Submitted to: American Dairy Science Association Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: July 26, 2000
Publication Date: N/A
Separate estimates of breeding value (BV) can be combined by selection index if combined analysis of all data is not possible or efficient. Computation is fast but not exact if reliabilities of separate estimates are approximate, if extent of overlap of data sets is unknown, or if selection has occurred across data sets. Selection index can be used to combine single-trait evaluations into approximate multitrait evaluations or single-country rankings into multicountry rankings. Iteration can be avoided by including parent evaluations as data and combining them before progeny evaluations. Some information would be lost because grandprogeny data from other countries would not contribute back to grandparents as a result of order in which information is combined. Exchange of parent evaluations could provide a closer connection between national and international evaluations and may be more accurate than the current sire- maternal grandsire model used internationally. Animal and parent reliabilities can be used to backsolve for daughter equivalents from parents vs. those from progeny and own records. An estimated BV can be adjusted separately for added parent or progeny information. If either parent is unknown, progeny Mendelian sampling is confounded with unknown- parent deviation from genetic group solution. Variance estimation can be made simpler by using sum of squared Mendelian sampling effects as the quadratic form instead of the relationship matrix inverse pre- and post- multiplied by BV's. The expectation involves sum of reliabilities, sum of reliabilities from parent average, and covariance of parent and progeny BV's. Selection index methods may allow current multiple-trait across- country evaluations for bulls to be improved and to be extended to cows.