Genomic Prediction of Heading Date and Crown Rust Resistance in Perennial Ryegrass

Perennial ryegrass (Lolium perenne L.) is the principal forage species grown in temperate regions worldwide. We have developed training populations with different mating structures and genotyped individual plants using a genotyping-by-sequencing approach for genomic selection. We determined predictive accuracies for heading date (0.84) and resistance to crown rust infection (0.63), and identified how these were affected by training population design, size, and marker density. Using variable importance measures we identified subsets of single nucleotide polymorphisms (SNPs) that could achieve predictive accuracies comparable with predictive accuracies achieved using the entire SNP set. Furthermore, using these smaller SNP sets improved predictive accuracies when predicting in plants that are unrelated to training set.