MAHALANOBIS’ D²-BASED GENETIC DIVERGENCE AND WARD’S CLUSTERING OF FINGER MILLET [Eleusine coracana (L.)] GENOTYPES IN A MULTI-LOCATION TRIAL
Abstract
Genetic divergence among twenty-four finger millet [Eleusine coracana (L.)] genotypes, comprising advanced breeding lines, released cultivars and checks, was assessed using Mahalanobis’ generalized distance. The genotypes were evaluated in an alpha lattice design with three replications at six locations (Almora, BAIF Nashik, Gaya, Palem, Ranichouri and Vizianagaram) during kharif, 2025 and best linear unbiased estimates across locations for nine quantitative traits formed the basis of the divergence analysis. Hierarchical clustering of the distance matrix by Ward’s minimum variance method grouped the genotypes into five clusters: Clusters I and IV were the largest (six genotypes each), followed by Cluster II (five genotypes), Cluster III (four genotypes) and Cluster V (three genotypes). Days to maturity (16.33%), days to 50% flowering (13.63%) and plant height (12.25%) contributed most to the total genetic divergence, together accounting for more than 42% of it. Intra-cluster distance ranged from 9.17 (Cluster I) to 22.60 (Cluster V), whereas inter-cluster distance was greatest between Clusters IV and V (23.69) and least between Clusters I and II (15.20), identifying Clusters IV and V as the most divergent pair. Cluster III recorded the most favourable cluster means for grain yield per plot and stover weight, together with earliness and reduced plant height. Crosses between genotypes of the divergent Clusters IV and V and between the high-yielding Cluster III and Cluster V, are suggested for generating desirable segregants in future finger millet improvement programmes.