Weighting Factors of Sire Daughter Information in International Genetic Evaluations

Abstract

International genetic evaluations of dairy bulls are currently based on national genetic evaluation results. Total number of daughters in a country is used to weight national information, but may not optimally reflect the precision of a sire's daughter contribution to international genetic evaluations. This study investigates the impact of alternative weighting factors on international evaluation results. A conventional progeny test scheme was simulated for two dairy cattle populations, with semen exchange at a fixed rate after each generation. True breeding values for both populations were generated as bivariate normal deviates. Each cow had three lactation records in one country only. After 10 generations of selection, all records were used in national breeding value prediction. National breeding values of bulls were used as input to international evaluations. Seven different weighting factors were evaluated: 1) total number of daughters; 2) total number of lactations; 3) as (one) also adjusted for finite contemporary group size; 4) as (three) also adjusted for distribution of daughters over contemporary groups; 5) effective daughter contribution considering finite contemporary group size and correlation between repeated records; 6) as (five) also considering the reliability of the daughter dam evaluation; and 7) as (five) also considering the reliability of the daughter female ancestors’ evaluations. Using the last two weighting factors yielded empirically unbiased estimates of sire variance. Using total number of daughters overestimated genetic variance by up to 7%. In general, international breeding values were marginally affected by choice of weighting factor. The effect was larger when different national evaluation models had been applied in the two countries. International reliabilities for the last two weighting factors were close to expectation, whereas using total number of daughters resulted in 1 to 4% negative bias. In practice, different countries apply a wide range of national evaluation models, and genetic ties may be weak between some populations, thereby increasing the potential effect of weighting factors on international comparisons. The weighting factor developed in this study, which considers contemporary group structure, correlation between repeated records, and reliability of dams of daughters, should replace total number of daughters in international genetic evaluations of dairy sires.