Gene Conversion and Cotton Allotetraploid Evolution

Cotton tetraploids are a valuable source of fiber for textiles and their genome evolution has not been fully characterized. We analyze the evolution of allotetraploid cotton using whole-genome resequencing data consisting of over 18 billion reads from over 30 lines and all 7 allotetraploid species. We mapped the polyploid reads to both the G. raimondii and G. arboreum reference sequences using GSNAP and PolyCat, eliminating or severely reducing mapping biases between species with information from ~25 million homoeo-SNPs. With this depth and breadth of data, we examine non-reciprocal homoeologous recombination (gene conversion). Past studies have been unable to confidently distinguish between gene conversion events and polymorphisms arising in diploid lines. With the broad sampling of our study, we show that roughly half of putative events reflect true non-reciprocal homoeologous recombination events. We provide a refined phylogeny based on whole-genome SNP data, including the A_T and D_T genomes of each species--including the newly described G. eckmanianum and G. stevensonii--and 10 extant diploid relatives. We identify introgression of G. barbadenseinto G. hirsutum and vice versa. Homoeo-SNPs between the A and D genomes and allele SNPs within and between tetraploid species are made available for free use at CottonGen.