Genotype × environment interaction of winter wheat (Triticum aestivum L.) in South Africa: II. Stability analysis of yield performance

Abstract

Thirteen winter and intermediate type bread wheat cultivars were evaluated for yield stability under dryland conditions over a four year period from 1991 to 1994 and over a total of 120 environments in the Western, Central and Eastern Free State wheat producing regions of South Africa. The following statistical analyses were conducted and procedures followed to determine yield stability: (i) Shukla's procedure of stability variance(*2j); (ii) Lin and Binns cultivar performance measure (Pi); (iii) Finlay and Wilkenson's regression analysis and coefficient (b); (iv) Eberhart and Russel's deviation from regression (S2d); (v) Wricke's ecovalence (W1); (vi) AMMI model. Since the AMMI model does not make provision for a quantitative stability measure, such a measure was developed to rank genotypes. Total correspondence for significance of Spearman's rank correlation coefficients for the different analysis procedures was noted over the three production regions. No significant rank correlation coefficients were found in the pairwise comparisons of both Lin and Binns' and Finlay and Wilkenson's procedures with the other procedures, nor in the comparison between the two mentioned procedures. This indicates that the Lin and Binns procedure, as well as the Finlay and Wilkenson procedure, differ significantly from the other procedures in stability determination and definition, and due to noted deficiencies are consequently not recommended for use. From the study it would appear that if a single method of describing the stability of a genotype had to be selected, the proposed AMMI Stability Value (ASV) AMMI model would be the most appropriate. Certain cultivars showed similar stability over the three regions, while others varied considerably over the three regions.