Tools to refine unknown parent groups definition.

Unknown Parent Groups model missing parentships according to breed, year, and pathway of selection. Genetic evaluations need a sensible definition of rules to form Unknown Parent Groups to ensure its accurate estimation. With more complex evaluations, there is a need for systematic rules to form Unknown Parent Groups. We propose several methods to ascertain the informativity and to join consecutive Unknown Parent Groups within breed and pathway. First, a pseudo-count of records based on descendants with record can be easily computed from a pedigree and a list of animals in records, by propagating fractions of one from the flagged animals to ancestors through the pedigree, from youngest to oldest. Then 2 rules are proposed, one joins Unknown Parent Groups forward in time until records appear, the second joins consecutive Unknown Parent Groups backward in time to achieve a minimum pseudo-count. Finally, a simplified model estimates the precision of contrasts of Unknown Parent Groups. Computations are feasible even for very large data sets. We illustrate with 2 examples from all-breed US traditional BLUP evaluation for yield and health traits. For yield traits, the number of pseudo-records is very high and joining Unknown Parent Groups occurs mainly in small breeds. Joining or not results in very similar estimated breeding values and genetic trends has no consequences. For health traits, joining Unknown Parent Groups occurs for all breeds, and more intensely for small breeds. Joining or not joining UPGs resulted, for some trait-breed combinations, in quite different genetic trends and estimated breeding values. The approximated contrasts across all pairwise Unknown Parent Groups show that the joining strategies achieve more precision, in form of lower standard errors.