Independent Domestication of Two Old World Cotton Species

Abstract

Domesticated cotton species provide raw material for the majority of the world's textile industry. Two independent domestication events have been identified in allopolyploid cotton, one in Upland cotton (Gossypium hirsutum L.) and the other to Egyptian cotton (Gossypium barbadense L.). However, two diploid cotton species, Gossypium arboreum L. and Gossypium herbaceum L., have been cultivated for several millennia, but their status as independent domesticates has long been in question. Using genome resequencing data, we estimated the global abundance of various repetitive DNAs. We demonstrate that, despite negligible divergence in genome size, the two domesticated diploid cotton species contain different, but compensatory, repeat content and have thus experienced cryptic alterations in repeat abundance despite equivalence in genome size. Evidence of independent origin is bolstered by estimates of divergence times based on molecular evolutionary analysis of f7,000 orthologous genes, for which synonymous substitution rates suggest that G. arboreum and G. herbaceum last shared a common ancestor approximately 0.4-2.5 Ma. These data are incompatible with a shared domestication history during the emergence of agriculture and lead to the conclusion that G. arboreum and G. herbaceum were each domesticated independently.

Keywords: Gossypium; crop plants; genome size; molecular evolution; repetitive DNA.

Fig. 1.

Morphological differences between fiber from wild and domesticated A-genome diploid cotton species. Shown are single seeds with single celled trichomes (fiber) from two cotton species, Gossypium herbaceum and Gossypium arboreum. Gossypium arboreum exists only as a cultigen.

Fig. 2.

Bar plot showing the abundance of the most common repeat types in the genomes of three Gossypium species. Species are color coded and indicated using genome designations (A1, Gossypium herbaceum; A2, Gossypium arboreum; and D5, Gossypium raimondii). Standard error bars are shown. Annotation abbreviations are as follows: RLG, Ty3/Gypsy retroelements; RLC, Ty1/Copia retroelements; RLX, unknown LTR retroelements; RXX, retroelement unknown superfamily; TXX, unknown DNA transposon; AT, AT-rich simple repeat.

Fig. 3.

Scatter plot of cluster abundance in Gossypium herbaceum (A1) and Gossypium arboreum (A2). Clusters that exhibit statistically significant difference in abundance between the two species are color coded as indicated. Clusters that do not exhibit and statistical difference are indicated in gray.

Fig. 4.

Cotton samples grouped by repeat content. Using the largest 1,000 clusters we assessed the similarity of repeat content on a per sample basis using hierarchical clustering, based on Euclidean distance. We identified natural groups (gated and numbered) using the ordihull and ordispider functions of the R package vegan. Importantly, Gossypium herbaceum (A1) and Gossypium arboreum (A2) are distinct in the first dimension.

Fig. 5.

Distribution of synonymous substitutions (K_s) between orthologs from Gossypium arboreum and Gossypium herbaceum. Alignment of over >7,000 genes in each comparison allowed the mean and median substitution rate between species to be estimated.