Environmental ranges estimated from species distribution models are not good predictors of lizard and frog physiological tolerances

Abstract

Using species ranges, in particular those derived from species distribution models (SDM), to obtain characteristics of the species’ niche such as temperature tolerances is tempting. Over the past decade the literature has seen the increase in the use of SDMs based on locality data and spatially explicit datasets (climate, vegetation etc.). Furthermore, several studies have explored climatic niche evolution and niche conservatism using temperature and precipitation extracted from the resulting models in a phylogenetic context. However, species´ fundamental niches (set of abiotic conditions in which a species can live) are often incompletely characterized in SDMs, reconstructed mainly based on spotty locality data (about species presence and rarely including absence data). Indeed, a species´ realized niche, the actually occupied conditions where a species live, may be a subset of their fundamental niche due to lack of habitat availability, constraints on dispersion, and biotic interactions. Here, we produced SDMs for 50 species of neotropical reptiles and amphibians and compared extreme temperature estimates extracted from the modelled area (model-inferred) with thermo-physiological estimates of critical temperatures (physiology-inferred). When comparing experimental critical thermal maximum and minimums with temperature values extracted from the estimated range, we found a general pattern of maximum temperatures experienced that are cooler than the species maximum tolerances, and minimum temperatures close to or even colder than their minimum tolerances. Characterizing niche traits from SDMs is dangerous because SDMs are not representing the fundamental niche of species as measured with thermal physiology limits and they are also not deviating from the fundamental niche in a predictable way.