Molecular phylogeny of Austrofundulus Myers (Cyprinodontiformes: Rivulidae), with revision of the genus and the description of four new species. Tomas Hrbek Donald C. Taphorn Jamie E. Thomerson Zootaxa 2005 825 1 39 36B5 52669 urn:lsid:zoobank.org:act:A12D02B2-18D2-450F-B6ED-A28984BAD110 Rivulidae Austrofundulus CoL Animalia Austrofundulus Myers 1932:159 Cyprinodontiformes 34 Chordata genus  DISCUSSION In the presented taxonomic revision of the genus Austrofundulus, we describe four new species and remove from synonymy an additional species. Some of the here presented taxonomic revisions have been anticipated in the most recent, but non-phylogenetic revision of the genus. In their revision, Taphorn and Thomerson (1978) predicted a close relationship of the RioAroa populations with those of the Rupununi. Due to the overall smaller body size of individuals from these populations, the authors furthermore predicted a close relationship to A. transilis. However, body shape and caudal fin morphology of the males suggested a relationship with A. limnaeus. Due to the sharing of features with both A. transilisand the other A. limnaeuspopulations, and the inability of morphometric and meristic data to distinguish the RioAroa and Rupununi populations from the other A. limnaeuspopulations, the RioAroa and Rupununi populations were conservatively assigned to A. limnaeus. However, these populations were considered an evolutionary transition between A. transilisand the other A. limnaeuspopulations. After the publication of Taphorn and Thomerson (1978), additional clues suggesting distinctness and possible species status of some of the Maracaibo populations came from difficulties of hybridization of the Guajira fish ( A. guajira) with RioMachango fish ( A. limnaeus). Despite repeated attempts to hybridize these fish, only two F1 hybrid offsprings of unknown fertility were produced (JET, pers. obs.). Due to rampant non-informative morphometric and meristic variation, a fine-scaled phylogenetic analysis of Austrofunduluswas possible only with the advent of modern molecular methods, resulting in a well supported phylogeny of the genus. The molecular phylogeny confirmed some, but not all of the previous revisers’ (Taphorn and Thomerson 1978) predictions. Additionally, the maximum likelihood molecular phylogeny is highly concordant with the geologic history of northern South America. The maximum parsimony phylogeny differs, but only in the phylogenetic position of A. myersi. Differences are most likely due to reconstruction artifacts associated with incomplete sequence data obtained from the 1958 A. myersiparatypes. Final resolution of the phylogenetic position of A. myersiwill require specimens more suitable for molecular analyses.  ShortOverview of northern South American Geology.- Although complex, the orogeny of northern South America is well documented and can be used as an additional source of support for the here proposed taxonomic revision of Austrofundulus. During the middle Eocene orogeny of Colombia, subduction of the Caribbean crust beneath the South American plate at the newly formed Sinu trench, caused a rapid uplift of the Cordillera Central and Cordillera Oriental respectively. This initiated the separation of the Colombian lowland, which, however, was not completed until the late Oligocene when major changes in direction of movement of the South American and Maracaibo plate with respect to one another caused the uplift of the Santa Marta massif, and the formation of the Sierra de Perija(Kellogg 1984). Uplift of the Sierra de Perijaconcluded at the early Pliocene. Further clockwise rotation of the Maracaibo basin initiated the orogeny of the Venezuelan Andes in the late Pliocene (Mattson 1984), resulting in the rapid rise of the Venezuelan Andes, and the separation of the Maracaibo basin from the present day Orinoco Llanos (Macellari 1984). Continuing clockwise rotation of the Maracaibo block together with the movement of the Caribbean and South American plates caused the rise of the Cordillera de la Costa and El Tigre highlands, respectively, effectively isolating the Tucacas lowlands and the RioUnare basin (Mattson 1984). A secondary uplift of the Guyana shield during the late Pleistocene isolated the Venezuelan Llanos from the Rupununi Savannah (Gibbs and Barron 1993). Based on this series of geologic events, the Colombian lowlands should have become isolated first, while the RioUnare basin should have separated from the RioOrinoco basin last. This orogenic series is concordant with the phylogenetic relationships within the genus Austrofundulus(Fig. 3). Geological areas that became separated more recently also contain species which branched off more recently from the ancestral Austrofunduluspopulation. The maximum likelihood phylogeny of Austrofundulusis thus highly concordant with the geological history of northern South America. Geological evidence provides additional support for the description of four new species of the genus Austrofundulus, and the resurrection of a fifth species from synonymy with A. limnaeus. Populations of A. transilisfrom the RioUnare basin are not recognized as a distinct species, since the separation of the Orinoco and Unare basins is very recent (Holocene), and populations from these two areas are morphologically indistinguishable and are not reciprocally monophyletic. This also applies to male coloration which otherwise clearly differentiates males of other species of Austrofundulus. Detailed biogeographical analysis will be presented elsewhere. When evaluated by traditional morphological criteria of species recognition and discrimination, our studies suggest that in areas of active orogeny that provide opportunities for allopatric speciation, much of the extant biodiversity remains and will remain unrecognized under many species concept criteria (Mayden 1997). However, the species in these different areas are not genetically interchangeable (Templeton 1981), although they may be ecologically interchangeable and morphologically indistinguishable simply due to stabilizing selection on morphological characters driven by the same set of ecological/environmental variables. This is the pattern we observe in northern South America. Moreover, this pattern is also prominent in the geologically active central Turkey, where, for example at least seven cryptic lineages of the killifish genus Aphaniusoccur (Hrbek et al. 2002). These lineages show complete or partial reproductive isolation (Villwock 1964), but only three species are scientifically recognized (Wildekamp et al. 1999; Hrbek and Wildekamp 2003). Thus, these species, whether occurring in northern South America or Anatolia, are real in that they embody an evolutionary process and form independent evolutionary lineages, but do not necessarily demonstrate a clear pattern of morphometric and meristic differentiation. It is important to remember that real evolutionary groups need not be morphologically distinct, whereas morphological categories are created as a direct function of their perceived distinction (Hey 2001). Therefore lack of morphological distinctness does not imply lack of real evolutionary lineages, i.e. species. Evolutionary lineages and morphological categories are not the same.