INTERBULL Routine Genetic Evaluation for Dairy Production Traits

November 2002

Introduction

The latest routine international evaluation for dairy production traits took place as scheduled at the Interbull Centre. Data from twenty five (25) countries were included in this evaluation. Results were distributed on November 7, 2002, to designated representatives who have signed a confidentiality agreement with the Centre, and on November 11, 2002, to representatives who still have not signed the agreement.

International genetic evaluations for milk, fat and protein yields of bulls from Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Estonia, Finland, France, Germany, Hungary, Ireland, Israel, Italy, Netherlands, New Zealand, Norway, Poland, Rep. of South Africa, Slovenia, Spain, Sweden, Switzerland, the United Kingdom, and the United States of America were computed. Ayrshire, Brown Swiss, Guernsey, Holstein, Jersey and Simmental breed data were included in this evaluation.

New data (national genetic evaluations) included in this evaluation

• Data from the most recent genetic evaluation for the Holstein breed in one country had to be excluded due to unresolved problems with data used for the national evaluation. Data provided for the May evaluation was used instead.
   
• Data from a joint evaluation of the Flemish part of Belgium and the Netherlands was submitted for the Holstein evaluation. The joint population is identified as NLD in the documentation. The evaluation results were based on a modified random regression test-day model including random herd curves. The same model was used for all breeds in the Netherlands.
   
• Data from the Walloon part of Belgium was evaluated separately with a multiple-trait, multiple-lactation random regression test-day model including random herd curves and submitted for this evaluation. This population is identified as BEL in the documentation.
   
• Data from a joint evaluation, based on a random regression test-day model, for Austria and Germany was submitted for the Brown Swiss and Simmental evaluations. The joint population is identified as DEA in the documentation.
   
• Data with corrected calculations of EDC was submitted by Hungary.
   
• Australia, Spain, and United Kingdom submitted data with a revised specification of publication status, which more accurately reflects the official publication policy in these countries.
   
• The international identification system for Brown Swiss and Simmental in Germany was changed and the new identifications were incorporated in the current Interbull evaluation. Further information about the new system, and cross-references between old and new identities, can be obtained from the Interbull Centre or BLT (Joerg Dodenhoff <jdodenh@blt.bayern.de>).
   
• Systems to identify non-AI and import (second proof) bulls in some countries are still in development stage; this may account for differences in national data set sizes from one evaluation to the next, in addition to young bulls first entering the system. In all such countries, progress is being made towards consistent definition systems. Non-AI bulls are not included in Interbull evaluations whereas import bull proofs are subject  to stricter editing criteria.
  
   

Changes in international genetic evaluation procedures and checks

• The requirement for inclusion of second country proofs in the evaluations was changed. Proofs that are based only on 2nd crop daughters (type=21) are included if they are:
  
  a) identified as meeting the standard national rules for official publication in the country sending the information (official publication of proof=Y)
  OR
  b) based on minimum 150/30/80 daughters in 50/10/20 herds (Holstein/Guernsey/other breeds).
  
  Number of second country proofs included in the evaluation increased for most populations, although it decreased in some populations.
   
• A weighted bending procedure (Jorjani et al. 2002, Interbull Bulletin No. 29) was implemented to obtain positive definite correlation matrices (number of common bulls was used as weighting factor).
   
• Improved data verification procedures following the ideas of Klei et al. 2002 (Interbull Bulletin No. 29) were introduced.
   

Data and method of analysis

Data were national genetic evaluations of AI sampled bulls with daughters in at least 10 herds. Table 1 presents the type and amount of data included in this Interbull evaluation. The difference between "No. Records" and "No. Bulls" is explained by common bulls with proofs in more than one country. The entry "Publishable Proofs" reflects bulls whose international evaluations were distributed to service customers. The difference between "No. Bulls" and "Publishable Proofs" is explained by bulls not meeting the minimum criterion for official publication in the country of origin.

Table 2 presents the current definition of the reference (genetic) base as supplied by each country in the dairy-production proof file.

Estimated genetic parameters are shown in APPENDIX I. Parameters are listed by breed, trait and country.

National proofs were first de-regressed within country and then analysed jointly with a linear model including the effects of evaluation country, genetic group of bull and bull merit. Different breeds were analysed separately. Heritability estimates used in both the de-regression and international evaluation were as in each country's national evaluation (Table 1).

Ancestor-bulls without own proofs were traced back two generations from the oldest bulls with proofs in order to increase across country connections and account for the effect of selection.

Genetic groups were defined according to unidentified parents by national origin, breed and birth year of the bull and path of selection (sire, maternal grand-sire, maternal grand-dam). Birth year grouping was by 1-5 year periods. Small groups (generally consisting of less than 10 bull) were combined.

Scientific literature

The international genetic evaluation procedure is based on international work described in the following scientific publications:

1. International genetic evaluation computation

L. Schaeffer. 1994. J. Dairy Sci. 77:2671-2678
W.F. Fikse and G. Banos. 2000. J. Dairy Sci. (submitted)

2. Genetic trend validation

D. Boichard et al. 1995 J. Dairy Sci. 78:431-437

3. De-regression and genetic parameter estimation

A. Sigurdsson and G. Banos. 1995. Acta Agric. Scand. 45:207-219
A. Sigurdsson et al. 1996. Acta Agric. Scand. 46:129-136
L. Jairath et al. 1998. J. Dairy Sci. Vol. 81:550-562
P.G. Sullivan. 1999. Interbull Bulletin 22:146-148.
H. Jorjani et al. 2002. Interbull Bulletin 29:1-3.

4. Time edits

K. Weigel and G. Banos. 1997. J. Dairy Sci. 80:3425-3430

5. International reliability estimation

B. Harris and D. Johnson. 1998. Interbull Bulletin 17:31-36

 Genetic correlation estimation procedure

Estimation of genetic correlations among countries takes place in test-runs only, when new or modified data are submitted from a country, according to the following procedure (as per Interbull technical workshop of April 1995, Uppsala, Sweden):

Step 1:

Several subsets of countries are analysed. At the most 10 countries at a time are included in each subset. Countries that are major link contributors (judged from the number and origin of common bulls with multiple national evaluations) are always included in these subsets. If multiple genetic correlation estimates are computed for a country pair, the highest estimate is kept, as per Sigurdsson et al (1996) showing that genetic correlations may be under-estimated but not over-estimated by the method used.

Step 2:

In some cases sufficient links between countries may be missing, resulting in close to zero genetic correlation estimates. If no reasonable correlations can be estimated via indirect links with third countries, one the the following procedures is followed:

a) Estimates from another breed for the country pair are used, if applicable

b) Product moment correlations of common bulls, adjusted for national evaluation accuracy are used (not a very frequent practice, since presence of common bulls will likely result in reasonable correlation estimates using the approximate REML method of Sigurdsson et al)

c) Estimates from the low end of the correlation distribution are assigned; these would normally range from .86 to .89 between two North Hemisphere countries and from .75 to .78 between a North and a South Hemisphere country; the model of national evaluation is also taken into consideration (countries with similar national evaluation models are assigned higher genetic correlation estimates)
 

Step 3:

Since genetic correlation estimates are not derived simultaneously, the full covariance matrix need to be bent in order to ensure it's positive definite.

Efforts to improve the procedure are currently under way. The use of covariance structure of models that include genetic and non-genetic (eg, national evaluation model, management practice etc) components in determining correlation estimates between weakly or non-linked countries/populations is being studied.

If a country is not linked to the other countries in the evaluation system, its data are not included in the international genetic evaluation.
 

Publication of INTERBULL evaluations

Results were distributed by the Interbull Centre to designated representatives in each country. The international evaluation file comprised international proofs expressed on the base and unit of each country included in the analysis. Such records readily provide more information on bull performance in various countries, thereby minimising the need to resort to conversions.

At the same time, all recipients of Interbull results are expected to honour the agreed code of practice, decided by the Interbull Steering Committee, and only publish international evaluations on their own country scale. Evaluations expressed on another country scale are confidential and may only be used internally for research and review purposes.

All recipients are also expected to follow the agreed guidelines for advertising genetic merit. The guidelines has been distributed to all members and is available on the Interbull homepage (http://www.interbull.org) under "Publications and Documentation/General information".

Next routine international evaluation

The next routine international evaluation for dairy production, udder health and conformation traits is scheduled for February 2003. New data for that run should reach the Interbull Centre not later than January 28, 2003, 17:00 Central European Time (CET); in any case, the most recently received data will be considered. The target time for distribution of results for internal review to designated representatives who have signed a confidentiality agreement with the Centre is February 6, 2003. Official release date of the results is targeted to be February 10, 2003.

Next test international evaluation

The next test international evaluation for dairy production, conformation, and udder health traits is scheduled for March 2003. Countries wishing to enter the system for the first time or planning to submit new information (modified national evaluation procedure, new breeds etc) for following routine evaluations must have their data tested in this test-run.

Deadline for sending data to the Interbull Centre for the next test-run is March 1, 2003.

Means of result distribution from the Interbull Centre

Electronic exchange of data is probably more efficient than anything else. Currently most countries in the service have established internet connections and receive international evaluation results on the day of their release. Subscribers to the service that do not already exercise this option are encouraged to consider it and establish an internet connection and ftp account. When such accounts are available, please provide their specifications to the Interbull Centre. Until then, data will be delivered on CD-ROM's.

For more information about the international genetic evaluation service please contact the INTERBULL Centre: address: SLU Box 7023, 750 07 Uppsala, Sweden;  fax: +46-18-672648;

e-mail: Ulf.Emanuelson_at_hgen.slu.se

e-mail: Thomas.Mark_at_hgen.slu.se

e-mail: Freddy.Fikse_at_hgen.slu.se

e-mail: Hossein.Jorjani_at_hgen.slu.se

e-mail: Jan.Philipsson_at_hgen.slu.se

e-mail: Dan.Englund_at_hgen.slu.se