Environmental associations of the NR1D1 cTNR
We retained the first four principal components for our ecological association RDA analyses, which explained ~86% of the variation in our environmental variable dataset (Table 3). According to the eigenvectors, the first principal component was most highly influenced by precipitation variables (precipitation of the wettest quarter, precipitation of the driest quarter and average snow cover; Table 4). Our second component was largely influenced by habitat (proportion of suitable and unsuitable lynx habitat; Table 4). The third component was influenced by both temperature (mean temperature of the warmest quarter) and average snow depth (Table 4). Our fourth component was explained by a combination of precipitation (precipitation of the wettest and driest quarters, and average snow depth) and temperature (mean temperature of the warmest quarter; Table 4).
The partial RDA models describing environmental and spatial effects onNR1D1 genetic variation (controlling for neutral genetic structure) were only significant in the eastern Canada lynx data subset (p=0.001 and 0.008 for space and environment, respectively; Table 5). In this data subset, the total explainable variation was 0.2128, of which the effects of purely neutral genetic structure accounted for the largest proportion of variance (63.20% of total explainable variation, p=0.001), followed by the joint (collinear) effects of neutral structure, environmental and spatial data (28.81% of total explainable variation, p=0.001), with purely environmental and spatial data contributing much smaller, but still significant proportions of variation (4.93% and 3.05% total explainable variation in space and environment, respectively; Table 5). In the full dataset, only the full model and partial RDA model describing the effects of neutral genetic structure (controlling for environmental & spatial data) were significant (p=0.001; Table 5). Here, the total proportion of explainable variation was lower (0.1592), and largely comprised of the effects of purely neutral structure (71.73%), and the joint effects of neutral structure and environment/space (26.01%), with environmental/spatial effects contributing small, insignificant amounts to explained variation (0.19% and 2.07% for space and environment, respectively; Table 5). No models in the western data subset were significant suggesting that neither neutral genetic structure nor environmental or spatial variation explain genetic variation at theNR1D1 locus in this dataset (Table 5).