Factors affecting calculation and use of conversion equations for genetic merit of dairy bulls.

Factors affecting calculation and use of conversion equations were reviewed. Methods of expressing reliability of converted evaluations were surveyed. Of 16 countries responding, 6 did not calculate reliability for converted evaluations, 5 accepted reliability from the exporting country, and 5 assumed genetic correlations of .6 to 1.0 with the US. Genetic correlations between the US and 8 other countries were estimated and generally were > or = .9; estimated correlations between the US and Canada were 1.0. Estimated correlations averaged .93 for milk, .89 for fat, and .92 for protein yields. Correlation estimates were lowest for countries differing most from the US in management conditions (Australia, New Zealand) or trait definition (Germany), which suggests that correlation estimates < 1.0 indicate differences in trait measurement as well as differences in biological expression. Conversion equations were computed from data of US and Canadian Holstein bulls with and against the gene flow. Equations against the gene flow generally had regression coefficients and intercepts lower than those calculated with the gene flow. Lower regression coefficients were explained by selection on the dependent variable. Lower intercepts were attributed to preferential treatment of daughters from imported semen, which would lower intercepts for equations against the gene flow and inflate intercepts with the gene flow.