Abstract
Physical proximity among alleles shaped by different sources of selection is a fundamental aspect of genetic architectures critical for predicting their evolution. Theory predicts that evolution in complex environments should select for modular genetic architectures with limited pleiotropy among modules. However, limited data exist to test this hypothesis because the field lacks consensus for how to control for intercorrelated climate variables. We aim to characterize the genetic architecture of adaptation to climate, including the modularity of the architecture (number of distinct climate factors), overlap among modules, and physical linkage among loci. We introduce a co-association network analysis, which parses loci into groups based on differing environmental associations, and use it to study the genetic architecture of local adaptation to climate in lodgepole pine (Pinus contorta). We identified several non-overlapping modules of genes associated with environmental factors (aridity, freezing, geography), which supports the hypothesis of modular environmental pleiotropy. Notably, we found moderate physical linkage among some candidate loci in different modules, which may facilitate or hinder adaptation depending on the multivariate trajectory of climate change. Moreover, we show that associations with environmental principal components would have missed candidates and resulted in a limited interpretation regarding the selective environment. Finally, simulations revealed that the propensity of co-association modules to arise under neutrality increased with demographic complexity, but also that true causal loci are more highly-connected within the module, which may be useful for prioritizing candidates.
