This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Molecular Phylogenetics and Evolution 50 (2009) 190–196 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev Phylogeography of the genus Epiplatys (Aplocheiloidea: Cyprinodontiformes) Glen E. Collier a,*, William J. Murphy b, Michael Espinoza a a b Department of Biological Science, The University of Tulsa, 800 S. Tucker Avenue, Tulsa, OK 74104-3189, USA Department of Veterinary Integrative Biosciences, College of Veterinary Medicine, Texas A&M University, Mail Stop 4458, College Station, TX 77843-4458, USA a r t i c l e i n f o Article history: Received 22 September 2008 Revised 7 October 2008 Accepted 7 October 2008 Available online 19 October 2008 Keywords: African biogeography Aplocheiloid Cyprinodont Mitochondrial DNA Phylogeny Epiplatys a b s t r a c t There are six major genera of aplocheiloid killifishes endemic to West Africa. Five of these are largely restricted to the two major blocks of rainforest. Two are found within the Eastern rainforest block (Nigeria to the central Congo) while three are found within the Western rainforest block (Sierra Leone to Togo). The sixth genus (Epiplatys) has a range that exceeds that of the combined area of the other five genera. Phylogenetically this genus is related to the Western rainforest taxa. Phylogeographic analysis of this genus suggests that it escaped the confines of the Western block by first expanding into lowland habitats exposed after a sea level drop and then dispersed along coastal habitats to the east. One lineage managed to penetrate the interior of the Eastern rainforest block and one derivative of this lineage penetrated the Congo basin. A second route out of the Western rainforest block was to the north through what is now savannah habitat. The greater phylogeographic range of Epiplatys is hypothesized to be due to retention of ancestral morphology related to a greater adaptability compared to the other five genera. Ó 2008 Elsevier Inc. All rights reserved. 1. Introduction The Order Cyprinodontiformes contains almost 800 species of freshwater fish distributed circumtropically (west of Wallace’s line) on all continents except Australia. These relatively small fish are commonly known as killifish, toothcarps or topminnows. Parenti (1981) divided the order into two suborders, the Cyprinodontoidei and the Aplocheiloidei. The aploicheiloids are found primarily in freshwater habitats throughout the Neotropics, midlower latitudinal Africa, Madagascar, the Seychelles, and India extending throughout the Indo-Malaysian archipelago. The major west African genera of the suborder Aplocheiloidei currently recognized are Aphyosemion, Fundulopanchax, Callopanchax, Scriptaphyosemion, Archiaphyosemion and Epiplatys (Lazara, 2000). Of these, the first five occupy rather discrete geographic areas. Aphyosemion is distributed from Nigeria southeast to Cameroon and from there south through Gabon and east to the central Congo basin. Fundulopanchax is distributed throughout Nigeria and coastal lowlands to the west and southeast of Nigeria. Callopanchax is restricted to the coastal lowlands of far Western Africa while Scriptaphyosemion and Archiaphyosemion are restricted to the interior area of the same far Western rainforest. Epiplatys is unique in that its distribution overlaps large portions of the areas occupied by each of the other five genera (Scheel, 1990; Wildekamp, 1996) (Fig. 1). Molecular phylogenies have provided strong support for the monophyly of each of these genera (Murphy and Collier 1997, * Corresponding author. Fax: +1 918 631 2762. E-mail address: glen-collier@utulsa.edu (G.E. Collier). 1055-7903/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2008.10.006 1999; Murphy et al., 1999) while clarifying the generic placement of some individual taxa. These results reveal that the relationships of the major groups within the suborder Aplocheiloidei as a whole are consistent with the vicariant breakup of Gondwanaland. The relationships among the six west African genera are consistent with a major vicariant event affecting freshwater fishes of west Africa. For a substantial period of the late Cretaceous and early Tertiary (95-53MYA) an epicontinental sea separated emergent west Africa from emergent central Africa. When this epicontinental sea regressed it left behind what has been called the Dahomy gap, a swath of savannah that separates the extant Eastern and Western block of African rainforest. The two genera largely restricted to the area east of the gap, Aphyosemion and Fundulopanchax, are sister groups while the taxa restricted to the west of the gap, Callopanchax, Scriptaphyosemion and Archiaphyosemion, are monophyletic. In turn these two clades are reciprocally monophyletic. It came as something of a surprise to discover that Epiplatys is sister to the Western clade of three closely related genera. The etymology of the generic name Epiplatys is ‘‘with flat upper back”. This is in reference to flat dorsal surface of the anterior half of the body. They are pronounced surface-oriented fish and their mouths are also directed upward. These two physical characteristics, and their underlying osteological basis, caused some previous workers to classify these fishes with similarly shaped species of Aplocheilus. This is difficult to accept on biogeographic grounds as the latter group is restricted to India, Sri Lanka and adjacent areas of southeast Asia while the former is restricted to west Africa. These physical features may represent convergence to the same surface feeding ecology or to the retention of an ancestral form Author's personal copy G.E. Collier et al. / Molecular Phylogenetics and Evolution 50 (2009) 190–196 191 Fig. 1. Distribution of the major aplocheiloid genera of West Africa. (A) The approximate ranges of the genera Aphyosemion, Fundulopanchax, Scriptaphyosemion, Archiaphyosemion and Callopanchax. (B) The approximate range of the genus Epiplatys. The approximate area of origin of each taxa sampled in this survey are indicated the placement of the three letter abbreviations for each species. The names are color-coded to correspond to the major divisions of the phylogenetic tree (Fig. 2) for the group. The generic ranges indicated are the cumulative range of the individual species as reported by Wildekamp (1993, 1996). The number of species in each genus (Aphyosemion [n = 80], Fundulopanchax [n = 19 ], Callopanchax [n = 3], Archiaphyosemion [n = 5 ]. Scriptaphyosemion [n = 13 ], Epiplatys [n = 35 ]) is taken from Lazara (2000). Placement of specific taxa in genera has been modified to be consistent with generic limits determined by molecular phylogenetic analyses (Murphy and Collier, 1997; Murphy et al., 1999; Murphy and Collier, 1999) (For interpretation of color mentioned in this figure, the reader is referred to the web version of this article.). abandoned by other related groups as they evolved to occupy slightly different or more specialized niches. Molecular data do not support a close relationship between Epiplatys and Aplocheilus (Murphy and Collier, 1997). A more recent phylogenetic analysis of the external anatomy and osteology of eight species of Epiplatys, two species assigned to two monophyletic genera closely related to Epiplatys and two outgroup taxa (Aarn and Shepherd, 2001) failed to resolve relationships among species of Epiplatys. This distribution of Epiplatys raises specific questions about their biogeography. If they are indeed the sister group to Callopanchax, Scriptaphyosemion and Archiaphysemion, then they too originated in the Western block of rainforest and their dispersal into regions east of this rainforest region must be a derived event. More specifically, we hypothesize that the Eastern taxa are derived and that the basal clades of Epiplatys are indeed rooted in the west. We predict that this will be evident in the phylogeographic pattern. 2. Materials and methods 2.1. Phylogenetic methodology Twenty-two aquarium stocks representing twenty one taxa (Table 1) were surveyed for DNA sequence variation for the same three segments of the mitochondrial genome (12S, 16S and cytochrome b) used in previous studies to assess relationships among the aplochieloids (Murphy and Collier, 1997, 1999; Murphy et al., 1999). Sequence alignment was performed with CLUSTAL X (Thompson et al., 1997). Mitochondrial rRNA segments were modified manually, using previous published alignments as guides (Murphy and Collier, 1999). The total analyzed alignment was 1145 bp in length after conservatively excluding 96 bp of sequence from the rRNA segments due to ambiguous homology. Phylogenetic analyses were performed in PAUP* 4.0b10 (Swofford, 2002) and Mr. Bayes (Ronquist and Huelsenbeck, 2003). Maximum parsimony (MP) and minimum evolution (ME) analyses were performed using 100 random addition replicates and TBR branch swapping; Starting trees were obtained by neighbor-joining for the maximum likelihood (ML) search due to computational burden. Nonparametric bootstrap analyses were performed using 100 heuristic replicates with TBR branch swapping and 100 random taxon additions. Settings for the model of DNA sequence evolution were estimated initially using the Akaiki Information Criterion implemented in Modeltest (Posada and Crandall 1998) and then optimized using multiple heuristic ML searches in PAUP until parameter values stabilized (Swofford, 2002). Bayesian phylogenetic analyses will be performed using the program MrBayes 3.0b4 (Ronquist and Huelsenbeck, 2003). Four simultaneous Markov chains were Author's personal copy 192 G.E. Collier et al. / Molecular Phylogenetics and Evolution 50 (2009) 190–196 Table 1 Taxa surveyed. Species Populationa E. E. E. E. E. E. E. E. E. E. E. E. E. E. E. CI, Guinea CI, Sierra Leone RL99, Liberia CI, Guinea Sangareyo, Guinea CI, Ghana CI, Malebo pool, Zaire Monroviae, Liberia Monrovia, Liberia Angona, Ghana CI, Togo Elan, Cameroon CI, Nigeria CI, Zaire GJS00/2, Massana Gabon GBN88/5, Gabon Sindara, Gabon Sangmelina, Cameroon Bangui, Republic of Central Africa N’sele, Zaire CI, Zaire guineensis fasciolatus roloffi lamottei hildegardae bifasciatus spilargyreius annulatus dageti chaperi togolensis infrafasciatus grahami singa ansorgii E. huberi E. multifasciatusd E. sangmelinensis E. mesogramma E. chevalieri E. duboisie nb 20 Geographyc 18 24 20 17 25 25 25 24 24 23–24 21 24 West West West West West Savannah Savannah West (coastal) West (coastal) West (coastal) East (coastal) East (coastal) East (coastal) East (coastal) East containing E. annulatus, E. chaperi, E. dageti, E. infrafasciatus, E. togolensis, E. grahami, E. singa, E. mesogramma, E. huberi, and E. multifasciatus and another containing E. lamottei, E. bifasciatus, E. spilargyreius, E. duboisi, E. roloffi, E. fasciolatus and E. guineensis. Minor differences in the MP results were observed in the first clade, with differences found primarily between the relationship of the E. singa + E. grahami and E. chaperi + E. dageti + E. annulatus clades relative to the other Epiplatys species. Bootstrap support values and Bayesian posterior probabilities were largely comparable for most of the remaining nodes in the topology. MP bootstrap support values were generally lower for deeper, short-internodes than ML, ME and BAYES results, most likely due to the difficulties of parsimony under the phylogenetic conditions observed: short-internodes and long branches. 4. Discussion 24 23 24 24 24 East East East East East East a CI stands for commercial import. The remaining names are references to collecting sites. The alpha-numeric collection codes can be found in Langton (2003). Frozen or ethanol preserved vouchers are retained in the collection at the University of Tulsa. b n is the haploid number of chromosomes. c The taxa are arranged in approximate geographic order from west to east. ‘‘East” and ‘‘West” simply refers to whether the taxa comes from an area east or west of the ‘‘Dahomy gap”. Those taxa restricted to lowland coastal rivers and swamps are indicated by ‘‘coastal”. ‘‘Savannah” refers to taxa widely distributed through savannah habitat to the north of both rainforest blocks and the area between them. d The representative of multifasciatus was collected near Sindara, Gabon and has circulated among aquarists as ‘‘boulengeri” (Heller, 1997). Wildekamp (1996) considers boulengeri to be synonymous with multifasciatus. We follow that convention here. However it must be noted that this represents a far Western population of this wide ranging ‘‘species”. e E. duboisi is often considered the sole representative of the monotypic genus Aphyoplatys. For reasons outlined in the discussion, this no longer tenable. We refer this species to the genus Epiplatys. run initially for 500,000 generations, sampling trees every 100 generations, employing a burn-in setting of 50,000 generations (based on evaluation of stationarity of the burn-in state graphically). Posterior probabilities for phylogenetic branches and parameters of the model of sequence evolution for at least two independent runs were examined for concordance. 2.2. Karyotypes The chromosome numbers published by Scheel (1972, 1990) were used in Table 1 except for E. lamotteii. Chromosome numbers were determined from slides prepared from gill epithelia by the method of Kligerman and Bloom (1977). 3. Results The phylogenetic relationships within Epiplatys (and including Aphyoplatys) (Fig. 2) were identical whether analyzed by ME, ML or Bayesian methodologies. The figure shown is the ML tree, based on the GTR + C + I model of sequence evolution selected in Modeltest. Model parameters used in the ML and ME (with ML distances) analyses were: rate categories: 2.863403, 10.032100, 3.957271, 0.579594, 30.853946, 1.000000; nucleotide frequencies: A = 0.31121 C = 0.21765 G = 0.18790 T = 0.28324; proportion of invariant sites = 0.47337; shape parameter (alpha) = 0.738516. These different analyses consistently reveal two clades of Epiplatys, one The taxa included in this survey include only slightly more than half of the described species, but do represent a broad sampling of the diversity represented by the genus Epiplatys. While addition of data from species not included might change the details of relationships, the broad outlines seem clear from these data. 4.1. Nomenclatural issues Two taxa included in this survey are morphologically distinct, diminutive species. Aphyoplatys duboisi is currently regarded as representing a monotypic genus. E. annulatus was previously placed in a monotypic genus Pseudepiplatys. The molecular data clearly places both of these taxa as members of a monophyletic group inclusively assigned to the genus Epiplatys. Recognition of either E. duboisi or E. annulatus as members of monotypic genera renders the remaining assemblage of species paraphyletic. The goal of rendering generic assignments compatible with phylogenetic reconstructions argues against the continued use of either monotypic designation for these taxa. 4.2. Western origin of the genus The extant distributions of the taxa in the two main clades are consistent with the interpretation that the group originated in the Western block of rainforest. One clade (the ‘‘Western/savanna clade”) contains species still restricted to the Western rainforest block (E. lamottei, E. roloffi, E. fasciolatus, E. hildegardae and E. guineensis). Two species (E. bifasciatus and E. spilargyreius) that are broadly distributed in savannah habitats near the rainforest block and to the north, northwest and northeast of it are derived from two early divergent lineages of this clade. The last member of this clade is an enigmatic, diminutive species (E. duboisi) from the central Congo region. The other clade (the ‘‘coastal clade”) contains taxa from west of the Dahomy gap, but these are from lowland swamps and streams (E. annulatus, E. dageti and E. chaperi). Other taxa in this clade are also from coastal habitats extending eastward and then south (E. togolensis, E. infrafasciatus, E. grahami and E. singa). The six taxa sampled (E. ansorgi, E. chevalieri, E. huberi. E. mesogramma, E. multifasciatus and E. sangmelinensis) that have penetrated the interior of the Eastern rainforest block are relatively recently derived members of a single lineage within this clade. The derived nature of the Eastern taxa and the fact that they are nested in clades with other taxa restricted to the west is consistent with a Western origin of the entire genus Epiplatys. Author's personal copy 193 G.E. Collier et al. / Molecular Phylogenetics and Evolution 50 (2009) 190–196 multifasciatus 99/99/88 1.00 100/100/100 1.00 huberi ansorgi <50/78/90 mesogramma 69/57/80 1.00 95/99/96 0.99 <50/<50/71 0.99 chevalieri 1.00 sangmeliensis togolensis 99/100/97 <50/67/<50 0.91 1.00 infrafasciatus grahami 100/100/100 1.00 51/66/89 singa 1.00 annulatus 100/100/100 1.00 68/74/84 annulatus 0.98 dageti <50/53/70 0.97 chaperi 100/100/100 99/100/99 1.00 74/78/84 82/97/94 64/84/86 <50/57/50 0.56 62/63/71 1.00 1.00 1.00 guineensis hildegardae fasciolatus roloffi 1.00 spilargyreius bifasciatus 0.97 100/100/100 1.00 76/85/91 1.00 duboisi duboisi lamottei 96/100/99 1.00 Callopanchax occidentalis Scriptaphyosmeion geryi Archiaphyosemion maeseni 0.01 substitutions/site Fig. 2. Phylogenetic relationships among Epiplatys species. The ML topology is shown, and is identical to the results of the minimum evolution and Bayesian analyses. Bootstrap support (in the order: MP, ME, ML) is shown adjacent to each internode, while Bayesian posterior probabilities are shown subtending the bootstrap values. The maximum parsimony tree differed at the nodes with bootstrap support of <50. Three species from the closely related genera Archiaphyosemion (A. maeseni), Callopanchax (C. occidentalis) and Scriptaphyosemion (S. geryi) were used to root the tree. Names are color-coded to facilitate comparison to geographic locations (Fig. 1b) (For interpretation of color mentioned in this figure, the reader is referred to the web version of this article.). 4.3. Color patterns These two clades also differ in fundamental features of coloration and fin morphology. Most members of the ‘‘coastal clade” are characterized by broad dark vertical cross bars on the sides. Taxa that lack them as adults (E. singa) do have them as juveniles. These taxa are also generally characterized by asymmetrically shaped caudal fins. The lower rays of the caudal are slightly elongated and look like a short ‘‘sword”. This is very pronounced in E. dageti and E. singa, but is evident in other taxa as well. However, this trait is variable and is hardly evi- dent in some populations. For example, this trait is hardly discernible in E. sexfasciatus and its near relatives E. infrafasciatus and E. togolensis. Members of the ‘‘Western/savannah” clade seldom have dark bars on their sides and when bars are present (i.e. E. spilargyreius and E. guineensis) they are thin and oblique rather than broad and vertical. The caudal fins are symmetrical in all taxa of this clade with no extension of the lower rays. In several species (i.e. E. roloffi and E. lamottei) submarginal bands of color are pronounced on both upper and lower caudal margins. Author's personal copy 194 G.E. Collier et al. / Molecular Phylogenetics and Evolution 50 (2009) 190–196 in the east. It is likely that both occupied even larger ranges prior to the southern expansion of the Sahara. The representatives of these taxa in this study came from two Western populations. A more comprehensive sampling of populations across this range may reveal substantial genetic variation within each of these taxa. However, if the taxa remain monophyletic relative to other named taxa in Epiplatys, the generalizations below should still hold. The nominal taxa from the Congo basin (i.e. E. multifasciatus and E. chevalieri) are also reported from an extensive area. Whether these extensive ranges are due to their recent expansion into this relatively new habitat or that these poorly studied nominal taxa are composites of multiple cryptic species will require further study. Among the better known taxa, the juxtaposition of ranges and relationships is striking for even this limited selection of taxa. For example, E. grahami and E. singa appear to be sister taxa and occupy adjacent coastal areas. E. grahami is found in the coastal lowlands from Benin, Nigeria to Cameroon. It is replaced to the south along the coastal lowlands of Gabon, Congo and Western Zaire by E. singa. The two species closely related to E. sexfasciatus are sister groups and occupy adjacent ranges along the coastal regions of Southeastern Ghana to Nigeria and southeast Nigeria and Cameroon. [These species, E. togolensis and E. infrafasciatus, have been only recently elevated to specific level by Wildekamp (1997). Prior to this they were considered subspecies of E. sexfasciatus]. E. roloffi is restricted to Northwestern Liberia and Southeastern Guinea. The sister taxa E. fasciolatus occurs to the south and the west, while the sister taxa to E. fasciolatus, E. guineensis replaces it to the far west. Finally, the data only weakly supports a branching order for E. chaperi, E. dageti and E. annulatus, but these taxa do replace one another in coastal lowland areas moving from Ghana in the east to Guinea in the west. 4.4. Chromosomes The aplocheiloids are cytogenetically exceptional in that there has been extensive chromosomal rearrangement and reduction of haploid chromosome number as a consequence of Robertsonian translocations. This has been quite pronounced in the genus Aphyosemion where the haploid chromosomes number ranges from 20 to 9. While there is some karyotypic variability in Epiplatys, it is rather modest in comparison. The haploid numbers reported by Scheel (1972) range from 25 to 17. E. lamottei was found to have n = 24 in this study. The haploid numbers are listed in Table 1. Three taxa (E. annulatus, E. dageti and E. chaperi) are unusual in that n = 25. It is unusual for the haploid number to exceed the basal number of the group (i.e. 24). Thus this could be considered a derived character uniting these taxa. The molecular data independently support this inference. The greatest reductions are in the two savannah species E. spilargyreius and E. bifasciatus and populations of E. fasciolatus. The two savannah species occupy immense ranges and only a few populations have been examined cytologically, so the range of karyotypic variability in these taxa is not yet fully known. The populations of E. fasciolatus vary with n = 18–20. 4.5. Ranges As the taxa within Epiplatys have become better known, apparent ranges of distribution have become better delineated (Wildekamp, 1996). Most taxa have rather discrete ranges on the order of 1000 km2 or less. The major exceptions to this rule are the two savannah species, E. bfiasciatus and E. spilargyreius, and some taxa from the Congo basin. The savannah species occur over an immense area from Gambia and Senegal in the west to Sudan New wave of dispersal into savannah habitat. As sea level recedes, there is dispersal into new lowland habitats. Lowland populations diverge. 1 2 At least two independent savannah species arise from this dispersal episode. Additional divergence within western rainforest area. bif and spi bif and spi One lineage enters what will become the Congo basin. 3 dub 4 As Congo basin develops, there is dispersal into it from coastal populations. Fig. 3. Hypothesized biogeographic events that led to this distribution of taxa of the genus Epiplatys. Author's personal copy G.E. Collier et al. / Molecular Phylogenetics and Evolution 50 (2009) 190–196 4.6. Biogeographic hypothesis The observations outlined above lead to a very specific biogeographic hypothesis to explain this distribution of taxa in the genus Epiplatys (Fig. 3). The ancestral population that gave rise to all modern Epiplatys lived in the upland Western rainforest block, probably close to the area of southeast Guinea and northwest Liberia now occupied by E. lamottei and E. roloffi. As the sea level dropped in the early Eocene, new coastal lowlands were exposed and eventually the entire area covered by the epicontinental seas was exposed. Some descendants of the ancestral population moved into these new lowlands. The early immigrants either continued dispersing eastward along the coast and/or were displaced eastward as succeeding waves of immigrants entered the lowlands. This inference is supported by the observation that the extant species of the coastal clade located farthest to the east and south, E. grahami and E. singa, are sister groups and diverged earliest. Dispersal to the west occurred later and the species found farthest to the west, E. annulatus, is the terminal member of this clade. This dispersal into the newly exposed lowlands was the defining event establishing the two major clades of Epiplatys. Later descendants of the early Western rainforest ancestors dispersed to the north, northwest and east into the surrounding savannah. This occurred at least twice and these lineages survive today as the widely distributed species E. bifasciatus and E. spilargyreius. A third extant species, E. duboisi, is currently restricted to the area of Eastern Congo and adjacent Western Zaire, but diverged earlier than the previous two lineages. Presumably the ancestors of the E. duboisi followed a similar savannah pathway to the central Congo area. The subsequent invaders of the savannah presumably displaced remnants of this lineage in the savannah. Approximately the same time as the savannah dispersal episode described above, one of the coastal clade ancestors succeeded in penetrating the interior of the Eastern rainforest block. All five taxa sampled from these areas in this study, E. ansorgii, E. chevalieri, E. huberi, E. multifasciatus (‘‘E. boulengeri”), and E. mesogramma form a monophyletic group. This clade is subdivided into two groups, one consisting of the taxa found just inland from the coastal lowlands (E. ansorgii, E. huberi and E. multifasciatus and a second consisting of the two taxa from the Congo basin proper (E. chevalieri and E. mesogramma). The last dispersal episode is the limited dispersal and proliferation of taxa in the Western rainforest block. The earliest divergence of this episode resulted in the extant taxa E. roloffi. The sister lineage to this dispersed to the south and west and gave rise to E. fasciolatus and a further westerly distributed taxa E. guineensis. 4.7. Tests of the hypothesis Addition of sequence data from taxa not yet sampled will test the phylogenetic hypothesis represented by Fig. 2. For many taxa there are no surprises expected. For example data from E. sexfasciatus (sensu Wildekamp, 1997) should cluster with E. togolensis and E. infrafasciatus. Similarly, E. esekanus shares a derived feature, an open frontal neuromast system, with these taxa and is also expected to cluster with them. In the Western rainforest area, E. olbrechsti, once considered a subspecies of E. fasciolatus, is expected to cluster with it when sequence data are available. For other taxa, predicted relationships are less obvious. E. longiventralis, from the lower Niger River and upper Niger delta, is similar to the E. sexfasciatus complex of species. Scheel (1974) reported fertile hybrids between E. longiventralis and what is now called E. togolensis. However, Radda (1975) considered E. longiventralis a relict species possibly related to E. spilargyreius. E. biafranus is another species of unclear relationship which occurs in an area that overlaps the range of E. longiventralis. Radda (1975) suggest 195 a relationship to E. spilargyreius, while Scheel (1990) suggested a relationship to E. grahami. These are situations which molecular data should easily resolve. Since all sequences used for this analysis were from segments of mitochondrial genes, we cannot rule out the possibility of past hybridization events and subsequent lineage sorting distorting some portions of the phylogenetic topology. Inclusion of nuclear sequences in future work will be necessary to assess this possibility. However, the central theme of this survey was to test the derived nature of Eastern taxa of Epiplatys. For the sample of taxa for which sequence data is presently available, this certainly appears to be true. However, addition of sequence data from E. phoeniceps and multiple populations of nominal E. multifasciatus and E. mesograma, all from the interior of the Congo basin, are critical to further testing the hypothesis that these are elaborated from a single invasive lineage derived from the coastal clade of the genus Epiplatys. 4.8. Contrasts in dispersal ability Both Fundulopanchax (Murphy and Collier, 1999) and Callopanchax (Murphy et al., 1999) represent independent invasions of the coastal lowlands, yet neither has been as successful as Epiplatys in geographic expansion. However, each of these genera contain many species that are characterized as annuals. The geographic restriction of the other five genera of west African aplocheiloids suggests that they are ecologically restricted to their respective rainforest blocks and each lacked the ability to break out of these regions as each genus diversified. Epiplatys is unique in two ways. First, it was able to escape the region of its origin and second, it had retained morphological features that caused it to be confused with the genus Aplocheilus, a genus of greater antiquity. These features may reflect an adaptable, generalist body plan that was lost by the ancestral stocks of the other five genera as they adapted to their respective regions of endemism. Acknowledgments The authors are indebted to the Epiplatys Subcommittee of the Killifish Conservation Committee of the American Killifish Association for donations of material and support for this project. Special thanks are due to W. Hammer, J. Heller, L. Hutchings, T. Klotz, C. Minnemeyer, C. Nunziata, T. Terceira and S. Young. References Aarn, A., Shepherd, M.A., 2001. Descriptive anatomy of Epiplatys sexfasciatus (Cyprinodontiformes: Aplocheilidae) and a phylogenetic analysis of Epiplatina. Cymbium 25 (3), 209–225. Heller, J., 1997. Epiplatys boulengeri. J. Am. Killifish Assoc. 30, 53–55. Kligerman, A., Bloom, S., 1977. Rapid chromosome preparations from solid tissues of fishes. J. Fish. Res. Board Can. 34, 266–269. Langton, R., 2003. Wild Collection of Killifish. American Killifish Association. Lazara, K., 2000. The Killifishes: An Annotated Checklist, Synonymy and Bibliography of Recent Oviparous Cyprinodontiform Fishes. [The Killifish Master Index, 4th Edition] American Killifish Association Publication. Murphy, W.J., Collier, G.E., 1997. A molecular phylogeny for aplocheiloid fishes (Atherinomorpha, Cyprinodontiformes): the role of vicariance and the origins of annualism. Mol. Biol. Evol. 14, 790–799. Murphy, W.J., Nguyen, T.N.P., Taylor, E.B., Collier, G.E., 1999. Mitochondrial DNA phylogeny of West African aplocheiloid killifishes (Cyprinodontiformes, Aplocheilidae). Mol. Phyl. Evol 11, 343–350. Murphy, W.J., Collier, G.E., 1999. Phylogenetic relationships of African killifishes in the genera Aphyosemion and Fundulopanchax inferred from mitochondrial DNA sequences. Mol. Phyl. Evol. 11, 351–360. Parenti, L.R., 1981. A phylogenetic and biogeographic analysis of cyprinodontiform fishes (Teleostei, Atherinomorpha). Bull. Am. Mus. Nat. Hist. 168, 335–537. Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14, 817–818. Author's personal copy 196 G.E. Collier et al. / Molecular Phylogenetics and Evolution 50 (2009) 190–196 Radda, A.C., 1975. Cyprinodontiden-Studien im sudostlichen Nigeria 2. Die Reliktarten des subgenus Epiplatys. Aquaria 22, 137–149. Ronquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574. Scheel, J.J., 1972. Rivuline karyotypes and their evolution (Rivulinae: Cyprinodontidae: Pisces). Zeit. Zool. Syst. Evol. 10, 180–209. Scheel, J.J., 1974. Rivulin studies. Taxonomic studies of rivuline Cyprinodonts from tropical Atlantic Africa (Rivulinae: Cyprinodontidae: Antheriniforems: Pisces). Ann. Musee Royal de l’Afrique Centrale Tervuren, Serie 8, Zoologie 211, p. 148. Scheel, J.J., 1990. Atlas of the Killfishes of the Old World. T.F.H. Publications, Inc., Neptune City, New Jersey. Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony (*and other methods). Version 4. Sinauer Associates, Sunderland, MA. Swofford, D. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876– 4882. Wildekamp, R.H., 1993. A World of killies atlas of the oviparous Cyprinodontiform fishes of the World, vol. I. Brian Watters (Ed.). The American Killifish Association. Wildekamp, R.H., 1996. A World of killies atlas of the oviparous Cyprinodontiform fishes of the World, vol. III. Brian Watters (Ed.). The American Killifish Association. Wildekamp, R.H., 1997. Taxonomic stability restored in three former subspecies of Epiplatys sexfasciatus. J. Am. Killifish Assoc. 30, 77–89.