Submitted to: Applied Statistics In Agriculture Conference Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 9/23/2000
Publication Date: N/A
Interpretive Summary: Most traits that are important to plant breeders in cotton are controlled by many genes each with small effects. These are called quantitative traits, i.e., the trait is expressed in a continuous or quantitative manner. There are also traits controlled by single genes which are called qualitative traits, i.e., the trait is either present or absent. Sometimes these single qualitative genes are linked to, or have a measurable effect, on important quantitative traits. One easily classified major qualitative gene in cotton controls leaf shape with either a normal or an okra leaf being expressed. We developed a statistical and mathematical model to separate single gene effects on quantitative traits from the available additive-dominance models currently used by plant breeders. With our model we measured the effect of the okra leaf gene on quantitative traits such as specific fiber properties of micronaire, elongation, or strength. Using our model we detected a significant and positive effect of the okra leaf gene on fiber strength and a significant negative effect on micronaire. Our model is general and can be used to determine the effect on any qualitative marker, whether morphological or molecular, and separate its effect from the traditional additive-dominance model effects of quantitative traits.
Technical Abstract: Separation of single gene and polygenic effects would be useful in crop improvement. In this study, additive-dominance model with a single qualitative gene based on diallel crosses of parents and progeny F1s (or F2s) was examined. The mixed linear model approach, minimum norm quadratic unbiased estimation (MINQUE), was used to estimate the variance and covariance components and single gene effects. Monte Carlo simulation was used to evaluate the efficiency of each parameter estimated from the MINQUE approach for this genetic model. The results of 200 simulations indicated that estimates of variance components and single gene effects were unbiased when setting different single gene effects for parents and F1s (or F2s). Results also indicated that estimates of variance and single gene effects were very similar for both genetic populations. Therefore, single gene effects could be effectively separated and estimated by this approach. This research should aid the extension of this model to cases that involve multiple linked or unlinked genes (or genetic markers) and other complex polygenic models. For illustration, a real data set comprised of eight parents of upland cotton (Gossypium hirsutum L.) with a normal leaf and one parent with okra leaf, and their 44 F2s was used to estimate the variance components and the genetic effects of the okra leaf gene on fiber traits.