Millets are hardy crops adapted to marginal lands in hot, drought-prone arid and semiarid environments. Among the small millets, barnyard (Echinochloa spp.), kodo (Paspalum scrobiculatum L.), and little (Panicum sumatrense roth ex roem. and Schult.) millets are the most underresearched crops in terms of useful genetic and genomic resources available to breeders for genetic enhancement in these crops. A core collection is an important strategy to enhance use of diverse germplasm with agronomically beneficial traits in applied breeding.

The entire germplasm collections of barnyard (736 accessions), kodo (656 accessions), and little (460 accessions) millets at ICRISAT were evaluated for 20 to 21 morphoagronomic traits in two to three rainy seasons at patancheru, India. Quantitative traits data were subjected to residual (or restricted) maximum likelihood analysis and the best linear unbiased predictors were obtained. Qualitative traits data and standardized data on quantitative traits were used to determine Gower distance matrix, which was subjected to hierarchical cluster analysis following Ward method at R square 0.75 to form distinct clusters. About 10% or a minimum of one accession from each cluster were selected to form core collections, which consisted of 89 accessions in barnyard, 75 in kodo, and 56 in little millets. Comparisons of means, variances, frequency distribution, diversity indices, and correlations indicated that the variation in the entire collection has been preserved in the core collections, which can be evaluated multilocationally to identify trait-specific diverse germplasm for use in genetic improvement of these crops.