METHODS OF IDENTIFYING USEFUL GERMPLASM Robert W. Allard University of California, Davis, U.S.A. Empirical data from three intensively studied crop groups will be examined to identify genetic changes that occurred as wild species were converted into land-races and then into elite cultivars. First to be examined are changes that occurred over generations in the frequencies of discretely recognizable alleles of Mendelian loci that govern various morphological, disease resiatante, allozyme and restriction fragment variants. The main findings are that the frequencies of such alleles are highly correlated with adaptedness, productivity and product quality: evidently natural and/or manguided selection for improvement in these attributes caused the frequencies of some of these alleles to reach high levels and the frequencies of most such alleles to reach low levels over generations. Fortunately the frequencies of discretely recognizable alleles can be determined quickly, with great precision, and relatively inexpensively by simple counts of numbers of different alleles in various sources of exotic germplasm (e.g. accessions in germplasm banks, genetically enhanced populations), as well as in current local breeding stocks and cultivars. Rapidly growing arrays of discretely recognizable variants, including restriction fragment variants, thus appear to offer outstanding opportunities for identifying promising alleles for introgression into local breeding stocks. In an allele is frequent in some area but not present in other areas, prudent genetic resource management calls for introducing such alleles into the areas where they are not present. If an allele is rare everywhere it is unlikely useful anywhere and cost-effective germplasm managers will direct their efforts elsewhere. The task of managing germplasm at the genotypic level is much more difficult than at the allelic level. This is because having superior alleles in breeding stocks is not enough -superior aleles must be aseembled into synergistic multilocus combinations that give wide adaptedness and high performance over the range of fluctuations which occur in local environments. This is a substantial complication because the numbers of possible genotypes increase exponentially with increasing numbers of loci and increasing numbers of alleles per locus. The consequences are that large nambers of cycles of segregations and recombination, carried out in large populations, are required merely to produce the most useful multilocus genotypes and further, that laborious and expensive testing under agricultural conditions is required to determine the real value in any given environment of novel genotypes. The concept which emerges is that preservation of biodiversity, as well as the utilization of genetic resources are evolutionary processes and that understanding of the underllying evolutionary mechanisms responsible for the genetic changes that have occurred over generations provides the most certain guide for development of effective management strategies for the future.