Application of Factor Analytic Mixed Model for Multi-environment Trial of Bread Wheat (Triticum aestivum L.) Genotypes in Ethiopia
Bayisa Asefa *
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Negash Geleta
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Demeke Zewedu
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Tarekegn Aregaw
Department of Biometrics, Ethiopian Institute of Agricultural Research (EIAR), Addis Ababa, Ethiopia.
Berhanu Sim
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Alemu Dabi
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Rut Dhuga
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Habtemriam Zegeye
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Gadisa Alemu
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Tafesse Solomon
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Abebe Delesa
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
Abebe Getamesay
Kulumsa Agricultural Research Center, Ethiopian Institute of Agricultural Research (EIAR), Ethiopia.
*Author to whom correspondence should be addressed.
Abstract
This study was carried out using dataset consisted of 15 multi-environment trials (MET) in Alpha lattice design with two replications arranged in plot arrays of rows and columns conducted in Ethiopia during 2021 and 2022 main seasons. The objective of this study was to identify promising wheat genotypes that might suite diverse agro-ecology of the country through analysis of multi-environment trials (MET) data using factor analytic mixed models. The result of the study revealed that estimates for genetic variance components ranged from 0.049 to 1.036 and 0.33 to 1.915 for error variance. By ranking average best linear unbiased prediction (BLUPs) within clusters, the fifteen bread wheat environments were clustered into five mega environments (C1, C2, C3, C4 and C5) for grain yield. Thus, factor analytic linear mixed model can be fitted to large and complex MET datasets using a large and highly unbalanced MET dataset where there is a factorial treatment structure. This method is used as a selection indicator, assisting in screening superior and adaptable genotypes. The predicted performance of genotypes based on BLUP values averaged across correlated trails after eliminating C4 and C5 due to low genetic correlation with the other trials and low genetic variation. In addition, the results of the factor analysis for considering relationships among measured traits were confirmed through the cluster analysis. Based on these clusters, the genotypes EBW202104, EBW202058, EBW202057 and EBW202088 were identified as potential genotypes in Bread wheat improvement programs. Moreover, about 58.33% of the genotypes had average grain yield above grand mean; accordingly these genotypes might be selected for subsequent study in bread wheat breeding activities. The examined FA models have also better data fitting, which significantly improves heritability. Therefore, increasing the application of this efficient analysis method will improve the selection of superior bread wheat genotypes. Our study also supports the usefulness of this statistical tool to interpret MET data results and assist decision-making for its routine use in Bread wheat breeding programs.
Keywords: Cluster, factor analysis, linear mixed model, multi-environment trials (MET)