Genotype-environment interactions and stability analyses for wheat and barley and classifying testing sites using multivariate and univariate techniques

Judgment of phenotype to evaluate the genetic potential of a genotype is difficult when genotype x environment interactions exist. Identifying the optimum environment where the greatest differentiation between genotypes relative to their genetic potential can be explored is essential. The objectives of this research were to estimate genotype x environment interactions and stability parameters of wheat and barley cultivars tested over 27 years.

This research was conducted on yield performance data of winter wheat and barley genotypes tested at several locations within the state of Tennessee during 1955-1981. Analysis of variance showed that all sources of variation and their interactions were highly significant except the genotype x year interactions for barley. In wheat, the genotype x year component of variance was much larger than that for genotype x location, suggesting that environmental effects associated with years were more important in determining differential genotypic responses.

Stability adaptation analyses indicated that 22% of the wheat genotypes had a deviation from regression close to zero and a regression coefficient not different from unity. Southern-Belle, Roy, and McNair 3001 were the most stable genotypes with broad adaptation and the best yielding capacity. Twenty-eight percent of barley genotypes were highly stable. Maury, Henry, Tenn 60-38, and Surry were highly stable barley genotypes with broad adaptation and high mean yields.

Factor analysis revealed that Jackson should be considered as a unique location due to its lack of association with other locations. Factor loading indicated interrelationships among Spring Hill, Springfield, Milan, Martin, and Knoxville. Regression techniques indicated that Crossville provided the least desirable conditions for wheat genotypes to reach their genetic potential and be differentiated, while Springfield, Milan, and Greeneville were suggested as the best testing sites, and Milan and Spring Hill were recommended as the most desirable selection sites for wide adaptation.

In barley, regression techniques revealed that Knoxville was the least desirable location for selection and testing purposes. However, subjecting the barley data to factor analysis revealed that Spring Hill, Knoxville, Springfield, and Jackson were loaded heavily on the same factor indicating the existence of interrelationships among them.