Graph pan-genome illuminates evolutionary trajectories and agronomic trait architecture in allotetraploid cotton

Upland cotton (Gossypium hirsutum), one of the world's major fiber crops, faces challenges from the genetic homogeneity of modern varieties. Here we present 107 gold-standard genome assemblies spanning the wild-to-domesticated continuum, revealing six large-scale structural variations, including a chromosomal reciprocal translocation and five inversions tracing the evolutionary history of cultivated cotton in the Americas. This history also involved continuous introgression from Gossypium barbadense, shaping the genetic diversity of G. hirsutum landraces and cultivars. Leveraging the graph pan-genome, we capture the sequence and structural diversity of nucleotide-binding site-leucine-rich repeat genes, uncovering pathogen-driven selection signatures and loci associated with disease resistance. A presence-absence variation genome-wide association study (GWAS) identified previously overlooked loci for key fiber traits, complementing single-nucleotide polymorphism-GWAS findings. Additionally, we construct a detailed map of large inversions, offering insights into hybridization dynamics and strategies to mitigate linkage drag. This study enhances our understanding of cotton evolution and domestication while delivering a valuable resource to enhance breeding.