Abstract
Abstract
Genotype-environment association (GEA) approaches identify genetic variants that likely contribute to adaptation along environmental gradients. These associations can inform predictive models of maladaptation risk and evolutionary rescue in future climates. However, because population structure and climate parameters often co-vary, tradeoffs arise between accounting for population structure to reduce false positive associations and removing true positives. Here, we examine whether conducting GEA analyses at multiple spatial scales provides more informative associations by mitigating confounds between population structure and environment, and we evaluate what factors influence how congruent GEA results are across regions. We performed GEA analyses for climate parameters on genome resequencing data obtained for 110 annual populations of the common monkeyflower,
Mimulus guttatus
. We find that more genetic variation was specifically attributable to environmental variation in regional GEA models compared to a full-range model, and that agents of selection differ considerably in importance among regions and range-wide. Few genetic targets of selection shared across regions were detected, yet all regions possess ample standing genetic variation for most environment-associated variants, suggesting that regional differences in climate breadth may be a primary cause of this incongruence. In combination with a genome-wide association study for several climate-associated phenotypes, we also find that regulators of flowering repressors in the
FLC/MAF
gene family often harbor adaptive variation in this species. Together, our results suggest that GEA analyses of widespread species considerably benefit from examining both range-wide and regional trends to more fully capture and understand the agents and targets of climate adaptation.