Genotype-environment associations across spatial scales reveal the importance of putative adaptive genetic variation in divergence

Identifying areas of high evolutionary potential is a judicious strategy for developing conservation priorities in the face of environmental change. For wide-ranging species occupying heterogeneous environments, the evolutionary forces that shape distinct populations can vary spatially. Here, we investigate patterns of genomic variation and genotype-environment associations in the hermit thrush (Catharus guttatus), a North American songbird, at broad (across the breeding range) and narrow spatial scales (at a hybrid zone). We begin by building a genoscape or map of genetic variation across the breeding range and find five distinct genetic clusters within the species, with the greatest variation occurring in the western portion of the range. Genotype-environment association analyses indicate higher allelic turnover in the west than in the east, with measures of temperature surfacing as key predictors of putative adaptive genomic variation rangewide. Since broad patterns detected across a species' range represent the aggregate of many locally adapted populations, we investigate whether our broadscale analysis is consistent with a finer scale analysis. We find that top rangewide temperature-associated loci vary in their clinal patterns (e.g., steep clines vs. fixed allele frequencies) across a hybrid zone in British Columbia, suggesting that the environmental predictors and the associated candidate loci identified in the rangewide analysis are of variable importance in this particular region. However, two candidate loci exhibit strong concordance with the temperature gradient in British Columbia, suggesting a potential role for temperature-related barriers to gene flow and/or temperature-driven ecological selection in maintaining putative local adaptation. This study demonstrates how patterns identified at the broad (macrogeographic) scale can be validated by investigating genotype-environment correlations at the local (microgeographic) scale. Furthermore, our results highlight the importance of considering the spatial distribution of putative adaptive variation when assessing population-level sensitivity to climate change and other stressors.