The genetic architecture and genomic context of glyphosate resistance

Much of what we know about the genetic basis of herbicide resistance has come from detailed investigations of monogenic adaptation at known target-sites, despite the increasingly recognized importance of polygenic resistance. Little work has been done to characterize the broader genomic basis of herbicide resistance, including the number and distribution of genes involved, their effect sizes, allele frequencies, and signatures of selection. In this work, we implemented genome-wide association (GWA) and population genomic approaches to examine the genetic architecture of glyphosate resistance in the problematic agricultural weed, Amaranthus tuberculatus. A GWA was able to correctly identify the gene targeted by glyphosate, but when we statistically controlled for two target-site genetic mechanisms, we found an additional 250 genes across all 16 chromosomes associated with non-target site resistance (NTSR). The encoded proteins had functions that have been linked to non-target site resistance (NTSR), the most significant of which is response to chemicals, but also showed pleiotropic roles in reproduction and growth. The architecture of NTSR was enriched for large effect sizes and low allele frequencies, suggesting the role of pleiotropic constraints on its evolution. The enrichment of rare alleles also suggested that the genetic architecture of NTSR may be population-specific and heterogeneous across the range. Despite their rarity, we found signals of recent positive selection on NTSR-alleles by both window- and haplotype-based statistics, and an enrichment of amino-acid changing variants. In our samples, genome-wide SNPs explain a comparable amount of the total variation in glyphosate resistance to monogenic mechanisms, even in a collection of individuals where 80% of resistant individuals have large-effect TSR mutations, indicating an underappreciated polygenic contribution to the evolution of herbicide resistance in weed populations.