1978 Volume 28 Issue 4 Pages 343-358
The paper describes a method for analyzing the action and interaction of several major genes controlling a quantitative trait by means of diallel crosses, using a genetic model based on the factorial model in the design of experiments. A situation where lines having one of the major genes concerned and the line with no major genes are chosen as parents in diallel crosses was considered. The situation is specific, but seems to be one of the simplest and most realistic ones in major gene system. First, the utility of Hayman's diallel analysis based chiefly on (Vr, Wr) graph was investigated. It was found that (1) (Wr, Vr) is not constant and varies with dominance effect of individual major gene. Hence, test for epistasis from scatter of points (Vr, Wr) around the line of regression of Wr on Vr or from departure of the slope of regression from 1 becomes invalid. (2) Average degree of dominance estimated from the W. intersection of the regression line is subject to a marked over- or underestimation, depending on the average sign of dominance effect. (3) Position of the parental points on the regression line is influenced by the possible difference in dominance effect of major genes to a greater extent than in a polygenic case, although it still reflects the number of dominant genes the parent possesses. Thus, the (Vr, Wr) graph seems to give little information in the analysis of major gene effects.