Inferring current and Last Glacial Maximum distributions are improved by physiology-relevant climatic variables in cold-adapted ectotherms

Abstract

Aim

Ecological niche-based models (ENM) frequently rely on bioclimatic variables (BioV) to reconstruct biogeographic scenarios for species evolution, ignoring mechanistic relations. We tested if climatic predictors relevant to species hydric and thermal physiology better proximate distribution patterns and support location of Pleistocene refugia derived from phylogeographic studies.

Location

The Western Palaearctic.

Taxon

Vipera berus and Zootoca vivipara, two cold-adapted species.

Methods

We used two sets of variables, that is physiologically meaningful climatic variables (PMV) and BioV, in a multi-algorithm ENM approach, to compare their ability to predict current and Last Glacial Maximum (LGM) species ranges. We estimated current and LGM permafrost extent to address spatially the cold hardiness dissimilarity between both species.

Results

PMV explained more accurately the current distribution of these two cold-adapted species and identified the importance of summer temperature and solar radiation that constrain activity in cold habitats. PMV also provide a better insight than BioV predictors on LGM distribution. By including notably, the permafrost extent, PMV-based models gave parsimonious putative arrangement and validity of refugia for each clade and subclade in accordance with phylogeographic data. Northern refugia were also identified from 48 to 52° N for V. berus and from 50 to 54° N for Z. vivipara.

Main Conclusions

Our hybrid approach based on PMV generated more realistic predictions for both current (biogeographical validation) and past distributions (phylogeographic validation). By combining constraints during the activity period (summer climatic niche) and those inherent to the wintering period (freeze tolerance), we managed to identify glacial refuges in agreement with phylogeographic hypotheses concerning post-glacial routes and colonization scenarios.