Introduction
Dairy cattle breeding has, since quite a number of years, been characterised by extensive internationalisation. The possibility to export frozen semen and embryos made it relatively easy to introduce foreign genetic material to breeding programs in individual countries. In several cases this was rather massive and a popular example is the so-called "Holsteinisation" of several of the old Friesian populations of Europe.
The need to identify and select genetically superior individuals regardless of their national origin, or commercial exposure, was early recognised. Results from national genetic evaluations are, however, not easily comparable due to differences in:
- genetic levels among cattle populations;
- recording and evaluation procedures in various countries (e.g. 1st vs. all lactations, single-trait vs. multiple-trait);
- breeding value expression (e.g. reference base definition, unit of expression);
- animal performance under variable environmental conditions and production systems.
The awareness of the problems and needs led to the formation of Interbull, the International Bull Evaluation Service, in 1983 and its "acting body" the Interbull Centre in 1991. The history of this process is thoroughly described by Philipsson (1998), and it suffices here to recognise that organisations like the European Association for Animal Production (EAAP), the International Dairy Federation (IDF), International Committee for Animal Recording (ICAR), and the United Nations’ Food and Agriculture Organisation (FAO) were active participants.
The objectives of this presentation are to review the current role and activities of Interbull, to identify some possible effects of the activities, and to outline some future challenges.
Role and activities
The objectives of Interbull are to:
- facilitate international communication through meetings, workshops, surveys, publications, etc.;
- conduct research and development work in the area of international genetic evaluation;
- provide international evaluations/comparisons of dairy cattle populations;
- provide the member countries with technical support on all matters related to the genetic evaluation of cattle populations.
The third objective, providing international evaluations, is the one that has probably attracted most attention and has had most impact on the international exchange of genetic material (at least bull semen). The Interbull Centre, located in Uppsala, Sweden, performs these evaluations where results of national genetic evaluations from various countries are combined using Multiple(-trait) Across Country Evaluation (MACE). Performance in each country is considered a different trait, allowing for different genetic parameters in different countries and genetic correlation of less than unity among countries (Schaeffer, 1994). The evaluation system of Interbull has been developed and established at the Interbull Centre as result of collaborative and/or independent research conducted at different institutes worldwide. Such results have been gradually integrated at the Interbull Centre into the routine international genetic evaluation process.
The MACE approach uses information on all bulls from all countries, i.e. not only a sample of selected bulls with potentially biased genetic evaluations, as used in methods based on conversion equations. Another advantage is that the procedure uses information on all known male relationships between bulls, both within and across populations, as well as daughter performance. For example, bulls with daughters in several different countries provide information about their own genetic merit and provide comparisons with other bulls.
A major advantage of the MACE method to calculate international genetic evaluations over older methods, is that different country scales and genotype by environment interaction can be taken into account. MACE thus allows for the possibility of animals re-ranking between certain countries. This occurs when animals perform better in certain environments than they do in others. For this reason, a separate set of results is calculated for every participating country. This process is demonstrated in Figure 1.
A separate list of international genetic evaluations for all traits and all bulls evaluated is computed for each participating country, expressed in their own units and relative to their own reference base group of animals. This provides the advantage that individual countries are able to identify those animals from around the world that will perform best under their own unique farming conditions, and in a form their farmers and sire analysts are accustomed to.
Figure 1 shows that the international genetic evaluations calculated for bulls from countries A and B and their subsequent rankings can be different from one country to the other. Such difference is caused by genetic correlations less than unity between countries, and is exemplified by the change in ranking for bulls B5 and B6 in the figure. Bulls also usually compare better in their home country than abroad when correlations are less than unity (exemplified by bulls A3 and B6 above), since more weight is put on local information. When the correlation between countries is unity the ranking of bulls is exactly the same in each country.
In general, a low correlation between countries leads to less weight on foreign information, more re-ranking, and reduced usefulness of international genetic evaluations. However, genetic diversity will be higher when correlations between countries are low. There may be many good reasons why correlations are low: different selection objectives, different scoring, true interactions between genotype and environment, etc. Whatever the reason, a major feature of international evaluations is the fact that such interactions are taken into consideration.
The first step in the routine evaluation process is collection of national data from various countries. Checks of incoming data, and on animal identification, are performed. Using a comprehensive cross-reference list, animals receiving different registrations in various countries are identified and assigned a unique ID number. De-regression of national genetic evaluations to remove effects that will be subsequently included in the international evaluation model is the next step in the process and are performed for each trait independently. At the same time, genetic variances are estimated with country (Sigurdsson et al., 1996; Klei and Weigel, 1998; Sullivan, 1999). International genetic evaluations are then computed based on the de-regressed national results and the estimated genetic parameters. Files containing results for all bulls, in the base and units for all countries, are produced and returned to national evaluation centres for distribution to breeders and farmers; individual countries may only publish Interbull evaluations expressed on their own official scale and base. For example, in 2002 about 70,000 Holstein Friesian bulls from around the world received international genetic evaluations for protein yield expressed in PTA lbs on the US base, in EBV kg on the Dutch base, in RBV on the Swedish scale and so on. It should be noted that Interbull does not have the right to provide genetic evaluation results or ranking of bulls directly to the "end-user", since the results are the property of each member.
Routine international genetic evaluations are performed four times per year (February, May, August, November) and each evaluation has a turnaround time of 10 days. Test-runs are performed, in addition to routine evaluations, in March and September. Such test-runs have a turnaround time of about 1 month and their purpose is to provide the opportunity for new countries or traits to be included, to check modified national evaluations, and to introduce modifications to the international genetic evaluation process.
The first routine international genetic evaluation results were released in August 1994. Only four countries (Denmark, Finland, Norway and Sweden) and two breeds (Ayrshire and Holstein) were included. The service has been expanded considerably since then and 25 countries and six breeds participated in the latest routine evaluation for production traits in November 2002 (Table 1).
Table 1. Breeds, number of populations and number of bulls in Interbull evaluations for production traits in November 2002
| Breed |
No. of populations |
No. of bulls |
| Holstein |
26 |
72,404 |
| Ayrshire |
11 |
9,941 |
| Jersey |
10 |
6,184 |
| Brown Swiss |
9 |
5,703 |
| Simmental |
9 |
19,710 |
| Guernsey |
6 |
803 |
Data from Holstein populations dominates the services, both in terms of number of populations and number of bulls. There is, however a very large variation in the size of the various Holstein populations with the United States contributing most bulls (>18,000) and Slovenia least (~100). European countries are in majority, but most parts of the world are represented, with notable exceptions being Asia and South America (Table 2).
Table 2. Holstein populations participating in Interbull evaluations for production traits in November 2002
| Country |
No. bulls |
|
Country |
No. bulls |
| Australia |
3398 |
|
Israel |
583 |
| Belgium |
430 |
|
Italy |
4338 |
| Canada |
5408 |
|
Netherlands |
6739 |
| Czech Rep. |
1599 |
|
New Zealand |
3149 |
| Denmark |
4726 |
|
Poland |
3851 |
| Denmark, RHOL |
155 |
|
Rep. South Africa |
628 |
| Estonia |
268 |
|
Slovenia |
98 |
| Finland |
648 |
|
Spain |
826 |
| France |
8344 |
|
Sweden |
1293 |
| France, RHOL |
109 |
|
Switzerland |
538 |
| Germany (+Austria) |
11067 |
|
Switzerland, RHOL |
617
|
| Hungary |
1350 |
|
Unit.Kingdom |
3058 |
| Ireland |
682 |
|
Unit. States |
18193 |
Genetic correlations for milk production traits between countries are generally high, ranging from 0.85 to 0.96 among countries in the Northern hemisphere, around 0.90 among countries in Oceania, and from 0.75 to 0.84 between Northern and Southern hemispheres. However, the correlations are still sufficiently low to change the international bull ranking from country to country.
International genetic evaluations were originally only provided for milk production traits, i.e. milk, fat and protein yield. However, research showed that international genetic evaluation for conformation traits was also feasible, and conformation trait evaluations for the Holstein breed were introduced in 1999. This service attracted 10 countries in the first official evaluation in August 1999, but has since expanded and 20 populations participated in the latest routine evaluation in November 2002. In 2001 conformation evaluation was also performed for the Jersey breed, and in 2002 conformation evaluation for the Brown Swiss and Guernsey breeds started. The large number of traits (18) and countries means that this service is computationally very demanding and part of it is therefore outsourced to a North-American consortium where Holstein Association-USA is the "acting body".
Genetic correlations for conformation traits between countries are generally lower than for production traits, but also much more variable. Correlations for teat length and foot angle are the extremes in this respect with correlations for teat length ranging from 0.88 to 0.99 and for foot angle from 0.01 to 0.88. These variable correlations probably to some extent reflect the degree of trait harmonisation, but may also be due to a true genotype by environment interaction. The low correlations between countries for some conformation traits, limit the usefulness of the international evaluation for these traits.
Up to now, emphasis of the international genetic evaluation at Interbull has been on production, although more recently also on conformation traits. However, many studies have shown antagonistic relationships between production on one side and health and fertility on the other; common estimates of genetic correlations have been in the range from 0.2 to 0.4. It is therefore important that Total Merit Indexes include health and fertility traits, in order to optimise economic profit, and that international genetic evaluation for functional traits becomes available.
Interbull has initially focused on clinical mastitis and milk somatic cells as potential functional traits for international genetic evaluations. Results from international research have been presented (Mark et al., 2000ab; Mark et al., 2001) and routine international genetic evaluations for udder health traits were first conducted in May 2001, and include now data from 5 breeds and 18 populations.
Genetic correlations between milk somatic cells in different countries were generally high and ranged between 0.75 and 0.97, while genetic correlations between milk somatic cells in one country and clinical mastitis in a different country were generally lower and ranged from 0.42 to 0.80. International genetic evaluations for udder health traits have been shown to be feasible and will probably prove valuable, despite that correlations are lower than for milk production traits.
It is likely that international genetic evaluations for other economically important traits will be added in the future. Conformation trait evaluations for more breeds, as well as calving performance and longevity are now being investigated, and female fertility is likely to come as well, since national evaluations for such traits are increasingly common. Key factors in the continuing expansion of number of traits in international genetic evaluations are data availability and harmonisation of trait definition and national evaluation methods, and surveys have shown that improvement in these areas is called for (Banos, 1999).
Consequences
Interbull routinely receives pedigree data on active sire populations for six dairy breeds as an integrated part of the international genetic evaluations, and is thus in an excellent position to monitor trends in breeding practices, population structures and genetic relationships on a global level. This is a task that could be made into a continuous process as part of the routine activities at the Interbull Centre, and relevant research has been initiated.
Effects on genetic gain
Effects of international genetic evaluations on long term selection responses in a global population have been investigated in simulation studies. Results indicate that numerically smaller and genetically poorer populations would benefit the most from international selection (Banos and Smith, 1991), but large, traditionally superior exporting populations would also benefit, although to a lesser extent (Wickham and Banos, 1998). The intensity of international selection will be higher than in any national breeding scheme, leading to faster progress, since the number of international selection candidates is much larger than bulls sampled locally in any country. For example, about 6,000 Holstein bulls are progeny tested worldwide each year in the countries actively participating in international genetic evaluations. In each country separately, this number varies from less than 10 to more than 1,400.
Selection differentials for protein yield in the Holstein breed, based on data from the routine Interbull evaluation of August 2001, were calculated as in Mark et al. (2001), but using only bulls born in 1994 and 1995 for the national average. Selection differentials ranging from 0.39 to 1.49 genetic standard deviations would be achieved by selecting the top 10 bulls nationally, while the corresponding range when selecting top 10 bulls internationally would be from 1.39 to 2.68. The gains from international selection were between 0.03 to 1.49 genetic standard deviations, with a median of 0.51. These selection differentials can not be compared directly with those of Wickham and Banos (1998), since the data selection and methods used differ to some extent. Still, they both endorse that almost all countries can make substantial gains in genetic progress from international selection. When comparing the gain in selection differentials in the data used by Wickham and Banos (1998), i.e. data from 1997, with the most recent data it can be seen that the gain has decreased for most, but not all, countries. Similar selection differentials and gains from international selection can be seen in the other breeds included in the international genetic evaluations of Interbull.
Selection differentials for milk somatic cells were also shown to be higher when selecting globally rather than locally for all populations in a study of the Holstein breed (Mark et al., 2001), although the differences were generally smaller than for protein yield.
Effects on genetic diversity
Globalisation of dairy cattle breeding has increased the selection intensity, but has also led to increased relationships between animals and thus higher risk for inbreeding. Wickham and Banos (1998) concluded that average genetic relationships increased noticeably between 1970 and 1990 for Holsteins, and that the ratio of effective to actual population size decreased correspondingly. More recent preliminary calculations indicate, however, that the decreasing trend in population size may have been broken and even improved slightly.
It can also be argued that the current international genetic evaluation procedure, that acknowledges genetic correlations less than unity, actually should offset some of the negative trend. The current procedure generates bull rankings that are specific for each country and re-ranking, due to low correlations between countries, mean that not the same bulls will be in the "top 100" list in all countries. Thus, selecting the top 100 bulls on each country scale would lead to a selection of between 120 and 300 different bulls globally depending on breed (Table 3). This would, in turn, lead to avoiding depletion of the global diversity, compared to a situation with a correlation of unity, as was assumed in the conversion equation era, leading to the same ranking in all countries.
Table 3. The impact of using correlations less than unity between countries on total number of different bulls among top 100 bulls selected for protein yield (data from the routine evaluation of Interbull, August 2001)
Breed
|
No. of
populations |
r=1 |
r<1 |
| Ayrshire |
10 |
100 |
156 |
| Brown Swiss |
9 |
100 |
169 |
| Guernsey |
5 |
100 |
124 |
| Holstein |
27 |
100 |
312 |
| Jersey |
10 |
100 |
158 |
| Simmental |
8 |
100 |
165 |
The introduction of international genetic evaluations for traits other than production also enables countries and individual farmers to base their selection decisions on broader selection criteria, thus increasing the total number of bulls selected and further improve genetic diversity. The proportion of common bulls within the "top 100" in different countries may be less for functional traits than for production, since genetic correlations are generally lower. There were, for instance, 273 different bulls appearing in the different "top 100" lists for milk somatic cells in the Holstein breed although only 12 populations participated in the study by Mark et al. (2001).
Challenges
Areas of improving international genetic evaluations include continuous refinements of the current methodology and data quality assurance. The major factor affecting the suitability of the current methods for international genetic evaluations is availability of appropriate genetic parameters. The current method of genetic correlation estimation presupposes well-connected data and genetic links between countries, and such may not always be present. Also, the latest test-run included 27 Holstein populations, thus creating the need to estimate 378 genetic parameters, with some of them likely close to the border of the parameter space. Estimating that many parameters may not only be an arduous task, but may also explore and reveal the limitations of the current methodology. It is therefore urgent to develop more robust methods and such methods are currently being investigated both at the Interbull Centre and elsewhere.
One limitation of the current methodology is that it can only handle one trait from each country, i.e. evaluating one trait at a time ignoring possible genetic correlations between traits. Many national genetic evaluations, providing data to MACE, employ multiple-trait evaluation methods and it would naturally be desirable if such structures could be accommodated in the international evaluations. A multiple-trait MACE would likely also be advantageous for international genetic evaluations of functional traits where information on correlated traits could improve breeding value estimations. A multiple-trait MACE, however, not only adds to the computational load but also to the complexity of the evaluation method since residual correlations would have to be taken into consideration at every step in the process.
Input to MACE is pre-processed data from national evaluation systems, and it is obvious that the quality of international genetic evaluations relies heavily on the quality of such data. Recent work has shown that there is a large variation among Interbull member countries in the pre-evaluation, evaluation and post-evaluation steps (Interbull Bulletin 24). Although MACE can accommodate such differences through genetic correlations less than unity between countries, a greater degree of harmonisation of data recording and handling would certainly increase the usefulness of the international genetic evaluations. Interbull is near completion of a new set of guidelines for national genetic evaluation systems, mainly directed towards production traits in dairy cattle. In addition, tools for validation of national evaluation systems are further developed.
At present, Interbull evaluations depend on national genetic evaluation results from various countries with all the advantages and disadvantages of the method, as mentioned above. New developments in national genetic evaluation procedures add to the variety of existing models and introduce differences among national evaluation results that do not always reflect different expressions of biology. Investigation of alternative methods based on analysis of individual cow performance records instead of national genetic evaluations of bulls could provide a viable approach to alleviating this problem. Such an approach could provide opportunities to better handle true genotype by environment interactions, possibly through some herd-clustering techniques, rather than the "genotype by evaluation model" that the current method has to rely on. In addition, an evaluation based on individual performance records would also give international genetic evaluations of cows. Such types of evaluations entail whole new opportunities, but also new challenges and issues to deal with.
The list of potential areas for research on international genetic evaluations can be made very long. Many developments in international genetic evaluations have been initiated by studies performed at various research institutes and this tradition needs to be kept up, since Interbull does not have in-house resources to investigate all pertinent questions. Interbull has recently initiated a process to identify the research needs in the area and to list projects with high priority, where active participation of external resources is sought. The finalised list of projects is available on the web-site of Interbull Centre at http://www.interbull.org.
Final comments
The presence of Interbull as an independent non-profit organisation providing international breeding values has facilitated better global use of dairy cattle genetics and wiser selection decisions. This is made possible with the availability of evaluation methods that recognise the fact that different genotypes may be best suited for different environmental conditions. Thus, current international genetic evaluations identify the best bulls for the breeding environment in each member country. Importers may directly make better decisions as to what genetic material should be brought in to contribute to the improvement of their population, and exporters can use this information to identify foreign markets where their bulls are expected to compete most successfully. What is equally important is that local breeding goals can be serviced, especially if international genetic evaluations for functional traits become more common, thus avoiding the need to compromise with objectives due to lack of information.
Appendix – Discussion about Interbull Evaluations and their Impact on International Trade
Does international genetic evaluations disrupt trade of bull semen?
No, I don't think so! Of course Interbull results greatly influence trade of semen, but I would not call this a disruption. International breeding values associated with a reliability that reflects the degree of uncertainty of the breeding value are the best and most objective, way to rank bulls from different sources. Anything else would be based on speculation and guesses. International genetic evaluations make a fair international competition possible. Of course, bulls with no or very little information will never have the same opportunity to rank as highly as a bull with much more information. Some might argue that this isn't fair. But using that as an argument for not using the best information available (e.g. stop doing Mace for conformation) would certainly make it more difficult to identify superior bulls for conformation and consequently decrease genetic gain for these traits. International genetic evaluations are necessary in order to achieve as much genetic progress as possible! Obviously, with the appearance of international breeding values for a larger range of traits, it has become more difficult to market bulls since the buyer have more information to take into account (not only production). But it gives the dairy breeders a better opportunity to select optimal bulls and enables higher total merit genetic progress. The fact is that trade of semen will continue, so why not use the best information available to ensure as optimal a trade (in terms of genetic progress) as possible!?
Are (low) genetic correlations only a curse (e.g. something that should make Interbull stop doing MACE for conformation traits)?
No. It is correct that international exchange becomes less valuable when correlations are low, but we have to remember that low correlations are low for a reason, e.g. some genes may be more important in one environment/country compared with another one. However, international EBV's breeding values still give guidance about what to expect from a foreign bull, and that is more than you would get by any other way. As long as genetic correlations are different from zero then international genetic evaluations are beneficial, because then information from other countries can be used to predict performance in the country of question. However, I think that there is room for improvement in terms of trait harmonization, especially for some conformation traits in some breeds and this could lead to slightly higher correlations and an even more increased usefulness of international evaluations.
References:
Banos, G., 1999. From research to application: A summary of scientific developments and possible implementation to the genetic improvement for functional traits. Interbull Bulletin 23, 65-74.
Banos, G., Smith, C., 1991. Selecting bulls across countries to maximize genetic improvement in dairy cattle. J. Animal Breeding and Genetics 108, 174-181.
Philipsson, J., 1998. Global use of bulls and the "INTERBULL system". Acta Agric. Scand. Sec. A, Animal Sci., Suppl. 29, 98-107.
Klei, L., Weigel, K.A., 1998. A method to estimate correlations among traits in different countries using data on all bulls. Interbull Bulletin 17, 8-14.
Mark, T., Fikse, W.F., Sigurdsson, A., Philipsson, J., 2000a. Feasibility of international genetic evaluations of dairy sires for somatic cell count and clinical mastitis. Interbull Bulletin 25,154-162.
Mark, T., Fikse, W.F., Banos, G., Emanuelson, U., Philipsson, J., 2000b. Summary of Mace pilot-runs for somatic cell count and clinical mastitis. Interbull Bulletin 26, 43-52.
Mark, T., Fikse, W.F., Emanuelson, U., Philipsson, J., 2001. International genetic evaluations of Holstein sires for milk somatic cell and clinical mastitis. Paper presented at EAAP, Budapest, Hungary, August 26-29, 2001.
Sigurdsson, A., Banos, G., Philipsson, J., 1996. Estimation of genetic (co)variance components for international evaluation of dairy bulls. Acta Agric. Scand. Sect. A, 46, 129-136.
Sullivan, P.G., 1999. Appendix: REML estimation of heterogeneous sire (co)variances for Mace. Interbull Bulletin 22, 146-148.
Wickham, B.W., Banos, G., 1998. Impact of international evaluations on dairy cattle breeding programmes. Proceedings 6th World Congress on Genetics Applied to Livestock Production 23, 315-322.
Related Links:
Interbull Home Page
|