Genotype x Environment Interaction and Stability Analysis in Upland Rice (Oryza sativa L.) Varieties in Ethiopia

Abstract

In Ethiopia, rice variety development program was started recently but many rice varieties were recommended for production. However, little information is available about the yield potential of varieties over diverse environments. This research was conducted to assess the magnitude of environment (E) and genotype x environment interaction (GEI) effects on yield and yields related traits, stability of the rice varieties for yield and to evaluate and stratify rice testing environments in Ethiopia. Twenty rice varieties were evaluated at six sites during 2016/17 in randomized complete block design with three replications. The combined analysis of variance over environments showed significant differences among genotypes (G), environments (E) and significant effect of GEI on phenology, growth traits, yield components and grain yield of varieties. Analysis of variance from Additive Main Effects and Multiplicative Interaction (AMMI) model revealed that G, E and GEI contributed 3.03 to 33.58, 25.69 to 72.02, and 7.09 to 30.36 %, respectively, to the total sum squares of traits. The E and GEI had a higher contribution than G to most of the traits, and GEI had larger contribution than G and E to the variations of varieties for yield. The interaction principal component axis one (IPC1) and two (IPC2) contributed 37.94 to 88.85 and 8.92 to 87.13% to sum squares of GEI, respectively. The varieties had mean yield of 3.509 t ha-1, and 9 and 11 varieties had mean yield lower and higher than the average yield of genotypes, respectively. The stability parameters of regression coefficient (bi) and deviation from regression (S2di) from Eberhart and Russell's (ER’s), IPCA1, IPCA 2, AMMI stability value (ASV) from AMMI model, GGE biplot, Francis and Kannenberg's coefficient of variability (CV), Lin and Binns's cultivar performance measure (Pi), Shukla's stability variance (δ2i), Wricke's ecovalence (Wi), Finlay and Wilkinson's (bi), Nassar and Huehn’s mean Absolute rank difference (Si1) and variance of ranks (Si2) were used evaluate the stability of genotypes for yield. NERICA-3, Hidasse and Chewaqa varieties had 15.99, 15.7 and 15.42% higher yield than average yield of genotypes, respectively, and identify as responsive to changing environments, selected as the first four best genotypes across poor to most favorable environments and GGE biplot identified as ideal and stable varieties suggested to be used for production across environments. NERICA-12 and ADET varieties had 12.85 and 8.29% higher yield, respectively, than average yield of genotypes and identify as most stable by most of the stability parmeters and considered as wide adaptable varieties. NERICA-4, NERICA-13 and Getachew varieties had 11.97, 10.83 and 5.7% yield advantages over the average yield of genotypes, identified as stable varieties by most of the stability parameters and selected as the first four best genotypes at poor to average environments suggested to be used in these environments. The GGE biplot identified Pawe and Tepi as most discriminating environments, Manbuk and Maitsebri identified as average environments and Woreta as representative environments that could be used as test environments to develop varieties at different stages of breeding activities. The stability parameters from ER's and AMMI models had positive and significant rank correlations with most of the stability parameters and yield and considered as sufficient to evaluate the stability of genotypes for yield. In conclusion, it is suggested to conduct the experiment over seasons to certain the results due to the significant effect of GEI and its larger contribution to the variations of varieties for yield and other traits.