SAMPLING STRATEGIES FOR COMPOSING A CORE COLLECTION OF CULTIVATED BARLEY (HORDEUM-VULGARE S-LAT) COLLECTED IN CHINA

Using 96 accessions of Chinese barley landraces with reliable passport data, alternative methods for composing a core collection have been c ompared. The relative success of the different methods was assessed by counting the number of isozymic variants in the resulting core collec tions; the larger the number of alleles in the core collection, the mo re successful the method. Clustering on the basis of the location of t he collection site was compared with clustering on the basis of qualit ative and quantitative morphological characteristics. The collection s ite data proved to be the best basis for clustering, followed by the q ualitative descriptive data, i.e., row number, presence of awns or hoo ds, and kernel covering. Clustering based on quantitative data, i.e., heading date and plant height, did not improve sampling efficiency. Th e optimal number of clusters obtained using the collection site data w as eight, and corresponds to the number of geographical regions where the germplasm was collected. Alternatives for allocating the accession s to the groups after clustering were also compared; but seemed to hav e little effect; in general, proportional and logarithmic allocation s cored better than constant representation, but this was not always the case.