Incorporating environmental information in an analysis of genotype by environment interaction for seed yield in perennial ryegrass

Abstract

Seed yield in perennial ryegrass was analysed for cultivar by environment interaction. Nine cultivars were evaluated in 12 trials at two locations over a 3-year period. Earlier attempts to describe the significant cultivar by environment interaction using a regression on the environmental mean or relationships with year, soil type, harvest method, or crop age, were unsuccessful. In this paper, therefore, meteorological data were introduced as explanatory variables. Three types of analysis were used. First, residuals from the cultivar by environment two-way table corrected for main effects were regressed on the explanatory variables for each cultivar separately. Secondly, the explanatory variables were used as concomitant variables for the environmental factor in a two-way analysis of variance of genotypes by environments. Finally, the matrix of residuals from additivity was subjected to a singular value decomposition, after which environmental scores were related to values of the explanatory variables using regression and a recently developed method to calculate confidence intervals for scores. All methods led to comparable conclusions about the importance of different variables in the interaction. Of equal importance were minimum temperature in the period before ear emergence, temperature sum in the period from the beginning of anthesis until peak anthesis, and mean and maximum temperature in the period from the end of anthesis until harvest. The major component of interaction was identified as a contrast between early and late cultivars. A minor component was due to cultivars that performed relatively well in the worst environment and relatively badly in the best environment. The usefulness of so-called AMMI models is discussed and compared with that of the more traditional regression on the environmental mean model. © 1993 The Genetical Society of Great Britain.