Statistical analysis of sweetpotato trials in different agroecological zones in Ghana using the additive main effects and multiplicative interaction (AMMI) model

Pages 368-377

J.A. Otoo, A. Missah, C. Osei, A.G. Carson, E. Okai, R. Sagoe and A.G.O. Dixon


A multilocational trial to evaluate seven sweetpotato clones, comprising five improved/exotic and two local clones in sixteen environments (location and year combinations) for three years was carried out in four agroecological zones in Ghana. The hybrid model, Additive Main Effects and Multiplicative Interaction (AMMI) model was used to examine the pattern of interaction of the clones over different environments and select stable genotypes. The model revealed highly significant (P<0.01) environment (E), genotype (G) and G x E interaction. Improved genotypes gave better yields than the local types. G1 (TIS 3017) and G3 (TIS 8266) gave the highest unstable yield of 15.1 and 15.7 t/ha respectively, with G7 (TIS 86/0350) and G6 (TIS 84/0320) yielding good stable yields of 14.1 and 14.4 t/ha respectively. Most of the environments were stable, so with limited resource allocation a few of the environment within the same cluster (positive interactions) could be selected. The best environments were Fumesua, Nyankpala and Cape Coast. The highest yielding environment E3 (Nyankpala, 1995) was quite stable and the best stable environment was E2 (Fumesua, 1995). Site specific physical differences and seasonal factors conditioned the G x E interaction pattern in the data set. Since AMMI analysis reflects the expected G x E interaction for different environments, the clones TIS 84/0320, TIS 86/0350 and Sauti are recommended for release at 40% selection pressure.

Leave A Comment