Designing dairy cattle breeding schemes under genomic selection: a review of international research

High rates of genetic gain canbe achieved through (1) accurate predictions of breedingvalues (2) high intensities of selection and (3) shorter generation intervals. Reliabilities of ~60% are currently achievable using genomic selection in dairy cattle. This breakthroughmeans that selection of animals can happen at a very early age (i.e. as soon as aDNAsample is available) and has opened opportunities to radically redesign breeding schemes. Most research over the past decade has focussed on the feasibility of genomic selection, especially how to increase the accuracy of genomic breeding values. More recently, how to apply genomic technology to breeding schemes has generated a lot of interest. Some of this research remains the intellectual property of breeding companies, but there are examples in the public domain. Here we review published research intobreeding schemedesignusinggenomic selection andevaluatewhichdesigns appear tobepromising (in termsof rates of genetic gain) and those that may have unfavourable side-effects (i.e. increasing the rate of inbreeding). The schemes range from fairly conservative designs where bulls are screened genomically to reduce numbers entering progeny testing, to schemes where very large numbers of bull calves are screened and used as sires as soon as they reach sexual maturity. More radical schemes that incorporate the use of reproductive technologies (in juveniles) and genomic selection in nucleus herds are alsodescribed.Themodels used are either deterministic andmore recently tend tobe stochastic, simulatingpopulationsof cattle. A key driver of the rate of genetic gain is the generation interval, which could range from being similar to that in conventional testing (~5 years), down to as little as 1.5 years. Generally, the rate of genetic gain is between 12% and 100% more than in conventional progeny testing, while the rate of inbreeding tends to be lower per generation than in progeny testing becauseMendelian sampling terms can be estimatedmore accurately. However, short generation intervals can lead to higher rates of inbreeding per year in genomic breeding programs. Received 3 June 2011, accepted 1 November 2011, published online 9 December 2011