Evaluation of five strategies for obtaining a core subset from a large genetic resource collection of durum wheat

Abstract

The use of plant genetic resources contained in a large collection may be enhanced by specifying subsamples, called core samples. Five strategies for selecting a core sample from a collection of 3000 durum wheat accessions were applied and evaluated using four qualitative and eight quantitative spike characters. Each of the following strategies generated about 500 accessions for the core sample: random, random-systematic according to chronology of entry of the accessions into the collection, stratified by countryof-origin, stratified by log frequency by country-of-origin, and stratified by canonical variables. The first three strategies produced samples representative of the whole collection, but the remaining two produced the desired effect of increasing frequencies from less-represented countries-of-origin for several characters. The stratified canonical sample increased phenotypic variances. The quality of core samples is dependent upon good passport and evaluation data to partition the collection. The multivariate approach is extremely useful, but requires considerable data from the whole collection. Ecogeographic origin may be used in the absence of evaluation data on several characters to select useful core samples.