Usefulness of Descriptors in Phenotyping Germplasm Collections.

A large number of crop germplasm collections are maintained within the U.S. National Plant Germplasm System (NPGS). For each of these crop collections, Crop Germplasm committees (CGC), crop curators, and collection staff have established extensive lists of descriptors or phenotypic traits by which to characterize and differentiate accessions. Descriptor data is maintained on the Germplasm Resources Information Network (GRIN) and utilized by requestors to select accessions that would best meet their research needs. This presentation will evaluate the descriptors for crops maintained at the Plant Genetic Resources Conservation Unit (PGRCU), Griffin, GA, to determine the usefulness of descriptors in properly characterizing accessions within each crop germplasm collection. For each crop descriptor, the amount of characterization data, the percentage of the collection characterized, the environmental impact on the trait, the measure of control over environmental variation, and environmental confounding will be determined. Many descriptors on GRIN have very little data or may not be useful in helping requestors select the best accessions for their needs. Other descriptors are routinely taken during seed regeneration in separate years with no replication or other control of the environment. These data have little value due to extensive differences between years that mask true differences between accessions. Research needs change over time and descriptors should be added in response to current needs of users. CGCs and crop curators should carefully review their crop descriptors to eliminate those that are of little use to requestors and add ones for new research areas. Efforts need to be made in characterizing core collections and whole crop collections with control of environmental variation (replications or years) to provide more useful phenotypic data on accessions within these valuable crop collections. These efforts could also stimulate more use of accessions that currently have little or no characterization data on GRIN.