Polyploidy is an important driving force in speciation and evolution; however, the genomic basis for parallel selection of a particular trait between polyploids and ancestral diploids remains unexplored. Here we construct graph-based pan-genomes for diploid (A₂) and allotetraploid (AD₁) cotton species, enabled by an assembly of 50 genomes of genetically diverse accessions. We delineate a mosaic genome map of tetraploid cultivars that illustrates genomic contributions from semi-wild forms into modern cultivars. Pan-genome comparisons identify syntenic and hyper-divergent regions of continued variation between diploid and tetraploid cottons, and suggest an ongoing process of sequence evolution potentially linked to the contrasting genome size change in two subgenomes. We highlight 43% of genetic regulatory relationships for gene expression in diploid encompassing sequence divergence after polyploidy, and specifically characterize six underexplored convergent genetic loci contributing to parallel selection of fiber quality. This study offers a framework for pan-genomic dissection of genetic regulatory components underlying parallel selection of desirable traits in organisms.

多倍体是物种形成和进化的重要驱动力;然而，多倍体和祖先二倍体之间平行选择特定性状的基因组基础仍未得到探索。在这里，我们为二倍体 （A₂） 和异体四倍体 （AD₁） 棉花物种构建了基于图的泛基因组，由 50 个遗传多样性种质的基因组组装实现。我们描绘了四倍体栽培品种的马赛克基因组图谱，说明了半野生形式对现代栽培品种的基因组贡献。泛基因组比较确定了二倍体棉和四倍体棉之间持续变异的同线和超发散区域，并表明一个正在进行的序列进化过程可能与两个亚基因组中对比鲜明的基因组大小变化有关。我们强调了二倍体中 43% 的基因表达遗传调控关系，包括多倍体后的序列分歧，并特别表征了有助于纤维质量平行选择的六个未充分探索的收敛遗传位点。本研究为生物体中理想性状平行选择的遗传调控成分的泛基因组解剖提供了一个框架。