Identification of Factors Affecting Trait Stability via Factorial Regression and Environmental Variance

Abstract

Stability and genotype-by-environment interaction (GEI) are important for plant breeders. Our objective is to develop a method combining environmental variance and factorial regression to characterize genotype stability. To do this, we have examined kernel protein percentage of 16 winter wheat varieties grown in 13 environments. Genotype stability was assessed by environmental variance. Twenty-five covariates (24 environmental and 1 genotypic) were available. The method consisted of three steps. Firstly, the environmental variances were grouped according to their level defining six stability groups. Secondly, GEI was modelled with factorial regressions, using one or two environmental covariates. For each regression, the genotypes were characterized by their stability group and their slope sign(s) on the covariate(s). The third step identified the model that could best explain the interaction sum of squares and with slope signs of all the stable varieties opposite to those of the unstable varieties. Two environmental covariates, nitrogen accumulation rate per kernel and nitrogen mr2 at anthesis, clearly separated the most stable from the most unstable genotypes, giving new information for breeding stable varieties.