Analyzing Genotype-Environment interaction (GxE) is useful for exploring the opportunities and limiting the drawbacks of the effects. An investigation was conducted on 12 environments (six locations and two years) to study GxE and stability of 18 tef varieties, and to identify desirable environments. The experiment was laid in a randomized complete block design with three replications. The result of AMMI analysis of variance showed that tef grain yield was significantly (P < 0.001) affected by environments (E), genotypes (G) and GxE. Environment, G and GxE explained about 90.23%, 1.03% and 8.74% of the total sum of squares of treatments, respectively. The partitioning of the GxE by AMMI analysis showed that two of the Interaction Principal Component Axes (IPCAs) were highly significant (P < 0.001). The two IPCAs explained 66.06% of the total GxE in grain yield of the tef genotypes. AMMI1 showed that genotypes G and D had small interaction effects. Likewise, environments MR06 and KB07 had the highest interaction effects whereas SR06, KB06, JM06 and JM07 had smaller interaction effects. AMMI2 also showed that environments MR06, KB07 and CH07 exerted higher interaction effects; however, KB06, JM06 and JM07 exerted lower interaction effects. The GGE biplot identified three mega-environments: The first mega-environment is composed of environments Kobo, Jari and SR07 with genotype Tseday as a winner; genotype Ziquala represented the second mega-environment containing Jamma, Chefa and SR06; the third environment, made up of Mersa, was represented by genotype Asgori. Tseday was the most desirable variety; while Mersa was the least desirable environment.
Adaptability, AMMI, Eragrostis tef, G x E, GGE, Stability