A comparison of different stability models in wheat.

Eight different varieties of wheat, namely S-308, WG-377, K-227, PV-18, HD-1941, HD-1949, C-306 and C-281, were tested under twenty-four environments over two successive years. The models of Eberhart and Russell (1966), Perkins and Jinks (1968a), and Freeman and Perkins (1971) were applied to study genotype-environment interactions. The correlations between different environments were also determined and the partitioning of sum of squares due to genotype X environment interaction attributable to each variety was done as suggested by Wricke (1962). The term 'ecovalence' was used to signify this parameter. The major findings of the study were as follows: (i) The models of Eberhart and Russell, and Perkins and Jinks, produced similar results with respect to both responsiveness (b) and stability [Formula: see text]. The pattern of b-values in Freeman and Perkins' model was also similar to its pattern using these two models. (ii) The pattern of correlations between environments (r) for various genotypes showed high similarity with the pattern of b-values obtained in various models. The varieties having high b-values were found to have high environmental correlations and vice versa. (iii) Ecovalence and Freeman and Perkins' model were quite similar to each other for determining the stability of a genotype, but the rank correlations between ecovalence and Eberhart and Russell's model were low. It was, however, observed that the most stable varieties could be detected by using any of these models. Keeping in view the computational convenience, the use of correlation between environments (r) and ecovalence was suggested for predicting responsiveness and stability of genotypes, respectively.