Interpretation and implications of genotype by environment interactions in advanced stage sugarcane selection trials in central Queensland

Abstract

Data from 6 series of routine advanced stage selection trials were used to study genotype response across 'test' environments (locations and crop-years) for sugar yield in central Queensland. The objective was to determine whether genotype × environment (G × E) interactions are present for sugar yield in central Queensland and, if so, to interpret the nature of the interactions as well as determine their implications for the selection program. The relative magnitude of the G × E interaction components, namely genotype × location (G × L), genotype × crop-years (G × C), and genotype × location × crop-years (G × L × C), was studied using variance component analysis. In addition, environments were classified based on similarity with which they discriminate amongst genotypes (pattern analyses). The study revealed substantial G × E interactions for sugar yield. The magnitude of variance attributable to the second-order interaction effect, G × L × C, was higher than that of the first-order interaction effects, G × L and G × C, in a considerable number of cases. The pattern of genotypic response across environments was not consistent among the different series of trials. These results indicate that the major contributing factor or pattern underlying G × E interactions for sugar yield in central Queensland may be complex and unpredictable, making it difficult to effectively exploit G × E interactions in the breeding program. Based on logistics and resources, the current practice whereby a manageable number of genotypes (40–50) is evaluated in a subset of locations (3–4 out of 6 possible) for 2 crop-years, and only the elite ones are re-evaluated in further locations and crop-years, appears appropriate.