Estimation of genotype x environment interaction for yield in Ethiopia maize (Zea mays L.)

English: The study was undertaken to assess the performance of 10 maize genotypes across 15 maize growing environments of Ethiopia. The study was conducted from 1999 to 2001. The grain yields of these genotypes were analyzed using different statistical procedures to determine their G X E interactions and grain yield stability. The main objective of this study was to investigate the G X E interactions and stability performance of genotypes in various environments by applying different statistical methods of analysis in order to make useful recommendations for future utilization. Separate and combined analyses of variance across locations and years and five types of stability parameters were performed, using the AGROBASE 2000 program. In order to perform the stability analyses, data of 10 maize genotypes tested across five locations and three years were analyzed using the procedures of Finlay and Wilkenson (1963), Eberhart and Russel, (1966) for the joint regression, Wricke (1962) for ecovalence, Shukla (1972) for stability variance and (Gauch and Zobel, 1988) for the AMMI stability model. Separate trial analyses for the three years showed highly significant (P<0.01) differences among genotypes and locations for grain yield. In the year 1999, BH-670 was the best performer, followed by (A-7016 x G-7462) x 142-1-e and (A-7032 x G-7462) x 142-1-e with average yields of 9.59, 9.51 and 9.14 t ha" respectively. This ranking changed during 2000 and 2001, due to the presence of interactions. Across locations and years, (A-7032 x F-7215) x 144-7-b ranked first, followed by (A-7032 x G-7462) x 142-1-e and BH-670. All are three-way hybrids with mean yields of 8.93, 8.79 and 8.74 t ha" respectively. Among the locations the highest yield of 8.80 t ha" was obtained from Awassa, followed by Bako and Jimma over the three years, indicating the high potential of these sites for maize production. The results also showed yield variations over locations and years, confirming the presence of G X E interactions. The average of ANOV A components over the three years indicated that about 42% of the total variance was accounted for by genotypes and 13% by blocks. This confirmed variability between genotypes in their response to environmental factors. Combined analyses of variance across locations found highly significant (P<0.01) differences among locations (L) and genotypes (G) for grain yield. There was a .differential response of genotypes over locations, mainly due to edaphic and climatic factors. About 34% of the variance components were attributed to locations, while 16% of the variance components were attributed to genotypes and 12% to their interactions over the three years. This confirms the effect of environmental factors and thus the necessity of stability analyses for the appropriate genotypes. The combined analyses across locations, years and their interaction indicated highly significant differences (P<0.01) among the genotypes for grain yield, which suggests differential responses of genotypes to their environments. Significant G X E interaction makes the genotype selection processes difficult, which create problems in cultivar characterization. Stability analyses with appropriate statistical methods are therefore required to overcome this problem. Most of these interactions were highly significant due to abiotic and biotic factors, which need in-depth studies for better understanding. Generally, when G X E interaction is mainly caused by unpredictable environmental factors, breeing efforts should be aimed at the development of stable varieties with a relatively good performance under a range of environments. When the interaction is however due to predictable environmental factors the aim should be to develop either different varieties for different environments or broadly adapted varieties for a range of environments. The joint regression model for grain yield indicated highly significant differences between the genotypes. The joint regression model identified (A-7032 X G-7462) X 142- l-e as the most stable genotype, followed by (A-7032 X F-7215) X 144-7-b and (A-7033 X F-7189) X 142-1-e. These last two genotypes were the best yielders across all environments and both are three-way hybrids. Wricke's (1962) ecovalence considered BH-660 (one of the popular hybrids) as the most stable genotype, followed by (A-7033 X F-7189) X 142-1-e and Gibe-l (an openpollinated variety). BH-660 is the most popular hybrid currently under production in the country and Gibe-l is a newly released open-pollinated variety (OPV). (A-7032 X G- 7462) X 142-1-e and (A-7032 X F-7215) X 144-7-b were categorized as intermediate in stability, unlike Kulani and BH-140, which were found to be unstable according to this stability measurement. According to Shukla's stability variance (1972), BH-660 followed by (A-7033 X F-7189) X 142-I-e and Gibe-l were the most stable genotypes, whereas Kulani and BH-140 were considered as the least stable genotypes. BH-660, the popular three-way hybrid was the most stable genotype as measured by both ecovalence and the stability variance. Joint regression was also in agreement with these results with only slight differences. Additive main effects and multiplicative interactions (AMMI) stability values, and scores of the interaction principal component analysis (lPCA) showed that BH-660 was the most stable genotype followed by (A-7032 X F-7215) X 144-7-b and (A-7033 X F-7189) X 142-1-e, whereas Kulani and BH-140 were considered to be unstable. AMMI gave the same results as the ecovalence and Shukla in identifying the stable genotypes.