Introduction
The promise of an early and accurate prediction of genetic merit for an individual has been the subject of many years of research and development and became possible with the introduction of low cost and high density genotyping tools in 2008. Since then the rapid uptake of the technology has already transformed the global dairy breeding industry and was incorporated into the UK dairy genetic evaluation systems in April 2012. Nearly two years on from the launch of Holstein male genomic evaluations, have these early predictions lived up to their high expectation?
Genomics – The basics
Small differences in the genetic makeup of individuals provide the basis for genetic variation between individuals. Although much has yet to be learned about the exact mechanisms underlying the genetic expression, scientists have developed high throughput tools to enable us to track genetic differences on the chromosomes using Single Nucleotide Polymorphic markers (SNPs). Different density reading devices (so called SNP-chips) are available today and are now routinely used to provide a genotype profile for an individual.
In 2011 DairyCo made international genotype exchange agreements with the USA, Canada and Italy which provide the UK access to a large bull reference population. These data enable the UK to interpret the effects these SNP markers have on UK Holstein performance. Today we use information on around 23,000 bulls, which have both daughter performance evaluations and genotypes, to establish the key to
unlock these genotype differences on around 50,000 SNP positions along the chromosome. These data are used for all the traits for which we have traditional genetic evaluations. DairyCo, with its service provider EGENES at SRUC in Edinburgh, provide Holstein genomic evaluations for production, SCC, fertility and longevity and Holstein UK provide genomic evaluations for the type traits.
The genomic information is added to the traditional pedigree, performance and progeny data to provide a single genomic Predicted Transmitting Ability (PTA) on the UK base and scale, which can be used in the same way as non-genomic PTAs.
Validation of Genomic predictions
Prior to the launch of male genomic evaluations in 2012, the genomic predictions were subject to extensive internal validation studies. The evaluations were also submitted to, and successfully passed, an independent international validation test which is provided by INTERBULL in Sweden.
The validation studies indicated that significant reliability gains over traditional parent average reliabilities could be made and ranged from 32% to 41% reliability gains. These gains are equivalent to about 19 daughters contributing to a production PTA, or 300 daughters with Calving Interval data. The reason for the much larger gains for the fertility traits is because these traits have a lower heritability.
This means more daughters are needed to obtain the same level of reliability as a trait with a higher heritability, such as production traits.
How well have early predictions lived up to expectation?
In order to demonstrate the accuracy of results since the launch of UK genomic evaluations, the April 2012 official national evaluation results are compared with the December 2013 official evaluations. For the analysis, two subsets were created. The first set considered only bulls that had a genomic prediction without daughter information contributing to their PTA in April 2012, and which had daughters contributing in December 2013. (These data are referred to later in the text as Young Sire dataset.) The second dataset that was created was for bulls that had daughter proofs in both the April 2012 and December 2013 evaluations. This second dataset was further restricted to bulls that were born after 1999 and changed by more than 5% reliability for production. (These data are referred to later in the text as Proven Sire dataset.) The first dataset contained 3001 bulls, and the second set contained 1254 bulls.
The correlations from the data presented in Figure 1 for the Proven sire dataset and Young sire dataset respectively were 0.93 and 0.88, indicating that for both datasets changes to the published £PLI’s occurred. This is in line with expectation as PTAs are not published with 100% reliability. The level of reliability is a guide for the confidence that the prediction is a true estimate of the animals’ genetic merit.
The higher the reliability, the less change is expected as more data is added, as the initial estimate is already predicted to be a closer resemblance of the animal’s true genetic merit. Reliability estimates for the Proven Sire dataset and Young Sire datasets were 77% and 69% respectively in April 2012 and changed to 85% and 81% for the two datasets; or an increase of 8% and 12% respectively. These changes in reliability go together with changes in £PLI. However, the average of the predictions in April 2012 and December 2013 did not change for the daughter proven sires and averaged £54 and £54. The same statistic for the Young Sire dataset was £100 and £101 respectively. This level of consistency indicates that there is no bias in either of the two datasets, and that genomic young sire predictions are not overestimating the genetic merit of this group as some had feared.
In Figure 2 the amount of change around the mean is given for the Young Sire dataset. These data show that the majority of sires stay very close to their initial £PLI estimate, and were within the ranges anticipated given their reliability. Just over 50% were within 15 points of their April 2012 predictions, 90% within 37 points, with 5% performing below their initial prediction by -£38 or more, and the remaining 5% showing an increase of £38 point £PLI or more.
Use of Genomic Young sires
Based on insemination data recorded through the milk recording organisations NMR, CIS and UDF we can track how much young sire semen is being used in UK dairy herds, and as part of that analysis also track the genetic merit of these bulls. These data show that historically young sire usage in the UK has always been rather low and fluctuated between around 10–15% of the total number of Holstein services. However, since the introduction of genomic indexes in April 2012 we have seen a marked increase in usage and in September 2013 for the first time ever it represented over 25% of the total number of Holstein inseminations. What’s more, the genetic merit of this group of young sires is also considerably better than their proven bull counterparts and averages £188 and £146 £PLI for the two groups respectively. This £42 higher genetic merit is considerably more than the difference between the two groups used to be prior to genomic information being available. Prior to April 2012 the young sires averaged only a £12.5 £PLI point advantage in the time period between January 2000 until March 2012. This lower advantage prior to the availability of genomics was simply a reflection of the lower level of reliability of the traditional parent average index, resulting in a lower selection differential between the proven sires and young sires which were used at that time.
With genomics, the quality of young sires which AI companies are procuring is also ever increasing as a result of more rapid genetic gain of the elite breeding animals, but also because AI companies continue to increase the number of candidates they screen prior to purchase. This increased selection intensity means that only the very best candidates are entering the AI stations, and only those will ultimately be made available for use. Taking a closer look at the average £PLI of young sires that have been available in the UK since April 2012, provides further proof of this increased quality. Table 1 shows that the genetic advantage today is £45 points £PLI, which is three times the superiority of the same comparison when genomics was first launched in the UK in April 2012.
The combination of increased usage of young sires in the UK and their genetic advantage is ultimately contributing to a significant acceleration of genetic gain that the national herd is experiencing as a result of genomics. The results presented in this paper on stability of young sire predictions confirm that this growing confidence and use of young sires in breeding programmes is not misplaced and indications are that the uptake of young sires as part of a herd’s breeding plan is continuing to increase. Similar statistics of young sire use in North America indicate that producers there are already serving around 50% of their cows to young sires.
The level of use of young sires as part of a herd’s breeding plan is a choice each producer will have to make for themselves, based on their attitude towards risk and reward. The slightly lower reliability for young sires will have to be weighed up against their genetic superiority. The choice of bull should first and foremost be dictated by some simple criteria; firstly, only pick bulls which are of high genetic merit and secondly, use the same criteria for both proven and young sires. This pre-selection should shortlist a group of possible candidates to choose from and will likely contain a mix of both daughter-proven and young sires. How much semen to buy from each of these will subsequently depend on the reliability of their indexes and Figure 3 can be used as a guide. The final recommendation is to always use a team of 4 or 5 young sires, and never gamble on one single ‘hot’ prospect.
Breeds other than Holstein
The large population size of the Holstein is providing a distinct advantage for the breed in establishing a reliable reference set to calibrate the genomic information on. However, other breeds are also making use of the technology and DairyCo is actively looking for partners to cooperate with in establishing UK genomic evaluations for non-Holstein breeds. Research is on-going to establish if British Friesian genomic evaluations can be launched during 2014, and in parallel, representatives of the Brown Swiss breed are in discussion with InterGenomics, which is a genomic evaluation service offered through Interbull for their breed.
Future
The application of genomic technologies has only just begun and with the price of genomic testing still falling, the likelihood is that the uptake of testing both males and females will become routine practice in the not-too-distant future. The spin-off benefits beyond breeding are vast and include amongst others, pedigree validation, traceability, selection of calves for rearing and screening for genetic abnormalities. It is also not unthinkable that other farm advisers such as feed consultants and vets will become interested in the genomic profile of a herd or an individual, in order to provide more tailored advice on either feeding or drug application.
Summary
Genomics is already having a dramatic impact on the dairy breeding industry and its importance is set to continue. The demonstrated stability and genetic superiority of genomic young sires highlights that their use is justified and to be encouraged. However, the rapid acceleration in genetic gain also comes with a need to closely monitor the direction and level of change. The DairyCo Genetics Advisory Forum, which is made up of industry stakeholders, will therefore continue to monitor both genetic and phenotypic performance of the national herd.
The benefits genomics can offer the UK industry in future is expected to be great, but in order to ensure the UK industry capitalises fully on the benefits genomics can deliver, a joined-up approach is needed and continued national recording schemes are essential to maintain.