Aim To test alternative biogeographic hypotheses related to the diversification of a montane mammal (Zapus hudsonius luteus) endemic to the American Southwest.
Location South-western United States.
Methods We used statistical phylogeographic analyses of mitochondrial DNA (1512 bp; two genes) from 93 individuals from six geographic regions to test diversification hypotheses. Species distribution models of climate and fossil records were integrated to assess contemporary and historical distributions and barriers to gene flow. We calculated dates of divergence and examined historical demography using coalescent simulations.
Results We documented monophyly of Z. h. luteus represented by 19 segregated haplotypes. Predicted current distribution generally coincided with known localities, while predicted paleodistributions suggested that this lineage was widespread throughout lower elevations of the American Southwest and on the Edwards Plateau (as documented by the fossil record). Population size did not change substantially during a westward shift in range that occurred in the last 100 k generations. Results supported fragmentation of a common ancestor during the Holocene as the most plausible explanation for genetic structure.
Main conclusions Monophyletic Z. h. luteus reflects fragmentation of a common ancestor with recent (Holocene) upslope colonization of disjunct montane areas. We refute the hypotheses of in situ divergence or origins from a Colorado Piedmont ancestor. Instead, westward colonization from the Edwards Plateau during the Wisconsin followed by Holocene fragmentation, which serves as a generalized biogeographic hypothesis for species associated with mesic graminoid habitats in the American Southwest. Further exploration of these signatures using independent nuclear DNA is warranted. Key conservation implications are (1) Z. h. luteus is a monophyletic lineage on an independent evolutionary trajectory; (2) Z. h. luteus shared a recent common ancestor with Z. h. pallidus (not Z. h. preblei); (3) mtDNA does not reflect recent population declines; and (4) coalescent simulations and species distribution models reflect Holocene fragmentation.