The success of the AMMI model in predicting lucerne yields for cultivars with differing dormancy characteristics

The AMMI model, which incorporates both additive and multiplicative components into an integrated, powerful, least squares analysis, is used for the analysis of cultivar trials at different localities. These trials are a major experimental component in crop production and plant breeding research. Multilocation trials have as objectives the assessment of the success of yield predictions, the grouping of sites for evaluation and the interpretation of genotype-environment interactions. The data used originated from the National Lucerne Evaluation Program and the lucerne cultivars were representative of dormancy groups 4–9, i.e. semi-, intermediate-, non- and highly non-dormant. Expected trends in results from the yield trials were not obvious using the additive main effects model (ANOVA), while the application of the AMMI model resulted in rankings of cultivars in different environments which could readily be explained by the breeding history and dormancy of the cultivars.