Atlas: Mining Genetic Resource Collections for Useful Traits Using the Focused Identification of Germplasm Strategy (Figs) by Abdallah Bari

Abstract

As a source of useful agronomic trait variation, plant genetic resources have the potential to help meet the continuously increasing demand for food crops. Plant improvement depends largely on a combination of 50-60000 trait loci of the plant genome. The challenge, however, is how to find these “important” alleles as there are 6 million accessions (plant samples) worldwide stored in national genebanks and at international centers (CGIAR). These accessions are assembled in collections as repositories representing hundreds or thousands of years of selection and adaptation to local conditions. Searching these collections for important traits can be a daunting process since most of the accessions in these collections lack evaluation. FIGS as a trait-based and user-driven approach was conceived to provide indirect evaluation of germplasm for specific useful traits using, as a surrogate, the environment; based on the hypothesis that the assembled accessions are likely to reflect the selection pressures of the environment from which they were originally sampled. The FIGS approach combines both the development of a priori information based on the quantification of the trait-environment relationship and the use of this information to define a best-bet subset of accessions with a higher probability of containing the sought after agronomic trait variation. The FIGS approach has led to the identification of several novel trait genetic variations in recent years. To define the a priori information FIGS uses passport data of the accessions in combination with environmental data. Passport data is a standard set of descriptive information for each recorded accession in the collections including the geographical location where the accession was first sampled. This data is mostly non-experimental type data containing non-replicated observations (unique accessions). The passport data can be linked to environmental data using the geographical coordinates. The environmental data consist of a large number of variables ranging from climate data to soil data. FIGS explores the trait-environmental relationship using these environmental variables as well as GIS derived variables that might have the potential as correlates of trait genetic variation expression. The results have led to the detection of trait for resistance to yellow rust relationships with the climatic variables using parametric and non- parametric modeling techniques including learning-based techniques. The results also suggest a number of other issues that may improve predictive performance of the different modeling techniques. The paper discusses the results as well as the potential of FIGS in light of the availability of environmental data to enhance further use of genetic resources including identification of hot-spots for the traits of interest.