Development of Core Sets of Dolichos Bean (Lablab purpureus L. Sweet) Germplasm

Abstract

Development of a reduced representative set of a large base collection enables greater use of genetic resources in crop improvement programs. Efficiency of standard stratified clustering (SSC) and heuristic approaches were compared for developing core sets of Dolichos bean from a base collection of 648 accessions on the basis of data on 21 qualitative and 20 quantitative traits. The SSC approach with a combination of two core sizes (10% and 15%), two sampling strategies (proportional and logarithmic), and two allocation strategies (random and preferred) were used to develop eight core sets. Two additional core sets of 10% and 15% sizes were developed following a heuristic approach. Similarity of classes on the basis of qualitative traits of 10 core sets with the base collection was examined using Chi-square test, Shannon-Weaver diversity index, and ‘class coverage’ statistics. Univariate statistics, based on quantitative traits, such as mean and variance and multivariate statistics, standardized mean difference (SMD %), coincidence ratio (CR %), variance difference (VD %), and variable difference (VR %) were also used to assess the representativeness of core sets. Heuristic approach-based core sets retained higher CR%, VD%, and VR% based on quantitative traits. The study suggested that core sets developed based on heuristic method are better than those developed based on SSC approach relative to their representativeness of the base collection.