| Abstract: | The highland fish fauna of eastern North America consists of Appalachian and Ozark centers of endemism separated by the intervening Glacial Till Plains. Clades within these areas are more closely related phylogenetically to each other than to clades occurring in the intervening formerly glaciated region, suggesting that the Pleistocene glaciations fragmented a widespread highland region and its associated fauna. Alternatively, it is possible that these faunal assemblages predate the glaciations or that recent dispersals may have been more important than vicariance in determining faunal compositions. We examined the relationships among mitochondrial DNA (mtDNA) haplotypes within five clades of highland fishes, each with a distribution suggestive of a Pleistocene vicariance event. Darters of the subgenera Litocara and Odontopholis have distributions and mtDNA relationships that are consistent with the Pleistocene integration and burial of the Teays-Mahomet valley, a major drainage of the early Pleistocene. The distribution and mtDNA relationships among subspecies of Erimystax dissimilis are not consistent with Pleistocene vicariance, but relationships among Appalachian haplotypes are consistent with the late Pleistocene integration of the modern Ohio River system. Both Cottus carolinae and the Fundulus catenatus species group have representatives in the Mobile basin consistent with pre-Pleistocene divergences. Three haplotype clusters were found in C. carolinae, corresponding to the Appalachian, Ozark, and upper Kanawha River populations. However, Appalachian and Ozark F. catenatus populations are paraphyletic with respect to each other. This, coupled with a relatively low degree of sequence divergence, suggests that no long-term barriers to gene flow exist for C. carolinae and F. catenatus. These three distinct phylogeographic patterns indicate that Pleistocene vicariance is not the only explanation for the Appalachian-Ozark distribution of highland fish communities. |
