THE EFFECTS OF COMPONENT TRAIT SELECTION ON GENETIC GAIN OF LINT YIELD IN UPLAND COTTON BREEDING PROGRAM INVESTIGATED USING SIMULATION

Component traits have been commonly used in selecting for improved performance of a composite trait. The selection efficiency depends on the inheritance (heritability, gene action model) of and genetic correlations among traits. In this study, computer simulation was conducted to investigate the use of component traits (boll number, boll weight, and lint percentage) for the improvement of lint yield in upland cotton. Trait heritability and genetic correlations were chosen based on the current literature. Three inheritance models (additive, digenic epistasis, and trigenic epistasis) were considered. In total, 36 gene-environment systems were generated (two heritabilities x three inheritance models x six genetic correlations) to cover a range of situations in practice. Forty-six selection methods, which are different in the number of traits selected, selection mode, and selection intensity allocation among traits, were compared. All simulated genetics and breeding factors significantly affected the genetic gain of lint yield. However, the relative ranks of selection methods were very stable across heritability levels and inheritance models. The number of traits selected and the allocation of selection intensity among traits had relatively little impact on genetic gain of lint yield. If the improvement of lint yield is the sole breeding objective, selecting for increased performance of any of the three component traits (lint percentage, boll number, or bull weight) or their combinations are effective. The genetic correlations among traits were the most important factors affecting the effectiveness of a selection method. No selection method performed best across all correlation combinations.