Genotype × environment interactions and environmental adaptation. I. Pattern analysis — application to soya bean populations

Abstract

The application of numerical classificatory and ordination procedures to the analysis of adaptation of populations, cultivars, or lines is discussed. Examples given are seed yield and seed protein percentage for a population of soya bean lines evaluated across environments in south-eastern Queensland. For each line, p attributes (mean performance in each of p environments) were defined, and were considered to represent a p-dimensional space. The performance of the lines defined their position in this space, and the attributes were used to classify the lines into groups which differed in the magnitude and/or direction of their environmental responses. Within groups, lines exhibited variations across environments which were similar in both direction and magnitude. The relative proximity of lines in the space was obtained by ordination, and this reflected the degree of similarity of their performance across environments. The actual environmental responses of a large number of lines may be examined by these methods, both individually and as members of groups of lines having similar responses. Unlike other analyses of adaptation, these methods require no prior assumption regarding the distribution and suitability of a particular environmental response. Consequently, the analysis is designed to allow hypothesis generation. If desired, a 'suitable' or 'ideal' response surface may be defined, but this occurs subsequent to, and may be independent of, the identification of the actual responses of the lines and groups of lines.