Introduction

The biodiversity and geology of Sulawesi, the largest island in Wallacea located at the boundary between Asia and Oceania (Whitten et al. 2002; Hall 2011, 2012), have attracted considerable research attention since Alfred Wallace’s expeditions (Wallace 1876). The central parts of this island are characterized by ancient lake systems, which are considered to have been formed by Pliocene collisions of tectonic subdivisions of the island (Brooks 1950; Ahmad 1978; Russell et al. 2020). Two tectonic lake systems, Lake Poso and the Malili Lake system, are especially well known for their endemic aquatic fauna and unique adaptive radiation and/or speciation (e.g., von Rintelen et al. 2012; Pfaender et al. 2016; Kakioka et al. 2021; Mandagi et al. 2021; Wasiljew et al. 2021; Hilgers et al. 2022). However, taxa living in these ancient lakes remain largely undescribed (e.g., Herder et al. 2006; Poettinger and Schubart 2014; Klotz et al. 2021).

The Malili Lake system comprises three large (Towuti, Mahalona, and Matano) and two satellite (Lantoa and Masapi) tectonic lakes (Fig. 1) which are connected by rivers to form a single water system. Ricefishes of the family Adrianichthyidae are a main component of the fish faunas in this lake system, and four valid species [Oryzias matanensis (Aurich 1935), Oryzias hadiatyae Herder and Chapuis 2010, Oryzias profundicola Kottelat 1990b, Oryzias marmoratus (Aurich 1935)] occur here. Of these, the first three are endemic to Lakes Matano, Masapi, and Towuti, respectively, while the last is more widely distributed and occurs in Lakes Towuti, Mahalona, and Lantoa (e.g., Kottelat 1990b, 2013; Kottelat et al. 1993; Herder and Chapuis 2010).

Fig. 1
figure 1

Map of locations of material examined from the Malili Lake system (Map by T. von Rintelen, modified). Stars indicate localities of Oryzias loxolepis sp. nov.: black holotype and two paratypes (Tanjung Timbala), white 37 paratypes (Lengkona). Solid circles represent locations of non-type congener specimens collected from 2013–2018

Full size image

Oryzias marmoratus, originally named Aplocheilus marmoratus by Aurich (1935), was described from 22 syntypes from Lakes Towuti and Mahalona, but type specimens were lost during World War II (Kottelat 1990b). Kottelat (1990b) designated a neotype of A. marmoratus based on a specimen newly collected from a small creek flowing into Lake Towuti (Fig. 2) and reported this species also from Lake Lantoa (as Lake Wawontoa). He also described substantial morphological variation in fishes from Lake Towuti and suggested that more than one species, or O. marmoratus × O. profundicola hybrids might exist there. Genetic variation in O. marmoratus within Lake Towuti was also reported using mitochondrial and nuclear phylogenies and population genetic structure based on microsatellite markers (Mokodongan and Yamahira 2015; Ansai et al. 2021; Yamahira et al. 2023). Herder and Chapuis (2010) described morphometric variation in O. marmoratus between Lakes Towuti and Lantoa. Geometric morphometric analysis by Mandagi et al. (2021) also revealed O. marmoratus body shape to differ among the three lake populations, and analyses of population genetic structure also indicated that these three lake populations had clearly separated microsatellite and genome-wide single nucleotide polymorphism (SNP) markers (Mandagi et al. 2021; Yamahira et al. 2023).

Fig. 2
figure 2

Neotype of Aplocheilus marmoratus Aurich 1935 (ZRC 38449, male, 36.3 mm SL). Upper, lateral view; lower, radiograph (by H. H. Tan)

Full size image

Despite recognized within- and between-lake morphological and genetic variation in specimens attributed to O. marmoratus, no detailed taxonomy of these populations has been conducted. Accordingly, we revise the taxonomy of the “Oryzias marmoratus complex”, i.e., the three lake populations of O. marmoratus and O. profundicola. We first define diagnostic characters of the “O. marmoratus complex”. Second, after re-describing O. marmoratus, we describe a new species, Oryzias loxolepis. Third, we demonstrate that O. loxolepis sp. nov. is phylogenetically distinct from other “O. marmoratus complex” species based on genome-wide single nucleotide variants (SNVs). Finally, we review the taxonomy of specimens previously reported as “O. marmoratus”.

Materials and methods

Material collections. For morphological and molecular analyses, we examined specimens of five Malili species (Oryzias loxolepis sp. nov., O. marmoratus, O. profundicola, O. hadiatyae, O. matanensis) collected by ourselves from 2013–2018 (Fig. 1). For each specimen, the right pectoral fin was removed and stored in 99% ethanol, and the remaining body was preserved in 70% ethanol after fixation in 10% formalin. Two specimens (MZB 22749 and 22750) were wholly fixed in 99% ethanol and stored intact. Several live young O. marmoratus individuals were transferred from Lake Lantoa and Lake Mahalona to the World’s Medaka Aquarium (WMA), Nagoya Higashiyama Zoological Park, Nagoya, in Japan in December 2018, and two specimens of Lantoa O. marmoratus (NSMT-P 143823 and 143824) and two specimens Mahalona O. marmoratus (NSMT-P 143799 and 143804) were obtained after keeping in aquariums for approximately one year; they were preserved in 70% ethanol after fixation in 10% formalin. We also examined morphologies of the primary type specimens of O. marmoratus (ZRC 38449), O. matanensis (ZRC 38450), O. profundicola (MZB 5868) and O. hadiatyae (MZB 18491).

Morphology. Counts and measurements follow Kottelat (1990a) and Parenti (2008), excepting: transverse scales (or series), which we count from the dorsal-fin origin extending downward and backward to the base of the anal fin, excluding the scale immediately in front of the dorsal-fin origin and irregular sets of small scales around the anal-fin origin; postorbital length, which we measure from the posterior edge of the eye to the posteriormost point of the opercle (including membrane); body depth, which we measure at points of both anal- and dorsal-fin origins; and anal-fin length, which we measure from the anal-fin origin to the posteriormost tip of the anal fin (usually the last ray tip). Counts of unpaired fin rays and vertebrae (total = abdominal + caudal) are made from radiographs. The last two dorsal and anal rays articulating on the last pterygiophore are separately counted. The hypural centrum was counted as a caudal vertebra. Other meristics were recorded directly from each specimen, using a dissecting microscope (Leica, MZ-75). Morphometric and meristic data were obtained from specimens fixed in formalin, except for the four specimens of O. marmoratus that were reared at WMA (NSMT-P 143799, 143804, 143823, and 143824). We report measurements and meristics as ranges, with the value for the primary type presented in square brackets. Descriptions of fresh color are based on photographs taken immediately after fixation for specimens, and/or when they were alive. Neuromasts were observed for one individual of each species by staining with 2-Di-4-ASP (Nakae et al. 2012). The terminology of cephalic and lateral superficial neuromast rows follows Ishikawa (1994) and Seleit et al. (2021) with the following minor modifications: AG, anterior groove; CNC, caudal neuromast cluster; DG, dorsal groove; dl, dorsal line; dll, dorsolateral line; PG, posterior groove; mll, midlateral line; vla, ventral line from anterior of anal-fin origin; vlla, ventrolateral line; vllp, ventrolateral line; and vlp, ventral line from posterior of anal-fin origin.

In the subsection “Oryzias marmoratus complex” below, morphometric and meristics data obtained above for the five Malili species are combined with data from Kottelat (1990a) and Herder and Chapuis (2010), to ensure that reported differences based on our data remained when incorporating more data from former descriptions. However, in subsections describing O. marmoratus and O. loxolepis, we use only our morphometric and meristics data to avoid potential measurer bias in comparisons with other species in the O. marmoratus complex. For non-Malili congeners, morphological data and body color descriptions from the following previous studies are used in comparisons (Kottelat 1990b; Parenti 2008; Magtoon 2010; Parenti and Hadiaty 2010; Herder et al. 2012; Parenti et al. 2013; Mokodongan et al. 2014; Mandagi et al. 2018; Roberts et al. 2021; Gani et al. 2022; Utama et al. 2022), excepting the hitherto unreported ‘anal-fin base length’ of Oryzias sarasinorum (Popta 1905) [novel data were obtained from two museum specimens (BMNH 1914.2.13.26–27: 13.4–22.0% SL)]. Although the generic-level taxonomy of the Adrianichthyidae is controversial (Roberts et al. 2021), we tentatively follow generic definitions proposed by Parenti (2008).

Museum collection acronyms follow Sabaj (2022) except for Wallacean Collection of Natural History, Manado City Museum, Sulawesi Utara (WMSU). Identifications of fish species collected along with Oryzias are based mainly on Kottelat et al. (1993), Herder et al. (2006), Larson et al. (2014), and Hoese et al. (2015).

Observation of reproductive traits. Each of the Lantoa and Mahalona O. marmoratus individuals transferred to WMA was maintained as closed colonies. Using these closed colony individuals, we observed mating behaviors under aquarium conditions (26 °C, 13L:11D) in August 2021. We also obtained data on the number of fertilized eggs, fertilized egg diameter, the number of days before hatching, and larval size soon after hatching.

Phylogenetic analysis using genome-wide single nucleotide variants. Genomic DNA was extracted from the pelvic fin of the O. loxolepis holotype (MZB 26367) and two paratypes (NSMT-P 143828, 143829) using a Maxwell RSC Blood DNA Kit (Promega, Fitchburg, WI, USA). Whole-genome sequencing (WGS) was performed with an Illumina NovaSeq6000 platform at Novogene Japan K.K., where the library was constructed using a NEBNext Ultra II DNA Library Prep Kit (Cat No. E7645). Raw sequences of these three specimens have been registered in the DNA Data Bank of Japan Sequence Read Archive (DRA) DRA015327. We also retrieved WGS short reads of seven Malili species/populations and two outgroup species (Oryzias dopingdopingensis Mandagi et al. 2018 from the Doping-doping River and Oryzias soerotoi Mokodongan et al. 2014 from Lake Tiu) in Ansai et al. (2021) from DRA010665 (one individual per species/population) and short reads obtained from double-digest restriction-site-associated DNA sequencing (ddRAD-seq) of six species/populations in Mandagi et al. (2021) from DRA010303 (20 individuals per species/population). Although the identifications of “O. marmoratus (Towuti 1)” of Ansai et al. (2021) and “O. marmoratus Towuti F03” of Mandagi et al. (2021) are based on the same individual, we treat them separately in our analysis to confirm that reads obtained from WGS and those from ddRAD-seq were successfully merged.

Reads were adapter-trimmed, and low-quality (mean quality < 5) and short (< 50 bp) reads were filtered out using fastp version 0.21.0 (Chen et al. 2018). Remaining reads were then mapped to an Oryzias celebensis (Weber 1894) reference, OryCel_1.0 (Ansai et al. 2021), with the BWA-MEM algorithm in BWA version 0.7.17 -r1118 (Li 2013). Generated SAM files were converted to sorted BAM files using samtools version 1.13 (Danecek et al. 2021), with and without removing PCR duplicates for files from WGS and those from ddRAD-seq, respectively. Variant calling was then conducted separately for each individual sequence using elprep version 5.1.1 (Herzeel et al. 2021). The resultant genomic variant call format (GVCF) files from all individuals were jointly genotyped across the O. celebensis reference genome using Genome Analysis Tool Kit (GATK) version 4.2.0.0 (van der Auwera and O’Connor 2020). In GATK, information for each chromosome was separately treated and stored into GenomicsDB format using the “GenomicsDBImport” tool. After the genotyping process was conducted using the “GenotypeGVCFs” tool, all chromosomal files were concatenated as a single VCF file by the “GatherVcfs” tool. SNVs were then extracted from this VCF file using BCFtools version 1.13 (Danecek et al. 2021), in which SNV positions with extremely low (DP < 10) or high read depth (DP > 200) in one or more individuals were removed and no missing data were allowed. Finally, we had 5,867 SNVs from 72 individuals from nine populations/species (seven Malili populations/species including the new species, O. dopingdopingensis, and O. soerotoi).

A maximum likelihood (ML) phylogeny was estimated using O. dopingdopingensis and O. soerotoi as outgroups. The VCF file containing 5,867 SNVs was converted to a PHYLIP file using vcf2phylip version 2.8 (Ortiz 2019). Invariant sites were removed using a script “raxml_ascbias” (https://github.com/btmartin721/raxml_ascbias), resulting in 1,723 SNVs [Electronic Supplementary Material (ESM) S1]. Analysis was performed with RAxML-NG version 1.1.0 (Kozlov et al. 2019) using the codon-specific GTR + G model, in which a transfer bootstrap expectation analysis of 1,000 replicates was performed to calculate branch support, and ascertainment bias was corrected by Lewis’s method.

Specimens used in Mandagi et al. (2021) were reidentified based on images taken soon after collection (ESM S2).

Results

Taxonomy

Oryzias marmoratus complex


Included species. Oryzias marmoratus, O. profundicola, and O. loxolepis sp. nov.

Differential diagnosis. This complex of three species differs from all 16 Sulawesi congeners in the following combination of characters: presence of dark blotches on the lateral body (vs. completely absent in Oryzias nigrimas Kottelat 1990a, Oryzias orthognathus Kottelat 1990a, Oryzias nebulosus Parenti and Soeroto 2004, O. soerotoi, Oryzias woworae Parenti and Hadiaty 2010, Oryzias asinua Parenti et al. 2013, and Oryzias wolasi Parenti et al. 2013, and vs. presence of a black stripe on posterior half of trunk in O. celebensis); 27–31 total vertebrae (vs. 32 or 33 in O. nigrimas, 33 in O. orthognathus, and 34 in O. sarasinorum); 9–13 dorsal-fin rays (vs. 8 in O. woworae); 20–29 anal-fin rays (vs. 18 or 19 in O. woworae, 17–19 in O. asinua, 17 or 18 in Oryzias eversi Herder et al. 2012); 30–35 lateral scales (vs. 41–47 in O. matanensis, 45–57 in O. orthognathus, 70–75 in O. sarasinorum, 36–39 in Oryzias bonneorum Parenti 2008, and 61–67 in Oryzias kalimpaaensis Gani et al. 2022); shorter head length (22.4–29.6% SL vs. 31–32% in O. bonneorum and 30.1–33.7% in O. kalimpaaensis); shorter snout length (6.6–9.5% SL vs. 11% in O. sarasinorum); larger body depth at anal-fin origin (24.7–35.2% SL vs. 17.9–25.7% in O. hadiatyae, 17.9–22.2% in O. nigrimas, 17.1–21.9% in O. orthognathus, 13–15% in O. sarasinorum, 17–20% in O. bonneorum, 18.9–24.6% in O. eversi, and 16.0–22.2% in O. kalimpaaensis); and longer anal-fin base length (31.4–43.4% SL vs. < 30% in the above 13 congeners excepting O. celebensis, O. matanensis, and O. orthognathus). From two congeners in the Malili Lake system (O. matanensis and O. hadiatyae), males of the three species in this complex differ in having an anal fin with a proximal dark purple blotch on membranes between some soft rays (vs. no blotch: Fig. 3c, d).

Fig. 3
figure 3

Specimen photographs immediately after fish fixation from the Malili Lake system: Oryzias profundicola (a NSMT-P 143841, male, 42.6 mm SL; b NSMT-P 143843, female, 43.4 mm SL), O. matanensis (c NSMT-P 143866, male, 37.0 mm SL; d NSMT-P 143871, female, 34.8 mm SL), and O. hadiatyae (e NSMT-P 143859, male, 40.5 mm SL; f NSMT-P 143862, female, 38.9 mm SL)

Full size image

Among non-Sulawesian congeners defined in Parenti (2008), Oryzias luzonensis (Herre and Ablan 1934) and Oryzias timorensis (Weber and de Beaufort 1922) also have dark blotches on the body, but members in the O. marmoratus species complex differ from them in dorsal-fin ray counts (9–13 vs. 5–7 in O. luzonensis) and anal-fin ray counts (20–29 vs. 17–19 in O. timorensis; 15–19 in O. luzonensis).

Remarks. The three species in the O. marmoratus species complex co-occur in Lake Towuti. Oryzias marmoratus and O. loxolepis sp. nov. were formerly treated as a single species (see discussion). We separately describe them below. No taxonomic revision is necessary for O. profundicola; a detailed description was provided by Kottelat (1990b).

Oryzias marmoratus (Aurich 1935)


(English name: Marmorated Ricefish)

(Figs. 1, 2, 4–10, 16, 17; Tables 1, 2, 4)

Fig. 4
figure 4

Oryzias marmoratus immediately after fixation from Lake Towuti (a NSMT-P 143825, male, 27.9 mm SL), Lake Lantoa (b NSMT-P 143806, male, 39.2 mm SL; c NSMT-P 143808, male, 25.2 mm SL; d NSMT-P 143807, female, 31.6 mm SL), and Lake Mahalona (e NSMT-P 143796, male, 29.7 mm SL; f NSMT-P 143797, female, 23.8 mm SL)

Full size image
Table 1 Counts and morphometrics of select Oryzias marmoratus specimens from Lake Towuti (NSMT-P 143825–143827), Lake Lantoa (NSMT-P 143806–143820; MZB 26344, 26346, 26348, 26357, 26365), and Lake Mahalona (NSMT-P 143787–143798), expressed as percentages of standard length
Full size table
Table 2 Frequency distribution of select meristic counts of Oryzias species from the Malili Lake system
Full size table
Fig. 5
figure 5

Live adult Oryzias marmoratus from Lake Towuti (a male, not preserved), Lake Lantoa (b NSMT-P 143824, male; c NSMT-P 143823, female), and Lake Mahalona (d NSMT-P 143799, male; e NSMT-P 143804, female) (photo by N. Hashimoto)

Full size image
Fig. 6
figure 6

Preserved Oryzias marmoratus from Lake Towuti (a NSMT-P 143825, male, 27.9 mm SL), Lake Lantoa (b NSMT-P 143810, male, 33.2 mm SL; c NSMT-P 143814, female, 33.0 mm SL), and Lake Mahalona (d NSMT-P 143796, male, 29.7 mm SL; e NSMT-P 143797, female, 23.8 mm SL)

Full size image
Fig. 7
figure 7

Arrangement of superficial neuromasts in Oryzias marmoratus (NSMT-P 143820, male, 29.6 mm SL, from Lake Lantoa). AG anterior groove, CNC caudal neuromast cluster, DG dorsal groove, dl dorsal line, dll dorsolateral line, PG posterior groove, mll midlateral line, vla ventral line from anterior of anal-fin origin, vlla ventrolateral line, vllp ventrolateral line, vlp ventral line from posterior of anal-fin origin. a Dorsal, b lateral, and c ventral views

Full size image
Fig. 8
figure 8

Upper dorsal bodies of Oryzias marmoratus (a NSMT-P 143806, male, 38.8 mm SL), O. loxolepis sp. nov. (b NSMT-P 143829, male, 37.2 mm SL), and O. profundicola (c NSMT-P 143850, male, 43.6 mm SL). White broken lines indicate scale edges. White arrows indicate the point of dorsal-fin origin. Red dots indicate intercepts of the outer scale edges and the midlateral dark stripe

Full size image
Fig. 9
figure 9

Localities of Oryzias marmoratus and O. loxolepis sp. nov. from the Malili Lake system: a Tanjung Timbala, Lake Towuti; b Lengkona, Lake Towuti; c Lake Lantoa; d Lake Mahalona; e in situ underwater photograph of O. marmoratus in Lake Lantoa

Full size image
Fig. 10
figure 10

Relationship between standard length and a caudal peduncle length, b body depth at anal-fin origin, c body depth at dorsal-fin origin, d preanal length, e anal-fin base length, and f predorsal length for Oryzias species from the Malili Lake system. Symbols: closed males, open females; with black margin primary types

Full size image

Aplocheilus marmoratus Aurich 1935: 102–103, fig. 1B [type locality: Lingkoburanga in Lake Towuti (by neotype designation); original type locality: Lakes Towuti and Mahalona)].


Oryzias marmoratus: Kottelat 1990b: 155–159, figs. 2, 5 (designated neotype); Kottelat et al. 1993: 125, pl. 43 [Lakes Towuti, Mahalona, and Wawontoa (= Lantoa)); Parenti 2008: 569–571, fig. 45 [Lakes Towuti, Mahalona, and Wawontoa (= Lantoa)]; Herder and Chapuis 2010: 269 (partim: Lake Lantoa); Mokodongan and Yamahira 2015, Gani et al. 2022 [partim: Lakes Towuti (Lake Towuti A), Lantoa, and Mahalona]; Sumarto et al. 2020 (Lake Mahalona); Ansai et al. 2021 [partim: Lake Towuti (Towuti 2)]; Mandagi et al. 2021, Utama et al. 2022 (partim: Lakes Lantoa and Mahalona); Yamahira et al. 2021, 2023 [partim: Lake Lantoa (Lantoa-type)].


Neotype. ZRC 38449 (formerly ZSM 27172), male, 36.3 mm SL, small rivulets flowing into lake at Lingkoburanga, about 6 km south of Timampu, Lake Towuti, 22 Jun. 1988, coll. M. Kottelat.

Non-type specimens. 40 specimens, 23.0–40.0 mm SL. TOWUTI: 4 specimens, 23.5–28.6 mm SL. MZB 22749 (alcohol-fixed), male, 23.5 mm SL; NSMT-P 143827, male, 28.6 mm SL, Lengkona, northern coast of Lake Towuti (2°40′37″S, 121°41′26″E), 10 Dec. 2013, coll. D.F. Mokodongan and K. Yamahira; NSMT-P 143825–143826, 2 males, 27.5–27.9 mm SL, southern coast of Tanjung Timbala, western shore of Lake Towuti (2°41′24″S, 121°25′20″E), 29 Nov. 2018, coll. H. Kobayashi, R. Kakioka, S. Ansai, K.W.A. Masengi and K. Yamahira. LANTOA: 22 specimens, 25.2–38.8 mm SL. MZB 26345, 26346, 26348, 26357, 26365, 3 males and 2 females, 26.0–35.8 mm SL, southern shore of Lake Lantoa (2°40′29″S, 121°42′59″E), 9 Dec. 2013, coll. D.F. Mokodongan and K. Yamahira; NSMT-P 143806–143820 (Fig. 2), 7 males and 8 females, 25.2–38.8 mm SL, southern shore of Lake Lantoa (as for MZB 26343–26366), 4 Dec. 2018, coll. H. Kobayashi, R. Kakioka, S. Ansai, K.W.A. Masengi and K. Yamahira; NSMT-P 143823, 143824, 1 female and 1 male, 38.9 and 32.6 mm SL, collected with NSMT-P 143806–143820, reared at WMA and fixed on 4 Jan. 2020. MAHALONA: 14 specimens, 23.0–40.0 mm SL. NSMT-P 143787–143798, 6 males and 6 females, 23.0–29.7 mm SL, southern shore of Lake Mahalona (2°34′46″S, 121°28′59″E), 1 Dec. 2018, coll. H. Kobayashi, R. Kakioka, S. Ansai and K. Yamahira; NSMT-P 143799, 143804, 1 male and 1 female, 38.1 and 40.0 mm SL, collected with NSMT-P 143787–143798, reared at WMA and fixed on 4 Jan. 2020.

Diagnosis. Oryzias marmoratus differs from O. profundicola and O. loxolepis sp. nov. in the following character combination: usually 12 abdominal vertebrae, in total 29–31 vertebrae; 9 or 10 dorsal-fin rays; 22–25 anal-fin rays; 10 or 11 transverse scales; body depth at anal-fin origin 24.8–30.1% SL; body depth at dorsal-fin origin 20.2–25.3% SL; caudal peduncle length 12.0–16.3% SL; distinct nuchal concavity; lateral scales more-or-less aligned parallel to midlateral stripe; elongated shape of male dorsal fin without yellow margin.

Description. Detailed morphometric and meristic characters are provided in Tables 1 and 2. Maximum specimen size 38.8 mm SL (wild), 40.0 mm SL (aquarium) (Figs. 2, 4–6). Body compressed laterally, elongate, body depth at anal-fin origin 24.8–30.1% [26.4%] SL, at dorsal-fin origin 20.2–25.3% [24.2%] SL. Mouth superior; lower jaw slightly projecting beyond upper jaw, with 1–3 irregular rows of conical teeth on each. Males with irregular external small conical teeth on upper and lower lips, especially at mouth corner. Head moderate, length 24.8–29.6% [25.6%] SL. Dorsal surface of head slightly convex just anterior to orbit; nape with well-marked concavity. Dorsal and ventral body profile gently arching from head to points of dorsal and anal-fin origin; without pronounced abdominal concavity between pelvic and anal fins. Snout length 7.4–9.5% [7.4%] SL, eyes large, diameter 9.1–11.2% [9.6%] SL; orbit not projecting beyond dorsal surface of head. Infraorbital space broad, proximally more than half eye diameter [slightly more than half]. Caudal peduncle moderate, length 12.0–16.3% [12.9%] SL; depth 9.7–12.2% [11.0%] SL. Male with short, slightly conical, tubular urogenital papilla; female with bilobed urogenital papilla.

Scales of moderate size, cycloid, deciduous; 30–33 [32] along lateral midline; 10 or 11 [11] along transverse series extending backwards from points of dorsal to anal-fin origin. Dorsal-fin rays 9 or 10 [10], elongate and filamentous, numbers 3–6 [5] being longest in males. Anal-fin rays 22–25 [24]; anterior rays elongate and filamentous, numbers 3–9 [7] longest, without bony contact organs in males; straight or slightly concave in females. Dorsal-fin point of origin over base of anal-fin ray bases 12–14 [13]. Pectoral-fin rays 9–11 [10]. Pelvic-fin rays 6, last ray connected to body via membrane along proximal half. Caudal fin truncate, with principal caudal-fin rays i,4/5,i. Procurrent-fin rays 4–6/5 or 6 [4/5]. Vertebrae 29–31 [30] (11 or 12 [12] + 17–19 [18]); first pleural rib on third vertebra.

Sensory canals and rows of the lateral line system of a non-type specimen (NSMT-P 143820, 29.6 mm SL, from Lantoa Lake) are illustrated in Fig. 7. Circumorbital sensory canal in open bony groove (AG, DG, and PG). Infraorbital and snout with small and numerous neuromast rows aligned between orbit and groove. Neuromasts in posterior infraorbital region and behind lower jaw lip not observed because of damage (possibly small rows like O. loxolepis sp. nov.). Lower opercle with two, and dorsal pectoral-fin base with five neuromast clusters. Anterior body from point of anal-fin origin with small transverse rows horizontally aligned (vlla and vla): ventrolateral line (vlla) with 7 rows; ventral line (vla) with 6 rows along throat and pelvic insertion, with right and left rows coalescing as 2 clusters in center of inter-pelvic region to point of anal-fin origin. Posterior body with 4 short, major, horizontally aligned transverse rows (dll, mll, vllp and vlp): dorsolateral line (dll) with 3 rows, slightly above midlateral line (mll) with 8 rows, ventrolateral line (vllp) with 6 rows, and ventral line above anal-fin base (vlp) with 8 rows. In center of predorsal region, dorsal line (dl) aligned backward from nape with 12 short rows. Each lateral row and cluster comprises 3–7 neuromast papillae. Central part of caudal fin with a neuromast cluster (CNC) slightly separated into two.

Color in life. Non-type Towuti (NSMT-P 143825) and Lantoa (NSMT-P 143806–143808) specimens with body and head entirely grayish-brown (Figs. 4, 5). Trunk with diffuse dark brown blotches, including a row of 5–9 larger blotches dorsal to midaxial stripe; small dark blotches dispersed in large male; broad and vertical dark blotches on midaxial stripe in young males from Lake Mahalona. Dorsal proximal half of head and body brown, brilliant silver and blue along opercle to ventral half of belly. Midlateral brilliant grayish-silver with pale yellow reticulation along scale edges. Eyes blue, infraocular region white. Dorsal fin with black distal margin, filaments hyaline, yellow. Anterior half of anal fin of male with distinct yellow margin, posterior half with black distal margin; anal fin of female with diffuse black margin. Male anal-fin rays hyaline or yellow along margin; in females yellow hyaline along margin. Dorsal and anal fins of neotype and most other male specimens sometimes with single proximal dark purple blotch on some interradial membranes. Dorsal and ventral margin of caudal fin bright yellow; with 2–4 longitudinal dark purple stripes on proximal half of membranes between median caudal-fin rays. Male pelvic fin yellow, female pelvic fin hyaline. Pectoral fin hyaline.

Color in alcohol. Ground color yellowish-gray (Figs. 2, 6). Diffuse brown spots on trunk remain, with diffuse oval spot above upper edge of opercle. Dorsal surface of head and dorsal and lateral surfaces of body with dense dark brown chromatophores. Melanophore row from dorsal head surface to point of dorsal-fin origin diffuse; midlateral line from head to caudal-fin base black, diffuse, and across approximately central part of lateral scales (Fig. 8a). Female: all fins hyaline. Male including neotype: pectoral fin hyaline; pelvic, dorsal and anal fins dusky; dorsal- and anal-fin rays with one brown blotch on some interradial membranes; anal-fin margin white anteriorly, black posteriorly; caudal-fin rays dusky, with dark line extending along proximal half of membrane between median caudal-fin rays. Urogenital papilla of females and males immaculate or with several melanophores.

Sexual dimorphism. Males and females attain comparable maximum size but differ in live color (as above). Males with filamentous dorsal and anal-fin rays. Large males with external conical teeth on each jaw. Females with bilobed genital papilla; males with short, conical, tubular papilla. Males lack modified anal-fin rays.

Reproduction. Aquarium observations reveal mating to occur in pairs. The male wraps his dorsal and anal fins around the female, while the female releases eggs, which are then instantly fertilized. Females are not pelvic brooders; eggs are carried on a genital pore for approximately 1 h (observed in Mahalona population: 34–120 min, mean 59 min, n = 7) before being deposited onto submerged yarn.

Reproductive traits observed in an aquarium at 26 °C for individuals from Lakes Lantoa and Mahalona, respectively, include: number of fertilized eggs in a clutch 2–5 (mean 3.3, n = 3) and 5–19 (mean 10.0, n = 4); fertilized egg diameter 1.3–1.4 (mean 1.35, n = 10) and 1.3–1.5 (mean 1.36, n = 52) mm; number of days before hatching 19–27 (mean 24, n = 4) and 11–14 (mean 12.3, n = 33); larval size (total length) soon after hatching 4.3–5.1 (mean 4.7, n = 4) and 4.8–5.8 (mean 5.2, n = 33) mm.

Distribution. Lakes Towuti, Lantoa, and Mahalona, southern Sulawesi Province, Indonesia.

Habitat. In 2013 and 2018, we collected and observed specimens only during the day, never at night. At Tanjung Timbala (Fig. 9a), slightly south of the type locality, and at Lengkona (Fig. 9b) in Lake Towuti, specimens were collected with O. loxolepis sp. nov. and O. profundicola. These three species form schools in gently sloping, shallow (< 1.5 m) shore areas with large boulders and submerged wood over sand and gravel; additional taxa occurring in samples included Paratherina spp., Glossogobius flavipinnis (Aurich 1938), Glossogobius intermedius Aurich 1938, Glossogobius sp., Mugilogobius rexi Larson 2001, Mugilogobius latifrons (Boulenger 1897), and Nomorhamphus megarrhamphus (Brembach 1982). In Lake Lantoa (Fig. 9c), O. marmoratus inhabits gently sloping, shallow shores (1 m depth) with a large area of plants on a muddy substratum; it swims in schools (Fig. 9e) and was collected in association with Telmatherina celebensis Boulenger 1897 and Glossogobius sp. (an undescribed lake endemic: HK, pers. obsv.). Some individuals were infested with parasitic copepods. In Lake Mahalona (Fig. 9d), O. marmoratus inhabits gently sloping, shallow shores (1 m depth) with large boulders over a sand and gravel substratum, covered with aquatic vegetation; it also usually forms schools; co-occurring taxa include Tominanga aurea Kottelat 1990c, Glossogobius sp. (as for Lake Towuti), Glossogobius mahalonensis Hoese et al. 2015, G. intermedius, M. rexi, and Mugilogobius cf. hitam Larson et al. 2014.

Comparison. In the O. marmoratus complex, O. marmoratus differs from the other two species in having a longer caudal peduncle (12.0–16.3% SL vs. 8.3–11.7% in O. profundicola; 8.0–11.2% in O. loxolepis sp. nov.; Fig. 10a), more abdominal vertebrae (usually 12, rarely 11 vs. only 11 in O. profundicola and O. loxolepis sp. nov.; Table 2); an elongated posterior edge of the male dorsal fin (vs. rounded), superior mouth (vs. terminal), and better-developed nuchal concavity (vs. opposed to reduced). Oryzias marmoratus also differs from O. profundicola and O. loxolepis sp. nov. in the trunk scale pattern, the scales of which line up approximately parallel with the lateral line (Fig. 8a) (vs. aslant to the lateral line: Fig. 8b, c).

Oryzias marmoratus differs from the sympatric O. profundicola in Lake Towuti in having fewer dorsal-fin rays (9 or 10 vs. 11–13; Table 2) and transverse scales (10 or 11 vs. 13–15; Table 2), a more shallow body at anal-fin origin (24.8–30.1% SL vs. 30.4–35.2%; Fig. 10b) and dorsal-fin origin (20.2–25.3% SL vs. 28.0–23.7%; Fig. 9c), and in fresh coloration (Fig. 3a, b), with the former being grayish-brown trunk (vs. golden yellow), the opercle being silver (vs. golden) in both sexes, and in live males having a diffuse black dorsal fin (vs. remarkable yellow).

For allopatric congers in the Malili Lake system, our measurements of O. marmoratus reveal it to differ from O. hadiatyae (Fig. 3e, f) in having a deeper body at anal-fin origin (24.8–30.1% SL vs. 17.9–24.0%; Fig. 10b), shorter preanal length (53.4–59.2% SL vs. 60.3–63.5%; Fig. 10d), and longer anal-fin base (32.0–37.9% SL vs. 24.4–28.9%; Fig. 10e); it also tends to differ from O. matanensis (Fig. 3c, d) in having a shorter predorsal length (71.8–76.9% SL vs. 76.7–79.8%; Fig. 10f). A specific comparison with O. loxolepis sp. nov. is presented below.

Oryzias loxolepis Kobayashi, Mokodongan and Yamahira, new species


(New English name: Towuti Ricefish)

(Figs. 1, 8, 9, 10, 11, 12, 13, 14, 15, 16 and 17; ESM S2; Tables 2, 3 and 4)

Fig. 11
figure 11

Oryzias loxolepis sp. nov. immediately after fixation: a, b MZB 26367, holotype, male, 40.1 mm SL; c NSMT-P 143829, paratype, male, 37.2 mm SL; d NSMT-P 143828, paratype, male, 36.9 mm SL

Full size image
Fig. 12
figure 12

Live adult Oryzias loxolepis sp. nov.: upper, male; lower, female (not preserved, photo by K. Mochida)

Full size image
Fig. 13
figure 13

Preserved Oryzias loxolepis sp. nov.: a MZB 26367, holotype, male, 40.1 mm SL; b NSMT-P 143833, paratype, male, 40.2 mm SL; c NSMT-P 143837, paratype, female, 35.4 mm SL

Full size image
Fig. 14
figure 14

Arrangement of superficial neuromasts in Oryzias loxolepis sp. nov. (NSMT-P 143831, paratype, male, 34.1 mm SL). AG anterior groove, CNC caudal neuromast cluster, DG dorsal groove, dl dorsal line, dll dorsolateral line, PG posterior groove, mll midlateral line, vla ventral line from anterior of anal-fin origin, vlla ventrolateral line, vllp ventrolateral line, vlp ventral line from posterior of anal-fin origin. a Dorsal, b lateral, and c ventral views

Full size image
Fig. 15
figure 15

In situ photograph of Oryzias loxolepis sp. nov. with congeners at Lengkona, Lake Towuti: O. marmoratus (m), O. loxolepis sp. nov. (c), and O. profundicola (p)

Full size image
Fig. 16
figure 16

Relationship between standard length (SL; mm) and a eye diameter, b upper jaw width, c snout length, d head length, e postorbital length, and f dorsal-fin base length (all as % of SL) in Oryzias species from the Malili Lake system. Symbols: closed males, open females; with black margins primary types

Full size image
Fig. 17
figure 17

Maximum-likelihood phylogenetic tree for seven species/populations of Oryzias from Malili Lake system with two outgroups, based on 1,723 single nucleotide variant sites shared between ddRAD and whole-genome sequencing data. Sequence data sources: no asterisk Mandagi et al. (2021), single asterisk new sequence data, double asterisk Ansai et al. (2021)

Full size image

Oryzias marmoratus (not Aurich): ?Herder and Chapuis 2010 (partim: Lake Towuti); Mokodongan and Yamahira 2015 [partim: Lake Towuti (Lake Towuti B)]; Ansai et al. 2021 [partim: Lake Towuti (Towuti 1)]; Mandagi et al. 2021 (partim: Lake Towuti); Yamahira et al. 2023 [partim: Lake Towuti (Towuti lineage)].


Holotype. MZB 26367, male, 40.1 mm SL, Indonesia, Sulawesi Island, Sulawesi Selatan, Lake Towuti, south shore of Tanjung Timbala, western lake (2°41′24″S, 121°25′20″E), 29 Nov. 2018, coll. H. Kobayashi, R. Kakioka, S. Ansai, K.W.A. Masengi and K. Yamahira.

Paratypes. 39 specimens (27 males and 12 females), 29.1–40.2 mm SL. NSMT-P 143828, 143829, 143831, 3 males (34.1–37.2 mm SL), Tanjung Timbala, Lake Towuti (2°42′56″S, 121°26′50″E); following 37 paratypes collected from Indonesia, Sulawesi Island, Sulawesi Selatan, northern Lake Towuti, Lengkona (2°40′37″S, 121°4126″E), 10 May 2013, coll. K. Yamahira and D.F. Mokodongan: MZB 22750 (alcohol-fixed), male, 32.7 mm SL; MZB 26368–26383, 13 males and 3 females, 30.2–36.9 mm SL; WMSU 15–17, 2 males and 1 female, 30.7–37.9 mm SL; NSMT-P 143832–143840, 4 males and 5 females, 32.4–40.2 mm SL; KPM-NI 69933–69934, 1 male (36.7 mm SL) and 1 female (29.8 mm SL); URM-P 49723–49726, 2 males and 2 females, 29.1–38.4 mm SL; ZRC 62613, 1 male (34.6 mm SL) and 1 female (33.7 mm SL).

Diagnosis. Oryzias loxolepis sp. nov. is distinguished from O. marmoratus and O. profundicola by the following character combination: 11 abdominal vertebrae; 9–12 dorsal-fin rays; 22–27 anal-fin rays; 12 or 13 transverse scales; head length 24.4–27.2% SL (24.4–26.2% in male, 24.6–27.2% in female); eye diameter 10.8–12.5% SL; snout length 7.9–9.3% SL; body depth at point of anal-fin origin 26.1–30.7% SL (27.5–30.7% in male, 26.1–28.5% in female); body depth at point of dorsal-fin origin 24.1–28.4% SL (25.0–28.4% in male, 24.1–26.8% in female); caudal peduncle length 8.0–11.2% SL; terminal mouth; reduced nuchal concavity; scales diagonally aligned aslant to midlateral stripe; male with round-shaped dorsal fin without yellow margin.

Description. Detailed morphometric and meristic characters are provided in Tables 2 and 3. Maximum size 40.2 mm SL. Body compressed laterally, deep, with depth at points of anal-fin origin 26.1–30.7% [30.2%] SL, and dorsal-fin origin 24.1–28.4% [28.4%] SL (Figs. 11–13). Mouth terminal, subequal or with upper jaw projecting slightly beyond lower jaw. Upper and lower jaws with 1–3 irregular rows of conical teeth. Males with irregular external small conical teeth on upper and lower lips, especially at mouth corners. Head length 24.4–27.2% [25.4%] SL. Dorsal surface of head slightly convex prior to orbit; nape with or without slight concavity. Dorsal and ventral body profile gently arching from head to points of dorsal and anal-fin origin; without pronounced abdominal concavity between pelvic and anal fins. Snout length 7.9–9.3% [8.0%] SL, eyes large, diameter 10.8–12.5% [11.2%] SL, orbits slightly projecting beyond dorsal head surface. Infraorbital space narrow, 26.2–40.5% of eye diameter. Caudal peduncle short, length 8.0–11.2% [10.2%] SL; depth 10.6–12.4% [10.7%] SL. Male with short, slightly conical, tubular urogenital papilla; female with bilobed urogenital papilla.

Table 3 Counts and morphometrics for type series of Oryzias loxolepis sp. nov. (excepting MZB 22750) and select O. profundicola (NSMT-P 143841–143855, MZB 26399–26402) specimens, expressed as percentages of standard length
Full size table

Scales of moderate size, cycloid, deciduous and diagonally aligned to transverse axis (Fig. 8b); 31–33 [32] along lateral midline, and 12 or 13 [13] along transverse series extending backwards from points of dorsal and anal-fin origin. Dorsal-fin rays 9–12 [10], rounded and filamentous; rays 3–6 [5] longest in males. Anal-fin rays 22–27 [24]; anterior rays elongated and filamentous, rays 3–9 [5] longest in males; straight or slightly concave in females; without bony contact organs in males. Dorsal-fin origin over anal-fin ray 11 or 12 [12]. Pectoral-fin rays 10 or 11 [11]. Pelvic-fin rays 6, last ray connected to body via membrane along proximal half. Caudal fin truncate, with principal caudal-fin rays I,4/5,i. Procurrent-fin rays 4–6/5 or 6 [4/5]. Vertebrae 27–29 [28] (11 + 16–18 [17]); first pleural rib on vertebra 3.

Lateral line system sensory canals and rows from paratype NSMT-P 143831 (34.1 mm SL) illustrated in Fig. 14. Circumorbital sensory canal an open bony groove (AG, DG, and PG). Infraorbital and snout with small and numerous neuromast rows aligned between orbits and grooves; lower opercle with 2 and dorsal pectoral-fin base with 5 neuromast clusters. Anterior body from anal-fin origin with small transverse horizontally aligned rows (vlla and vla): ventrolateral line (vlla) with 9 rows; ventral line (vla) with 8 rows along throat and pelvic insertion, right and left rows coalescing as 4 clusters in center of inter-pelvic region to point of anal-fin origin. Posterior body with 4 short, major, horizontally aligned transverse rows (dll, mll, vllp and vlp): dorsolateral line (dll) with 14 rows, midlateral line (mll) with 8 rows, ventrolateral line (vllp) with 8 rows, and ventral line above anal-fin base (vlp) with 11 rows. Central predorsal region with dorsal line (dl) aligned backward from nape with 11 short rows. Each lateral row and cluster comprises 3–14 neuromast papillae. Central part of caudal fin with a neuromast cluster (CNC) slightly separated into two.

Color in life. Body and head grayish-brown, trunk with diffuse, randomly dispersed dark brown blotches (Figs. 11, 12). Dorsal half of head and body brown, opercle and belly brilliant silver and sky blue, with oval spot on opercle projection. Midlateral brilliant silver with pale yellow reticulation along scale edges. Eyes diffuse blue, infraocular white. Dorsal fin with distinct black distal margin, filaments in male hyaline. Male anal fin with narrow yellow distal margin (either fully, or with the posterior half changing to black under anesthesia, as in holotype); black in females; filamentous anal-fin rays in male hyaline or pale yellow along margin, with proximal dark purple blotch between median rays. Dorsal and ventral margin of caudal fin pale yellow; one or two longitudinal dark purple stripes on proximal half of membranes between median caudal-fin rays. Pelvic fin hyaline or pale yellow, with diffuse dark margin. Pectoral fin hyaline.

Color in alcohol. Ground color yellowish-gray. Irregular spots on trunk remain, supra-operclar-projection oval spot prominent (Fig. 13). Dorsal surface of head and dorsal and lateral surfaces of body with dense dark brown to black chromatophores. Melanophore row from dorsal surface of head to point of dorsal-fin origin diffuse; midlateral black line from head to base of caudal-fin, and across upper or lower part of lateral scales, diffuse (Fig. 8b). Pectoral fin and female pelvic fin hyaline; male pelvic fins with diffuse black margin. Dorsal and anal fins hyaline, excepting distinct black margin; rays of males dusky. With proximal row of black spots on membranes between anal-fin rays in males. Caudal-fin rays of males dusky, with dark line along proximal half of membrane between median caudal-fin rays. Urogenital papilla immaculate or with several melanophores in males; dense small, dark gray chromatophores in females.

Sexual dimorphism. Males and females of comparable maximum size, of different coloration above described. Males with filamentous dorsal and anal-fin rays. Large males with external conical teeth on each jaw. Females with bilobed genital papilla; males with short, conical, tubular papilla.

Distribution. Known only from Lake Towuti, southern Sulawesi Province, Indonesia.

Habitat. All specimens were collected and observed at day (none at night). In Tanjung Timbala (Fig. 9a) and Lengkona (Fig. 9b), Lake Towuti, this species occurs on gently sloping, shallow shores (1.5 m depth) with large boulders and submerged woods over sand and gravel, where it usually schools with O. profundicola and O. marmoratus (Fig. 15). Paratherina spp., Glossogobius sp., G. flavipinnis, G. intermedius, Mugilogobius rexi, M. latifrons, and Nomorhamphus megarrhamphus co-occur at each locality.

Etymology. The specific name “loxolepis” is derived from Ancient Greek “loxos”, meaning “oblique” and “lepis” meaning “scales”, referring to the diagonally aligned scales of this species.

Comparisons. Historically, O. loxolepis sp. nov. and O. marmoratus have been confused, but the former differs from the latter in having a shorter caudal peduncle (8.0–11.2% SL vs. 12.0–16.3%; Fig. 10a), more transverse scales (12 or 13 vs. 10 or 11; Table 2), in having a rounded posterior edge of the male dorsal fin (vs. sharp elongated), a terminal mouth (vs. superior), reduced nuchal concavity (vs. well-marked), and body scales aslant to the lateral line (vs. aligned in parallel).

Despite some overlap, O. loxolepis sp. nov. differs from O. marmoratus in having fewer abdominal vertebrae [11 vs. 11 or 12 (usually 12) in O. marmoratus; Table 2], caudal vertebrae (16–18 vs. 17–19), and total vertebrae (27–29 vs. 29–31) (Table 2). Oryzias loxolepis sp. nov. also tends to differ from O. marmoratus in having a deeper body at the dorsal-fin origin (24.1–28.4% SL vs. 20.2–25.3%; Fig. 10b), a larger eye diameter (10.8–12.5% SL vs. 9.1–11.2%; Fig. 16a), a narrower upper jaw (7.1–8.4% SL vs. 8.0–9.8%; Fig. 16b), and longer anal-fin base (37.6–43.4% SL vs. 32.0–37.9%; Fig. 10b).

Oryzias loxolepis sp. nov. shares abdominal vertebral number, and dorsal fin, terminal mouth, and nuchal concavity shapes with O. profundicola. However, O. loxolepis sp. nov. differs from O. profundicola in having a longer snout (7.9–9.3% SL vs. 6.8–7.8%; Fig. 16c); fewer lateral (31–33 vs. 33–35) and transverse (12–13 vs. 13–15) scales; caudal (16–18 vs. 17–19) and total (27–29 vs. 28–30) vertebrae; dorsal-fin (9–12 vs. 11–13) and anal-fin (22–27 vs. 25–29) rays, respectively. The two species also differ in fresh coloration (brilliant beige trunk and silver opercle in O. loxolepis sp. nov. vs. golden yellow trunk and golden opercle in O. profundicola) (Fig. 3a, b).

Male O. loxolepis sp. nov. and O. profundicola also differ in having a shorter head length (24.4–26.4% SL vs. 22.5–24.3%; Fig. 16d), larger eye diameter (10.8–12.5% SL vs. 9.1–10.3%; Fig. 16a), a shallower body depth at the anal-fin origin (27.5–30.7% SL vs. 30.5–34.4%; Fig. 10b) and dorsal-fin origins (25.0–28.4% SL vs. 28.2–33.3%; Fig. 10c), and a black margined dorsal fin (vs. remarkable yellow; Fig. 3a), respectively. Female O. loxolepis sp. nov. and O. profundicola differ in having a shallower body at points of anal-fin origin (26.1–28.5% SL vs. 30.4–35.2%; Fig. 10b) and dorsal-fin origin (24.1–26.8% SL vs. 28.0–32.2%; Fig. 10c), and they tend to have a larger eye diameter (10.8–12.4% SL vs. 9.7–11.0%; Fig. 16a), respectively.

From the Malili O. hadiatyae, O. loxolepis sp. nov. has shorter head length (24.4–27.2% SL vs. 28.9–32.0%; Fig. 16d), snout length (7.9–9.3% SL vs. 10.1–11.8%; Fig. 16c), postorbital length (6.5–8.7%SL vs. 11.0–12.6%; Fig. 16e), narrower upper jaw width (7.1–8.4% SL vs. 8.7–11.0%; Fig. 16b), greater body depth at the anal-fin origin (26.1–30.7% SL vs. 17.9–24.0%; Fig. 10b) and dorsal-origins (24.1–28.4% SL vs. 15.5–20.6%; Fig. 10c), in having shorter predorsal length (71.0–75.3% SL vs. 75.6–78.8%; Fig. 10f), preanal length (53.4–59.2% SL vs. 60.3–63.5%; Fig. 10d), and caudal peduncle length (8.0–11.2% SL vs. 11.8–15.7%; Fig. 10a), and in having a longer anal-fin base (32.0–37.9% SL vs. 24.4–28.9%; Fig. 10e), respectively. From the Malili O. matanensis, O. loxolepis sp. nov. has more abdominal vertebrae (11 vs. 12; Table 2), a greater body depth at the dorsal-fin origin (24.1–28.4% SL vs. 18.6–23.1%; Fig. 10c), shorter predorsal length (71.0–75.3% SL vs. 76.8–79.9%; Fig. 10f) and caudal peduncle length (8.0–11.2% SL vs. 11.8–15.5%; Fig. 10a), and longer dorsal-fin (11.1–14.9% SL vs. 7.9–10.2%; Fig. 16f) and anal-fin bases (37.6–43.4% SL vs. 30.6–34.4%; Fig. 10e), respectively.

Phylogeny

The ML phylogeny based on 1,723 genome-wide SNVs revealed six major clades within the Malili Lake system, each with > 79% ML bootstrap support (Fig. 17). The holotype and two paratypes of Oryzias loxolepis sp. nov. formed a clade with “O. marmoratus Towuti” individuals of Mandagi et al. (2021), indicating that “O. marmoratus Towuti” of Mandagi et al. (2021) represents O. loxolepis sp. nov. The O. loxolepis clade was sister to the O. profundicola clade. In contrast, the “O. marmoratus (Towuti 2)” individual of Ansai et al. (2021) clustered with “O. marmoratus” individuals from Lake Lantoa. Individuals of “O. marmoratus” from Lake Mahalona formed another clade which was separated from both the O. loxolepis sp. nov. and Towuti–Lantoa “O. marmoratus” clades.

Discussion

Validity of Oryzias loxolepis as a phylogenetic and biological species. We report O. loxolepis sp. nov. to differ morphologically from O. profundicola. Although these two species cannot be presently distinguished using mitochondrial DNA and microsatellite markers (Mandagi et al. 2021; Yamahira et al. 2023), this may reflect incomplete lineage sorting, with O. loxolepis sp. nov. and O. profundicola being too young for many of the loci (including mitochondrial genes and microsatellite loci) to have been fixed (see also Mandagi et al. 2021). Because our genome-wide SNV analysis detected several sites which differ or are almost fixed between O. loxolepis sp. nov. and O. profundicola (see ESM 1), they are phylogenetically clearly separated from each other in our tree (Fig. 2). Moreover, because O. loxolepis sp. nov. and O. profundicola (and O. marmoratus) usually swim together in shallow rocky lake habitat (Fig. 15), their phylogenetic distinction in the absence of habitat separation suggests the existence of strong reproductive isolation. We therefore conclude that O. loxolepis sp. nov. and O. profundicola represent both valid phylogenetic and biological species.

Taxonomic status of Oryzias marmoratus. Both O. marmoratus and O. loxolepis sp. nov. differ morphologically and phylogenetically, even in Lake Towuti where they both occur, reconfirming that O. marmoratus occurs in Lakes Lantoa, Mahalona, and Towuti. Diagnostic traits of specimens presently re-identified as O. marmoratus, including vouchers used in phylogenetic analysis, are generally consistent with those of the neotype of Aplocheilus marmoratus, although some between-lake variation exists. The Mahalona population has unique phenotypes, most of which are shared with O. matanensis, including a broad vertical midlateral spot on the male. Based on admixture analyses using genome-wide SNPs, Mandagi et al. (2021) demonstrated that the Mahalona population was of hybrid O. marmoratus × O. matanensis origin, which may explain the phylogenetic differences between Mahalona and other O. marmoratus, and possibly this population’s unique phenotypes. We only tentatively treat all three O. marmoratus allopatric populations as one species.

Review of O. marmoratus from Lake Towuti in previous genetic studies. Because O. marmoratus and O. loxolepis sp. nov. in Lake Towuti have been historically confused (Table 4), we re-identify those specimens formerly attributed to O. marmoratus from this lake. Mokodongan and Yamahira (2015) reported two individuals as “O. marmoratus Towuti A” and “O. marmoratus Towuti B” with distinct mitochondrial haplotypes; the former forming a clade with O. marmoratus from Lakes Lantoa and Mahalona, and the latter monophyletic with O. profundicola. After re-examination of these individuals, we re-identify the former as O. marmoratus (MZB 22749) and the latter as O. loxolepis sp. nov. (paratype: MZB 22750). We also examined the 40 “O. marmoratus Towuti” specimens used for microsatellite analyses by Yamahira et al. (2023) and identify the “Lantoa-type” individual as O. marmoratus (NSMT-P 143827) and others as O. loxolepis sp. nov. (some of our paratypes). All 10 “O. marmoratus Towuti” of Mandagi et al. (2021) (and reused in Utama et al. 2022), which formed clades with O. profundicola in both earlier works and in our own phylogenies using genome-wide SNVs analyses, are referred to O. loxolepis sp. nov., based on images taken soon after field collection (ESM S2). A phylogeny based on more than 10,000 nuclear genes (Ansai et al. 2021) also reported “O. marmoratus (Towuti 1)” and “O. marmoratus (Towuti 2)” to be phylogenetically distant, and to form clades with O. profundicola and O. marmoratus in Lake Lantoa, respectively. The individual referred to as “O. marmoratus (Towuti 1)” was used also by Mandagi et al. (2021) (tag ID “O_marmoratus_Tow-F03”; see Mandagi et al. 2021: supplementary table S2), indicating that it was O. loxolepis sp. nov. (ESM S2: F03). The “O. marmoratus (Towuti 2)” individual referred to as MZB 22749 in Mokodongan and Yamahira (2015) is O. marmoratus. MZB 22749 was also referred to as “O. marmoratus” in Yamahira et al. (2021) and Seleit et al. (2021).

Table 4 Identification of previous records of species of Oryzias from Lake Towuti, Mahalona, and Matano
Full size table

Although we have not examined those specimens referred to “O. marmoratus” by Herder et al. (2012), assuming that O. loxolepis sp. nov. forms a clade with O. profundicola, they probably include representatives of both O. loxolepis sp. nov. and O. marmoratus. The “O. marmoratus” individual of Takehana et al. (2005) could not be identified in the same way because of poor taxon sampling. However, we confirm that this individual is O. marmoratus by directly comparing its mitochondrial sequence with that of MZB 22749 (Y. Takehana, pers. comm.).

Conclusions. As above, O. marmoratus and O. loxolepis sp. nov. have historically been confused under the name “O. marmoratus”. Because these two species, and O. profundicola, are sympatric in Lake Towuti, ricefishes in the Malili Lake system may have experienced a more complex evolutionary history than is currently recognized. Future taxonomic and evolutionary biological studies for the genus Oryzias in the Malili Lake system are needed to improve our understanding of how biodiversity in this region has developed.

Revised key of Oryzias species in the Malili Lake system

  1. 1a

    Lateral scales 42–47; body with dense vertical bands or blotches … O. matanensis (Lake Matano)

  2. 1b

    Lateral scales 27–35; body with diffuse lateral blotches or spots … 2

  3. 2a

    Preanal length long, 60.3–63.5% SL; snout long, length 10.1–11.8% SL … O. hadiatyae (Lake Masapi)

  4. 2b

    Preanal length short, 53.4–59.1% SL; snout short, length 6.8–9.5% SL … 3

  5. 3a

    Caudal peduncle long, length 12.0–16.3% SL; transverse scales 10 or 11 … O. marmoratus (Lakes Towuti, Mahalona, and Lantoa)

  6. 3b

    Caudal peduncle short, length 8.0–11.7% SL; transverse scales 12–15 … 4

  7. 4a

    Snout length short, 6.8–7.8% SL; female body deep, 30.4–35.2% SL; male dorsal fin with yellow margin … O. profundicola (Lake Towuti)

  8. 4b

    Snout relatively long, length 7.0–9.3% SL; female body slender, depth 26.1–28.5% SL; male dorsal fin with black margin … O. loxolepis sp. nov. (Lake Towuti)

Comparative materials examined

INDONESIA, Sulawesi Selatan (Malili Lakes): Oryzias profundicola: holotype: MZB 5868 (formerly ZSM/LIPI 12), male, 37.2 mm SL, Lake Towuti, Tanjung Posombuwang; 35 nontypes: NSMT-P 143841–143845, 3 males and 2 females, 39.5–47.5 mm SL, Lake Towuti, Tanjung Timbala; MZB 26384–26402, 19 males, 35.9–44.1 mm SL, NSMT-P 143839, 143846–143855, 5 males and 6 females, 35.2–45.8 mm SL, Lake Towuti, Lengkona; Oryzias hadiatyae: holotype: MZB 18491, male, 35.6 mm SL; 21 nontypes: MZB 26403–26413, 5 males and 6 females, 31.2–37.5 mm SL, NSMT-P 143856–143865, 5 males and 5 females, 35.4–47.2 mm SL, Lake Masapi; Oryzias matanensis: neotype: ZRC 38450 (formerly ZSM 27368), male, 45.0 mm SL, Lake Matano; 21 nontypes: MZB 26414–26424, 5 males and 6 females, 33.5–40.6 mm SL, NSMT-P 143866–143875, 5 males and 5 females, 33.7–42.3 mm SL, Lake Matano. Sulawesi Selatan (others): Oryzias celebensis: lectotype of Haplochilus celebensis Weber 1894: ZMA 112.585, female, 28.5 mm SL, Maros River; 4 nontypes: NSMT-P 143876, 1 male, 30.7 mm SL, River in Maros; NSMT-P 143877–143879, 2 males and 1 female, 27.1–30.4 mm SL, Walanae River. Oryzias dopingdopingensis: holotype: MZB 23873, male, 32.7 mm SL; 9 paratypes: MZB 23874–23882, 4 males and 5 females, 31.8–34.7 mm SL, District Luwu Timur, Doping-doping River. Oryzias eversi: holotype: MZB 20780, male, 35.8 mm SL; 1 paratype: MZB 20781, female, 28.0 mm SL, Tana Toraja, stream near Tilanga. Sulawesi Tengah: Oryzias bonneorum: holotype: MZB 15499, male, 52.7 mm SL; 3 paratypes: ZMA 123.863, 2 males and 1 female, 38.2–44.9 mm SL, Lake Lindu; Oryzias kalimpaaensis: holotype: MZB 26462, male, 41.9 mm SL, Lake Kalimpa’a; Oryzias nebulosus: 10 nontypes: NSMT-P 143885–143894, 5 males and 5 females, 27.7–39.9 mm SL, Dumulanga, Southwestern coast of Lake Poso; Oryzias nigrimas: holotype: MZB 5859 (formerly ZSM/LIPI 1), male, 42.2 mm SL, Lake Poso; 2 nontypes: NSMT-P 143880–143881, 1 male (38.8 mm) and 1 female (52.4 mm), Lake Poso, Tentena; Oryzias orthognathus: holotype: MZB 5870 (formerly ZSM/LIPI 3), female, 49.2 mm SL, Lake Poso; 3 nontypes: NSMT-P 143882–143884, 2 males and 1 female, 46.4–47.9 mm SL, Lake Poso, Tentena; Oryzias sarasinorum: 6 syntypes of Haplochilus sarasinorum Popta 1905: RMNH 7664, 3 males, 51.6–57 mm SL, ZMA 100.648, male, 54.3 mm SL, BMNH 1914.2.13.26–27, 2 males (54.0, 54.3 mm SL), Lake Lindu; Oryzias soerotoi: holotype: MZB 21377, male, 27.1 mm SL; 6 paratypes: NSMT-P 114680–114685, 1 male and 5 females, 25.6–31.0 mm SL, Lake Tiu. Sulawesi Tenggara: Oryzias asinua: 2 paratypes: NSMT-P 111645, 1 male (20.0 mm SL) and 1 female (20.2 mm SL), District Asinua; Oryzias wolasi: 4 paratypes: NSMT-P 111646, 3 males and 1 female, 17.5–26.0 mm SL, District Wolasi; Oryzias woworae: 3 paratypes: BMNH 2009.5.27.1–3, 2 males and 1 juvenile, 19.8–22.4 mm SL, Muna Island, District Parigi, Fotuno Fountain. Other congeners: Oryzias javanicus (Bleeker 1854): 1 syntype of Aplocheilus javanicus Bleeker 1854: BMNH 1866.5.2.101, 24.8 mm SL, Java Island, Perdana, Panimbang River; 4 nontypes: NSMT-P 67703, 3 males, 15.7–23.5 mm SL, 3 November 2003; NSMT-P 67730, 1 male, 19.8 mm SL, Bali Island; Oryzias timorensis: 6 paralectotypes of Aplocheilus timorensis Weber and de Beaufort 1922: ZMA 120.761, 2 males and 4 females, 16.6–30.3 mm SL, Timor Island, Mota Talau.