The analysis and prediction of some two factor interactions in grass breeding

Abstract

SUMMARY A method of analysis of two factor experiments is given. This involves a development of the regression analysis used by Finlay & Wilkinson (1963) and others, and allows for the regression of interaction components onto both main effects. The usefulness of the single joint regression parameter for prediction is outlined. The applicability of the analysis to three situations commonly found in grass breeding is illustrated by means of examples. It is concluded that the model may frequently describe variation due to genotype-environment interactions, and among diallel arrangements of binary mixtures of genotypes, but is likely to be of little utility for genetic diallel or other mating schemes unless the genes have a strongly correlated distribution among the parent plants.