Cultivar stability analysis using a discontinuous bi-segmented model: unbalanced experiments

Abstract

Groups of cultivar experiments carried out at different environments are often unbalanced since some cultivars are not tested in all the locations. An analysis of variance and phenotypic stability by the discontinuous bi-segmented model with corrections due to errors in the variables, for the unbalanced cases, is presented in this paper. It was observed, by simulation, that the higher the level of imbalance the lower is the accuracy of the estimates. However, losses of up to 25% of environments are tolerable and do not preclude a good description of the behavior of the cultivars under environmental variation if the given cultivar is present in at least two favorable and two unfavorable environments.