Abstract: 【Objective】The loci, elite alleles and candidate genes associated with yield component traits, such as boll weight, lint percentage, number of bolls per plant and seed index, were explored using a genome-wide association analysis (GWAS), which provided a theoretical reference for the molecular breeding of cotton yield.【Method】The GWAS based on a mixed linear model was performed on 408 upland cotton accessions grown in six different environments using the Cotton SNP 80K chip for the four yield component traits, and the significant SNP loci (SNPs) and elite allele were also detected. Finally, on the basis of the gene expression levels of the transcriptome, candidate genes related to the target traits were mined within a 1 Mb genome range of the flanking sequences of the significant SNPs. 【Result】The four yield component traits showed wide phenotypic variations in different environments, with the maximum coefficient of variation for number of bolls per plant being 16.67%-22.66%. The heritability of each trait was between 48.4% and 92.2%. The correlations among traits were significant or highly significant, except between boll weight and lint percentage. A total of 23 significant SNPs distributed in seven different genomic regions associated with the four traits were identified across the 408 cotton accessions in the BLUP. The numbers of loci associated with boll weight, lint percentage, number of bolls per plant and seed index were 5, 1, 9 and 8, respectively, and three loci (TM21094, TM21102, and TM57382) were associated with multiple target traits simultaneously. Seven elite allele types, TM21099(TT), TM57382(GG), TM78920(CC), TM53448(TT), TM59015(AA), TM43412(GG) and TM69770(AA), were identified. A total of 158 candidate genes potentially related to yield formation were selected through an analysis of gene expression patterns in RNA-Seq data. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes analyses indicated that the functions and metabolic pathways of most genes were varied.【Conclusion】In this study, 23 significant SNPs associated with four yield component traits were identified across 408 cotton accessions, and 158 candidate genes were predicted using RNA-Seq.

Key words: upland cotton, yield components, genome-wide association analysis, candidate genes