Genome-environment associations in sorghum landraces predict adaptive traits

Improving environmental adaptation in crops is essential for food security under global change, but phenotyping adaptive traits remains a major bottleneck. If associations between single-nucleotide polymorphism (SNP) alleles and environment of origin in crop landraces reflect adaptation, then these could be used to predict phenotypic variation for adaptive traits. We tested this proposition in the global food crop Sorghum bicolor, characterizing 1943 georeferenced landraces at 404,627 SNPs and quantifying allelic associations with bioclimatic and soil gradients. Environment explained a substantial portion of SNP variation, independent of geographical distance, and genic SNPs were enriched for environmental associations. Further, environment-associated SNPs predicted genotype-by-environment interactions under experimental drought stress and aluminum toxicity. Our results suggest that genomic signatures of environmental adaptation may be useful for crop improvement, enhancing germplasm identification and marker-assisted selection. Together, genome-environment associations and phenotypic analyses may reveal the basis of environmental adaptation.