Improvements to Genomic Evaluations

“Bryan Kingston (ICBF) discusses the improvements to dairy genomic evaluations this Spring”

Genomic selection has contributed significantly to the increased rate of genetic gain in EBI, through the more accurate identification of genetically elite males and females. The current rate of genetic gain in EBI is €10/year. 

How does genomics work?

Genomics is breeding using DNA to help better predict how an animal will perform in the future. DNA is passed from parents to offspring and is therefore central to breeding. The DNA profile of an animal is analysed and is compared to the DNA profiles of proven animals in the training population and looks for similarities. Performance data, ancestry data and genomic data are combined on the animal itself generating a more accurate prediction of the animal’s genetic make-up.

Updated Genomic Training Population

To generate genomic evaluations, a training population needs to be established. The training population is a large population of genotyped animals with accurate performance information such as milk yield or fertility. The associations between the DNA and performance measures are then developed from this population. Previously, there were in the region of 10,000 informative sires within the Irish Holstein-Friesian training population. The criteria for including animals to this population has also remained the same since dairy genomics was first launched over 10 years ago. After several years of research involving ICBF and Teagasc, the size and criteria of the training population has now been updated. There are now over 45,000 and 50,000 animals in the training population for milk and fertility traits respectively.  Increasing the size of the refence population is essential to ensure genomic predictions are accurate.

Addition of Females and Multiple breeds

Genotyped females and other dairy breeds such as Jerseys and Norwegian Reds are now included in the Irish training population for the January2020 evaluations. Many other countries have also started to include informative females into their reference populations.  Incorporating females to the training population is adding a significant amount of information and is a crucial change to the genomic predictions. The addition of genotyped cows with good quality phenotypic data to the training population will further increase the accuracy of genomic predictions for traits such as female fertility (Table 1). Dairy genomic evaluations were previously limited to Holstein-Friesians, which don’t have more than 6.25% of any other breed. However, the inclusion of multiple breeds to the training population now means that genomic evaluations can be published for other dairy breeds such as Jerseys, Norwegian Reds, Montbeliardes and Ayrshires.

ICBF have now developed a system to routinely add male and female genotypes and their records to the training population which will further increase the reliabilities of genomic predictions. A further improvement to genomic predictions is the continuous genomic blending approach. This means the system is constantly integrating an animal’s own performance records with its genomic data.  This impact will result in fewer sudden changes to individual bulls from one evaluation to the next.

Accuracy of Predictions

A recent validation exercise revealed the accuracy of genomic evaluations is 16-44% more accurate than evaluations based only on non-genomic proofs.  The validation was based on data from the current evaluations for 262 sires born after 2010 with at least 50 daughters in milk. The accuracy has improved for all traits which can be seen in Table 1. This clearly shows the benefit of including genotyped cows in the training population.

Table 1.  Accuracy of predictions (Range 0 – 1) for 262 proven bulls born after 2010.


Impact of Improvements on Active AI sires

The addition of new data to the training population for milk traits is very similar to the previous information for all Holstein-Friesian sires. For the fertility traits, there is a much larger difference. The impact of adding females to the training population has shifted the distribution of the training population because the females added are more reflective of the current population of dairy cows. The new improvements will bring about change in sire rankings on the Active bull list. Table 2 shows the average change for 1,118 Active sires for the EBI and the Milk and Fertility index and associated reliabilities. Figure 1 shows the distribution of the change (for example the central category with the highest sire count (180 sires) are sires that moved between -€12 and -€2 euro). 50% of sires moved between -€23 and +€8. 80% of sires moved between -€46 and €26.

Index Nov-19 Jan-20 Difference
EBI €156 €147 -€9
EBI reliability 73% 79% 6%
Milk SI €57 €55 -€2
Milk reliability 83% 91% 8%
Fertility SI €57 €57 €0
Fertility reliability 66% 73% 7%

Table 2. Active Bull list averages Nov 2019 v Jan 2020.

Figure 1. Distribution of change in EBI for Active sires

Despite the recent increases in reliability with genomic selection, it’s vital to know that the EBI of an individual animal can still change over time as more information gathers. It’s important to minimise risk associated with genomic selection by using a large team of bulls. For further queries call the ICBF HerdPlus team on 023 8820452.

In Short                                                                                                                                          

  • ICBF have updated the size and criteria of animals included in the training population which will increase the accuracy of EBI and generate greater genetic gain in the future.
  • New updates will have minimal impact on the milk sub-index but a larger impact on the fertility sub-index.
  • Genomic evaluations are now available for multiple dairy breeds.
  • Improvements to genomic predictions will give more confidence in breeding decisions especially for young GS bulls.
Close Menu