ASSESSMENT OF GENOTYPE BY ENVIRONMENT INTERACTION, YIELD STABILITY AND INTERRELATIONSHIPS AMONG DIFFERENT STABILITY MEASURES OF BREAD WHEAT GENOTYPES

doi:10.12816/ejpb.2019.256171

ASSESSMENT OF GENOTYPE BY ENVIRONMENT INTERACTION, YIELD STABILITY AND INTERRELATIONSHIPS AMONG DIFFERENT STABILITY MEASURES OF BREAD WHEAT GENOTYPES

Document Type : Original research

10.12816/ejpb.2019.256171

Abstract

The objectives of this study were to compare nine parametric and 5 nonparametric stability measures and to identify high yielding and stable bread wheat genotypes in twelve variable environments during 2014-15 and 2015-16 growing seasons. The wheat genotypes comprised ten local cultivars and five exotic Syrian genotypes. They were grown in a split plot design arranged in a randomized complete blocks with 3 replications. Combined analysis of variance for grain yield (kg m^-2) indicated that the genotypes, environments and their genotype x environment interaction effects were highly significant. The parametric stability statistics; deviation from regression (S²_d) , ecovalence (W_i), stability variance (σ_i²), coefficient of determination (R²) and mean variance component for a pairwise G x E interaction (P⁵⁹) revealed that the local cultivar Sakha 93 was the most stable genotype, while cultivar’s superiority index indicated that the local cv. Shakha 94 was considered the most stable genotype. The nonparametric stability statistics indicated that cv. Giza 168 exhibited the smallest changes in rank (S_i⁽²⁾) and thus was the most stable genotype. According to the two nonparametric stability measures (TOP and RS), the two local cvs.; Sakha 94 and Sids 1 were the most stable genotypes for grain yield. As for comparing the fourteen stability measures, the S²_d, W_i, σ_i², P⁵⁹and the nonparametric stability measures (S_i⁽²⁾, S_i⁽³⁾ and S_i⁽⁶⁾) were nearly similar in assessing the relative stability of genotypes, whereas remaining stability measures deviated from others. The rank correlation analysis indicated that most nonparametric statistics were significantly correlated with parametric measures and therefore can be used as alternatives due to its simplicity.