Accumulating evidence on ecogeographic variation in the mega-diverse central corridor of the Brazilian Atlantic Forest motivates us to revisit the vanishing refuge model (VRM) of speciation by Vanzolini and Williams (1981), testing hypotheses on mechanisms of tropical divergence. The VRM reconciles ecogeographic speciation with a modification of the forest refuge model in an attempt to explain patterns of distribution where sister or closely related species inhabit contrasting habitat types. The model evokes vicariance and divergent selection, leading to genetic divergence and phenotypic disparity among populations, eventually leading to speciation caused by climate-driven fragmentation of forested areas that vanish and give place to drier and more open habitats. Thirty years after its publication, this model remains relevant for integrating both geographic and ecological mechanisms of divergence and speciation, and it can be straightforwardly transferred to species in other vanishing biomes. The VRM is faithful to the tone of the earliest discussions about speciation being a multifarious process, yet it is in accordance with most recent studies that emphasize the ecological aspects of divergence. Here we refine the VRM and use examples of the Brazilian Atlantic Forest herpetofauna, spanning across a range of phylogenetic depths, to (1) make clear predictions on patterns of genetic structure, phylogenetic relationships, habitat and climate histories, and phenotypic disparity resulting from this model, and (2) consider alternative hypotheses on drivers of divergence and speciation that result in diversity patterns similar to those expected under the VRM. We also discuss how additional phenotypic and molecular data may help to elucidate mechanisms of divergence. We provide several examples of the Atlantic Forest herpetofauna, illustrating distinct stages of the VRM. Some Atlantic Forest reptiles show two patterns that are expected under the VRM, representing different stages during the process of lineage divergence and speciation. First, sister taxa split between forests and open formations, as predicted when the speciation process is completed under the VRM. Second, as expected during the divergence process, geographically isolated populations inhabiting high elevation forest enclaves surrounded by contrasting habitats (the “brejos de altitude”) are deeply genetically divergent and phenotypically distinct from forest populations in the climatically stable coastal forested areas. Yet, the pattern is not ubiquitous. Some taxa will be more likely than others to diversify in response to the shrinking and vanishing refugia, especially those with the ability to respond quickly to changes in climate and habitat structure. In this context, we predict that additionally to the presence of pre-adaptations, divergent selection is aided by high lability in tolerance traits, in such taxa. The patterns of distribution and phenotypic variation predicted by the VRM, however, are also compatible with alternative scenarios of speciation, including founder events and divergence along an ecological gradient. Our findings support the view that divergent selection acting in populations from geographic isolates in climatically dynamic areas can be a powerful force driving population divergence and, potentially, speciation. This mechanism, we argue, may be important in other vanishing habitats – going beyond forest fragmentation. Our results also highlight how populations in peripheral isolates show potential to adapt locally. We conclude by suggesting that studies on population divergence and speciation will benefit from more integrated analyses of molecular and phenotypic divergence.
Damasceno, R., Strangas, M. L., Carnaval, A. C., Rodrigues, M. T., & Moritz, C. (2014). Revisiting the vanishing refuge model of diversification. Frontiers in Genetics, 5(OCT). https://doi.org/10.3389/fgene.2014.00353