Evaluating the role of inbreeding depression in heterozygosity‐fitness correlations: how useful are tests for identity disequilibrium?

Heterozygosity‐fitness correlations (HFCs) have been observed for several decades, but their causes are often elusive. Tests for identity disequilibrium (ID, correlated heterozygosity between loci) are commonly used to determine if inbreeding depression is a possible cause of HFCs. We used computer simulations to determine how often ID is detected when HFCs are caused by inbreeding depression. We also used ID in conjunction with HFCs to estimate the proportion of variation (r²) in fitness explained by the individual inbreeding coefficient (F). ID was not detected in a large proportion of populations with statistically significant HFCs (sample size = 120 individuals) unless the variance of F was high (σ²(F) ≥ 0.005) or many loci were used (100 microsatellites or 1000 SNPs). For example, with 25 microsatellites, ID was not detected in 49% of populations when HFCs were caused by six lethal equivalents and σ²(F) was typical of vertebrate populations (σ²(F) ≈ 0.002). Estimates of r² between survival and F based on ID and HFCs were imprecise unless ID was strong and highly statistically significant (P ≈ 0.01). These results suggest that failing to detect ID in HFC studies should not be taken as evidence that inbreeding depression is absent. The number of markers necessary to simultaneously detect HFC and ID depends strongly on σ²(F). Thus the mating system and demography of populations, which influence σ²(F), should be considered when designing HFC studies. ID should be used in conjunction with HFCs to estimate the correlation between fitness and F, because HFCs alone reveal little about the strength of inbreeding depression.     

Population structure and genetic basis of the agronomic traits of upland cotton in China revealed by a genome‐wide association study using high‐density SNPs

Gossypium hirsutum L. represents the largest source of textile fibre, and China is one of the largest cotton‐producing and cotton‐consuming countries in the world. To investigate the genetic architecture of the agronomic traits of upland cotton in China, a diverse and nationwide population containing 503 G. hirsutum accessions was collected for a genome‐wide association study (GWAS) on 16 agronomic traits. The accessions were planted in four places from 2012 to 2013 for phenotyping. The CottonSNP63K array and a published high‐density map based on this array were used for genotyping. The 503 G. hirsutum accessions were divided into three subpopulations based on 11 975 quantified polymorphic single‐nucleotide polymorphisms (SNPs). By comparing the genetic structure and phenotypic variation among three genetic subpopulations, seven geographic distributions and four breeding periods, we found that geographic distribution and breeding period were not the determinants of genetic structure. In addition, no obvious phenotypic differentiations were found among the three subpopulations, even though they had different genetic backgrounds. A total of 324 SNPs and 160 candidate quantitative trait loci (QTL) regions were identified as significantly associated with the 16 agronomic traits. A network was established for multieffects in QTLs and interassociations among traits. Thirty‐eight associated regions had pleiotropic effects controlling more than one trait. One candidate gene, Gh_D08G2376, was speculated to control the lint percentage (LP). This GWAS is the first report using high‐resolution SNPs in upland cotton in China to comprehensively investigate agronomic traits, and it provides a fundamental resource for cotton genetic research and breeding.