Ancestry and Environmental Adaptation in North American Feral Cannabis

Understanding how different populations respond to environmental variation is fundamental to breeding climate resilient crops. In this study, we integrate three diverse georeferenced Cannabis datasets comprising North American feral populations and Eurasian samples (n=909) to resolve population structure, infer evolutionary relationships and quantify adaptive responses to climate across a broad environmental gradient. Phylogenetic analysis rooted to Humulus lupulus shows North American feral populations are more closely related to basal and hemp-type lineages than to drug-type or Iranian populations, a pattern supported by ancestry and PCA analyses. By combining these datasets, we capture a wider range of climatic variation and gain new insight into the adaptive potential of cannabis germplasm. Using environmental genomic selection (EGS), we identified nine bioclimatic traits with prediction accuracies exceeding 0.5 across the combined datasets (12,030 SNPs; 909 samples; training = 191). When analyzing the North American dataset alone, EGS revealed 14 traits with prediction accuracies greater than 0.85 (22,852 SNPs; 760 samples; training = 310). Genomic estimated adaptive values (GEAVs) revealed population specific climatic responses, particularly for precipitation related traits, with northeastern North American populations (Indiana and New York) showing signatures consistent with adaptation to wetter and cooler environments. Climate projections under a high emission scenario (SSP585) indicate that [~]34 % of sampled locations are expected to transition to different Koppen-Geiger climate classes by 2050, with distinct shifts observed among North American and Iranian populations. Genome environment association (GEA) analysis identified replicated temperature and precipitation associated loci across multiple chromosomes. Using genotype-phenotype data, genomic selection for cannabinoid traits revealed that only CBD achieved prediction accuracies exceeding 0.5, consistent with a polygenic architecture extending beyond the CBDAS locus. Collectively, these results demonstrate that feral and landrace Cannabis populations harbor substantial adaptive variation and represent an underutilized reservoir for climate resilient breeding and allele discovery, with relevance for pre-breeding efforts under future climate scenarios.