Vole samples and climate variables
Tissue samples of a total of 275 voles representing 12 widely dispersed populations (Figure 1) with 21-24 individuals per population (Table S1) were collected by the authors, covering a distance of 3,200 km from the northeast in  N Finland (25.9°E 68.0°N) at the northern distribution limit to the southwest in France (0.8°E 43.2°N) and 2,700 km from the southwest (France) to the southeast in Romania (25.1°E 46.6°N). For each population, values of 10 bioclimatic variables were downloaded from the WorldClim V2 dataset (Fick & Hijmans, 2017). Climate at the sampling sites ranged for mean annual temperature from -2°C to 12.5°C, for temperature seasonality from 500 to 1000 (SD*100), for mean diurnal temperature range from 5.5°C to 9.8°C, for mean annual precipitation from 480mm to 1080mm, and for precipitation seasonality from 11% to 50% (Figure 1). To account for correlation among climate variables (Table S9), principal component analysis (PCA) was used to reduce dimensionality in R v3.4.4 using the prcomp function (R Core Team, 2018)⁠ [2] (Figure S1). This resulted in two climate-based principal components that together explained 80% of the total variation. PC1 explained 62.5% of the variation and was mainly associated with temperature variables, while PC2 explained 17.1% of the variation and was mainly associated with precipitation variables (Table S2).