doi:10.1111/j.1420-9101.2009.01691.x Lacustrine radiations in African Synodontis catfish J. J. DAY,*, R. BILLSà & J. P. FRIEL§ *Department of Genetics, Evolution and Environment, University College London, London, UK Department of Zoology, The Natural History Museum, Cromwell Road, London, UK àSouth African Institute for Aquatic Biodiversity, Grahamstown, South Africa §Cornell University Museum of Vertebrates, Ithaca, NY, USA Keywords: Abstract brood parasitism; cichlid fishes; lacustrine diversification; Lake palaeo-Makgadikgadi; Lake Tanganyika; molecular dating. Synodontis catfish are a species-rich, tropical pan-African genus that predominately occur in fluviatile environments, but which also form a small radiation within Lake Tanganyika (LT). Here we estimate Synodontis relationships, based on mitochondrial and nuclear DNA, greatly expanding previous sampling. Data were analysed using different methods of phylogenetic inference: Bayesian (also testing compositional heterogeneity), likelihood and parsimony, in order to investigate biogeographic history and the extent of intralacustrine speciation within this group. Bayesian-relaxed clock analyses were used to estimate timings of radiations. Our analyses reveal a single origin of the LT flock with the inclusion of the nonendemic S. victoriae, and that these taxa evolved relatively recently (5.5 Ma), considerably later than the formation of LT (9–12 Ma). Two internal endemic clades diversified at a similar time (2–2.5 Ma), corresponding to a period of climate change, when lake levels dropped. We find evidence for a further species flock, composed of riverine southern African taxa, the diversification of which is very rapid, 0.8 Ma (95% HPD: 0.4–1.5) and infer a similar scenario for the diversification of this flock to southern African serrachromine cichlids in that they radiated in the now extinct lake Makgadikgadi. We also reveal that the biogeographic history of Synodontis catfish is more complex than previously thought, with nonmonophyletic geographic species groupings. Introduction Lakes, analogous to islands, are notable in presenting settings conducive for evolutionary radiations (Turner, 1999) and thus offer fruitful systems in which to study processes that generate biodiversity. In particular, the lakes of the East African rift have received considerable attention due to the diverse endemic faunal radiations, most pronounced in cichlid fishes (Fryer & Iles, 1972). Radiations of noncichlid faunas are reported from ancient lake settings (lakes > 100 000 years), with the highest levels of diversity and endemicity occurring in the oldest of the great lakes, Lake Tanganyika (LT, 9– Correspondence: Julia J. Day, Department of Genetics, Evolution and Environment, University College London, Gower Street, London WC1E 6BT, UK. Tel.: +44 2076792660; fax: +44 2076797096; e-mail: j.day@ucl.ac.uk 12 Myr, Cohen et al., 1993). Here, numerous unrelated species flocks occur including catfish, spiny-eels, gastropods, crabs and ostracods, for example (Coulter, 1991). Cichlids and gastropods aside, these radiations are on a much smaller scale (10–15 species) and at a lower taxonomic rank. Until recently (West & Michel, 2000; Wilson et al., 2004; Day & Wilkinson, 2006; Koblemüller et al., 2006; Marijnissen et al., 2006, 2008) the evolutionary significance of such faunas has been overlooked; yet, they provide important comparative systems that may help us to better understand processes responsible for clade diversification and generalities across diverse taxa. Synodontis catfish (Siluriformes: Mochokidae) are one of the most species-rich African fish genera (120 species, Poll, 1971). High species richness and their subSaharan distribution similar to that of cichlids make them a useful comparative system. However, relatively little is ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY 805 806 J. J. DAY ET AL. known about their evolutionary history, although two recent studies have begun to address this, focussing on their radiation within LT (Day & Wilkinson, 2006; Koblemüller et al., 2006). This species flock comprises 10 endemic, largely cryptic species (Wright & Page, 2006) as well as the nonendemic S. victoriae (Day & Wilkinson, 2006). Independent molecular studies of this radiation (Day & Wilkinson, 2006; Koblemüller et al., 2006) aimed to test the monophyly, and their colonization history of LT. These studies, including similar taxonomic sampling, but utilizing different mitochondrial (mt)DNA markers (entire cytochrome b, Day & Wilkinson, 2006; partial ND6 and partial control region, Koblemüller et al., 2006), found conflicting results regarding the evolutionary relationships of the LT flock, and because of differences in calibrations, subsequent interpretation of the colonization of LT. Although both studies support a nonmonophyletic endemic flock, a finding that mirrors LT cichlid fish evolution (Salzburger et al., 2002), this is a consequence of differences in putative relationships between nonendemic taxa (Fig. 1). Koblemüller et al. (2006) interpret the overall lack of phylogenetic signal between the six East African lineages as evidence of relatively rapid speciation. Conversely, regardless of phylogenetic analysis, Day & Wilkinson (2006) recover two wellresolved East African clades, one of which comprises all Tanganyika endemics and the inclusion of the nonendemic S. victoriae from Lake Victoria. These authors estimate a divergence time for the LT flock (including S. victoriae) to post-date the origin of LT, largely coinciding with the onset of full lacustrine conditions, which occurred approximately 5–6 Ma (Cohen et al., 1997), with individual endemic clades (Fig. 1a and b) diversifying relatively recently (< 3 Myr), as has also been (a) Outgroups (b) S. nigromaculata * reported in LT crabs (Marijnissen et al., 2006). This is a very different scenario from that of Koblemüller et al. (2006), whose estimated dates for these clades are considerably older at 12 and 6–7.8 Myr, respectively, with the implication that the former clade diversified in nonlacustrine conditions. As ultimately we may wish to incorporate these data into a broader picture investigating evolutionary processes within this environment, it is important to qualify these results and test alternate hypotheses concerning their evolution. In contrast to cichlid fishes, the majority of Synodontis diversity is within fluviatile habitats, reaching a maximum in central Africa, principally the Congo drainage basin (Poll, 1971). However, it has recently been shown that past environments have played an important role on current cichlid diversity. Genetic evidence reveals that current diversity of riverine cichlids in southern Africa is the product of adaptive radiation in a now extinct lake basin (Joyce et al., 2006). That Synodontis have radiated within LT suggests they may be more prone to lacustrine diversification. In the light of cichlid research, it is possible that this mechanism may also be invoked to account for the present genetic diversity of southern African species. It is unknown whether southern Africa taxa (nine described species, Skeleton, 2001) form a monophyletic grouping, or not, and although initial studies infer a close relationship between the southern species S. nigromaculata and S. zambezensis, and those from Lakes Tanganyika and Malawi (Day & Wilkinson, 2006; Koblemüller et al., 2006), this has not been thoroughly tested. To test this hypothesis and re-examine the origins of the LT species flock, we present a molecular phylogeny based on both mtDNA and nuclear (nc)DNA of Synodontis catfish, greatly expanding taxonomic sampling of previous studies. All geographic areas are included with multiple species, so that we are also able to test their monophyly and the importance of past vicariance events. Methods S. njassae Taxon selection and DNA methods S. granulosa i * * S. multipunctata * S. lucipinnis * * ii S. polli ** * S. irsacae * S. petricola S. victoriae Fig. 1 Consensus based on taxa common to the studies of (a) Day & Wilkinson (2006) and (b) Koblemüller et al. (2006). Support values (Bayesian posterior probabilities) indicated by *(> 95%), **(90–95%). Taxon names for LT endemic species are revised following Wright & Page (2006) and are highlighted in bold. A total of 65 Synodontis samples representing 40 species (33% of known species diversity) and two outgroup taxa (mochokids) are included in the phylogenetic analyses (Appendix 1). Taxa were sampled from the majority of major African drainage basins, with particular emphasis on East and southern species. All LT species are included with the exception of S. dhonti (only one specimen known, Wright & Page, 2006) and the nonendemic S. melanostrictus (Wright & Page, 2006). The southern African species, S. macrostoma, is omitted from this study due to the degraded nature of DNA from our only sample. Genomic DNA was extracted from white tissue and fin clips using DNA purification kit (Qiagen Ltd, UK). As mtDNA and ncDNA are inherited in different ways, we ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY Lacustrine radiations in African Synodontis catfish targeted genetic markers from each genome, in order to independently validate our phylogenetic hypothesis. Cytochrome b (mtDNA) provided good resolution in a previous study of LT Synodontis (Day & Wilkinson, 2006). Although ncDNA is more slowly evolving, the ribosomal protein-coding gene S7 (rpS7) has proved useful in studies of relatively young faunas at a comparable taxonomic level (e.g. Feulner et al., 2006; Schelly et al., 2006). Published primers were used to amplify cytochrome b and tRNA (Hardman & Page, 2003; Hardman, 2005) using varied primer annealing temperatures (49–55 C). rpS7 (Chow & Hazama, 1998) was amplified (826 bp in Synodontis), with a primer annealing temperature of 55 C, with only the first intron sequenced (530 bp). Samples were analysed using an ABI 3730xl sequencer (Applied Biosystems, UK). All sequences are deposited in GenBank. We further use published mtDNA sequence data (Day & Wilkinson, 2006; Koblemüller et al., 2006). See Appendices 1 and 2 for all GenBank accession numbers. 807 software (p4). Branch support was evaluated using Bayesian posterior probabilities (MR BA Y E S and p4) and ML nonparametric bootstrapping (1000 pseudoreplicates, G A R L I ). Further Bayesian (MR BA Y E S ) and ML analyses, were performed on single gene sets (65 samples mtDNA; 54 samples ncDNA) to determine taxon validity (mtDNA) and congruence of data sets. The same settings were employed as before and the models closest to the optimal models, GTR + I + G (mtDNA data) and HKY + G (ncDNA data), selected in M O D E L T E S T implemented. Finally, a subset of taxa (13 species, see Appendix 2) common to the data sets of Day & Wilkinson (2006) and Koblemüller et al. (2006), incorporating all genetic data: mtDNA (cytochrome b + tRNA, partial ND6, partial D-Loop) and ncDNA (rpS7 intron 1, this study), was subsequently analysed using ML and maximum parsimony (M P , P A U P * Swofford, 2002) to test alternative hypotheses regarding the evolution of the LT radiation (Fig. 1) and evaluate the strength of phylogenetic signal. Divergence time estimation Sequence alignment and phylogenetic analyses Presumed orthologous DNA sequences were aligned in CL U S T A L X (Thompson et al., 1997) using default parameters and checked manually. Data were analysed using maximum likelihood (ML), implemented in G A R L I (Zwickl, 2006) and Bayesian Markov Chain Monte Carlo (MCMC) analyses were conducted using the programmes MR BA Y E S v3.1.2 (Huelsenbeck & Ronquist, 2001) and p4 v0.84. r158 (Foster, 2004). The latter programme has the advantage of assessing compositional heterogeneity across data sets (Foster, 2004), as well as employing the polytomy proposal (Lewis et al., 2005). Optimal model and parameter values were obtained using M O D E L T E S T v3.7 (Posada & Crandall, 1998). All methods were performed for the combined data set (54 taxa) as follows: ML analyses were performed both by estimating parameters and by specifying the optimal model and parameter values (TVM + I + G) based on the AIC criteria. The model GTR + I + G was implemented for both Bayesian programmes as the model closest to that selected using M O D E L T E S T . MR BA Y E S analyses were run for 2 000 000 generations (four chains, 0.2 temperature) partitioning these data based on individual genes as well as an additional analysis also enabling the third codon a partition-specific model. Optimal composition analyses were performed using p4 and run for 1 500 000 generations (four chains, 0.05 temperature), with inclusion of a strong prior for polytomies (i.e. C = log(10)). Composition chi-squared test using predictive posterior simulations conducted in p4, revealed homogeneity of the ncDNA data, but significant heterogeneity of the mtDNA data. For the mtDNA data to be regarded as homogenous (P = 0.99), it was necessary to apply two composition vectors. Convergence of the MCMC to a stationary phase was determined using TR A C E R (MR BA Y E S ) and G N U Timing of speciation events were estimated using Bayesian-relaxed clock methods as the assumption of a molecular clock was rejected after comparison of likelihood scores for clock and nonclock ML trees using a likelihood ratio test (LR = 132.30, d.f. = 54, P > 0.05). We used the programme B E A S T v1.4.6 (Drummond & Rambaut, 2007), selecting an uncorrelated log-normalrelaxed clock rate variation model and a Yule speciation tree prior (Drummond et al., 2006), which assumes a constant rate of speciation per lineage. We selected a similar model to that implemented in MR BA Y E S for the combined data set (GTR + I + G), partitioning loci and codons (first + second vs. third). Different tree priors (exponential and uniform) were implemented to determine if these would greatly affect subsequent results of divergence dates. For each tree prior, two independent MCMC analyses were run, starting from randomly chosen trees, for 1 000 000 generations, to ensure individual runs had converged and determine burn-in (25%). We report results implementing a uniform prior, as analyses were largely congruent. We calibrate the ingroup using the oldest occurrence of Synodontis. Fossils assigned to this genus date from the Early Miocene (Burdigalian) of Egypt (Priem, 1920) and Kenya (Greenwood, 1951), although the Egyptian site (Wadi Moghara) is marginally older, dated at c.18–17 Myr (Miller, 1999). Although it is considered better practice to use more than one calibration (Benton & Donghue, 2007), fossil Synodontis cannot reliably be assigned to extant species (Pinton et al., 2006), despite them being not uncommon in the fossil record (Stewart, 2001). Furthermore, use of geographical dates, such as the maximum ages of the great lakes, is not an independent means of calibration (McCune, 1997), particularly as we wish to estimate age of colonization of lacustrine ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY 808 J. J. DAY ET AL. environments. To make allowances for the imprecision of fossil dating, the ingroup (i.e. Synodontis) is calibrated using the maximum and minimum dates for the Burdigalian period (16.3–21.5 Myr). Results Phylogenetic analysis The combined data matrix consists of 1697 sites. A total of 1660 bp (cytochrome b, 1058 bp; tRNA 65 bp; rpS7 intron 1, 538 bp) were sequenced for all species, whereas only cytochrome b + tRNA was sequenced for multiple individuals in order to test species validity. The alignment of rpS7 intron 1 included up to 37 indels (up to 22 indels discounting S. gambiensis). Data were missing for rpS7 intron 1 for S. katangae and S. nigromaculata (CU 91094). Cytochrome b, contains the most parsimony informative (PI) sites (175), whereas rpS7 intron 1 (538 bp) has 20 ⁄ 32 PI sites (gaps coded as missing data ⁄ 5th state). The matrix for a subset of taxa (13 species, Appendix 2) common to the data sets of Day & Wilkinson (2006) and Koblemüller et al. (2006) consists of 2530 bp (cytochrome b + tRNA, rpS7 intron 1, partial ND6 and partial control region). PI sites for ND6 (426 bp) include 46 sites, with the control region (407 bp) containing 128 ⁄ 387 sites (gaps coded as missing data ⁄ 5th state). Trees obtained under BI (MR BA Y E S ) and ML methods are highly congruent (Fig. 2a), with minimal differences in relationships. Both methods consistently recover two main clades (A and B, BPP > 75 to < 85; BS 55) and several internal clades (C–K), which are well supported, with the exception of clade H. The BI (p4) tree obtained under a compositional heterogeneity model (Fig. 2b) differs in that clade B is not recovered. Instead, clades D and C are recovered as successive sister groups to clade A, with clade E, plus (S. contractus, S. nigriventris), as sister to these, although support for this set of relationships is not strong. Evaluation of alternative phylogenetic trees using the approximately unbiased (AU) test (Shimodaira, 2002), implemented in the programme C O N S E L (Shimodaira & Hasegawa, 2001), found no tree as significantly worse fit to the data (P = 0.51–0.49). BI (MR BA Y E S ) trees based on different data sets (single gene: cytochrome b, rpS7, combined) were also evaluated using the AU test, with the tree constructed from cytochrome b data rejected as significantly worse fit (P > 0.05). There are several differences in taxon ⁄ clade placements between this tree and those generated from the rpS7 and combined data set. These include the placement of S. gambiensis and S. clarias, recovered as basal members of clade A and B, respectively, and the alternate placements of clade C and the clade (S. contractus, S. nigriventris) within clade B. The instability of S. gambiensis and S. clarias may be caused by the long branches of these particular taxa. Clade A, is composed of two subclades (H and I), the former and its constituent clades (J and K) are well supported, as largely are the internal nodes. Support for the latter (H) is less convincing, although this can be attributed to the placement of S. gambiensis and S. clarias. There is strong internal support within clade H for the placement of the West African S. thysi, with East African ⁄ Nilotic members (S. frontosus, S. serratus) and a species (S. af. punctulatus) assumed to have been collected from Tanzania. The LT endemics, plus the nonendemic S. victoriae are recovered as a single clade (K, Fig. 2). Clade J, which includes the Lake Malawi endemic (S. njassae), is otherwise composed of riverine species from the Zambezi and Okavango drainage basins (including the widespread S. nigromaculata species complex). Inclusion of the majority of all described southern African species reveals a polyphyletic assemblage (Fig. 2), the placements of which occur in both of the main clades (A and B). However, the majority of species form a monophyletic grouping (F), with maximum support (100 BS ⁄ 100 BPP) under all analyses and are most closely related to the Congolese taxa, within clade B. The remaining southern African taxa (e.g. S. nigromaculata species complex and S. zambezensis) occur within clade A as previously discussed. Discussion Increased taxonomic sampling (50% more than previous studies) and additional data, including the first inclusion of ncDNA, provide an improved estimate of phylogenetic relationships within Synodontis. Although sampling is far from complete, the results of this study are decisive regarding alternate evolutionary hypotheses of the LT radiation. Furthermore, this study identifies a novel southern African radiation, which irrespective of actual dates, reveal a more rapid diversification of this grouping than the LT radiation. More broadly, the phylogeny highlights the nonmonophyletic nature of geographic regions, and thus a considerably more complex history of Synodontis evolution. Lake Tanganyika radiation revisited Irrespective of analyses of the combined data set, the LT endemics, plus the nonendemic S. victoriae are recovered as a single clade (K, Fig. 2) supporting the hypothesis of Day & Wilkinson (2006) over that of Koblemüller et al. (2006). One notable difference in our analyses, although weakly supported, is that the tree recovered using compositional heterogeneity places S. victoriae as basal to all LT endemics (Fig. 2b), rather than internally within the LT endemic radiation (Fig. 2a). Support for a monophyletic LT endemic clade, plus S. victoriae is largely due to the stronger signal of the cytochrome b data, as separate gene analysis reveals inconclusive evidence from ncDNA data to support this grouping. Although ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY Lacustrine radiations in African Synodontis catfish (a) 100 1.00 100 1.00 94 0.51 98 1.00 100 F 1.00 1.00 53 0.67 64 E 0.55 55 B 0.83 G 1.00 100 98 1.00 C 1.00 0.54 100 D 0.91 100 1.00 91 1.00 82 0.84 100 1.00 56 H 0.88 0.58 86 100 0.99 1.00 84 * * * 98 0.98 S. soloni S. angilica S. congica S. pleurops ** S. greshoffi S. unicolor S. af.afrofisheri RM S. afrofischeri LV S. afrofischeri LA S. nigrita S. rebeli S. waterloti 1 S. waterloti 2 S. frontosa 1 S. frontosa 2 S. frontosa 3 100 1.00 55 A 57 0.63 0.76 73 0.99 1.00 J 100 1.00 I 1.00 K ii 71 0.87 1.00 100 1.00 91 1.00 83 0.99 100 1.00 100 i 88 100 87 0.77 71 0.66 84 0.96 * * S. zambezensis S. sp. nov. S. njassae S. nigromaculata 1 S. nigromaculata 2 S. nigromaculata 3 S. nigromaculata 4 S. granulosa S. multipunctata S. grandiops S. lucipinnis S. af.tanganaicae S. irsacae S. polli S. af.ilebrevis S. petricola S. victoriae * * * * * ** * * * * * S. af.punctulata 1.00 79 0.96 (b) Microsynodontis sp. S. contracta S. nigriventris S. decora S. brichardi S. katangae 1 S. katangae 2 S. woosnami S. macrostiga S. leopardina S. nebulosa S. af.thamalakanensis S. vanderwaali S. clarias S. gambiensis S. thysi S. serrata 1 S. serrata 2 809 * * * * ** * ** ** * * * ** ** 0.01 changes Fig. 2 Phylogenetic hypotheses of Synodontis relationships inferred from mtDNA (cytochrome b + tRNA) and ncDNA (rpS7 intron 1). (a) Consensus of maximum likelihood tree and Bayesian (MR BA Y E S ) consensus. Support values above 50% are shown for bootstrap (above branches) and Bayesian posterior probabilities (below branches). (b) Bayesian consensus based on P4 analysis employing compositional heterogeneity model and a strong polytomy prior. Support values > 81 indicated by **(81–90), *(91–100). Southern African species are highlighted by grey taxon names. ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY 810 J. J. DAY ET AL. rpS7 intron 1 and 2 have been shown to be useful for resolving species-level relationships (e.g. Schelly et al., 2006), it is probable that this marker is too slowly evolving and therefore less informative in resolving relationships for such a young radiation (see below). Analysis of a subset of taxa (13 species, Appendix 2) common to the data sets of Day & Wilkinson (2006) and Koblemüller et al. (2006), incorporating all genetic data (2530 bp), again strongly supports the inclusion of S. victoriae within the LT endemic clade (as depicted in Fig. 1a), with the clade (S. njassae, S. nigromaculata) as sister. Separate gene analysis of cytochrome b + tRNA and also ncDNA support this grouping. By contrast, analysis of partial ND6 does not support this grouping, although support for alternate relationships is weak. Analysis of partial control region data finds no resolution between this set of taxa. Divergence dates in this study are similar to Day & Wilkinson (2006), despite the implementation of different dating methods. The endemic LT flock, plus S. victoriae is estimated to have diverged at 5.5 Myr (95% HPD, highest posterior distribution: 4.0–7.3), whereas LT endemic clades i and ii are dated at 3.0 Myr (95% HPD: 1.9–4.4) and 2.5 Myr (95% HPD: 1.6–3.5) respectively (Fig. 3). In contrast to Koblemüller et al.’s (2006) estimates for these endemic clades (11.2–14.8 and 5–8 Myr respectively), the dates generated here are relatively young, being similar to those estimated for the platythelphusid crab radiation (Marijnissen et al., 2006) and similarly implying rapid diversification of endemic species in situ. LT has had a complex geological and climatic history, with researchers linking these episodes to speciation events (e.g. Sturmbauer et al., 2001; Day et al., 2008). As previously proposed (Day & Wilkinson, 2006), the inferred timing of the Synodontis radiation corresponds to the deepening of LT at around 5–6 Myr (Cohen et al., 1997), with the two endemic clades diversifying during a time when LT experienced lower lake levels through climate change (Cane & Molnar, 2001). Recently, two timescales have been suggested for cichlid evolution based on fossil and Gondwanan break-up calibrations (Genner et al., 2007), leading to disparate age estimates for the divergence of the LT cichlid flock at 12 and 18 Myr, respectively. It is therefore of interest that the evolution of the LT Synodontis catfish flock is comparable with the younger timescale, with divergences of all the major cichlid tribes estimated to have occurred between 6.6 and 3.1 Ma (Day et al., 2008). It would appear that during this time frame, when a fully formed LT was subjected to lake-level fluctuations, a number of groups including catfish and crabs were diversifying, possibly in response to environmental factors (Day & Wilkinson, 2006; Marijnissen et al., 2006). In the context of adaptive radiation theory (Schluter, 2000), it is noteworthy that although LT Synodontis exhibit recent divergence, this clade displays little morphological disparity. In contrast to the phenotypically diverse cichlid fish (e.g. lamprologine cichlids, Day et al., 2007) and Platythelphusa crab radiations (Marijnissen et al., 2006), they are instead largely cryptic, and display no sexual dimorphism. Ecological niche segregation has been demonstrated to play an important role in the divergence of crabs (Marijnissen et al., 2008) and cichlids (e.g. Hori, 1991), but has not yet been investigated in Synodontis. Although the placement of S. victoriae is less robust (also see Day & Wilkinson, 2006), it appears reasonable, judging on additional data, that this taxon nests within the LT radiation. Considering that this species principally occurs in Lake Victoria, which is considerably younger than the age of this species, dated at 4.4 Myr (95% HPD: 3.0–6.0), it is plausible to infer that S. victoriae evolved within the LT basin and has subsequently emigrated. The alternate hypothesis that the two endemic clades are independent radiations would imply that the ancestor of these clades made it to LT during a dry period in this region’s history (Cane & Molnar, 2001), when presumably aquatic environments would have shrunk, making mobility for aquatic taxa less probable. It is likely that the nonendemic LT taxa, S. melanosticta, is closely related to S. nigromaculata being closely allied with this species on morphological grounds (Wright & Page, 2006), and therefore it is reasonable to infer that it invaded the lake independently to the LT endemics. Differences in dating between studies appear to be calibration specific, as opposed to being methodological. In particular, rather than dating the ingroup using the oldest fossil, Koblemüller et al. (2006) calibrate their East African clade using the oldest fossil from that region (Kenya). Although this ignores a similarly aged fossil from a non-East African locality (Priem, 1920), it also assumes a monophyletic East African clade. However, on the basis of their limited sampling this is not an assumption that is supported by the current analysis. Moreover, dating of lacustrine endemics using maximum lake ages by these authors does not enable independent means of estimating their divergence, although these internal calibrations did not appear to impact on estimates, with similar dates obtained when these calibrations are excluded. Koblemüller et al. (2006) attribute lack of phylogenetic signal between ‘East African’ lineages (clade I) to rapid speciation and compare this clade with other examples of this phenomenon (e.g. Joyce et al., 2005). However, artefactual causes, indicative of a soft polytomy, should first be eliminated (Maddison, 1989) before concluding this scenario. One of the principal causes for the lack of resolution is insufficient data (characters and taxon sampling), although other factors, such as data evolving appropriately for the question, can also be problematic (Poe & Chubb, 2004). Our results reveal that there is little evidence for a ‘rapid’ speciation event and that the lack of phylogenetic signal reported by Koblemüller et al. (2006) is in this case a result of lack of data. The lack of resolution ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY Lacustrine radiations in African Synodontis catfish 811 Microsynodontis sp. S. rebeli S. waterloti S. nigrita S. soloni S. congica S. angilica S. pleurops S. af.thamalakalensis S. nebulosa S. vanderwaali S. leopardina S. woosnami S. macrostiga S. decora S. brichardi S. katangae S. contracta S. nigriventris S. af.afrofischeri S. afrofischeri S. unicolor S. greshoffi S. thysi S. serrata S. af.punctata S. frontosa S. clarias S. gambiensis S. nigromaculata 1 S. nigromaculata 2 S. nigromaculata 3 S. sp. nov. S. njassae S. zambezensis S. grandiops S. multipunctata S. granulosa S. af.ilebrevis S. petricola S. polli S. irsacae S. lucipinnis S. af.tanganaicae S. victoriae Fig. 3 Chronogram inferred from Bayesian dating analysis (B E A S T ) of the combined data set. Node bars indicate 95% posterior distributions. Grey boxes indicate lacustrine radiations. Light grey: East African taxa; Mid-grey: southern African taxa; Dark grey: central and West African taxa. Time in Myr. from the control region regarding the deeper splits between the main lineages within clade F is unsurprising, considering the age of this clade, 8.0 Myr (95% HPD: 6.0– 10.4) as the mtDNA control region is rapidly evolving. Brood parasitism Among LT species, S. multipunctata has been shown to be a brood parasite of mouthbrooding cichlids from the tribe ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY 812 J. J. DAY ET AL. Tropheini (Sato, 1986), the first time this behaviour has been recorded in a nonavian vertebrate group. The catfish eggs incubated in the cichlid mouths along with those of the hosts, hatch earlier, and subsequently feed on their host’s eggs. The extent of brood parasitism in LT species is unknown, however, although not substantiated in the scientific literature or reported in the wild, S. petricola and its ‘dwarf form’ (S. lucipinnis, Wright & Page, 2006) also exhibit this behaviour in aquarium conditions (aquarium web blogs). We might also presume that the newly described S. grandiops (Wright & Page, 2006) is also a brood parasite. This species (if indeed distinct) is very similar to S. multipunctata in terms of morphology and genetics, with specimens previously catalogued together under the name, S. multipunctata. Although speculative, it appears that brood parasitism could have evolved independently in the two endemic LT lineages. Notably, brood parasitism has a dual origin in old world cuckoos (Aragón et al., 1999). Koblemüller et al. (2006) suggest that S. multipunctata diverged at around the same time as their hosts. However, their estimated date of 2.37 Ma is erroneous, being the time of origin between the particular population(s) of this species that they sampled. The correct age of origin of this species (judging from their timescale) is the time it diverged from its sister species, S. granulosa, which in their analysis is 12.0 Ma. However, our date estimates for both endemic clades (see previous section) do correspond to newly estimated dates for the Tropheini 3.4 Ma (95% HPD: 2.7–4.5, Day et al., 2008) assuming the younger calibration for the LT cichlid radiation and therefore Koblemüller et al.’s hypothesis is thus supported by our analysis. Evidence for a southern African species flock Despite the nonmonophyly of the southern African species (Fig. 1), the majority form a species flock (clade F), which has evolved from Congolese ancestors. The recovery of clade F agrees with the preliminary analysis of partial cytochrome b and allozyme data (Bruwer et al., 2000), although the monophyly (with the inclusion of S. macrostoma) was evaluated against only several species, which include S. nigromaculata, S. njassae, S. petricola and S. zambezensis in their study. The most notable observation of clade F is that the internal branches are exceptionally short (Figs 2b and 3), suggesting rapid cladogenesis. The age of this clade is estimated at 0.89 Myr (95% HPD: 0.42–1.4), and irrespective of calibration (Loader et al., 2007), it is considerably younger than the LT radiation. Identification of many of these species is notoriously difficult (J.J. Day & R. Bills, personal observation) despite field keys (e.g. Skeleton, 2001). Thus, the young age of these species might explain why morphological characters are not yet totally fixed and seemingly labile. As Synodontis have been shown to hybridize in laboratory conditions (S. zambez- ensis · S. nigromaculata, Bruwer & van der Bank, 2002), it is therefore possible that there is ongoing hybridization, particularly as this is a very young species flock (Seehausen, 2004). It has been proposed that extant southern African riverine serranochromine cichlids are the result of an adaptive radiation in a now extinct palaeo-lake, Makgadikgadi (Joyce et al., 2005). This lake, one of Africa’s largest water bodies (120 000 km2), began to form in the Pleistocene (315 000–460 000 years), drying up in the last few thousands years, and would have been located in Okavango delta region (Moore & Larkin, 2001). Irrespective of calibrations used, the cichlid radiation falls within this time frame (Genner et al., 2007), and although the estimated age for the Synodontis species flock is older, it also falls within the bounds of the upper confidence limit. As such, it possible that the flock arose within lacustrine conditions, or at least in conditions associated with the evolution of this hydrological feature. Members of the flock subsequently spreading through southern African rivers, in a similar scenario as hypothesized for the cichlids. Unlike these cichlids, members of the Synodontis flock have not spread out across Africa to the same degree, with the majority of species occurring sympatrically in the Upper Zambezi, Okavango, Kafue and Cunene river systems, whereas S. nebulosus occurs in the Middle and Lower Zambezi and Buzi systems (Skeleton, 2001). Although recent evidence from African weakly electric fish (mormyrids) reveals that adaptive radiation may proceed in rivers (Feulner et al., 2006), it seems less likely that the Synodontis species flock could have arisen in fluviatile conditions considering that we find no evidence for such rapid diversification in other river systems (e.g. Congo or Nile) for this group. In the light of a probable flock that evolved in a now extinct lake basin, it is of interest that there are no Synodontis species flocks in either Lakes Malawi or Victoria. The environment of Lake Victoria is similar to Palaeolake Makgadikgadi, being a large shallow young lake, estimated to have formed approximately 400 000 years ago (Johnson et al., 1996), with a probable complete desiccation period in the late Pleistocene and subsequent refilling 12 400 years ago. It contains two unrelated nonendemic species (S. afrofischeri and S. victoriae). Conversely, Lake Malawi, dated at 8.6 Myr, with deep water conditions attained 4.5 Ma (Ebinger et al., 1993; Delvaux, 1995) is a rift environment similar to that of LT. Cichlids and gastropods aside, it is therefore somewhat surprising that LM lacks the multiple species flocks associated with LT, with a single described endemic species (S. njassae), possibly two (Snoeks, 2004). This could be as a consequence of environmental factors, in that LM experienced periodical desiccations estimated to have occurred as recently as 570 000 years ago (Delvaux, 1995). However, if Synodontis can radiate within this time frame, then their absence in LM may be due to lack of available niche space, possibly because of cichlid fishes. It ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY Lacustrine radiations in African Synodontis catfish is notable that although all three great lakes have sizable cichlid radiations, LT is by far the smaller flock with less than half the species estimated for LM and LV, yet it contains greater overall biodiversity. Synodontis biogeography Day & Wilkinson (2006) found good support for the existence of an eastern ⁄ central African split in Synodontis faunas, the timing of which corresponds to East African rifting events. This has also been observed in other, albeit terrestrial, taxa (e.g. chameleons, Matthee et al., 2004). The phylogenetic hypothesis presented here also recovers two main clades (A and B), and although the reality of clade B is questionable, the divergence of these clades are within a similar time frame: clade A 15.3 Ma (95% HPD: 12.5–18.7) and clade B 16.3 Ma (95% HPD: 13.5–19.7). However, increased taxonomic sampling to previous studies (Day & Wilkinson, 2006; Koblemüller et al., 2006) highlights a complex biogeographic pattern of this catfish clade, with no geographic region strictly monophyletic. African cichlid fishes also show a complicated biogeographic history (Genner et al., 2007). Our data reveal that although clade A contains a greater proportion of East African taxa, it is by no means exclusively so, containing representatives from West, and central, and southern river systems (Fig. 3). With the exception of the ‘southern African’ flock and the East African S. afrofischeri, clade B is composed of West and central African taxa. Notably divergence estimates for these taxa are relatively young, suggesting that colonization into different geographic regions ⁄ river systems has occurred relatively recently. Thus, although rifting events may have been responsible for the initial divergence within Synodontis, dispersal has been dominant in their subsequent evolutionary history. Our phylogenetic hypothesis reveals that Synodontis have colonized southern Africa from two different lineages and that judging from the chronogram (Fig. 3) this has occurred more recently in their history. Divergence of the ‘southern African’ flock from its most recent common ancestor (MRCA) occurred 7.7 Ma (95% HPD: 5.8–10.3), whereas S. nigromaculata and S. zambezensis diverged from their MRCA 6.5 Ma (95% HPD: 4.7–8.5). The timings of these two invasions, particularly taking into account 95% confidence intervals, lend support that colonization occurred at a similar time. The MRCA of the former clade are Congolese taxa, and it is known that sections of the river system south of the Congo basin (e.g. Zambezi and Okavango networks) have belonged to the palaeo-congo network (Stankiewicz & de Wit, 2006). Colonization from East to southern Africa is more complicated, although the occurrence of the S. nigromaculata species complex in southern Africa as well as LT drainage, suggests fish-accessible waterways throughout this region (Koblemüller et al., 2006). Clearly, denser sampling of Synodontis along with quantitative biogeo- 813 graphic analyses will shed more light upon their phylogeographic history. Acknowledgments JJD is supported by a Dorothy Hodgkin Royal Society Fellowship and this study was financed by NERC grant NE ⁄ F000782 ⁄ 1. We thank C. Cox (NHM) for help and implementation of P4 software. JJD thanks the Fisheries Resources Research Institute (FIRI), Jinja, Uganda, in particular J. Balirwa and G. Namulemo and M. Wilkinson (NHM) for field assistance. I. Payne and R. Wakeford (MRAG, Ltd) are thanked for tissue samples. References Aragón, S., Møller, A.P., Soler, J.J. & Soler, M. 1999. Molecular phylogeny of cuckoos supports a polyphyletic origin of brood parasitism. J. Evol. Biol. 12: 495–506. Benton, M.J. & Donghue, P.C.J. 2007. Paleontological evidence to date the tree of life. Mol. Biol. Evol. 24: 26–53. Bruwer, E.E. & van der Bank, F.H. 2002. Biochemical genetic markers (hybrizymes) to identify Synodontis zambezensis · S. nigromaculatus hybrids (Teleostei: Mochokidae). Biochem. Syst. Ecol. 30: 641–649. Bruwer, E.E., van der Bank, F.H., Steyn, G.J. & Wink, M. 2000. Phylogenetic relationships between Synodontis species: some preliminary results. Afr. J. Aquat. Sci. 25: 89–92. Cane, M.A. & Molnar, P. 2001. Closing of the Indonesian seaway as a precursor to east African aridification around 3–4 million years ago. Nature 411: 157–162. Chow, S. & Hazama, K. 1998. Universal PCR primers for S7 ribosomal protein gene introns in fish. Mol. Ecol. 7: 1255–1256. Cohen, A.S., Soreghan, M.J. & Scholz, C.A. 1993. Estimating the age of formation of lakes – an example from Lake Tanganyika, East African rift System. Geology 21: 511–514. Cohen, A.S., Lezzar, K.-E., Tiercelin, J.J. & Sorgehan, M. 1997. New palaeogeographic and lake-level reconstructions of Lake Tanganyika: implications for tectonic, climatic and biological evolution in a rift lake. Basin Res. 9: 107–132. Coulter, G.W. 1991. Lake Tanganyika and its Life. Oxford University Press, Oxford. Day, J.J. & Wilkinson, M. 2006. On the origin of the Synodontis catfish species flock from Lake Tanganyika. Biol. Lett. 2: 548– 552. Day, J.J., Santini, S. & Garcia-Moreno, J. 2007. Phylogenetic relationships of the Lake Tanganyika cichlid tribe Lamprologini: the story from mitochondrial DNA. Mol. Phylogenet. Evol. 45: 629–642. Day, J.J., Cotton, J.A. & Barraclough, T.G. 2008. Tempo and mode of diversification of Lake Tanganyika cichlid fishes. PLoS ONE 3: e1730. Delvaux, D. 1995. Age of Lake Malawi (Nyasa) and water level fluctuations. Musée royal de l’Afrique Centrale (Tervuren), Départment de Géologie et Minéralogie. Rapport Annuel 1993– 1994: 99–108. Drummond, A.J. & Rambaut, A. 2007. B E A S T : Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7: 214. Drummond, A.J., Ho, S.Y.W., Phillips, M.J. & Rambaut, A. 2006. Relaxed phylogenetics and dating with confidence. PLoS Biol. 4: e88. ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY 814 J. J. DAY ET AL. Ebinger, C.J., Deino, A.L., Tesha, A.L., Becker, T. & Ring, U. 1993. Tectonic controls on rift basin morphology – evolution of the Northern Malawi (Nyasa) Rift. J. Geophys. Res. 98: 17821–17836. Ferraris, C.J. 2007. A checklist of catfishes, recent and fossil (Osteichthyes: Siluriformes) and catalogue of siluriform primary types. Zootaxa 1418: 1–628. Feulner, P.G.D., Kirschbaum, F., Mamonekene, V., Ketmaier, V. & Tiedemann, R. 2006. Adaptive radiation in African weakly electric fish (Teleostei: Mormyridae: Campylomormyrus): a combined molecular and morphological approach. J. Evol. Biol. 20: 403–414. Foster, P.G. 2004. Modeling compositional heterogeneity. Syst. Biol. 53: 485–495. Fryer, G. & Iles, T.D. 1972. The Cichlid Fishes of the Great Lakes of Africa: Their Biology and Evolution. Olive & Boyd, Edinburgh. Genner, M.J., Seehausen, O., Lunt, D.H., Joyce, D.A., Shaw, P.W., Carvalho, G.R. & Turner, G.F. 2007. Age of cichlids: new dates for ancient lake fish radiations. Mol. Biol. Evol. 24: 1269– 1282. Greenwood, P.H. 1951. Fish remains from Miocene deposits of Rusinga Island and Kavirondo Province, Kenya. Ann. Mag. Nat. Hist., 12: 1192–1201. Hardman, M. 2005. The phylogenetic relationships among nondiplomystid catfishes as inferred from mitochondrial cytochrome b sequences; the search for the ictalurid sister taxon (otophysi: Siluriformes). Mol. Phylogenet. Evol. 37: 700–720. Hardman, M. & Page, L.M. 2003. Phylogenetic relationships among bullhead catfishes of the genus Ameiurus (Siluriformes: Ictaluridae). Copeia 2003: 20–33. Hori, M. 1991. Feeding relationships among cichlid fishes in Lake Tanganyika: effects of intra- and interspecific variations of feeding behavior on their coexistence. Ecol. Int. Bull. 19: 89– 101. Huelsenbeck, J.P. & Ronquist, F. 2001. MR BA Y E S : Bayesian inference of phylogeny. Bioinformatics 17: 754–755. Johnson, T.C., Scholz, C.A., Talbot, M.R., Kelts, K., Ricketts, R.D., Ngobi, G., Beuning, K., Ssemmanda, I. & McGill, J.W. 1996. Late Pleistocene desiccation of Lake Victoria and rapid evolution of cichlid fishes. Science 273: 1091–1093. Joyce, D.A., Lunt, D.H., Bills, R., Turner, G.F., Katongo, C., Duftner, N., Sturmbauer, C. & Seehausen, O. 2005. An extant cichlid radiation emerged in an extinct Pleistocene lake. Nature 435: 90–95. Koblemüller, K., Sturmbauer, C., Verheyen, E., Meyer, A. & Salzburger, W. 2006. Mitochondrial phylogeny and phylogeography of East African squeaker catfish (Siluriformes: Synodontis). BMC Evol. Biol. 6: 49. Lewis, P.O., Holder, M.T. & Holsinger, K.E. 2005. Polytomies and Bayesian phylogenetic inference. Syst. Biol. 54: 241–253. Loader, S.P., Pisani, D., Cotton, J.A., Gower, D.J., Day, J.J. & Wilkinson, M. 2007. Relative timescales reveal multiple origins of parallel disjunct distributions of African caecilian amphibians. Biol. Lett., 3: 505–508. Maddison, W.P. 1989. Reconstructing character evolution on polytomous cladograms. Cladistics 5: 365–377. Marijnissen, S.A.E., Michel, E., Daniels, S.R., Erpenbeck, D., Menken, S.B.J. & Schram, F.R. 2006. Molecular evidence for recent divergence of Lake Tanganyika endemic crabs (Decapoda: Plathelphusidae). Mol. Phylogenet. Evol. 40: 628–634. Marijnissen, S.A.E., Michel, E., Kamermans, M., Olaya-Bosch, K., Kars, M., Cleary, D.F.R., van Loon, E.E., Rachello Dolmen, P.G. & Menken, S.B.J. 2008. Ecological correlates of species differences in the Lake Tanganyika crab radiation. Hydrobiologia 615: 1573–5117. Matthee, C.A., Tilbury, C.R. & Townsend, T. 2004. A phylogenetic review of African leaf chameleons: genus Rhampholeon (Chamaeleonidae): the role of vicariance and climate change in speciation. Proc. R. Soc. Lond. B 271: 1967–1975. McCune, A.R. 1997. How fast is speciation: molecular, geological and phylogenetic evidence from adaptive radiations of fishes. In: Molecular Evolution and Adaptive Radiation (T. Givnish & K. Sytsma, eds), pp. 585–610. Cambridge University Press, Cambridge. Miller, R. 1999. Faunal correlation of Wadi Moghara, Egypt: implications for the age of Prohylobates tandyi. J. Hum. Evol. 36: 519–533. Moore, A.E. & Larkin, P. 2001. Drainage evolution in southcentral Africa since the breakup of Gondwana. S. Afr. J. Geol. 104: 47–68. Pinton, A., Fara, E. & Otero, O. 2006. Spine anatomy reveals the diversity of catfish through time: a case study of Synodontis (Siluriformes). Naturwissenschaften 93: 22–26. Poe, S. & Chubb, A.L. 2004. Birds in a bush: five genes indicate explosive evolution of avian orders. Evolution 58: 404–415. Poll, M. 1971. Révision des Synodontis africains (famille Mochokidae). Mus. R. Afr. Centr. Tervuren (Belg) Zool. 191: 1–497. Posada, D. & Crandall, K.A. 1998. M O D E L T E S T : testing the model of DNA substitution. Bioinformatics 14: 817–818. Priem, R. 1920. Poissons fossiles du Miocène d’Egypte (Burdigalien de Moghara, ‘‘Désert lybique’’). In: Contribution à l’étude des vertébrés miocènes de l’Egypte (R. Fourtau, ed), pp. 8–15. Government Press, Cairo. Salzburger, W., Meyer, A., Baric, S., Verheyen, E. & Strumbauer, C. 2002. Phylogeny of the Lake Tanganyika cichlid species flock and its relationships to the Central and East African Haplochromine cichlid fish faunas. Syst. Biol. 1: 1–23. Sato, T. 1986. A brood parasitic catfish of mouthbrooding cichlids in Lake Tanganyika. Nature 323: 58–59. Schelly, R., Salzburger, W., Koblmüller, S., Duftner, N. & Sturmbauer, C. 2006. Phylogenetic relationships of the lamprologine cichlid genus Lepidolamprologus (Teleostei: Perciformes) based on mitochrondrial and nuclear sequences, suggesting introgressive hybridization. Mol. Phylogenet. Evol. 38: 426–438. Schluter, D. 2000. The Ecology of Adaptive Radiation. Oxford University Press, Oxford. Seehausen, O. 2004. Hybridization and adaptive radiation. TREE 19: 198–207. Shimodaira, H. 2002. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51: 492–508. Shimodaira, H. & Hasegawa, M. 2001. C O N S E L : for assessing the confidence of phylogenetic tree selection. Bioinformatics, 17: 1246–1247. Skeleton, P. 2001. Freshwater Fishes of Southern Africa. Struik Publishers, South Africa. Snoeks, J. 2004. The non-cichlid fishes of the Lake Malawi system: a compilation. In: The Cichlid Diversity of Lake Malawi ⁄ Nyasa ⁄ Niassa: Identification, Distribution and Taxonomy (J. Snoeks, ed.), pp. 20–26. Cichlid Press, El Paso, TX. Stankiewicz, J. & de Wit, M.J. 2006. A proposed drainage evolution model for Central Africa – did the Congo flow east? J. Afr. Earth Sci. 44: 75–84. ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY Lacustrine radiations in African Synodontis catfish 815 West, K. & Michel, E. 2000. The dynamics of endemic diversification: molecular phylogeny suggests an explosive origin of the thiarid gastropods of Lake Tanganyika. In: Ancient Lakes: Biodiversity, Ecology and Evolution (A. Rossiter & H. Kawanabe, eds), pp. 331–354. Adv. Ecol. Res., 31, Academic Press, London. Wilson, A.B., Glaubrecht, M. & Meyer, A. 2004. Ancient lakes as evolutionary reservoirs: evidence from the thalassoid gastropods of Lake Tanganyika. Proc. R. Soc. Ser. B. 271: 529–536. Wright, J.J. & Page, L.M. 2006. Taxonomic revision of Lake Tanganyikan Synodontis (Siluriformes: Mochokidae). Florida Mus. Nat. Hist. Bull. 46: 99–154. Zwickl, D. 2006. Genetic algorithm approaches for the phylogenetic analysis of large biological sequence datasets under the maximum likelihood criterion, PhD thesis. University of Texas at Austin, Texas. Stewart, K.M. 2001. The freshwater fish of Neogene Africa (Miocene–Pleistocene): systematics and biogeography. Fish Fish 2: 177–230. Sturmbauer, C., Baric, S., Salzburger, W., Rüber, L. & Verheyen, E. 2001. Lake level fluctuations synchronize genetic divergences of cichlid fishes in African lakes. Mol. Biol. Evol. 18: 144–154. Swofford, D.L. 2002. P A U P * 4.0: Phylogenetic Analysis Using Parsimony (*and other Methods). Beta Version 4.0b4a. Sinauer Associates, Sunderland, MA. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. & Higgins, D.G. 1997. The CL U S T A L X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 24: 4876–4882. Turner, G.F. 1999. Explosive speciation of African cichlid fishes. In: Evolution of Biological Diversity (A.E. Magurran & R.M. May, eds), pp. 113–129. Oxford University Press, Oxford. Appendix 1 Provenance, voucher and GenBank accession data for samples (59 taxa) used in phylogenetic analyses. Accession numbers GenBank Species First described Country Locality Water body Vouchers Cyt b rpS7 Microsynodontis sp. Chiloglanis asymetricaudalis Synodontis East Africa S. afrofischeri S. afrofischeri S. af. afrofischeri S. af. ilebrevis S. irsacae S. frontosa1 S. frontosa2 S. frontosa3 S. frontosa4 S. frontosa5 S. frontosa6 S. grandiops Boulenger 1903 De Vos 1993 C TZ Cuvette-Ouest Kigoma Lékénie river Mukuti Bridge CU 89213 CU 90429 DQ886604 FM878873 – – U U TZ TZ ZM U U U U U U TZ Lake Victoria Lake Albert Malagrasi river Lake Tanganyika Lake Tanganyika Lake Albert Lake Albert Lake Albert Lake Albert Lake Albert Lake Albert Lake Tanganyika BMNH 2006.3.6.34* (5550) BMNH 2007.8.29.31 (5547) CU 90419 BMNH 2006.3.6.25 (5153) BMNH 2006.3.6.20 (5053) BMNH 2007.8.29.19 (5540) BMNH 2007.8.29.20 (5541) BMNH 2007.8.29.21 (5542) BMNH 2007.8.29.22 (5543) BMNH 2007.8.29.23 (5544) Tissue only CU 91902 DQ886618 FM878842 DQ886616 DQ886644 DQ886653 FM878851 FM878850 FM878852 FM878853 FM878854 FM878876 FM878846 FM878890 FM878917 FM878889 FM878902 FM878900 FM878913 FM878912 FM878914 FM878915 – – FM878897 TZ TZ ZM MW U Jinja Wanseko Kigoma Sibwesa Mpulungu Wanseko Wanseko Wanseko Wanseko Wanseko Wanseko Jakobsen’s Beach Ikola Kigoma Mpulungu Nkhata Bay Wanseko Lake Lake Lake Lake Lake UF 160945 BMNH 2006.3.6.15 (5126) BMNH 2006.3.6.9 (5143) BMNH 2006.3.6.1 (4766) BMNH 2007.8.29.24 (5546) DQ886651 DQ886631 DQ886625 DQ886620 FM878855 FM878898 FM878903 FM878896 FM878907 FM878880 Borodin 1936 Boulenger 1906 ZM ZM TZ? U U MZ TZ U Mpulungu Mpulungu n.a. Lake Albert Wanseko Niassa Area Kigoma Jinja Lake Tanganyika Lake Tanganyika n.a. Lake Albert Lake Albert Lugenda river Lake Tanganyika Lake Victoria BMNH 2006.3.6.29 (5145) BMNH 2006.3.6.21 (5100) BMNH 2007.8.29.25 (5537) (Tr) BMNH 2007.8.29.26 (5545) Tissue only SAIAB 73898 CU88758 BMNH 2006.3.6.14 (5518) DQ886638 DQ886645 FM878847 FM878849 FM878848 FM878859 DQ886658 DQ886657 FM878901 FM878899 FM878909 FM878911 FM878910 FM878906 FM878904 FM878908 Schilthuis 1891 Poll 1959 C C n.a. n.a. n.a. n.a. BMNH 2006.3.6.35 (Tr) BMNH 2006.3.6.36 (Tr) DQ886605 DQ886606 FM878882 FM878883 S. S. S. S. S. granulosa lucipinnis multipunctata njassae nigrita S. petricola S. polli S. af. punctulata S. serrata1 S. serrata2 S. sp. nov S. af. tanganaicae S. victoriae Central Africa S. angelica S. brichardi Cuvier 1817 Hilgendorf 1888 Wright & Page 2006 Wright & Page 2006 Vaillant 1895 Wright & Page 2006 Boulenger 1900 Matthes 1959 Boulenger 1898 Keilhack 1908 Cuvier & Valenciennes 1840 Wright & Page 2006 Matthes 1959 Günther 1889 Rüppell 1829 Tanganyika Tanganyika Tanganyika Malawi Albert ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY 816 J. J. DAY ET AL. Appendix 1 (Continued) Accession numbers GenBank Vouchers Cyt b rpS7 n.a. Mambili river Lékénie river Lékoli river Congo river Luapula river Chambeshi river Cuvette-Ouest Lékoli river Cuvette-Ouest Mambili river Sud Lobé river Cuvette-Ouest Congo river Luapula Lake Mwueru CU 91408 CU 88252 CU 87469 CU 88152 AMNH 235728 CU 91002 CU 91003 CU 87539 CU 87571 CU 89984 AMNH 235730 CU 91011 FM878875 DQ886607 DQ886608 DQ886609 DQ886610 FM878856 FM878857 DQ886611 DQ886612 DQ886613 DQ886614 FM878874 FM878925 FM878885 FM878881 FM878884 FM878894 – – FM878887 FM878886 FM878877 FM878888 FM878891 BW BW MZ MZ Boulenger 1905 BW ZM ZM ZM BW Skeleton & White 1990 BW BW Boulenger 1911 BW Peters 1852 MZ Okavango Okavango Manica Area Manica Area Okavango Luapula Luapula SAIAB 68779 SAIAB 68679 SAIAB 67669 SAIAB 67674 SAIAB 52275 CU 91094 CU 91092 CU 91099 SAIAB 52277 SAIAB 68674 SAIAB 68675 SAIAB 68721 SAIAB 64838 FM878860 FM878867 FM878862 FM878863 DQ886615 FM878845 FM878843 FM878844 FM878861 FM878865 FM878864 FM878866 FM878858 FM878918 FM878923 – FM878920 FM878895 Okavango Okavango Okavango Okavango Maputo area Thoage river Okavango river, Maunachira Buzi river Buzi river Thoage river Lake Bangwuelu Luapula river Chambeshi river Thoage river Okavango river, Boro Okavango river, Boro Maunachira river Maputo river Günther 1864 Poll 1971 Magburaka Magburaka Magburaka Bumbuna Bumbuna Rokel Rokel Rokel Rokel Rokel Tissue only (5765) FM878868 Tissue only (5530) FM878872 Tissue only (5531) FM878871 BMNH 2007.8.29.6 (5535) FM878869 BMNH 2007.8.29.7 (5536) FM878870 Species First described Country Locality S. clarias S. congica S. contracta S. decora S. greshoffi S. katangae S. katangae S. nigriventris S. pleurops S. rebeli S. soloni S. unicolor Southern Africa S. leopardina S. macrostiga S. nebulosa S. nebulosa S. nigromaculata1 S. nigromaculata2 S. nigromaculata3 S. nigromaculata4 S. af. thamalakalensis S. vanderwaali S. vanderwaali S. woosnami S. zambezensis West Africa S. gambiensis S. thysi S. thysi S. waterloti1 S. waterloti2 Linné 1758 Boulenger 1901 Vinciguerra 1928 Boulenger 1899 Schilthuis 1891 Boulenger 1901 CAR C C C C ZM ZM C C CR C ZM David 1936 Boulenger 1897 Holly 1926 Boulenger 1899 Boulenger 1915 Pellegrin 1914 Boulenger 1911 Peters 1852 Daget 1962 SL SL SL SL SL Water body Nana-Grébiz Cuvette-Ouest Cuvette-Ouest Cuvette-Ouest Cuvette-Ouest Luapula river river river river river FM878892 FM878893 FM878919 – FM878921 FM878922 FM878905 FM878924 FM878916 FM878878 FM878879 Superscript numbers denote individuals of the same species. Locality codes. B, Botswana; CAR, Central African Republic; CR, Cameroon; C, Congo; MW, Malawi; MZ, Mozambique; SL, Sierra Leone; TZ, Tanzania; U, Uganda; ZM, Zambia; n.a., not available. Repository codes. AMNH, American Museum of Natural History; BMNH, The Natural History Museum, London; CU, Cornell University Museum of Vertebrates; SAIAB, South African Institute for Aquatic Biodiversity; UF, Florida Museum. Numbers in parenthesis = JJD field tags. Tr = fish acquired through aquarium trade. Note on taxonomy. Following Wright & Page (2006), the ‘dwarf form’ of S. petricola (Day & Wilkinson, 2006) is now referred to as S. lucipinnis, whereas S. af. petricola is referred to here as S. petricola. Specimens described as S. dhonti in previous studies are now referred to as S. irsacae, as S. dhonti is only known from a single specimen. It should also be noted that we follow the taxonomic names outlined in Ferraris (2007), in which species names are now feminine. ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY Lacustrine radiations in African Synodontis catfish 817 Appendix 2 Locality and GenBank accession data for samples common to the data sets of *Day & Wilkinson (2006), Koblemüller et al. (2006) and this study. GenBank accession numbers Taxon Locality Cytochrome b Control region ND6 S7 intron 1 S. S. S. S. S. S. S. S. S. S. S. S. Congo river, n.a.*; CAR Congo river, n.a.*; CAR Congo river, C*; CAR Thorage river, BW*; Lake Mweru, ZM LM LV LT LT LT LT LT LT DQ886605 DQ886606 DQ886611 DQ886615 DQ886620 DQ886657 DQ886651 DQ886653 DQ886631 DQ886625 DQ886638 DQ886645 DQ662944 DQ662943 DQ662941 DQ662954 DQ663012 DQ663013 DQ662962 DQ662997 DQ662989 DQ662963 DQ662977 DQ662987 DQ663033 DQ663035 DQ663030 DQ663043 DQ663077 DQ663059 DQ663053 DQ663056 DQ663051 DQ663046 DQ663050 DQ663052 FM878882 FM878883 FM878887 FM878895 FM878907 FM878908 FM878898 FM878900 FM878903 FM878896 FM878901 FM878899 angelica brichardi nigriventris nigromaculata njassae victoriae granulosa irsacaeà lucipinnis§ multipunctata petricola polli Note that voucher specimens from independent author studies are not the same. With the exception of S. nigromaculata, specimens are from similar localities. àS. dhonti in previous studies, §S. petricola (Day & Wilkinson, 2006); S. sp. nov. (Koblemüller et al., 2006). Received 4 September 2008; revised 8 December 2008; accepted 18 December 2008 ª 2009 THE AUTHORS. J. EVOL. BIOL. 22 (2009) 805–817 JOURNAL COMPILATION ª 2009 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY