EFFECTS OF NOISE ON AMMI AND HIERARCHICAL-CLASSIFICATION ANALYSES

The Additive Main effects and Multiplicative Interaction (AMMI) model is used for the analysis of trials in which cultivars are tested at a different number of locations. The data used in this study originated from the South African National Lucerne Evaluation Program. The application of the AMMI model resulted in rankings of cultivars at different environments which could readily be explained by their breeding history and dormancy, while this was not obvious using the ANOVA model. The cultivar by environment means were then subjected to a hierarchical classification (HC) to group the cultivars into homogeneous groups. It was found that, in contrast to the results of HC, the AMMI biplot presents an agronomically interpretable pattern as far as the grouping of the cultivars and environments is concerned. In a simulation experiment conducted to determine the influence on pattern recovery under multiplicative noise, AMMI seemed to focus more on pattern than HC.