Search Results
Abstract
Improved methods of breeding, selection, and testing for yield can be developed with information on the magnitude and nature of genotype–environment interactions. Cultivar trials of processing tomatoes (Lycopersicon esculentum Mill.) grown in Ontario for 2 years at 5 locations each year were studied for genotype–environment interactions. Cultivars were evaluated for phenotypic stability and desirability using regression coefficients, mean square deviations from linear regression, and t test comparisons of genotype means with environment means. Genotype-environment interactions were significant for yield of marketable fruit each year and in a combined analysis across years. Regression analysis indicated that low-yielding genotypes had above-average yield stability across environments, while several high-yielding genotypes were unstable. Several cultivars were found to be desirable because they had a high mean yield and did not have lower yields than the test mean in any of the 5 environments. Regression analysis alone could result in misleading conclusions about the performance of high-yielding tomato genotypes. Large genotype-environment interaction variances relative to genotype variances were detected. The interaction variance components involving year were large relative to the genotype-location interaction variance, indicating the need for multiyear evaluation and selection for stability even when breeding for a limited geographic region.