Simultaneous testing of genotype x environment interaction and stability for more than one trait

This problem of simultaneous testing of stability for more than one trait is tackled here with the multivariate analysis of variance (MANOVA) approach considering the three important traits of sugarcane, viz. yield, ccs and sucrose simultaneously and by extending the perkins and jinks' model [1–3] of univariate. Results indicated significant variety x location interaction for all the traits, whether it is tested by the univariate or multivariate analysis. Regarding the significance of regression coefficients and deviation from regression under univariate (Perkins and Jinks approach) and multivariate testing the analyses showed that in most of the cases, if all the regression coefficients are significant then regression matrix is also significant. In other cases the multivariate testing reflects the majority position (significant or otherwise) under univariate testing; with a few exceptions i.e., except Co 87021, Co 7219 and Co 6304 regression coefficients were found to be non-significant in both cases either tested univariately or multivariately. In these genotypes regressions were found to be not significant for two traits only and resulted not significance for Co 87021 and Co 6304 but significant for Co 7219 when tested by multivariate analysis. The same can be said about the agreement in the multivariate and univariate testing of deviation mean squares. Deviation was observed to be not significant for Co 6304 only by both the procedures i.e., univariate as well as multivariate analysis.