GMACE Weighting Factors

Abstract

Statistical indicators to quantify information in national GEBV are required for optimal weighting of these GEBV as input observations for GMACE.  The weighting factors are also used when approximating reliabilities of GMACE output, the international GEBV.  The present study considered a change in approach from providing a recipe to national centers for deriving the weighting factors, to instead having Interbull apply the recipe internally and in a guaranteed consistent way for all countries.  The main impacts of this change in approach were on selected estimates of genomic variance and GMACE reliabilities, most notably for specific estimates that had previously been questioned as erroneous.  Most estimates of variance and reliability were essentially unaffected by the change in approach, because most weighting factors were only slightly different than before.  An advantage of the new approach was that weighting factors and GMACE model specifics became perfectly aligned, and thus GMACE reliabilities were aligned without exception to the national values.  All GMACE reliabilities were equal or higher than national values.  All increases in reliability with GMACE were as expected, only for bulls that had input GEBV from more than one country, and with smaller increases if the multiple GEBV were from countries that share data for national genomic predictions.