Skip to main content

17 Feb 2021

Do genomic breeding values work in practice?

Today, more and more farmers are interested in testing their heifers genomically, Christian Bengtsson, PhD student at The Swedish University of Agriculture (SLU) and one of VikingGenetics’ breeding specialist, has given answers to this trending question: Do genomic breeding values work in practice?

The study took place in 2019 and included three breeds: Red Dairy Cattle (RDC), Holstein, and Jersey. In his research, Bengtsson compared the abilities of virgin heifer genomic enhanced breeding values (GEBV) and parent average (PA) to predict future cow performance. (Parent average is also known as pedigree index).


Genomic breeding values are beneficial

“These findings are an important step to show farmers that genomic breeding values are beneficial for use on their herds for selection decisions,” Bengtsson says.

Bengtsson explains that the traits that were considered for the study are those that farmers consider the main ones on a daily basis in their work. The traits range from milk yield and fertility to conformation, to reflect how genetic values work in practice.

The PhD project required the analysis of 12 traits in first parity:

  • Three milk production traits (milk yield, fat yield, and protein yield)
  • Two udder health traits (Somatic Cell Score and occurrence of clinical mastitis)
  • One fertility trait (interval, in days, from first to last service)
  • Two conformation traits (udder, and feet and legs)
  • One calving trait (calving ease)
  • One survival trait (survival to second calving, survival 1–2)
  • One claw health trait
  • One general health trait

“The GEBV and PA used were based on the breeding values estimated closest in time to when the animal reached one year of age because the breeding values should not be affected by their own phenotype,” the researcher explains.

“Another argument is to reflect the breeding values at first insemination for virgin heifer. For example, at that age, the farmer can combine genomic selection with decisions about sexed and beef semen, as suggested in other studies,” Bengtsson says.

Main results

The study used the phenotypes and correlations between GEBV and PA to quantify their differences. The correlations with adjusted phenotypes were 38-136% higher for GEBV than for PA in Red Dairy Cattle, 42-194% higher for GEBV in Holstein, and 11-78% higher for GEBV in Jersey.

Traits with low heritability, such as fertility, clinical mastitis, calving ease, claw health, and general health, gained relatively more accuracy from the inclusion of genomic information than the highly heritable traits like production did.

“It´s positive to be able to show that genomic breeding values don´t only work in theory but also in practice on real farms. Farmers can have confidence in using genomic technology on their herds for selection decisions,” he says.


Important management component 

Genotyping of cows and virgin heifers in DFS (Denmark, Finland and Sweden) started on a large scale in 2012. In the early years of genomic selection, mainly bulls were tested, but genotyping of virgin heifers has become more interesting as the cost of the tests decreases.

At herd level, genomic test results can be used to: 

1. Find the best females for breeding and replacement

2. Identify females for embryo transfer or in vitro fertilization

3. Correct parentage assignment

4. Control monogenic traits

5. Avoid inbreeding through genomic-assisted mating plans

These findings are an important step to show farmers that genomic breeding values are beneficial for use on their herds for selection decisions.

Christian Bengtsson,
PhD student at The Swedish University of Agriculture (SLU) and one of VikingGenetics’ breeding specialist

Scientific-based and reliable information

Three breeds in the DFS countries have genomic breeding schemes: RDC, Holstein, and Jersey. In 2018, close to 12% of females born in DFS were genomically tested, compared with approximately 2% in 2012. To date, over 250,000 females have been genotyped, and phenotypic information from over 100,000 of these animals has been recorded.

Approximately 85% of farms in DFS are enrolled in the national milk recording schemes. This enables validation of GEBV with phenotype data on a large scale.

“These numbers mean that we have enough information to draw relevant conclusions,” says Bengtsson.


Materials and methods:

  • Phenotype data was collected from the Danish, Finnish and Swedish milk recording schemes for these three breeds: RDC, Holstein, and Jersey.
  • The total number of genotyped females in the study period was 20,274 RDC, 23,910 Holstein, and 9,312 Jersey.

Table 1. Relative change in percent between phenotypes and respective parent average breeding values (PA) and genomically enhanced breeding values for RDC, Holstein and Jersey.




Relative change


Relative change


Relative change

















Clinical mastitis












Feet and legs




Calving ease




Claw health




General health




Survival 1-2