Genetic variation, conditional analysis, and QTL mapping for agronomic and fiber traits in upland cotton

A recombinant inbred (RI) population containing 188 RI lines and two parental lines were used in this study. Variance components and covariance components were estimated for 18 agronomic and fiber traits by a mixed linear model approach. The proportion of genotypic variance to phenotypic variance was greater than 20% for all traits except fiber perimeter. Predicted genotypic values showed that agronomic traits could be influenced by additive x additive epistatic effects. A recursive approach was presented for constructing a new random vector that can be equivalently used to analyze multivariable conditional variance components and conditional effects. Lint yield and its three component traits were measured and analyzed by the recursive approach. The data showed that boll number per unit area made the largest contribution to genotypic and genotype x environment (G x E) variations in lint yield. Both boll number and lint percentage, and boll number and boll size jointly made more than 70% of the contributions to genotypic and G x E variations in lint yield. Ninety-nine percent of the genetic and phenotypic variation in lint yield could be explained by these three component traits. The data also showed that boll number and boll size interacted to affect the performance of lint yield. Boll retention rate for this RI population was analyzed by both ANOVA and logistic models. Estimates for boll retention were similar for both models; however, the logistic regression model gave higher precision for the estimates than the ANOVA model. Fifty-two SSR markers were used to construct linkage maps and to conduct quantitative trait locus (QTL) mapping for agronomic and fiber traits. A genetic model with additive, and additive by additive interaction effects was employed. Twenty-six QTLs with significant effects were mapped for nine agronomic and fiber traits. Most significant QTLs indicated that additive by additive interaction effects influenced most of traits with significant QTLs.