Much of the debate around speciation and historical biogeography has focused on the role of stabilizing selection on the physiological (abiotic) niche, emphasizing how isolation and vicariance, when associated with niche conservatism, may drive tropical speciation. Yet, recent re-emphasis on the ecological dimensions of speciation points to a more prominent role of divergent selection in driving genetic, phenotypic, and niche divergence. The vanishing refuge model (VRM), first described by Vanzolini and Williams (1981), describes a process of diversification through climate-driven habitat fragmentation and exposure to new environments, integrating both vicariance and divergent selection. This model suggests that dynamic climates and peripheral isolates can lead to genetic and functional (i.e., ecological and phenotypic) diversity, resulting in sister taxa that occupy contrasting habitats with abutting distributions. Here, we provide predictions for populations undergoing divergence according to the VRM that encompass habitat dynamics, phylogeography, and phenotypic differentiation across populations. Such integrative analyses can, in principle, differentiate the operation of the VRM from other speciation models. We applied these principles to a lizard species, Coleodactylus meridionalis, which was used to illustrate the model in the original paper. We incorporate data on inferred historic habitat dynamics, phylogeography and thermal physiology to test for divergence between coastal and inland populations in the Atlantic Forest of Brazil. Environmental and genetic analyses are concordant with divergence through the VRM, yet physiological data are not. We emphasize the importance of multidisciplinary approaches to test this and alternative speciation models while seeking to explain the extraordinarily high genetic and phenotypic diversity of tropical biomes.