Genomic Prediction in Tetraploid Ryegrass

Perennial ryegrass is an outbreeding forage species, which is one of the most widely used forage grasses in temperate regions. Compared with diploid ryegrass, tetraploid ryegrass has better palatability, digestibility, and also hold a higher yield potential. Genomic selection has successfully been implemented in diploid ryegrass breeding during the past six years. The genomic selection is based on genotyping-by-sequencing (GBS), which is particularly well suited for genotyping plants that are bred in families. Similarly, statistical prediction models have been developed, which rely on allele frequencies that can easily be derived from GBS data. The aim of this study is to investigate the possibility of implementing genomic prediction in tetraploid perennial ryegrass and study the effects of different coverage depth.

A total of 1,148 F2 tetraploid ryegrass families from the breeding program at DLF Seeds A/S were included in the study. The traits considered were forage dry matter yield (DM), crown rust resistance (RUST) and heading date (HD) recorded between 2004 and 2014. Different SNP sets were created with minimum depth ranging from 1 to 50. The genomic prediction accuracy was estimated by using a “leave one single F2 family out” cross validation scheme, and the predictive ability was assessed with regards to accuracy and bias by comparing the corrected phenotypes and genomic breeding values.

The results indicate that genomic prediction can help to optimize the breeding of tetraploid ryegrass while bias needs to be further investigated by exploring possible missing of factors in the model or alternative scaling of the genomic relationship matrix.