Predicting Complex Phenotypes for Genome-Enabled Crop and Livestock Breeding

Agricultural genetics is currently being revolutionized by technological developments in genomic and statistical research. High-throughput genotyping technology delivers hundreds of thousands of single nucleotide polymorphism markers and has become available for many crop and livestock species. The technological feasibility of obtaining full genome sequences at reasonable costs for a large number of individuals is within striking distance. Thus, genetic analysis of quantitatively inherited traits and prediction of the genetic predisposition of individuals based on molecular data are rapidly evolving fields of research in agricultural genetics.

This talk presents quantitative genetic approaches that make it possible to extract knowledge on the genetic value of individual plants or animals from high-dimensional genomic data and to predict the genetic makeup of individual offspring. Statistical methods for prediction of genetic values and phenotypes from genome-wide molecular marker data will be introduced, and challenges arising from the large number of predictors and their high degree of collinearity will be addressed.

The efficiency of genome-enabled prediction will be demonstrated with experimental studies on grain yield and insect resistance in maize (Zea mays L.). Estimates of prediction accuracies achieved in these studies are encouraging with respect to the usefulness of genome-enabled prediction in practical breeding programs. On this basis, optimum scenarios for exploiting knowledge from high-dimensional molecular data in breeding schemes will be discussed.