practice because of
insect damage, but that may change with new insect-resistant hybrids. For soybean, April 15
sowing date and MG 8 cultivar were optimal under rainfed and irrigated conditions.
For the Patancheru, India location under current weather, the optimum soybean maturity
group was MG 8, as shorter cycle cultivars had lower yield, and longer cycle cultivars had
similar high yield, but left less residual soil water. The optimum sowing date for soybean and
maize at Patancheru would be as early as possible after onset of reliable monsoon. Under
rainfed conditions, later sowing caused lower yield and also less residual water in the soil.
Under irrigated conditions, maize yield was increased with later sowing, because growth was
shifted to a cooler time of year, even under current weather. Soybean yield was increased
under irrigation, but yield was reduced with later sowing because of daylength effects to
shorten the life cycle.
It is important for future global climate change assessments, that crop models be improved,
tested, and be well-parameterized for temperature and CO2 effects on photosynthesis, grainset, grain-filling rate, and grain-filling duration, especially for use under temperature
conditions of future climate. The models also need to include termination of growth due to
freeze kill or terminal water deficit. The different model formulations tested here, illustrate
the value to consider multiple viable crop models in such assessments, rather than just one
selected model per crop. Just as the various global climate models give different climate
projections, different models for the same crop will also vary in their predicted crop response
to a given climate scenario, thus providing information to improve crop models and
understand uncertainties.
References
Boote, K.J., et al., 1998. In: Eds R.M. Peart & R.B. Curry, Agricultural Systems Modeling
and Simulation. Marcel Dekker, Inc, New York, pp. 651-692.
Jones, J.W., et al., 2003. Europ. J. of Agronomy 18: 235-265.
407
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A minimum-data approach to ex-ante assessment of climate change
adaptation strategies in resource-poor countries
L. Claessens, J.J. Stoorvogel, J.M. Antle, R.O. Valdivia, M. Herrero
International Potato Center, P.O. Box 25171, 00603 Nairobi, Kenya
Contact: l.claessens@cgiar.org
Introduction
The changing climate is exacerbating existing vulnerabilities of the poorest people who
depend on semi-subsistence agriculture for their survival (Slingo et al., 2005; IPCC, 2007).
Sub-Saharan Africa (SSA) in particular is predicted to experience considerable negative
impacts of climate change (e.g., Thornton et al., 2006). The IPCC Fourth Assessment
emphasizes that adaptation strategies are essential and these must be developed within the
broader economic development policy context (IPCC, 2007). Addressing adaptation in the
context of small-scale, semi-subsistence agriculture in SSA raises special challenges that
cannot be addressed adequately by the approaches taken thus far in most studies (Adger,
2003). Most of the existing research has focused on impacts of climate change and adaptation
to climate change in the agricultures of industrialized countries. In the relatively few studies
conducted in Africa, agricultural research has either focused on individual crops(e.g., Jones &
Thorton, 2003), has used aggregated data and models (e.g., Winters et al., 1998, Mendelsohn
et al., 2000), or used statistical analysis too general to be useful for site-specific adaptation
strategies (e.g., Kurukulasuriya & Mendelsohn, 2006). One of the important constraints to
carrying out this type of research is that the data demands are high, because site-specific biophysical and economic data are required, typically obtained from costly multi-year farm-level
surveys. The development and application of relatively simple and reliable methods for exante evaluation of adaptation strategies at the household and system levels are needed to
provide timely assessments of the potential impacts in the context of climate change.
Methods
This paper describes and applies a new approach to ex-ante impact assessment that produces
locally useful, site-specific results that can also be aggregated for regional policy analysis.
The methodology makes use of the kinds of data that are more often available, especially in
resource-poor countries. The stochastic approach uses and integrates available socioeconomic and biophysical data on farmers’ land use allocation, production and input and
output use. Spatially heterogeneous characteristics of the agricultural system regarding
resources and productivity are analysed and compared for both current climate conditions and
predicted climate changes. A variety of possible adaptation strategies is then assessed for their
capability to overcome or reduce the adverse effects of climate change. A static expected
profit maximization model is used to characterize the opportunity cost of adaptation (Antle &
Valdivia, 2006). The model represents the impact of climate change as the ‘compensating
variation’, i.e., the loss in income that producers experience relative to the base climate
scenario.
Results and discussion
We apply the methodology to the mixed crop-livestock system of Vihiga district in western
Kenya. After characterizing the current agricultural system with actual climate data, the
effects of a perturbed climate on biophysical and economic indicators are analysed and a
variety of adaptation strategies (agricultural technologies and policies) are tested (results from
earlier work in Machakos, Kenya, are given in Figure 1). Despite the limitations, the
minimum-data approach offers a flexible framework for evaluating adaptation strategies using
scarce data of resource-poor countries in SSA and other parts of the world. It allows a rapid
408
Session C2: Modelling adaptation strategies to cope with climate change
integrative analysis for timely advice to policymakers and for exploration of technology and
policy options.
Figure 1. Impact of climate change and adaptation strategies on farmers in Machakos, Kenya.
References
Adger, W.N., et al., 2003. Progress in Development Studies 3: 179-195.
Antle, J.M. & R.O. Valdivia, 2006. American J. of Agricultural Economics 88: 1174-1180.
IPCC (Intergovernmental Panel on Climate Change), 2007. Climate Change 2007: Impacts,
Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change, Cambridge University Press,
Cambridge, UK.
Slingo, J.M., et al., 2005. Philosophical Transactions of the Royal Society, London, Series B
360: 1983-1989.
Jones, P.G. & P.K. Thornton, 2003. Global Environmental Change 13: 51-59.
Kurukulasuriya, P. & R. Mendelsohn, 2006. Crop Selection: Adapting Climate Change in
Africa. CEEPA Publication 26. http://www.ceepa.co.za/docs/CDPNo26.pdf
Mendelsohn, R., et al., 2000. Climatic Change 45: 553-569.
Thornton, P.K., et al., 2006. Mapping climate vulnerability and poverty in Africa. ILRI,
Nairobi, Kenya, May 2006, 200 pp.
Winters, P., et al., 1998. Environmental and Resource Economics 12, 1-24.
409
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Sensitivity of southern African maize yields to the definition of onset under
conditions of climate change
O. Crespo, S. Hachigonta, M. Tadross
Climate Systems Analysis Group, EnGeo Department, University of Cape Town,
Private Bag X3, Rondebosch 7701, Cape Town, South Africa
Contact: olivier.crespo@csag.uct.ac.za
Introduction
Climate change has led national and international decision makers to look for adaptation
guidelines and decision tools regarding its potential impacts. The likely impacts can be
simulated using climate sensitive models, which simulate the system of interest, subject to
future hypothetical climate conditions. With respect to agricultural impacts previous work has
simulated the variability of yield and soil organic nitrogen levels over three South African
climate regions (Walker & Schulze, 2008), whilst Lobell et al. (2008) studied crop adaptation
over southern Africa using multiple General Circulation Models (GCMs). Here we present a
simulation-based method (using the FAO agrometshell crop model) aimed at analysing the
sensitivity of yield (as represented by the Water Satisfaction Index - WSI) to the definition of
climate-related adaptation options (i.e. rainfall onset) under statistically downscaled control
and future climate scenarios.
Ultimately the work will be expanded to include a variety of crop models and simulate the
sensitivity to other management options such as application of irrigation and fertilizer, as well
as including additional uncertainty attributable to the parameterization of crop growth within
the models. The objective of the work is to highlight which adaptation options are likely to
generate the most effective changes in yield given the current climate and uncertainty in the
projections of future climate conditions in each region.
Materials and methods
The preliminary stage of this study uses the Agrometshell crop model, developed by the Food
and Agriculture Organisation’s, Environment and Natural Resources Service (Agrometshell,
2004). The model is based on crop-specific water balances and simulates the behaviour of
crops over southern Africa, forced by statistically downscaled climate data from 6+ General
Circulation Models (GCMs)
The initial step is to run the crop model 1000 times for each year, varying the onset
definition (due to the assumed total rainfall needed before and after planting) in each case.
The sensitivity part of the method is based on the Fourier amplitude sensitivity test as
introduced by Saltelli et al. (1999). This allows us to understand how the amount of rainfall
before and after planting, contributes to the WSI (yield) variance. The sensitivity of yield (to
definition of onset) under present climate conditions highlights regions where adaptation
decisions about changing sowing dates and definitions significantly alter yields. By repeating
this using the long-term future (2050s) scenarios of climate change, we can evaluate if these
adaptation decisions/options are robust given change in future climate.
Results and discussion
Figure 1 shows the change in Maize WSI when applying the statistically downscaled climate
scenario (GFDL model) to the Agrometshell crop model, as an example of the information
that is generated. Maize WSI calculations for each individual year are based on a matrix of
1000 simulations defining the start on the growing season. A total of 20 years is used for
control (1979–1999) and future (2046–2065) scenarios.
The model outputs show crop reduction (approx. –9%) for maize in southern parts of
Zimbabwe and some slight changes (approx. +5%) are observed over region around Lesotho
410
Session C2: Modelling adaptation strategies to cope with climate change
(Figure 1a). Maize WSI is projected to increases in most parts of Zambia, parts of east South
Africa and northern Mozambique. Figure 1b shows the change in the contribution of rainfall
before planting to the final WSI between the current and future climate as simulated by the
GFDL model. The model shows a significant increase in the sensitivity to pre-planting rainfall
around Lesotho (approx. +20%) and a significant decrease in sensitivity over Malawi (about 15%). Although the projected changes in sensitivity to onset definition and total WSI differs
between models, similar changes emerge over some regions, e.g., central Zimbabwe.
(a)
(b)
Figure 1. (a) Change in mean WSI over the region, and (b) % change in mean X1 (rainfall in
first 10 days after planting date) sensitivity.
References
Agrometshell, 2004. Food and Agriculture Organisation of the United Nations, Environment
and Natural Resources Service.
Lobell, D.B., et al., 2008. Science 319: 607-610.
Saltelli, A., et al., 1999. Technometrics 41: 39-56.
Walker, N.J. & R.E. Schulze, 2008. Agriculture, Ecosystems & Environment 124: 114-124.
411
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessment of nitrogen and green house gas fluxes at the European scale in
response to land use and livestock change
W. De Vries, J. Kros, G.J. Reinds, H.J.J. Wieggers, G.L. Velthof, D.A. Oudendag,
G.J. Nabuurs, M.J. Schelhaas, M. Pérez-Soba, W.P. De Winter, J. Van den Akker,
M.M. Bakker, P.H. Verburg, J. Liski, B. Eickhout, A.F. Bouwman
Alterra, Wageningen University and Research Centre,
P.O. Box 47, 6700 AA Wageningen, The Netherlands
Contact: wim.devries@wur.nl
Introduction
Studies on responses of N and GHG emissions in Europe to global change strongly focused
on the impacts of climate change and land use change on carbon exchange (e.g., Smith et al.,
2005). Moreover, considerable attention has been given to the combined impact of climate
change and forest management on the European (forest sector) carbon budget (e.g., Nabuurs
et al., 2002). A study on the combined impacts of land use change, agricultural change
(management) and forest management on all major N and GHG fluxes (NH3, NO, N2O, CO2
and CH4 emissions and N leaching and runoff) at the European scale is still missing. This
paper describes such a study, using an integrated model called INTEGRATOR (De Vries et
al., 2008), based on a detailed schematization to assess N and GHG emissions from both
agricultural and natural terrestrial ecosystems, including grassland, arable land, forests,
heathlands and peat lands. The INTEGRATOR model is presented and its predictions of past,
present and future N and GHG fluxes in response to (i) reconstructions of land use and land
management changes in the period 1970–2000 and (ii) projections for the period 2000–2030
under the IPCC A1 and B2 scenario.
Methods
INTEGRATOR links various modules, calculating N and GHG emissions from industrial
sources, housing and manure storage systems, agricultural and non-agricultural soils and
surface waters, while accounting for the interaction between different sources through an
emission-deposition model for NH3 and NOx. To assess the impact of scenarios and policies
(e.g., measures related to various EU directives) on future N and GHG emissions,
INTEGRATOR is coupled with models that predict changes in land cover and agricultural
management and climate in response to such scenarios and policies. The INTEGRATOR
concept is based on an appropriate balance between model complexity and data availability
by: (i) using relatively simple and transparent model formulations based on the use and
adaptation of available simple model approaches, (ii) including empirical model approaches,
using statistical relations between model outputs and environmental variables, and (iii)
focusing on the derivation of high resolution spatially explicit input data. INTEGRATOR
includes sub models, to predict:
• NH3, NOx, N2O and CH4 emissions from housing and manure storage systems and
agricultural soils (the adapted MITERRA-Europe model, Velthof et al., 2008);
• NOx, N2O and CH4 emissions from non-agricultural terrestrial systems (empirical
relationships);
• CO2 emissions from agricultural and non-agricultural terrestrial systems (the YASSO soil
model, Liski et al., 2005), in combination with (i) EFISCEN (Nabuurs et al., 2002) for
forest, (ii) MITERRA-Europe for agriculture, and (iii) empirical relationships for
peatlands;
• N2O emissions from ground water and surface waters;
• N deposition (an emission-deposition matrix for NH3 and NOx, accounting for the
interaction between agricultural and non-agricultural soils.
412
Session C2: Modelling adaptation strategies to cope with climate change
To derive a complete N and GHG budget, background emissions and energy emissions are
also included in INTEGRATOR, based on literature data and IMAGE model calculations.
Results and discussion
At the conference results will be given of temporal changes in NH3, NOx, N2O, CH4 and CO2
emissions for the period 1970–2030 in response to (i) reconstructions of land use and land
management changes in the period 1970 and (ii) projections for the period 2000–2030 under
the IPCC A1 and B2 scenario. As an example, we present the NH3 and N2O emissions form
agricultural systems for the year 2000 as calculated with INTEGRATOR (using the adapted
MITERRA sub-model) for the year 2000 and standard approaches such as the IPCC reports
by the various countries and emission factor models, in view of N emissions, such as
EDGAR, GAINS and IMAGE (Figure 1). Results of total emissions appear to be quite
comparable for NH3 but for N2O there is considerable more scatter.
Figure 1. A comparison of country emissions for NH3, N2O and NOx within EU 27 as derived
with INTEGRATOR/MITERRA, compared to OECD/IPCC, EDGAR, GAINS and IMAGE
for the year 2000.
References
De Vries, W., et al., 2008. INTEGRATOR: A modelling tool for European-wide scenario
assessments of nitrogen budgets and greenhouse gas emissions. Calculation procedures and
application methodology of a prototype. Alterra Report, Wageningen.
Liski, J., et al., 2005. Ecol. Model. 189(1-2): 168-182.
Nabuurs, G.J., et al., 2002. Change Biol. 8: 304-316.
Smith, J.U., et al., 2005. Glob. Change Biol. 11(12): 2141-2152.
Velthof, G.L., et al., 2008. Integrated assessment of nitrogen emissions from agriculture in
EU-27 using MITERRA-EUROPE. Environ. Pollut. (in press)
413
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
SimKat: A virtual laboratory to explore the impact of climate change
scenarios on the Western Australian wheat-belt
A. Dray, S. Asseng, P. Perez, S.P. Charles, B. Bates
RMAP/RSPAS, Coombs Building, Australian National University,
Canberra ACT 0200, Australia
Contact: anne.dray@anu.edu.au
Introduction
The impact of climate change on an entire agricultural region is often not clear (IPCC Report,
2007) due to the complex interactions between individual farmers’ behaviour with the biophysical landscape, the large range of multiple external and internal factors and the further
complication of continuous changes to climate variables in time and space. As a striking
example, the wheat-belt of Western Australia is one of the most vulnerable regions to climate
change in Australia. Rainfall has already declined by more than 15% in the last decades and it
is projected to further decline. Farm numbers are plummeting and the natural resource base is
threatened by various soil degradation processes including salinity. The agricultural future of
the region is highly unpredictable due to the complex and adaptive nature of human-landscape
interactions. Nevertheless, policymakers relentlessly ask experts for such predictions to assist
in anticipating upcoming issues and to take immediate decisions to influence future socioeconomic and environmental settings of the region for the better. It is fair to recognize that, to
date, scientific research has provided answers to their questions that are only partly adequate
and the best available biophysical science is insufficient. It is now of critical importance to
understand the long-term consequences of climate change on these already threatened socialecological systems and to anticipate ways for local farmers to adapt.
Methods
This paper builds on a simulation model prototype that focussed on salinity changes in the
wheat-belt region of Western Australia (Asseng et al., 2008). The model used an agent-based
modelling framework and was developed with the CORMAS platform (Bousquet et al.,
1998). It combined simplified biophysical processes of paddock cover with an extension to
include CO2 impact, dry-land salinity changes and rainfall. Simulated farmers (agents) in the
model made individual land use decisions based on the performance of their past land cover
productivity and market returns. In addition, farmers in the model could display various
attitudes towards market signals and salinity mitigation.
In this extended version, we use the model as an exploratory tool to focus on likely climate
change scenarios and their impact on the viability of an agricultural region. We explored the
impact of the worst case scenario for CO2 atmospheric concentration trend given by the IPCC
Report 2007: the A1F1 storyline which corresponds to very rapid economic growth and
reliance on fossil intensive energy as currently the most likely scenario (Raupach et al.,
2007). Additionally, we overlap climate change trend impacting upon maximum yield grain
with technological trends influencing farmers’ ability to crop (Ewert et al., 2005). As input
data for rainfall, we use 50 generated rainfall series covering the 2001–2050 period. These
were generated stochastically using a downscaling technique that relates changes in
atmospheric predictors from a GCM (in this case the CSIRO Mk3 GCM) to multi-site daily
rainfall (Charles et al., 1999). The stochastic nature of the generated series accounts for the
variability in the timing of daily rainfall sequences resulting from natural climate variability
and long-term climate change.
414
Session C2: Modelling adaptation strategies to cope with climate change
Results and discussion
Simulated scenarios will be discussed related to the impact of rainfall variability, drought
sequences and atmospheric CO2 increase on individual and regional farm viability (farm
survival and size) and environmental sustainability (i.e. salinity extension). The scenarios will
provide means to closely analyse the resilience of a region to potential impacts of climate
uncertainty (based on ensemble data) and prolonged period of extreme dry periods on an
agricultural region. The model does not aim at predicting but rather delineating the range of
possible outcomes resulting from coupling climate change rainfall scenarios with likely CO2
and technology trends.
We aim to expend the potentialities of this model into an interactive modelling tool of
climate change consequences at a much deeper level to inform future agricultural policies. An
iterative and participatory process with policymakers and relevant stakeholders will allow for
discussion on available sustainable land use options or conditions of social and economic
resilience of farming communities. This approach is seen as a first step towards participative
citizenship mediated by computer-assisted tools in order to build mutual understanding, trust
and respect amongst stakeholders and to secure effective decisions at different levels of
implementation. In the context of the Western Australian wheat-belt, this approach might be
the only way to anticipate the potential success or failure of policy interventions in relation to
climate change impact and adaptation and therefore contribute to safeguarding the social,
ecological and economic fabric of the region.
References
Asseng, S., et al., 2008. Human-landscape interactions in a salinity-prone agricultural region
of the Western Australian Wheat-Belt. Submitted to Ecological Modelling.
Bousquet, F., et al., 1998. Lecture Notes in Artificial Intelligence 1416: 826-837.
Charles, S.P., et al., 1999. Journal of Geophysical Research – Atmospheres 104(D24): 3165731669.
Ewert, F., et al., 2005. Agriculture, Ecosystems and Environment 107: 101-116.
IPCC Report, 2007. In: Eds B. Metz et al., Climate Change, Cambridge Press, UK and New
York, USA.
Raupach, M.R., et al., 2007. PNAS 104(24): 10288-10293.
415
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
GHG mitigation through bio-energy production versus avoided
deforestation and afforestation: A quantitative analysis
P. Havlík, U.A. Schneider, E. Schmid, H. Böttcher, G. Kindermann,
S. Leduc, M. Obersteiner
IIASA, Forestry Program, Schlossplatz 1, 2361 Laxenburg, Austria
Contact: havlik.petr@gmail.com
Introduction
By first principles, land is a fixed production factor where a change in the extent of one land
use type will impact on other land uses. Concerns about the environmental effects of direct
and indirect land use changes due to bio-energy production have been recently expressed, e.g.,
by Petersen (2008). These effects do not concern only biodiversity conservation, water
quality, and other natural resources, but might negatively influence also the total greenhouse
gas (GHG) balance of biofuels deployment (Searchinger et al., 2008). Already Righelato &
Spracklen (2007) argued that where competition between biofuels and forests arises,
afforestation and forest restoration (A/R) may be more efficient mitigation strategy than
biofuel production. Marland et al. (2007) commented then that this result is not general and
will depend among others on the specific site. In this paper, we present recent developments
and refined results of GLOBIOM – a global partial equilibrium model integrating the
agricultural, bio-energy and forestry sectors (Havlík et al., 2008). The main aim is to study
the land use implications and resulting GHG effects of the deployment of different bio-energy
production technologies. Next we analyse the relative cost efficiency of bio-energy
production and A/R under various economic and policy scenarios.
Methods
The applied model contains detailed description of both the agricultural and forestry sectors.
Within the agricultural sector all the major crops and the pertinent management alternatives in
terms of fertilization and irrigation are represented. The livestock sector covers explicitly the
main animal categories and the pertinent production systems. Also within the forestry sector,
choice between several management strategies differentiated mainly by the rotation period is
available for each site. Both crop yields and mean annual forest harvests are estimated for the
different management strategies by means of biophysical models, like EPIC, which are
applied geographically explicit on sub-national level (homogeneous response units, HRU).
Finally, three bio-energy production pathways are taken into account in the model: (1)
biofuels based on conventional feedstocks (sugar cane, maize, soybeans and rapeseed),
sometimes referred to as biofuels of the first generation (ethanol and biodiesel); (2) biofuels
based on woody feedstock, representing the second generation technologies assuming
gasification technologies; and (3) heat and power generated by direct combustion of woody
biomass Optimization occurs in the model through maximization of the market surplus under
technological and resource constraints. Prices and international trade flows are endogenously
computed for 27 world regions.
Results and discussion
Current results show that there is strong competition between traditional forests and biofuel
production. In 2030 scenarios of no avoided deforestation policies and 10 percents fossil fuel
substitution by ethanol, some additional 100 million hectares of forest would disappear due to
agricultural land expansion. A similar area of traditional forests would be converted to short
rotation forest plantations if second generation technologies were deployed. Yet, second
generation appears to be preferable with respect to the total greenhouse gas balance, and food
security in least developed countries, Figure 1. Under avoided deforestation scenarios we find
416
Session C2: Modelling adaptation strategies to cope with climate change
that a carbon tax of 15 USD per ton of CO2 would be necessary to prevent deforestation
induced by biofuel expansion. Work is ongoing on the analysis of direct heat and power
production, and on chains using biomass from forest residues rather than forest plantations.
Figure 1. Prevalence of malnutrition in Sub-Saharan Africa [%], in relation to the global
biofuel target by 2030 depending on the transformation pathway.
References
Havlík, P., et al., 2008. Avoided deforestation and biofuel production: A global analysis. Yale
Student Chapter of ISTF 2008 Conference, March 27-29, 2008, New Haven, USA.
Marland, G., et al., 2007. Science 318: 1066-1068.
Petersen, J.E., 2008. European Review of Agricultural Economics 35: 385-408.
Righelato, R. & D.V. Spracklen, 2007. Science 317: 902.
Searchinger, T., et al., 2008. Science 319: 1238-1240.
417
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Impact assessment of climate change on agriculture:
Econometric-process analysis under partial equilibrium
Iddo Kan, Ayal Kimhi, Kyrill Shraberman
Department of Agricultural Economics and Management,
The Hebrew University of Jerusalem, P.O.B. 12, Rehovot 76100, Israel
Contact: kan@agri.huji.ac.il
Introduction and methods
The objective of integrated assessment studies of climate-change impacts is to simulate
behaviors of natural-economic systems outside the range of observed behaviors. Due to the
large set of factors influencing agricultural production, the challenge of predicting farmers'
adaptation to climate changes attracted many researchers and stimulated the development of
various methodological approaches. A class of studies relies on mathematical programming
models, in which production technologies are represented explicitly (e.g., Howitt et al., 2003;
Kan et al., 2007). This approach was criticized since technological parameters are generally
derived from experimental-based data rather than from representative samples of the farming
population. An alternative methodology, known as the Ricardian approach (Mendelsohn et
al., 1994), applies econometric techniques to real-world farming data in order to estimate
reduced-form equations, which explain spatial-temporal variations in agricultural returns by
differences in economic and environmental variables. A major drawback of this technique is
attributed to the inability to link the outcomes to general- or partial-equilibrium models,
because production is not explicitly represented by the estimated return functions.
Consequently, prices must be considered exogenous throughout simulations of climatechange scenarios (Cline, 1996). Antle & Capalbo (2001) developed the econometric-process
modelling approach, in which econometrically estimated agricultural production models are
integrated with biophysical and economic equilibria models. We adopt this approach for
assessing climate-change impacts on agriculture under partial equilibrium in the vegetative
agricultural outputs markets. The following procedure is applied: first, village-level data are
used for estimating equations explaining crop acreages by climate factors and output prices;
then, a production model translates these land-use equations into production units and
aggregates across villages to obtain regional or nationwide agricultural-products supply
functions; finally, the resultant supply functions are incorporated into a partial-equilibrium
model to be used for applying the impact assessment with respect to climate changes. The
study compares between climate-change impacts assessed under the constant-prices
assumption versus the case of endogenous partial-equilibrium prices.
The study contributes to the integrated assessment methodology by developing an
innovative estimation procedure which takes into account the presence of corner solutions.
While in aggregated regional-scale data all crops are likely to be grown in all regions, farm or
village level data may exhibit a wide range of cropping portfolios. This phenomena matters
since the impact of explanatory variables on the land allocation among crops may depend on
the chosen set of crops. We develop an estimation procedure based on a sample selection
model where selection is specified as a multinomial logit model. Specifically, we modify the
method developed by Bourguignon et al. (2007).
Application
The analysis is applied to the case of Israel, utilizing both the sharp spatial climate gradient
throughout the country and the fact that the Israeli farming sector is technologically advanced;
hence, representing modern adaptation to various climate conditions. Furthermore, the
isolation of the Israeli vegetative agricultural markets, mainly due to high import taxes,
supports the assessment under partial equilibria. A panel data of 746 villages (about 85% of
418
Session C2: Modelling adaptation strategies to cope with climate change
all in Israel) over the years 1992–2002 is used for estimating acreage functions of four groups
of crops, which are treated as composite goods: deciduous, other plantations, field-crops and
vegetables. Acreage shares are translated into production quantities using state-wide
production data, and composite-good demand functions are derived using demand elasticities
estimated by Hadas (2001) for the main Israeli vegetative agricultural products. Climate
change forecasts for 2041–2050 are from Alpert et al. (2008) for the A1B IPCC-2001
scenario. These forecasts predict a 10% reduction in annual precipitation, increases in winter
temperatures and in the frequency of extreme temperatures during the summer.
Results and discussion
The following table compares between the impact of climate change on the producer surplus
of the Israeli vegetative agricultural sector, assessed by the aforementioned EconometricProcess model for the cases of partial equilibrium (endogenous prices) and constant-prices.
Also presented is an evaluation obtained from a Ricardian analysis based on the same data
set. Values are in million dollars per year in terms of 2002 dollars.
Base Period (1992-2001)
260
Future Period (2041-2050)
Ricardian Model
Econometric-Process Model
Partial Equilibrium
Constant Prices
220
107
425
Considering prices endogenously in the Econometric-Process model lessens the forecasted
impacts on farmers’ surpluses from a reduction of 59% to only 15%. The Ricardian model, on
the other hand, predicts an increase in farming incomes of 64%. These differences reveal the
sensitivity of the evaluated impacts on the agricultural sector to the assessment methodology,
which in turn casts on the recommendations for governmental intervention.
We claim that the estimates derived by the partial equilibrium econometric-process
analysis are preferential due to the relative comprehensiveness of the model. Computable
consumer surpluses are increased by $0.94 millions per year; hence, the overall effect on
consumers’ and producers’ surpluses amounts to a reduction of $41 millions per year. The
model indicates that the reduction in precipitation constitutes the most significant climatechange effect, to which farmers react by reducing the land share of field crops – the mostly
rain-fed based crops. If the government will cut water quotas for irrigation by 20% in order to
secure water for the urban sector, farmers would react by replacing some vegetable lands with
plantations, resulting in price changes that ultimately increase farmers’ and consumers’
surpluses by $7 millions relative to the un-intervention scenario. On the other hand, if the
government would keep irrigation quotas by desalination, while increasing water prices
correspondingly such that the overall production costs will rise by 10%, farmers will increase
field-crops areas, leading to a reduction of $154 millions in farmers' plus consumers' surplus.
References
Alpert, P., et al., 2008. Global and Planetary Change 63: 163-170.
Antle, J.M. & S.M. Capalbo, 2001. Amer. J. Agr. Econ. 83(2): 389-401.
Bourguignon, F., et al., 2007. Journal of Economic Surveys 21: 174-205.
Cline, W.R., 1996. American Economic Review 86: 1309-1311.
Hadas, E., 2001. Economic Aspects of Agricultural Trade Among Israel, Gaza and the WestBank. PhD thesis, The Hebrew University of Jerusalem. (in Hebrew)
Howitt, R., et al., 2003. Impacts of Global Climate Change on California’s Agricultural Water
Demand. Department of Agricultural and Resource Economics, UC, Davis, California.
Kan, I., et al., 2007. Assessing Climate Change Impacts on Water, Land-Use and Economic
Return in Agriculture. Available at SSRN: http://ssrn.com/abstract=1020562
Mendelsohn, R., et al., 1994. American Economic Review 84: 753-771.
419
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Counter measures of agricultural sector for post-Kyoto Protocol
Chang-Gil Kim
Korea Rural Economic Institute
4-102 Hoegidong Dongdaemoonku, Seoul 130-710, Korea
Contact: changgil@krei.re.kr
Introduction
The comprehensive consideration of the internal and external changes in conditions pertaining
to the post-Kyoto Protocol scheme indicates that Korea is highly likely to be mandated to
reduce greenhouse gas emission from the second commitment period (2013–2017). This
paper is to analyse the impact of the post-Kyoto Protocol implementation on the agricultural
sector and to propose systematic countermeasures in preparation for the implementation of the
Protocol. The analytical results show that the participation of the agricultural sector in the
emission trading program is found to increase earnings and boost national economy. With
regard to future implementation of the Kyoto Protocol, the agricultural sector can take it as a
good opportunity because its greenhouse gas emission tends to decrease and it may contribute
to emission reduction by providing greenhouse gas sinks. On the other hand, when mandatory
greenhouse gas reduction is imposed resulting in the enforcement of relevant policies such as
emission trading scheme, carbon tax and regulations on energy use, it would have significant
impacts on the cost of agricultural products, the distribution of agricultural resources and the
farming household income. Therefore, it is necessary to develop appropriate measures to
comply with the post-Kyoto Protocol.
Methods
In order to measure the impacts of 3 emission reduction scenarios on the agricultural sector
when the reduction commitment level at 2013 is set to 5% below 2000 emission levels,
dynamic CGE (computable general equilibrium) model, a general equilibrium approach was
used. Three types of emission reduction scenarios to comply with the UNFCCC were
analysed: Scenario 1 that implemented emission reduction objective individually; Scenario 2
in which all industrial sectors participated in emission trading system (ETS); and Scenario 3
in which only non-agricultural sectors participated in ETS. In addition, greenhouse gas
reduction potential (reduction capacity) is calculated by subtracting annual emissions when
the reduction measures are not taken from those when the measures are taken. When there are
emission reduction measures in force and more powerful measures can be put into force
additionally, more emission is reduced and reduction potential increases more.
Results and discussion
Comparing economic effects of emission reduction by each scenario, the economic effects of
emission trading system (Scenario 2) over individual implementation (Scenario 1) appeared to
be USD 74.3 million in the arable sector, USD 550 million in the livestock sector and USD
7.7 million in the agriculture-related industry, totaling USD 632 million in the entire
agricultural sector. Also, economic effects in the non-agricultural industries amounted to USD
41,724 million in the manufacturing and service industry, USD 2,802 million in the fossil fuel
industry, totaling USD 45,144 million. So, it was estimated that if all industrial sectors
including the agricultural sector participated in ETS, an additional economic effect of USD
4,746 million would be generated
Emission reduction capacity of greenhouse gas reduction measures in the arable and the
livestock sectors in 2020 appeared to be 3,675 × 103 t CO2, about 25.3% of total emissions in
the agricultural sector. It was expected that emissions after greenhouse gas reduction would
be 10,828 × 103 t CO2 in 2020, which would be about 31% less than 2000 level (15,693 × 103 t
420
Session C2: Modelling adaptation strategies to cope with climate change
CO2). When 5% below baseline (2000) is decided as allowable emission level in 2013 under
the Kyoto Protocol, surplus emissions would exceed by 44 × 103 t CO2the emission cap when
greenhouse gas reduction measures are not in force in 2013, and thus reduction objective
would be imposed. However, it is estimated that if greenhouse gas reduction measures
continue to be in force, surplus emissions of about 875 × 103 t CO2would be generated in
2020. Though surplus emission before reduction in 2020 was 406 × 103 t CO2, it increased 10
times to 4,081 × 103 t CO2 when reduction measures were taken. Therefore, if emission
trading system is put in force among domestic industries, it would be serve as an important
opportunity for the agricultural sector.
Figure 1. Estimation of reduction potential in agricultural sector. Unit: 103 t CO2
As for basic policies for establishing measures to comply with the Kyoto Protocol in the
agricultural sector, it is suggested in this paper that the government should take this
opportunity for establishing a sustainable agricultural system, combining and consolidating
agricultural policies and greenhouse gas policies, positively and actively coping with the
domestic and international negotiations, scientifically analysing the amounts of greenhouse
gas emission and capture, and adapting to global warming. The implementation strategy for
the agricultural sector will be approached in three stages with 2030 as the target year:
foundation establishment stage (2008–2012), take-off stage (2013–2018) and settlement stage
(2019–2030). A stage-by-stage road map would be presented to establish an environmentfriendly low-carbon agricultural production system through programme implementation in
each sector of greenhouse gas reduction, capture and adaptation.
References
Beach, R.H., et al., 2006. Mitigation Potential and Cost for Global Agricultural Greenhouse Gas
Emissions. Contributed Paper at the IAAE Conference.
Kim, Chang-Gil, et al., 2006. Impacts of the UNFCCC on Agricultural Sector. Research
Report R520. Korea Rural Economic Institute.
Kim, Chang-Gil, et al., 2007. Korean Journal of Agricultural Economics 48(2): 1-30.
Schneider, U. & B.A. McCarl, 2002. The Potential of U.S. Agriculture and Forestry to
Mitigate GHG Emissions: An Agricultural Sector Analysis. CARD Working Paper 02-WP
300, Iowa State University, USA.
421
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Analysis of land use combination to minimize global warming potential
from farmland and farmland surplus Nitrogen and to maximize income
using eco-balance analysis
S.D. Kimura, Y. Toma, Z. Mu, R. Hatano
Graduate School of Agriculture, Tokyo University of Agriculture and
Technology 3-5-8 Saiwaicho Fuchu, Tokyo, 183-8509 Japan
Contact: skimura@cc.tuat.ac.jp
Introduction
To meet the needs of growing population in the 21st century, we need to create a sustainable
nutrient management system that finds a balance between food/feed production and
environmental loads. To provide a criteria for choosing the most environmental friendly
management methods to farmers and policy makers, the relation of productivity and
environmental load must be analysed comprehensive and quantitatively. A challenge is to
reflect the regional specific condition of soil, climate and land use methods to the criteria. In
this study, monitoring results of green house gas emissions (GHG) from different land uses
conducted in a watershed scale were parameterized to calculate the GHG emission at a
watershed scale. Management methods and productivity were obtained by farmer’s inquiry
and the result was analysed using a method defined as ‘eco-balance’ analysis (Kimura et al.,
2007). The objective of this study was to quantify GHG emission from different agricultural
land uses and to quantify the total global warming potential (GWP) and nitrogen (N) flow at a
watershed scale. The quantified results were used to analyse the land use combination which
maximize the productivity and minimize the GWP and farmland surplus N.
Methods
The study was conducted at the Ikushunbetsu River watershed in central Hokkaido.
Greenhouse gas fluxes were measured at 21 fields (9 land use types, totally 47 year × place)
using closed chamber method from 2002 to 2007. Carbon dioxide (CO2) flux from soil
organic carbon (C) decomposition was measured on bare soil, nitrous oxide (N2O) and
methane (CH4) fluxes were measured on planted soils. Annual cumulative flux of each gas
was parameterized using soil, climate and management parameters. The GHGs were
converted into GWP (IPCC, 2001). Carbon (C) sequestration (kg C ha–1 yr–1) was calculated
as below:
C sequestration = net primary production + C in manure – soil organic C decomposition –
C in harvested products
Nitrous oxide flux (kg N ha–1 yr–1) of an upland field was estimated using emission factors
(EF: proportion of N2O emission to applied nitrogen) on the basis of the following assumption
(Toma et al., 2008):
N2O flux = EF chemical fertilizer × applied chemical fertilizer + EF
organic matter + background N2O emission
organic matter
× applied
The CH4, CO2 and N2O flux for a paddy rice field showed a high correlation with the applied
amount of straw and were expressed as a function of amount of straw residues (Naser et al.,
2007). Farmland surplus N (kg N ha–1 yr–1) was calculated based on the following equation:
Farmland surplus N = N input to farmland – N exported from farmland in products –
denitrification – NH3 volatilization from chemical fertilizer and manure – N2O
emission
422
Session C2: Modelling adaptation strategies to cope with climate change
Farmer’s inquiry and ground survey analysis were conducted in 2002, 2005 and 2007 to
obtain the management practices and land use distribution in the study area. Amount of yield
was multiplied by the ratio of main product to by product to calculate the net primary
production. All inputs were converted into C and N units. Values not obtained from our
monitoring were obtained from literature (see for more detail Kimura et al., 2007).
Productivity was evaluated by the price of sold product per area for each year.
paddy rice
soybean
onion
grass
wheat
fallow
vegetable
200
150
100
grass
wheat
50
fallow
0
500,000
1,000,000 1,500,000
soybean
paddy rice
-50
0
vegetable
onion
(kg N ha-1 yr-1)
16
14
12
10
8
6
4
2
0
-2
-4
-6
Farmland surplus N
(Mg CO2 eq ha-1 )
Global warming potential
Results and discussion
The average annual income per ha ranged from 250,000 to 1,060,000 Yen for 2002, 330,000
to 1,330,000 Yen for 2005 and 360,000 to 1,170,000 for 2007. Due to different amount of
yield and cost per kg yield, the land use with the highest income was different among the
years. On the other hand, land uses with low or high GWP and farmland surplus N were
always the same under the present management method. The income was compared to the
GWP and farmland surplus N for each land use and the present situation in 2002 (Figure 1).
Changing the land use proportion in the watershed, the average value can move only inside
line connecting the situation where one land use is 100%. The area marked with the slashed
lines can reduce the GWP and farmland surplus N lower than the present situation while
maintaining the income by changing the land use combinations. The land use combinations
were analysed by changing the land uses combinations from 0 to 100% in 10% interval. The
average GWP and surplus N reduced to 39–87% and 75–82%, respectively, for the 3 years.
The average income for all the possible combination increased by 20%, 24% and 14% for
2002, 2005 and 2007, respectively. All major land uses can be conducted in the present
management methods. In case a land use with high environmental impact is desired, land uses
with low environmental impact must be combined.
0
500,000 1,000,000 1,500,000
Income (Yen ha-1 )
Figure 1. Relation of global warming potential (GWP) and farmland surplus N in relation
to income. The gray big circle represents the watershed average in 2002, the black small
circles relation if only one land use occupies 100% of the watershed. The dotted lines
indicate the value of income, GWP and farmland surplus in 2002, the area with the slashed
lines the possible area where GWP and farmland surplus N can be reduced from the
present situation while maintaining the income by changing the land use combination.
This study showed how monitoring data can be used to screen land use combinations with
less environmental impact. The target can be changed according to the environmental and
economic condition of the area. Using this analysis, incentives to choose land use
combinations with lower GWP and farmland surplus N can be created.
References
IPCC, 2001. The scientific basis. Cambridge University Press, Cambridge, UK.
Kimura, S.D., et al., 2007. Soil Sci. Plant Nutr. 53: 373-386.
Naser, H.M., et al., 2007. Soil Sci. Plant Nutr. 53: 95-101.
Toma, Y., et al., 2008. Soil Sci. Plant Nutr. 53: 692-703.
423
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Modelling the costs of climate change on grazing systems
M. Lieffering, P. Newton
AgResearch Grasslands, Tennant Drive, Palmerston North, New Zealand
Contact: mark.lieffering@agresearch.co.nz
Introduction
For countries such as New Zealand that derive a large proportion of their GDP from
grasslands, an authoritative assessment of the potential responses to the main components of
anthropogenically induced global change (increasing levels of atmospheric CO2 and changes
in climatic factors such as temperatures and rainfall) is an essential first step in determining
where adaptations will be necessary, how these might be achieved and their costs. Long-term
effects such as changes in botanical composition and changes in the availability of plant
nutrients are of central importance in determining the influence of global change on grassland
ecosystems. Any satisfactory simulation of the future must take account of these higher-order
effects and feedbacks and to this end we have developed an ecosystem model that simulates
annual and seasonal pasture growth responses satisfactorily (EcoMod; described in Johnson et
al., 2008). EcoMod mechanistically simulates plant, soil and animal processes in grazed
grasslands, enabling us to examine future climate and management changes that are beyond
our current experience. An assessment of responses of grazing systems to climate change
must consider the effects on not only direct ecosystem services such as pasture production,
but also other inputs/outputs such as nutrient leaching, soil C sequestration and greenhouse
gas (GHG) emissions. In this paper, we illustrate the use of EcoMod to determine the
potential effects of climate change on pastoral ecosystem inputs/outputs in an area of New
Zealand that is projected to be 0.7 °C warmer and 15% drier by 2050.
Methods
We modelled a single paddock of ryegrass and white clover on a hypothetical N.Z. dairy
farm. We ran simulations from 1966 to 2006 using actual climate data and from 2006 to 2050
using climate change projections superimposed on cycled historical climate data. The
projections used were annually incremented increases in minimum (+0.8 °C) and maximum
(+0.6 °C) temperatures by 2050 to 6.7 and 17.0 °C, respectively, while rainfall decreased 15%
from 714 mm to 606 mm per year. Atmospheric CO2 increased in annual increments to 450
ppmv by 2050. Cows (40 cows ha–1) began grazing the paddock when pasture dry matter
(DM) was 2800 kg DM ha–1 and were taken off when the residual reached 1600 kg DM ha–1.
Cows weighed a minimum of 400 kg, had a lactation length of 270 days and a potential peak
daily milk production of 35 litres d–1. All energy for milk production was from the pasture
grown (i.e. no supplementary feed was supplied) and all pasture management was done using
grazing. Soils were parameterized based on the soil type typical of this region and N fertilizer
was added whenever soil test values fell below a threshold. All other nutrients were assumed
to be non-limiting and it was assumed that dung and urine were homogeneously returned to
the soil. In order to smooth out interannual variation, results are presented as 20 year means
centered around and referred to as ‘1990’ (present) and ‘2040’ (climate change). To explore
the effects and potential costs and benefits of a simple adaptation strategy, we ran simulations
with and without irrigation. Irrigation occurred when soil moisture levels caused more than a
20% decrease in plant growth; this resulted in 6 irrigation events of about 50 mm per year
under present conditions and 7 irrigations under climate change.
Results and discussion
Under non-irrigated conditions, climate change resulted in decreased pasture intake and milk
production (Table 1). The majority of growth occurred in spring though the largest decrease
424
Session C2: Modelling adaptation strategies to cope with climate change
in growth rate was in autumn due to increased soil moisture stress. As a result of lower
pasture growth, simulated fertilizer inputs decreased, though because clover content increased
(data not shown), N fixed increased leading to an overall increase in N inputs and greater NO3
leaching. Due to lower pasture intake and animal production, enteric CH4 emissions decreased
while soil C changed little.
Table 1. The modelled effects of climate change on some key pasture inputs and outputs of a
hypothetical grazed dairy farm in Canterbury, N.Z. All data are on an annual 20 year average
basis.
Non-irrigated
Irrigated
1990
2040 %change
1990
2040 %change
Pasture intake (t ha–1)
Growth rates (kg ha–1 d–1)
Spring
Summer
Autumn
Winter
7.3
6.6
-10
17.4
18.8
+8
37.3
20.6
9.2
12.2
33.7
19.9
5.2
13.2
-10
-4
-43
+8
67.3
80.2
28.9
14.6
75.3
86.7
29.0
16.4
+12
+8
0
+12
Milk production (l ha–1)
8073
7229
-11
18757
20543
+10
Fertilizer N (kg N ha–1)
N fixed (kg N ha–1)
Total N inputs (kg N ha–1)
21.4
116.0
137.4
12.1
138.9
151.1
-43
+20
+10
130.7
68.8
199.5
147.9
78.7
226.5
+13
+14
+14
70.3
79.3
+13
46.1
25.9
-44
3.0
2.7
-10
7.3
7.9
+8
181.5
177.2
-2
169.2
153.4
-9
NO3 leached (kg ha–1)
CH4 emitted (t CO2e ha–1)
soil C (t ha–1)
An obvious mitigation strategy to lower rainfall with climate change is irrigation. Under
irrigated conditions, the warmer temperatures in 2040 resulted in greater pasture intake and
milk production. However, to sustain this extra growth total N inputs were greater but NO3
leaching decreased while enteric CH4 increased because of the higher animal production.
As shown here, simulation modelling enables us to explore future changes that are beyond
our current experience and look at the potential trade-offs of implementing adaptation
strategies. For example, with irrigation, CH4 emissions increased under climate change while
NO3 leaching decreased. Because all of the inputs/outputs can be assigned at least nominal
monetary values (e.g., installing an irrigation system, the value of the pasture, GHG and NO3
emission charges), by integrating these biophysical results with farm business models and
upscaling to regional levels, the costs of climate change nationally can be explored.
Reference
Johnson, I., et al., 2008. Australian Journal of Experimental Agriculture 48(5): 621-631.
425
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Testing adaptation options in a +2 °C warmer climate
M. Moriondo, R. Ferrise, M. Bindi
Department of Agronomy and Land Management,
P.le Cascine 18, CAP 50144 Firenze, Italy
Contact: marco.moriondo@unifi.it
Introduction
The results of recent global climate monitoring as well as the simulations of general and
regional circulation models (GCM and RCM, respectively) stressed out that the future climate
will be significantly different than that experienced in the past, resulting in significant impact
on different economic sectors (e.g. agriculture, forestry, energy consumptions, tourism, etc.)
(Hanson et al., 2006).
A particular attention should be devoted to the study of climate change impact in
agriculture. Since this economic sector, representing the major land use across the globe, is a
major economic, social, and cultural activity, and it provides a wide range of ecosystem
services. Importantly, agriculture in its many different forms and locations remains highly
sensitive to climate variations, since crop yield is largely determined by the weather
conditions during plant life cycle and even with minor deviations from the normal weather,
management practices and yield are seriously threaded.
As a consequence, understanding the potential impacts of climate change on the agriculture
has become increasingly important and is of a main concern especially for the sustainability
of agricultural system and for adopting reliable adaptation options to cope with.
Building on these premises, this work aims at assessing the effect of different adaptation
strategies in crop managing in a +2 °C warmer climate.
Methods
The general framework of this paper provides the use of climate future data, as simulated by a
GCM, as input of CropSyst model in a period corresponding to a +2 ºC global warming
(2026–2060; New, 2005) respect to pre-industrial level. Since the results of a typical GCM
simulations may be inappropriate, especially for impact assessment at local scale in
agriculture, a simple GCM statistical downscaling software (LARS WG) was used to
reproduce both present and future climate (Tmin, Tmax, rainfall and global radiation) at a
spatial resolution suitable for impact assessement on a regional scale (50 × 50 km).
According to LARS WG procedure (Semenov & Barrow, 1997), available observed daily
weather data for a given site were used to determine a set of parameters for probability
distributions of weather variables as well as correlations between them (calibration stage).
This set of parameters is then used to generate both the synthetic weather time series
describing the present period and as a baseline to be perturbed using forcing factors derived
from the GCM.
In this work observed daily data (including Tmin, Tmax, rainfall and radiation) for the
period 1975–2005 and spatially interpolated at a resolution 50 × 50 km over EU (provided by
MARS project www.) where used in the calibration phase of the stochastic weather generator.
After calibration, 100 years of synthetic daily weather data were produced for each grid point
to represent the current baseline 1975–2005.
The results of HadCM3 for A2 scenario in 2030–2060, over the European domain, were
used to derive the forcing factors for the downscaling procedure. These factors were
computed for each GCM grid point as monthly average differences of Tmin, Tmax, rainfall
and radiation respect to the reference period (1975–2005). For temperature and rainfall the
relative change in standard deviation and in duration of wet and dry spell were also
calculated.
426
Session C2: Modelling adaptation strategies to cope with climate change
CropSyst model was run for +2 °C scenario to simulate growth and development of barley,
wheat, sunflower, soybean and maize in business as usual condition (BAU) (using common
agricultural practices including sowing dates, fertilization, rainfed conditions) and using
different adaptation strategies. These included early and late sowing date (respectively –15
and +15 dd with respect to BAU), short and long cycle variety (–20% and +20% with respect
to BAU), irrigation, fertilization (+20% with respect to BAU).
According to A2 scenario, CO2 air concentration was set to 550 ppm for the considered
period and consequently crop biomass accumulation in CropSyst was set to increase by 18%
with respect to present conditions (350 ppm).
Results and discussion
The general picture simulated by HadCM3 for a +2 °C warmer climate showed a clear pattern
in the change of rainfall distribution with a general annual rainfall increase in Northern
Europe above 55 °N, no changes in central Europe roughly between 45° and 55 °N and a
sensible decrease over the Mediterranean basin. Annual maximum and minimum temperature
increased following a strong longitudinal gradient from –10° to 40 °E. Both increased
temperature and the changed rainfall pattern had significant impacts on selected adaptation
strategies efficiency. As expected the use of irrigation increased crop yield respect to BAU
especially at lower latitudes where the decrease in rainfall is more evident. In contrast higher
fertilization did not increase crop growth probably because the BAU fertilization is close to
the optimum threshold indicated in CropSyst. The use of longer cycle varieties resulted in a
increased yield due to the lengthening of time for biomass accumulation. This strategy
resulted more effective in northern region where no changes or even increase in rainfall rate
were simulated. However, the effectiveness of this strategy was reduced at lower latitudes due
to the projected decrease in rainfall especially in summer. On the other hand, shorter cycle
varieties resulted in a decrease of yield especially in areas with better growing conditions (i.e.
Northern Europe). Changing sowing dates gave not significant results but in southern regions
where an earlier sowing allowed the crops to escape the spring-summer drought.
shorter
cycle
longer
cycle
fert
irrig
earlier
sowing
later
sowing
50
40
Delta change (%)
30
20
10
0
-10
-20
Germany
Greece
-30
-40
-50
Figure
1.
Relative
change (%) in crop yield
considering
different
adaptation options at
higher (Germany) and
lower (Greece) latitudes.
References
Hanson, C.E., et al., 2007. Clim. Change 81: 163-177.
New, M., 2005. In: Ed. L.D. Rosentrater, Evidence and implications of dangerous climate
change in the Arctic. WWF International Arctic Programme.
Semenov, M.A. & E.M. Barrow, 1997. Climatic Change 35: 397-414.
427
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Simulation of climate scenarios and the bio-economic assessment of
different agricultural management systems in the Marchfeld region
F. Strauss, E. Schmid, E. Moltchanova
University of Natural Resources and Applied Life Sciences, Institute of Sustainable Economic
Development, Feistmantelstraße 4, 1180 Vienna, Austria
Contact: franziska.strauss@boku.ac.at
Introduction
The aim of this analysis is to assess the impacts of climate change on crop yields, soil organic
carbon stocks, and Nitrogen leaching, as well as on the profitability of different crop
production systems in the Marchfeld region. Rischbeck (2007) used downscaling methods
from global climate models to predict climate change for the Marchfeld region. In this study
we develop regional climate scenarios for the next 30 years using a linear regression model
based on daily weather observations from 1975 to 2006 at the weather station in Grosz
Enzersdorf. The climate scenarios are integrated together with other site specific data such as
soil types, two crop rotations (corn-winter wheat-sunflower-winter wheat-spring barley, and
sugar beet-winter wheat-field peas-winter wheat-spring barley) and different crop
managements systems (conventional, reduced or minimum tillage with or without irrigation,
as well as with or without straw removal) in the biophysical process model EPIC
(Environmental Policy Integrated Climate; Williams, 1995; Izaurralde et al., 2006). The most
important modules in EPIC are weather simulation, hydrology, erosion and sedimentation,
Nitrogen-, Phosphor-, Potassium- and Carbon-cycles, plant growth, soil qualities and tillage
operations. Hence, panel data of crop yields, soil organic carbon stocks and Nitrogen leaching
were simulated with EPIC depending on the site data including 30 climate scenarios, and the
twelve crop production systems. Moreover, the simulated crop yields are used to assess the
economic profitability of the different production systems in the region.
Data and method
The data source for the climate scenarios are daily weather observations from 1975 to 2006 in
Grosz Enzersdorf. Trends for temperature, solar radiation, relative humidity, wind and
precipitation were calculated by means of linear regression, in which the residuals were
reallocated randomly by retention of the monthly sequences. This process of reallocation was
repeated 30 times to capture the variability and uncertainty in the climate scenarios.
EPIC simulations over 64 years (1975 to 2038) were performed to analyse the impact of
climate change and of different crop production systems on crop yields, soil fertility and
Nitrogen leaching. Furthermore, the variable production costs were calculated for each crop
and production system (BMLFUW, 2008).
In the profitability analysis we compare stochastic producer prices, based on historical
price time series, and average variable production costs. The historical price time series
provide a basis for the generation of stochastic prices for the two periods from 1975 to 2006
and from 2007 to 2038. We presume that prices are subjects to normal distribution from
which prices are taken randomly (no trends are accounted in this analysis). Frequency
distributions show, how often the average variable costs are above or below the stochastic
producer prices. The redistribution of frequency gives information about likely impacts of
climate change on the profitability of crop production in the Marchfeld region.
Results and discussion
The simulation results of both crop rotations show that soil organic carbon stock reaches its
maximum under minimum tillage and without straw removal. On average, the soil organic
carbon stock is 109.93 t/ha in this region. Nitrogen leaching is lowest with minimum tillage
428
Session C2: Modelling adaptation strategies to cope with climate change
and with straw removal and is on average 22.14 kg ha–1. Under c.p. both ecological indicators
are decreasing over time on average. The decrease of soil organic carbon stock is related to
the increase of soil temperature and thereby to higher CO2 respiration. Moreover, an increase
in annual precipitation leads to higher sediment transportation rates. The combination of an
average annual increase of temperature by 1.3 °C with an increase of CO2 concentration to
443 ppm in 2038 and an increase of annual precipitation by 100 mm leads to less Nitrogen
leaching on average in this region. This is mainly due to higher biomass production and
therefore higher Nitrogen uptakes by crops. All crop yields show a positive trend over time on
average. Corn yields increase by about 0.5 t ha–1, and spring barley and sunflower by about
0.02 to 0.07 t ha–1 (dry matter). Using irrigation, the variability of crop yields as measured by
the standard deviations is somewhat reduced in most cases.
The profitability analysis (Table 1) indicates that production of corn is most profitable with
minimum tillage and irrigation. The percentages describe how often average variable costs are
below stochastic producer prices. The production of sugar beets is most profitable in this
region. The production of winter wheat, sunflower, field peas and spring barley is most
profitable with minimum tillage and without irrigation. Not removing straw from the field has
positive crop yield effects for winter wheat and sunflower. The production of field peas does
not seem to be economically viable in this region, however, within a crop rotation system
other benefits account obviously more. The comparison between the periods 1975 to 2006 and
2007 to 2038 shows that the profitability of corn is increasing without irrigation and
decreasing with irrigation, but the production of corn is profitable in both periods. The
profitability of winter wheat, spring barley and field peas is slightly increasing whereas the
profitability of sugar beet and sunflower remains nearly unchanged between the two time
periods. The major conclusion of this analysis is that minimum tillage represents a profitable
and ecological sound production system, especially under the impact of climate change.
Table 1. Profitability in % (SB=sugar beet, C=corn, W=winter wheat, P=field peas,
SF=sunflower, SB=spring barley) for different tillage operations (conventional, reduced or
minimum) and alternatives in management (I=irrigation, nI=no irrigation, S=straw removal,
nS=no straw removal).
SB 1975-2006
SB 2007-2038
C 1975-2006
C 2007-2038
W 1975-2006
W 2007-2038
P 1975-2006
P 2007-2038
SF 1975-2006
SF 2007-2038
SB 1975-2006
SB 2007-2038
I/S
100
100
81
74
54
46
0
2
39
27
2
6
Conventional
nI/S
I/nS
97
100
98
100
55
81
67
74
70
55
74
47
2
0
9
2
55
39
57
27
28
2
33
6
nI/nS
97
98
55
68
70
74
2
9
55
59
28
33
I/S
100
100
82
79
61
57
0
3
34
23
3
9
Reduced
nI/S
I/nS
97
100
98
100
57
82
69
80
74
62
79
62
6
0
12
3
54
36
53
33
41
3
43
9
nI/nS
97
99
58
69
74
80
5
12
55
61
41
43
I/S
100
100
82
81
67
65
2
4
34
25
8
15
Minimum
nI/S
I/nS
97
100
99
100
59
83
70
82
76
69
83
69
9
2
16
4
56
36
56
36
47
6
51
15
nI/nS
97
99
59
70
76
84
9
16
56
64
47
51
References
Bundesministerium für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft, 2008.
Deckungsbeiträge und Daten für die Betriebsplanung, Berger, Horn (BMLFUW).
Izaurralde, R.C., et al., 2006. Ecological Modelling 192(3-4): 362-384.
Rischbeck, P., 2007. Der Einfluss von Klimaänderung, Bodenbearbeitung und Saattermin auf
den Wasserhaushalt und das Ertragspotential von Getreide im Marchfeld. Dissertation at
University of Natural Resources and Applied Life Sciences, Vienna, 1-155.
Williams, J.R., 1995. In: Ed. V.P. Singh, Computer Models of Watershed Hydrology. Water
Resource Publications, Highlands Ranch, Colorado, pp. 909-1000.
429
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Energy use efficiency and greenhouse gases emissions of dairy farms of an
isolated territory: Reunion Island from 2000 to 2007
M. Vigne, J. Vayssières, J.L. Bochu, P. Lecomte
CIRAD, UPR Systèmes d’Elevage, Pole Kappa, 97410 St Pierre, La Reunion
Contact: mathieu.vigne@cirad.fr
Introduction
Criteria to assess agricultural systems have widely evolved during the past few decades.
Technical and economic indicators are associated with environmental criteria such as nutrient
or energy balance. In a context of external inputs dependence, Reunion Island, a French
territory (2,500 km²) situated in the Indian Ocean, is an interesting field to study the links
between local development policy, farm functioning and energy use. The aim of this study is
to assess non-renewable energy (NRE) consumptions and greenhouse gases emissions of
Reunion dairy farms and to characterize farm management impact on environmental
indicators, more particularly on greenhouse gases (GHG) emissions.
Methods
Based on a techno-economic survey on dairy farms carried in 2000 by Taché (2001), the
energy and GHG balances on 31 dairy farms for 2000 and 14 for 2007 have been established
according to PLANETE methodology adapted to local context (Vigne, 2007). Based on Life
Cycle Analysis (Bochu, 2007), PLANETE is a national French tool to quantify energy consumptions and GHG emissions at the farm scale. Direct and indirect energy use is estimated
according to standards all along the cycle and translated in fuel equivalent litre (EQF). GHG
emissions, carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O), are estimated and
summarized in tons of equivalent CO2 (teqCO2) based on their global warming potential.
Evolution of indicators are analysed along time and farm management options are discussed.
Results and discussion
Average results for NRE and GHG emissions are listed in Table 1.
Environnemental indicators of dairy production in Reunion Island have been improved
between 2000 and 2007. The energy efficiency increased by 13% whereas the energy
consumption per 100 liters of milk decreased by 4.2 EQF (16%). Moreover, emissions of
greenhouse gases decreased from 2.32 to 1.73 teqCO2 for 1000 liters of milk produced on the
farm (–25%). We can observe an increase of the proportion of CO2 (+22%) in the total
emission while the contributions of CH4 and N2O decreased.
All these variations are in conformity with management changes and the sustainability
objectives of the dairy sector: enhancing milk productivity, farmers income and reducing
environmental impacts from this activity.
Decrease of total energy consumption and GHG emissions are mainly linked to the
improvement of the overall transformation of the farm imported feed supplements into milk
(R² - 0.80;). Due to a better management, average efficiency of dairy farms to transform this
major input into milk has significantly increased from 1.25 to 1.32 L kg–1.
Increase of CO2 emissions is linked to changes in feeding management. During 2000 and
2007, average farm input of concentrates summarized per cow per day has increased from
11.5 to 14.6 kg. These importations, subsidized by local policies, led to an increase of CO2
emitted for production and transport (+21%; +32%, respectively). On the other hand, increase
of the concentrates in the ration contributed to the reduction of animal CH4 emissions by 17%.
Studies have shown that due to fermentation profile modification, higher proportion of
supplements tend to lower animal CH4 emissions (Giger-Reverdin et al., 2000).
430
Session C2: Modelling adaptation strategies to cope with climate change
Table 1. Energy use and GHG emission in dairy farm.
Indicators
2000 (n = 31)
Average SD
Energy Efficiency
Energy use (EQF/100 l of milk)
GHG emission (teqCO2/103 l milk)
Proportion of each gas (%) :
Emission of CO2
Consuming direct energy
Mineral fertilizers production
Feeds production
Buildings and machinery
Transports to supply inputs to the island
Emission of CH4 by animals
Emission of N2O
Fertilizer production
Soil N fertilization
Manure
2007 (n=14)
Average SD
Variation
(%)
0.39
26.2
0.09
6.8
0.44
22.0
0.05
2.9
+13
–16
2.32
0.57
1.73
0.21
–25
4.5
1.9
0.4
4.3
3.0
3.3
1.7
3.4
0.7
2.1
1.1
+22
+23
–53
+21
+11
+32
–17
–9
–42
–20
–16
44.9
4.7
1.5
18.4
4.4
15.9
27.8
27.3
1.9
12.8
12.6
4.8
1.3
0.7
3.7
1.9
2.9
3.1
3.6
1.1
2.2
1.5
54.7
5.8
0.7
22.3
4.9
21.0
23.2
22.1
1.1
10.2
10.6
Valorization of manure as organic fertilizer in substitution of mineral fertilizer is one of the
main recommendation from local technical support to improve N balance and to enhance the
income of farmers. Results of this study shows a decline of the nitrogen (N) fertilizer input,
from 165 unit N per ha in 2000 to 111 unit N per ha today. This decrease has a direct effect on
emissions of GHG. CO2 and N2O emissions for fertilizer production declined of 53% and
23%, respectively. N2O emissions resulting of N fertilizer applied on the soil (volatilization
and denitrification) declined of 20 %.
Applied to annual accounting databases LCA tools are useful to assess effect of local
support and development policies, and more particularly, the effect of local technical
support’s message on climate change. Indeed, results show local development policy for dairy
sector in Reunion Island have been effectively diffused and had a positive effect on
environmental impact from 2000 to 2007.
In addition, this analysis highlights the need to address issues to decrease impact of local
productions on environment. For example, transport of inputs from metropolitan France to
Reunion Island (10,000 km) is responsible for 25% of total energy use. Despite strong
political and socio-economic relationship, this high energy use underlines the interest to shift
to closer Indian ocean regional markets for inputs, which would reduce impact on the
environment of transport. An other opportunity could be to use manure in biogas process to
take advantage from CH4 valorization and reduce emissions during soil application of
manure.
References
Bochu, J.L., 2007. Synthèse 2006 des bilans PLANETE. Rapport scientifique ADEME, 248
pp.
Giger-Reverdin, S., et al., 2000. Modélisation empirique des facteurs de variations des rejets
de méthane par les ruminants. VIIème Rencontres autour de la Recherche sur les
Ruminants. Paris, France, pp. 187-190.
Taché, C., 2001. Diagnostic des exploitations laitières de l’île de La Réunion. Rapport de
stage ISTOM, 96 pp.
Vigne, M., 2007. Evaluation du bilan énergétiques des exploitations bovines laitières à la
Réunion. Rapport de Stage CIRAD-UM II, 76 pp.
431
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Mujika community climate: Cropping systems modelling exercise
S. Walker, D. Nanja, P. Gondwe, J. Dimes, et al.
University of the Free State, P.O. Box 339,
Bloemfontein 9300, South Africa
Contact: walkers.sci@ufs.ac.za
Introduction
In the semi-arid regions of SADC there is widespread dependence on subsistence farming
which results in a high level of poverty among community members. There is a high
variability in the rainfall received, both in amount and distribution through the growing
season as well as large differences between seasons. These factors have contributed to the
poor coping strategies and few practical adaptations in the cropping systems in these areas.
The main staple crop is maize and it is grown during the summer rainfall season, mainly from
November to March. The Zambian Department of Meteorology has been working with this
community and has been disseminating seasonal and other weather forecasts via radio for the
past few years. The purpose of this exercise was to bring together the community’s
perceptions of their natural resources and the potential for crop production in this area using a
cropping systems model.
Methods
This community exercise was part of a larger capacity building climate change IRDC project
in both Zambia and Zimbabwe (ICRISAT, 2008). Mujika is situated at approximately 16° S
latitude and 27.5° E longitude and is at an altitude of 1020 m in the Southern Province of
Zambia receiving an annual rainfall of about 800 mm.
The programme for this particular intervention was to meet with some members of the
community on the first day and then do some model runs on the second day. During the third
day the model outputs were to be shared with the community (Carberry et al., 2004).
Community members who are farmers from villages in Mujika and Njola districts were
selected by the camp extension officer to attend the meeting. Following the introductions and
a brief outline of the vision of the project on the first day, the group divided into their village
groups for the natural resources mapping exercise. Each village first drew a map of their own
area – including houses, roads, fields and sources of water. Then each farmer proceeded to
draw an individual map of their own farm, again including the homes and fields. Using a
dialogue method with translation into Tonga, the farmers then also added specific details of
the natural resources including soil type and water resources. As part of the exercise was to
compare the farmers production with the model output, other information about the past two
season’s production was also requested and written on their maps – such as land preparation
method, planting dates, fertilization, weeding and grain production. As a task for homework,
they were requested to consider what variations in management or adaptations they might
make if the season was to be drier or wetter than normal. On the second day the researchers
met and discussed the individual maps and farm management activities. After compiling the
available climate data into the necessary format for APSIM, the model was run for several
management scenarios. These outputs were then transferred into simple graphs for
presentation to the farmers on the third day. The first activity on the third day was to receive
the report back on the possible intervention that the farmers could envisage implementing.
The latest seasonal forecast from the Zambian Department of Meteorology was given to the
famers. This was followed by reports from three individual farmers, and then an introduction
to crop modelling. Some of the model results from two particular farmer’s resource maps
were shown. Following this, a discussion ensued around the proposed mother-baby trials for
the coming season (Snapp, 2002; Rusike et al., 2006).
432
Session C2: Modelling adaptation strategies to cope with climate change
Results and discussion
A total of 40 farmers (equal numbers of male and female) attending the meeting were from
the following villages – Bulimo, Homooyo, Kayumba, Malomo, Nkabika and Sikaula. From
the farmers’ maps is became clear that there were two dominant soil types, namely a sandy
loam and a clay soil occurring mainly in the low lying parts of the landscape. The farmers
plant a wide variety of different maize varieties (up to 19) with most planting medium length
varieties (500 and 600 series) and they usually only weed their lands once per season. The
farmers use either cattle manure or inorganic fertilizer as a basal and/or top dressing. The
tillage is done either by hand hoe or with an animal drawn plough and some form small basins
as a water harvesting technique.
The seasonal rainfall forecast was for normal rainfall amount during the Oc-Nov-Dec
season and normal to above normal for the January-February-March season. The farmers said
that they would continue to plant the 500 and 600 series varieties after the first rains had been
received and try to do the weeding and top dressing at the optimal time. They all expected a
good yield this season as the forecast was for normal to above normal rainfall.
The rainfall variability was discussed using the graph of the daily rainfall from the
previous season, showing that it rained almost every day during December 2007, which the
farmers clearly remembered. This had resulted in waterlogged fields and poor maize production. The variation between seasons was also illustrated by looking at the graph of the last 11
years of seasonal rainfall amounts and the various low and high rainfall seasons were
discussed. This graph was then compared with the model predicted yield graph, which
showed that in one high rainfall year the yield was still low and in one low rainfall season the
yield was average. The farmers also compared the yields that they could remember from the
various years, agreeing that the model had done a good job of simulating the yields over the
last 11 years.
One of the farmers, Constantine Manianga from Homooya village, who had applied
manure to a certain field in two consecutive years then, commented on the model simulations
of his particular field. The model had been run with the fertilizer and manure applications as
reported on his farm resource map. In the last two seasons (06/07 and 07/08) the model had
accurately predicted his yields as one year his reported yield was slightly higher and the other
slightly lower that the predicted values. He said that this gave him confidence in the use of the
model to show the effect of various interventions.
The effect of additions of the recommended basal and top dressing fertilizer on the grain
produced on the clay soil was then illustrated with a graph comparing the predicted yield with
and without fertilizer over the past 11 years. Most of these yields were above those reported
by the farmers, so they were encouraged to include fertilizer treatments in the mother-baby
trials in the coming season. There was much discussion around which treatments were of
most concern, including different varieties; different tillage practices; planting dates and
fertility. However, it was not possible to come to a consensus on which were the most
important to be included in the trials. Unfortunately, due to lack of time the discussion had to
be cut short and will be resumed at the next meeting of the farmers with the researchers and
extension officer.
References
Carberry, P., et al., 2004. In: Eds R.J. Delve & M.E. Probert, Modeling Nutrient Management
in Tropical Cropping Systems. ACIAR Proceedings 114: 32-46.
ICRISAT, 2008. Building adaptive capacity to cope with increasing vulnerability due to
climate change. Inception Workshop Report, 7-8 June 2007, Bulawayo, Zimbabwe.
Rusike, J., et al., 2006. Int. J. Agric. Sustain. 4(2): 1-17.
Snapp, S., 2002. In: Eds M.R. Bellon & J. Reeves, Quantitative analysis of data from
participatory methods in plant breeding. CIMMYT, Mexico.
433
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme C
Case studies and application of tools and
empirical methods
Session C3: Application of integrated assessment tools
Session organizers:
Marijke Kuiper
LEI, Wageningen UR, The Netherlands
Jacques Wery
UMR System (Agro.M-CIRAD-INRA), Montpellier, France
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Effects of trade liberalization on EU agriculture:
A typical macro level application of SEAMLESS-IF
M. Adenäuer, M. Kuiper, K. Louhichi
Nussallee 21, 53115 Bonn, Germany
Contact: marcel.adenaeuer@ilr.uni-bonn.de
Introduction
EU trade policy is determined in Brussels by EU institutions but has ramifications that may
differ for EU member states and regions within member states which affect the political
feasibility of the agreement reached by the EU commission. Furthermore (economic) trade
policy may have environmental and social impacts which could support or conflict with other
policies at EU, national or regional level. Furthermore EU trade policies have to be in line
with WTO commitments.
In this paper, we assess the impact of a reduction in tariffs and EU export subsidies along
the lines of the proposed modalities for agriculture of July 2008 of the World Trade
Organization (WTO) negotiations (Doha Development Round). These modalities include
reductions in tariffs (with exceptions for sensitive products) and removal of export subsidies.
The modalities are not yet fully specified due to the breakdown of the negotiations in July.
We explore different potential negotiation outcomes by assessing the impact of not arriving at
a provision for sensitive products (one of the most contested elements in the agricultural
modalities), and varying tariff reductions to both the upper and lower bounds in the
modalities. These additional three scenarios provide an indication of the sensitivity of our
conclusions to variations in negotiation outcomes.
Apart from providing an integrated assessment of the July 2008 agricultural modalities in
terms of its economic and environmental impact, a second objective of our paper is to
evaluate to what extent an integrated assessment that accounts for detail in agricultural
production adds insights that cannot be derived from other existing models. To this end we
compare the results of a chain of models to the results from applying the same scenarios in the
existing CAPRI model.
Methods
We make use of the model chain FSSIM-EXPAMOD-SEAMCAP-FSSIM, reaching from the
field to the global market level, as realized in SEAMLESS-IF (Van Ittersum et al., 2008).
Producer response is modelled using farm level simulation models (FSSIM) developed in the
SEAMLESS project (Louhichi et al., 2007). These models are available only for a subset of
the EU regions and therefore an extrapolation procedure is employed to determine supply
response at NUTS2 level using the EXPAMOD model (Bezlepkina et al., 2007). The
resulting elasticities of agricultural supply are then used in SEAMCAP, a special version of
the agricultural sector model CAPRI (Britz et al., 2007), to arrive at a consistent modelling of
agricultural production from field to market level. In SEAMCAP we apply the trade
liberalization shock and obtain a new price quantity equilibrium. The new equilibrium prices
are then transferred back to FSSIM in order to obtain feedback on farm type level to the trade
liberalization. This provides an assessment of the July 2008 agricultural modalities from farm
to market level.
To achieve our second aim we also compare the results to an analysis of the same
scenarios, but with the standard version of CAPRI in order to asses the value added of an
integrated assessment with SEAMLESS-IF.
436
Session C3: Application of integrated assessment tools
Results and discussion
The first analysis with the stand-alone CAPRI model reveals considerable impacts of trade
liberalization on the European agricultural sector. On the economic side we observe
increasing partial welfare due to consumers benefiting from lower prices. This comes at the
cost of farmers who will suffer from those price changes. We further find out that the animal
sector is more heavily affected than the crop sector mainly due to meat markets currently
being subject to stronger border protection compared to most of the major crop products. This
in turn leads to a quite diverse picture of income changes across European NUTS2 regions
and farm types.
In terms of environmental indicators the analysed scenarios reveal positive aspects. Lower
prices across the whole agricultural sector tend to reduce the production of agricultural
commodities and for example animal herd sizes what particularly reduces methane emissions
and nitrate surplus from manure application.
First findings indicate that different results will be obtained with the SEAMLESS-IF model
chain than with the same application in the standard stand-alone CAPRI model, since the
agricultural supply response in the model chain differs from that in the standard CAPRI
model. A different supply response implies that the changes in agricultural production will
vary between the SEAMCAP in SEAMLESS-IF and the stand-alone CAPRI model, which
will affect all other components of these two models (like welfare of consumer and market
prices). Furthermore, a deeper look into changes occurring at farm type level including
corresponding environmental impacts is possible when using the SEAMLESS-IF model
chain.
References
Bezlepkina, I., et al., 2007. EXPAMOD: Component to statistically extrapolate from FSSIM
models to other farm types and regions including aggregation to NUTS2: motivation,
description and prototype, PD3.6.11.2, SEAMLESS integrated project, EU 6th Framework
Programme, contract no. 010036-2. www.SEAMLESS-IP.org, 28 pp.
Britz, W., et al., 2007. Description of the CAPRI Modeling System, Final report of the
CAPRI-DYNASPAT project; Universität Bonn, available on http://www.ilr1.unibonn.de/agpo/rsrch/capri/Final_Report_Model_Description.pdf
Louhichi, K., et al., 2007. A generic Template for FSSIM for all farming systems, PD3.3.11,
SEAMLESS integrated project, EU 6th Framework programme, contract no. 010036-2,
www.SEAMLESS-IP.org, 85 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
437
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Ex-ante institutional compatibility assessment of policy options:
Insights from empirical applications and interactions with policy experts
L. Amblard, C. Mann, C. Schleyer
Cemagref, UMR Métafort, 24 avenue des Landais, 63 172 Aubière, France
Contact: laurence.amblard@cemagref.fr
Introduction
The Procedure for Institutional Compatibility Assessment (PICA) has been developed in the
EU-project SEAMLESS-IF (Van Ittersum et al., 2008) as a formalized methodology to assess
ex-ante the compatibility between policy options and different institutional contexts
(Theesfeld et al., 2008). In 2007 and 2008, PICA was applied to the implementation of the
policy option ‘EU Nitrate Directive’ in the regions of Auvergne and Midi-Pyrénées in France.
Additionally, interactions with policy experts as one important group of end-users of PICA
were tested in order to define their preferred modes of interaction. This paper presents the
results of the application of PICA to the implementation of the EU-Nitrate Directive in
Auvergne and Midi-Pyrénées. The contribution of the tool to integrated ex-ante policy
assessment within SEAMLESS-IF (S-IF) is highlighted by the combination of the PICA
statements and the results of the ex-ante assessment of the EU Nitrate Directive carried out by
the other SEAMLESS models in Midi-Pyrénées. Further, the implications of the empirical
applications for methodological improvement of PICA are discussed. Finally, the testing
activities provide insights with regard to the use of PICA from the point of view of the future
'integrative modeller’ running the tool as well as from the perspective of policy experts.
Methods
Within a first testing, PICA was applied ex-ante to the implementation of the policy option
‘EU Nitrate Directive’ in the ‘département’ Puy-de-Dôme in Auvergne, France. For
methodological purposes, two additional analyses were carried out in the neighbouring
‘département’ Allier, where the Nitrate Directive has been implemented since 1994. A
‘simulation’ of running PICA before the actual implementation of the EU Nitrate Directive as
well as an ex-post evaluation of the process were realized. Empirical methods used include
literature reviews, statistical data analysis and interviews with scientists and stakeholders
involved in the implementation process of the EU Nitrate Directive in Allier and Puy-deDôme. The second application was conducted on the case of the implementation of the EU
Nitrate Directive in Midi-Pyrénées. An external expert with a background in institutional
economics carried out the PICA assessment within a given timeframe of two months. The
requests and experiences of the external expert were documented to improve the quality of the
already available information on PICA and for the development of training materials for
future applications of this tool. The interactions with policy experts regarding the use of PICA
were also tested, using meetings between Cemagref researchers and policy experts from the
regional services of the French Ministries of Agriculture and Environment as platforms.
During workshops carried out by the SEAMLESS-PICA team, the perception of PICA by
policy experts and their preferred modes of interaction were evaluated.
Results
The empirical application of PICA in Puy-de-Dôme shows that the most important factors
potentially affecting the implementation of the EU Nitrate Directive in this ‘département’ are
related to the influence of farmers’ organizations. The restrictions on fertilization which could
be imposed by the Directive in the vulnerable zone are likely to affect the capacity of farmers
to fulfil the terms of the contracts for high quality products with the two large cooperatives
present in the area. As a result, these two cooperatives may seek to use their high bargaining
438
Session C3: Application of integrated assessment tools
power to influence the choice of the measures in the action programme to be implemented in
the vulnerable zone. Also, the opportunity costs borne by farmers are likely to constraint the
implementation of the policy. Indeed, the farmers under contract with the cooperatives will
incur income losses if they have to change their fertilization practices so that they can not
fulfil the requirements agreed on in these contracts. Then, one could expect a low degree of
compliance with the mandatory rules in case these rules prevent farmers to fulfil the terms of
the contracts they depend on economically.
In contrast, the opportunity costs borne by crop farmers to change their practices in MidiPyrénées do not seem to be high, thus not necessarily hampering the implementation process.
This finding is in line with the aggregated result of the CropSyst-FSSIM assessment that the
loss of income induced by the adoption of alternative activities is marginal (Louhichi et al.,
2008). However, while the modelling of cross-compliance in FSSIM rests on the assumption
of full monitoring such that any non-compliance is detected, the PICA assessment reveals that
the low resources devoted to the administrations in charge of monitoring as well as the high
information asymmetry between the administrations and the farmers in Midi-Pyrénées are two
important factors which may prevent the detection of non-compliance. Moreover, beyond
economic reasons and controlling issues, psychological factors were found to have a strong
influence on farmers’ behaviour in the region, which may also lead them not to comply with
the EU Nitrate Directive. Given these institutional features, the compliance of crop farmers
with the EU Nitrate Directive measure and thus the decrease in nitrate leaching in MidiPyrénées may be not as significant as predicted by the model chain.
Discussion
The two first empirical applications of PICA proved to bring relevant results with regard to
the potential institutional constraints to the implementation of the EU Nitrate Directive in
Auvergne and Midi-Pyrénées. Relating PICA statements to the model chain results illustrates
the added value of taking into account an institutional perspective within SEAMLESS-IF exante policy assessment. Further, the two assessments helped to evaluate the procedure from a
methodological perspective. The ability of PICA to ‘predict’ ex-ante the institutional factors
potentially affecting the implementation of a policy option and the tool’s capacity to account
for differences between distinct institutional contexts were validated. Regarding the empirical
methods used, the applications brought valuable insights on the combination of qualitative
and quantitative data, the choice of qualitative methods and the integration of stakeholders’
points of view for the collection of the information needed. Finally, the application of PICA
by an external expert offered the opportunity to define the conditions for successful future
applications in terms of competencies, timeframe and information/training to be provided
while the analysis of the perception of policy experts helped to identify improvements of the
communication on the tool and to specify further the modes of interaction with end-users.
References
Louhichi, K., et al., 2008. Application of FSSIM in two Test Case regions to assess agroenvironmental policies at farm and regional level - PD6.3.2.2, SEAMLESS integrated
project, EU 6th Framework programme, contract no. 010036-2, www.SEAMLESS-IP.org,
62p.
Theesfeld, I., et al., 2008. Ex-ante policy assessment from an institutional perspective. A
Procedure for Institutional Compatibility Assessment (PICA). ICAR Discussion Paper,
Humboldt University Berlin, 25 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
439
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Challenges in the application of multi-criteria assessment (MCA) tool for
the evaluation of irrigation management alternatives: PLEIADeS project
Paula Antunes, Rui Santos, Vanja Karadzic
CENSE - Ecological Economics and Environmental Management Group,
Faculty of Sciences and Technology, New University of Lisbon,
Campus da Caparica, 2829-516 Caparica, Portugal
Contact: mpa@fct.unl.pt
Irrigated agriculture remains the dominant single water user at the global level. Hence, it is
the key strategic focus for efficient water use. The existent literature and on-field experience
provide information on promising water saving projects, dealing with the water use efficiency
and water productivity, irrigation system performance, use of non-conventional water sources,
focus on participatory management, etc. Even though most of these programmes substantially
contribute to increase of irrigation efficiency, they have been sector oriented and restricted to
certain pilot areas, hence missing an integrated management approach. To address this issue
PLEIADeS (Participatory multi-Level EO-assisted tools for Irrigation Water Management
and Agricultural Decision-Support) STREP-F6 research project was developed with the
overall objective of improving the technical, environmental and economic performance of
irrigation schemes by means of New Technologies (NT): Earth Observation (EO) satellites,
Geographical Information Systems (GIS) and Information and Communication Technologies
(ICT) that can facilitate integrated management (including the integration of different spatial
scales) and active stakeholder participation. Within the scope of the project, Multi-criteria
Assessment (MCA), as an integrated assessment tool is used for evaluation of different
irrigation management options, combining economic, technical, environmental, and social
aspects. In this paper we describe a distinct MCA framework; developed taking into
consideration differences in the assessment scope, involvement of stakeholders’ and
information needs, thus aiming for application beyond the scope of the PLEIADeS project.
The overall evaluation is based on approach derived from the concept of social multi-criteria
evaluation - SMCE (Munda, 2004), involving an institutional analysis and the active
participation of stakeholders in several stages of the process, namely in the identification of
the alternative actions, in the definition of the evaluation criteria, on the assessment of tradeoffs among alternatives/criteria and on the definition of an aggregation procedure. Taking into
account differences in the irrigation management setting in 12 PLEIADeS pilot areas
worldwide, the general MCA framework is applied for two levels of assessment:
1. Basic, not driven by any of the pilot areas, but based on a general approach to assess
irrigation efficiency, environmental performance as well as for the economic analysis of
irrigation systems. The result of the analysis is just an overall assessment of the
alternatives, and therefore does not need to accommodate a complex decision structure.
This very general approach may enable a certain level of direct comparisons across the
different pilots in PLEIADeS.
2. Advanced, ‘tailored approach’ to pilot area specific irrigation management setting.
Focus is on evaluation and comparison of alternatives, designed with stakeholders in a
participatory setting (workshops). Alternatives should be constructed as a group of actions
that better inform the actors taking part in a decision process.
For each level the proposed MCA framework defines: (a) necessary level of information
detail; (b) intensity and impact of stakeholders’ involvement; (c) aggregation rule and d)
presentation technique.
440
Session C3: Application of integrated assessment tools
Table 1. How MCA differs between basic and advanced levels of assessment.
Differences addressed
Levels of assessment
BASIC
ADVANCED
INFORMATION USED
Existing information and data (e.g.,
background of irrigation scheme,
technical annual reports,
historical and institutional
analysis). Information used based
on qualitative assessments
STAKEHOLDERS’
INVOLVEMENT
(who and how?)
Informal meetings and consultation
with the project partners that
could facilitate understanding the
problem and constructing
alternatives Stakeholders’ not
directly involved in the
assessment process.
AGGREGATION RULE
No aggregation
PRESENTATION TECHNIQUE
General impact matrix with
qualitative criteria scores
Graphical display
Full stakeholder analysis resulting
in the pilot area report,
including precise description of
the irrigation management
setting. Alternatives designed
considering relevant
stakeholders’ and experts’
contributions
Qualitative & quantitative data
and information coming from
scientific and technical reports,
interviews, consultation, etc.
Interviews and consultations:
mapping views, preferences and
values
Workshops:
1st round: framing the problem
and discussion including
different perspectives (farmers,
irrigation scheme managers,
river basin authorities’)
2nd round: generation of
alternatives & criteria;
3rd round: presenting and
discussion of results
Combination of different
aggregation methods
Graphical: rank bars, diagrams
SMCE – dendogram of coalitions;
ranking of alternatives
Combination with GIS
On the base level, two types of questionnaire surveys were distributed to PLEIADeS pilot
areas’ stakeholders’, used to explore their values, awareness and knowledge about irrigation
efficiency, environmental problems, economy (water pricing), society (culture, history,
tradition) and technology (available irrigation technologies). All interviews were recorded and
transcribed, facilitating construction of the first set of irrigation management alternatives and
criteria. We selected the case study of Caia, Guadiana river basin, Portugal for the MCA
advanced level. The assessment process in this case is combined with other participatory
techniques (Antunes et al., 2006, Kallis et al., 2006), i.e. workshop, where participants are
directly involved in alternative and criteria selection, weighting and aggregation procedure.
Ultimately, MCA application will aim to address the specific question of evaluating the effect
of NT-assisted tools on water productivity and performance of pilot irrigation systems using
an extended evaluation.
References
Antunes, P., et al., 2006. Land Use Policy 23: 44-52
Kallis, G., et al., 2006. Environment and Planning C: Government and Policy, 24: 215-234.
Munda, G., 2004. European Journal of Operational Research 158: 662-677.
441
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A crop-farm-indicators modelling chain to assess farmer’s decision in
response to socio-economic scenarios
H. Belhouchette, K. Louhichi, O. Therond, J. Wery, G. Flichman
UMR-SYTEM, 2, Place Pierre Viala,
34060 Montpellier Cedex 1, France
Contact: Hatem.Belhouchette@supagro.inra.fr
Introduction
Bio-economic models can be used to assess the impact of policy and environmental measures
with economic and environmental indicators. Focusing on agricultural systems, the strategies
adopted by farmers in term of cropping systems and the associated crop managements at field
level are essential in such studies. The objective of this paper is to present a study using such
models to assess impacts of the nitrate directive in the Midi-Pyrenees region (France) by
analysing, at farm and regional level, the farm income and three environmental indicators:
nitrate leaching, erosion and water consumption. In addition, several intermediate variables
are calculated and analysed at field level for a better comprehension of the farmer’s strategies.
Methods
Two scenarios, the 2003 CAP reform (reference scenario) and the Nitrate Directive (policy
scenario), with a time horizon of 2013, have been built and compared for two representative
arable farm types in the Midi-Pyrenees region. The main structural difference between the
two farm types is the irrigable land which is 37% and 13% of total land, respectively, for farm
1 (FT1) and 2 (FT2). For this purpose, the modelling chain: CropSyst-FSSIM-Indicators
developed in the SEAMLESS project has been used (Louhichi et al., 2008). Different types of
data characterizing the biophysical context in the region (soil×climate), the current cropping
systems (rotation, crop management) and farm resources (irrigated land, labour) have been
collected to calibrate and run the models. In addition, different alternative activities with low
nitrogen fertilization, in compliance with the Nitrate Directive, have been described. Current
and alternative activity productions and externalities have been assessed with the CropSyst
model. For each alternative activity an additional cost of 5% of total current activity cost is
added as transaction cost in order to represent the additional knowledge and workload needed
by farmers to acquire information on these activities. In addition, 3% of cross-compliance
restriction is implemented if the farmer doesn’t respect the Nitrate Directive.
% of area variation
spring crops
maïze
oats
barley
D. wheat
442
S. wheat
Results and discussion
60
Figure 1 shows the difference of crop pattern for policy
40
and reference scenarios. More cereals and irrigated grain
maize are cultivated in FT1 when the Nitrate Directive is
20
implemented. However, the area of spring crops is reduced
0
by almost 50%. In FT2, characterized by a low irrigable
-20
area, the irrigated maize area remains marginal, with no
change of spring crops and a slight increase of cereal area
-40
(data not shown). To understand those modifications and
their environmental consequences we compared (Table 1):
- activity share: The model results show that few Figure 1. Policy to reference
alternative activities (low N fertilizer) have been adopted scenario: % of area variation
when the policy scenario was implemented. This implies in FT1.
that the penalty of 3% is not enough to force farmers to
respect the cross-compliance condition. In fact, in both farm types only 5ha (less than 10% of
total cereal area) of cereals were cultivated with alternative nitrogen fertilization.
Session C3: Application of integrated assessment tools
- Farm income: no significant changes are observed for farm income when the policy
scenario is compared to the reference one. For FT1 the reduction of total premium induced by
the decrease of spring crops (–4700 euros) is compensated by the increase of the total
premium allocated mainly for wheat (durum and soft) and for maize (+ 5000 euros). The same
tendency is observed for FT2.
- Nitrate leaching:
Table 1. Economic and environmental indicators at farm scale.
The impact of policy
FT1
FT2
scenario on the total
Reference Policy Reference Policy
nitrogen leaching for
–1
835
830
761
761
both farm types is low Farm income (euros ha )
Premium
(euros)
36143
35057
38767
37600
(+4% for FT1 and –8%
–1
3
9
3
3
for FT2). The partial Labour (h ha )
–1
43
45
54
51
adoption of alternative Nitrogen leaching (kg ha )
–1
Soil
erosion
(t
ha
)
1.9
1.6
2.6
1.9
activities seems to be the
41
60
8
5
origin of this low Water use (mm)
variation. The reduction of the total nitrogen leaching caused by the decrease of spring crops,
barley and oats is replaced by the increase of wheat and maize (data not shown). In addition,
more nitrogen was leached with maize in the policy scenario because maize is cultivated as a
mono-crop (producing 41 kg ha–1 of N leached), than in the reference one, where almost 50%
of the area is cultivated with maize/soybean rotation (with 31 kg ha–1 of N leached).
- Soil erosion: The major changes occurred on soil erosion which was reduced by 16 to 29%,
respectively, for FT1 and FT2, mainly because of the reduction of spring crops (sunflower,
soybean, and peas) to the benefit of maize and winter wheat, thereby reducing the bare soil
area during winter. The total soil erosion induced by maize seems to be inconsistent because,
the total soil erosion induced by this crop decreases despite the increase of its area (data not
shown). This result comes from the fact that all maize, in the reference scenario, is cultivated
on clay-loam soil (with 1.2 t ha–1 yr–1 of erosion); while, in the policy scenario, 50% of maize
is produced on clay calcareous soil, which is less sensitive to soil erosion (with 0.3 t ha–1 yr–1
of erosion) than the clay-loam soil.
- Water use: The total water use increased in FT1 (+19 mm) and slightly dropped in FT2 (–3
mm) when the policy scenario is implemented. The area allocated to irrigated maize, which
partially depends on the irrigable area, in both farm types seems to be the origin of water use
variation. In FT2, the marginal decrease of the irrigated area reserved for maize is
compensated by the increase of the area of irrigated peas (+1.5 ha) which need few water in
comparison to maize. As a consequence of a such strategy, the total labour by farm type
increased by 6 h ha–1 for FT1 and by 1 h ha–1 for FT2.
These results show that this modelling chain can be functional for complex scenarios
combining economic and environmental drivers with technological changes, and provided
relevant results at farm and regional scale (data not shown), when discussed with local
experts. Results of this study indicate that the modification of environmental regulations may
lead to several changes in crop rotations on each soil type and by consequence in the crop
management options and their spatial allocation. So, even if the farm profitability (farm
income) did not change, several modifications occurred at field scale depending of the
biophysical context and the economic and social constraints. This work highlights some key
methodological aspects for future improvements and further uses of the meso backbone
modelling chain of SEAMLESS-IF: (i) organization of iterative and cyclical interactions with
local experts, (ii) development of sound methodologies for model calibration and validation at
field, farm and regional levels.
Reference
Louhichi, K., et al., 2008. Application of FSSIM. www.SEAMLESS-IP.org, 67 pp.
443
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Land-Use Management using a soil survey geographic database for the
Viisoara-Aiton hills (Transylvanian Plain-Romania)1
H. Cacovean, T. Man, T. Rusu, D.C. Weindorf, I. Fodorean, C. Moldovan, V. Puiu, A. Pop
O.S.P.A., 1 Fagului street, 400483 Cluj-Napoca, Romania
Contact: turda75@yahoo.com
Introduction
In Romania, the determination of the best management practices (BMPs) for land resource
management is complicated by a lack of modern means for land evaluation and a lack of soil
quality. Quantifying agricultural landscape patterns and their change is essential for
monitoring and assessment of the ecological consequences of recent transformations in
agriculture of the Viisoara-Aiton Hills (Irvin et al., 1997; Wu, et al., 1997). Using GIS-based
land-use data and pedological data we have constructed an analysis of the surrounding of the
urban center of Turda with landscape metrics to quantify the spatial patterns of the
agricultural transformations in the Viisoara-Aiton Hills. Modernization of the agricultural
practices is a major force driving land-use change, which inevitably affects the structure,
function and dynamics of the land-use cover and ecosystem.
Material and methods
The Viisoara-Aiton Hills, located in the southern part of the Transylvanian Plain (Romania),
was chosen for this study. This region has a surface of 8,670 ha and represents the most
important agricultural area in the province of Cluj. This province is well developed and
economically important in Romania. Agriculture is an important economic branch in this
geographic area, with corn (Zea mays) and wheat (Triticum aestivum) being the major crops.
Existing soil maps of the area contain 4 classes and 26 types of soil, not including significant
inclusions that could also occur within each map unit. Each soil layer has unique
physicochemical properties, and an area-weighted number was computed, representing the
value of the property for each map unit (Richardson et al., 1977). In this study, we considered
organic matter (OM) content, clay content, soil reaction and soil texture that are the main
variables to asses the land-use management of that area (Fix & Burt, 1996). From these
properties, interpolated maps were generated using geographic information system (GIS)
software (Blaszczynski, 1997). The topographical map, as obtained from the Military
Topography Department (1996), was used to realize the physiographic layer and because the
physiographic region of Viisoara-Aiton Hills is closely related to soil cover. This region
consist of one physiographic region and by referencing the physiographic map of the
Transylvanian Plain we merged all soil types within the same region and formed a final
physiographic layer. The four thematic maps were coupled with aerial photos and used to
evaluate land use/management in relation to quantified soil properties (Lark, 1999; Park et al.,
2001. The objectives of this study were to: (1) identify land-use change; (2) evaluate the
influence of soil cover on land-use; and (3) assess land-use potential and present management
alternatives.
Results and discussion
To evaluate the importance and influence of OM content, clay content, soil reaction and soil
texture on land-use patterns, we correlated land-use distribution with soil types/properties.
Corn farmland had the lowest OM content, while wheat and alfalfa farmland approached
average values. In terms of clay content, grassland showed the lowest value of all land-use
types. The percentage of each textural class of the surface horizon that was used for wheat,
1
In cooperation with University of Babes-Bolyai, Romania; USAMV, Romania; Louisiana State Univ., USA.
444
Session C3: Application of integrated assessment tools
corn and grassland. We find four textural classes in that region and without any difference for
the wheat and corn. Land capability classes are generally used to make land-use assessments,
determine erodible areas and review land-use potential. An examination of land suitability
class for a forest steppe conditions is important in the determination of land-use alternatives.
By overlaying the suitability class with the land-use map, we obtained a map that highlighted
those surfaces with lower suitability classes: ~1,256 ha of Class II, 3,478 ha of Class III,
2,145 ha, and 1,791 ha of Class IV. Those lands with suitability Class IV or higher are
unsuitable for cultivation, and should be used for conservation practices such as grassland.
References
Blaszczynski, J.S., 1997. Photogrammetric Engineering and Remote Sensing 63: 183-191.
Fix, R.E. & T.P. Burt, 1996. Earth Surf. Processes Landforms 20: 817-827.
Irvin, B.J., et al., 1997. Geoderma 77: 137-154.
Lark, R.M., 1999. Geoderma 92: 141-165.
Park, S.J., et al., 2001. Geoderma 103: 249-272.
Richardson, A.J., et al., 1977. Eng. Remote Sensing 43: 207-216.
Wu, J., et al., 1997. J. Soil Water Cons. 52: 352-358.
445
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Organic crop rotations evaluated using LCA (Life Cycle Assessment)
L. Coletto, S. Bona, S.G. Sandrini, A. Brigi, G. Florio, P. Sambo
Dipartimento di Agronomia Ambientale e Produzioni Vegetali,
AGRIPOLIS - Facoltà di Agraria - Università di Padova,
Viale dell’Università n. 16 - 35020 Legnaro (PD), Italy
Contact: lucia.coletto@unipd.it
Introduction
Frequently, when dealing with cropping system evaluation, the focus is only on individual
effects such as nitrate leaching or ammonia volatilization (Bach & Becker, 1995) although
agricultural production systems may have a wide range of environmental impacts (e.g.,
climate change, acidification, eutrophication, etc.). The analysis of individual effects does not
permit an overall evaluation from an environmental point of view. The life cycle assessment
(LCA) methodology was designed to study all environmental impacts connected to an entire
production system (Consoli et al., 1993). In the case of cropping systems this includes not
only on-field activities but also all impacts related to the production of farm inputs, such as
emissions and resource consumption due to the production and transport of fertilizers.
The aim of this work was to investigate the environmental impact of different cover crops
inserted in different crop rotations. We utilize LCA for analysing the different cover crops as
this is a valid tool for comparing similar products and allows the identification of possible
improvements of products and processes during their whole life cycle.
Methods
The LCA methodology has been applied to the crops of the rotations conducted in the
Experimental Farm of Padova University (Project financed by: Progetto FISR - SIMBIOVEG). Crops included in the LCA calculation were: the cover crops, oat-vetch-pea (OVP),
rye–vetch (RV) and forage sorghum (FS); the crops, soybean (SO), maize (MA) and einkorn
(EK), in comparison with bare soil (BS).
The main objective was to evaluate the environmental impacts of cover crops placed in
different 3-year crop rotations representative of the current systems of the region (Table 1).
The three crop rotations under control differ only for the cover crops not for the crop
sequence, allowing to focus on the analysis of the effect of cover crops (in S1 and S2)
compared to bare soil in the same rotation. The data represents the values accumulated since
the start of field trial (September 2005) until autumn of 2008. As we worked on 3-year crop
rotations and as all phases of all crop rotations were present each year, we averaged the values
obtained from each rotation.
The results refer to the impact categories proposed by SETAC (Consoli et al., 1993).
Table 1. Crop rotations under investigation in the organic field trial.
S1 = EK – FS – OVP – SO – OVP – MA
S2 = EK – FS – RV – SO – RV – MA
Tn = EK – BS – SO – BS – MA
Results and discussion
As expected, the lack of cover crops (Figure 1) in a crop rotation induces a decrease in the
number of tillage which leads to a decrease in consumption of fossil fuels (for the
construction and utilization of machineries and tractors).
Despite the significant use of fossil fuel, emission of greenhouse gases of all types of cover
crops were negative, which means that crops uptake greenhouse gases (the largest amount is
446
Session C3: Application of integrated assessment tools
attributable to CO2). It is interesting to note that even in the case of bare soil, there is a storage
of CO2.
Ozone layer
Fossil fuels
7000
8.0E-07
6000
7.0E-07
6.0E-07
5000
5.0E-07
4000
4.0E-07
3000
3.0E-07
2000
2.0E-07
1000
1.0E-07
0
0.0E+00
S1
S2
TN
S1
S2
TN
Climate change
Carcinogenics
S1
8.0E-03
S2
TN
0.0E+00
7.0E-03
-5.0E-04
6.0E-03
-1.0E-03
5.0E-03
-1.5E-03
4.0E-03
-2.0E-03
3.0E-03
-2.5E-03
2.0E-03
-3.0E-03
1.0E-03
-3.5E-03
-4.0E-03
0.0E+00
S1
S2
TN
-4.5E-03
Figure 1. Amount of ‘fossil fuels’ used (MJ per hectare), gases affecting ‘ozone layer’ and
‘carcinogenics’ into the atmosphere (data expressed as years of life lost by humans) and
substances affecting the ‘climate change’ (data expressed as years of life lost by humans) as a
function of cover crop per hectare and per year of cultivated area.
Carcinogenic and ozone layer substances emissions (it must be note that in this category all
tractors emissions and those used for constructing the tractors, for transporting, etc., are
included), in the case of the two types of cover crops, were higher than those of the bare soil.
This can be explained mainly by an increase of machinery utilization.
The evaluation crop production is very often limited at field level and it does not include
all the processes which contribute to the emissions in all the compartments. This is especially
true for organic productions. The introduction of cover crops induces positive effects on the
climate change emissions (CO2 storage in the soil) but not in terms of carcinogenics for the
more elevated number of tillages.
References
Bach, M. & R. Becker, 1995. Regional differenzierte Abschätzung des Nitrateintrages aus
diffusen Quellen in das Grundwasser-Untersuchung für die Bundesrepublik (Regionally
differentiated estimation of nitrate immissions to groundwater from non-point sources –
investigations in Germany). Institut für Landeskultur, University of Gießen.
Consoli, F., et al. (Eds), 1993. Guidelines for Life-Cycle Assessment: A ‘Code of Practice’.
Society of Environmental Toxicology and Chemistry (SETAC), Brussels.
447
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Using MASC to evaluate the sustainability of cropping systems:
Application to multi-criteria assessment of organic cropping systems in
Southern France
B. Colomb, F. Angevin, J.E. Bergez, M. Blouin, F. Carpy-Goulard, A. Glandières, L. Prieur
UMR 1248 AGIR, BP 52627, 31326 Castanet Tolosan, France
Contact: colomb@toulouse.inra.fr
Introduction
Cropping systems can be defined as ‘a set of management procedures applied to a given,
uniformly treated area, which may be a field, part of a field or a group of fields’ (Sebillotte,
1990). Despite the importance of this spatio-temporal scale, very few tools are available to
assess the sustainability of such an entity. Sustainability assessment is a typical decisionmaking problem requiring multi-criteria decision-aid methods (Sadok et al., 2008). The
availability of proper methodologies and tools to identify innovative and sustainable cropping
systems is a major request from agricultural advisors. The MASC (Multi-attribute Assessment
of the Sustainability of Cropping Systems) model was developed to perform a sustainability
assessment of cropping systems using a multi-criteria qualitative analysis (Sadok et al., 2007,
2009). The objective of this paper is to assess the ability of the model to be used by a group of
organic farming advisors and to discriminate the economic performances, the social
acceptance and the environmental impacts of cropping systems currently practiced on organic
arable farms in southern France.
Methods
MASC has been programmed through the DEXi technology developed by Bohanec et al.
(2008). The model is based on a decision tree that breaks the decisional problems of the
sustainability assessment down into simpler units, referring to the economical, social and
environmental dimensions of sustainability. The aggregation processes throughout the
decision tree are based on ‘if-then’ qualitative rules, entered by the user (Sadok et al., 2008).
Theses rules reflect scientific knowledge as well as expert representations and preferences,
and result in a weighting pattern of the basic attributes to calculate the aggregated ones. As
input data the model requires a vector of 32 elementary criteria characterizing cropping
systems (e.g., profitability; health risks; pesticides emissions; energy consumption;
phosphorus use autonomy; crop diversity …). The assessment process involves the
calculation of these criteria and their homogenization into qualitative information for input
into the model. The user group has to specify how qualitative criteria should be assessed,
taking into account the specificities of the regional context and of the crop sequences under
evaluation. Threshold values, derived from local benchmarking of performances or impacts,
should be specified to get qualitative value (such as ‘low’ or ‘acceptable’) from quantitative
measurements or estimates (e.g., energy efficiency ratio).
As part of a multi-disciplinary and participative research project designed to identify ways
to promote organic farming in the Midi-Pyrenees Region, a group a five advisors was set up
to test the model. Two members of the MASC designer group joined the advisors group to
help them to fit the model to their own objectives and contexts. Fifty crop sequences were
then analysed with MASC. Six crop sequences (six year long, 2001–2007) were chosen from
the Experimentation and Research Center on Organic Farming located in Auch (Prieur, 2006),
affiliated to the ITAB (National Organic Farming Institute). Forty-four sequences (4 year
long, 2002–2005) were chosen by the advisors amongst 20 representative stockless organic
farms located in the Midi Pyrenees region. The intensification levels of these crop sequences
widely differed, depending on the use of off-farm N sources on cereals (bread wheat, rye),
and on irrigation of pulse crops.
448
Session C3: Application of integrated assessment tools
Results and discussion
It took four full-day work sessions for the group to fully grasp and properly parameterize the
model. The economical and the social dimensions both required one day each. The other two
days were devoted to the understanding and parameterization of the three components of the
environmental dimension: abiotic resources conservation, biodiversity conservation and
environmental quality (impacts on waters, soils or the atmosphere). For each criteria,
discussions aimed at: (i) developing a common understanding of which performance or
impact is under evaluation, (ii) establishing a detailed calculation procedure of the assessment
indicators either qualitative or quantitative, and (iii) proposing a set of threshold values as a
basis for a nominal scale. Even if it can be seen as a time-consuming process this step is
fundamental as it allows participants to share the same representation of the cropping systems
and of how sustainability is described.
MASC discriminated efficiently the different stockless organic cropping systems, as a
function of the intensification level used by the farmers, thanks to the fine tuning of the model
done by the advisors (Figure 1). The graphical representation was done using an ordinal scale
to associate numbers to the qualitative
Non Fertilized- Non
assessments of the different indicators (1 for
Irrigated
Fertilized-non irrigated
Economic
very low up to 5 for very high). Economic
Sustainability
Non Fertilized -Irrigated
performances ranged from low to high,
Fertilized-Irrigated
depending on market prices (wheat and
soybean for human consumption being the
Biodiversity
Social Acceptance
Conservation
most profitable crops) and the amount of
European, national and local subsidies that
could be captured by the organic farmers.
Social acceptance for the farmers varied from
Resources
Environmental quality
low to very high, depending mainly onto the
Conservation
labour requirement (higher in irrigated crop
Figure 1. Comparative assessment of stockless sequences, since irrigation implies more
organic cropping systems in Southern France with mechanical interventions for weed control).
MASC, as a function of the intensification level.
The environmental sustainability varied from
medium to high, the energy balance being the main driving factor of the observed variability.
The advisors gained confidence in the model as a screening tool to rank existing cropping
systems and to deliver sound assessments of their sustainability. They are willing to use it in a
collaborative approach with farmers to design and evaluate ex-ante innovative cropping
systems. However, the advisors group worried about the effect of the uncertainty attached to
the assessment of the basic criteria, of the location/depth of some criteria in the tree, and of
the weighting pattern used for aggregation on the final rating. To tackle these issues and to
better understand the model itself, different sensitivity analyses are in progress. This
experiment additionally showed that an additional companion tool is needed to help
calculating and managing the 32 criteria required by MASC.
5
4
3
2
1
References
Bohanec, M., et al., 2008. http://www-ai.ijs.si/MarkoBohanec/dexi.html
Prieur, L., 2006. Organic eprints. Orgprints ID Numbers 6745.
Sadok, W., et al., 2007. In: Eds. M. Donatelli, et al., Farming Systems Design 2007, Int.
Symposium on Methodologies on Integrated Analysis on Farm Production Systems, 10-12
September 2007, Catania (IT), pp. 193-194.
Sadok, W., et al., 2008. Agronomy for Sustainable Development 28: 163-174.
Sadok, W., et al., 2009. Agronomy for Sustainable Development. (accepted)
Sebillotte, M., 1990. In: Eds. L. Combe & D. Picard, Un point sur les systèmes de culture.
INRA éditions, Paris (FR), pp. 165-196.
449
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Building crop producers’ capacity to adapt to climate change and climate
variability with Yield Prophet
Zvi Hochman, James Hunt, Bill Long, Dean P. Holzworth, Anthony Whitbread,
Stephen Van Rees, Peter DeVoil
CSIRO Sustainable Ecosystems, 306 Carmody Road,
St Lucia Qld 4067, Australia
Contact: Zvi.Hochman@csiro.au
Introduction
Much of Australia’s grain cropping zone has experienced a decade of below average seasons.
Despite a growing awareness of the scientific evidence for climate change, there is still a great
deal of uncertainty in farming communities about the relative roles of climate change and
climate variability in explaining this extended period of hot and dry seasons. This uncertainty
is hindering farmers from developing strategies to deal with climate change. In this paper, we
describe how Yield Prophet® (www.yieldprophet.com.au; Hochman et al., 2008), an Internet
enabled user interface to the cropping system simulator APSIM (Keating et al., 2003), has
been further developed to provide farmers and agronomic advisers with a tool that allows
them to investigate adaptation options in response to the simultaneous impacts of climate
change scenarios and climate variability on future crop production in their fields.
Methods
The effect of an increase in atmospheric CO2 concentration and temperature on wheat and
barley crop growth is captured in the APSIM wheat model with two functions that influence
key model parameters (Radiation Use Efficiency and Transpiration Efficiency) that were
derived from literature (Van Ittersum et al., 2003). Monthly mean climate change scenarios
for the year 2030 are obtained from the OzClim website (www.csiro.au/ozclim/home.do) for
the location of user defined patched point data set (PPD) met stations. IPCC climate change
scenarios used for the report are A1F1 emissions using the GFDL-CM2.1 model with high
climate sensitivity (‘worst case’) and B1 emissions using the NCAR:CCSM3 model with low
climate sensitivity (‘best case’). In order to capture the changes in extremes of the distribution
of temperatures, past trends were investigated using M-quantile regressions (Breckling &
Chambers, 1988) for reference meteorological data sets in each region. These transformations
were then applied in conjunction with the mean monthly temperature changes (from IPCC
scenarios above) to historic climate data. The historic daily rainfall series is adjusted by the
monthly mean change.
Yield Prophet® subscribers enter paddock specific information such as soil type and
nearest PPD meteorological station into the Yield Prophet® website. The Climate Change
Report uses the user’s current field’s sowing date, cultivar and plant density and runs crop
yield simulations comparing 100 years of historical data with 100 years of data transformed as
described above for the best and worst case scenarios. The Climate Adaptation Report enables
users to enter ‘adaptation’ strategies by changing sowing rules, cultivar, plant density, N
fertilizer, tillage and stubble retention under a climate change scenario of their choosing (‘best
case’, ‘worst case’ and a ‘moderate’ scenario represented by IPCC A1B emission scenarios
using the Max Planck ECHAM5/MPI-OM model and medium climate sensitivity). Farmers or
their advisers initiate reports and the information is used to generate one hundred year
continuous wheat simulations in APSIM, describing wheat yield probabilities for the historic
data, and the chosen scenarios.
450
Session C3: Application of integrated assessment tools
Results and discussion
A summary of regional differences in the Australian cropping zone in both climate change
impacts and in efficacies of various adaptation strategies will be presented. Outputs generated
by the climate change reports include an assessment of observed climate change to date
through presentation of (1) a time series and 10 year running mean of grain yield potential for
the last 100 years given current crop management, and (2) a comparison of yield potential
probabilities of the past 30 years compared to the previous 70 years. A forward-looking view
is presented by contrasting past climate with the best and worst case scenarios for 2030
(Figure 1), which in this example illustrates the importance of a ‘stereoscopic’ view of
climate change and climate variability. Contrasting yield potential of past years with that of
the best case scenario for 2030 reveals a similar median yield but with greater variability and
hence risk. The worst case scenario for this field is driven by lower rainfalls and higher
temperatures in August, September and October, which approximately corresponds to the
period of cereal crop growth from floral initiation to grain fill.
Figure 1. Yield Prophet Climate Change report output showing wheat yield implications of
climate change scenarios compared with the historically based expectation for a specific field
in Victoria.
References
Breckling, J. & R. Chambers, 1988. Biometrika 75: 761-771.
Hochman, Z., et al., 2008. Re-inventing model-based decision support with Australian
dryland farmers: 4. Yield Prophet® helps farmers monitor and manage crops in a variable
climate. Agricultural Systems. (in review)
Keating, B.A., et al., 2003. European Journal of Agronomy 18: 267-288.
Van Ittersum, M.K., et al., 2003. Agriculture, Ecosystems and Environment 97: 255-273.
451
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A decision support to choose between changes of agricultural practices:
A spatially distributed Costs/Effectiveness assessment framework
J.M. Lescot, P. Bordenave
Cemagref, ADER Research Unit, 33612 Cestas Cedex, France
Contact: jean-marie.lescot@bordeaux.cemagref.fr
Introduction
Compliance with the Water Framework Directive implies a reduction of the impact of
agricultural pressures on the environment. In France within the Adour-Garonne River Basin
District, in most of the territories of the Pilot River Basin entitled Gascogne Rivers, pollution
of drinking water by pesticides and nitrates originating from agriculture is the most important
issue. Nitrogen surplus range from 30 to 80 kilograms of Nitrogen per year and for pesticides
65% of water quality controls present at least one molecule with concentration higher than 0.1
microgrammes per liter. The work from which results are presented here has been initiated
first in the whole Gers river basin and is being refined in the upstream part of this basin
throughout the European Life Environment project Concert’Eau1. Here we focus mainly on
the environmental and economic assessment in a spatially distributed Cost/Effectiveness
(C/E) analysis framework for comparison of measures or scenarios (combination of measures)
dealing with changes of agricultural practices.
Methods
Simulation and Assessment of the scenarios: a Cost Effectiveness Analysis framework
Measures considered for changing agricultural practices were already considered in the
French Agro environmental programmes or totally new ones as defined by the actors.
Amongst these scenarios, we can notice planting hedges around farming plots and on
watercourses banks, grass strips or good agricultural practices (catch crops, long rotations, no
plowing, mechanical weeding). Only results concerning the latest measures will be considered
in this paper.
Effectiveness assessment - Hydrological modelling
Effectiveness of the measures on the reduction of Nitrogen and pesticides in the outlets was
assessed first when measures were applied on the whole upstream part of the River Gers basin
and second within this upstream part, only on zones with priority (defined for their
vulnerability). Eleven standard crop rotations with management sequences were first defined
in the area by Cluster Analysis and only the main crop rotation assigned to a district.
Assessment was carried out using the spatially distributed hydrological SWAT model2.
The comparison of the hydrological effectiveness was carried out and considered in terms of
relative reduction of a particular pollutant (main herbicides and fungicides used for cereals,
maize and sunflower within the basin) with the measure implementation.
Costs assessment - Economic modelling
Costs were assessed with a an aggregate-level Linear Programming modelling approach that
simulates the agricultural land use at the watershed level calculating the economic returns and
the costs that would result if particular measures were applied. Such model which is a widely
1
The objective of the Concert’eau project was to develop a Collaborative Technological Platform as a decision support based
first on the definition of scenarios of changes of agricultural practices by the actors, second on the economic, social and
environmental simulations and evaluations of the scenarios and finally on a comparison of the scenarios regarding their
economic, social and environmental efficiency assessment.
2
The Soil and Water Assessment Tool model is a watershed scale model developed to predict the impact of land
management practices on water, sediment, and agricultural chemical yields in watersheds with varying soils, land use, and
management conditions over long periods of time.
452
Session C3: Application of integrated assessment tools
accepted means of modelling large area (Norton & Schieffer, 1980) is developed in mix
integer linear programming using GAMS software (Rosenthal, 1998). The economic
assessment is achieved by modelling at the district level. Costs calculation is based on the
assumption that the levels of incentive linked with the measure needed to make it appear in
optimal modelled solutions represent the direct costs of its implementation. Transition from
farm to district/watershed level analysis is achieved by aggregating the resources of districts
and modelling the aggregated variables as a single large farm. Thus, only one standard crop
rotation (as with the hydrological modelling) is assigned to a district. For construction of the
technical coefficients matrix, technical data that consist of input and output flows were
provided from expertise on the farming systems in the area and from technical and
economical regional references on crops and livestock. Ratios between outputs and inputs
have been assumed constant (deterministic model) as well as their prices using their mean
value for the current year. Optimization runs for the aggregate-level watershed show the trade
offs and abatement cost curves illustrating the relationship between costs and implementation
of each measure for the different districts.
Results and discussion
Results of the environmental effectiveness calculations demonstrate that for the conditions
prevailing in the upper stream of the Gers river, Nitrogen and pesticides loss may in some
case be decreased effectively by implementing measures. For a given measure, costs of implementation may vary extremely between districts because of the distribution of crops and
farming systems. Similarly effectiveness varies greatly between Hydraulic Response Units
because of spatially variable catchment properties (topography and soils defining channel
processes) and agricultural activities (crops and management practices) related to farming
systems. For example, a longer crop rotation measure for districts with a maize as single crop
farming systems will be implemented at a much higher costs than districts with already more
diversified systems were costs are very low or null. Effectiveness of the measure will depend
of the changes of practices, crops and land cover differently according to the various locations
within the river basin. Uncertainty of models outcomes, by itself a strong concern, could be
partially overcome when C/E ratios are used in comparison and not with their absolute values.
A combined C/E approach could help to indicate the optimal spatial allocation of variants of
farm good management practices such that desired regional and sub-regional nitrate and pesticides concentrations are obtained at minimum regional costs. Furthermore, such measures
collectively defined by actors could be supposed to be more easily accepted by farmers.
Conclusion
The framework presented here could be useful for policies analysis to shed light on resource
allocation problem for river basin management. The policies are envisaged here in terms of
alternative allocations of resources, the objective being to find the changes of crops and
practices that will contribute most to goals achievement at the minimum costs. Such approach
could help water districts better target implementation of measures and financial
compensation. When choosing between measures, policies should have then to consider the
trade-off between costs and environmental effects, as highest environmental effects of
measures could entail larger costs too.
References
Norton, R.D. & G.W. Schieffer, 1980. Eur. Rev. Agric. Econ. 7: 229-265.
Rosenthal, R.E., 1998. GAMS user’s guide. GAMS Development Corporation, 1217 Potomac
Street, N.W., Washington, DC 20007, USA.
453
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Quantification of the impacts of Cross Compliance measures on
European air, soil and water quality
J.P. Lesschen, J. Kros, G.L. Velthof, D.A. Oudendag, M. Kempen, W. De Vries
Alterra, Soil Science Centre,
P.O. Box 47, 6700 AA Wageningen, The Netherlands
Contact: janpeter.lesschen@wur.nl
Introduction
The EU Cross Compliance instrument implies that farmers receive payments subject to
meeting Statutory Management Requirements (SMRs) related to environment, food safety,
animal and plant health and animal welfare, as well as standards of good agricultural and
environmental conditions (GAECs). Within the Cross Compliance Analysis Tool (CCAT)
project (http://www.ccat.nl/UK/) an analytical tool is being developed that enables the
integrated assessment of the impact of cross compliance at different geographical scales
ranging from the European (markets) to regional and farm levels. Impacts assessed by the tool
include effects on agricultural markets, producer’s income, land use, soil, water, air, climate,
biodiversity and landscapes, as well as food safety, animal welfare and health.
This paper describes the approach and application of the environmental impact tool at
European scale to evaluate cross compliance measures on environmental effect indicators.
The overall objective of the modelling framework is to assess the impact of cross compliance
measures on air, soil and water quality in terms of:
- Atmospheric emission of ammonia and greenhouse gases (air quality and climate);
- Accumulation or release of soil organic carbon (organic matter), phosphorous and heavy
metals balances (chemical soil quality);
- Nitrogen balance and leaching and runoff of nitrogen (water quality).
Methods
For the environmental impact assessment at the European level the MITERRA-Europe model
is used in combination with the agricultural sector model CAPRI (Heckelei & Britz, 2001).
The MITERRA-Europe model was developed to assess the effects and interactions of policies
and measures in agriculture on N losses on a regional level in the EU-27 countries (Velthof et
al., 2008). MITERRA-Europe is partly based on the existing models CAPRI and RAINS,
supplemented with a N leaching module and a measures module. It is a deterministic and
static N cycling model which calculates N emissions on an annual basis, using N emission
factors and N leaching fractions. The model has been extended with a module to calculate
changes in soil organic carbon stocks. The CAPRI model in combination with MITERRAEurope outputs (manure excretion and fertilizer use) predicts: (i) cost of the measures, (ii) the
changes in animal numbers and (iii) the change in crop shares, whereas MITERRA-Europe in
combination with CAPRI outputs predicts the environmental indicators (see Figure 1).
The environmental impact indicators for the atmosphere are: emissions of ammonia and
greenhouse gases N2O and CH4; for soil quality: change in soil organic carbon content in the
topsoil, the balances of N and P as well as balances for the metals Cd, Pb, Cu and Zn, and for
water quality the NO3 leachate fluxes and concentrations. This paper will focus on the
impacts of the cross compliance measures from the Nitrate Directive, which are balanced N
fertilizer application, maximum manure N application standard, limitation to N application in
winter and wet periods, limitation to N application on sloping grounds, manure storage with
minimum risk on leaching, appropriate application techniques, buffer zones and growing
winter crops.
454
Session C3: Application of integrated assessment tools
CAPRI
Livestock, land use
Base line 2005
CC measures
Excretion module
MITERRA
Excretion
C, N, P, Cd, Pb, Cu, Zn
housing module
MITERRA
manure/fertilizer
module MITERRA
Uptake module
MITERRA
input to soil
C, N, P, Cd, Pb, Cu, Zn
uptake
N, P, Cd, Pb, Cu, Zn
Soil module
MITERRA
housing emission
NH3, N2O, CH4
soil emission
NH3, N2O, CH4
N leaching,
runoff
Figure 1. Approach to predict air, soil and water quality indicators within CCAT.
Results and discussion
At the AgSAP conference the quantified impacts of Cross Compliance measures on European
air, soil and water quality will be presented. The results will focus on the application with
measures from the Nitrate directive. The impacts of cross compliance measures on air, soil
and water quality will be presented based on calculated cross compliance levels for the Nitrate
Directive for 2005 (Elbersen et al., 2009). In addition, the environmental impact will be
simulated for scenarios with different cross compliance levels.
References
Elbersen, B.S., et al., 2009. Compliance and costs of compliance: A procedure to obtain
estimates at a disaggregated level. Abstract for the Integrated Assessment of Agriculture
and Sustainable Development Conference.
Heckelei, T. & W. Britz, 2001. In: Eds Th. Heckelei et al., Agricultural Sector Modeling and
Policy Information Systems, Proceedings of the 65th European Seminar of the European
Association of Agricultural Economists (EAAE), March 2000, Bonn. Wissenschaftsverlag
Vauk Kiel KG, pp. 281-290.
Velthof, G.L., et al., 2008. Integrated assessment of nitrogen emissions from agriculture in
EU-27 using MITERRA-Europe. Journal of Environmental Quality. (in press)
455
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessment of flood prevention contribution of multi-functional
agricultural land: Results of a case study in Lower Austria
J. Neuwirth, K. Wagner, H. Janetschek
Marxergasse 2, 1030 Vienna, Austria
Contact: Julia.neuwirth@awi.bmlfuw.gv.at
Introduction
Agricultural land fulfils a broad range of functions, e.g., production of food, forage and raw
material, habitats for different species, space for recreation and protection against hazards.
During the last years the increase of extreme weather incidents as well as flood disasters
caused serious damages. In consequence this called forth increasing demands on agricultural
land in terms of flood prevention and its usage for retention purposes. An ongoing discussion
has started about types of land use adapted to flooding and related compensation payments.
Furthermore, the new Directive on the Assessment and Management of Floods together with
the existing Water Framework Directive stipulate flood risk management plans by 2015.
Since suitable basics in spatial planning with respect to agricultural land are missing so far,
this paper assesses the multi-functionality of agricultural lands extended by flood prevention
contribution and flood sensitivity.
Methodology
The basics for our analysis are the results derived from the ILUP project1, which has assessed
the multi-functionality of agricultural land (Wagner, 2007). In the project “Agriculture and
Flooding”2 we adapted the underlying methodology of the ILUP project to analyse the
contribution of agricultural land to flood prevention (Wagner et al., 2008). In the following
we overlapped the results of both projects with each other in an exemplary model region to
visualize the variety of society’s demands on agricultural lands.
In order to asses the multi-functionality the model community of “Seitenstetten” in Lower
Austria was subdivided into functional areas due to geomorphologic conditions and
homogeneity in agricultural land use. Six main functions of agricultural land were identified:
production, hazard protection, resource protection, diversity, recreation and spatial
structuring. An evaluation model was developed for each function. The results of each
function were transformed into a common scale from 1 (unimportant) to 5 (highly important)
to facilitate a comparison of the different functions. Information for the assessment of multifunctionality originates from the Austrian digital soil map, data of the Integrated
Administration and Control System (IACS), the Austrian hazard risk map, aerial pictures, the
statistical census and various Austrian basic maps.
In order to assess the flood prevention contribution and the flood sensitivity of agricultural
land indicators for natural properties, the flooding frequency and the type of agricultural land
use were defined considering the regional data availability. For the purpose of linking these
results with the results of the assessment of multi-functionality we also applied a scale from 1
(very low flood prevention and very low flood sensitivity, respectively) to 5 (very high flood
prevention and very high flood sensitivity, respectively). Required information stem from the
Austrian digital soil map, the digital flood risk map and from the IACS land-use data as well
as from data about available water capacity and erosion (Murer et al., 2004; Strauss, 2007).
An overlaying of all results in the model community “Seitenstetten” using GIS reflects the
broad range of functions fulfilled by agricultural land.
1
Interreg III B project “ILUP - Integrated Land Use Planning and River Basin Management”.
This is a sub-project of the project “Flood Risk II” by the Austrian Federal Ministry of Agriculture, Forestry,
Environment and Water Management.
2
456
Session C3: Application of integrated assessment tools
Results and discussion
The community was divided into three functional areas depending on geologic units and
predominant land use (see Figure 1). The agricultural land of the southern area (area 1) is
mostly characterized by grassland for agricultural production as well as resource protection.
Diversity and recreation are at medium level while flood prevention and flood sensitivity are
at lower levels. The production function dominates in the north-western area (area 2) with
resource protection, diversity, recreation and spatial structuring at medium level. The
contribution of agricultural land to flood prevention is low to middle, but flood sensitivity is
low. The north-eastern area (area 3) also has a very high production function and diversity,
recreation, hazard and resource protection show medium values. In this area the flood
prevention contribution is on a middle level and flood sensitivity is low to middle.
The results show that the management of agricultural land in the sample area is already
adapted to the regional flood risk: The agricultural usage of land with a higher flood
sensitivity targets more on flood prevention than in areas with flood sensitivities on the lowest
level. Values of flood sensitivity equal or higher than the respective values of flood
prevention would have given a hint that changes in land use are urgently needed.
The developed method is an appropriate tool for assessing, comparing and visualizing
various functions of agricultural land. An illustration of the multi-functionality of agricultural
land is useful for setting priorities and for the development of future management plans on a
regional scale. Future research of greater river catchment areas can provide indications of
functional relations among different functions of agricultural lands.
Results of area 3
Results of area 2
Overall assessment of agricultural lands
Scale of bars from 1 to 5 units
Results of area 1
Prod uction
Wood, bui ldi ng land
Hazard protection
Grassland
Resource protection
Arable land
Diversity
Risk crops on arable land
Recreation
Border of functional area
Spatial structuring
River
Flood prevention
Flood sensitivity
0
0,5
Kilometers
1
±
2
Figure 1. Results of the
assessment
of
different
functions in the community of
Seitenstetten.
References
Murer, E., et al., 2004. Die nutzbare Feldkapazität der mineralischen Böden der landwirtschaftlichen Nutzfläche Österreichs. Serial publication of the Austrian Federal Agency for
Water Management, Volume 20, Vienna, pp. 72-78.
Strauss, P., 2007. Flächenhafter Bodenabtrag durch Wasser, in Hydrological Atlas of Austria,
3. Edition, Map Board 8.4. Austrian Federal Ministry of Agriculture, Forestry, Environment and Water Management.
Wagner, K., 2007. Landwirtschaftliche Grundlagen für eine Integrierte Regionalplanung.
Serial publication of the Austrian Federal Institute of Agricultural Economics, Vienna.
Wagner, K., et al., 2008. Projektforschungsbericht Landwirtschaft und Hochwasser. Austrian
Federal Institute of Agricultural Economics, Vienna.
457
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Silvo-arable agro-forestry in Europe: Systems for meeting food and
industry crop requirements with improved resource-use efficiency
J. Palma, A.R. Graves, P.J. Burgess, W. Van der Werf, F. Herzog
FORCHANGE – Forest Ecosystems Management under Global Change,
Centro de Estudos Florestais, Instituto Superior de Agronomia,
Universidade Técnica de Lisboa, Tapada da Ajuda, 1349-017 Lisboa, Portugal
Contact: joaopalma@isa.ist.pt
Introduction
European agriculture is facing competing demands for food and bio-energy crops. Silvoarable agro-forestry (SAF) integrates trees and arable crops on the same area of land. An
important benefit of integrated tree and crop systems is that they can make more efficient use
of resources and produce more biomass, then when grown separately. However, performance
in Europe is variable and methods that combine environmental and economic indicators are
needed to evaluate in what circumstances, SAF offers a viable alternative.
Method
Data on nineteen Landscape Test Sites (LTS) were randomly selected in Spain, France and
The Netherlands and used with the YieldSAFE model (Van der Werf et al., 2007) to predict
the yield of integrated crops and trees in silvo-arable systems. Crops were grown in rotations
typical of each LTS and five tree species were selected including poplar (Populus spp), wild
cherry (Prunus avium), oak (Quercus ilex), pine (Pinus pinea) and walnut (Juglans hybr.).
Tree density was assumed to be 113 trees ha–1 and yields were predicted for 10% and 50% of
both the least and most productive portions of each LTS. The Land Equivalent Ratio (LER,
Eqn 1) was calculated for each tree and crop combination to assess biophysical benefits in
comparison with arable and forestry systems, and the infinite net present value was calculated
assuming no-payments, pre-2005 CAP payments, and post-2005 CAP payments, to assess
economic benefits (Graves et al., 2007).
LER =
Tree silvoarable yield
Crop silvoarable yield
+
Tree monoculture yield Crop monoculture yield
Eqn 1
Soil erosion, nitrogen leaching, carbon sequestration and landscape biodiversity were also
assessed (Palma et al., 2007a) and the environmental and economic data were then integrated
in a Multi Criteria Decision Analysis (MCDA) with the outranking method ‘promethee II’ to
assess the sustainability of each system (Palma et al., 2007b).
Results and discussion
The predicted LER were consistently greater than 1, indicating that growing crops and trees in
integrated SAF systems was more productive than separating them in arable and forestry
systems. The LERs formed a convex arc with maximum values obtained when the trees and
crops had similar relative yields, and minimum values obtained when the tree or crop
components were dominant (Figure 1A). The greater productivity of SAF is primarily due to
its capacity to capture more light, water, and nutrient resources per unit area than the
respective monoculture systems (Cannell et al., 1996). Thus, the higher LER for the SAF
systems modelled here, were found to be partly due to complementary capture of light by
trees and crops when they were combined on the same area of land (Graves et al., 2007).
The integrated environmental and economic analysis (Figure 1B) showed that the
performance rank of arable systems increased when CAP payments were included, whilst the
opposite was true for the SAF, indicating that CAP payments favour arable systems. Although
the post-2005 CAP reform corrects the payment distortion evident in the pre-2005 CAP
458
Session C3: Application of integrated assessment tools
(A)
(B)
Status Quo
1.4
(n=19)
All50%
LTS
SAF113 Best Land
Poplar
50% SAF113 Worst Land
Cherry
Oak
1.0
Pine
0.8
10% SAF113 Best Land
10% SAF113 Worst Land
4
3
Walnut
0.6
0.4
0.2
Performance
Relative tree yield
1.2
2
1
0
-1
-2
-3
0.0
-4
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
No payments
pre 2005
post 2005
Relative crop yield
Figure 1. (A) Land equivalent ratio of silvo-arable agro-forestry plot scale results for different
tree species in the system design. (B) Results from multi-criteria analysis showing
performance of economic and environmental indicators under different system designs and
payment scenarios in Spain, France and the Netherlands.
payment scenario, subsidies still favor conventional arable systems. Nevertheless, the
evaluation for all three countries showed that SAF is preferable to the ‘Status Quo’ under all
payment schemes, and that it would also be preferable to maximize use of SAF on the better
land within each LTS (Figure 1B: 50% SAF113 Best Land). With its capacity to increase
biomass productivity because of greater resource capture, SAF could be an important means
of addressing the increased demand for bio-energy crops in Europe, at the same time
delivering environmental benefits. National governments in Europe should, therefore, make
use of new EU regulations that allow support for the establishment of SAF systems
References
Cannell, M.G.R., et al., 1996. Agrofor. Syst. 34: 27-31.
Graves, A.R., et al., 2007. Ecological Engineering 29: 434-449.
Palma, J., et al., 2007a. Modelling Agriculture Ecosystems & Environment 119: 320-334.
Palma, J., et al., 2007b. Ecological Economics 63: 759-767.
Van der Werf, W., et al., 2007. Ecological Engineering 29: 419-433.
459
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated impact assessment of agro-environmental schemes on soil
erosion and water quality
M. Perugini, M. Toderi, G. Seddaiu, R. Orsini, G. De Sanctis, P.P. Roggero
Dep. of Environmental and Crop Sciences,
Polytechnic University of Marche, Italy
Contact: m.perugini@univpm.it
Introduction
The EU demands to the National authorities of the European members an ex-post evaluation
of Rural Development Plans (RDPs). The EU Commission has provided a set of indicators
and evaluation criteria to be adopted for the RDPs assessment (e.g., Docs. STAR
VI/43517/02, VI/12004/00, VI/8865/99). An interdisciplinary team of the Department of
Environmental and Crop Sciences of the Polytechnic University of Marche was mandated to
carry out a quantitative evaluation of agro-environmental measures of the Marche RDP 200006. The aim of the quantitative evaluation was to analyse the effects of the application of the
agro-environmental measures F1 (low-input) and F2 (organic farming) of the RDP on the
main biophysical processes which control the impact of cropping systems on water and soil
quality. Moreover this work should provide an assessment of different tools and
methodologies useful for a generalization of the results at regional level.
Methodology
The assessment was focussed on the analysis of some agro-environmental indicators on soil
erosion, nitrate pollution of runoff water and soil fertility. Previous research experiences in
the Marche Region showed the complexity of processes which control environmental impacts
and the inadequacy of EU indicators (e.g., nitrogen surplus) to explain and evaluate these
processes. Cropping systems, come from interviews to farmers of the whole area, and water
runoff were continuously monitored in the decade 1998-2008 in two micro-catchments (60–
80 ha) of the hilly rural area of Serra de’ Conti (Ancona) and in an irrigated low-land farm in
a NVZ (nitrate vulnerable zone).
In the two micro-catchments we quantified soil erosion and nitrate leaching in runoff and
in subsurface flow by an automatic sampler combined with a displacer meter at the close
section of micro-catchments. In the NZV case study only the nitrate leaching was monitored.
One of the two micro-catchments is characterized by a high diversification of the land use
with the presence of an organic farmer who grows legumes and perennials like vineyard. The
second one is more intensive with a biennial rotation between durum wheat and a spring crop.
Cropping system were also monitored by surveys, in particular crops, N inputs and outputs
were determined to calculate the nitrogen surplus, one of the key indicators for impact
assessment in the EU STAR documents.
The results of the monitoring systems were also used to calibrate and validate two
mathematical models, EUROSEM and DSSAT for soil erosion and N leaching simulation
under different cropping systems adopted within the RDP.
The agro-environmental schemes evaluation aimed also to assess the relationship between
cropping systems and plant and animal biodiversity through the application of an innovative
methodology based on the quantification of several bio-indicators such as carabides beetles
populations and maturity indexes of the vegetation. These aspects are not reported here.
In order to achieve a deeper analysis of the complex system of interest and to take into
account the social, political and human dimension of the studied environmental issues, a
system thinking methodology as Soft System Methodology (Checkland & Scholes, 1999) was
explored throughout the assessment.
460
Session C3: Application of integrated assessment tools
Results
The synthesis of the results of the impact assessment of the agro-environmental measures on
soil erosion and water quality are reported in Table 1.
Table 1. Main results of the impact assessment of the agro-environmental schemes on water
quality and soil erosion.
Environmental issue
Main results
Water nitrate
pollution
Soil erosion
- the more diversified was the catchment (in terms of number of crops) the
lower was the N surplus at the same environmental conditions
- organic farming leaded to a reduction of N surplus up to 200% compared
to the traditional farming systems at catchment scale
- in low-land intensive farming systems, low impact farming practices
leaded to have less than 10 kg ha–1 of N surplus
- according to model simulations, in low-land intensive farming systems,
organic farming could lead to higher N leaching risks than low impact
practices (+52%) during autumn due to the mineralization of the buried
crop residuals and simultaneous soil water surplus and bare soil
- the percentage of bare soil during sensitive periods is a key factor to
influence N leaching as well as crop diversification at catchment scale
- the more diversified was the catchment the lower was the soil erosion at
the same environmental conditions
- soil erosion could be reduced by –33%÷–21 in relation to crops spatial
distribution in the catchment
- soil erosion was not influenced by N fertilization reduction
- soil erosion occurred occasionally in relation to intense rainfall and
proportion of bare soil at catchment scale
- permanent crops could lead to a reduction of soil losses up to 90%
compared to a bare soil at catchment scale
Conclusions
The integrated impact assessment of the agro-environmental schemes of the Marche RDP
2000–06 showed that one of the key factors controlling soil erosion and water pollution by
nitrate is the degree of temporal and spatial diversification of cropping systems, particularly in
hilly areas and in complex systems such as organic farming. However, a better distribution of
same crops in time and space is still not taken into account in the more recent agroenvironmental schemes (e.g., RDP 2007–2013) and also by EU. In fact the incentives
promote the substitution by crops which are considered to have less impact on the
environment. Interventions favouring temporal and spatial diversification of cropping systems
would guarantee a more efficient use of funding resources and achievement of the schemes
objectives. Mathematical models application at field level and single crop as Marche Region
asked, showed that organic farming practices for durum wheat and corn within the agroenvironmental scheme would lead to a higher environmental impact than low impact
measures, since it did not consider the overall changes in farming practices, included changes
in crop rotations, related to the adoption of organic farming. The monitoring at catchment
scale allowed to take into consideration the changes associated to organic farming adoption
showing a general positive benefit of their application. Therefore, a stand-alone model-based
approach, could lead to a misinterpretation of the whole impact of agro-environmental
measures.
Reference
Checkland, P. & J. Scholes, 1999. Soft systems methodology in action. J. Wiley & Sons,
West Sussex, England.
461
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing the effects of Cross Compliance obligations
on European biodiversity and landscapes
J.J. Oñate, B.S. Elbersen, P. Martinez Alonso, B. Peco, J. Roos-Klein Lankhorst
Universidad Autónoma de Madrid, Ciudad Universitaria de Cantoblanco, 28049-Spain
Contact: juan.onate@uam.es
Introduction
Despite the increasing consideration of environmental issues in the successive reforms of the
European Common Agricultural Policy (CAP), the dual process of intensification and
abandonment driven by its economic incentives has continued to entail profound changes in
the European agro-ecosystems over the last decades. The Structural and Sustainable
Development Indicator ‘Farmland Birds’ (EC, 2005), continues to worsen (EEA, 2006) also
despite the spatial importance of those agricultural systems usually characterized as extensive
(EEA, 2004). In this paper we assess the effects on biodiversity and landscape that could be
expected from the most recently introduced CAP instrument targeting environmental
preservation, namely the Cross Compliance policy (hereafter CC).
CC was introduced as part of the 2003 CAP Reform (Regulation 1782/2003). It involves
Member States making receipt the Single Farm Payment (SFP) conditional on farmers
meeting two sets of obligations: (a) the Statutory Management Requirements (SMRs), related
to 19 pieces of EU environmental, public, animal and plant health and animal welfare
legislation (listed in Annex III of the Regulation); (b) the Good Agricultural and
Environmental Conditions (GAECs), related to the appropriate management of soils and the
minimum maintenance of agricultural land and its features (Annex IV). Since SMRs and
GAECs obligations apply to all farmers claiming the SFP and to the entire farm holding, CC
is likely to impact, to a greater or lesser degree, on most of the processes related to European
agriculture which are affecting biodiversity or landscape. This paper presents a methodology
to qualitatively assess these impacts and shows preliminary results of this assessment.
Methods
Past experience in scientifically assessing the impact of agriculture on biodiversity and
landscape has relied in the approach Before-After Control-Impact (BACI approach, e.g., Bro
et al., 2004). This consists in the comparison of trends in biodiversity on treatment fields and
control fields both before and after implementation of the considered treatment. This ex-post
approach has been applied to the impact assessment of agri-environmental schemes, yet on
the basis of pressure (e.g., Primdahl et al., 2003) or state (e.g., Kleijn et al., 2006) indicators.
However, the application of BACI approach to the assessment of CC effects is not feasible
at the moment because baseline data on the state of biodiversity at the field level before CC
implementation are rarely available and because there are not fields without treatment
available to perform the comparison, since CC obligations are compulsory for all farmers in
any given region. Alternatively, we have applied an ex-ante assessment approach through the
use of pressure indicators on the basis of literature analysis and expert knowledge (e.g., Llusia
& Oñate, 2005; Petit & Elbersen, 2006).
Our analysis focuses on those SMRs and GAECs targeting the preservation of landscapes
and biodiversity (Birds and Habitats Directives, and GAECs targeted on habitat/landscape
preservation, including, e.g., measures against soil erosion). The assessment is performed in
the following steps: (1) Collection of concrete CC obligations promulgated by Member States
or regions; (2) Identification and qualitative assessment of the potential effects of each
obligation; (3) Weighting of potential effects according to UAA; (4) Translating the potential
effectiveness to an expected effectiveness per region; (5) Aggregation of results to compute
the final expected effectiveness.
462
Session C3: Application of integrated assessment tools
For the identification of potential effects, we have considered both, direct effects on
species and/or habitats listed in the Birds and Habitat directives, and indirect effects on
biodiversity and landscape through changes in the structural and/or functional characteristics
of the agro-ecosystems. The potential effect of each obligation has been categorized in one
out of four semi quantitative classes (0, 1, 2, 3) and applied to the impact fields biodiversity
and landscape separately. Then the effect of each obligation has been weighted by the
regional UAA where it is to be implemented to obtain a potential effectiveness, under the
logic that the greater the targeted UAA, the higher the effectiveness will be. Both, share of
UAA and absolute hectares of UAA at NUTS2 level are considered as weighting factors,
reflecting respectively the magnitude and the extent of the potential effects.
To come from a potential effectiveness to an estimate of the expected effectiveness, the
potential effectiveness is multiplied by the level of compliance with the evaluated obligation,
collected from Member States. Different scenarios are considered, including the baseline
compliance situation in 2005, and 75% and 100% compliance.
The expected effectiveness is then expressed in aggregated terms for each impact field
(biodiversity or landscape), according to the expression:
i=n
ExpEffj = Σ ((PEi * UAAi) * CLi)
i=1
Where, ExpEff is expected effectiveness for impact field j, i is each considered obligation, PEi
is the potential effect for obligation i, UAAi is the share or ha of UAA targeted by obligation i,
and CLi is the compliance level reported for obligation i.
Subsequent aggregations of results are computed adding the expected effectiveness values
for different impact fields under each directive or GAEC groups, and finally in a total figure
of CC expected effectiveness per region and scenario.
Results and discussion
Expected effectiveness of CC obligations on the preservation of landscapes and biodiversity
varies strongly among regions in Europe. This is especially caused by differences in the way
national and regional authorities have defined SMRs and GAEC obligations in their
implementation of CC. Nevertheless, remarkable differences in effectiveness among SMRs
and GAECs related obligations appear even inside a single region. These differences are
discussed and recommendations are made.
References
Bro, E., et al., 2004. Journal of Applied Ecology 41: 846-857.
EC, 2005. Sustainable Development Indicators to monitor the implementation of the EU
Sustainable Development Strategy. SEC(2005) 161 final. Commission of the European
Communities. Brussels.
EEA, 2004. High nature value farmland. Characteristics, trends and policy challenges.
European Environment Agency, Office for Official Publications of the European
Communities. Luxembourg.
EEA, 2006. Species diversity (CSI 009) - Assessment DRAFT created May 2006. European
Environment Agency, Available at: http://themes.eea.europa.eu/indicators/
Kleijn, D., et al., 2006. Ecology Letters 9: 243-254.
Llusia, D. & J.J. Oñate, 2005. Ardeola 52: 31-42.
Petit, S. & B.S. Elbersen, 2006. Ambio 35: 297-303.
Primdahl, J., et al., 2003. Journal of Environmental Management 67: 129-138.
463
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated model of paddy rice production and
related environmental effects
Hiroki Sasaki
OECD Directorate for Trade and Agriculture
2, rue André-Pascal 75775 Paris Cedex 16, France
Contact: hiroki.sasaki@oecd.org
Introduction
Lankoski & Ollikainen (2003) have developed an analytical framework for analysing joint
production of commodity and non-commodity outputs as well as negative externalities under
heterogeneous land quality and this is the point of departure for this modelling analysis.
In the model, the optimal land use allocation and nitrogen application were considered
under a representative Japanese farm that consists of paddy field and upland production.
Methods
The model consists of a quadratic nitrogen response function for cultivated crops and
exponential nitrogen runoff (purification) function, damage functions from nitrogen runoff
and benefit functions from nitrogen purification. By assumption, there are 60 land quality
classes consisting of parcels of size of 10a each, so that the overall amount of the agricultural
land is 6 ha. The sums of consumer and producer surplus are maximized under exogenous
crop prices and input costs. The social welfare function for agriculture can be expressed as
SW = ∫
1
0
∑π
i
+ 6563z1 − 650 z2
for i = 1, 2
where, πi denotes farmer’s profit function, z denotes the nitrogen runoff (purification) and
social value of marginal benefit which is estimated on the basis of Shiratani et al. (2004).
Monetary value of nitrogen runoff and purification per kg are given by –650 JPY kg–1 and
+6563 JPY kg–1, respectively.
Results and discussion
Preliminary results reveal that under the private optimum fertilizer use is higher than under
the social optimum. Under the social optimum, more land is allocated to paddy fields due to
their positive (i.e. reduced negative) environmental externalities. Paddy fields even remove N
when the N application is below 11.2 kg per 10a, according to the estimated N runoff
functions. It is well known that paddy fields could function as an N purification type of land
use, but this depends on the types of agricultural activities and nitrogen concentrations of
irrigation waters.
Table 1. Input use and land allocation: comparing private and social optima.
Runoff damage/ Social
Nitrogen runoff Profit
Optimum
Parcels
purification
welfare SW/SO
(kg)
(000JPY)
(000JPY)
(000JPY)
Rice Wheat Rice
Private optimum 44 16
–1
Social optimum 60
0
–48
Wheat
153
0
464
1873
1823
–92
310
1781
2133
0.84
1.00
Session C3: Application of integrated assessment tools
In the next step, additional externalities, such as GHG emissions and sequestration and
biodiversity (measured by indices), will be incorporated into the model. Then the effects of
alternative agri-environmental policies will be analysed.
Japan is now undertaking market-oriented agricultural policy reforms and accelerating the
implementation of agri-environmental policies. The present farm level model could provide
valuable new insights into the design for mixed-policies.
50000
45000
40000
35000
30000
a0
25000
1
/Y
P
J
20000
Rice Profits
Wheat Profits
Rice Social Returns
Wheat Social Returns
15000
10000
5000
0
0
10
20
30
40
50
60
parcels
Figure 1. Private profits and social returns.
References
Lankoski, L. & M. Ollikainen, 2003. European Review of Agricultural Economics 30: 51-75.
Shiratani, E., et al., 2004. Journal of Japan Society on Water Environment 27(7): 491-494.
(in Japanese with English summary)
465
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Can regional policies mitigate the environmental effects of price raise?
A typical application of SEAMLESS framework at meso level in Auvergne
N. Turpin, J.P. Bousset, G. Bigot, E. Josien, E. Perret
Nadine Turpin, UMR 1273 Métafort, 24, Avenue des Landais, 63172 Aubiere Cedex, France
Contact: nadine.turpin@cemagref.fr
Introduction
Growing concern for the local competition for resource allocation between economic
development and natural resource restoration highlights the difficulties of coordinating
agricultural and environmental policies at EU, national and regional levels. More precisely,
local policy makers have to face the general evolution of economic forces (like prices
changes, increased competition between countries for labour and resources use), they have to
coordinate their action with EU wide policies, and operate through specific regional stakes.
This paper explores the ways a regional agro-environmental policy can improve the
sustainability of the farming systems in a region, in a context of general food price increase
and related environmental threats. The newly developed SEAMLESS integrated framework
(Van Ittersum et al., 2008) provides interesting features to address this multi-scale issue.
Methods
The region of the Mountains and Uplands of Auvergne has been selected because of its specific
livestock production, essentially milk production for typical cheeses and grass calves
production, produced with labels like Protected Designation of Origin. The main stakes for this
region are the preservation of the high level of biodiversity, the natural landscapes and the
water quality with a stabilization of the rural population.
We assume in this paper that the level of increase for agricultural products prices and for
production costs modifies the relative profitability of agricultural production over Europe.
Following Bigot et al. (2008), the expected effects in France are a shift of breeding activities
towards the western part of the country, where breeding activities are highly competitive, and
towards mountainous areas where it is possible to supply high quality cheese and meat (and
where crops become less and less profitable because of climate and mechanization difficulties).
FADN Auvergne
354 exploitations
SMALL
8 exploitations
MEDIUM (2…)
133 exploitations
LOW (21..)
41 exploitations
LARGE (3…)
213 exploitations
MEDIUM (22..)
92 exploitations
HIGH
Aucune exploitation
LOW (31..)
44 exploitations
ARABLE SYSTEM
1 exploitation
ARABLE SYSTEM
5 exploitations
ARABLE SYSTEM
7 exploitations
DAIRY CATTLE
2 exploitations
DAIRY CATTLE
56 exploitations
DAIRY CATTLE
7 exploitations
BEEF & MIXED CATTLE
29 exploitations
With
PERMANENT GRASS (2109)
29 exploitations
With
PERMANENT GRASS (2205)
56 exploitations
MIXED LIVESTOCK
1 exploitation
HIGH (33..)
8 exploitations
ARABLE SYSTEM
32 exploitations
With CEREAL (3201)
21 exploitations
BEEF & MIXED CATTLE
31 exploitations
With FALLOW
1 exploitation
With
PERMANENT GRASS (3109)
31 exploitations
BEEF & MIXED CATTLE
26 exploitations
With
PERMANENT GRASS (2209)
26 exploitations
SHEEP & GOATS
8 exploitations
MEDIUM (32..)
161 exploitations
SHEEP & GOATS
3 exploitations
With SEPCIALISED CROPS
6 exploitations
SHEEP & GOATS
2 exploitations
MIXED LIVESTOCK
3 exploitations
With OTHERS CROPS
4 exploitations
DAIRY CATTLE
51 exploitations
With
PERMANENT GRASS (3205)
51 exploitations
MIXED LIVESTOCK
1 exploitation
HORTICULTURE
1 exploitation
BEEF & MIXED CATTLE
36 exploitations
With
PERMANENT GRASS (3209)
36 exploitations
SHHEP & GOATS
12 exploitations
Figure 1. Modelled farm types (shaded types) inside the
taxonomy of all farms in Auvergne.
466
MIXED FARMS
12 exploitations
MIXED LIVESTOCK (3219)
18 exploitations
Session C3: Application of integrated assessment tools
In mountainous areas, increasing the breeding activities may hamper the environment quality,
so we explored the possibility of designing agro-environmental schemes that enable breeding
intensification with respect to the environment quality. The policy option takes the form of a
regional premium targeted to farmers who limit their inputs, such as fertilizers and fuel.
So far, the farms in Auvergne are represented using eight farm types (see Figure 1),
including one arable farm and seven breeding farms (dairy and beef farms that differ in their
intensity).With the Farm System Simulator (FSSIM, see Louhichi et al. (2007) for a complete
description of the model), we compared several levels of price increase, of production costs
increase and of premium levels, associated with various levels of constraints for the farm, and
we analyse which FADN farm types commit themselves to the agro-environmental scheme.
The simulations are formulated on the period 2003–2013 with the application of the 2003
CAP reform.
Results and discussion
The results highlight that the modelled farms commitment to the agro-environmental scheme
is strongly dependant on the level of farm production prices, the relative level of production
costs, and of course the premium level (see Figure 2 for the adoption scheme for the arable
farm type).
N leaching (kg N/ha)
minimal premium level for
adoption
40
50
38
40
35
30
33
20
0
20 regional premium
(euros/ha)
35
0
50
100
150
200
250
300
350
400
400
50
450
300
200
250
100
wheat price change (%)
350
0
150
0
wheat price change (%)
50
30
10
450
Figure 2. Wheat price change and minimal
level of the regional premium for the arable
farm type to commit
Figure 3. Consequences of the arable
farm type commitment on N leaching.
More important, the farm types that adopt the regional agro-environmental scheme (for a
given premium) are not the same depending on the relative increase of prices for products and
production costs. The environmental effects are thus very sensitive to these assumptions, and
the modelling framework enables analysing the trade-off between farm income and
environment effects of the policy options. Last, the modelling outputs can help policy makers
to design efficient policies as soon as they rely on accurate spatial farm typologies.
References
Bigot, G., et al., 2008 A typical application of SEAMLESS-IF at Meso level: the Green
Intensification scenario applied in Auvergne, SEAMLESS integrated project, EU 6th
Framework Programme, contract no. 010036-2, www.SEAMLESS-IP.org, 34 pp.
Louhichi, K., et al., 2007 A generic Template for FSSIM for all farming systems, PD3.3.11,
SEAMLESS integrated project, EU 6th Framework programme, contract no. 010036-2,
www.SEAMLESS-IP.org, 82 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
467
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme C
Case studies and application of tools and
empirical methods
Session C4: Systems analysis informing policy:
Case studies from agricultural systems and rural regions
in the face of change
Session organizers:
Brian Keating
CSIRO, Sustainable Ecosystems, Brisbane, Australia
Pablo Tittonell
CIRAD, Montpellier,France
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Systems analysis informing policy1
David Pannell
School of Agricultural & Resource Economics, Univ. Western Australia,
Crawley 6009, Australia
Contact: David.Pannell@uwa.edu.au
Good policy needs good science, particularly good systems analysis. In practice, science is
often lacking, and systems analysis can be particularly lacking. For example, in agrienvironmental policy, the science used often relates to the level and timing of environmental
threats, but much less often is there a systems analysis conducted to assist with understanding
of the impacts of alternative agricultural management practices on desired policy outcomes.
A large number of reasons have been identified why it can be hard for science (in general, not
just systems analysis) to connect with Policy, including the following:
• Non-scientific considerations matter, such as legal mandates, societal desires, economic
benefits and costs, rights, distributional equity and procedural fairness.
• There may be political or bureaucratic objectives unrelated to the public interest, so that
research that seeks to advance the public interest is not wanted.
• Policy attention tends to be directed to certain issues with high currency, especially crises,
and this may leave little scope for research in other areas to influence policy at that time.
• “The capacity of science to provide information may require more time than policymakers
are willing to accept, especially for politically hot issues” (Clark et al., 1998).
• Policymakers often rely primarily on locally based and trusted experts with whom they are
familiar.
• There may be suspicion about motivations of scientists, so that they are treated as just
another interest or lobby group.
• “University reward systems rarely recognize inter-disciplinary work, outreach efforts, and
publications outside of academic journals” (Jacobs, 2002).
• Research findings may be communicated in ways that policymakers cannot understand,
using jargon, technical language, or mathematics.
• In some situations there is rapid turnover or movement of staff in government policy
agencies, leading to lack of expertise by responsible staff.
Systems analysis can contribute to overcoming a number of these barriers to uptake. For
example, the following common features of systems analysis are valuable in this regard:
• Taking a solution-oriented approach.
• Working with intended users. This will help to ensure that the solution being proposed is in
fact practical and sufficiently simple and can a researcher’s credibility with policymakers.
• Distinguishing between knowledge and values.
• Working as a multidisciplinary team. A team that includes ‘integrators’ can span the
disciplines and draw them together in a way that is relevant to policy.
On the other hand, there are some requirements of good policy engagement that are not
necessarily met by systems analysts, and to which they need to pay particular attention:
• Understanding the policymaker’s perspective.
• Ensure communication is brief, with clear messages, using simple language, free of jargon.
• Developing relationships with policymakers to establish mutual understanding and trust.
• Keeping communications and the solutions that one offers as simple as possible.
1
Keynote presentation
470
Session C4: Systems analysis informing policy: Case studies
•
•
•
•
•
•
Being pragmatic and accepting compromise.
Being patient and persistent.
Being resilient. Numerous problems, frustration and setbacks will arise.
Being prepared to respond quickly to requests for information.
Identifying and cultivating a champion for your work within the policy organization.
Avoiding any appearance of vested interest.
Change is a constant reality in the world of agri-environmental policy. Economic, social,
political and physical conditions change over time, and policy must adapt. Systems analysis is
a particularly valuable tool under conditions of change, providing a means to integrate
existing knowledge and predicted changes in a problem-solving framework that can assist
policymakers to consider the changes properly. On the other hand, the time frame for policy
development is often short, and sometimes is extremely short, providing little opportunity for
new analyses to be conducted to address current pressing policy issues. This highlights the
need for analysts to be ahead of the game if possible, or at least well tuned into current policy
concerns and ready to respond rapidly.
An Australian case study is outlined, where systems analysis is used as a core element of a
strategy to attempt to influence agri-environmental policy. Since 1990, the main
environmental programs for rural Australia have been very poorly informed by science, and
by systems science in particular. Since 2000 the author has been attempting to alter this. The
efforts have involved various forms of systems analysis, from complex to simple, with the
simple tools proving to be particularly important for policy engagement. For example,
detailed farm-level bio-economic modelling was used to investigate the economic
attractiveness to farmers of a range of farming practices needed to improve management of
dryland salinity. A catchment scale model was used to estimate trade-offs between river
salinity, water yield, and farm economics. Simple graphical tools were developed to assist
policymakers to understand the benefits of salinity mitigation in different circumstances, and
to assist with the selection of appropriate policy tools.
The author had intensive involvement with both environmental managers and with
policymakers at national, state and local levels, and attempted to put into place all of the
strategies described above. The success of these efforts has been encouraging, and is still
growing. These experiences will be described and lessons highlighted.
References
Clark, R.N., et al., 1998. Integrating science and policy in natural resource management:
Lessons and opportunities from North America. Gen. Tech. Rep. PNW-GTR-441.
Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research
Station, 22 pp.
Jacobs, K., 2002. Connecting science, policy and decision-making: A handbook for
researchers and science agencies. NOAA Office of Global Programs, Boulder, Colorado,
25 pp.
471
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing poverty-sustainability trade-offs in semi-subsistence agricultural
systems: A case study of policy options in Machakos, Kenya
J.M. Antle, J.J. Stoorvogel, R.O. Valdivia
Montana State University, Bozeman, MT 59717 USA
Contact: jantle@montana.edu
Introduction
There is a growing demand for quantitative policy analysis that can address environmentpoverty trade-offs in the poorest regions of the world (World Bank, 2008). This paper uses an
econometric-process simulation model (Antle & Capalbo, 2001) with the Trade-off Analysis
software (Stoorvogel et al., 2004) to implement spatially-explicit integrated assessment
models of semi-subsistence agricultural systems. We use this model to investigate the
potential impacts of technology and policy options being considered by the Government of
Kenya to address poverty and sustainability in the agricultural systems of Machakos, Kenya.
Methods
Modelling semi-subsistence agricultural systems (SSAS) poses special challenges that have
not been addressed adequately in the recent literature. SSAS typically involve complex
patterns of inter-cropping, exhibit high rates of crop failure, use non-essential inputs such as
mineral fertilizer. Often SSAS involve dynamically interacting crop and livestock subsystems, as crop residues are fed to livestock and manure is used to recycle organic matter
and nutrients back to crops. To be able to analyse impacts on poverty, a whole-farm approach
is needed that integrates crops and livestock and also accounts for off-farm income.
In this paper, we utilize a spatially-explicit econometric-process simulation model to
implement policy analysis of SSAS. The case study utilizes data collected by the Nutrient
Monitoring (NUTMON) project (De Jager et al., 1998). In related work, procedures were
developed to extract data from the NUTMON system (Vallejo et al., 2009) so that they could
be utilized with the Trade-off Analysis (TOA) modelling system. The TOA software is used
to simulate the system by integrating spatially-referenced data with site-specific biophysical
models (crop growth, nutrient balance) and economic models.
Results and discussion
We consider technology policy interventions identified by the NUTMON research team’s
interactions with stakeholders ranging from farmers to government officials, as well as the
Government of Kenya’s official policy documents. The policy options include enhancing
prices of key cash crops, such as maize and vegetables, making mineral fertilizer available to
all farmers at lower cost (e.g., by reducing import impediments and improving transportation
and marketing infrastructure, credit availability, etc.), and increasing organic nutrient intensity
and per capita incomes through policies that increase farm size (e.g., by promoting rural
development and off-farm employment opportunities).
An illustrative result in presented in Figure 1. These curves show the relationship in the
Machakos region between the poverty rate and soil nitrogen depletion (kg N season–1,
estimated by the NUTMON model) constructed by varying the maize price above and below
its base value (the mid-point represents the base price), for the base scenario, for a fertilizer
use scenario, for a rural development scenario, and for the two scenarios combined. In the
fertilizer scenario, fertilizer prices are reduced by 50% and all farmers use a positive quantity
of fertilizer as determined by their fertilizer demand functions (as compared to only 30%
using a positive amount in the base population). The rural development scenario assumes that
household size is reduced by 25% and that average farm size is doubled. The results show a
trade-off between poverty and sustainability (except at high maize prices, where nutrient
472
Session C4: Systems analysis informing policy: Case studies
applications are high enough to offset export of nutrients through grain harvested). The
fertilizer use and rural development scenarios have substantially different impacts on poverty
and nutrient depletion. The two policies combined clearly have the greatest impact on both
dimensions.
Figure 1. Relationship in the Machakos region between the poverty rate and soil nitrogen
depletion (kg N season–1, estimated by the NUTMON model).
Discussion
This paper uses a spatially-explicit integrated assessment model to investigate the impacts of
technology and policy options on poverty-sustainability trade-offs in Machakos, Kenya.
Results suggest substantially different impacts of fertilizer use and rural development policy
interventions on poverty and sustainability, and indicates that a multi-faceted development
policy would be most effective at reducing poverty and enhancing sustainability. Analysis
also suggests a positive, win-win outcome (lower poverty and nutrient depletion) would be
possible with policies that would improve vegetable prices and increase investment in
irrigation for vegetable production.
References
Antle, J.M. & S.M. Capalbo, 2001. American Jour. of Agric. Economics 83(2): 389-401.
De Jager, A., et al., 1998. Agriculture, Ecosystems and Environment 71(1-3): 37-48.
Stoorvogel, J.J., et al., 2004. Agricultural Systems 80(1): 43-66.
Vallejo, A.M., et al., 2009. Agricultural, Ecosystems and Environment. (revision submitted)
473
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Performance analysis on the changes of farmland property right system:
An explanation of the diversification of Chinese farmland system in
transition
Zhigang Chen, Futian Qu, Xianjin Huang
Department of Land Resources and Tourism Sciences, Nanjing University, Hankou Road 22,
Nanjing 210093, P.R. China
Contact: zgchen@nju.edu.cn
Introduction
Since the household contracted responsibility system was carried out across China, various
innovations in farmland tenure system aiming at improving the household contracted
responsibility have appeared in different areas of China. Generally speaking, ‘the two-field
system’ is carried out in most areas, the scale management and stock cooperation system are
implemented in coastal developed areas, the auction of the use right of ‘four wastes’
(including barren hills, barren hillsides, barren hillocks and desolated beaches) is put into
practice in the less developed north-western and south-western China regions which have vast
land and small population (Zhang, 2001). These innovative forms of the farmland system with
different emphases on the rights of the farmland have made great achievements in their
respective representative regions. However, they do not have generality in China whose
regional differences are so great, but often they are the institutional arrangements in a specific
period and region. Why do these systems have eminent performance in one area but little
performance or even negative effects in another area. This paper tries to explain the diversity
of the farmland tenure systems through the performance analysis on the changes of farmland
property right system.
Methods
According to theories of neo-institutional economics, the emergence of an institution is
inevitably in accordance with people’s demand for the institution which will produce the
corresponding performance (North, 1990). Therefore, a theoretical hypothesis is made that in
different areas with different economic development, the demands of the people for the use
right, transfer right and benefit right of the farmland are different and this difference is the
main reason for the innovation of the farmland tenure system with different emphases on the
rights of the farmland.
In order to test the above theoretical hypothesis, an econometric analysis on changes of the
farmland system and farmland performance is made. The basic model chosen to conduct the
empirical analysis is Cobb-Douglas production function with variables representing the
farmland property right. The econometric model can be written as follows:
lnY=C+a1lnL+a2lnF+a3lnM+a4USE+a5TRAN+a6TAX+ei
where, Y means the output of farmland, L, F and M mean the labour force and capital input
(including fertilizer and agricultural machinery), respectively. The use right (USE) is reflected
by the left time limit of the land use right, which can represent the peasants’ confidence in the
reliability of the use right. The transfer right (TRAN) adopts dummy variables and the value is
1 from 1995 and the previous years are all 0. Benefit right can be measured by the burden of
the agricultural tax (TAX). C is the constant term, and ei is the random disturbance term.
The data for model estimation are the panel data of several provinces of China in recent 22
years. Therefore, this paper estimates the above econometric model of the areas with different
economic development respectively using the fixed effects model and the random effects
model of the panel data.
474
Session C4: Systems analysis informing policy: Case studies
Results and discussion
The estimated results show that the fixed effects model is better than the random effects
model. Table 1 lists a part of estimated results of the fixed effects model. It is concluded that
the peasants in different areas have different demands with regard to the use right, the transfer
right and the benefit right of the farmland. It is just this different demand that the
improvement of various rights in the changing process of the farmland system has different
degree of influences on the performance of the farmland. Specifically, no matter developed
areas or underdeveloped areas, the improvement of the use right of the farmland is very
important to the peasants; the improvement of the benefit right of the peasants is more in
agreement with the will of the peasants in the less developed areas; the evolution of the
transfer right has a more significant influence on the performance of the farmland in the
developed areas. So the research conclusions of this paper to some extent give a reasonable
explanation to the diversification of the innovation of Chinese farmland system.
Table 1. The estimated results of the fixed effects model of different areas.
Explanatory Expected Developed areas
Less-developed areas
variable
symbol
coef.
standardized coef.
coef.
standardized coef.
USE
+
0.001
0.073
0.002
0.070
(0.002)
(0.002)
TRAN
+
0.074
0.198
–0.019
–0.029
(0.047)
(0.043)
TAX
0.009
0.036
–0.084
–0.115
(0.039)
(0.047)
R2
0.557
0.750
F value
18.16
28.43
Note: the value in the brackets is the standard deviation of the corresponding coefficient.
It is worth mentioning that the research conclusions of the present paper also provide some
policy implications for the reform of Chinese farmland system in the future. China has a vast
territory and the regional differences are great, so the reform of the farmland system must
consider the real situation of different areas and proceeds emphatically and hierarchically. It is
assumed that the reform scheme with optimal cost-income ratio should be arranged according
to different needs of the peasants.
References
North, D., 1990. Institutions, Institutional Change and Economic Performance. Cambridge
University Press, UK.
Zhang, H.Y., 2001. The political economy of land tenure system reform in China. PhD thesis
of Southwest Agricultural University of China. (in Chinese)
475
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Assessing the sustainability of Irish agriculture
E.J. Dillon, T. Hennessy, S. Hynes
Rural Economy Research Centre, Teagasc, Athenry, Co. Galway, Ireland
Contact: Emma.Dillon@teagasc.ie
Introduction
Indicators encompassing the multi-dimensional nature of sustainability (economic,
environmental and social) are developed here using National Farm Survey (NFS) data over a
ten-year period (1996–2006) to assess the overall sustainability of Irish agriculture. This is the
first such study undertaken for Ireland and the results show great change over the decade. The
effect of a number of policy scenarios on the sustainability of Irish farming is then examined,
in particular the possible impact of milk quota expansion on agricultural sustainability.
Economic viability was found to be generally in decline over the ten-year period, however,
when individual farming systems were taken into account, some were found to perform better
than others. Unsurprisingly the more intensive farming systems (primarily dairy) were found
to pollute more on average, and a case is then made for the potential trading of emissions
permits across farms. Irish agriculture is experiencing a period of fundamental change, not
least in terms of the ever-changing rural demographic; the challenge therefore lies in ensuring
that farms remain economically, environmentally and socially sustainable in the long-run.
Methods
Indicators of Irish agricultural sustainability are designed and outlined here using Irish NFS
data (1996–2006), which is collected as part of the FADN dataset. It is a random sample of
1,200 farms representing approximately 115,000 farms nationally. Within the NFS, the farm
system variable is broken down into six different categories as follows: Dairying, Dairying
and Other, Cattle rearing, Cattle Other, Mainly Sheep and Tillage Systems (Connolly, 2007).
The system titles refer to the dominant enterprise in each group (another enterprise could be
present on the farm also). Indicators calculated here give a benchmark measure of agricultural
sustainability in the country. Indicators were chosen according to their overall suitability
(based on the general literature) and the availability of data. A number of indicators are
illustrated in Table 1 below with information on how some of these are calculated and how
they change over the period discussed in more detail in the following section.
Table 1. Sustainability indicators.
Sustainability areas
Economic
Environmental
Social
air quality
water
quality
Indicators
Viability
Direct payments as a % of
Gross Output
Market Return
Methane emissions
Organic nitrogen
Organic phosphorous
Demographic viability
Isolation
476
Measurement units
%
%
€/farm
kg per farm
kg per ha
%
%
Session C4: Systems analysis informing policy: Case studies
Results and discussion
In terms of economic indicators, the economic viability of farms is a pertinent issue. Based on
the work of Hennessy (2004) and Frawley & Commins (1996), an economically viable farm
is defined as having (a) the capacity to remunerate family labour at the average agricultural
wage, and (b) the capacity to provide an additional 5% return on non-land assets. In the
absence of an average Irish agricultural wage, the minimum wage for agricultural workers as
set by the Labour Court annually is used here. A viability threshold is then calculable with
those farms rising above this value deemed economically viable and those falling below it
unviable. A poor degree of viability across all systems is reported upon in general between
1996 and 2006 with between 28% and 41% of farms only, classified as economically viable
over the ten-year period. On the whole, it can be seen throughout the period that the dairying
and tillage systems tend to have a relatively higher proportion of viable farms compared to
other systems. Direct payments as a proportion of gross output can be thought of as another
important economic indicator. There is no threshold value as such here; however the relative
change over the ten-year period examined is of interest. The influence of direct payments is
seen to be greater in 2006 than in 1996 for all systems. Such payments are evidently of huge
significance to Irish farmers and therefore any future reform should prove central to the future
sustainability of Irish farming.
In terms of the environmental indicators chosen, methane emissions are the most important
pollutant arising from Irish agriculture and can, therefore, be thought of as a valuable
indicator of environmental sustainability. Based on livestock emission factors, methane
emissions (kg ha–1) can be calculated (see O’Mara et al., 2007). Emission factors represent
the quantity of gas produced by an animal over a specific period of time and by multiplying
emissions factors for each animal type by the size of the herd, total emissions from a
particular source category can be generated for each farm in the NFS. Methane emissions
were found to be, as expected, much higher for dairying than for other systems. A case could
then be made then for the future trading of emissions permits across systems in order to
reduce greenhouse gas emissions and thus improve the environmental sustainability of Irish
farms in general going forward.
Undoubtedly, the task of choosing the most relevant indicators of social sustainability
using variables available in the NFS proved most difficult. Taking into account the percentage
of farm households which have at least one household member below 45 years of age (i.e.,
those defined as demographically viable), a slight decline is found over the ten-year period
examined here, across all systems. This indicator can perhaps be thought of as an indicator of
succession (with the likelihood of someone taking over the farm worsening slightly over the
period). Those at risk of isolation (i.e., one-person farm households, many of whom are
elderly) can be thought of as another indicator of social sustainability. There was relatively
little change in this indicator over the period 1996–2006.
Finally in terms of future agricultural sustainability in an Irish context, the possible impact
of milk quota expansion on the sustainability of Irish farming is examined.
References
Connolly, L., 2007. Overview of the changing structure of Irish agriculture monitored by the
National Farm Survey. Presentation made to the Agricultural Economics Society of Ireland
AGM, Rural Economy Research Centre, 30th October 2007, Teagasc, Athenry.
Frawley, J.P. & P. Commins, 1996. The Changing Structure of Irish Farming, Trends
Prospects. Rural Economy Research Series No. 1, Teagasc, Dublin.
Hennessy, T.C., 2004. Projecting Farm Numbers. Report of the Agri Vision 2015 Committee.
Dept of Agriculture and Food.
O’Mara, F., et al., 2007. Climate Change – Estimation of Emissions of Greenhouse Gases
from Agriculture and Strategies for their Reduction. EPA Report.
477
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Ex-ante assessment of the abolishment of the EU set aside policy:
Results from a bio-economic farm analysis
A. Kanellopoulos, H. Hengsdijk, S. Janssen, M.K. Van Ittersum
Wageningen University, P.O. Box 8130, 6706 KN Wageningen, The Netherlands
Contact: Argyris.Kanellopoulus@wur.nl
Introduction
The set-aside policy was introduced by the European Union (EU) in 1988 to (i) reduce the
large and costly cereal surpluses produced under the guaranteed price system of the Common
Agricultural Policy (CAP), and (ii) to provide environmental benefits following considerable
damage to agro-ecosystems and nature as a result of the intensification of agriculture.
Although the implementation of the set aside policy differs across the EU, in general the
measure entails the obligation for leaving a proportion of farm land uncultivated or put to
non-food purposes for a certain period in exchange for subsidy payments. The area under setaside in the EU totaled 3.8 million ha in 2007.
Since the tightening of the cereal markets and associated increase in global cereal prices in
2007, the EU decided to abolish temporarily the obligation for set aside. There is strong
political pressure to permanently abandon the set aside measure in the future. This could have
major implications for arable farming systems in the EU.
The objective of this study is to evaluate ex-ante the possible consequences of abandoning
the set aside policy on farm income, cropping pattern and environment. For this purpose we
use the Farm System Simulator (FSSIM), which is a generic bio-economic farm model to
simulate the behaviour of farm types across the EU25 and to enable the ex-ante evaluation of
a broad range of policy measures. FSSIM has been developed within the 6th framework EU
project ‘System for Environmental and Agricultural Modelling; Linking European Science
and Society’ (SEAMLESS) (Van Ittersum et al., 2008).
Methods
FSSIM maximizes the farmer’s utility, defined as the total gross margin minus risk, subject to
a set of resource and policy constraints (Louhichi et al., 2007). The model is calibrated using
Positive Mathematical Programming (PMP) which guarantees exact reproduction of activity
levels observed in the base year (Howitt, 1995).
The large number and wide variety of farming systems in the EU25 do not allow for
application of FSSIM to individual farms. Therefore, a farm typology has been developed
combining socio-economic and biophysical farm characteristics (Andersen et al., 2007). In
total 40 different arable farm types were identified for 11 regions across the EU25. The model
has been calibrated using data from 2003 (base year) and is applied to explore possible effects
in the year 2013 using expected prices estimated with the Common Agricultural Policy
Regional Impact Analysis model (Britz et al., 2007). In addition, the levels of area subsidies
in 2003 were recalculated according to the CAP reform. In the base year, the observed set
aside area is used as a constraint in FSSIM, and farmers receive a region-specific subsidy. In
the exploration, this constraint is relaxed and the set aside subsidy is set to 0 € ha–1.
Data from the Farm Accounting Data Network (FADN) are used to quantify available farm
resources while production activities (crop rotations) and associated technical coefficients are
characterized through a dedicated survey (Borkowski et al., 2007).
Results and discussion
The simulated economic and environmental indicators and the cropping patterns of six farm
types (each with a set aside area greater than 12.5% of the farm area) in four EU regions are
presented in Table 1.
478
Session C4: Systems analysis informing policy: Case studies
Table 1. Simulation results of FSSIM for 2003 and 2013 of 6 farm types in four regions.
Brandenburg
Denmark
Size of farm type*:
Large
Medium
Large
Economic indicators
2003 2013 2003 2013 2003 2013
Utility (1000€)
232.9 219.0 31.2 41.0 125.2 133.7
Income (1000€)
234.3 222.6 32.0 42.3 125.2 133.7
Subsidy share (%)
50
23 32
7
24
9
Environmental indicators
Water use (m)
N-use (kg N/ha)
92
112 116 158 127 145
Cropping pattern (aggregated crop groups)
Barley (ha)
35.4 24.4 13.1 0.6 26.7 17.7
Grass (ha)
- 1.2 13.2 3.8 17.5
Maize (ha)
44.4 59.3 2.7 11.8 5.6 11.1
Oil seed (ha)
68.9 193.5 3.2 9.4 8.0 21.5
Other cereals (ha)
106.1 87.5 3.5 1.0 6.2
5.6
Protein crops (ha)
36.5 24.4 0.9
- 6.2
4.5
Root crops (ha)
1.0
0.4 0.3 0.1 4.5
0.6
Set aside (ha)
116.8 27.5 7.1 3.6 18.1 13.7
Wheat (ha)
52.7 44.6 7.9 0.3 41.1 28.2
Total farm size (ha)
462
462 40 40 120 120
*
Castilla y Leon
Midi-Pyrénées
Medium
Large
Large
2003 2013 2003 2013
2003
2013
53.1 52.8 135.2 128.3
72.1
88.2
53.6 53.1 140.5 131.7
73.7
91.6
13 14
32
18
44
29
8
65
5
52
4
56
5
43
3
117
3
104
2.0 2.7 6.8 25.8
0.5 1.0 5.1 5.6
4.0 0.6 1.2 0.8
12.1 12.5 71.3 75.1
10.0 11.9 79.1 56.1
28.6 29 163 163
1.6
25.6
17.7
3.6
18.9
23.7
91
29.7
37.1
0.9
15.6
7.8
91
Large farm: >40 ESU, Medium farm: 16 ESU ≤ size ≤ 40 ESU, Small farm:<16 ESU.
Abolishment of the set aside policy results in a reduction of the set aside area in most farm
types. Nevertheless, in none of the regions all set-aside land is put into production. The main
reasons are the limited management and machinery capacity and the risk averse attitude of the
farmer. In Castilla y Leon, the set-aside area even increases slightly because of the 40%
decrease in sugar beet price and subsidy reduction of irrigated soft wheat. The areas of both
crops are reduced substantially and the excess land is allocated to sunflower and to set-aside.
Except for the medium farm type in Castilla y Leon, there is a shift from cereal crops to oil
seed crops (sunflower, rape seed) and maize. The main reason for this change is the large
reduction of subsidies which reduces the marginal gross margin of cereals in year 2013
despite the expected increase of their market prices. The share of subsidies in farm income
reduces substantially for most farm types. In the medium farm type of Denmark and in the
large farm type of Midi-Pyrenees, total farm income increases with 32 and 24%, respectively.
For the other farm types, income remains the same or slightly decreases i.e. the maximum
income decrease is 6%.
Average nitrogen use increases in Brandenburg and Denmark and decreases in Castilla y
Leon and Midi-Pyrénées. This is associated with the type of oil seed crop grown, i.e. rape
seed in Brandenburg and Denmark and the more nitrogen demanding sunflower in Castilla y
Leon and Midi-Pyrénées. The environmental effects of abolishing the obligatory set-aside
policy can be different between regions and depend on the existing alternative technologies
and land uses. Environmental policies that account for particularities of each region are more
appropriate for achieving environmental goals at EU level.
References
Andersen, E., et al., 2007. Journal of Environmental Management 82: 353-362.
Borkowksi, N., et al., 2007. FSD 2007: An international symposium on Methodologies for
Integrated Analysis of Farm Production Systems, Catania, Italy, Vol. 1, pp. 33-34.
Britz, W., et al., 2007. SEAMLESS integrated project PD 3.5.1.1, 116 pp.
Howitt, R.E., 1995. American Journal of Agricultural Economics 77: 329-342.
Louhichi, K., et al., 2007. SEAMLESS integrated project PD 3.3.11, 82 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
479
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Direct and indirect impacts of CAP changes on the economy of rural areas:
Integrated assessment of farm households’ adaptation strategies in Eastern
and Western Germany
Andrej Lange, Annette Piorr, Rosemarie Siebert, Davide Viaggi
Leibniz-Centre for Agricultural Landscape Research (ZALF),
Eberswalder Str. 84, 15374 Müncheberg, Germany
Contact: andrej.lange@zalf.de
Introduction
A lot of recent research underlines the impact of the CAP reform on various types of farms
across Europe. However, there is little research on interdependencies between CAP policies,
agriculture, entrepreneurial activities in other economic sectors and related land use changes.
This is particular of interest for remote areas which have less long-term economic
development alternatives. Farm size and specialization, land use characteristics and socioeconomic situation determine the scope and direction of the adaptation potential of rural farm
households. The CAP-IRE project (Assessing the multiple Impacts of the Common
Agricultural Policies (CAP) on Rural Economies), funded under the 7th Framework
Programme, addresses these issues. The research project provides valuable results for a better
understanding of described interdependent processes and assessment of CAP policy effects.
The CAP-IRE project includes eleven case study regions in nine European countries. This
contribution will focus on two German regions, the Lahn-Dill district in Hessen and the
district Ostprignitz-Ruppin in the East German state of Brandenburg.
Variability in farm structure and socio-economic situation between the two case study
areas provide for a sharper picture of how changes in CAP measures are reflected and reacted
upon.
The case study approach focuses on policy impacts of current and future CAP development
on farm households and their adaptation strategies towards these changes. Objectives are to
identify key determinants of development in selected regions and to assess trends of change.
An analysis of farm household reactions towards different scenarios of CAP changes provides
the basis for developing concepts and tools for future policy-making, based on improved
understanding of long-term socio-economic mechanisms.
Methodology
Based on a detailed regional analysis, socio-economic key indicators for both regions are
gathered, compared and reflected in terms of economic structure and performance,
demographic development and policy aspects. Main target of the research is the farm
household as central connecting agent between policy and socio-economic changes as well as
between agriculture and activities in other economic sectors.
The analytical approach links future CAP scenarios to empirical data on farm-level
adaptations, including structural change, investment management decisions, off farm work,
chain effects and networking. Results are analysed and interpreted at regional scale, assessing
the developments´ impacts on the socio-economic situation of rural and partially remote
regions. The comparison of an Eastern German and a Western German area provide insight
into the particularities of local farmers’ perceptions and projections of different CAP payment
schemes in relation to farm structure and socio-economic background. Respondents are
confronted with future CAP policy scenarios with a time frame of ten years, including a
complete phasing out of CAP payments and a shift from market and income support to rural
development measures.
The analysis focuses on local experts’ assessment in a two-tier procedure: detailed in-depth
face-to-face interviews with selected farm households provide the basis for a larger survey
480
Session C4: Systems analysis informing policy: Case studies
among farm businesses in both regions. Since participatory procedures are emphasized, strong
stakeholder involvement accompanies the analysis. Local participatory network groups
comprising of actors in regional administration, agriculture, regional planning and regional
entrepreneurial representatives assess the relevancy of research questions for their region,
validate the survey findings and are consulted throughout the process.
Results
Local experts emphasize the role of farm enterprises for climate protection, food and energy
security, and regional development as well as the need for more transparency and information
dissemination. The effects of climate change on regional rural economies are likely to play a
larger role in the future. CAP is perceived to exert strong influence on farm household
structure, with some negative effects on family farming. Links to other regional economic
sectors such as recreation industry, energy production and food processing and will increase
in importance for farm businesses.
The paper presents results of the in-depth farm household surveys, addressing farmers’
assessment of CAP changes as well as their reaction to incentives for economic
diversification. Differences between the Western German and the Eastern German region with
regard to perceived socio-economic changes and adaptation abilities of farming households
will be elaborated upon. Specific socio-economic characteristics of both regions and their
development will be compared and analysed in terms of their reaction to CAP payment
changes.
Outlook
The essential findings from this survey will in a second step guide the large surveys of the
CAP-IRE project on CAP scenarios and farm household adaptation strategies. These surveys
will focus on environmental sustainability, interactions between rural communities and the
rest of the world and governance issues.
References
Brouwer, F., 2006. Main trends in agriculture. Agriculture for sustainable development: a
dialogue on societal demand, pressures and options for policy. SASSPO-SSP4-022698.
Gallerani, V., et al., 2007: Report on the empirical analysis on farmers' investment behaviour
in conventional and emerging farming systems under different policy scenarios 1503692005 F1SC IT – Final report, Bologna.
Happe, K., et al., 2006. German Journal of Agricultural Economics 55: 268-279.
Piorr, A., et al., 2007. In: Eds Ü. Mander, et al., Multifunctional land use: meeting future
demands for landscape goods and services. Springer Verlag, Berlin, pp. 167-182.
Piorr, A., et al., 2009. Integrated assessment of future CAP policies: land use changes, spatial
patterns and targeting. In: Environmental Science and Policy. Special Issue: Sensitive
Region Policy. (in prep.)
Ray, C., 2001. Sociologia Ruralis 41: 280-295.
Shucksmith, M., et al. (Eds), 2005. The CAP and the regions. The territorial impact of the
common agricultural policy. CABI Publishing, Wallingford.
Uthes, S., et al., 2008. Ökologisches Wirtschaften 1: 30-34.
481
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Alpine grazing trends in Switzerland:
Economic importance and impact on biotic communities
Gabriele Mack, Christian Flury, Thomas Walter
Agroscope Reckenholz-Tänikon ART Research Station, 8356 Ettenhausen, Switzerland
Contact: Gabriele.mack@art.admin.ch
Introduction
Alpine grazing fulfils important functions for the environment, agriculture and the regional
economy as well as society (cf. Rudmann, 2004). In Switzerland, a decline in summering has
been observed since 2000 (Table 1). In addition to this, grazing has been dropped in marginal
areas. On easily accessible summering grounds, however, alpine grazing is predominant.
Differences in grazing intensity affect the existence and variety of species. The aim of this
paper is to highlight the economic importance of summering for Swiss agriculture, to forecast
numbers and herds of summered animals and to estimate the impact on flora and fauna in
grassland.
Methods
The Swiss Agricultural Sectoral Information and Forecasting system (SILAS) is used to
forecast summering trends and analyse their economic significance for Swiss agriculture
(Mack et al., 2008). The SILAS model is based on regional farms assigned to production
zones defined by increasingly difficult production conditions. It optimizes regional land use
and regional animal stocks, including summering, by maximizing the factor income according
to the economic accounts for agriculture. Summering is represented in the model by the
amount of available pasture sustaining livestock units during an average alpine grazing season
and by summered animal activities, these encompassing a production segment in the region of
origin and an alpine grazing season of 100 days. The economic significance of summering is
estimated on the basis of its contribution to the factor income. Summering herd changes
impact grazing intensity, which influences the number of common, rare and threatened
species. Trends within high nature-value flora and fauna communities are estimated by a
heuristic analysis.
Results and discussion
Forecasts using the SILAS model show that herds of summered animals are still in decline
and that the current trend will continue, despite the fact that direct payments for summering
animals will be increased by 10% from 2009 on (Table 1). Compared with 2005 the number
of summered animals will decline by 8.4% until 2011, with the trend varying depending on
animal category. The decline in summered animals is significant to the extent that total
roughage consuming livestock units (RCLU) will increase slightly between 2005 and 2011.
Table 1. Forecast of summered RCLU and total RCLU (Change from 2005 to 2011).
Animal category
Dairy cows
Suckling cows
Yearlings
Sheep
Goats
Total
Trend 2000/2005
summered RCLU
total RCLU
–5%
–8%
+53%
+74%
–10%
–8%
–10%
+7%
+16%
+37%
–5%
–4%
Forecast 2005/2011
summered RCLU
total RCLU
–16%
+2%
–5%
–4%
–0%
+2%
–9%
–9%
0%
–1%
–8%
+1%
Source: Own calculations (for further information on data sources see Mack et al., 2008).
482
Session C4: Systems analysis informing policy: Case studies
Summering is of major economic importance to Swiss agriculture despite declining animal
numbers. In 2007, the factor income from summering was CHF 282 mill., equivalent to 11%
of total factor income. Direct payments for summered animals were the major income source
(CHF 202 mill.) in 2007. By 2011 the factor income will drop to CHF 271 mill. Here it is
worth noting its economic importance to the mountain region, where roughly one third of
total factor income is linked to summering. Therefore direct payments for summering
contribute substantially to securing the survival of mountain agriculture and are hence also
relevant for income-related and social policy considerations. In spite of this high income
contribution, the forecasts show that the payments can not prevent a decrease of summered
animals. Additional calculations show that policymakers need to increase direct payments for
summered animals by 25% to maintain summering in future.
In addition to land utilization, changes affecting summered animals also influence the
nature-value of grassland. Different trends are conceivable here, depending on the location
and land configuration of alpine farms: (1) Less land will be stocked as intensively as before.
In this case, primarily high-yielding and easily accessible areas will continue to be farmed.
Low yielding and less accessible marginal areas will be abandoned. In the medium to long
term a decline in high nature-value grassland is anticipated due to the succession towards
forests (Sc. 1 in Figure 1). (2) Less land will be stocked more intensively than before. This
trend reinforces the effects discussed in point 1, with the proportion of land no longer grazed
increasing by the intensification factor. The areas used on a moderately intense level today
will also lose nature-value if use is intensified (Sc. 2 in Figure 1). (3) The area will be less
intensively stocked than today (Sc. 3 in Figure 1). This would have a positive effect on high
nature-value grassland. To realize the third scenario (Sc. 3), policymakers need to change the
direct payment system from payments linked to summered livestock units to payments linked
to pastured area.
Number of pla nt species / 25 m 2
Number of common s pecies
100
Sc
80
60
40
.1
Sc
an
c
dS
Number of additional rare
and/or threate ned s pecies
.3
.2
S c. 2
20
Woodland
and natural
mountain
societies
very
extensive not very moderately intensive
very
transition stage
extensive
intensive intensive
intensive
following
abandonment of
farming
Succession following
abando nment of farming
Grazing intensity
Figure 1. Plant species (indicating high nature-value) at varying grazing intensities and after
abandonment of farming. The arrows show the changes anticipated for each scenario. Source:
derived from Dietl (1995).
References
Dietl, W., 1995. Ökologie und Naturschutz 4: 239-249.
Mack, G., et al., 2008. Entwicklung der Alpung in der Schweiz: Ökonomische Bedeutung und
ökologische Auswirkungen. Yearbook of Socioeconomics in Agriculture 2008, 259-300.
Rudmann, C., 2004. Langfristige Sicherung der Funktionen der schweizerischen Alpbetriebe:
Ein Beitrag zur Umsetzung des Nachhaltigkeitskonzeptes. Diss. ETH No. 15680, 267 pp.
483
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Impact Assessment of policy changes on the arable sector using the FSSIM
model: Case study of the Zachodniopomorskie NUTS2 region
E. Majewski, A. Was, H. Belhouchette, K. Louhichi, I. Mouratiadou
Warsaw University of Life Sciences,
Nowoursynowska 166, 02-787 Warsaw, Poland
Contact: Edward_Majewski@sggw.pl
Introduction
Agricultural policy reforms and changes in the market environment have a direct impact on
farmers’ decision-making. A process of continuous adjustment to such changes can be
observed in the Zachodniopomorskie test case region, since the transformation of Poland into
a market economy in 1989 and the introduction of the EU Common Agricultural Policy
(CAP) in 2004. An interesting question is how farmers are likely to react to expected further
policy developments and changing market conditions and macroeconomic environments.
The FSSIM-MP (Louhichi et al., 2007) farm-level bio-economic model was used to assess
the impact of a number of policy scenarios regarding different sugar market regimes and a
policy promoting biofuel production. The main crops included in the analysis are cereals, oilseed rape and sugar beet, which are the dominant crops in Zachodniopomorskie.
Methods
The most common FADN farm types in the region, namely arable (FT1), livestock mixed
(FT7) and mixed (FT8), were modelled, so as to explore the impacts of the following policy
scenarios:
1. Baseline 2013 – a reference for comparison to further policy scenarios. It assumes the
2004 CAP regime as introduced in Poland and the continuation of the 2006 sugar reform.
All economic parameters regarding prices and costs changes as estimated for the year 2013
were incorporated.
2. Quota Restructured 2013. Under this scenario the A quota1 was cut by 50% and the B
quota was increased by the double of the A quota reduction. Sugar beet prices and compensation in accordance with the 2006 sugar reform remained as in the Baseline scenario.
3. No quota 2013. Complete liberalization of the sugar market was assumed. The quotas were
removed, and the sugar beet price was set equal to the double of the present C quota price.
4. Biofuel Policy. It has been assumed that a policy fostering the use of grains and canola for
biofuel production would result in a growing demand for those crops and consequently
higher prices. A 20% price increase of potential biofuel crops was assumed.
The FSSIM-MP used for modelling is a mathematical programming model with a non-linear
objective function. The model maximizes utility function considering assumed level of risk:
U = P’x + S’x – d’x – x’Qx/2 – K – Φ σ
Subject to: Ax≤ B; x≥0
where, U is the utility, P - vector of gross margin for each agricultural activity, S - vector of
subsidies, d - vector of parameters of the cost function, Q - symmetric, positive (semi-)
definite matrix of the cost function (the estimation of the vector d and the matrix Q depends
on the calibration approaches), x - vector of the level of agricultural activities, K – farm level
fixed costs (including annuity), A - matrix of technical coefficients, B - vector of available
resource levels, Φ - scalar for the risk aversion coefficient, σ - standard deviation of income
according to states of nature defined under two different sources of variation: yield and prices.
1
Sugar quota is a total quantity of sugar allocated to a country. A quota – quantity which can be distributed to
the EU market, B quota – can be traded with export subsidies, C quota – can be exported at world prices beyond
the EU markets (without export subsidies) (Buysse et al., 2007).
484
Session C4: Systems analysis informing policy: Case studies
Agricultural activities are defined in FSSIM-MP as a combination of crop rotation, soil type,
production technique and production orientation. The model was run twice: (1) including only
activities that are currently practiced in the area, and (2) introducing also alternative activities,
regarding rotations and crops that farmers might consider in the future. The first model run
shows the impacts of the policy scenarios on production patterns and farm incomes. The
second model run allows also the examination of further adjustments of production structures
and technologies to different policy environments.
Results and discussion
Table 1 shows the economic results for FT1. Increased farm incomes can be observed, in
comparison to the Base 2004 situation. This is partly due to the growing amount of direct
CAP payments, which rise from 55% of the negotiated rates in 2004 to 100% of the rates in
2013, in accordance with the scheme of the introduction of CAP payments in Poland. This is
also why the premiums as a share of farm incomes increase.
Table 1. Economic results for the arable farm.
Specification
Farm income (Euros)
Premiums (Euros)
Gross production (Euros)
Total costs (Euros)
Total labour use (Hours)
Premium share of income (%)
Base 2004
26098
5671
30084
21740
765
22
Baseline
Quota
No Quota
(Sugar Reform) Restructured (Liberalization)
30137
33241
27561
11192
11662
10693
35705
38339
37926
27290
27290
29470
779
779
934
37
35
39
Biofuel
Policy
35645
10957
41883
28293
782
31
Changes in the level of farm incomes
correspond with shifts in the cropping
80%
structure (Figure 1). The most
60%
significant change is an increase of
40%
land dedicated to sugar beet
production under the No Quota sce20%
nario. The share of cereals and oil0%
seed rape do not increase much
between the Baseline and the Biofuel
Wheat
Rye
Other cereals
Canola
Sugar beets
Potatoes
Policy scenarios, although these crops
are much more profitable under the
Figure 1. Cropping structure in FT 1 under different latter. The reason is that the initial
share of land occupied by these crops
policy scenarios.
was already very high. Other model
results include changes in the supply of different commodities, and in environmental
indicators such as nitrogen input, nitrate leaching, and organic matter balance. Connecting
FSSIM-MP to the entire model chain, comprising also a biophysical model APES, allowing a
broader range of agricultural policy aspects and a larger set of indicators.
Cropping structure (% of arable land)
100%
Base
Baseline (Sug ar
Reform)
Quota Restructured
No Quota
(Liberalization)
Bio-fuel Policy
References
Buysse, B., et al., 2007. Eur. Rev. Agric. Econ. 34(1): 21-52.
Louhichi, K., et al., 2007. A generic template for FSSIM for all farming systems,
SEAMLESS PD3.3.11, 82 pp.
485
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Introducing flat rate direct payments taking into account social concerns
and ecological non-commodities
F.L. Marchand, D. Claeys, L. Lauwers, J. Buysse, B. Van der Straeten, G. Van Huylenbroeck
Institute for Agriculture and Fisheries Research – Social Science Unit,
Burg. Van Gansberghelaan 115 bus 2, 9820 Merelbeke, Belgium
Contact: Fleur.marchand@ilvo.vlaanderen.be
Introduction
Direct payments to producers were decoupled since the June 2003 reform of the European
Common Agricultural Policy (CAP). Farmers get a single farm payment by activating
payment entitlements with eligible area. In Flanders the single farm payment is calculated
according the historical model (based on individual entitlements in the reference period 2000–
2002). Nowadays the European Commission wants to fine tune the modernized CAP, well
known as ‘Health Check’. One of the proposals of the European Commission proposes is
moving away from historical payments. The effects of a flat rate introduction (equal payments
per hectare of cultivated land) on shifts in farm activity and impacts on farmers’ income of
arable, dairy and cattle farms were investigated by Marchand et al. (2008). The impact
analysis of direct payments’ reallocation was merely financially oriented. The aim of current
research is to make the direct payments more goal-orientated and more acceptable to the
public by taking into account also social concerns and ecological non-commodities of
agriculture. We examine how flat rate payments are distributed in relation to some social an
ecological aspects and which effects this redistribution may cause. How many payments do
retiring farmers receive at the expense of the active farmers? When retiring farmers are
excluded, does redistribution have an effect according to farm size, region and farm type? Are
ecological non-commodities rewarded at this time? With respect to this question, we analyse
if areas where agriculture is important to policy purposes in the scope of biodiversity and
valuable areas favored or not? Depending on the results of this more sustainable-oriented
analysis, suggestions to redistribute direct payments will be discussed.
Methods
The farm-based sector model SEPALE, system for evaluation of agricultural and agroenvironmental policies (Buysse, 2006) is used for assessing the impact on production and
income of arable, dairy and cattle farms (cfr. Marchand et al., 2008). SEPALE is a positive
mathematical programming (PMP) that maximizes profit at farm level. As farms can be
selected according to farm size, region and farm type, simulations can be run for specific
subsectors, size classes or regions. A representative sample of FADN farms is used as input.
Marchand et al. (2008) simulated the flat rate scenarios FR1 and FR2. For the scenario
FR1 the total value of payment entitlements and coupled subsidies in 2005 are grouped and
divided by the total available eligible area, excluding potatoes, vegetables, fruit and
permanent cultures. These crops are not allowed to be cultivated with payment entitlements.
The scenario FR2, based on the reform of the common market organization for fruit and
vegetables, does also take potatoes, vegetables, fruit and permanent cultures into account.
Following assumptions are made: farm cannot start new activities, neither payment
entitlements nor milk quota can be traded and the total area can be activated.
In comparison with Marchand et al. (2008) two new flat rate scenarios FR3 and FR4 were
simulated analogous to FR1 and FR2, but in FR3 an FR4 retiring farmers are excluded from
the beneficiaries. This option can be seen (i) as a sensitivity analysis to the major policy
concerns that are incorporated in FR1 and FR2 and (ii) as a socially adjusted scenario which
has been revealed by some stakeholders.
486
Session C4: Systems analysis informing policy: Case studies
A distinction between areas where agriculture is important to policy purposes in the scope
of biodiversity and valuable areas will be calculated by comparing the amount of direct
payments in the different agricultural regions in Flanders (Kempen, Leemstreek, Polders,
Zandstreek, Zandleemstreek) and the importance of biodiversity and valuable areas according
Biological Evaluation Map of Belgium (De Blust et al., 1985).
Results and discussion
The scenarios FR3 and FR4 are socially adjusted scenarios of FR1 and FR2 by the exclusion
of retiring farmers to receive farm payments. They mainly cause a smaller decrease or a
bigger increase in the specific farm activities or may even turn a negative shift in a positive
one. Only in a few cases (e.g., silage and grain corn), the decrease is slightly bigger, however,
negligible. Without any exception, all the changes in income were less negative or more
positive in FR3 and FR4 compared to FR1 and FR2, respectively (Figure 1). Only for cattle
farms in both scenarios and for dairy farms with other cattle in scenario FR4, the loss on
income persisted.
Figure 1. Change in income of Flemish firms, categorized in farm types, caused by two flatrate scenarios (FR 1; FR 2: with CMO-reform = areas of potatoes, vegetables and fruits
included; FR 3 with retired farmers excluded; FR 4 with CMO-reform and retired farmers
excluded) with and without possible shift in farm activities.
We could thus observe in the model results that almost all farm activities shift in a similar but
more positive way when retiring farmers are excluded from receiving payments. Furthermore,
all effects on income were more positive and in many cases a negative effect on income was
turned into a positive effect. As the funds are redistributed within a smaller section of the
farmers, these results are logic. As no surprises in the socially adjusted redistribution are
expected an as it is an issue in the stakeholders’ debate, this option should be at stake at
policy level. The simulations according to the regional biodiversity patterns are not yet
finished. Depending on the results of the flat rate payment distribution in relation to
ecologically important policy issues, redistribution will be simulated and results will be
discussed.
References
Buysse, J., 2006. Farm-level mathematical programming tools for agricultural policy support.
PhD Dissertation, University of Ghent, Belgium, 177 pp.
De Blust, G., et al., 1985. Biologische waarderingskaart van België: algemene verklarende
tekst. Ministerie van Volksgezondheid en Gezin- Instituut voor Hygiëne en Epidemiologie.
Brugge, Belgium, 98 pp.
Marchand, F.L., et al., 2008. In: EAAE Seminar “Modelling Agricultural and Rural
Development Policies”, presented in Sevilla, 2008.
487
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Investigating the economic and water quality effects of the 2003 CAP
reform on arable cropping systems: A Scottish case study
Ioanna Mouratiadou, Cairistiona F.E. Topp, Graham Russell
SAC, King’s Buildings, West Mains Road, Edinburgh, EH9 3JG, UK
Contact: I.Mouratiadou@sms.ed.ac.uk
Introduction
The 2003 Common Agricultural Policy (CAP) Reform aimed to promote the socio-economic
and environmental sustainability of agricultural systems. An important question is how far the
Reform has indeed encouraged farmers to contribute to achieving broad economic and
environmental goals. The effects of the Reform on economic decision making and associated
viability of farms can be explored by analysing data on current farmers’ decisions. However,
these represent the combined effects of all the changes that took place during that period.
Mathematical Programming Modelling provides an alternative for policy assessment that
allows the impact of different factors affecting agricultural production to be studied
separately. Water quality is a major environmental factor that is affected by agricultural
production. The effects of the Reform on water quality are not easy to predict, as they are the
result of the interaction of changes in farmers’ production decisions with biophysical factors
and there are significant time lags between the cause and effect of the environmental
problems. The effects of farmers’ decisions on water quality can be simulated using
appropriately constructed biophysical models to estimate production and pollution functions.
The overall aims of the paper are (1) to explore the economic and water resource effects of
the 2003 CAP Reform on arable cropping systems in Scotland and (2) to evaluate the
methodology used with special reference to agricultural and environmental policy assessment.
Methods
The analysis uses the case study area of the Lunan catchment, a representative catchment of
Eastern Scotland. June Census Data (JCD) were analysed to quantify the changes in land use
after the Reform. The data set consists of information on cropping areas of different crops for
the individual farms in the area, for the years 2000–2007. The JCD were then multiplied by
nitrogen (N) input and nitrate leaching coefficients, to explore changes of nitrate leaching.
The nitrate leaching coefficients were estimated with the NDICEA model (Van der Burgt et
al., 2006), a process-based simulation model which simulates soil water dynamics, N
mineralization and inorganic N dynamics on a weekly time-step. Finally, the outputs of
NDICEA were incorporated into the Mathematical Programming component of the Farm
Systems Simulator (FSSIM-MP) (Van Ittersum et al., 2008; Louhichi et al., 2007), which was
used to identify the effects of a range of scenarios on the average general cropping and
average cereal farms. FSSIM-MP is a bio-economic model developed under the EU FP6
Project SEAMLESS. The model is based on profit maximization and risk aversion and
specifies agricultural activities in terms of rotations, soil types, and management techniques.
The results of the JCD analysis and FSSIM-MP modelling on land use for contrasting
scenarios (Table 1) were compared to identify the drivers of change and to test model
reliability.
Results and discussion
The analysis of both the JCD and modelling results show only small changes in the cropping
pattern of the two farm types. Overall, a shift towards higher yielding and therefore more
profitable crops was observed, partly as a result of the CAP Reform but mainly due to
changes in crop prices. The most significant change (Figure 1) was a decrease in the area of
barley and an increase in the area of wheat. Scenario 1 captured the direction of changes for
488
Session C4: Systems analysis informing policy: Case studies
Table 1. Scenarios (Modelling and JCD Analysis).
Base year
JCD
Scenario 1
Scenario 2 Scenario 3
(2001–2003)
(2006–2007)
Scenario
JCD analysis
JCD analysis
Source
FSSIM-MP
FSSIM-MP modelling
Assumed
2001–2003
2001–2003
2006–2007
Prices
EU CAP
Agenda 2000
2003 CAP Reform
Measures
None
None
Cross-compliance N quota
None
most crops regarding the general cropping farm, and showed an increase in the area of all
crops after a reduction of set-aside for the cereal farm. These results were further augmented
and much closer to the changes shown by the JCD analysis under Scenarios 2 and 3. Farm
incomes did not decline significantly after the introduction of the Reform. Indeed,
significantly higher incomes were achieved after the introduction of new crop prices. Cereal
farms appear to have lower income per hectare, to be more dependant on premiums and to be
more sensitive to price changes than general cropping farms. Overall N use and nitrate
leaching do not differ significantly between the scenarios for either of the two farm types. The
average N use is only above the quota for Scenario 3 on the cereal farm. However, although N
use is higher in this case, the leaching does not seem to increase due to higher uptake by N
intensive crops. This also explains why nitrate leaching did not change from 2000–2007,
suggesting that more restrictive input quotas will yield no major improvements while
discouraging farmers from growing profitable but high N demanding crops.
Bio-economic modelling provides a framework for simultaneously analysing the economic
and environmental impacts of farmers’ decision-making, as it enables agricultural systems to
be represented in economic and biophysical terms. In spite of a lack of detailed input data,
this approach can help in explaining the drivers of changes. Although the comparison of
model predictions with actual data constitutes a form of model testing and increases
confidence in the model outcomes, not all potential modelling scenarios have been tested and
model predictions should not be accepted blindly. Rather, the outcomes should be considered
as hypotheses that become the input to further discussions with experts and policymakers.
70
Baseyear
JCD
Scenario 1
Scenario 2
Scenario 3
60
LandUse %
50
40
30
20
10
0
Barley
Wheat
Winter OSR
Set-aside
Oats
Main Crop
Potatoes
Seed
Potatoes
Figure 1. Modelling results and current levels - cereal farm.
References
Louhichi, K., et al., 2007. A generic Template for FSSIM for all Farming Systems, PD3.3.11,
SEAMLESS integrated project, EU 6th Framework Programme, contract 010036-2, 82 pp.
Van der Burgt, G.J.H.M., et al., 2006. Nutrient Cycling in Agroecosystems 74: 275-294.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
489
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Systems analysis underpinning development of management action targets
to meet water quality objectives in sugarcane production systems
P.J. Thorburn, J. Brodie, R. Milla, E. Shannon, S. Attard, T. Anderson, A. Davis
CSIRO Sustainable Ecosystems, 306 Carmody Rd, St. Lucia Qld 4068, Australia
Contact: Peter.Thorburn@csiro.au
Introduction
In response to concerns about the health of the Great Barrier Reef (GBR), governments in
Australia have developed the Reef Water Quality Protection Plan to significantly reduce
discharge of pollutants from industries in catchments draining into the GBR lagoon. These
reductions will be achieved through development and implementation of Water Quality
Improvement Plans (WQIPs) in reef catchments. A WQIP requires identification of
management practices and targets for their adoption (management action targets) to meet
stated regional water quality objectives and provide progressive (quantified) reductions in
pollutant loads. Agriculture is an important land use in GBR catchments and has been
identified as an important source of diffuse pollution. Thus agricultural industries and local
catchment authorities are challenged with developing management practices and adoption
targets to meet water quality objectives.
Agricultural industries in these catchments, especially crop production, do not have a long
history of research into environmental impacts of different farming systems. Thus modelling
and systems analysis have an important role to play in identifying both management practices
and targets for their adoption to meet water quality objectives. Furthermore, given the
importance of adoption, it is necessary to have wide stakeholder involvement in and
acceptance of the process and its outputs. This paper describes the process used to
characterize the water quality impact of different management practices for sugarcane
production in the lower Burdekin catchment, and explore the consequences of and reach
agreement amongst stakeholders for different levels of adoption of the various practices. This
participative process involved staff from agricultural advisory agencies, farmers and the
systems analysts, to define adoptable management practices for the Burdekin WQIP.
Identification of management practices
Nitrogen (N) had been identified as the main pollutant from sugarcane production in the
Burdekin region (Brodie & Bainbridge, 2008), so management practices concentrated loss of
N via runoff and deep drainage. Workshops were held with local farmers and extension
officers to explore possible management practices to reduce long-term N losses. These
practices classified the range of current management practices in the region and those under
development that had potential to deliver water quality benefits. Five classes of management
practices, termed E to A (from ‘bad’ to ‘good’), were identified combining decreasing tillage
intensity, reducing N application rate improved irrigation scheduling. Classes E to C
represented practices common in the Burdekin region; Class B was similar to the currently
promoted ‘best practice’; with Class A being the possible future best practice that is currently
under investigation. The long-term N losses (Figure 1) and productivity of these management
classes was simulated with APSIM-Sugarcane for the dominant soil types in the four main
districts (Delta fine textured, Delta coarse textured, Mona Park and Mulgrave) in the region,
based on experimental studies of water quality in sugarcane production in these districts.
Extension officers were interviewed to estimate the proportion of farmers practicing these
classes in each of the four districts. The majority (50–60%) of farmers were practicing Class
D, with fewer practicing Classes E and C. From these adoption estimates and the areas of the
districts, the regional average N fertilizer use (211 kg ha–1) was determined. This N use
compared well with data on actual use in the past five years (220 kg ha–1), suggesting that the
490
Session C4: Systems analysis informing policy: Case studies
distribution of management practices were plausible. The water quality impacts of the
adoption estimates and N loss predictions was also plausible: The recent N load estimations in
the region (3000–4500 t yr–1, Brodie & Bainbridge, 2008) compare favourably with predicted
long term N loads (5500 t yr–1) that are based on under current conditions, rather than
historical (and variable) N usage and areas under sugarcane.
80
40
N in Runoff
N in D.Drainage
120
80
40
A
B
C
D
E
Management Class
200
Mona Pk
N in Runoff
N in D.Drainage
120
80
40
0
0
0
160
N Lost (kg N / ha)
N in Runoff
N in D.Drainage
120
160
200
Delta(c)
N Lost (kg N / ha)
160
200
Delta(f)
N Lost (kg N / ha)
N Lost (kg N / ha)
200
A
B
C
D
B
C
D
E
Management Class
Management Class
M ulgrave
N i n R unof f
N i n D . D rainage
120
80
40
0
A
E
160
A
B
C
D
E
Management Class
Figure 1. Predicted long-term annual losses of N via runoff and deep drainage at the four sites
under five classes of management practices designed to meet water quality targets.
Setting management action targets
The hydraulic connections in the region are simple, with transport of water and chemicals to
the river and creeks rapid so processes such as in-stream denitrification negligible. Reflecting
this simplicity, a simple regional ‘calculator’ was constructed to allow participative
exploration of the relationship of between different patterns of adoption and regional N loads
(Figure 2). The calculator was used in a facilitated workshop with local farmers and extension
officers to determine targets for the adoption of the different practices to meet regional water
quality objectives; i.e. predicted N loads reduced 4400 t yr–1 by 2013. This target would be
met by having a net shift of 10% of farmers from Class E to Class B practices.
Figure 2. Screen from the regional
nitrogen load calculator used
participatively with stakeholders to
assess the water quality benefits
from
adoption
of
various
management practices.
Discussion
The analyses undertaken and the process used has provided the local catchment management
authority, Burdekin Dry Tropics NRM, with quantitative and agreed targets for the adoption
of management practices for sugarcane production in the lower Burdekin catchment. This will
underpin funding of incentive schemes facilitating practice change, the means of assessing the
effectiveness of those funds, and the water quality improvement resultant from them. This
provides a sound basis for implementing the Reef Water Quality Protection Plan.
Reference
Brodie, J. & Z. Bainbridge, 2008 ACTFR Report No. 06/01. Aust. Cent. Tropical Freshwater
Research, Townsville.
491
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Efficiency of environmental policies in agriculture in the Czech Republic:
The case of dairy cattle
M. Havlikova, C. Kroeze
Wageningen University, Environmental Sciences,
Environmental Systems Analysis Group, P.O. Box 47,
6700 AA Wageningen, The Netherlands
Contact: carolien.kroeze@wur.nl
Introduction
Dairy cattle production is one of the traditional sectors in the Czech Republic, and an essential
part of Czech agriculture. It is also a source of various air, soil and water pollutants. The
impact of agriculture is, therefore, increasingly considered in environmental policy making.
Current policies in the Czech Republic focus on minimizing emissions of nitrate and
ammonia. It can be questioned, however, how effective it is to primarily reduce these two
pollutants.
We developed a model (DAIRY) that can be used to analyse the cost-effectiveness of
policies for the simultaneous abatement of acidification, terrestrial and aquatic eutrophication,
human toxicity and global warming caused by dairy production in the Czech Republic. In this
paper we evaluate the current and future environmental impact of dairy cattle in the Czech
Republic. To this end, a number of scenarios are developed and analysed with the DAIRY
model. Several environmental targets are used as model input. These include reduction targets
for ammonia emissions and for the overall environmental impact (OEI). The scenarios also
differ with respect to the underlying assumption on changes in animal management and
animal numbers. We present future environmental trends, and the costs associated with costoptimal sets of reduction measures as selected by the model.
Dairy model description
The DAIRY model considers seven pollutants: ammonia (NH3), nitrous oxide (N2O),
methane (CH4), particulate matter (PM2.5, PM10) and nitrate (NO3) emitted by dairy cattle.
Several processes give rise to emissions: grazing, milking, housing, storage, application,
indirect emissions and leaching. The model distinguishes between nine study regions within
the Czech Republic which differ in dairy cattle intensity, environmental sensitivity and
population density. Fifteen emission reduction measures are included to abate the above
mentioned emissions. Six are primarily aimed at reducing ammonia emissions, five at
reducing greenhouse gas emissions such as nitrous oxide and methane, and four at reducing
nitrate leaching. In addition, the model considers unintended side-effect of the measures on
other pollutants. Emissions are calculated by study region, based on emission factor
approaches and from a process-based model. The potential environmental impact is
calculated using so-called characterization factors, which are region specific for human
toxicity, country specific for acidification, and for terrestrial and aquatic eutrophication.
They are generic for global warming. The individual impact categories are aggregated into an
overall environmental impact (OEI) indicator by means of normalization and valuation
factors as usual in Multi-Criteria Analysis. The model is modified from Brink et al. (2005).
For a more detailed description of the model see Havlikova et al. (2008) and Havlikova &
Kroeze (2009).
Model results: Discussion and conclusions
We present DAIRY model results for three scenarios (Table 1). These scenarios use either
reduction targets for ammonia or for the overall environmental impact. In the AMMONIA
Scenario, the restriction is that the model can only select from measures aimed at reducing
492
Session C4: Systems analysis informing policy: Case studies
ammonia, while in the EUPLAN and IMPACT Scenarios the model can also select measures
aimed at reducing other pollutants that, as a side effect, also reduce ammonia emissions. The
AMMONIA and EUPLAN scenarios define targets for ammonia emissions alone, while the
IMPACT scenario aims to reduce the overall environmental impact (OEI).
The results indicate that the AMMONIA and EUPLAN scenarios (1–3% reduction in OEI;
costs 11–16 MEuro yr–1) are less cost-effective in reducing the OEI than the IMPACT
scenarios (21% reduction; 12 MEuro yr–1). We conclude that it seems more cost-effective to
aim at an overall reduction of the environmental impact, than on reducing ammonia emissions
alone. It is interesting that in all three scenarios stable adaptation and manure efficiency
improvement are selected as cost-effective measures by the model. This indicates that these
are robust solutions for environmental pollution problems caused by dairy cattle in the Czech
Republic. The policy implications would be that implementation of these measures may serve
multiple purposes.
Table 1. Overview of scenarios for the year 2020 as analysed in the present study.
Scenario
Aim of analysis
AMMONIAMinimizing total reduction costs in
the Czech Republic to reduce NH3
emissions by 15% relative to 2005
in each study region, model can
only select technical measures to
reduce NH3
EUPLAN Minimizing total reduction costs in
the Czech Republic to reduce NH3
emissions by 30% relative to 2005
in each study region, model
considers all technical measures.
IMPACT Minimizing total reduction costs in
the Czech Republic to reduce the
OEI emissions by 30%
relative to 2005 at the national
level
*
Measures
Total reduction
selected in costcosts
optimal solution
(MEuro/year)
11
- Stable adaptation
- Low nitrogen feed
- Covered manure stora
Change in
OEI (%)*
–1%
- Manure efficiency
improvement
- Stable adaptation
- Low nitrogen feed
16
-3%
- Manure timing
application
- Manure efficiency
improvement
- Stable adaptation
12
-21%
Relative to a No Control scenario for the year 2020, assuming no environmental policies for
agriculture in the Czech Republic (see Havlikova & Kroeze, 2009).
References
Brink, C., et al., 2005. Ecological Economics 53: 59-74.
Havlikova, M. & C. Kroeze, 2009. Cost-effective reduction of environmental impact of dairy
cattle in the Czech Republic. Submitted to the Environmental Modelling and Assessment.
Havlikova, M., et al., 2008. The Science of the Total Environment 396: 121-131.
493
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme D
Integrated assessment: Science and impact
Session D1: Stakeholder involvement in designing, use and
evaluation of integrated assessment models
Session organizers:
Sylvia Herrmann
Leibniz University, Hannover, Germany
Johanna Alkan Olsson
Lund University, Sweden
Elske Van de Fliert
The University of Queensland, Brisbane, Australia
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
496
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
The role of models in supporting processes of change towards sustainable
resource governance1
Claudia Pahl-Wostl
Institute of Environmental Systems Research Department of Mathematics & Computer
Science University of Osnabrück,
49069 Osnabrück, Germany
Contact: Pahl@usf.Uni-Osnabrueck.de
Numerous arguments have been put forward regarding the need for a major change in
resource governance to guarantee the sustainability of social-ecological systems, in particular
given the challenges of global change. Regarding the system metaphor governance of
resources has to be characterized as navigating complex adaptive systems in an uncertain
environment rather than as solving multi-objective optimization problems of predictable and
controllable systems. Goal oriented, instrumental approaches have to be complemented by a
procedural understanding which portrays sustainable development as a societal search and
learning process. Processes of social learning play a major role where stakeholders at different
scales are connected in flexible networks and where the capacity and trust is developed to
collaborate in a wide range of formal and informal relationships from formal legal structures
and contracts to informal, voluntary agreements.
Such a perspective suggests as well a new role of models in analysing and supporting
learning and decision-making processes. Participatory modelling and scenario building
approaches may be instrumental in supporting processes of social learning and in developing
innovative approaches in dealing with uncertainties. The paper will give an overview on
recent developments and illustrate potential and limitations with emphasis on challenges
posed by water management.
1
Keynote presentation
497
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Models as social learning tools in participatory integrated assessment
Joop De Kraker, Carolien Kroeze, Paul Kirschner
School of Science & Netherlands Laboratory for Lifelong Learning,
Open Universiteit Nederland, P.O. Box 2960, 6401 DL Heerlen, The Netherlands
Contact: joop.dekraker@ou.nl
Introduction
Integrated assessments (IA) are increasingly being conducted in interaction with stakeholders.
The major reasons for participation of stakeholders are to (1) account for diversity in
perspectives on the problem and possible solutions, (2) tap relevant non-scientific sources of
knowledge, and (3) enhance the basis of support for the outcomes of the assessment. The
degree of participation may vary from ‘informing’ to ‘co-production’ and ‘mutual learning’
(Van de Kerkhof, 2006). The role of learning in participatory integrated assessments (PIAs)
has received considerable recent attention. Studies on learning in PIA-projects make clear that
learning can take place in many ways (Tuinstra et al., 2008). Scientists can learn from
stakeholders and vice versa. When this concerns acquiring new technical knowledge and
insights on, for example, important constraints or effective strategies, we speak of
‘instrumental’ or ‘single-loop’ learning. Learning may also concern the underlying values,
beliefs and objectives that determine how stakeholders frame the problem. When a PIA
results not only in a change in knowledge, but also in a change in the stakeholders’
perspective on the problem, we speak of ‘social’ or ‘double-loop’ learning (Argyris & Schön,
1996). Whereas the emphasis used to be on instrumental learning in IA-projects aiming at
decision support for policymakers, attention has shifted to social learning in PIA-projects that
concern problems characterized by complexity, uncertainty, multiple stakeholders and
diverging perspectives (Pahl-Wostl, 2007). To deal with such ‘wicked problems’ in a
sustainable way, requires a shared understanding and reframing of the problem by the
stakeholders serving as a basis for the development of new, collectively supported and
pursued solutions. This new and shared perspective on the problem and its solutions is
precisely what social learning promises to offer (Wals, 2007).
This contribution reviews the recent literature on social learning in PIAs to assess whether
and how the integrative environmental models typically used in IAs can serve as tools to
support social learning among stakeholders.
Models as social learning tools: state of the art
From its inception, IA recognized that models could play an important role in supporting
stakeholder learning, including social learning (e.g., Rotmans, 1998). This role, however, has
rarely been an object of critical examination. An exception is the retrospective evaluation of
the role of models in major PIA-projects by Siebenhüner & Barth (2005), who concluded that
models had only supported instrumental learning, and who doubted whether models would
even have the potential to support social learning in PIAs. This conclusion is not surprising
given the mismatch they observed between the models used and the needs of the stakeholders
in the cases studied. As such, it is unlikely to be a generally valid conclusion. A mismatch
between the perspective of the model developers and the intended users was also the reason
for the limited potential of an integrated land-use model to support social learning in natural
resource management by farmer communities in the Sahel (Van Paassen, 2004). A study by
Sterk (2007) on the contribution of land-use modelling under conditions closer resembling a
PIA found proof of a positive effect on social learning, such as adapted problem definitions, a
changed solution space, and the formation of new coalitions to tackle the problem. All three
studies suggest the need to involve stakeholders in model development and/or the production
of model results to make models more effective as tools for social learning. A variety of such
498
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
participatory approaches have been proposed and to a certain extent have been developed and
implemented over the past five years. These include model-based interactive back casting
with an active role for stakeholders in scenario development (Robinson, 2003), participatory
assessment of model quality and uncertainty (Van der Sluijs et al., 2005), participatory
development of agent-based models by including stakeholder perspectives and objectives in
an interactive way (Tabara & Pahl-Wostl, 2007), and interactive combinations of modelling
with role playing or gaming (Valkering et al., 2008). The effectiveness of these approaches in
supporting social learning in the context of PIAs has not been evaluated thus far.
Conclusions
Only a small number of studies have assessed the role of models as social learning tools in
PIAs. Some of these studies were post-hoc, and none deeply studied the learning processes
involved. During recent years a variety of participatory approaches have been developed that
may enhance the effectiveness of models as social learning tools, but they have not yet been
evaluated as such. One could speak of a ‘paradox of model-supported social learning’ (cf.
Armitage et al., 2008): many authors have stressed its importance, but evidence is scarce and
validated guidelines on how the contribution of models could be enhanced are lacking. This
indicates the need to study the role of models as social learning tools in on-going PIAs,
enabling due attention for the processes by which social learning takes place. To identify how
models could be designed and deployed to better support these processes, such studies could
greatly benefit from the methods and insights of educational technology and collaborative
learning (e.g., Clark & Linn, 2003; Van Bruggen & Kirschner, 2003).
References
Argyris, C. & D. Schön, 1996. Organisational learning. A theory of action perspective.
Addison-Wesley Publishing Company, Massachusetts, USA.
Armitage, D., et al., 2008. Global Environmental Change 18: 86-98.
Clark, D. & M.C. Linn, 2003. The Journal of the Learning Sciences 12: 451-493.
Pahl-Wostl, C., 2007. Env. Modelling & Software 22: 561-569.
Robinson, J., 2003. Futures 35: 839-856.
Rotmans, J., 1998. Env. Modelling and Assessment 3: 155-179.
Siebenhüner, B. & V. Barth, 2005. Env. Impact Ass. Review 25: 367-389.
Sterk, B., 2007. A window of opportunities. The contributions of land use modelling to
societal learning. PhD thesis, Wageningen University, The Netherlands.
Tabara, J.D. & C. Pahl-Wostl, 2007. Ecology and Society 12, 3. (online)
Tuinstra, W., et al., 2008. Int. J. of Innovation and Sustainable Development. 3: 128-152.
Valkering, P., et al., 2008. MATISSE Working Paper 25.
Van Bruggen, J.M. & P.A. Kirschner, 2003. In: Eds. J. Andriessen, et al., Cognitions:
Arguing to learn. Kluwer Academic Press, Dordrecht, pp. 177-203.
Van de Kerkhof, M., 2006. The Integrated Assessment Journal 6: 7-34.
Van der Sluijs, J.P., et al., 2005. Risk Analysis 25: 481-492.
Van Paassen, A., 2004. Bridging the gap: computer model enhanced learning about natural
resource management in Burkina Faso. PhD thesis, Wageningen University.
Wals, A. (Ed.), 2007. Social Learning: Towards a more sustainable world. Wageningen
Academic Publishers, Wageningen, The Netherlands.
499
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Application of an integrated monitoring tool for sustainable dairy farming
in Flanders
K. De Mey, K. D’Haene, M. Meul
Institute for Agricultural and Fisheries Research – Social Sciences Unit,
Burg. Van Gansberghelaan 115 Box 2, 9820 Merelbeke, Belgium
Contact: karen.demey@ilvo.vlaanderen.be
Introduction
Modern society increasingly recognizes the importance of sustainable farming systems and
expects that the agricultural sector undertakes considerable efforts to produce in a social and
ecological sound way, without overlooking the economic liveability of a farm. In order to
convert the theoretical concept of sustainability into a tangible concept at farm level, an
indicator-based Monitoring Tool for Integrated Farm Sustainability (MOTIFS) was designed
in Flanders (Meul et al., 2008). MOTIFS (Figure 1) is a graphical tool which integrates
indicators for economic, ecological and social sustainability themes and sustainable
entrepreneurship. The tool allows for an immediate and holistic interpretation of the farm’s
overall sustainability level and gives an overview of the farm’s strengths and weaknesses. In
this paper, we describe two applications of MOTIFS. Both are used to introduce the tool into
practice and to validate its end-use value, which allows optimizing and continuously
improving the tool and its application.
theme (or indicator) scores vary
between 0 (not sustainable) and
100 (sustainable)
risk 100
internal social
sustainability
the segment width defines the
theme's (or indicator’s) weight
80
profitability
external social
sustainability
60
the segment color type defines the
theme's (or indicator’s) sustainability
dimension: red= social, green=
ecological, yellow= entrepreneurship,
blue= economic
40
20
disposable
income
productivity
and efficiency
The bold line represents the average
score of a reference group of
comparable farms
use of inputs
entrepreneurship
biodiversity
when a theme (or indicator) has not
(yet) been worked out, the name will
be in italics and the segment is left
blank
quality of natural
resources
Figure 1. MOTIFS, presented with a legend concerning its reading and interpretation.
Methods
We look at two types of application of MOTIFS on Flemish dairy farms: (1) an application in
a European Leader+ project, and (2) an application in two existing farmer groups from
Flemish farm consultancy agencies. These applications allow us to obtain an overview of the
opportunities and bottlenecks for future applications of MOTIFS.
First, MOTIFS has been applied on 20 Flemish dairy farms participating in the Leader+
project ‘Strong with Milk, 2006–2008’ with the aim to monitor sustainability and stimulate
communication and exchange of knowledge between farmers (Schoonhoven, 2008). For a
number of selected sustainability themes, the project leader regularly collected on-farm data
through farm accountings and direct farmer inquiries, calculated the indicators and discussed
the results with each farmer individually. Additionally, farmer discussion groups were
organized, in which the results for a specific sustainability theme were discussed, together
with an invited expert. For a first end-use validation of MOTIFS, a qualitative research
500
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
approach was used, consisting of semi-structured interviews with the potential end-users, i.e.
farmers and farm consultants.
In a second, on-going application, MOTIFS is applied by two groups of dairy farmers that
are member of two different farm consultancy agencies. Both groups consist of approximately
10 farmers who meet on regular basis to discuss their farm economic accounting under
guidance of a farm consultant. In close cooperation with the groups’ farm consultants, the
data already available in the farm accountings (economic and environmental farm figures) are
linked to a basic system for automatic calculation of MOTIFS. The discussion of the sustainability results is integrated in the regular farmer meetings that are directed by the farm consultant. The first meeting elaborated the entire tool and its principles, while following meetings
deal with a specific theme. Feedback on the end-use value of MOTIFS is gathered through
interviews and questionnaires with end-users and direct observations in farmer meetings.
Results and discussion
Both types of application of MOTIFS learned that both farmers and farm consultants find the
tool useful to give an objective overview of all important sustainability aspects of a farm and
its management. It is a good starting point to touch upon other than only economic and
technical issues, as is usually done in the regular farmer meetings. Particularly the farm
consultants indicate it as a practical tool to base their advice on and to steer farmers towards a
larger awareness of an integrated sustainable farm management. However, it becomes clear
from the use of MOTIFS that it takes time to become familiar with the type of presentation
and that for some indicators that are new to farmers and consultants, it is not easy to
immediately interpret the results. It is therefore important to assist potential users in the
application of MOTIFS and provide support in the interpretation of the results, especially in
the beginning of the implementation. The ‘Strong with Milk’ project demonstrated that
MOTIFS can be a valuable communication tool since it allows to compare individual farm
results and to discuss about tangible aspects of farm management in discussion groups.
However, additional information and hard figures are still necessary to give precise advice or
take decisions on specific management measures.
The cooperation with the farm consultancy agencies allows the development of a userfriendly electronic system that is integrated in their existing accounting programs in order to
easily calculate MOTIFS. For this purpose, data available in farm accountings is tailored to
the data necessary for calculating MOTIFS. As the cooperation with farm consultancy
agencies is still at an initial stage, the end-use value and willingness of dairy farmers and
consultants to use MOTIFS in practice need to be further examined. A preliminary finding is
that the use of MOTIFS by farm consultancy agencies is fairly different than its use in a
project like ‘Strong with Milk’. Farm consultants are mostly economic-technically oriented
with a strong focus on current farm income, and although they want to broaden their view and
incorporate other sustainability aspects, it is difficult to change their advising system.
Consequently, they try to deal only with the economic-technical aspects of MOTIFS.
As the implementation of MOTIFS on Flemish farms is, however, constantly evolving,
with a continuous feedback between research and practice, we will look in the next phase of
the research at the entire process of implementing a sustainability monitoring tool like
MOTIFS ‘from theory towards practice’. The purpose will be to look from a broader
perspective at critical success factors arising in the adoption process of sustainability
monitoring tools and sustainability measures.
References
Meul, M., et al., 2008. Agron. Sustain. Dev. 28: 321-332.
Schoonhoven, D., 2008. Sterk met Melk! Een handleiding om duurzaam melk te produceren.
Plaatselijke Groep Leader+ Meetjesland, 59 pp. (in Dutch)
501
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Analysis of stakeholder - researcher interactions for designing a generic
grassland model suited for a diversity of farmers’ practices
M. Duru, G. Martin
UMR1248 INRA-ENSAT, Chemin de Borde Rouge, BP 52627,
31326 Castanet Tolosan, France
Contact: mduru@toulouse.inra.fr
Introduction
In order to provide support to farmers, a lot of attention has been given by researchers to the
development of crop models, which are used to study the adaptation of farmers’ management
practices to new challenges, such as new policies. In practice, few models are effectively used
by farmers or advisors in spite of their ability to integrate a great deal of relevant knowledge.
Models designed for decision support systems are considered to have little relevance to real
world situations (Woodward et al., 2008), being too prescriptive and unable to account for
farmers’ management preferences. Participatory research is encouraged to avoid these
problems (Nowotny et al., 2001) but there are no methodological guidelines for this. In addition, there is such a range of farmers’ targets and constraints (Vanclay, 2004) that a generic
crop model might fail to meet the stakeholders’ expectations. To be adopted, even a technical
change needs to be satisfactory in terms of its wider consequences i.e. social (labour) and
environmental, giving rise to questions such as how to conduct participatory research making
use of crop models and how to design experiments to enrich a generic crop model.
We hypothesize that by coupling on-farm observations, analytical experiments, and
stakeholder-researcher workshops, it is possible to produce a generic crop model well fitted
for local situations. We draw some lessons based on the critical analysis of participatory
research projects concerned with different versions of a grassland model simulating herbage
growth and nutritive value. This model was targeted to learn about the effect of management
practices. To analyse these projects, we used the grid of David (2001) which suggests that
participatory research aimed at building cognitive or decision tools for management should
alternate phases of model formalization and model contextualization.
Materials and methods
The analysis is based on research projects focusing on grazing management in the Aveyron
plateau (1988–1997), and on management of species-rich grasslands on mountains (Pyrenees:
1997–2004 and Massif Central: 2005 onwards). Starting from a very simple model structure,
targeted experiments driven by on-farm observations provided the data for the model. Three
main successive developments of the model were the modelling of grazing intensity,
flexibility in management and biodiversity.
These projects combined different methods, successively and/or concomitantly: (i) onfarm observations and surveys for problem finding and model evaluation; (ii) targeted
experiments and integration into the grassland model (model formalization); (iii) stakeholder
surveys and training sessions (model contextualization).
Some lessons drawn from these case studies
We find that, as in other management domains (David, 2001), it may be advantageous to
combine successively or roughly simultaneously different methods for building a model
intended to provide relevant information for advisors and farmers at a local level.
On-farm observations are involved in model formalization and contextualization. They allow
key performance criteria used by farmers to be identified, but usually leave aside by
researchers, e.g., the timing of herbage growth on a seasonal time scale; flexibility, i.e. timing
of grassland use, or the possibility of bringing forward or delaying the use of a meadow
502
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
Table 1. Description of the three projects (columns) using the David grid (lines).
Aim and
innovation
required
Problem finding
model evaluation
Changes in
model
formalization
Process of model
contextualization
(who, when,
what)
References
(not listed)
Studying grazing
management practices by
modelling herbage
availability
On-farm surveys
and measurements
Added processes: leaf
senescence, stem
elongation
Advisory services
(operational training)
Before, during and after
Graphs for defining
herbage thresholds to
meet different
management aims
Cros et al., 2003;
Duru et al., 2000
Studying grazing
management practices by
modelling flexibility in
grassland use
On-farm soft
observations
Added agrienvironmental targets
(thresholds for minimum
and maximum herbage N
content)
Advisory services
(operational training)
After
Graphs for defining
defoliation regimes
allowing to remain within
the thresholds
Duru & Delaby, 2003;
Duru et al., 2007
Studying management
for agri-environmental
value by modelling
species-rich grasslands
On-farm deep
observations
Key model variables
for different plant
functional types (tissue
composition, leaf life
span etc.)
Advisory services
Interactions during the
process of knowledge
formalization
Typology grid to assess
agri-environmental
value
Duru et al., 2008
within a given time interval. Furthermore, on-farm observations may identify management
practices that look suboptimal over the short term and on small spatial scales, but that may be
valuable when considering longer time scales and higher organizational levels.
Experiments targeted by on-farm observations and surveys, as well as training sessions with
farm advisors, are a useful step to gather results to be integrated into the model. A new
paradigm can emerge, e.g., Duru (2009) for modelling feeding value.
Training sessions at each stage of the model design have shown that whereas the questions
which arise about grassland management are very diverse, it is an acceptable simplification to
choose a limited number of biophysical processes to put into the model and to parameterize
them for key driving factors. This may interest a large group of stakeholders and may be
applicable to a wide range of local farming situations. The model does not provide a complete
response to advisors’ questions, but it is a support for discussion, simulations giving
indications of what may happen to some key performance indicators when changing driving
factors. In this way, running the model for a given field or season gives partial but interesting
information for planning grassland management for the whole farm over a number of years.
From our experience, using our grassland model as a teaching aid during training sessions
with advisors provided an integrated view of the response of herbage growth to abiotic factors
and management practices. It was used as a cognitive tool highlighting how these driving
variables interact. Currently, the model is not used by the grassland experts themselves:
researchers need to act as intermediaries.
By being contextualized, this research produces socially robust knowledge supported by
certified knowledge (academic publications), but more than this, it incorporates the results of
multiple interactions and applications with stakeholders. However this is long-term research.
References
David, A., 2001. Les nouvelles fondations des sciences de gestion. Editions FNEGE, 213 pp.
Duru, M., 2009. Renouveler les questions de recherche par la recherche en partenariat. (in
press)
Nowotny, H., et al., 2001. Re-Thinking Science. Polity Press, Cambridge.
Vanclay, F., 2004. Aust. J. of Exp. Agric. 44: 213-222.
Woodward, S.J., et al., 2008. New Zealand J. of Agric. Res. 51: 235-252.
503
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Accounting for user’s uncertainty information needs in Integrated
Assessment models: The case of the SEAMLESS Integrated Framework
Silke Gabbert, M.K. Van Ittersum, F. Ewert, C. Kroeze, S. Stalpers, J. Alkan Olsson
Wageningen University, Environmental Economics and Natural Resources Group,
Hollandseweg 1, 6706 KN Wageningen, The Netherlands
Contact: silke.gabbert@wur.nl
Introduction
Integrated Assessment (IA) models are computer models which serve as tools to analyse
complex real world problems and to portray their social, economic, environmental and
institutional dimensions. Given the problems addressed and the complexity of IA models
themselves they are subject to various types and sources of uncertainties. To become useful
tools, therefore, assessments of uncertainties in IA models are inevitable.
Current practice is that IA models are predominantly designed from the modellers’
perspective, without consulting model users. Only little attention has been paid to the
question which uncertainty information is demanded by model users. This is surprising as
scientists have recognized the need to analyse the model users’ perspective of uncertainty
assessment in IA models (e.g., Gough, 1999; Walker et al., 2003; Gabbert, 2008), and as
model users have repeatedly stressed the need for a more user-oriented analysis of
uncertainties in these models (e.g., IIASA, 2002; CEC, 2004). This may also be attractive to
give uncertainty analysis focus, as the number of sources to be investigated is often
enormous.
The objective of our paper is, therefore, to suggest an approach for uncertainty analysis in
IA models that explicitly accounts for model user perspectives on uncertainties. Our intention
is to complement existing frameworks and analytical concepts for uncertainty analysis in IA
models in order to strengthen their usefulness as ‘science-policy interfaces’. As an illustrative
example, our approach is applied to the SEAMLESS Integrated Framework.
Approach
We define the general objective of uncertainty analysis as to identify and assess model
imperfections of any type and source (either in a quantitative or qualitative way) in order to
create confidence in model outcomes. Since Integrated Assessment is a highly interactive
process, involving different stakeholder groups, we combine the typology of sources and
types of uncertainties from Walker et al. (2003) with an investigation of perspectives of
model users. Model users can be analysts, i.e. researchers using an IA model for research
purposes, or public decision makers, who use IA models as scientific underpinning in
concrete decision contexts. The IA model, i.e. the (computerized) formalization of a given
(complex) problem transforms different inputs (assumptions, parameters, data, mathematical
relationships) into information that aids decision making on this problem.
The usefulness of IA models as decision-support tools crucially depends on whether the
model users feel confident with model outcomes. This requires to investigate whether the
uncertainty information provided by model builders and analysts adequately reflects the
“model users’ uncertainty information needs” (Gabbert, 2008). User needs can be identified
either from a theoretical perspective, applying assumptions on decision-maker’s preferences,
or empirically by using, for example, questionnaires, interviews, round tables, or workshops.
Which of these approaches is most appropriate depends on the IA model considered, the user
groups involved and the interaction between model builders and users.
The SEAMLESS Integrated Framework includes IA models for assessing and comparing
alternative agricultural and environmental policy options (Van Ittersum et al., 2008). The
SEAMLESS modelling team has been in close interaction with the intended users, in
504
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
particular with policy experts on the European and national level. Their uncertainty
information needs were identified by means of a questionnaire. The questionnaire investigated
the user perspective on uncertainty analysis in SEAMLESS-IF in a comprehensive way,
addressing the user’s perception of ‘uncertainty’ and ‘uncertainty analysis’, aspects that users
consider important for creating confidence in model outcomes, the location of uncertainties in
the model and on user’s preferences for uncertainty documentation. Ten completed
questionnaires were analysed.
Results and discussion
We conclude that various understandings of ‘uncertainty’ exist. This underlines the need for
model developers and model users to agree with model users on a consistent and commonly
accepted terminology before performing uncertainty analysis. While there has been general
awareness for the existence of different types and sources of uncertainties, we observe that
they do not seem equally relevant to model users. For policy experts using SEAMLESS-IF
uncertainties within the model seem to be more relevant than uncertainties associated with the
context of the problem (i.e. lying beyond model boundaries). Hence, instead of addressing
each potential type and source of uncertainties, it is probably sufficient to focus on selected
types and sources of uncertainties that are most relevant for model users.
In addition, users of SEAMLESS-IF wish to get insight into the impact of uncertainties on
key model outcomes. This indicates that uncertainty analysis is considered a means for better
understanding and interpreting model results. Hence, confidence in the outcomes of
SEAMLESS-IF can be created if uncertainty analysis responds to this need. Thus, the set of
user-relevant types and sources of uncertainties can be further limited if focusing on relevant
model outcomes. Finally, our results point to a high relevance of uncertainty communication.
Thus, technical uncertainty analysis needs to be accompanied by regular and transparent
information of model users about (planned) modelling activities.
References
CEC, 2004. Service contract for Review of the RAINS Integrated Assessment Model –
Reference ENV.C1/SER/2003/0079. European Commission, Brussels.
Gabbert, S., 2008. J. of Env. Ass. Pol. And Managem. (submitted)
Gough, C., 1999. Int. J. of Env. and Poll. 11: 462-478.
Harremoës, P. & R.K. Turner, 2001. Reg. Env. Change 2: 57-65.
IIASA, 2002. Uncertainty Treatment in Integrated Assessment Modelling: Conclusions from
a workshop held at IIASA in Laxenburg (Austria), 24-25 January 2002. IIASA,
Laxenburg, Austria.
Jakeman, A.J. & R.A. Letcher, 2003. Env. Mod. & Software 18: 491-501.
Janssen, P.H.M., et al., 2005. Water Sc. & Techn. 52: 125-131.
Sundqist, G., et al., 2002. Env. Science & Policy 5: 147-156.
Van Asselt, M.B.A., 2000. Perspectives on Uncertainty and Risk. The PRIMA Approach to
Decision Support. Kluwer Academic Publishers, Boston.
Van der Sluijs, J.P., 1997. Anchoring amid uncertainty. On the management of uncertainties
in risk assessment of anthropogenic climate change. CIF-Gegevens Koninklijke
Bibliotheek, Den Haag.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
Walker, W.E., et al., 2003. Integr. Assessm. 4: 5-17.
505
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Modellers, model results and local stakeholders – dynamics in combining
expert and lay knowledge in locally grounded policy formulation
A. Jonsson, J. Alkan Olsson, L. Andersson
Centre for Climate and Policy Research, Department of Water and Environmental Studies,
Linköping University, 601 74 Norrköping, Sweden
Contact: Anna.jonsson@tema.liu.se
Introduction
The involvement of stakeholders is today seen as crucial for the mitigation of environmental
problems (Jonsson et al., 2005). This is also acknowledged by the EU directive on water
(Council of the European Communities, 2000) which strongly emphasize the need for
increased stakeholder participation in water management. It is argued that public participation
may contribute to achieve several interrelated goals such as better-informed stakeholder
groups, transfer of knowledge from local stakeholder groups to regional and national planning
authorities, socially accepted mitigation measures (leading to higher efficiency in
implementation) and reduction of conflicts between stakeholder groups (Pahl-Wostl, 2002).
At the same time models are more and more frequently used in this natural resource
management which to a certain extent is a contradiction as models often lead to an increased
use of expert knowledge (Alkan Olsson & Berg, 2005).
The aim of this paper is to assess this possible contradiction and to discuss aspects of
combining different types of knowledge in the formulation of a locally suggested remedy plan
for water management along EU WFD ambitions. Specifically, the case concerns a project
aiming at the development of a methodology for participatory modelling (DEMO) as a way to
involve farmers and other local stakeholders in water resource management.
Methodology
The methodology is based on dialogues between stakeholders and researchers, where the setup, running and presentation of results of a catchment model (HBV-NP) (Arheimer & Brandt,
1998), served as a platform for discussion. Issues dealing with understanding present
conditions and formulating a locally suggested remedy plan to reduce nutrient transport to the
Kaggebo Bay in the Baltic Sea were in focus throughout the process. Included stakeholders
were farmers, rural households (permanent and summer cottages), municipal environmental
officers, hydrological modelling experts, and process leaders with a background in social
science. The paper presents results from a sequence of around 50 stakeholder meetings
involving around 100 different stakeholders as well as results from 45 individual interviews
with involved stakeholders. The focus is to understand whether involved stakeholders saw the
presented model results as useful or not, in what way and how this helped in the process of
formulating the remedy plan. Moreover, the resulting dynamics between involved experts, lay
stakeholders and the model itself are discussed.
Results and discussion
The involved stakeholders perceive the employed model as a useful in relation to three issues.
Firstly, they appreciated the models ability to give a general picture of the nutrient status in
the drainage area. Several interviewees argued that the discussions around the models gave a
clearer picture of what the problem and who the polluter is. Some interviewees also argued
that the discussion on the pollution status created a feeling that “this is our problem and we
have to solve it”. In the discussions on source-appointment it was found that the emission of
phosphorous from rural households not connected to wastewater treatment plants were
relatively high, this information changed the stakeholders’ understanding of who actually is
the polluter in the drainage area. This new insight was highly motivating for the involved
506
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
farmers who before had seen their agricultural activities as the major polluter. On several
occasions source-appointments provided by the model clarified the contribution of pollutants
from the traffic which by some stakeholders were seen as larger than it actually was.
The second issue is related to the models capacity to estimate the needed quantity of
reduction to achieve locally as well as nationally defined goals both at a sub-catchment level
as well as in the total drainage area. The participating stakeholders argued that the pollutant
load estimated by the model and compared with national defined goals helped them to create
a realistic and for them legitimate definition of local goals for reduction.
The third issue relate to the measure packages prepared using the model. In these packages
the needed reductions to achieve nationally and locally defined reduction goals were related to
actual measures. However, when it came to the selection or priority making of measures it
became obvious that there were many other factors influencing stakeholders’ decisions such
as the easiness to implement the measure, the cost of implementation as well as the possible
economic benefit for implementing it as some of the farm measures were linked to money
contributions for example buffer strips and the creation of wetlands. Moreover, equality
between stakeholders in the drainage area regarding the burden of implementing measures
was seen as more important than finding the most cost-efficient spot for implementation. The
exception was the construction of wetlands, were stakeholders signaled a will to discuss
environmentally efficient locations provided that they were technically and economically
feasible and that the concerned land owner(s) were willing to engage in such a project.
The model employed in this particular process has assisted in creating a consensus around
the nutrient pollution problem in the drainage area. It has put environmental changes in a
tangible spatial and temporal perspective which may serve as a basis for effective planning
and mitigation. It has also assisted in increasing the tolerance and understanding between
stakeholder groups, which indirectly may decrease future conflicts in relation to
implementation of measures. In this particular study it is evident that the decision on which
measures to implement and were to implement them were also influenced by factors outside
the scope of the model. Most stakeholders concluded that it was the process in itself that had
been influential on their understanding and possible future action. One major problem as to
the dynamic between the involved stakeholders, scientists and modellers were the timing of
the delivery of results. Stakeholders facing model results were often spurred to ask new
questions that often required immediate additional runs of the model which was not possible
in this case. Moreover as this particular model do not have a ready user interface consequently
a substantial amount of after preparation of model results was needed and therefore the time
issue was also important for the involved scientists. This indicates that the model use cannot
be separated from the employed participatory methodology or the policy context in which the
process takes place. Hence the usefulness of the model in aiding to formulate locally
suggested remedy plans is highly dependent on the ways in which the model is used as well as
how easily accessible the model result is. However compared to traditional use of models in
environmental decision making the experts’ role was radically transformed from a one-way
communication of final results to assistance in various steps of a participatory process.
References
Alkan Olsson J. & K. Berg, 2005. Ambio 7: 507-512.
Arheimer, B. & M. Brandt, 1998. Ambio 27(6): 471-480.
Council of the European Communities, 2000. Directive 2000/60/EG of the European
Parliament and of the Council of 23 October 2000 establishing a framework for
Community action in the field of water policy. Official Journal of the European
Communities, 22.12.2000, L327/1.
Jonsson, A., et al., 2005. Ambio 7: 489-494.
Pahl-Wostl, C., 2002. Integrated Assessment 3(1): 3-14.
507
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Transition towards more adaptive irrigation management: the impact of
innovative technologies: A case study in Caia, Guadiana River Basin
Vanja Karadzic, Paula Antunes, Rui Santos, Anna Osann
CENSE - Ecological Economics and Environmental Management Group,
Faculty of Sciences and Technology, New University of Lisbon,
Campus da Caparica, 2829-516 Caparica, Portugal
Contact: v.karadzic@fct.unl.pt
Irrigation management has to become more adaptive to the increasing pressures and
uncertainties coming from forces of major global climate, social, economic and political
changes. While ‘adaptation’ could be defined as a change within the given regime structure
and management paradigm, ‘transition’ refers to change in management paradigm,
accomplished through structural changes of dominant practices, rules, shared assumptions,
interests and beliefs that underlie political policy (Rotmans et al., 2001). In this paper, we
hypothesize that innovation technologies, in the case of irrigation management in Portugal,
could contribute to transition towards more adaptive water management (Bogliotti C. & M.
Todorovic, 2007).
The context for this work is a research and technological-development project PLEIADeS
(Participatory multi-Level EO-assisted tools for Irrigation Water Management and
Agricultural Decision-Support), developed with the overall objective of improving the
technical, environmental and economic performance of irrigation schemes by means of New
Technologies (NT). We selected the case study of Caia, Guadiana River Basin, Portugal,
because it was used as a pilot zone for many local and European projects, through which the
Associação de Beneficiários do Caia (Caia Water District Management Board) developed
technological skills in GIS and automatic management of irrigation and pumping systems.
Furthermore, within the scope of the PLEIADeS project local stakeholders have been
involved since the beginning in the design of the new irrigation technology, thus they are not
expressing confusion and aversion towards it.
Transition Management (TM) and Strategic Niche Management (SNM) are explored as
analytical frameworks for understanding potential transition dynamics. We use transition
multi-level model (Geels, 2002) to reflect on global pressures (climate change, EU water and
agricultural policies, cultural trends) influencing water management regime (dominant
practices, rules, procedures, embedded into infrastructures and institutions). As an example
climate change is inducing a new paradigm to water management: moving from responsive
mode to climate change concerns towards developing and investing in programs targeted at
climate change, opening opportunities for innovations and adjustment of organizational
structures. Other global trends are analysed in a similar way. SNM (Hoogma et al., 2002) is
used to analyse the introduction of PLEIADeS irrigation technology into the Caia pilot
irrigation management setting. We do this on the basis of three internal niche processes:
articulation of expectations, building of social networks and learning processes. According to
the SNM concept, the interaction between these processes forms the foundation for
understanding success or failure of innovation technology (Raven, 2005). Using internal niche
dynamics, characteristics of the PLEIADeS irrigation technology are explained, as well as its
contribution to the achievement of the ‘good level’ in these processes. We focus on this
connection (global pressures-innovations), while striving to design NT implementation
strategy on how to manage specific innovation project, with a focus on its learning potential
and design of reflexive learning processes (see Figure 1). Analysed impacts of innovation
technology would lead to the answer what kind of change is demanded on the regime level,
taking into account macro (global) and innovation niche pressures.
508
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
2.
Building social
networks
1.
Concrete expectations
Coherent vision
Connection: different
stakeholders
Initial work with stakeholders
Articulating expectations
3.
Information
gathering
Learning goals
NT specifications
NT user requirements
4.
Environmental
Social
Economic
Technical
Practical
application
Monitoring
Evaluation
Goals Revision
New user requirements
Technology established
Figure 1. Reflexive design of learning processes in the context of PLEIADeS / SPIDER
technology.
References
Bogliotti C. & M. Todorovic, 2007. Identifying research and development targets for the
management of limited water resources in the Mediterranean: Background and a brief
outline of the WASAMED case. Presentation at the Int. Conf. Water saving in Mediterranean agriculture & Future Research Needs, Valenzano, Italy. WASAMED Project.
Geels, F.W., 2002. Research Policy 31: 1257-1274.
Hoogma, R., et al., 2002. Experimenting for sustainable transport: the approach of strategic
niche management. Spon Press, London and New York.
Raven, R.P.J.M., 2005. Strategic niche management for biomass. PhD thesis, Technische
Universiteit, Eindhoven.
Rotmans, J., et al., 2001. Foresight 3: 1-17.
509
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Prospecting the behaviour of agro-ecosystems: experiences in using
FALLOW model with natural resource managers
Betha Lusiana, Desi Suyamto, Rachmat Mulia, Meine Van Noordwijk, Georg Cadisch
Institute for Plant Production and Agroecology in the Tropics and SubTropics,
University of Hohenheim, Garbenstr. 13, Stuttgart 70559, Germany
Contact: B.lusiana@cgiar.org
Introduction
Agro-ecosystems involve trade-offs between multiple functions related to food security,
environmental services, stakeholders’ interest, policy interventions and technological
innovations. Integrated Natural Resource Management requires a dynamic and efficient
approach to help natural resource managers and policymakers in assessing the various
plausible options in managing the landscape and finding ways to balance the immediate
human needs and ecosystems functions. Negotiation support tools can help facilitate
stakeholders to prospect and discuss the behaviour of agro-ecosystems. The use of simulation
models and scenario analysis in negotiation support tools is an efficient approach to conduct
such dynamic trade-offs analysis with scale-dependent properties. But the challenges remain:
(i) Can envisioning tools such as simulation models and scenario analysis help decision
makers and natural resource managers to explore plausible options effectively and
efficiently?, and (ii) What are the essential factors for a simulation model or model results to
be valuable for natural resource management in decision making?
Methods
Figure 1 presents the steps of activities conducted in this study. To explore the model users’
perceptions of a potential model to be used, surveys were carried out in Indonesia, Philippines
and Kenya focusing on salience, credibility and legitimacy of model simulation, scenarios
performed and output results (McNie, 2007).
Figure 1. Steps of activities to evaluate stakeholders’ perceptions of using simulation model
for natural resource management.
In step 2, in order to prospect the impacts of possible natural resource management options,
the landscape simulation model FALLOW was used together with potential model users
(natural resource managers/policymakers and researchers) in West Aceh, Sumatra, Indonesia.
The FALLOW model was developed as an impact assessment tool at landscape level to help
integrate understanding of landscape-mosaic resource interactions (Van Noordwijk, 2002).
The model was applied to explore the trade-offs between carbon stocks and livelihoods in
forested and non-forested landscapes (Van Noordwijk, et al., 2008).
A stakeholders’ discussion was then held to gain model users’ feedback and evaluation on
model results presented, including scenario analysis used.
510
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
Results and discussion
In step 1, survey result based on 115 respondents of potential model users (52% of
lecturers/researchers and 34% natural resource managers/policymakers) revealed that salience
and credibility of simulation model are considered as more important characteristics than
legitimacy (Table 1).
Table 1. Users’ perspectives of simulation model.
Rank Criteria
Clear and understandable theory and processes underlying the model
1
Model output is useful and applicable for managing natural resources
Model output has similar patterns to what is observed in the field
2
Model is easy to use and parameterize
3 Model output is attractive and easy to understand
4 Model was developed by well known scientist
5 Model has previously been used by policymakers
Salience and
Credibility
Legitimacy
We tested these findings further by using the model to simulate a study site of 315 km2 in
West Aceh and presented its results to stakeholders (step 2 and 3) We particularly focused on
results of land use dynamics and its impact on livelihood (farmers’ welfare, indicated by nonprimary expenses) and carbon sequestration based on baseline (‘business-as-usual’) condition
and with natural resource management interventions (Figure 2).
Figure 2. Relative increment
(compared to current baseline) of
carbon stocks and farmers welfare
under scenarios of improving
smallholder
rubber
systems
(Rubber), improving smallholder oil
palm plantation (Oil Palm) and
protecting 50% of existing forest
(Forest Conservation), based on 25
years on simulation.
This activity is on-going. The information we will gain from stakeholders are (i) how easy can
they understand the modelling results, (ii) how much do they trust the results, (iii) how
relevant do they find the scenarios applied are for them and (iv) how useful are the results in
making decisions for natural resource management. Results of this outcome will be an
essential basis for informed discussions among natural resource managers in land use
planning of West Aceh landscape as well as for further model development.
References
McNie, E.C., 2007. Environmental Science and Policy 10: 17-38.
Van Noordwijk M., 2002. Ecological Modelling 149: 113-126.
Van Noordwijk, M., et al., 2008. Agriculture, Ecosystem and Environment 126: 98-112.
511
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Learning by doing: Designing and conducting impact assessment studies for
citrus Farmer Field Schools in Vietnam
O. Nicetic, E. Van de Fliert, Ho Van Chien, R. Spooner-Hart, Z. O’Leary, D. Rae
University of Western Sydney, Hawkesbury Campus, Richmond, NSW Australia
Contact: o.nicetic@uws.edu.au
Introduction
The Farmer Field School (FFS) is not just an extension method; it rather serves as a platform
for adaptive research and experiential learning to address sustainable development of a
complex agro-ecosystem, like that in citrus production. Actors of the platform include a
variety of users: farmers; extension and technical personnel of government departments; nongovernment organizations and private companies; scientists from universities and research
institutions; and government officials involved in policy making. Consequently, assessment of
impact, and particularly sustainability aspects related to it, is very complex since it should not
focus on farmers involved in the process alone, but on a suite of beneficiaries and the broader
community. Ultimately, the results of the impact assessments and verdict on effectiveness of
FFS will depend on who was included as an ‘object’ of the assessment. A further dilemma is
who should conduct the evaluation: an ‘objective’ outsider with limited grounded knowledge
about the underlying principles of the approach, or an insider with sufficient insight but who
is potentially ‘subjective’ (Van den Berg & Jiggins, 2007). Another issue is precisely what
should/can be measured. While economic indicators are commonly used, it can be
extraordinarily difficult to identify and quantify all costs, as environmental and social
indicators are often limited due to difficulties in capturing change and the cost involved in
rigorous evaluation (Bartlett, 2005; Fleischer et al., 2004).
We could state many more uncertainties about how to conduct impact assessment that
leads us to conclude that there is no defined methodological protocol for assessing FFS.
Differences in impact assessment methodology rather than differences in impacts themselves
are a probable reason why several studies on impact and cost effectiveness of FFS had
positive results, while some others come to less positive conclusions (Feder et al., 2004).
This paper reports on impact assessment processes and outcomes of an AusAID CARD
funded project conducted in 2005 and 2006, which initial goal was to implement IPM in citrus
using the FFS approach. However, the actual interaction between project stakeholders
(including Vietnamese and Australian researchers, trainers and farmers) during the course of
FFS lead to a total reformulation of the citrus IPM strategies and practices. Farmers’ opinions
and experiences in the first cycle of the learning process seemed more appropriate under the
prevailing conditions than the methods determined by researchers in the initial FFS design.
These interactive participatory learning processes that all project stakeholders went through
became equally important in terms of capacity building as the implementation of the IPM FFS
itself. We sought to assess the effectiveness of FFS in capacity building of all stakeholders
taking in account impact of the process of development of new practices, not only the impact
of the changed practices itself.
Methods
Impact assessment was done by the project management team and includes three major parts:
(a) analysis of knowledge, attitudes and practices (KAP) of farmers, pre and post FFS
attendance, (b) analysis of economic, social and environment impact as a result of the
changed practices 12 months after farmers completed FFS and (c) impact of participating in
the FFS process on stakeholders themselves and their social environment. The KAP survey
was conducted with 710 farmers in 2005 and 1,659 in 2006 from 12 provinces in the Mekong
Delta and Central coastal provinces of Vietnam. Impacts were assessed using semi-structured
512
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
interviews and group discussions with farmers that allowed the team to identify changes in
their agricultural practices, major economic impacts and changes in their environment.
Farmers were also asked to describe impacts of participating in FFS on their family life and
interaction with community. A total of 60 farmers were interviewed from 9 provinces.
Originally, only KAP analyses were planned to be used in assessment of FFS effectiveness.
During participatory evaluation of the process at the end of the first year it was, however,
concluded that results of KAP analysis did not capture the essence of the FFSs impact on
farmers and their community. The project team together with trainers then developed other
assessments tools including semi-structured interviews and group discussion in an attempt to
improve the assessment.
Results and discussion
The KAP analysis showed that attitudes of farmers were influenced by participation in FFS
particularly in relation to pest control methods. Farmers increased their level of agreement
that pesticide can cause pest resurgence and agreed less to the statements that application of
pesticide will increase the yield, that pesticide are cheap and easy to use, and that advanced
farmers use a lot of pesticide. Farmers’ knowledge about pests and diseases significantly
improved as a result of participation in FFSs, with a significant increase in number of farmers
giving correct answers recorded across all provinces but with steeper increase at Central coast
due to the lower level of knowledge at the beginning of FFSs. As citrus is a perennial crop
with a year-long growing season, most practices were not possible to change within the
timeframe of FFS. Only change of practice recorded was reduction of number of sprays in
Mekong delta from 7 to 6.5 and from 7.7 to 6.0 in 2005 and 2006, respectively.
Major changes in practices recorded in interviews with farmers a year after FFS
completion were visible in the increased use of compost and manure followed by a change in
number of pesticide spray used (slight decrease) and a significant change from use of broad
spectrum pesticides (primarily synthetic pyrethroids) to less disruptive pesticide like mineral
spray oils and imidacloprid. Major economic impact was reduction of input costs. It was
estimated that cost of FFS per participant represented only 1.60% of their net annual profit.
Major environmental impacts were the increase of beneficial arthropods and an increase in
abundance of fish in canals.
Participation in FFS raised confidence in ability of participants to manage citrus agro-eco
system. It improves relationship between farmers who participated in FFS and increased their
influence in community. It increases activities in growers clubs that resulted in formation of
several cooperatives. Attendance in the FFS assisted in transition of farm management from
father to son (4 cases) and daughter (1 case) and from husband to wife (3 cases). Even though
by combining more objective and robust KAP analysis with more subjective and inherently
biased semi-structured interviews and group discussions, we captured a wide range of impacts
and developed rich picture of FFS impacts. These, however, could not be precisely quantified
which confirms the complexity of the assessment process and need for development of
interactive, participatory assessment tools that will be able to quantify impacts of FFS.
References
Bartlett, A., 2005. Farmer field schools to promote integrated pest management in Asia: The
FAO experience. Proceedings of the ‘Workshop on Scaling Up Case Studies in
Agriculture’, International Rice Institute, 16-18 August 2005, Bangkok.
Feder, G., et al., 2004. Review of Agricultural Economics 26: 45-62.
Fleischer, G., et al., 2004. In: Eds W. Rivera & G Alex, Demand-driven approaches to
agricultural extension: Case studies of international initiatives. Agricultural Development
Discussion Paper 10: 40-49.
Van den Berg, H. & J. Jiggins, 2007. World Development 4: 663-686.
513
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
SEAMLESS training material on concepts and models for Integrated
Assessment of agriculture and sustainable development
P. Reidsma, J. Alkan Olsson, I. Bezlepkina, S. Brogaard, J. Queste,
R. Fischer, M.K. Van Ittersum
Plant Production Systems, Wageningen University,
P.O. Box 430, 6700AK Wageningen, The Netherlands
Contact: pytrik.reidsma@wur.nl
Introduction
Concerns about environmental problems, food security and climate change have put
sustainability in agricultural development prominently on policy agendas. There has been a
shift from supporting agricultural production towards policies supporting sustainable (rural)
development in a broader sense. Hereby, important drivers of change in agricultural systems,
(e.g., globalization, liberalization, environmental policies, climate change) request integrated
analyses considering the full set of natural, economic, social and institutional aspects of
sustainability at multiple scales (i.e. field, farm, region, market and global levels). Such
integrated assessment of agricultural systems requires integration of knowledge from different
disciplines.
Over the past years, there tends to be an increase in large research projects with the aim to
develop integrated assessment tools for various domains. One example is a large European
research consortium that has developed an integrated modelling framework to support
analysis of relationships between agricultural systems and sustainable development:
SEAMLESS-IF (System for Environmental and Agricultural Modelling; Linking European
Science and Society – Integrated Framework; Van Ittersum et al., 2008). The SEAMLESS-IF
is a computerized integrated framework that assesses and compares ex-ante, alternative
agricultural and environmental policy options, allowing (1) analysis at the full range of scales
(farm to EU and global), (2) analysis of the environmental, economic and social contributions
of a multi-functional agriculture towards sustainable rural development and (3) analysis of a
broad range of issues, such as environmental policies and liberalization.
Integrated assessment projects require a new generation of scientists with strong
integrative skills, both conceptually, methodologically and technically. Such scientists may
have a strong disciplinary background supplemented with inter- and transdisciplinary skills or
they may have a mainly interdisciplinary training. Developing ‘T-shaped skills’ (broad
scientific overview combined with in-depth knowledge of specific subjects) allows for
scientific, communicative and co-operative flexibility (Bouma, 1997). We anticipate that new
courses and curricula are needed which on the one hand capitalize and disseminate
experiences and capacity from large integrated assessment projects and on the other hand train
a new generation of scientists. This contribution reports, as an example, modular training
material developed from the SEAMLESS project. This training material can be used for
undergraduates, postgraduates and experts from research and policy institutes, and parts of it
can also be useful for discussions with other stakeholders.
Training and courses
SEAMLESS training material is based on a modular structure (Figure 1; Reidsma et al.,
2008). The underlying principle is that several modules can be flexibly combined together and
some specific ones can be added depending on the course duration, intensity and the audience
to easily produce ‘à la carte’ courses. One of the courses developed is the post-graduate
course of one week, which can be considered as an example of how to use the SEAMLESS
training modules to create a course.
514
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
The objectives of the course are (1) to present concepts for integrated assessment of
agricultural systems, (2) to gain theoretical and practical understanding of the methods,
models and tools used in integrated assessment of agricultural systems, (3) to understand how
integrated assessment and modelling can support ex-ante impact assessment and decisionmaking processes and (4) to understand how own specific research relates to an integrated
assessment and modelling perspective.
In the course, SEAMLESS-IF and its research tools are used as an example to present how
concepts and models can be integrated to assess complex agricultural systems. The course is
problem orientated, so all lectures are linked to practical applications, such as environmental
policies within the Nitrate Framework Directive or trade policies.
The post-graduate course is an advanced course and can form a basis for researchers that
want to learn about methods for integrated assessment and/or that want to continue working
with the framework or with individual components (advanced level). When training policy
experts or other stakeholders the focus will be more on basic and intermediate level modules.
The courses and their set-up will be evaluated, which can improve training material and
identify gaps in SEAMLESS-IF.
4.1.1 Project definition
4.2.1 Model chains
4.3.1 Visualization
4.1.2 Indicator manager
4.2.2 Calculations
4.3.2 Interpretation
4.1.3 Experiment design
4.3.3 Sensitivity analysis
4.1 Pre-modelling
4.2 Modelling
4 Using SEAMLESS-IF
2 Theoretical approach
4.3 Post-modelling
3 Practical Illustrations
1 SEAMLESS in a nutshell
2.1 Sustainable Development
3.1 Trade liberalization
3.2 Nitrate Directive
2.2 Impact Assessment
2.3 Using models in policy making
2.4 Complex System Analysis
6 Data
5 Individual components
in general
5.1 CAPRI
6.1 Typologies
7 Indicators
7.1 Indicator
framework
5.2 EXPAMOD
7.2 Up-scaling
6.2 SeamDB
5.3 FSSIM
5.4 APES
5.5 GTAP
Policy experts
Policy experts /
Integrative modellers
Basic level
Intermediate level
5.6 Employment
model
5.7 Landscape
explorer
5.8 PICA
Agricultural
modellers
Advanced level
5.9 SEAM
FRAME
5.10 Ontologies
Figure 1. The modular
structure of the SEAMLESS
training material.
References
Bouma, J., 1997. Geoderma 78: 1-12.
Reidsma, P., et al., 2008. PD7.5.2, SEAMLESS Integrated Project.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
515
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Serious Gaming as a tool to create sustainability awareness, discussion and
negation among stakeholders in cattle systems, the case of Los Angelitos1
E.N. Speelman, L.E. García-Barrios
Biological Farming Systems Group, Wageningen University,
Marijkeweg 22, 6709 PG Wageningen, The Netherlands
Contact: erika.speelman@wur.nl
Introduction
Ever since the Brundtland Report, sustainability has been on many agendas in a wide range of
disciplines. In the field of Natural Resource Management Systems (NRMS), sustainability has
become a prime focus point of researchers, NGO’s and governments. However, reaching
sustainability is not an easy task. NMRS are usually highly complex systems in which a wide
range of stakeholders at different scales pursue different goals from the system and evaluate
system’s performance with a stakeholder-specific set of indicators. The social process of
NRMS has become increasingly complex, interdependent and uncertain (Gunderson, 1999).
Recently, new methods have been developed to search together with stakeholders for more
sustainable management, and to stimulate interaction among stakeholders to help to frame and
reframe problems. Simulation models, especially those that combine computer simulation and
role-playing have shown to stimulate learning and discussion (see www.commod.com;
Speelman & García-Barrios, 2006; García-Barrios et al., 2008). Role-playing can have
significant influence on the way players will behave in the future (Gurung et al., 2006). In
general, human beings have problems dealing with complexity and tend to use short-term,
adaptive and ad hoc problem solving strategies that are successful only within certain limits
(Dörner, 1997). These ‘serious games’ can help people to develop a better sense of scoping
and consensus building when dealing collectively with complex systems.
The state of Chiapas, Mexico is one of the biodiversity hotpots in Mexico. The Sepultura
reserve was created to preserve this natural richness. However, the establishment of the
reserve has limited farmers, who live in the reserve, in expanding their farming areal. In order
to maintain their livelihoods these farmers have now moved to highly unsustainable forms of
land use, i.e., intense cattle herding. Trees have virtually disappeared from farming areas. This
in combination with the high grazing pressure has lead to very little soil cover at the
beginning of the rainy season; resulting in very high erosion levels.
We have recently started developing a participatory modelling project in the region. The
project aims for bringing together the relevant stakeholders from different scales to discuss
the challenges that the area is facing and to develop (simulation) tools that aid the social
learning process. Within this effort, the modelling game named Los Angelitos (Speelman &
García-Barrios, 2008) was developed.
Methods
The model was developed in Netlogo 3.1 (Wilensky, 1999). A micro-watershed suitable for
the game was selected from an areal photograph of La Sepultura. The image was uploaded in
Netlogo to create the landscape of the game, as seen in the user interface (Figure 1). Simple
minimalistic modelling was used to develop the submodels, included some published plant
growth and grazing models, i.e. García-Barrios et al. (2008).
Los Angelitos combines role-playing with simulation and elements of board games. Users
take on the role of farmers. They select and appropriate an area; after which they decide their
land use, i.e., forest, maize or grass, and the number of cows. Users and the systems are
subject to perturbations in the form of pop-up cards.
1
This demo presentation will show the development of the Netlogo game ‘Los Angelitos’ in Los Angeles,
Chiapas, Mexico. The work is still ongoing.
516
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
(a)
(c)
(b)
Figure 1. User Interface of the simulation game ‘Los Angelitos’ with (a) buttons and sliders
controlling the simulation and the individual fields, (b) view of the virtual landscape (users
can select from a range of different views such as land use (shown here), elevation (d), runoff
(e), and erosion (f), and (c) graphs showing different outputs of the simulation.
Results and discussion
Some first model-stakeholder interactions showed that farmers from the local communities
were very positive about the game. They were instantly intrigued by how decisions of
neighbour stakeholders affected their fields and the system as a whole. As part of the project a
modelling workshop was organized in July, 2008. During this workshop (inter)national
researchers and local stakeholders interacted and discussed the future of serious games. The
workshop and especially the input of farmers profoundly helped the further development of
the model.
References
Dörner, D., 1997. The logic of failure. recognizing and avoiding error in complex situations.
Addison-Wesley, Reading, Massachusetts, 222 pp.
García-Barrios, L.E., et al., 2008. Ecol. Mod. 210: 115-126.
Gunderson, L., 1999. Conserv. Ecol. 3 (1): 7.
Gurung, T.R., et al., 2006. Ecol. Soc. 11 (2): 36.
Speelman, E.N. & L.E. García-Barrios, 2006. Proceedings Agent 2006, pp. 289-305.
Speelman, E.N. & L.E. García-Barrios, 2008. ‘Los Angelitos’, unpublished-software.
Wilensky, U., 1999. “Netlogo”, Center for Connected Learning and Computer-based
Modeling, Northwestern University. Evanston, IL.
517
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Understanding and modelling socio-ecological networks to support
multi-scale governance of sustainable dairy farming systems
E.N. Speelman, J.C.J. Groot, H. Van Keulen, K. Kok, W.A.H. Rossing
Wageningen University- Plant Sciences group, Biological Farming Systems Group,
Marijkeweg 22, 6709 PG Wageningen, The Netherlands
Contact: erika.speelman@wur.nl
Introduction
The performance of farming systems is increasingly evaluated for new functions and services
in addition to their primary function of producing food and fibres. In addition, with increased
pressure on resources and increased connectivity between scales, relationships between
stakeholders have become more apparent and more intense. The seemingly independent
actions of different stakeholders involved can produce unwanted and uncontrollable consequences for all (Bouwen & Taillieu, 2004), which contributes to conflict and separation
between social actors and groups of stakeholders. This has often led to conflicts (e.g.,
Senegal: D’Aquino et al., 2003; Mexico: Speelman et al., 2006; Bhutan: Gurung et al., 2006).
The knowledge required to search for alternative management is usually not held by one
social actor, instead each stakeholder has her or his own version of reality; joining these
partial realities aids to frame and reframe problems (Spector & Anderson, 2000).
(a)
(b)
Figure 1. (a) Conceptual framework for the analysis of linked social-ecological systems.
Local ecosystems are nested within other ecosystems. Management practices are embedded in
institutions, and these can be nested within other sets of institutions. Modified from Berkes &
Folke (2002) and Olsson (2003). (b) Hypothetical network of social (blue), farm (white) and
natural (green) entities.
The way by which stakeholders are connected and how stakeholders’ actions influence, affect
or even determine underlying natural entities and other stakeholders (Figure 1) is not yet fully
understood. Currently, an increased interest has risen in network theory and so far many types
of systems have been subject to investigation with methods based on network theory e.g.
proteins, food webs, organizations. Network characteristics, structure and motifs have been
identified in various studies. However, studies of socio-ecological network are still rare. A
fuller understanding of socio-ecological networks in farming systems (see Figure 1a) can aid
research and implementation of more sustainable management options. For instance,
understanding an innovation network can guide the dissemination process of promising
alternative management.
Recently, tools aiming for bringing relevant stakeholders together for social learning,
discussion and negotiation have been developed which combine tools from both social and
environmental sciences, e.g., role-playing and modelling. A site-specific approach
‘Companion modelling’ (Barreteau et al., 2001) and a generic tool “Negotiated design of
518
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
Sustainable Production Systems among Social Agents with Conflicting Interests” (GarcíaBarrios & Pimm, 2005) have been developed. Both approaches have proven through
widespread implementation (see www.commod.org and García-Barrios et al., 2008) to be
able to create an interactive setting in which stakeholders learn, discuss and negotiate
important issues and future management.
However, both approaches focus mainly at stakeholders of one scale, whereas moredifficult-to-solve conflicts often arise between stakeholders of different scales. As their partial
realities are more likely to differ dramatically, making communication more challenging.
Creating a setting in which stakeholders from different scales meet, discuss and play with a
simulation model is therefore very important. The current approaches demand for the
development of tools that are suitable for a wide variety of users from low to higher
governance levels, e.g., farmers and regional managers. Additionally, the current site-specific
approach is very costly and has a small reach whereas the generic tool is less costly and has a
larger reach, but is too abstract for use at lower governance levels. A new tool yet to be
developed should therefore be ‘medium-generic’, meaning that it should sufficiently reflect
stakeholders’ reality, but at the other hand sufficiently generic to be interesting and appealing
to other similar stakeholders in other areas who fight similar issues. The innovative aspect of
the proposed research is focused on filling this gap between the current approaches.
Methods
The proposed research consists of developing a ‘medium-generic’ simulation tool appealing
to a wide range of stakeholders aiming for social learning, information dissemination,
discussion and negotiation. As far as we know, no efforts of this kind have yet been made in
dairy systems nor in socio-ecological systems. The tool or toolbox to be developed will be
based on solid literature review as well as a case study, to ensure that it will address a more
wide-spread problem which is also present for analysis in a specific case. Therefore, this
study will focus on a case study of dairy systems in an area in the state of Michoacán,
Mexico. The research will consist of three main components: (a) data collection, (b)
participatory modelling, and (c) application of the modelling tool or toolbox. During the data
collection phase, information on (i) common problems in dairy farming systems, (ii) current
relevant modelling approaches will be gathered, and (iii) case study specific information
concerning biophysical and social networks. After which a participatory modelling process
with start in the case study area. The applicability outside the borders of the specific case
study will be assessed through its application in two more areas.
References
Barreteau, O., et al., 2001. J. Artif. Soc. Social Simul. 4(2): 5.
Berkes, F. & C. Folke, 2002. In: Eds L.H. Gunderson & C.S. Holling, Panarchy:
Understanding transformations in human and natural systems. Island Press, Washington,
DC, pp. 121-146.
Bouwen, R. & T. Taillieu, 2004. J. Commun. Appl. Social Psychol. 14: 137-153.
D’Aquino, P., et al., 2003. J. Artif. Soc. Social Simul. 6(3): 5.
García-Barrios, L.E. & M. Pimm, 2005. Negotiated design of sustainable production systems
among social agents with conflicting interests. Software.
García-Barrios, L.E., et al., 2008. Ecol. Mod. 210: 115-126.
Gurung, T.R., et al., 2006. Ecol. Soc. 11(2): 36.
Olsson, P., 2003. Doctoral Thesis, Stockholm University, Sweden, 37 pp.
Spector, M. & T. Anderson (Eds.), 2000. Integrated and Holistic Perspectives on learning,
Instruction and Technology: Understanding Complexity. Kluwer Academic Publishers,
Dordrecht, 300 pp.
Speelman, E.N., et al., 2006. Outlook Agric. 35: 57-64.
519
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Specifying impact assessment simulations in interaction with end-users:
The example of SEAMLESS-IF
M. Taverne, O. Therond, J.P. Bousset, J. Alkan Olsson, F. Ewert, M.K. Van Ittersum, J. Wery
Cemagref, 24 Avenue des Landais - BP 50085,
63 172 Aubiere Cedex, France
Contact: marie.taverne@cemagref.fr
Introduction
The SEAMLESS Integrated Framework (IF) aims at assessing ex-ante impacts of agricultural
and environmental policies and technical innovation with modelling chains that enable the
simulation of biophysical, economical and social changes in processes related to agriculture
(Van Ittersum et al., 2008). End-users and experts of SEAMLESS-IF will have to interact for
the definition of simulations to be performed within this integrated assessment and modelling
framework. Indeed, only the end-users can define the issue of the impact assessment at stake,
the changes in driving forces (policies or agricultural practices) to investigate, the concerned
contexts, the critical uncertainties and the domains requiring impact assessment (Alcamo,
2008). Due to the complexity of this framework, a thorough knowledge of the framework, its
model chains and the underlying assumptions are required to define the simulation
experiments. As a consequence, end-users are unlikely to be able to use this framework
without support from framework experts.
Several degrees of intensity of interaction with users are foreseeable. This is because the
experts of the framework are not equally skilled to understand the problems expressed by endusers, to make end-users’ assumptions and expectations more explicit and to share their
expertise of the framework. At the same time end-users do not have equivalent ability to
understand the way the framework works, including the model assumptions and variables
(inputs, outputs, parameters). And further, the translation of a problem defined by end-users
into simulation parameters (models inputs) by a framework expert could create a lack of
transparency if there are no guidelines and visualization techniques that are easy to use to
support the discussion (Alcamo, 2008).
The purpose of this contribution is to present tools, which may be useful to assist in the
process of interaction between end-users and experts of the framework, to specify simulations
in SEAMLESS-IF enabling well qualified impact assessments.
Methods
Two tools that can contribute to the joint specification of simulations were developed and
used during the project. One tool is a template to develop story lines, which was elaborated
during case studies for model evaluation. In impact assessment practices, story lines are used
to describe qualitative images of the future (qualitative scenarios) with expert knowledge.
They can be the only final outcome of an impact assessment or can precede model
simulations and be revised and combined with the quantified impacts (Alcamo, 2008). During
the development of SEAMLESS-IF, story line development was a step to specify the
simulations to be performed in case studies. Further, this exercise aims at helping common
understanding between end-users and framework experts about agricultural systems and
corresponding processes to be assessed. The tool uses a table template which allows to
describe, in a few words, changes in driving forces (e.g., policy or technology) to be assessed,
to define in a hierarchical way systems and sub-systems that may be affected by these
changes (e.g., market, territory, farms and fields) and to qualitatively describe the expected
impacts of each investigated change. The development of story lines makes end-users to
describe, more or less freely, the narratives underpinning their policy assessment questions
and the associated assumptions while the story line template helps SEAMLESS-IF experts to
520
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
capture narrative information necessary to set up an impact assessment within SEAMLESS-IF.
A second tool was specifically elaborated to support interactions between SEAMLESS
researchers and end-users during the evaluation of the framework in ‘real’ conditions. It is a
problem framing guideline, gathering information that requires specification to perform
simulations. Three versions of this guideline have been developed with different degree of
detail. The less directive version provides a set of pre-defined general questions to be asked
by SEAMLESS-IF experts to end-users. An intermediate version adds indications on
SEAMLESS-IF capabilities (scales, modelled processes, etc.), so that experts can check the
possibility to address end-user’s questions. The most directive version is a set of tables
providing in-depth guidance to SEAMLESS experts and end-users for collecting required
information. Tables are structured according to the topics to be discussed and to SEAMLESSIF capabilities. This guideline more directly matches the information to be specified in
SEAMLESS-IF than the story line template.
During the development of SEAMLESS-IF, interactions between experts and end-users to
specify simulations started in 2007. To lead these interactions, the experts had the choice
among the three versions of the problem framing guideline and the template for story line
development.
Results and discussion
Intensity of interactions varied a lot according to end-users’ willingness to interact with
experts to specify the problem to be assessed and also to the researchers’ confidence in their
skills to manage such discussions. Some of these interactions were detailed enough to specify
simulations to be performed. Some of the experts led these detailed interactions with tables
provided in the guideline because they preferred a well-defined structure. Other experts who
led detailed interactions preferred rather free discussions with end-users, i.e. they either asked
open questions inspired from the problem framing guideline to end-users (less directive
version) or asked questions enabling to fill out the story line template.
If precise enough, information collected with the problem framing guideline is sufficient to
fully specify the simulation. With information provided by story lines, the expert of the
framework can deduce the needs in terms of simulation parameters and can explain the limits
of the modelling framework to the end-user, providing he has a thorough knowledge of the
framework and is used to translate story lines into simulation parameters. Each end-user who
was asked to fill-in the concrete tables provided in the guideline or to develop a story line
found the exercise very useful to frame the problem and to stimulate interactions (even
without simulation perspective for some of them). This should improve the relevance,
credibility, legitimacy and transparency of the simulations.
We conclude that interactions between end-users and experts of the framework require
several accompanying tools, considering the diverse expectations and skills. Interactions
between end-users and experts of the framework will affect the understanding of the
framework by end-users, the chances of doing simulations that meet their needs and the
traceability and transparency of the whole process. Our observations will be further evaluated
once SEAMLESS-IF will be applied to answer end-users’ concerns and to support decisionmaking. The design of methods and tools to support interaction between experts of the
framework and end-users will be developed further in the future as the framework is used.
Observations of how problems are specified and a synthesis of these experiences will help to
improve the use of SEAMLESS-IF in close interaction with end-users.
References
Alcamo, J. (Ed.), 2008. Environmental futures: The practices of environmental scenario
analysis. Elsevier, Amsterdam, 197 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
521
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Agricultural decision support systems:
Tools to facilitate co-learning for stakeholders and scientists
P.J. Thorburn, E. Jakku
CSIRO Sustainable Ecosystems, 306 Carmody Rd, St. Lucia Qld 4068, Australia
Contact: Peter.Thorburn@csiro.au
Introduction
The development of decision support systems (DSSs) has been one way in which agricultural
scientists have attempted to help farmers deal with the complexity of optimizing farming
systems. DSSs can be described as knowledge intensive innovations, because they require
people to learn new knowledge and skills to operate the technology effectively and adopt its
‘outputs’. There has been a move to employing participatory action research (PAR)
techniques in the application of DSSs, in an attempt to increase their generally low impact.
The shift towards participatory application of agricultural DSSs opens opportunities for new
ways of thinking about application process. The sociology of both science and technology
describes how science and technology are produced through social relationships and practices,
and so might inform and guide participatory DSS application. In the paper we develop a
framework based on these sociological principles that describes the participatory process as a
co-learning venture. We illustrate the application of the framework through a case study of the
use of agricultural DSSs for improved management of sugarcane production systems.
Sociological framework
We propose that participatory DSS application is an interplay between three concepts in
science and technology studies; Interpretative flexibility (Hess, 1997), Technological frames
(Bijker, 1995) and Boundary objects (Cash, 2001); illustrated by Figure 1. A DSS can act as a
boundary object, creating a temporary bridge that promotes dialogue between the various
people involved in its application, while remaining flexible enough to be utilized by the
different parties for their own purposes. Interpretive flexibility allows the DSS to serve as a
boundary object, and highlights the different goals people have when working with DSSs.
Through the negotiation, cooperation and co-learning that the DSS-as-boundary-object can
facilitate, the parties involved may arrive at an increasingly shared understanding of the
problem. This works towards increasingly congruent technological frames. Obviously any
technology, including a DSS, is conditioned by, and embedded in (Figure 1) its external
social, cultural, political, economic and biophysical context which must be considered during
these interactions.
CONTEXT
Biophysical
context
Interpretative flexibility
and incongruent technological frames
Social and
cultural
context
Negotiation and
co-learning
Boundary
object
(e.g. DSS)
Economic
context
Figure 1. Theoretical framework of
the context, processes and outcomes
of participatory DSS development.
Outcome 1: Ongoing
use of technology
522
Potential for increasingly
congruent technological frames
Outcome 2: Management
recommendation
Political
context
Outcome 3: No reason to
change or use technology
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
Participatory DSS application leads to three potential outcomes (Figure 1): (i) Acceptance
by potential users of the need for, and value of ongoing use of the DSS, a process that
corresponds more to classical DSS development. (ii) A better understanding of the problem
and its context, leading to the development of a new management heuristic that can be
extended as a management recommendation for farmers. (iii) The stakeholders involved may
find that there is no opportunity for, or little benefit from changing current management
practice. If some parties, e.g., the scientists, disagree with this view, the technological frames
are obviously incongruent.
Case study
The framework was assessed through a case study in the Australian wet tropics, at Tully. A
group of farmers and extension officers wanted to increase sustainability of nitrogen (N)
fertilizer management within sugarcane farming in their region in response to concerns over
health of the nearby Great Barrier Reef. They thought that seasonal climate forecasting could
possibly help reduce environmental N losses, whilst increasing crop yields. These issues were
examined in a participatory approach, by simulating various management practices thought to
potentially increase sustainability of local sugarcane production systems with the APSIMSugarcane model. Semi-structured, in-depth interviews were conducted with all case study
participants and their comments compared with the elements of the framework.
The range of expectations the participants initially held about possible direction of the
study illustrates the interpretative flexibility that shaped their initial reactions to the study: …I
was curious to see how you could use the nitrogen application part of, with it, which I
understand we probably all have our own ideas through experience, about putting on
nitrogen in different wet years and dry years, and things like that. (Farmer) APSIM acted as a
boundary object, allowing the participants to explore nitrogen management and gained a
better understanding of the nitrogen cycle and the consequences of different nitrogen
management scenarios in that environment: I was just fascinated with the different responses
of the soil types ……I’m very enthusiastic about it, opened up a lot of possibilities, scenarios,
something like that. (Farmer) At the conclusion of the study there was a degree of
incongruence amongst the technological frames of the case study members: I don’t think I’ve
learnt much more … (Farmer) …We’re looking at (recommending) a lower N rate in ... wetter
conditions. (Extension officer)
Discussion
The framework describes the social context and processes that shape participatory DSS
application. The concepts of interpretative flexibility and technological frames reinforce the
importance of acknowledging the different perspectives held by people involved in the
participatory process, and then the need/opportunity to work towards a shared understanding.
When used as a boundary object, a DSS encourages co-learning between those involved in its
development. Appreciating the way in which a DSS can act as a boundary object recognizes
how cooperation and co-learning among those involved in that process can occur, despite the
fact that these people can hold diverse perceptions of a DSS or the issue it is designed to
address. The framework, and the concepts on which it’s based, highlight that participatory
DSS application is a valuable learning process leading to knowledge acquisition.
References
Bijker, W., 1995 Of bicycles, bakelites, and bulbs: Toward a theory of socio-technical
change. Massachusetts Institute of Technology Press, Cambridge, Massachusetts.
Cash, D., 2001 Sci. Technol. and Human Values 26: 431-453.
Hess, D., 1997 Science studies: An advanced introduction. New York University Press, New
York.
523
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Participatory modelling for Renewable Resource Management:
A critical assessment
Annemarie Van Paassen, Cécile Barnaud, Guy Trébuil
Communication science, Wageningen UR, Hollandseweg 1,
6706 KN Wageningen, The Netherlands
Contact: Annemarie.vanPaassen@wur.nl
Introduction
The objective of Integrated Assessment Models (IAM) of agriculture and sustainable
development is to support learning of societal actors about possible actions. Multi-Agent
System (MAS) models, integrate biophysical and socio-cultural dynamics in the assessment.
They assume that socio-cultural aspects only change slowly and that policymakers should
know what is biophysically feasible and social acceptable in a specific location. For MAS
modellers the challenge is to capture the local socio-cultural behaviour in their models so
many opt for a participatory modelling approach. In this article, we problematize the
collaboration between MAS modellers and the stakeholders involved: What are critical
aspects of participatory modelling for societal learning and action, and what could be an
effective role for the scientist? To answer this question, we studied Companion Modelling
(ComMod, http://www.commod.org) activities at Mae Salaep village in northern Thailand.
Methods
There are different forms of knowledge each with their assumptions about governance and
change, and on the role that scientists can play. Post-modern scientists who embrace the
theory elaborated by Funtowicz & Ravetz (1994) recognize the high complexity and
dynamics of socio-ecological systems. Society needs to opt for adaptive management. As
resource users have both high stakes and contextual knowledge, scientists are called to engage
in collaborative forms of research. Adaptive management refers to joint experiential learning
for ecologically sustainable development. Most of the research for adaptive management
concerns the exploration of socio-ecological system dynamics, dialogue and deliberation to
select the best alternative. However, deliberative approaches ignore the distorting effect of
dominant discourses, power relations and conflict. As a result, we now see a variety of
research aims and intended outcomes:
- the generation of socially robust knowledge for effective policy making;
- enhancing social learning and capacity building for sustainable development;
- learning, empowerment and advocacy for transformation for more equitable sustainable
development.
To gain insight in the actual role of ComMod, we use reflexive research. Fischer (1995)
developed a critical reflexive evaluation method to assess the effectiveness of an intervention
and link it to a wider socio-political and ideological perspective. Here we reflect, together
with ComMod designers, on the goal and the theory of action (espoused theories and tacit
theories-in-use; Argyrus, 1992) and the effects of the ComMod activities on the stakeholders.
Results and discussion
ComModians recognize the high uncertainty of Renewable Resource Management (RRM)
and the existence of multiple legitimate perspectives. They use the co-construction of models
to facilitate learning about a RRM problem at stake. Core activities of a ComMod process are
Role Playing Games (RPGs) and MAS computer simulations. In this case, scientists started
with a preliminary problem analysis, to develop a first conceptual model and a related RPG.
The local stakeholders were invited to play the RPG, to enact and share their knowledge and
perspectives, to criticize and validate the model. ComMod discussed the problem situation
524
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
and options for improvement within homogeneous sub-groups and at a plenary meeting.
Scientists integrated the proposed options into a MAS model, to jointly explore the long-term
effects of the proposed RRM options. In this way, ComMod scientists intend (a) to generate a
rich multiple perspective of the socio-ecological system and its dynamics, and (b) to support
communication and negotiation for collective decision making.
In Mae Salaep village, ComMod scientists developed a RPG and a model to represent the
risk of soil erosion to test whether this enabled scientists and farmers to share knowledge for
mitigating land degradation. From observed game behaviour and individual interviews,
scientists learned that farmers like the RPGs as they were ‘fun’ and ‘resembled reality’. The
farmers’ understanding of the MAS model was mixed. Nevertheless, they proposed to use the
tools to explore what they considered an appropriate solution for the soil erosion: perennial
crops. They wanted to examine how rural credit and irrigation possibilities could support the
expansion of plantations. Scientists agreed and engaged themselves in an iterative process of
joint learning. To enhance engagement of the stakeholders, scientists formulated a new goal:
was it possible for ComMod to enhance communication and learning amongst participants?
This was the espoused theory. However, from the start of the 2nd ComMod cycle (on the
village credit system) the main designer also considered the diversity of interests, local power
relations and their influence on plenary discussions (Barnaud et al., 2008). Though
prioritizing local knowledge, she did not start with a plenary ‘conception workshop’ but opted
for individual interviews. The theory-in-use was to ensure that the conceptual model would
reveal the situation of the poor as well as the affluent. The subsequent RPGs and simulations
showed that the credit regulations prohibited the poor farmers to use formal credit for longterm investments in orchards or plantations. At first, more affluent participants remarked the
poor just had to rely on informal credit arrangements, but later they agreed that the formal
credit rules should cover the needs of everyone. Participants noted the game and simulations
gave them more insight in the credit system dynamics, and it stimulated ‘inclusive problemsolving’. Unfortunately, high-level authorities, not involved in the process, had to decide upon
a change of formal credit rules. For the next cycle on water sharing (Barnaud et al., 2007),
scientists again took care to integrate all interests in the model, and they organized extra subgroup discussions and participatory simulations to empower the poor for the plenary
discussions. To link up with political decision makers, a higher-level authority was invited to
attend the plenary discussion. This time the poor organized themselves around a charismatic
leader and started to persuade other farmers. In the end, the village representative agreed to
collaborate, but the local authority evaded public commitment. Despite the official discourse
of decentralization, the participatory bottom-up approach did not (yet?) fit the administrative
procedures and culture of decision making. The foreign scientists refrained from further
action; it was deemed inappropriate to get involved in local politics or advocacy.
This is a challenge for ComMod scientists. Formally, they aimed at an inclusive process
of knowledge exchange and social learning for sustainable development. However, by doing
so they also put the situation of the marginalized on the political agenda. This was accepted
by the village community. However, introducing participatory learning and decision making
in the wider context of an hierarchical bureaucracy proved difficult. To create space for
transformation, scientists need a long-term socio-political perspective and strategic alliances
with an influential NGO, policy research institutes, brokers, etc.
References
Argyris, C., 1992. On organizational learning. Blackwell Publishers, Massachusetts.
Barnaud, C., et al., 2007. Simulation and Gaming Journal 38: 398-420.
Barnaud, C., et al., 2008. Ecological Economics 66: 615-625.
Fischer, F., 1995. Evaluating Public Policy. Nelson-Hall Publishers, Chicago, 296 pp.
Funtowicz, S.O. & J.R. Ravetz, 1994. Ecological Economics 10: 197-2007.
525
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Using explorative models to enhance expert learning about farm context
and farmer rationalities in Burkina Faso
A. Van Paassen, N. De Ridder, L. Stroosnijder
Communication science, WUR, Hollandseweg 1,
6706 KN Wageningen, The Netherlands
Contact: Annemarie.vanPaassen@wur.nl
Introduction
In 1997, after a decade of agronomic research, the Antenne Sahelienne of the Wageningen
University decided to integrate this knowledge in a Multiple Goal Linear Programme (MGLP)
model to make it accessible for local users. As Burkina Faso was in the process of delegating
natural resource management (NRM) authorities to village committees, the so-called
SHARES models (Stroosnijder & Van Rheenen, 2001) focussed on the optimization of land
related activities within the village boundary. In this paper we present the action research that
agricultural staff of the Integrated Rural Development Project – PEDI undertook to test the
potential of SHARES in two villages to enhance communication and learning for NRM.
Methods
In an iterative process of action, reflection and learning, staff members tried the potential of
the modelling tool for their own learning and for extension purposes. First, local data
weregathered to run SHARES for two villages in the province Sanmatenga: a heavily
populated village in the South and a village with abundant land in the North. In the latter both
arable farming and extensive livestock rearing were possible. Both villages differed
considerably in access to, and use and management of their natural resources.
To assess the utility of the SHARES model for learning, a theory of learning was
elaborated. It is based on the work of Ajzen & Madden (1986) about Planned Action and
Argyrus & Schön (1996) about 1st order and 2nd order learning (Figure 1).
People engage in 1st order learning when they feel that things are different from the way
they want. They look for information and experiment to gain the knowledge necessary to
attain a desired situation. The motivation and focus of learning depends on the (perceived)
Figure 1. The learning actor (-network).
526
Session D1: Stakeholder involvement in designing, use and evaluation of assessment models
context and the internalized frame of reference. In social life, actors have different
perspectives and project goals. So, collective learning cannot do without 2nd order learning:
gaining insight in each other’s frame of reference to attain mutual understanding and
convergence of project goals. Models embody the perspective and projects goals of the
designers. Actors are interested to use modelling tools for 1st order learning, when they
perceive a match of perspective (system elements included, level of inquiry) and project goal
(David, 2001).
With this theoretical model, we first explored whether the SHARES model could answers
questions relevant to the farmers and/or agricultural extension officers (1st order learning).
Did the model match their perspective, envisaged projects and learning issues? Subsequently
the modelling tools were used to discuss the frame of reference of the scientists and the PEDI
staff, as embodied in the model, with those of various farmer categories (2nd order learning).
Results and discussion
SHARES ambition was to enhance learning about village NRM, but the first matching trials
proved that the model lacked several relevant components (on forestry, water and pasture
management). The prime focus was on arable farming, so it was decided to downscale the
level of inquiry to the farm. At the farm level, farmers and extension officers were in the
process of fine-tuning fertilization practices, but SHARES lacked the technical details to
support operational questions (1st order learning). MGLP models are summary models, made
to explore long-term orientations, and they may be used to trigger 2nd order learning. The
agricultural staff was interested to discuss farm orientations as they realized that farmers still
farmed differently than they expected.
SHARES generated a gamut of farming options for four distinctive farmer categories in the
two villages. This provided the agricultural staff with important insights. They learned more
about overall socio-ecological system dynamics; the intimate relation between crop choice,
livestock practices, and farm output. They were surprised by the low farming potential of the
heavily populated village: despite intensive farming, these farmers could hardly satisfy their
basic needs.
The model results were visualized in simple pictures to invite farmers to position
themselves with respect to their applied farm strategy. This triggered a debate about the
technical-economic optimization perspective of the experts, embodied in the model. Within
the village, there was social pressure not to optimize income from arable farming, but to stay
focussed on household food security. In addition, investment and optimization of income
occurred in less risky activities such as livestock rearing and petty trade. This resulted in
different farm-household dynamics than anticipated by the extension staff.
In sum, SHARES provided extension officers critical details about the farm system
dynamics and the biophysical production boundaries of different farm categories (context
variables) and proved to be a valuable tool to better understand farm frame of reference. This
latter aspect is important, because most extension workers focus on operational questions
without real knowledge of the farm potential and farmer rationality.
References
Ajzen, I. & J.T. Madden, 1986. Journal of Experimental Psychology 22: 453-474.
Argyris, C. & D.A. Schön, 1996. Organizational learning II: Theory, method and practice.
Addison-Wesley, Reading MA.
David, A., 2001. European Journal of Operational Research 1: 459-480.
Stroosnijder, L. & T. Van Rheenen (Eds), 2001. Agro-silvo-pastoral land use in Sahelian
villages. Advances in Geoecology 33. Reiskirchen, Germany, 408 pp.
527
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Theme D
Integrated assessment: Science and impact
Session D2: Role of integrated assessment for policy
making and evaluation
Session organizers:
Floor Brouwer
LEI, Wageningen UR, The Netherlands
Hayo Haanstra
Ministry of Agriculture, Nature and Food quality, The Netherlands
Insa Theesfeld
Leibniz Inst. of Agric. Development in Central and Eastern Europe,
Halle, Germany
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Participatory IA tool development:
Lessons learnt in the SEAMLESS project
J. Alkan Olsson, M.K. Van Ittersum, A.-K. Bäcklund, S. Brogaard, F. Brouwer
Lund University Centre for Sustainability Studies (LUCSUS),
P.O. Box 170, 221 00 Lund, Sweden
Contact: Johanna.alkan_olsson@lucsus.lu.se
Introduction
During the preparation of the World Summit for Sustainable Development a model for
science was established; interdisciplinarity, policy-relevance, holistic perspectives and
stakeholder involvement were its cornerstones. This scientific model has evolved and has
been given diverse names, including; reflexive science (Beck et al., 1994), Mode 2 science
(Gibbons et al., 1994) and sustainability science (Kasemir et al., 2003).
In 2003 the European Commission introduced a formal procedure for Impact Assessment
(IA) of its future policies. The political call for IA includes several of the characteristics of the
new model for science (EC, 2002). It is, therefore, no surprise that several large projects
funded by the European Union (EU) have included participatory elements in their setup. An
example of such a project is the SEAMLESS project that has developed a computerized
framework to assist in the impact assessments of new agricultural and environmental policies
across a range of scales (Van Ittersum et al., 2008).
Following the intentions in Mode 2 science the aim of the SEAMLESS stakeholder
interactions has been to collect user requirements and provide feedback into the project in
order to increase the policy relevance and user friendliness of the framework. The general set
up of the user interactions is depicted in Figure 1, however, the form and content has slightly
evolved over time.
Figure 1. General setup of user interactions in SEAMLESS.
The aim of the paper is to analyse and discuss lessons learnt from a development process in
interaction with potential users. This will be done by analysing, from the perspective of the
SEAMLESS project (i) the accomplishment and problems with the user interaction set up; (ii)
the accomplishment and problems with uptake of collected user information by the project.
Methods
The analysis is based on the experiences made by the scientists engaging in discussions with
users. It is also based on minutes from various types of stakeholder meetings, i.e. in this paper
the notes from the User Forum and targeted meetings with experts in the Commission
services. The achievements of the SEAMLESS user interactions will be analysed in terms of
how the form of the interactions has facilitated the collection of requirements and how the
nature of the discussed issues and the communication structure in the project have influenced
the uptake of these requirements.
530
Session D2: Role of integrated assessment for policy making and evaluation
Results and discussion
Since the beginning the User Forum, which gathered every 6 months, was the core activity.
The establishment of the UF was an achievement in itself – it is not trivial that representatives
from the Commission DGs commit themselves to closely interact with individual research
projects throughout their lifetime. Interviews with representatives from DGs responsible for
IA were also an early activity. The so-called targeted meetings were developed later as a
response to the need of discussing specific issues with specific DG units or persons.
By help of the UF it was envisaged in which way the framework could contribute to the
impact assessment work in the DGs. The limited experience of direct work with modelling
tools in most of the DGs formed the understanding that the system should have two different
user interfaces. The possibility to run the system should be restricted to people with
knowledge in modelling. But the policy officers should be provided with an interface where
they can alter parameters in already modelled issues to test alternative answers by help of
minor changes. This interface was also adapted to the structure of the Commission’s work
procedures. The scientists in charge of the user interactions were primarily responsible for
feeding back the user requirements into the project. This was not always an ideal set up;
sometimes direct interactions between users and developers would have been more effective.
However, at most of the meetings developers were present and could answer technical
questions, as well as more directly bring ideas of users back into the development process.
A recurring theme on the agenda of interactions was the selection of indicators to be
included in the indicator library of SEAMLESS-IF. Generally this was very effective, in the
sense that the project was able to both include lists of indicators that are already adopted in IA
processes and to tap indicators from newly developed lists by for instance the European
Environment Agency. Also, indicators proposed by the scientists from SEAMLESS were
discussed. A continuous challenge was to balance the desired indicators with the possibilities
of the models included in SEAMLESS-IF to actually assess the indicators. Clearly not all
desired indicators can be assessed by the models presently included in the framework. This
was compensated by the fact that the framework was designed such that new model
components allowing computation of new indicators can be added at a later stage.
The interaction with potential users has contributed to a more policy relevant SEAMLESSIF, where design of the workflow of an assessment and the corresponding interface for the
policy experts are clearly shaped as a result of the participatory process. Scientists and coordination of the project have learned that participatory development is a time and resource
consuming process, which is hard to plan before the start of the project. There is a clear
tension between the request (from the funding agency) to define a clear proposal before the
start of the participatory process, detailing the scientific approach and the consortium,
resulting in restricted flexibility to adapt to issues emerging from the participatory process. It
is therefore important for a research project to define the ambition of the participatory process
at an early stage.
In order to meet the Mode 2 scientific requirements, the interaction has to be built into the
research from the beginning and relevant stakeholders appointed in the same manner as the
technical components for a project is chosen.
References
Beck, U., et al., 1994. Reflexive modernization: Politics, tradition and aesthetics in the
modern social order. Polity Press, Oxford, UK.
EC, 2002. Communication on Impact Assessment, COM 2002(276).
Gibbons, M., et al., 1994. The new production of knowledge. SAGE Publ., London, 174 pp.
Kasemir, B., et al. (Eds), 2003. Public Participation in Sustainability Science. Cambridge
University Press, Cambridge, 281 pp.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
531
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Multiple agendas and cultures in a participatory development of
computerized IA tools: The experience of SEAMLESS
A.-K. Bäcklund, J. Alkan Olsson, S. Brogaard, M. Taverne, J.P. Bousset,
J. Wery, M.K. Van Ittersum
Department of Social Geography, Lund University,
P.O. Box 170, 22100 Lund, Sweden
Contact: Ann-Katrin.Backlund@keg.lu.se
Introduction
Integrated Impact assessment (IA) is gradually making inroad in European policy making.
The European Commission places great aspirations on the IA system as a way to achieve
better policy and law making but also as a tool for improved legitimacy of government and
increased consensus in European politics. This is reinforced by opening up for the possibility
of stakeholders to influence every step of the work process. In order to increase the
credibility and legitimacy of the assessments, which have been questioned, there is a call for
using more science based methods. As a result there is an increasing interest in using
modelling tools to support assessment work (Bäcklund et al., 2009). However, the use of
science based models in a decision process is not always straightforward. The reasons for this
are multiple, but mainly linked to the different agendas, cultures and dynamics of the political
respectively scientific communities. The SEAMLESS project has developed a computerized
framework to assist impact assessments of agricultural and environmental policies across a
range of scales (Van Ittersum et al., 2008). To develop this tool the researchers in the project
have engaged in a participatory process with potential users. This included a more formal
way through the so called User Forum meeting every six months and specific evaluation
sessions in the test cases throughout the duration of the project as well less formal ways, i.e.
through personal interviews, e-mails, and targeted meetings on specific issues. The aim of
this paper is to, based on the experiences gained in the interactions, identify policy/science
interfaces and discuss the different approaches to the use of modelling tools between policy
developers and tool developers.
Methods
The empirical material consists of comments and reactions expressed at the SEAMLESS User
Forum and during individual interviews with personnel mostly from DGs of the European
Commission as well as during interactions at regional level when evaluating the framework,
in France, Poland and Mali.
When analysing the interaction between science & technology and the policy process the
criteria of credibility, salience, and legitimacy are frequently used (Cash et al., 2003).
Credibility concerns the scientific adequacy of evidence and arguments. Salience concerns the
relevance, appropriateness, usefulness and timing of the information. Legitimacy is achieved
when the production of knowledge has been conducted in an unbiased way and has treated
opposing views and interests in a fair manner. The paper will use these concepts to analyse
the stakeholders’ comments on the developed tool.
Results and discussion
Stakeholders’ comments were of two types, i.e. requests concerning the technical
performance of the tool and strategic comments. The technical comments were often possible
to meet by development of the components of the tool, workflow for IA and the Graphical
User Interface. The strategic comments have a more profound political dimension
highlighting the differences in how modelling results are used in the political and scientific
communities and were therefore not as easily transformed into the tool. When analysing the
532
Session D2: Role of integrated assessment for policy making and evaluation
stakeholder strategic comments using the concepts Credibility, Salience and Legitimacy it
becomes evident that these attributes are tightly coupled and efforts to enhance one may lead
to trade offs with the others.
Credibility The issue of uncertainty of the modelling outcome is very important to the users,
particularly when dealing with politically hot issues. A participant stated: “If I do not get
precise information how could I otherwise motivate the results to an angry stakeholder
phoning me up”. To deal with this demand we engaged with users to assess their perspectives
as to uncertainty analysis (Gabbert et al., 2009).
Salience The approached policy developers also argued that is important that tools give
answers to policy relevant questions. Similarly, it was argued that it is important that these
tools do not produce too much or irrelevant information. The continued stakeholder
interactions have been a way to identify this demand. However, the models used in
SEAMLESS have a particular scope which to a certain extent can be enlarged but for some
issues additional models are needed or science simply lacks the methods to provide an
analysis.
Legitimacy In order to meet the demand for legitimacy a tool has to be flexible as to what to
assess. It must be possible to incorporate stakeholders’ views and be sensitive to the political
process. It was also essential that a tool assisting in IA is transparent, i.e. it should be easy
understandable how the assessment has been done and what the underlying assumptions are.
The participants repeatedly expressed their concern for lack of transparency of the modelling
system. One official argued that “Scientists are lining up arguing that they have a new model
that can assist me in assessing this or that. If I am not able to understand the underlying
assumptions of a model how could I judge which model to use?” Legitimacy is achieved
when the production of knowledge has been conducted in an unbiased way and has treated
opposing interests in a fair manner. If the modelling system is a black box, where assumptions
and other critical parameters are not clear and understandable, the outcome might not be
perceived as legitimate. We have dealt with this demand in SEAMLESS through spending
significant resources on a transparent user interface, extensive documentation of each of the
models, a possibility to use and assess each of the models standalone and providing access to
intermediate results of an analysis performed through a model chain.
However, assessment tools cannot only be regarded as technical/scientific applications but
also as tools for communication between science and policy (Bäcklund et al., 2009).
Therefore, to improve the usability of these types of tools it is essential that scientists increase
their understanding of the social and institutional dynamic and conflicting interests and
cultures of policy developers.
References
Bäcklund, A.-K., et al., 2009. In: Eds F. Brouwer & M.K. Van Ittersum, Environmental and
agricultural modelling: Integrated approaches for policy impact assessment. Springer,
Dordrecht. (forthcoming)
Cash, D.W., et al., 2003. Proc. of the National Acad. of Science of USA 100(14): 8086-8091.
Gabbert, S., et al., 2009. AgSAP Conference 2009, these proceedings.
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
533
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated assessment for the development of Progressive Coordinative
Action (PCA) plan: An approach for sustainable agriculture
B.M. Dahal
Environment and Public Health Organization (ENPHO),
P.O. Box 4102 KTM, Kathmandu, Nepal
Contact: binodm_dahal@yahoo.com
Introduction
This paper focuses on how to achieve the sustainable agricultural production through an
integrated assessment at a given time. Integrated assessment (IA) is considered as a ‘higherorder approach’ tool to analysis of any system with interconnectedness between all the
possible factors for the sustainable development of a community (Gough et al., 1998).
Integrated assessment seeks to recognize and incorporate the uncertainty and divergent
interests that characterize current situation regarding the sustainable development of a specific
sector. IA specialists emphasize the importance of on going participation in the process by
both scientific experts and all stakeholders, such as policy-makers and the general public.
Thus, Gough et al. (1998) and Schneider (1997) dispute that the emerging practices of IA can
promote broad and justifiable participation in the value- laden, complex, and uncertain
decisions that we increasingly face. But the history of the developing countries regarding to
sustainable development of any discipline is difference that the planning and policy only can
not meet the target. Therefore, the IA or IAM should be able to develop a Progressive
Coordinative Action (PCA) which can contribute to confer techno-social solution for the
sustainability of an agricultural system.
The concept of PCA was developed after the several series of discussion and interview
with farmers of two communities of Nepal. Although the communities have difference
characteristic in terms of social and economic conditions, both of them gave an emphasis on
coordinative action to over come the problems regarding the sustainability of agriculture.
However, this concept has not been implemented yet; the following theoretical outline has
been drawn for its execution.
Integrated Assessment
of an agri-system
Progressive Coordinative
Action Plan
Sustainable Agriculture
Techno-social solution
Figure1. Sustainable development cycle for agriculture and role of IA and PCA.
534
Session D2: Role of integrated assessment for policy making and evaluation
What is Progressive Coordinative Action (PCA)?
It is proposed the phrase ‘Progressive Coordinative Action Plan’ to describe an output of
integrated assessment of an agricultural system. The PCA plan helps to find techno-social
solution for sustainability of agriculture on the basis of information generated from Integrated
Assessment (Figure 1). Therefore, IA is a root action for development of PCA which enforce
to generate a techno-social solution for the improvement of all related factors.
In such case, The PCA plan mainly focuses on Social, Natural and technical sector to get
optimum productivity, to reduce environment and economic problems. Further more it gives
prominence to develop coordinative project.
Expected achievements of PCA plan
The PCA plan focus to activate all factors related to the agricultural production system on the
basis of Integrated Assessment Information. Therefore, it can achieve the tangible agricultural
sustainability (Figure 2). The system can self sustained because the key stakeholders are main
implementer of the plan. Further more, the PCA play an important role to reduce the
environmental risk and health risk of farmers, individual farmer attitudes toward public
concerns therefore, low risk of acceptance of adopted technology etc.
PCA
phases
Time
Figure 2. Tangible achievement
of sustainable agriculture through
PCA plan.
Conclusion
PCA can be a tool for sustainable development of agriculture in the context of development
countries which is possible to achieve only through an Integrated Assessment.
References
Gough, C., et al., 1998. Environ. Mod. Assess. 3: 19-29.
Schneider, S.H., 1997. Environ. Mod. Assess. 2: 229-248.
535
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Better regulation by the use of Integrated Assessment?
SEAMLESS integrated framework and European policy making
Robert Fischer, Insa Theesfeld
Humboldt-University of Berlin,
Department of Agricultural Economics and Social Sciences,
Division of Resource Economics,
Philippstr. 13, 10099 Berlin, Germany
Contact: Robert.Fischer@staff.hu-berlin.de
Introduction
For three reasons there will be great demand for external integrated assessments (assessing
environmental, economic, social and institutional impacts of policy options) within the
European Union: First due to the lack of resources of the European Commission, second
because it is now compulsory for the Commission to do an ex-ante impact assessment at the
European level and third because the EU has committed itself to a strategy for sustainable
development.
Methods and theoretical concept
The paper provides an institutional analysis for the use of integrated assessment tools. It sheds
light on institutional constraints immanent in the procedure of ex-ante impact assessment in
the European Union political context. In doing so, the paper focuses on the requirements for
the production of scientific expertise for policy making.
The paper argues that scientific knowledge within impact assessments has two functions
that are both unavoidable and legitimate: The technocratic function is to increase quality of
policy proposals and inform policymakers, the Machiavellian function is to increase
legitimacy and to persuade the public. Thus integrated assessments could be used to assess
economic, ecologic and social impacts but they could also be used as a rhetorical device and
political tool to gain or maintain power.
The use of integrated assessment tools heavily depends on an appropriate institutionalization of the science-policy relationship and smooth interaction of this science-policy
interface with the political environment. From an institutional perspective a crucial reason
hindering a successful uptake of integrative assessment tools could be a misfit between the
knowledge produced by scientific experts and the requirements within the political system. If
there could be a misfit or not will be illustrated using the case of SEAMLESS-IF as an
example. SEAMLESS-IF is a computerized framework to assist in the impact assessments of
new agricultural and environmental policies across a range of scales that has been developed
in the SEAMLESS project (Van Ittersum et al., 2008).
On the one hand, the produced knowledge of integrated assessment tools may not fit
because the knowledge itself may be insufficient, unsound, intransparent, incomprehensible,
biased, or useless. On the other hand, it could be that the knowledge is pretty reliable and
comprehensible but, nevertheless it does not fit into the policy-making context because of bad
timing, insufficient administrative procedures, improper organizational linking, ignoring of
political power play or disturbing balanced institutional equilibriums. In this latter case the
question for any integrated assessment would be how to cope with the institutional constraint
of being used or misused in the political process. All ex-ante policy assessment studies do
face inherent and unavoidable political characteristics concerning the provision of the inputs
and the use of the outputs: (a) during the data collection phase studies are often drawn into the
political process by providing false or misleading information by relevant actors and
interviewee, (b) during the assessments, the public authority who commissioned the ex-ante
assessment may convey certain expectations to the evaluators and, (c) during the
536
Session D2: Role of integrated assessment for policy making and evaluation
interpretation of the results: the studies may be used as political instrument to justify
preconceived legislative actions.
Concerning the latter, the policy experts who will use integrated assessment tools will have
their own interests and expectations about the outcome of the assessment. If the impact
assessment fails to meet these expectations it is not of much use for the civil servants and
politicians and it is likely that they will take no notice of the produced knowledge. On the
other hand if it meets the expectations of civil servants and politicians it will be part of the
political process and it will be perceived by the public as intertwined with it. From this it
follows that the scientific knowledge produced within an assessment procedure will be
contested by different stakeholders and the public. Therefore, this scientific expertise requires
additional efforts compared to normal academic science organized in disciplines at
universities. In this context of scientific expertise for policy making there is an increasing
demand for interdisciplinary, transparency, usability, legitimacy, completeness and
credibility. Remarkably the European Commission has picked up these principles in its
guidelines and documents on impact assessment. However, the newly established demand for
impact assessments could lead to some allied perils. It could increase different expert
opinions, counter assessments and different policy recommendations and with this plurality it
might even decrease credibility of integrative modelling tools.
Results
From the perspective of science and technology studies there is a difference between scientific
knowledge and expertise for policy making. There are additional requirements for the
production of expert knowledge that result from these differences. From an institutional
economics perspective the introduction of impact assessments and the use of integrative
modelling tools has to meet the demands of the institutional environment already in place
before doing an assessment.
References
Fischer, R., 2008. Uptake of SEAMLESS-IF: Reflections on demands and institutional
implications of the implementation and application of ex-ante assessment within the
European Union, PD7.2.3, SEAMLESS integrated project, EU 6th Framework
Programme, contract no. 010036-2, www.SEAMLESS-IP.org.
Fischer, R., 2009. Democratising scientific advice: New dimensions of science-based policymaking in the European Commission? In: European Integration Online Papers.
(forthcoming)
Van Ittersum, M.K., et al., 2008. Agricultural Systems 96: 150-165.
537
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
A multi-criteria approach to environmental policy
N. Hasnain, H. ApSimon, T. Oxley
Centre for Environmental Policy, Imperial College London, SW7 2AX London, UK
Contact: n.hasnain@imperial.ac.uk
Introduction
There is growing emphasis on the need to control ammonia emissions in Europe. Arising
mainly from agriculture, ammonia causes eutrophication and acidification of eco-systems and
can also form secondary particulates which are of concern for human health. As a result of
this concern, ammonia became one of the pollutants addressed under the Convention on
Long-Range Transboundary Air Pollution of the UN/ECE, where integrated assessment
modelling, in particular the RAINS model of IIASA provided a solid, scientific basis to the
negotiations leading up to the Gothenburg protocol (Hordijk & Amann, 2007). The same
modelling approach was subsequently adopted by the European Commission to set emission
ceilings for ammonia and other pollutants under the National Emission Ceilings Directive.
Meanwhile at the UK level, more detailed assessment of ammonia emissions reflecting
national farming practices and conditions has been linked to a national integrated assessment
model, UKIAM, exploring implementation of emission abatement strategies at a much finer
scale (Oxley et al., 2003).
Despite these success stories of integrated assessment models, the models remain narrowly
focused and do not take into account many stakeholder interests which could give rise to
difficulties at the implementation stage of a policy. This study explores the use of multicriteria decision analysis as a complementary assessment tool to help address environmental
problems more diversely (Hasnain, 2009).
Methodology
An initial assessment of tools and studies undertaken under the UN/ECE and at the national
level in the UK was carried out to identify the current state-of-the-art. These tools include
integrated assessment models such as RAINS, its extension GAINS (Greenhouse gas Air
pollution INteraction and Synergies) (Klaassen et al., 2004), and UKIAM. The models use
information on projected emissions, dispersion and deposition of pollutants, geographical
variations in environmental impact, and finally, abatement costs to advise policymakers on
cost-effective solutions. In addition, these models can be linked with other models to assess
impacts upon other components of the nitrogen cycle, such as nitrate leaching.
Despite having a sound scientific basis, these models focus only on technical measures and
do not address wider stakeholder issues. For instance, implementing any abatement policy
will impinge on several stakeholders, most notably farmers, who have to implement measures
and bear additional costs. There may be the issue of how this additional burden is supported
and whether consumers are willing to bear some of these costs. Another drawback of focusing
only on technical measures is the exclusion of animal welfare concerns that may bring into
question some abatement measures. For instance reducing dairy cattle numbers and meeting
demand by intensifying milk output from each animal may impinge upon animal welfare;
these non-technical concerns need to be factored into the analyses to ensure difficulties and
opposition do not arise at the implementation stage.
The first step taken is to identify relevant stakeholders. These are policymakers, scientists,
producers, consumers, together with NGOs concerned with biodiversity and animal welfare.
Criteria are identified for example, health, ecosystem protected, affordability, compliance etc.
to capture different environmental and stakeholder concerns. For each criterion, a suitable
indicator is selected, and when not available, developed. Using modelling data from
integrated assessment models, surveys and interviews with key stakeholders these criteria are
538
Session D2: Role of integrated assessment for policy making and evaluation
weighted and compared on a dimensionless scale for selected scenarios. Scenarios are chosen
to reflect changing consumer patterns, land use and application of abatement measures.
Results and discussion
The performance of each criterion with respect to individual scenarios is evaluated. In
addition, the criteria are weighted by different stakeholders (Table 1). These are then
normalized to give the relative weighting of each criterion on a dimensionless scale. The
performance of the criterion is combined with its normalized weighting to give the overall
value for a scenario (Figure 1). Figure 1 also gives the relative contribution of each criterion
to the value of the scenario.
Table 1. Stakeholder weighting of criteria between 0–100.
Stakeholder Compliance Affordability
Health
NO3
Ecosystem
leached protected N2O
WTP
NGO
Scientist
Public
Public
Dairy farmer
Farmer
DEFRA
DEFRA
EA
EU
100
15
20
10
10
10
20
2
10
50
60
25
10
5
5
10
5
4
15
30
50
0
20
50
5
5
5
7
2
20
80
10
10
5
8
20
25
1
25
100
30
0
15
5
60
40
20
6
8
10
90
25
18
20
7
10
10
3
25
50
70
25
15
5
5
5
15
5
15
40
Value
From our analysis, we conclude that despite significant improvements in integrated
assessment approaches over the years, boundaries of analysis of current integrated assessment
models needs to be widened to take into consideration wider stakeholder concerns. In the case
of ammonia abatement, this means firstly developing scenarios that reflect changing
consumer and land use patterns along with environmental protection and compliance, and
secondly use of suitable criteria that address stakeholder concerns. A multi-criteria approach
is suggested as a means of
0.7
supplementing
existing
0.6
policy
tools
as
it
enables
Compliance
the analysis of multiple
0.5
Health
criteria
under
one
Ecosystem Protected
0.4
framework,
highlighting
Willingness to pay (WTP)
0.3
Nitrous oxide
issues of concern.
0.2
Nitrate leached
0.1
Affordability
0
1
2
3
4
Figure 1. Public valuation
of criteria for each scenario.
5
Scenario number
References
Hasnain, N., 2009. A multi-criteria approach to environmental policy: a case study in
agricultural ammonia emissions. PhD Thesis, Centre for Environmental Policy, Imperial
College London.
Hordijk, L. & M. Amann, 2007. Environmental Policy & Law 37(4): 336-340.
Klaassen, G., et al., 2004. The Extension of the RAINS Model to Greenhouse Gases. IIASA
Interim Report IR-04-015, Laxenburg, Austria.
Oxley, T., et al., 2003. Integrated Assessment 4(4): 236-249.
539
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Exploring receptivity through innovative actions
V. Hernández Jiménez
Rural Planning Department, School of Agronomy
Polytechnic University of Madrid - Spain
Contact: veronica.hernandez.jimenez@upm.es
Introduction
Resolution of land use conflicts requires understanding the complex relationships between
social systems and territorial policies. However, this would always depend on the receptivity
of the systems. In this paper we describe on-going research about stakeholder receptivity and
its relationship with innovation, following ideas developed by Jeffrey & Seaton (2003). This
paper is carried out under the ISBP – Integrative Systems and the Boundary Problem –
research project (Winder, 2006), funded by EU FP6.
Methods
Stakeholders’ receptivity is characterized by using both direct features (based on interviews)
and indirect features (based on policy actions, public speeches, position documents and
others). Then, we can classify it into different categories, for example open receptivity, forced
receptivity and no receptivity. Discussion will bring up the following topics: how
stakeholders’ receptivity influence innovative actions, which (cultural) constrains imposed the
lack of receptivity, and, how receptivity could improve as part of a collaborative process.
Results
Qualitative methods have been carried out to better understand how the role of society could
be strength in the decision making process. The rich picture (Figure 1) is a technique of
mapping actors and conflicts which is based on building trust between researchers and the
Figure 1. A rich picture. Current situation of land use conflicts in Madrid region as it is
perceived by the researcher (Hernández Jiménez, 2007).
540
Session D2: Role of integrated assessment for policy making and evaluation
stakeholders involved. Although it might be difficult to understand due to the hand-writing,
this attempts to explain the need of simplify science in order to reach from local politicians
towards the national decision-makers.
Conclusions
We present work focused on the emergent process of developing agrarian parks to protect
highly productive land and agricultural activities in areas of intensive urban development. The
conclusions explore the reasons why similar land management options cannot be adopted
equally in every region. We describe the development process and compare stakeholder maps
in both areas to deal with these differences in an attempt of providing tools and guidelines for
the global and local decision making processes.
References
Hernández Jiménez, V., 2007. Participatory Planning in the region of Madrid (Spain): An
integrative perspective. PhD thesis, Newcastle University, UK.
Jeffrey, P. & R. Seaton, 2003. Environmental Sciences 1(3): 277-300.
Winder, N. (Ed.), 2006. ISBP Problem Statement. http://www.tigress.ac/isbp (July, 2007)
541
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Determinants of policy-oriented research design
B. König, K. Helming, K. Tscherning, H. König, K. Diehl
Leibniz-Centre for Agricultural Landscape Research (ZALF) e.V.,
Eberswalder Str. 84, 15374 Müncheberg, Germany
Contact: koenig@zalf.de
Introduction
Science, policy and the public have different logics in following the agenda for a sustainable
development. Therefore, scientists have to deal with the different perspectives of politics in
order to make scientific advances useful to enable political framing of the conditions for
sustainable development. At the same time scientists are ethically bound to their professional
standards. This can impose some challenges to research design, investigation and result
communication. What is more, engagement of scientists at the science-policy interface is
experienced in many cases as rewarding but also immensely frustrating (Pannell, 2008), given
the poor uptake of research results in policy-making processes.
Several authors have dedicated their exploratory interest to the understanding of the
different environments of science on the one side and politics on the other in order to explain
reasons for this low adoption phenomenon (e.g., Pielke, 2004; Lackey, 2007; Pielke, 2007).
In this paper, we argue that efficient science policy interface requires the consideration of
policy requirements already from the onset of the design of the research project. This
procedural view shall illustrate how practical challenges of science-policy-interaction occur
and how they can be addressed.
Method
This paper is based on an extensive literature review, which informed the policy-oriented
SENSOR project about principles of science-policy interaction. Based on this, the sciencepolicy interaction experiences of the SENSOR coordination team were theoretically reflected.
Results and discussion
With science policy on the one side and the key role knowledge should play today for
legitimizing political decisions on the other, science and policy are two societal fields
reflexively influencing the preconditions of action for each other. But what conclusions have
to be drawn for policy relevant research design?
We argue that a precondition for policy relevant research design is the understanding of the
political background, actors and their logics that should be informed by science. The
SENSOR project provides one example for incorporating this understanding and translating
its meaning into the research process (Tabbush et al., 2008; Thiel & König, 2008).
Politicians ask science to deliver information which helps to grasp the wider picture, to
reveal the central causal relationships and to help to make broad developments visible, to
enable them to evaluate with overview and orientation on future developments (Kropp &
Wagner, 2007). Sustainability, including the assessments of conflicting targets, is one field
where politicians ask for scientific expertise to balance emotional public debates by avoiding
risky and controversary decisions through neutral scientific evidence (Bogner & Torgersen,
2005). Science with a policy perspective should make a broad range of action options visible
for decision makers, providing state of the art scientific knowledge for each option (Pielke,
2004).
To date most research projects are designed by more or less interdisciplinary composed
group of scientists acting with the intention to close knowledge gaps in their particular field of
research. However, these knowledge gaps might not be identical with those identified in the
policy-making process. Moreover, policy making might require a different perspective to the
542
Session D2: Role of integrated assessment for policy making and evaluation
research issue and different methodological approaches. These perspectives and questions
have to be considered in order to make the research results exploitable for policy design. To
overcome this inconsistency, policymakers would ideally be directly involved in shaping the
research questions, setting up the research group and designing the research project. However,
this ‘ideal’ approach is hindered by numerous restrictions (see Pannell & Roberts, 2008).
Facilitation means are, therefore, warranted that support research designers in adopting the
perspectives of policymakers in the elaboration and execution of research projects. These
means should feed into the research design process as early as possible, preferably already at
the level of topic selection and definition of evaluation criteria at the research funding level,
and further from the onset of project design at the project coordination level. Research has to
consider the relevant spatial and temporal dimension associated to the policy question, has to
analyse how policy affects the particular issue under question – the policy sensitivity of the
underlying process.
The central focus of our contribution lies in the transfer functions (e.g., scaling, translation,
communication and timing) policy relevant research projects have to fulfill to close the gaps
at the science-policy interface and how these functions can be facilitated by research design
and management. By doing so, we want to draw meaningful conclusions in research project
design and management to meet adoption criteria such as legitimacy, relevancy, credibility,
evidence and the like
References
Bogner, A. & H. Torgersen, 2005. In: Eds A. Bogner & H. Torgersen, Wozu Experten?
Ambivalenzen der Beziehung von Wissenschaft und Politik. Verlag für
Sozialwissenschaften, Wiesbaden, pp. 7-29.
Kropp, C. & J. Wagner, 2007. In: Eds C. Kropp, et al., Die Zukunft der
Wissenskommunikation. Perspektiven für einen reflexiven Dialog von Wissenschaft und
Politik - am Beispiel des Agrarbereichs. Ed. Sigma, Berlin, S. 7-15.
Lackey, R.T., 2007. Conservation Biology 21(1): 12-17.
Pannell, D.J. & A.M. Roberts, 2008. Conducting and delivering integrated research to
influence land-use policy: An Australian case study. INFFER Working Paper 0803,
University of Western Australia, Perth. http://cyllene.uwa.au/~dpannell/dp0803.htm
Pielke Jr, R.A., 2004. Environmental Science & Policy 7 (2004), pp. 405-417.
Pielke Jr, R.A., 2007. The honest broker. Making sense of science in policy and politics.
Cambridge University Press, Cambridge, 188 pp.
Tabbush, P., et al., 2008. In: Eds K. Helming, et al., Sustainability Impact Assessment of
Land Use Changes. Springer, Heidelberg, pp. 35-54.
Thiel, A. & B. König, 2008. In: Eds K. Helming, et al., Sustainability Impact Assessment of
Land Use Changes. Springer, Heidelberg, pp. 55-74.
543
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Integrated impact assessment as a new institution:
The political dimensions of modelling-tool use
Martina Padmanabhan
Institut für Wirtschafts- und Sozialwissenschaften des Landbaus,
Landwirtschaft-Gärtnerische Fakultät,
Humboldt Universität zu Berlin, Germany
Contact: Martina.padmanabhan@agrar.hu-berlin.de
Introduction
Modelling tools are not pure and isolated instruments which simply evaluate the potential
outcomes of different policy options as information to policymakers. Rather this instrumental
perspective on the role of modelling tools ignores the intrinsic political characteristic of the
policy-making process. The aim of the paper is to reflect on the political nature of modellingtools in the process of policy making by analysing the rationales of policy developers in the
European Union in their use of ex-ante models with an institutional approach. Within the
highly dynamic environment of impact assessment and greater importance paid to
participation and transparency in the European Union, the current use of existing modelling
tools can provide knowledge for a better understanding of the policy-science interface. The
focus on the ex-ante evaluation of policies with the outspoken aim of sustainability in a crosssectoral manner poses questions of its institutional arrangements within the policy-making
process, which is by very nature a political one (Padmanabhan & Beckmann, 2009).
This paper reflects on the experiences collected during the formulation of SIAT
(Sustainability Impact Assessment Tool), developed for the ex-ante assessment of policy
outcomes on integrated sustainability issues concerning land use by the SENSOR consortium.
As Thiel and colleagues (Thiel et al., 2007; Thiel & König, 2008) mention any policy
assessment procedure encounters existing implicit or explicit policy assessment practices and
thereby reconfigures them, resulting in institutional change. Though ‘good’ Impact
Assessment is an integral to policy development, only those regarding salient new regulations,
substantial revisions or expenditure programmes require in-depth modelling according to the
principle of proportionate analysis. Salience depends on the type of issues and interests a
proposal impinges upon and their political significance at the European level. The quality of
the Impact Assessment is of increasing importance for getting a policy adopted by the
hierarchy and other European legislative bodies. As Thiel (2008) shows, individual desk
officers have different reactions to Impact Assessments. DGs and units that are intensely
involved in Impact Assessments recruit economists for Impact Assessments or establish units
that provide back-up capacity. Desk officers have a preference to get the Impact Assessment
procedure right in order to guide policies to adoption, motivated by a career advance which
implies gains in remuneration and prestige. Superiors often have less technical know-how but
more sensitivity for the political aspects of a proposal. Desk officers apply a heuristic of
criteria (partly codified in guidelines) that indicate how to steer policy development and
Impact Assessment:
Methods
To analyse the institutional setting for model end users in the European Commission,
qualitative interviews were conducted in different Direction Generale, building on a previous
round of expert interviews. The literature review, document and interview analysis builds on
the Institutional Analysis and Design Framework by Ostrom (2005). An overview to the fast
moving environment of impact assessment in the European Commission is given. The
literature section reviews the recent findings on evidence-based policy support tools.
544
Session D2: Role of integrated assessment for policy making and evaluation
Results and discussion
The findings are structured as follows: At the example of an integrated ex-ante assessment
tool, the institutional analysis focus on dimensions relevant to this cross-sectoral approach.
The political term of multi-functional agriculture provides a key rationale and indicates the
necessity to reflect on political colourings within research. In the same line, the concept of
land use as a central feature to bundle different use perspectives on resources is inherently
political, as it frames issues beyond organizational terrain. A further challenge and at the same
time an interesting entry-point into the discussion on norms and values, which translate into
political interests, is the outspoken policy goal of sustainability. While the overarching policy
goal is fixed, the increasing demand for stakeholder participation in the policy-making
process poses questions of influence on the assessment and vulnerability to manipulation.
Visualizing policy outcomes in a spatial manner of coloured maps on different levels is highly
sensitive and can bear political risks for users. The credibility of the data pool on which
results rest is linked to the reputation of researchers and organizations within the community
and therefore an outcome of a social process of recognition and denial open to political
interventions. Form these deliberations on former dimensions of an ex-ante impact assessment
modelling tool, consequences for the targeting of its end-users and their political
characteristic can be drawn.
As a conclusions the paper reflects on the political dimensions of the Sustainability Impact
Assessment Tool developed by SENSOR and on the institutionalizing of policy evaluation
instruments.
Reference
Ostrom, E., 2005. Understanding Institutional Diversity. Princeton University Press,
Princeton, NJ.
Thiel, A., et al., 2007. In: Eds K. Helming & H. Wiggering, SENSOR Report Series 2007/01,
www.sensor-ip.eu, ZALF, Germany.
Thiel, A., 2008. Changing administrative culture and political debate? A review of Impact
Assessment practices in the European Commission, in: Zeitschrift für Umweltpolitik &
Umweltrecht
Thiel, A. & B. König, 2008. In: Eds K. Helming, et al., Sustainability Impact Assessment of
land use policies. Springer, Heidelberg.
Padmanabhan, M. & V. Beckmann, 2009. Institutions and Sustainability: Introduction and
Overview. In: Eds V. Beckmann & M. Padmanabhan, Institutions and Sustainability.
Springer, Heidelberg. (forthcoming)
545
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Contribution of the development process to the actual implementation and
use of Integrated Spatial Decision Support Systems
H. Van Delden, R. Vanhout
RIKS, P.O. Box 463, 6200 AL Maastricht, The Netherlands
Contact: hvdelden@riks.nl
Introduction
Integrated Spatial Decision Support Systems (ISDSSs) are rapidly gaining traction in the
planning and policy-making community. They can have a high added value for planning
processes by bringing scientific knowledge to the decision makers’ table. However, only a
few ISDSSs are in actual use for the preparation of spatial policies. Several reasons for this
are recognized in academic literature, most notably a lack of transparency, inflexibility and a
focus on technical capabilities rather than on real planning problems (Vonk et al., 2005). In
order to deploy an ISDSS as an instrument for strategic policy making, it seems crucial that it
matches the perceptions, experiences and operational procedures (the tacit knowledge) of the
policymakers and that it enhances their current policy practices rather than replace existing
and well-embedded ones (see, e.g., Van Delden, 2003).
Based on practical experience, and more thoroughly discussed in earlier papers (e.g., Van
Delden et al., 2007), there are seven elements that seem to determine the success or failure of
the implementation of an ISDSS:
1) Strategic value: to what extent does the system provide an added value to the current
planning practice?
2) Availability of appropriate data and models: what is available at present or can easily be
collected?
3) Credibility of the system: do the users have faith in underlying assumptions?
4) Domain language of the system: does it fit the worldview of the end users and connect to
their perception?
5) Institutional embedment: where will the system be based in the organization and who is
actually going to use the system?
6) Culture: are people willing to adopt and use the system and is there commitment to give
the system a place in the planning process?
7) Ease of use: does the user interface allow a quick and simple use and provide easy access
to all relevant functionality?
Methods
This paper discusses the lessons learned in the design & development of spatial decision
support systems over the past decades. It builds on practical experience of ISDSS
development using the GEONAMICA software environment for spatial modelling and
(geo)simulation (Hurkens et al., 2008). GEONAMICA has been the basis for many integrated
spatial decision support systems that vary greatly in their application domain (urban and rural
areas, coastal zones, river basins) and spatial extent (cities, countries, EU-27), based on the
requirements of the user. Examples of ISDSSs developed with GEONAMICA are WadBos,
Environment Explorer (Engelen et al., 2003), MedAction (Van Delden et al., 2007), Xplorah,
Elbe-DSS, LUMOCAP and MOLAND (Barredo et al., 2003).
From the development of several ISDSS over the past decades important lessons for ISDSS
development processes can be learned. An ideal development process can best be described as
an iterative process of communication and social learning amongst three involved parties.
Such a process, that mirrors insights in other planning support domains, is depicted in Figure
1 and described in more detail in Hurkens et al. (2008).
546
Session D2: Role of integrated assessment for policy making and evaluation
Figure 1. Main parties, responsibilities and integration issues during the development of a
SDSS.
Results and discussion
In the introduction of this abstract we have posed seven elements that determine the success
or failure of the implementation of an ISDSS. After a closer investigation of the development
process of ISDSS, it becomes clear that this process plays a crucial role in the actual uptake of
an ISDSS. Through close interaction between the different parties involved (end-users,
scientists, IT-specialists and architect) the ISDSS under development can be greatly enhanced
on items related to its usefulness and usability. Moreover, an iterative development process
can also lead to an improved uptake of the system in the organizations and overcome some of
the problems mentioned regarding the institutional context and the willingness of people to
use and adopt it as part of their daily practice. The presentation will detail on the contribution
of the development process to each of the seven elements mentioned.
References
Barredo, J., et al., 2003. Sustainable urban and regional planning: The MOLAND activities on
urban scenario modelling and forecast. Office for Official Publications of the European
Communities, Luxembourg.
Engelen, G., et al., 2003. Integrated Assessment 4(2): 97-105.
Hurkens, J., et al., 2008. In: Eds M. Sànchez-Marrè, et al., Proceedings of the iEMSs Fourth
Biennial Meeting: ‘Integrating Sciences and Information Technology for Environmental
Assessment and Decision Making’. International Environmental Modelling and Software
Society, Barcelona, Spain.
Van Delden, H., 2003. Design and use of decision support systems for spatial planning.
Framing Land Use Dynamics: Integrating knowledge on spatial dynamics in socioeconomic and environmental systems for spatial planning in western urbanised countries.
Faculty of Geographical Sciences, Utrecht University, Utrecht, The Netherlands.
Van Delden, H., et al., 2007. Environmental Modelling and Software 22(2): 223-238.
Vonk, G., et al., 2005. Environment and Planning A, 37: 909-924.
547
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Design and institutional embedment of the meta-model
Sustainability Impact Assessment Tools (SIAT)
S. Sieber, P.J.F.M. Verweij, K. Müller, K. Tscherning, K. Helming
Leibniz-Centre for Agricultural Landscape Research e.V.,
Eberswalder Str 84, 15374 Müncheberg, Germany
Contact: stefan.sieber@zalf.de
Introduction
The European Union funded Integrated Project SENSOR develops ex-ante Sustainability
Impact Assessment Tools (SIAT) to support decision making on policies related to multifunctional land use in European regions (Helming et al., 2008). The meta-modelling approach
SIAT is the central product of the SENSOR project, which innovates sustainability impact
assessment towards an integrated perspective of region-explicit economic, environmental and
social trade-off impact analysis. This paper illustrates the process of end user involvements
and software prototyping to design the SIAT under limiting budget capacities. It concludes on
a feasible design process under specific project-funded conditions.
The meta-model SIAT
Current operational tools for Impact Assessment (IA) are compared to SIAT mostly restricted
to precise, quantitative sector information on aspects of economic, social or environmental
impacts that are mainly designed for ex-post analysis (Bartolomeo et al., 2004). Along with
institutional integration processes, these ex-ante impact assessment tools are accompanied
with increased requirements on system- and interdisciplinary complexity (Tamborra, 2002).
Figure 1. Design of prototype II.
The challenge of developing SIAT is the transformation of interdisciplinary knowledge
into a meta-model design that allows high (a) technical performance, (b) a high level of
integration and (c) compatibility and flexibility for further developments and assures (d)
standard criteria such as reliability, plausibility, accuracy and explicitness of simulation
results.
The integrated analysis by means of SIAT allows iterative policy scenario solving across
the six sectors agriculture, forestry, energy, transport, nature conservation and tourism related
to multi-functional land use in 570 European regions at the level of the EU 27 (Sieber et al.,
2008). End users are able to preselect policy variables (e.g., subsidy) and to choose intensities
(e.g., Million Euro) to be simulated by using pre-calculated response functions from a model
framework. 45 implemented impact indicators analyse trade-offs among sets of iteratively
solved scenarios in a broad sustainability view of ‘quick scan’ analysis.
SIAT development process
Software prototyping focuses on the design process towards the final model version, which
should be tailored for the specific use of end user groups. Typically, subsets of features
548
Session D2: Role of integrated assessment for policy making and evaluation
demonstrate functionalities of specific domains to be developed step-wise in collaborative
discussions with interdisciplinary researcher.
The SIAT prototype I was released in 2006 (Verweij et al., 2006). This version contained a
small subset of the features to demonstrate the functionality of simulating land use-related
policy impacts. Group discussions as well as individual interviews with potential end users
surveyed expectations and requirements for the operational prototyping. The feedback has
supported accurate estimates on given capacities for software specification and the level of
integration on thematic areas (variety on indicators, fact sheet information etc.). As a result,
the final version of SIAT is being developed until May 2009.
Concluding findings
The meta-model concept caused specific needs for knowledge integration by means of nonstandard technical solution finding. A balanced resource allocation between (i) technical
model advancements and (ii) increased levels of integration by covering an escalated number
of thematic areas for impact assessment and (iii) institutional embedment to survey and meet
end user requirements are predominant objectives. While the ratio between resource use for
(i) and (ii) is relatively pre-defined, institutional embedment seems to be more flexible and a
key factor for success for a high model acceptance and thus probability of further funding:
• Group discussions with end users is a valuable instrument, whereas key contacts to
potential end users and permanently alternate meetings are most crucial obstacle.
Technical requirements and desired analytical objectives could be surveyed.
• Targeted inputs of experts in bilateral interview form are essential, because ‘strategic
opinions’ in group discussion are by-passed and thus sincere options can be focused.
Design questions on anticipated, most likely research requirements could be answered.
• Understanding interactive development processes with end users helped to steer the
general model design towards a presumed, most likely acceptability; clear, detailed
requirements lacked often.
• Knowing the institution regarding its organizational structure is an empiric key for
efficient result-oriented end user collaboration on specific requirements of integrated
impact assessment models (Sieber et al., 2008).
Becoming acquainted the organizational structure and requirements through group
discussions and individual interviews is key for success, but often rather general and diverse
requests cause the need for presumptions. Strategic behaviour is major obstacle for honest
feedback. The presentation will present SIAT and conclude on risks and success factor as
well as best practices, which are newly elaborated and base on experiences of the
development process of SIAT. Explanations are made by means of applied economic theory.
References
Bartolomeo, M., et al., 2004. Approaches to Impact Assessment in six OECD countries and at
the European Commission. IQ-Tools Project report, WP1A.
Helming, K., et al., 2008. In: K. Helming, et al., Sustainability impact assessment of land use
changes. Springer, pp. 77-105.
Sieber, S., et al., 2008. In: K. Helming, et al., Sustainability impact assessment of land use
changes. Springer, pp. 107-128.
Tamborra, M., 2002. Socio-economic tools for sustainability impact assessment – the
contribution of EU research to sustainable development. Energy, environment and
sustainable development. EUR 20437.1 Office for official publications of the European
Communities, Luxembourg, 2002.
Verweij, P.J., et al., 2007. SIAT - a Sustainable Impact Assessment Tool for Understanding
the drivers in integrated impact assessment. International Conference IEMSS, Vermont,
USA, 6 pp.
549
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Bridging the gap between science and decision makers for sustainable
development of multi-functional land use in developing countries
H. König, K. Helming, P. Reidsma, L. Zhen, S. Feng, S. Sieber, F. Brouwer
Leibniz-Centre for Agricultural Landscape Research, ZALF e.V.,
Eberswalder Strasse 84, 15374 Müncheberg, Germany
Contact: Hannes.Koenig@zalf.de
Introduction
Sustainable land use is a key issue to improve food security, control land degradation and
reduce poverty, especially in developing and transition countries. Global changes and
growing demands on resources have resulted in dynamic changes of land and related
sustainability issues. Hence policy- and decision makers have increasing demand for ex-ante
assessment tools to support decision making processes in a sustainable way. The EU-funded
projects LUPIS (www.lupis.eu) and SENSOR-TTC (www.sensor-ip.org; Helming et al.,
2008) develop methods and tools for assessing policy- and/or alternative problem scenario
impacts on land use and sustainable development (SD) in a selected number of case study
regions in Africa, Asia and Latin America.
When developing scientific tools, the focus is often on specific scientific questions, while
the applicability and usability for stakeholders is often neglected. To have impact however,
communication with stakeholders and usability of tools is of major importance. Therefore, the
assessment, communication and visualization tool ‘Pro-Vision’ is developed. Pro-Vision is
used to support the impact assessment procedure considering participatory approaches and to
evaluate anticipated impacts on sustainable development issues.
Methodology
In Pro-Vision, the analytical chain is developed based on the Driver-Pressure-State-ImpactResponse (DPSIR) framework (OECD, 2003), to structure different case study regions and to
identify cause-effect relations between policies/problems, land use changes and regional
sustainability issues. The DPSIR framework is used to (a) develop regional specific land use
change scenarios, and to (b) evaluate possible impacts of land use changes on sustainable
development (Figure 1).
For the scenario development, Pro-Vision is first applied to elaborate and visualize
alternative land use change scenarios, e.g., during regional stakeholder workshops. Stakeholders and experts are asked to specify possible land conversion dynamics (type specifications, suitability and allocation dynamics), and to discuss possible scenario outcomes.
For the second part, Pro-Vision is used to evaluate possible land use change impacts on
sustainable development. The concept of Land Use Functions (LUFs) as developed by PérezSoba et al. (2008) is used in the Pro-Vision tool to assess in an integrative way the economic,
environmental and social impacts that land use changes have on sustainable development.
Land Use Functions are developed to illustrate most relevant sustainability issues at regional
level and are defined as goods and services associated with land use (e.g., economic: food
production; environmental: maintenance of ecosystem processes; social: provision of work).
For the evaluation of impacts, stakeholders and experts are asked to identify a key set of Land
Use Functions and to define causal-chain dependencies between land use and sustainability.
This is done by assigning region specific weights between land use types and Land Use
Functions.
In the final step, sustainable development can be interpreted using the LUFs framework, to
allow stakeholders, and scientists identifying at a glance those functions of the land which are
fostered or hindered under various scenarios of land use changes, and making it possible to
explore the trade-offs between them.
550
Session D2: Role of integrated assessment for policy making and evaluation
Figure 1. The methodological framework of the Pro-Vision tool is based on the DPSIR
framework, and is used to support (a) participatory scenario development and (b)
sustainability evaluation.
Results, discussion and conclusion
The two projects LUPIS and SENSOR-TTC aim at assessing alternative policy/problem
scenarios in different case study regions. The methodological framework of both projects
builds upon European assessment methods (Reidsma et al., 2008; Helming et al., 2008), and
is currently being adapted and tested in a selected number of case study regions in China and
Brazil to identify strengths and weaknesses of this approach, and also to provide a primer for
sustainability impact assessment in developing countries. A heterogeneous set of land use
related problems is assessed: water pollution in China, and side effects of infrastructure
projects and land conversion in Brazil.
By using Pro-Vision, integrated assessments of land use policies are possible that do not
depend on the use of complex tools, but by complementing quantitative and qualitative
approaches which are insightful for decision makers and other stakeholders. In order to bridge
the gap between scientists and decision makers, participatory communication and
visualization tools are needed to support the understanding of complex causal-chain
relationships between policies and their impacts on regional sustainability issues. The ProVision tool has been developed to support this process. Pro-Vision enables users and
stakeholders to identify causal-chain relationships between land use changes and regional
sustainability issues and improve their decision-making process.
References
Helming, K., et al., (Eds), 2008. Sustainability Impact Assessment of Land Use Changes.
Springer, Berlin.
OECD, 1993. Core set of indicators for Environmental Performance Reviews. Environmental
Monographs 83. OECD, Paris, France.
Pérez-Soba, M., et al., 2008. In: Ed. K. Helming, et al., Sustainability impact assessment of
land use changes. Springer, Berlin, pp. 375-404.
Reidsma, P., et al., 2008. D4.1, LUPIS Integrated Project, Wageningen University.
551
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Barroso, F., 100
Baruth, B., 350, 352
Bates, B., 414
Bathgate, A., 270
Baudet, D., 202
Baveco, J.M., 118
Beça, P., 36
Beletskaya, O., 310
Belhouchette, H., 130, 194, 196, 198, 200,
442, 484
Bellon, S., 178
Ben El Ghali, M., 250
Benedetti, L., 136
Berbel, J., 96, 182
Berg, E., 286
Berges, R., 78
Bergez, J.E., 198, 212, 226, 232, 326, 448
Berkhoff, K., 18
Beusen, A.H.W., 146
Bezabij, Y.G., 404
Bezlepkina, I., 56, 58, 66, 130, 138, 140,
176, 514
Bianchi, F.J.J.A., 118
Biberacher, M., 312
Bigot, G., 262, 466
Billen, G., 146
Binder, C.R., 172
Bindi, M., 426
Blanco, M., 256
Blandford, D., 86
Blaser, S., 334
Blazy, J.M., 294
Blouin, M., 448
Bobojonov, I., 286
Bocardo, P.J., 364
Bochu, J.L., 430
Bockstaller, C., 66, 130, 140, 176, 178
Bocquého, G., 252
Bona, S., 366, 446
Bonin, M., 294
Boote, K.J., 406
Bordenave, P., 452
Borkowski, N., 256
Bossio, D., 22
Botreau, R., 322
Böttcher, H., 416
Bouraoui, F., 240, 392
Bourges, J., 326
Bousset, J.P., 100, 176, 466, 520, 532
Bouwman, A.F., 146, 158, 412
Brady, M., 288
Authors Index
Achuo, G., 152
Adam, M., 194, 196, 198, 200, 342
Adamowicz, M., 174
Adams, D., 34
Adenäuer, M., 24, 52, 436
Adiku, S., 218
Adriaensen, K., 396
Agabriel, J., 264, 266
Aguerre, V., 60
Ahuja, L., 324
Akkas Ali, M., 248
Alary, V., 268, 272
Albrecht, A., 40
Alcamo, J., 378
Alkan Olsson, J., 6, 66, 130, 140, 176, 178,
504, 506, 514, 520, 530, 532
Amblard, L., 438
Andersen, E., 6, 56, 130, 320, 350, 352,
354
Anderson, T., 490
Andersson, L., 506
Andrieux, P., 294
Angenendt, E., 310
Angevin, F., 448
Antle, J.M., 48, 68, 110, 318, 408, 472
Antrop, M., 112
Antunes, P., 36, 440, 508
ApSimon, H., 538
Arnauld de Sartre, X., 356
Arriaza, M., 16, 84
Ascough, J., 324
Asseng, S., 414
Astover, A., 372, 386
Athanasiadis, I.N., 140, 320, 330, 344
Attard, S., 490
Auclair, D., 296
Ayanu, Y.Z., 28
Aznar, O., 302
Bäcklund, A.-K., 530, 532
Baethgen, W.E., 402
Bagstad, K., 120
Baijukya, F., 40
Bakker, F., 102
Bakker, M.M., 412
Bamière, L., 252
Banse, M., 162, 384
Barbier, J.M., 180, 294
Barnaud, C., 524
553
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Crespo, O., 410
Cruz, J.C.S., 364
Cullen, B.R., 236
D’Haene, K., 500
Dahal, B.M., 534
Dahlmann, C., 222
Daniell, T., 32
Das, S., 242
David, O., 324
Davis, A., 490
De Blaeij, A., 88
De Cara, S., 254
De Castro, V.M., 102
De Koeijer, T.J., 108
De Kraker, J., 498
De la Flor Tejero, I., 14
De Lange, W.J., 338
De Mey, K., 500
De Ridder, N., 40, 526
De Sanctis, G., 460
De Smedt, P., 128
De Visser, C.L.M., 394
De Vries, S.C., 394
De Vries, W., 74, 298, 412, 454
De Winter, W.P., 412
Debaeke, P., 326
Debortoli, N., 102
Deffuant, G., 64
Delmotte, S., 180
DeVoil, P., 450
Dhillion, S.S., 90
Diehl, K., 542
Dietz, F.J., 116
Dillon, E.J., 476
Diman, J.L., 294
Dimes, J., 432
Dirkx, G.H.P., 92
Dixon, J., 22
Dogliotti, S., 60
Dolédec, S., 356
Donatelli, M., 206
Dos Santos Rocha, A., 368
Dray, A., 414
Duru, M., 502
Dury, J., 62
Dux, D., 334
Dzotsi, K., 218
Eickhout, B., 148, 164, 384, 412
Elbersen, B.S., 54, 74, 94, 240, 350, 354,
462
Elzen, B., 38
Breukers, A., 290
Brigi, A., 366, 446
Bringezu, S., 376
Brisson, N., 222, 224
Britz, W., 50, 52, 144, 154, 184
Brodie, J., 490
Brogaard, S., 514, 530, 532
Broll, G., 312
Brouwer, F., 6, 64, 114, 130, 138, 530, 550
Brun, J.-P., 322
Bryon, K., 258
Buckwell, A., 82
Buis, S., 326
Bulatewicz, T., 292
Bunce, R.G.H., 94
Burgess, P.J., 458
Burghard, M., 64
Bursztyn, M., 102
Buysse, J., 160, 486
Cacovean, H., 444
Cadisch, G., 26, 28, 510
Cai, X., 242
Calgaro, A., 366
Carberry, P.S., 192
Carlson, J., 324
Carpy-Goulard, F., 448
Casellas, E., 194, 196, 198, 200, 204, 228,
232
Causeret, F., 294
Cavigelli, M.A., 230
Celette, E., 196, 204
Ceroni, M., 120
Champciaux, P., 322
Charles, S.P., 414
Charron, M.-H., 202
Chen, Z., 474
Christen, O., 20
Ciaian, P., 42
Claessens, L., 110, 408
Claeys, D., 160, 486
Clastre, P., 326
Clavel, L., 202
Coletto, L., 366, 446
Colomb, B., 448
Comunello, E., 158
Cools, J., 136
Cooper, M., 242
Corbeels, M., 40, 196, 204, 238
Corporaal, A., 108
Corre-Hellou, G., 222
Costamagna, A.C., 118
554
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Hachigonta, S., 410
Hallett, P., 32
Hansen, J.W., 216, 402
Happe, K., 64, 308
Hasnain, N., 538
Hatano, R., 422
Hauggaard-Nielsen, H., 222
Havlik, P., 148, 156, 370, 416
Havlikova, M., 492
Hawes, C., 32
Hazeu, G.W., 350, 352
He, J., 218
Heckelei, T., 6, 56, 58, 130, 154, 168, 256
Hecker, M., 256
Hein, L., 124
Heinemann, A.B., 214
Helming, K., 14, 132, 542, 548, 550
Hengsdijk, H., 200, 232, 256, 262, 478
Hennessy, T., 476
Hernández Jiménez, V., 540
Herrero, M., 22, 40, 148, 408
Herrmann, S., 18, 188
Hertel, T., 144
Herzog, F., 458
Hilger, T., 26, 28
Ho Van Chien, 512
Hochman, Z., 450
Hoffmann, C., 306
Hoffmann, V., 26
Holzworth, D.P., 192, 450
Hood, A., 34
Hoogenboom, G., 406
Hoogervorst, N., 164
Howart, R.W., 376
Huang, X., 474
Huber, D., 330
Hubert, B., 356
Hugenschmidt, C., 28
Hunt, J., 450
Hutchinson, W.G., 112
Huth, N.I., 192
Hynes, S., 476
Ines, A.V.M., 216, 402
Jacquet, F., 252, 254, 370
Jafarzadeh, A.A., 304
Jakku, E., 522
Janetschek, H., 456
Janssen, S., 66, 130, 140, 232, 256, 262,
320, 330, 332, 344, 350, 478
Jansson, T., 24, 162
Jensen, E.S., 222
Emmett, B.A., 106
Erdélyi, É., 208
Evans, C.D., 106
Everaert, P., 258
Ewert, F., 6, 56, 66, 70, 130, 140, 176,
200, 504, 520
Fallot, A., 370
Fantke, P., 210
Fatondji, D., 218
Feng, S., 138, 550
Ferrise, R., 426
Fischer, R., 514, 536
Flichman, G., 56, 246, 256, 442
Florio, G., 446
Flury, C., 482
Fodorean, I., 444
Franz, J., 286
Freeman, H.A., 22, 148
Fricke, K., 132
Friedrich, R., 210, 310
Fritz, S., 148
Gaasland, I., 86
Gabbert, S., 504
Gaillard, G., 134, 334
Garcia-Barrios, L.E., 516
Gardiner, M.M., 118
Garrod, G., 114
Gary, C., 180, 228, 294
Geniaux, G., 178
Gerard, B., 22
Gerten, D., 148
Geter, F., 324
Giannoccaro, G., 96, 182
Gies, T.J.A., 298
Giller, K.E., 40, 62, 238
Glandières, A., 448
Goddard, L., 402
Godeschalk, F.E., 358
Gondwe, P., 432
Gooding, M.J., 222
Goulevat, G., 254
Grabaum, R., 34
Grant, T., 388
Graves, A.R., 458
Greene, A.M., 402
Griffon, S., 296
Groot, J.C.J., 518
Guérif, M., 326
Guillaume, S., 212
Guille, T., 370
Guo, Z., 220
555
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Lamadon, A., 322
Lamers, J.P.A., 286
Landis, D.A., 118
Lange, A., 480
Langeveld, H., 394
Lankoski, J., 374
Laplana, R., 64
Launay, M., 222, 224, 326
Lauwers, L., 160, 258, 278, 486
Lavelle, P., 356
Lebonvallet, S., 224
Lecharpentier, P., 326
Lecomte, P., 268, 272, 280, 430
Leduc, S., 416
Leenhardt, D., 202
Lefebvre, A., 152
Leffelaar, P.A., 70
Legg, W.J., 4
Leip, A., 184, 376
Leneman, H., 260
Lescot, J.M., 452
Lesschen, J.P., 74, 158, 454
Lherm, M., 264, 266
Li, H., 256
Liebhart, P., 234
Lieffering, M., 236, 424
Linderhof, V., 88
Lindoso, D., 102
Lippe, M., 26, 28
Liski, J., 412
Long, B., 450
Longo, A., 112
López-Baldovín, M.J., 84
Lotze-Campen, H., 148
Louchart, X., 294
Louhichi, K., 104, 130, 246, 256, 262, 276,
436, 442, 484
Lubbers, M., 268, 272
Lucas, P., 254
Lusiana, B., 510
MacGregor, R.J., 152
Machado, C., 100
Mack, G., 482
Mailhol, J.C., 226
Majewski, E., 484
Man, T., 444
Mangum, R.A., 230
Mann, C., 438
Mapfumo, P., 40
Marchand, F.L., 160, 486
Marohn, C., 28
Jeuffroy, M.H., 254
Joel, A., 336
Johnson Jr, G.W., 120
Johnson, I.R., 236
Jones, J.W., 218, 406
Jones, M.L.M., 98
Jongeneel, R., 54
Jonsson, A., 506
Jonsson, B., 130, 330, 344
Josien, E., 178, 466
Justes, E., 212, 326
K’ungu, J., 40
Kaizzi, C., 40
Kan, I., 418
Kanellopoulos, A., 58, 256, 478
Kansou, K., 228
Karadzic, V., 440, 508
Karanja, S., 40
Kasnakoglu, H., 348
Kassahun, A., 328
Kasynova, E., 222
Keating, B.A., 192
Kellermann, K., 288
Kempen, M., 54, 74, 354, 454
Kephaliacos, C., 250
Khanh Hoang, L., 136
Kim, C.-G., 420
Kimhi, A., 418
Kimura, D.S., 364
Kimura, S.D., 422
Kindermann, G., 156, 416
Kirschner, P., 498
Knapen, M.J.R., 130, 140, 330, 344
Koch, J., 378
Koduru, P., 242
Kok, K., 518
Kölking, C., 378
Kondoh, K., 364
König, B., 542
König, H., 138, 542, 550
Koo, J., 220
Kopeva, D., 64
Krivov, S., 120
Kroeze, C., 492, 498, 504
Kros, J., 74, 298, 412, 454
Kruitwagen, S., 116
Kruska, R., 22
Kuhn, A., 210
Kuiper, M., 24, 130, 150, 436
Kukk, L., 372, 386
Kummer, U., 310
556
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Nicetic, O., 512
Nidumolu, U.D., 268
Nieuwenhuizen, W., 340
Nolte, S., 160
Nonhebel, S., 380
Noormets, M., 372, 386
Notenbaert, A., 22, 148
Nowicki, P., 14
Nwage, D., 40
O’Connell, D., 388
O’Leary, Z., 512
Obersteiner, M., 156, 416
Ollikainen, M., 374
Oñate, J.J., 462
Oomen, R.J., 200, 232
Orsini, R., 460
Osann, A., 508
Osterburg, B., 300
Osuch, A., 308
Oude Lansink, A., 58, 256, 290
Oudendag, D.A., 412, 454
Oxley, T., 538
Ozier-Lafontaine, H., 294
Pacini, C., 40, 98
Pacini, G.C., 60
Padmanabhan, M., 544
Pahl-Wostl, C., 497
Palma, J., 458
Pannell, D., 470
Paracchini, M.L., 42, 98
Peco, B., 462
Peepson, A., 54
Pérez Dominguez, I., 58, 154
Perez, P., 414
Pérez-Soba, M., 98, 352, 412
Perny, P., 322
Perret, E., 114, 262, 466
Perugini, M., 460
Petersen, J.-E., 362
Peterson, J.M., 292
Philips, A., 34
Picchi, P., 100
Pingali, P., 11
Pinto-Correira, T., 100, 176
Piorr, A., 78, 308, 480
Pistón, J.M., 182
Polman, N., 88
Pop, A., 444
Popp, A., 148
Prieur, L., 448
Prins, A.G., 122, 384
Martin, G., 502
Martinez Alonso, P., 462
Martius, C., 286
Maul, J., 230
McRoberts, N., 30, 32
Meers, E., 396
Meinke, H., 400
Mendes da Silva, M., 368
Mérot, A., 204, 226
Metay, A., 198, 228
Metzger, M.J., 352
Meul, M., 186, 500
Meuter, E., 256, 332
Meyer, B., 34
Michelin, Y., 100
Mikk, M., 54
Milla, R., 490
Mirsky, S.B., 230
Moldovan, C., 444
Molla, M.S.H., 248
Moltchanova, E., 428
Momin, M.A., 248
Montfort, F., 254
Monti, M., 222
Moore, A., 270
Moore, R.V., 316
Morand, H., 152
Moriondo, M., 426
Mosnier, C., 264, 266
Mouratiadou, I., 484, 488
Mourits, M., 290
Moyer, B., 242
Msangi, S., 22, 148, 378
Mu, Z., 422
Mücher, C.A., 352
Mulia, R., 510
Muller, C., 148
Müller, K., 548
Müller, M.,154
Mwale, M., 40
Mynett, A., 136
Naab, J., 218
Nabuurs, G.J., 412
Nanja, D., 432
Neef, A., 26
Nekhay, O., 16
Nelson, R., 400
Nemecek, T., 134, 334
Neuwirth, J., 456
Newton, P., 424
Nguyen, G., 250
557
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Sasaki, H., 464
Satger, S., 222
Sattler, C., 308
Sauer, T., 156
Schaldach, R., 378
Schauppenlehner, T., 274
Schelhaas, M.J., 412
Schellhorn, N.C., 118
Schipper, R., 272
Schleyer, C., 302, 438
Schmid, A., 172
Schmid, E., 234, 274, 416, 428
Schmidt, T.G., 300
Schmitt, J., 242
Schneider, U.A., 156, 234, 416
Scholz, J., 188
Schönhart, M., 234, 274
Schoorl, J.M., 110
Schrijver, R.A.M., 108, 260
Schuler, J., 382
Schuler, U., 28
Seddaiu, G., 460
Shahbazi, F., 304
Shahbazi, M.R., 304
Shannon, E., 490
Shraberman, K., 418
Sieber, S., 132, 548, 550
Siebert, R., 480
Sijtsma, F.J., 116
Sintanyeho, R., 404
Slegten, V., 280
Smaling, E.M.A., 158
Smart, S., 106
Smarzewska, A., 174
Smidt, R., 298
Smith, J., 40
Snow, V.O., 236
Soudais, J., 202, 326
Spargo, J.T., 230
Speelman, E.N., 516, 518
Spoelstra, S.F., 38
Spooner-Hart, R., 512
Squire, G.R., 32
Staggenborg, S., 292
Stalpers, S., 504
Stapleton, L., 64, 114, 176, 178
Staritsky, I., 350, 354
Stehfest, E., 164, 384
Steinberger, J.K., 172
Steward, D.R., 292, 338
Stockstad, G., 256
Pristeri, A., 222
Prosperi, M., 96
Prosperi, P., 240
Puiu, V., 444
Qu, F., 474
Queste, J., 514
Quiroz, C., 40
Rae, D., 512
Raffaillac, J.-P., 224
Raley, M., 176, 178
Rapidel, B., 276
Rees, R.M., 390
Reidsma, P., 138, 514, 550
Reinds, G.J., 412
Renetzeder, Ch., 352
Reynaud, A., 254, 264, 266
Ridier, A., 250
Rienks, W., 14
Righi, E., 60
Ringler, C., 148
Rio, P., 294
Ripoche, A., 180, 228
Rizzoli, A.E., 130, 206, 320, 330, 344
Robertson, M.J., 192, 270
Robiul Alam, M., 248
Rodrigues-Filho, S., 102
Roe, J.L., 242
Roggero, P.P., 460
Romeo, M., 222
Romstad, E., 58, 104
Rood, T., 164
Roos-Klein Lankhorst, J., 340, 462
Roostalu, H., 372
Roscoe, R., 158
Rosegrant, M., 148
Rosenzweig, C., 406
Rossing, W.A.H., 60, 518
Rotmans, J., 8
Rowe, E.C., 106
Rudrum, D.P., 108
Ruffini, F.V., 306
Rufino, M.C., 40, 62
Ruinelli, L., 130, 344
Russell, G., 200, 206, 232, 488
Rusu, T., 444
Salazar, O., 336
Salgado, P., 272
Sambo, P., 446
Sandrini, S.G., 366, 446
Sanogo, O.M., 40
Santos, R., 36, 440, 508
558
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Van der Velde, M., 240, 392
Van der Werf, W., 110, 290, 458
Van Diepen, C.A., 350, 352
Van Doorn, A., 94
Van Eetvelde, V., 112
Van Egmond, P., 116
Van Eupen, M., 340
Van Evert, F., 342
Van Griensven, A., 404
Van Griensven, S., 136
Van Hinsberg, A., 116
Van Huylenbroeck, G., 160, 278, 486
Van Ittersum, M.K., 6, 14, 56, 130, 138,
200, 232, 256, 262, 332, 478, 504, 514,
520, 530, 532
Van Keulen, H., 70, 138, 268, 272, 518
Van Meensel, J., 258, 278
Van Meijl, H., 162
Van Noordwijk, M., 510
Van Os, J., 74
Van Paassen, A., 524, 526
Van Passel, S., 44, 186
Van Rees, S., 450
Van Slycken, V., 396
Van Wijk, M.T., 40, 62
Van Zeijts, H., 122
Vangronsveld, J., 396
Vanhout, R., 42, 546
Vanlauwe, B., 40
Vårdal, E., 86
Vayssières, J., 280, 430
Veiga, I., 356
Veissier, I., 322
Velthof, G.L., 412, 454
Verburg, P.H., 124, 412
Verhoog, D., 350, 358
Verweij, P.J.F.M., 340, 548
Viaggi, D., 480
Vigne, M., 280, 430
Villa, F., 120, 344
Volkery, A., 14
Voltr, V., 176
Von Braun, J., 2
Von Fragstein, P., 222
Voogd, J.C., 298
Wagner, K., 456
Wagner, S., 210, 310
Walker, R.L., 32
Walker, S., 432
Wallach, D., 200, 326, 342
Wallman, P., 344
Stomph, T.J., 72
Stoorvogel, J.J., 68, 110, 284, 318, 408,
472
Strauss, F., 428
Streifeneder, T., 306
Stroosnijder, L., 526
Stuczynski, T., 42
Sulser, T., 148
Suuster, E., 372, 386
Suyamto, D., 510
Svensson, M., 344
Tack, F.M.G., 396
Tadross, M., 410
Tagliafierro, C., 112
Taverne, M., 520, 532
Teasdale, J.R., 230
Temme, A., 110
Terluin, I.J., 350, 358
Terres, J.M., 240
Thai Minh, T., 26
Theesfeld, I., 302, 536
Theloke, J., 310
Therond, O., 56, 66, 130, 140, 176, 194,
198, 200, 232, 442, 520
Thewys, T., 396
Thorburn, P.J., 388, 490, 522
Thornton, P.K., 22, 148
Tittonell, P., 40, 62, 238
Toderi, M., 460
Toma, Y., 422
Topp, C.F.E., 30, 32, 390, 488
Traoré, B., 276
Trébuil, G., 524
Truchet, S., 100
Tscherning, K., 542, 548
Turpin, N., 56, 64, 66, 100, 114, 130, 140,
176, 466
Uckert, G., 382
Uthes, S., 308
Valdivia, R.O., 68, 318, 408, 472
Van Bussel, L.G.J., 70
Van de Fliert, E., 512
Van de Giessen, N., 72
Van de Steeg, J., 22, 148
Van de Ven, G.W.J., 394
Van Delden, H., 42, 546
Van den Akker, J., 412
Van den Brink, L., 394
Van der Heide, C.M., 116
Van der Heide, M., 176, 178
Van der Straeten, B., 160, 486
559
AgSAP Conference 2009, Egmond aan Zee, The Netherlands
Walter, T., 482
Warnecke, S., 312
Was, A., 484
Watson, C.A., 32
Watson, D., 34
Wauters, E., 44
Weindorf, D.C., 444
Weiss, F., 184
Welch, S.M., 242, 292
Wery, J., 6, 130, 180, 194, 196, 198, 200,
226, 228, 442, 520, 532
Wesström, I., 336
Westhoek, H., 122, 164
Whitbread, A., 450
Wieggers, H.J.J., 412
Wien, J.J.F., 6, 130, 316, 320, 330, 344
Wilczek, A., 242
Willemen, L., 124
Witters, N., 396
Wolf, J., 130
Woltjer, G.B., 162, 164, 166, 384
Wood, S., 22, 148, 220
Wriedt, G., 392
Wustenberghs, H., 44
Yang, X., 292, 338
Yeboah, E., 40
Young, M., 32
Zakayo, S., 76
Zander, P., 104, 256, 308, 382
Zasada, I., 78
Zeddies, J., 310
Zhen, L., 550
Zhang, W., 118
Zimmermann, A., 168
Zingore, S., 62
560