Genotype × environment interactions for two-row barley grain yield and implications for selection of test locations

Abstract

The response of barley grain yield to Canadian prairie environments was studied to evaluate genotype × environment interactions, and to group locations according to genotype response, which identifies locations whose removal would not significantly affect the validity of conclusions. The data were also used to illustrate a method for handling a large genotype × location × year data base with few common entries. Information from 20 test locations with 11–19 annual entries over 7 yr was used. Analyses of variance of data sets with three to seven common entries in adjacent years indicated significant genotype × location × year interactions for grain yield. The structure of the genotype × location interaction was studied using cluster analysis within each year and summarized over years. Cluster analysis using individual years allowed more test entries and should increase the reliability of the conclusions compared to that using average over years with few entries. Clustering identified six locations with dissimilar genotype yield responses. In the 7 yr, 19 pairs of locations usually clustered together. The 19 pairs involved eight of the 20 locations and most were in the same geographical region. Some of the eight locations could be eliminated without significant loss of reliability. Rankings of test entries for grain yield at locations within a cluster were generally similar when genotype effects were larger. Key words:Barley, Hordeum vulgare L, genotype-environment interaction, grain yield