TY - JOUR
T1 - Most of the benefits from genomic selection can be realised by genotyping a proportion of selection candidates
AU - Henryon, Mark
AU - Berg, Peer
AU - Sørensen, Anders Christian
AU - Ostersen, Tage
AU - Nielsen, Bjarne
PY - 2012/12
Y1 - 2012/12
N2 - We reasoned that there are diminishing marginal returns from genomic selection as the proportion of genotyped selection candidates is increased and breeding values based on a priori information are used to choose the candidates that are genotyped. We tested this premise by stochastic simulation of breeding schemes that resembled those used for pigs. We estimated rates of genetic gain and inbreeding realized by genomic selection in breeding schemes where candidates were phenotyped before genotyping and 0-100% of the candidates were genotyped based on predicted breeding values. Genotypings were allocated to male and female candidates at ratios of 100:0, 75:25, 50:50, 25:75, and 0:100. For genotyped candidates, a direct-genomic value (DGV) was sampled with reliabilities 0.10, 0.50, and 0.90. Ten sires and 300 dams with the highest breeding values after genotyping were selected at each generation. Selection was for a single trait with heritability 0.20. We found that the marginal returns did diminish as the genotyping proportion was increased, while the rate at which the returns diminished slowed as the DGV became more reliable. With DGV reliability 0.10, genotyping as little as 5% of the selection candidates realized 86% of the additional genetic gain and 67% of the reduction in inbreeding that was realized by genotyping 100% of the candidates. All of the genetic gain and reduction in inbreeding was realized by genotyping 40 and 50% of the candidates. When the reliability was increased to 0.90, genotyping 20% of the candidates was required to realize 76% of the genetic gain and 85% of the reduction in inbreeding. Genotyping 50% of the selection candidates with DGV reliability 0.90 realized 91% of the genetic gain and 94% of the reduction in inbreeding. Regardless of the DGV reliability, returns at low genotyping proportions of 0.5-10% were maximized when only male candidates were genotyped. At the high genotyping proportions of 20-50%, returns were maximized by genotyping both males and females. Our findings indicate that, provided a priori information is available, only 5-20% of the selection candidates need to be genotyped to realize most of the benefits from genomic selection. At these genotyping proportions, it is best to target males in schemes when the selection intensity for males is greater than females. Our findings should be welcomed by breeders because they suggest that large investments in genotyping are not required to reap most of the benefits from genomic selection
AB - We reasoned that there are diminishing marginal returns from genomic selection as the proportion of genotyped selection candidates is increased and breeding values based on a priori information are used to choose the candidates that are genotyped. We tested this premise by stochastic simulation of breeding schemes that resembled those used for pigs. We estimated rates of genetic gain and inbreeding realized by genomic selection in breeding schemes where candidates were phenotyped before genotyping and 0-100% of the candidates were genotyped based on predicted breeding values. Genotypings were allocated to male and female candidates at ratios of 100:0, 75:25, 50:50, 25:75, and 0:100. For genotyped candidates, a direct-genomic value (DGV) was sampled with reliabilities 0.10, 0.50, and 0.90. Ten sires and 300 dams with the highest breeding values after genotyping were selected at each generation. Selection was for a single trait with heritability 0.20. We found that the marginal returns did diminish as the genotyping proportion was increased, while the rate at which the returns diminished slowed as the DGV became more reliable. With DGV reliability 0.10, genotyping as little as 5% of the selection candidates realized 86% of the additional genetic gain and 67% of the reduction in inbreeding that was realized by genotyping 100% of the candidates. All of the genetic gain and reduction in inbreeding was realized by genotyping 40 and 50% of the candidates. When the reliability was increased to 0.90, genotyping 20% of the candidates was required to realize 76% of the genetic gain and 85% of the reduction in inbreeding. Genotyping 50% of the selection candidates with DGV reliability 0.90 realized 91% of the genetic gain and 94% of the reduction in inbreeding. Regardless of the DGV reliability, returns at low genotyping proportions of 0.5-10% were maximized when only male candidates were genotyped. At the high genotyping proportions of 20-50%, returns were maximized by genotyping both males and females. Our findings indicate that, provided a priori information is available, only 5-20% of the selection candidates need to be genotyped to realize most of the benefits from genomic selection. At these genotyping proportions, it is best to target males in schemes when the selection intensity for males is greater than females. Our findings should be welcomed by breeders because they suggest that large investments in genotyping are not required to reap most of the benefits from genomic selection
KW - animal breeding
KW - genotyping
KW - genetic gain
KW - genomic selection
KW - inbreeding
U2 - 10.2527/jas.2012-5158
DO - 10.2527/jas.2012-5158
M3 - Journal article
C2 - 23087087
SN - 0021-8812
VL - 90
SP - 4681
EP - 4689
JO - Journal of Animal Science
JF - Journal of Animal Science
IS - 13
ER -