Genomic Selection In Perennial Ryegrass

Genomic selection (GS) for crop improvement makes use of genome-wide molecular marker information available in more and more species. The aim is to enable accurate prediction of breeding values for target traits of a quantitative nature by determining the effect on phenotype of all the markers simultaneously. GS has proven its usefulness in animal breeding programmes, but its impact in plant breeding is only just emerging. GS can potentially increase the speed of the breeding cycle, reduce the cost and effort of phenotyping, and achieve faster selection of candidates for crossing programmes and production of synthetic varieties. Phenotypic predictions are based on models developed in a training population for which both phenotypic and genotypic data are available. The size of the training population and the density of molecular marker coverage are key constraints for prediction accuracy. The perennial ryegrass recurrent selection breeding programme at IBERS is based on a relatively small founder population, theoretically decreasing the number of molecular markers needed to achieve a certain level of prediction accuracy. Evidence will be presented for estimation of the extent of LD in this population. This will be followed by a description of how we aim to integrate GS into the existing breeding programme. Some initial cross-validation results will be presented using existing phenotypic data and molecular marker data of approximately 3000 markers, largely based on a ridge-regression best linear unbiased prediction model