Submitted to: Journal of Dairy Science
Publication Type: Abstract Only
Publication Acceptance Date: February 24, 2011
Publication Date: June 30, 2011
Citation: Olson, K.M., Van Raden, P.M., Null, D.J. 2011. Impacts of inclusion of foreign data in genomic evaluation of dairy cattle. Journal of Animal Science 89(E-Suppl. 1)/Journal of Dairy Science 94(E-Suppl. 1):164-65(abstr. 35).
The accuracy of genomic predictions tends to increase with larger numbers of animals in the training data set. Because of this, many countries have combined data to increase the size of their training population. The objective of this study was to investigate different methods of combining the data. Methods included using the domestic data, a combined domestic and foreign training population, and a domestic and foreign multi-trait method. Foreign bulls with U.S. daughters were considered domestic in this study. The combined foreign and domestic training data set was comprised of bulls and cows that were proven (had daughter or own information) as of August 2007 and totaled 9,874 Holsteins and 1,473 Brown Swiss. The domestic training animals included U.S. cows and bulls with U.S. daughters and totaled 8,674 Holsteins and 741 Brown Swiss. The foreign training data set had 1,200 Holsteins and 732 Brown Swiss. Over 90% of the foreign data from Holsteins were from Canada which has a very high genetic correlation to the U.S. A majority of the Brown Swiss data were from Germany and Switzerland, but 6 other countries were represented by ID number all with moderate to high genetic correlations. The validation data sets consisted of U.S. bulls that were unproven as of August 2007 and proven with daughters in at least 10 herds as of December 2010. There were 3,094 and 115 Holstein and Brown Swiss validation bulls, respectively. Genetic correlations for the multi-trait method were computed as the weighted average of countries genetic correlation with the U.S. based from Interbull. Results show a general increase of about 2% from inclusion of foreign data over domestic data only in the Holstein population and an increase of 5% in the Brown Swiss. Multi-trait methodology was beneficial for Brown Swiss and most traits gained reliability and some traits gained as much as 6%. Multi-trait was not advantageous in Holsteins, probably due to the high genetic correlation between Canada and the U.S. It is recommended for countries with high genetic correlations to be treated as one population. Diverse populations may benefit from the implementation of multi-trait methodology.