Multiple Parameters for Ascertaining Yield Stability of Upland Cotton Varieties Tested Over Number of Environments

Abstract

Thirteen upland cotton varieties were evaluated in 12 different environments of the Sindh province, Pakistan, so as to arbitrate their stability in yield performance. The regression coefficient (b) parameter was used as a measure of varietal adaptabil- ity, whereas the sum of squared deviations from regression (s2d) and coefficient of determination (r2) were implied as the measure of stability. The regression coefficients (b) of all the varieties, though did not deviate significantly from a unit slope (b = 1.0), yet varieties FH-1000, VH-142, BH-147 and FH-945 exhibited (b) values very close to a unit slope suggesting their better adaptation to the test environments. Varieties CRIS-467, DNH-57 and FH-945 displayed lower s2d and higher r2 values implying that these varieties were relatively more stable in yield performance than others in the test environments. Generally, not all the stability and adaptability parameters simultaneously favoured the same variety except FH-945, which was thus considered more stable, based on majority of the parameters. Principal component analysis (PCA) revealed that latent vectors of first two components, i.e., PCA-1 and PCA-2 accounted for about 91.24 % of the total variation. The eigen vectors of first PCA-1 were smaller and all were positive, which further suggested that the test varieties were quite adaptive to all the test sites. However, in PCA-2, some varieties gave positive and some negative eigen values, yet varieties FH-1000, CIM-499, CRIS-467 and FH-945 expressed smaller and positive PCA-2 scores suggesting less genotype-environment interactions for these particular varieties.