Assessment of parametric and non-parametric methods for selecting stable and adapted durum wheat genotypes in multi-environments

Abstract

Seventeen parametric and non-parametric methods for grain yield of 5 cultivars and 20 advanced durum wheat genotypes evaluated across 10 environments during the 2004-2007 growing seasons were used to assess performance stability and adaptability of the genotypes as well as to study interrelationship among these methods. Biplot analysis based on the rank correlation matrix indicated that most non-parametric methods were significantly inter-correlated with parametric methods. The results also showed that stability methods could be classified into four groups based on biplot analyses. The group related to the dynamic stability concept and strongly correlated with mean grain yield included the parameters of regression coefficient (bi), alpha (αi), TOP (proportion of environments in which a genotype ranked in the top third), environmental variance (Si2), coefficient of variation (CVi), Di2, Si(3) and Si(6). The second group included Wricke's ecovalence (Wi2), the Huehn's parameters [Si(1) Si(2)], Shukla's stability variance (σ2i), Plaisted and Peterson's parameter (P59) and Tai's model (λi) which were influenced by both yield and stability simultaneously. The third group included Kang's parameter (RS) and superiority index (Pi), which only measures stability. Genotypes 18, 16 and 2 were most stables based on parametric and non-parametric stability methods used.