Introduction
Genomics has revolutionized dairy cattle breeding by dramatically shortening the generation interval and increasing the rate of genetic progress. The US–Canadian database includes over 850,000 genotypes, which are used to evaluate Holsteins, Jerseys, Brown Swiss and Ayrshires. Parentage validation based on genotypes improves the accuracy of pedigrees and, therefore, traditional evaluations. Once a genotype has been obtained, it can be used in genomic evaluations for all traits. To maintain prediction accuracy in current conditions, continued collection of performance records is needed for existing traits so that the effects of genetic markers can be re-estimated frequently. The cost of genotyping has decreased as new genotyping chips have become available. Later versions of existing chips usually include more markers at the same price. In the future, genomic evaluation will be available for more traits, which will allow better tracking of profitability. Weekly genomic evaluations have become available in the Netherlands, Germany and the US to support efficient culling (Wiggans et al., 2015). Accuracy of genomic evaluations will improve as causative genetic variants are discovered and included in the calculations. Variants discovered in the Holstein and Jersey breeds may be useful for breeds with few genotyped animals. Tracking of individual variants holds great promise for avoiding harmful effects and promoting beneficial ones.
US genomic evaluation program
In the US, the Council on Dairy Cattle Breeding (CDCB; www.cdcb.us) is responsible for the genomic evaluation program. USDA provides research support through the Animal Genomics and Improvement Laboratory (formerly the Animal Improvement Programs Laboratory, aipl.ars.usda.gov). CDCB has authorised 16 organisations (nominators) to submit animals for genomic evaluation. Nominators include breed registry associations, AI organisations and genotyping laboratories. Four genotyping laboratories provide most of the genotypes. Figure 1 shows the number of animals submitted monthly for genomic evaluation for the past year. Each genotype is compared with every other one to confirm or discover parent-offspring relationships. Breed and sex also are validated. The identity of the maternal grandsire is checked, and bulls are suggested if the grandsire is unknown or the reported grandsire appears to be incorrect. This validation minimizes the chance that the genotype was collected from the wrong animal and enables correction of pedigree information.
The genomic evaluations are based on 60,671 genetic markers called single nucleotide polymorphisms (SNP). Most of the genotypes are from genotyping chips with a lower density of SNP, which reduces the cost of genotyping. The current low-density genotyping chips provide 6,787 to 13,218 SNP that are used in evaluations. Any missing SNP are filled in through a process called imputation, which uses genotypic information from ancestors and offspring to infer the missing SNP. Bulls that have both genotypes and traditional evaluations are the most informative source of information for estimating the difference in performance from having one variant (allele) versus the other. Cows with US traditional evaluations and genotypes also are used to estimate these SNP effects. The alleles are represented as A and B. The direct genomic component of an evaluation is calculated by summing the number of A alleles for each SNP and then multiplying by the estimated SNP effect. Between 10 and 15% percent of genetic merit is assumed not to be captured by SNP effects, and that portion is estimated as a polygenic (controlled by more than one gene) effect. To create the final evaluation, the estimates for SNP and polygenic effects are combined with information from traditional evaluations that is not captured by genomics.
In the US, genomic evaluations are calculated for yield, functional, calving and type traits. Yield traits include fat and protein percentages as well as milk, fat and protein yields. Functional traits include mastitis resistance (somatic cell score), fertility (heifer and cow conception rates and daughter pregnancy rate) and longevity (productive life). Calving traits include calving ease and stillbirth. The specific type traits included vary by breed. In addition to genomic evaluations for Holsteins, Jerseys, Brown Swiss and Ayrshires, the possibility of calculating them for Guernsey’s is being investigated.
Predictions for traits affected by single genes are generated as part of the imputation process and can be used to track carrier status. These include haplotypes (DNA sequences inherited from one parent) for recessive conditions that affect fertility and other traits (Cole et al., 2014). For Holsteins, the haplotype tests include bovine leucocyte adhesion deficiency (BLAD; haplotype HHB), complex vertebral malformation (CVM; haplotype HHC), deficiency of uridine monophosphate synthase (DUMPS; haplotype HHD), mulefoot (syndactyly; haploptype HHM), polledness (haplotype HHP) and red coat colour (haplotypes HBR, HDR and HHR). Brown Swiss haplotype tests include spinal dismyelination (SDM; haplotype BHD), spinal muscular atrophy (SMA; haplotype BHM) and Weaver Syndrome (haplotype BHW). Two Jersey haplotypes (JH1 and JH2) affect fertility, and Ayrshire haplotype (AH1) affects conception rate.
Generation interval
Genomic evaluations have been widely accepted by dairy cattle breeders. In the US, bulls with only a genomic evaluation are used to breed over half of the cow population. Additionally, AI organisations frequently use genomic bulls and virgin heifers as parents of the next generation of bulls. These changes have led to a progressive reduction in the generation interval (Figure 2), which has led to an increased rate of genetic gain.
Accuracy
The confidence that breeders have in genomic evaluations is supported by investigation of historical data. Bulls with a high rank for their genomic evaluations generally retain their rank when they receive an evaluation based on progeny. Net merit in December 2012 and December 2014 was compared for 642 Holstein bulls that received their first evaluations based on daughter records in August 2014 and had 50 daughters or more in December 2014. For the top 100 bulls in 2012, the average change in rank was 9.6. For all 642 bulls, the 2012 and 2014 evaluations were correlated by 94%. The net merit differences between bulls in December 2012 generally were similar in December 2014.
Genomic mating programs
Mating programs for genomic selection can minimize genomic inbreeding by comparing genotypes of potential mates. In the US, files of genomic relationships between genotyped females and bulls likely to be used for breeding are available so that breed associations and AI organisations can use them in mating programs. In addition, methods have been developed to consider dominance effects of individual markers when assigning mates to improve offspring merit further (Sun et al., 2013). Mating programs that include genomic relationships are more effective than those using pedigree relationships because they improve the expected value of offspring as well as decrease expected offspring inbreeding. The expected decrease in inbreeding currently is worth over $3 million annually for US Holsteins. That economic value will grow as more cows are genotyped.
DNA sequencing
The technology for determining the exact identity of DNA nucleotides (sequencing) has improved in recent years, and a bull’s DNA sequence now can be determined for around US$1,000. This affordability provides great opportunities for research because the actual causative genetic variants can be discovered for traits of interest. That knowledge should aid in improving the accuracy of genomic evaluation because the current problem that results from decay in the association between the genetic marker and the causative mutation would be overcome. Another possible benefit of DNA sequencing is that across-breed evaluations would be more practical because the problem of breed differences in the association of markers with causative variants would be eliminated.
Whole-herd genotyping
In the US, over 80% of genotyped animals are female because many herd owners genotype all their animals. The genomic information allows them to classify their heifers into (1) those to flush, (2) those to breed with sexed semen, (3) those to use as embryo recipients or breed to beef bulls and (4) those to cull. This use of the information for management decisions justifies the cost of genotyping. The evaluation system overall benefits from wide-spread genotyping because pedigree errors are corrected, which makes traditional evaluations more accurate. The development of low-density genotyping chips was critical for the increase in genotyping of females. Investigation continues on ways to reduce the cost of genotyping, specifically through developing chips with approximately 5,000 SNP. A reduction in cost would increase the number of herds that find whole-herd genotyping profitable.
The future
Genotyping provides a powerful tool for determining the genetic potential of dairy cattle. Economic performance can be improved while avoiding expression of harmful recessives. With increased evaluation accuracy and reduced generation intervals, dairy cattle can be adapted for changing environmental conditions, such as increasing temperatures as the result of climate change. Precision selection for milk with characteristics required by niche markets will become more practical. Genomics increases the value of phenotypic data while removing the need for its collection just to get an evaluation. The industry must provide incentives to generate phenotypic data for new traits (such as feed efficiency) as well as current traits, which still need up-to-date information so that SNP effect estimates remain accurate for the current population under current conditions. Genomic mating programs offer a way to determine the best bull to use for a specific cow with interactions of their genotypes considered and harmful recessives avoided.
Improvements in genotyping technology are expected to provide low-cost genotyping chips for first-stage screening, intermediate-density genotyping chips with more SNP for the same cost and full-sequence data to support discovery of more informative SNP. Because a genotype provides information for all traits, it can be used to estimate genetic merit of traits for which the animal does not have a performance record. Thus, the benefit of recording performance for additional traits that affect economic merit is increased because evaluations can be generated for all genotyped animals. All relevant traits can be included in an index to enable selection of the most profitable animal. As understanding of the value of phenotypic data increases, some herd owners may specialize in the collection and sale of data. For example, some traits (such as feed intake) have such a high cost for data collection that it cannot be justified for the management of a single herd.
Conclusions
Genomic evaluation has been very successful in providing accurate predictions of genetic merit. Participation in the US increases nearly every month. Ongoing research along with larger predictor populations for estimating effects of genetic markers is expected to increase accuracy. The discovery of causative genetic variants will improve avoidance of undesirable recessives, contribute to accuracy of genomic evaluation and possibly make across-breed evaluation more practical. Using genomic information in mating programs has the potential for significant financial gain. Continued improvement in genotyping technology may reduce the cost of genotyping and make whole-herd genotyping financially attractive for most dairies. The extension of genomic evaluation to more traits will enable more accurate selection for overall genetic merit and for traits of interest to niche markets. The capability that genomics provides for more rapid genetic change will enable more rapid adaptation to changing requirements.
References
Cole, J.B., VanRaden, P.M., Null, D.J., Hutchison, J.L., Cooper, T.A. and Hubbard, S.M. (2014). Haplotype tests for recessive disorders that affect fertility and other traits. AIP Res. Rep. Genomic3 (09–13). http://aipl.arsusda.gov/reference/recessive_haplotypes_ARR-G3.html.
Sun, C., VanRaden, P.M., O’Connell, J.R., Weigel, K.A. and Gianola, D. (2013). Mating programs including genomic relationships and dominance effects. J. Dairy Sci. 96: 8014–8023.
Wiggans, G.R., VanRaden, P.M. and Cooper, T.A. (2015). Technical note: Rapid calculation of genomic evaluations for new animals. J. Dairy Sci. 98: In press.