Search

Insights into the causal mechanisms that limit species distributions are likely to improve our ability to anticipate species range shifts in response to climate change. For species with complex life histories, a mechanistic understanding of how climate affects different lifecycle stages may be crucial for making accurate forecasts. Here, we use mechanistic niche modeling (NicheMapR) to derive "proximate" (mechanistic) variables for tadpole, juvenile, and adult Rana temporaria. We modeled the hydroperiod, and maximum and minimum temperatures of shallow (30 cm) ponds, as well as activity windows for juveniles and adults. We then used those ("proximate") variables in correlative ecological niche models (Maxent) to assess their role in limiting the species' current distribution, and to investigate the potential effects of climate change on R. temporaria across Europe. We further compared the results with a model based on commonly used macroclimatic ("distal") layers (i.e., bioclimatic layers from WorldClim). The maximum temperature of the warmest month (a macroclimatic variable) and maximum pond temperatures (a mechanistic variable) were the most important range‐limiting factors, and maximum temperature thresholds were consistent with the observed upper thermal limit of R. temporaria tadpoles. We found that range shift forecasts in central Europe are far more pessimistic when using distal macroclimatic variables, compared to projections based on proximate mechanistic variables. However, both approaches predicted extensive decreases in climatic suitability in southern Europe, which harbors a significant fraction of the species' genetic diversity. We show how mechanistic modeling provides ways to depict gridded layers that directly reflect the microenvironments experienced by organisms at continental scales, and to reconstruct those predictors without extrapolation under novel future conditions. Furthermore, incorporating those predictors in correlative ecological niche models can help shed light on range‐limiting processes, and can have substantial impacts on predictions of climate‐induced range shifts. [ABSTRACT FROM AUTHOR]