Little is known about how historical climatic events have shaped patterns of diversity in tropical islands of the Caribbean Sea. Pleistocene climate change may have had a large influence on the evolution of the biota in Puerto Rico, an island with high levels of biological diversity and endemism. Using ecological niche models of historical climatic conditions and detailed analyses of population genetic data, we tested alternative hypotheses of how two ecologically distinct species of frog responded to Late Pleistocene (21,000 ybp) climate change. We generated paleodistributional maps by projecting ecological niche models that predict the current distribution of each species onto a model of past climatic conditions. Eleutherodactylus portoricensis (Mountain Coquí) is restricted to montane rainforest in Puerto Rico. E. antillensis (Red-eyed Coquí) has a widespread distribution that encompasses Puerto Rico and the U.S. and British Virgin Islands, and is a habitat generalist that occurs in humid and xeric forests, forest edges, pastures, and suburban gardens. The paleodistribution models provided estimates of historical distributions and past barriers to gene flow for E. antillensis and E. portoricensis.  We tested these models of population history with genetic data (555 bp of mtDNA for 140 individuals of each species).  

Results/Conclusions

Paleodistributional models indicated that E. portoricensis and E. antillensis responded differently to Late Pleistocene environmental changes. The paleodistributional model for E. portoricensis predicted increased suitable habitat during the Late Pleistocene, with populations in two disjunct mountain refugia: Sierra de Luquillo and Sierra de Cayey. Phylogenetic analyses indicated that individuals of E. portoricensis from the two refugia form reciprocally monophyletic clades. Population genetic analyses and coalescent simulations revealed historical population stability, high genetic diversity, and significant differentiation between these two hypothesized refugia. In contrast, the paleodistributional model for E. antillensis predicted highest habitat suitability in southeastern Puerto Rico, with the current ecological niche model also showing suitable habitat in northeastern Puerto Rico. Strong genetic structuring of E. antillensis in northeastern Puerto Rico suggests that gene flow was more restricted during the Late Pleistocene due to low habitat suitability. Shallow differentiation between Sierra de Luquillo and Sierra de Cayey populations supports the hypothesis that E. antillensis was historically widespread in the lowlands between the two mountain ranges. Coalescent simulations do not support the existence of two refugial populations of E. antillensis. These collective findings improve our understanding of the factors that promote population divergence and that ultimately produce regional patterns of biodiversity in a neotropical island.