Forecasting the Future of Genomic Selection

By George Wiggans, Ph.D.

The use of genomic evaluations has transformed dairy cattle breeding in the United States and worldwide. Bulls are used widely as sires based on the analysis of their DNA before they have any milking daughters. Many dairy producers genotype all their heifers so that they can select among a range of breeding and management strategies. As this is today’s reality, where is genomic selection heading for the future?


Current Snapshot

The popularity of the genomic evaluation program has resulted in an amazing growth in genotyping of dairy cattle. When genomic evaluation began in 2008, the focus was on bulls. Genotyping of females has grown rapidly in recent years, and nearly half a million genotypes were received by the Council on Dairy Cattle Breeding (CDCB) in 2016.


Figure 1: Genotypes Submitted for Use in U.S. Genomic Evaluations


Genomic selection has also dramatically changed the artificial insemination (AI) industry. The market share of genomic young bulls has increased steadily and in 2016 was two-thirds (67%) of recorded inseminations. Nearly all progeny-tested bulls have been genotyped.


Figure 2: Service-Sire Profiles for Recent Breedings


Most animals are genotyped with chips that have around 21,000 (21K) single-nucleotide polymorphisms (SNPs). However, some producers use the less expensive 7K chip. The chips with more than 21K SNPs are used for bulls considered for marketing or used in research projects. In some cases, animals are re-genotyped with higher-density chips so that they can qualify for evaluation. For lower-density genotypes, the breed-purity requirement is stricter than for higher density because imputation – the process of filling in missing SNPs – is breed specific and has a larger role.


Figure 3: Chip Densities for Genotypes Received in 2016 for Use in Genomic Evaluation



When evaluations are available shortly after birth, breeding and herd management decisions can be made sooner. If a calf is going to be culled, that decision should be made as soon as possible to reduce feeding and rearing costs.  Generation interval is measured by the age of parents when offspring are born and affects the rate of genetic improvement. The shorter the generation interval, the faster progress can be made. A few years after genomic evaluations were introduced in 2008, parent ages for bulls began to drop dramatically and are now near the biological minimum.


Figure 4: Holstein Generation Intervals


Genomic selection has produced a large increase in genetic trend as indicated by the average net merit of bulls being marketed. The $80 annual trend for Holstein bulls that entered AI since 2011 is nearly double that of the previous 6 years ($47), which was already a substantial improvement over the $18 average gain in 2000–2004.


Figure 5: Gain in net merit (NM$) for marketed Holstein bulls


The CDCB goal is to provide the most useful information for breeding and management decisions so that the efficiency of dairy production can be optimized. Although we’ve seen significant progress to date, we can expect further improvement in future genomic evaluations.


Increase accuracy

Genomic evaluation is successful because we have enough DNA markers available for tracking the segments of chromosomes associated with high performance across generations. SNPs are used as markers because of their reliability and low cost. However, the linkage decays over time between SNPs and genomic regions associated with high performance because of ordinary DNA rearrangement from crossing over in chromosomes (recombination). To counteract this, new data are needed so that the SNP effects can be re-estimated to maintain or improve evaluation accuracy. Now that the cost of whole-genome sequencing is lower and thousands of animals can be sequenced, it should be possible to find SNPs that are more closely associated with the alleles (genetic variants) that cause differences in performance (or even the causative variants themselves). In the next few months, CDCB intends to increase the number of SNPs used in evaluations, which will improve the ability to track the causative variants and provide a small increase in the reliability of genomic evaluations.

Genomic evaluations generally do not change much between releases because they are derived primarily from an animal’s own genotype and estimates of marker effects based on the entire predictor population (animals with both genotypes and traditional evaluations). An exception is when the pedigree changes because this affects the parental contribution from traditional evaluations. To reduce changes after an animal’s evaluation has been released, animals with an unlikely grandsire will not receive evaluations beginning in September. For registered animals, genotyping the parent (usually the dam) will be recommended. If an unlikely grandsire cannot be resolved, the parent or grandsire can be designated as unknown. An evaluation with an unknown parent or grandsire is more accurate than an evaluation based on incorrect pedigree information.


Evaluate more animals

Because of genetic differences among breeds, genomic evaluations have been separate by breed and limited to animals with a very high percentage of their DNA from within that breed. Many herds have animals with varied breed ancestries (crossbreds), and those herd owners would like genomic evaluations for all their animals. The scientists at the Animal Genomics and Improvement Laboratory (AGIL), part of USDA’s Agricultural Research Service, have developed an evaluation procedure that blends the SNP effect estimates from the various breeds of an animal according to the percentage of each breed (breed base representation). Genomic evaluations for crossbreds will be calculated on an all-breed base, blended across breeds and then adjusted to an individual-breed base. CDCB is considering implementation in 2018.

Evaluate more traits

For traditional genetic evaluations, an animal’s evaluation for a trait can only be more accurate than its parent average if that trait is observed for the animal or its offspring. With genomics, evaluations can be generated for all genotyped animals if enough animals have genotypes and traditional evaluations for the trait. AGIL and CDCB have been collaborating to research and develop evaluations for several new traits. In August, CDCB plans to release genetic evaluations for gestation length – of particular interest for grazing herds, for which a short calving season is preferred. A December release is planned for evaluations of resistance to six health conditions: hypocalcemia (milk fever), displaced abomasum, ketosis, mastitis, metritis and retained placenta. AGIL scientists also are working on a genetic evaluation for feed efficiency by building on feed intake data collected through a joint project by several U.S. universities. CDCB is investigating the possibility of continued data collection by the project contributors so that the database can be updated and expanded.


Figure 6: Recent and upcoming traits
Cow livability (August 2016)

Gestation length (August 2017)

Health traits (December 2017?)

Feed efficiency (under development)


Increase the predictor population

Early in the development of the U.S. genomic evaluation system, arrangements were made to share genotypes between countries to increase the size of the predictor population. From the beginning, all U.S. genotypes have been shared with Canada. For Holsteins, sharing is ongoing with Italy, the United Kingdom, Switzerland and Germany. The predictor population for Holsteins is so large that the value of adding older animals has declined, but the addition of younger animals is still beneficial. New arrangements for sharing genotypes will be made with other countries primarily for marketing benefits, but the additional data will also add to evaluation accuracy. The greatest benefit may come from sharing phenotypes, especially for feed efficiency because of the high cost of data collection.

Improve mating decisions

CDCB currently provides information on the genomic relationships between potential dams and currently marketed bulls. These data report the actual portion of genetic variants (alleles) in common and not the average based on relationships. This allows avoiding inbreeding more precisely. CDCB also provides predictions for a number of recessive conditions so that likely carrier-to-carrier matings can be avoided even without testing an animal. The haplotypes that affect fertility are conditions discovered through genomics and can now be considered in matings. In the future, mating programs may also consider the effects of dominance, which causes some sire-maternal grandsire combinations to do better than expected and others to do worse.


Better and more data

Herds with high levels of genotyping generally have fewer misidentified sires in their data that contribute to traditional evaluations. With better data, less information is lost from the extensive checks done by CDCB to eliminate unreliable and inconsistent data. Genotypes are checked against all other genotypes to ensure they are assigned to the correct animal and that the parents are correctly identified. Data problems can include reporting the same ID for different cows, not identifying that sexed semen was used for an insemination and reporting the transfer of an embryo to a recipient as a normal breeding. Greater knowledge of data usage should help providers better understand the importance and benefit of accurate data collection.

In addition to data accuracy, comprehensive reporting is important. Although genomics provides evaluations for animals without an observed trait, someone must still provide the phenotypic data. For the health trait evaluations that CDCB intends to release later this year, their accuracy will benefit from much broader data submission. Because the entire dairy industry benefits from more data, additional recognition of those that contribute may be justified to increase the data available. Currently, CDCB fees are lower for those that contribute more, and CDCB support for dairy records processing centers varies by quantity and completeness of data. CDCB plans to invest further in data collection by covering part of the cost of collecting feed efficiency data.



Dairy genetics have been transformed by genomics. Genomic evaluations allow determining the value of animals at a much earlier age and have contributed to a dramatic increase in the rate of genetic improvement. A continuing stream of improvements are planned to increase accuracy and comprehensiveness of genomic evaluations. Success requires a partnership between data suppliers and users to generate the most effective information for all.


George Wiggans, Ph.D., is a technical advisor for the Council on Dairy Cattle Breeding (CDCB). Contact him at 240-334-4164, ext. 322, email:


Additional Resources

Council on Dairy Cattle Breeding website:

Animal Improvement Program website:

Wiggans, G.R., J.B. Cole, S.M. Hubbard and T.S. Sonstegard. 2017. Genomic selection in dairy cattle: The USDA experience. Annual Review of Animal Biosciences 5:309–327.

Cole, J.B., P.M. VanRaden, D.J. Null, J.L. Hutchison, T.A. Cooper and S.M. Hubbard. 2016. Haplotype tests for recessive disorders that affect fertility and other traits. AIP Research Report Genomic3 (09-13), updated May 2, 2016.

Chesnais, J., T.A. Cooper, G.R. Wiggans, M. Sargolzaei, J. Pryce and F. Miglior. 2016. Using genomics to enhance selection of novel traits in North American dairy cattle. Journal of Dairy Science 99:2413–2427.

VanRaden, P.M., and T.A. Cooper. 2015. Genomic evaluations and breed composition for crossbred U.S. dairy cattle. Interbull Bulletin 49:19–23.