Comparison of stability models in multi-environment rice trials in the Somali Regional State of Ethiopia

Abstract

Twelve upland rice varieties, deemed NERICA (Oryza sativa x O. glaberrima) were planted at three locations of the Somali Regional State of Ethiopia in April 2010 with the objective of identifying the most stable varieties and suitable stability models. Data were collected on eight yield and yield-related phenological and agronomic traits. The analysis of variance across locations for grain yield indicated that genotypic variations were highly significant to significant in all locations. The combined analysis of variance using Analysis of Variance (ANOVA) indicated that location, genotype, and genotype x location (G x L) mean squares were highly significant. The six stability models considered in the experiment (ANOVA, Eberhart and Russell, Additive Main Effects and Multiplicative Interactions (AMMI), Wricke’s ecovalence; Nassar and Hühn’s; and Shukla’s) identified different genotype(s) to be the most stable ones. On the other hand, comparison among the stability models using the criteria of Zobel et al. (1988) and Spearman’s rank correlation coefficients indicated, AMMI and Eberhart and Russell’s model outsmarted ANOVA and AMMI’s Interaction Principal Component Axis (IPCA); Wricke’s ecovalence (Wi); Nassar and Hühn’s absolute rank difference (S₁); and Shukla’s stability variance (σi²) were in best correspondence with the ranking of the genotypes, respectively.

Keywords/phrases: Correlation, Ecovalence, Genotypes, NERICA, Stability models

Ethiop. J. Biol. Sci. 11(1): 43-55, 2012