• Anurans;
  • brazil;
  • phylogeography;
  • correlative models;
  • hylidae;
  • microhylidae;
  • mitochondrial


We assess whether correlative paleoclimatic models of species ranges accurately predict genetic diversity patterns in species of distinct life histories traits in the Atlantic Forest (AF) of Brazil. To this end, we use sequences of the mitochondrial gene ND2 from Dendropsophus elegans and Chiasmocleis carvalhoi – summarized in the shape of phylogenies and population genetic statistics – and maximum entropy models of species distributions under current, 21 kya BP and 120 kya BP climatic reconstructions. The two target species have distinct ranges, habitat tolerances, rates of reproduction and dispersal abilities, yet are endemic to the AF. Although the more restricted and semi-fossorial Ccarvalhoi is associated with forested habitats and thought to be a poor disperser, the widely ranged arboreal Delegans inhabits open areas such as pastures and human-impacted regions of the AF, and is easily found perched on herbaceous vegetation in inundated areas. We had anticipated that correlative distribution models of the broadly distributed D. elegans would perform better then models of the narrowly ranged C. carvalhoi, thus better predicting current patterns of genetic diversity. The results demonstrate poor predictive ability of climate-based models of C. carvalhoi under current climatic conditions, suggesting that factors such as biotic interactions or dispersal ability may be playing a central role in defining this species distribution – both now and in the recent past. Models under current climate are nonetheless accurate in the broadly ranged D. elegans. As a corollary, paleoclimatic models accurately predicted patterns of diversity of the ND2 mitochondrial gene in Delegans, but not in Ccarvalhoi. We attribute these distinct responses to the poor explanatory power of paleodistributions models when applied to species that violate the basic assumption of the environment as main driver of distribution patterns. This calls for a careful use of distribution models for the purpose of evolutionary biogeographical inference. Like C. carvalhoi, other species whose ranges are not yet at equilibrium, or which are impacted by competitor, parasite or pathogen presence, may not be suitable to the combined use of paleoclimatic-model based phylogeographic inference, as here implemented – despite relatively high area under the curve values.