Developing Genomic Selection in White Clover (Trifolium repens L.)

Intensively grazed pastoral grasslands based on perennial ryegrass (Lolium perenne L.) grown in a mixed sward with white clover (Trifolium repens L.) are a primary source of animal nutrition in temperate agriculture. These forage species are adapted and bred as heterogeneous populations of heterozygous individuals – offering challenges and opportunities for plant breeders. Genomic selection (GS), used widely in animal breeding, utilises marker and phenotype data and is being developed in forages. Uptake of GS has been enabled by the availability of genomic resources, including reference genomes, and the advent of low-cost high-density genotyping platforms, such as genotyping-by-sequencing (GBS).

As part of Pastoral Genomics, a New Zealand Industry:Government-funded research consortium developing GS in forages, we derived and cross-validated genomic-estimated breeding values (GEBVs) for 242 white clover plants whose phenotypes were inferred by performance of corresponding half-sib families evaluated in a four-year field trial. GEBVs for traits such as dry matter production (DM) and stolon density (SD), were derived from Ridge Regression-BLUP, Random Forest regression, GBLUP and KGD-GBLUP (Dodds et al. BMC Genomics 2015 16:1047) using up to 20,861 GBS markers. Depending on GS model and marker dataset, cross-validation accuracy ranged from 0.05 to 0.29.

Divergent GEBV-directed polycross populations are being developed by selecting individuals from remnant half-sib family seed based on trait index GEBVs for both DM and SD using a Smith-Hazel Index - which enables simultaneous selection of multiple traits weighted on their individual value. These populations will be assessed in an upcoming field trial to evaluate response to selection.