Genome-wide association study of yield components and fibre quality traits in a cotton germplasm diversity panel

• DOI: 10.1007/s10681-017-1855-y
• Publication Date: 2017-02-10T13:44:24Z
• Authors: Washington Gapare; Warren Conaty; Qian-Hao Zhu; Shiming Liu; Warwick Stiller; Danny Llewellyn; Iain Wilson
• Keywords: Germ plasm; Genome-wide Association Study; Linkage Disequilibrium; Genetic Association; Association mapping; Genetic architecture

A genome-wide association study (GWAS) was conducted on a diversity panel of 103 cotton accessions over three seasons to determine genetic contributions to a range of cotton yield components including fibre quality, plant architecture and stomatal conductance traits. The accessions covered breeding lines, released cultivars and some obsolete cultivars that contributed significantly to modern breeding pools. They were genotyped with Illumina’s CottonSNP63 K single nucleotide polymorphism (SNP) assay. Broad-sense heritability was low for yield component traits ( $$h_{B}^{2}$$ = 0.14–0.43), except for gin turnout and boll weight ( $$h_{B}^{2}$$ ) = 0.74 and 0.59, respectively), and low to high for fibre quality traits ( $$h_{B}^{2}$$ = 0.26–0.89). Population structure analysis revealed extensive admixture and cryptic relatedness amongst the accessions. Genome-wide linkage disequilibrium (LD) analyses showed LD decayed, on average, within a physical distance of 5 Mbp and reduced to 2 Mbp at r 2 ≥ 0.2, suggesting that few markers are necessary for association mapping in cotton. A mixed linear model accounting for population structure and cryptic relatedness identified 17 and 50 significant SNP associations for fibre length and micronaire, respectively. GWAS failed to detect significant associations in other traits, with the contribution of any single SNP to the phenotypic falling below 5%. This may be due to the low level of DNA polymorphism in cotton and/or insufficient resolution provided by the cotton SNP chip. Whole genome sequencing combined with whole genomic selection approaches that do not require prior knowledge about the effect or function of individual SNPs may be better suited than GWAS for trait dissection and prediction in cotton breeding.