The germplasm base collection (BC) of wild emmer wheat,Triticum dicoccoides, from Jordan consists of over 2,000 accessions. Data on 10 quantitative and 8 qualitative traits were analyzed using multivariate statistical analysis procedures. A stratified and proportional sampling procedure was used to select 10 and 5 of the BC to form two core collections: CC10 and CC5, respectively. Phenotypic variation in the BC, as estimated by the phenotypic diversity index, hs,j, was unevenly distributed among accessions, among populations, and among ecogeographical regions. Moreover, its level differed significantly among and within quantitatively and qualitatively inherited traits. This characteristic led to successfully selecting a core collection which represents all levels of hierarchy in the base collection. Examination of the data for the same 18 traits in the BC, CC10, and CC5 indicated that CC10 and CC5 retained, on average, 86.0 and 74.3 of the variation in the BC, respectively. Percent discrete phenotypic classes, with frequency 0.10 (i.e., common widespread and common localized phenotypes). recovered in CC10 and CC5 were 79.6 and 51.5, respectively. Results of hierarchical analyses of variance showed that percent contribution by traits to total variance in hs,j. estimates increased from 64.0 in the BC to 78.0 in CC10, and to 96.3 in CC5. Nevertheless, and as theoretically expected, CC10 was more efficient than CC5 in retaining the amount and composition of the variation originally found in the base collection.
展开▼
