Comparison of parametric and non-parametric stability statistics for selecting stable chickpea ('Cicer arietinum' L.) genotypes under diverse environments

In order to study phenotypic stability of 17 chickpea genotypes and comparison among parametric and non-parametric stability procedures, a randomized complete block design with four replications was carried out in five different research stations of Iran across two years (2004- 2005). Principal component analysis (PCA) indicated that the first four PCAs explained 88% of the variance of original variables. Biplot analysis of stability measures revealed that stability measures can be classified into four groups. Group 1 included the mean yield with non-parametric measures Ysi and TOP (proportion of environments in which a genotype ranked in the top third). Group 2 consisted of non-parametric measures (Si (1), Si (2), Si (3) and NPi (1)) and parametric measures (bi, S2xi, ASV, S2di and Wi2), which were correlated with mean yield (except for bi and NPi (1)). Group 3 included the non-parametric measures Si(6), NPi(2), NPi(3) and NPi(4) which were not correlated with mean yield and group 4 consisted of parameters R, Ri2, RSM and Pi that were negatively associated with most of the stability methods and mean yield, hence they can be excluded as suitable stability indices. The results of biplot analysis based on rank correlation showed that ASV, S2xi, W2i, Si(1), Si(2), Si(3) and NPi (1) can be used for evaluating the responses of chickpea genotypes to changing environments. Most of the stability methods indicated that the genotype G13 (FLIP 97-114) was the most phenotypically stable with high mean yield.