Assessment of a genomic design for a French meat sheep breeding program

Abstract

The majority of French meat sheep populations use both insemination (AI) and natural mating sires. Usually, AI sires are progeny tested and then the best are used as proven sires to produce male candidates. A breeding program based on genomic selection would be an alternative. Using a stochastic model, where both individuals and their genomes were simulated, we assessed a genomic breeding program design. The reference population was based on sires genotyped with a medium density panel (MD = 50K SNPs), including two to ten generations of sires born before the implementation of the genomic scheme and all sires born thereafter. For sire replacement, newborn progeny were first preselected on parent average genomic estimated breeding values (GEBV) and then genotyped with a very low density panel (VLD = 1K SNPs). MD genotypes of candidates were imputed using the software Fimpute and GEBV computed with a single step Genomic BLUP animal model using the software Blupf90. Males selected for replacement were then genotyped with the MD panel to update the reference population. We assessed the sensitivity of genetic gain to various sizes of the initial reference population and compared the genetic gain of genomic and classical breeding program designs at a fixed total cost with three different price levels for VLD genotyping. Within the range of values assessed for the size of the initial reference population and the price level of VLD genotyping, no significant differences between genomic breeding schemes was observed. At a fixed total cost, the annual genetic gain was higher for genomic designs (+18%) than for the classical design.