Genomic Selection in grass breeding

Genomic Selection in grass breeding does not come without challenges. Like many other grasses, perennial ryegrass is an  outbreeding plant. it cannot self-pollinate and has to receive pollen from genetically different plants to become fertile. This feature has to major implications for genomic selection:

1. Grasses are bred in families of similar, but not identical, sibs, and therefore all DNA samples must be prepared from bulks on several plants from one family.

2. With a long history of outcrossing events, bits and pieces of the ryegrass genome has been shuffled around and recombined in many ways over time. As a consequence, the configuration between two distant points in one genome may not be the same i another genome. This is called a low LD (linkage disequilibrium) and means that if you want to identify a certain pattern (for example associated with yield), which is consistent over many ryegrass genomes, your DNA marker profiles have to be really dense (> 100.000 markers).

Once we have established a solid association between our DNA marker profiles and the traits we wish to breed for, we can start feeding the computer model with DNA profiles for new breeding material. The model will then compute a genomics estimated breeding value (GEBV) which can be used to either shortcut the breeding process or pre-select what is continuing through the breeding pipeline. Several traits can be estimated with great accuracy: