| Nine maize (Zea mays L.) composites grown at seven different locations in Nepal in 1982 and 1983 were studied to evaluate their yield performance in terms of stability and potential yield. A statistical model developed by Eberhart and Russell was utilized for computation of three parameters, mean yield, regression coefficient, and deviation from regression. Large deviations from regression limited the use of regression coefficients as predictive parameters.;The combined analysis of variance exhibited highly significant first and second order interactions indicating a need to select genotypes more specifically for different agroclimatic regions in the country. The stability parameters were used to show the existence of wide genotypic differences in response to environments and they are useful in describing and predicting genotype performances.;Genotypes having higher yield over 14 environments were Mankamna-2, Rampur-2, Ganesh-2, and Khumal Yellow. However, Mankamna-2 and Khumal Yellow had significantly greater deviation from regression. Rampur-2 and Ganesh-2 had higher mean yield, regression coefficients of one and had smaller deviation (near zero) from regression, therefore were classified as stable genotypes. In contrast a local cultivar which had low mean yield and highest deviation from regression was designated as the most unstable source. Rampur composites which had a low mean yield, a regression coefficient significantly less than unity and nearly zero deviation from regression was classified as a genotype suitable for less favorable environmental conditions.;Other agronomic characters including plant height and days to 50% silking were also evaluated using the Eberhart and Russell (1966) model. Only deviation from regression parameters was considered in this case since means and regression coefficients were nearly the same for all sources. Ganesh-2 was most stable in days to 50% silking. For plant height, Ganesh-2, Khumal Yellow, Rampur composites, Rampur-1, and Rampur-2 were rated as most stable. Although genotypes showed marked differences in mean performance, they were relatively nonsensitive to environment for their traits and the most important stability parameter appeared to be deviation from regression. |
