Fig. 5

PAV selection signals during cotton domestication and improvement. a Gene number among the G. hirsutum landrace and improved accessions. The Wilcoxon rank-sum test (P < 0.001) was used for the significant statistics. b PCA analysis of 1020 accessions based on shell PAVs. c Maximum-likelihood phylogenetic tree and population structure with different number of clusters (K = 2, 3, and 4) in 1020 G. hirsutum accessions using 3803 shell PAVs. The population structure is sorted according to the phylogenetic tree. d, e Comparison of significant gene presence frequency between the landrace versus GhImpUSO group (domestication) and GhImpUSO versus GhImpCHN group (improvement) (FDR < 0.001, two-sided Fisher’s exact test). f Numbers of favorable and unfavorable genes during domestication and improvement. g, h PAV presence frequency of favorable and unfavorable genes during domestication and improvement. i, j GO enrichment analysis of favorable gene (i) and unfavorable gene (j) gain and loss during domestication and improvement