Factors affecting genetic correlation estimates from dairy sires' genetic evaluations to assess genotype-environment interaction

Effects of trait heritability (0.05 or 0.25), effective daughters number (30 to 500), sires number (10 to 500), and sire selection (selecting or not the top 50% sires), were evaluated based upon standard error (SE) and bias of genetic correlations (r sub(G)) between countries estimated from Calo's method (r sub(G)) using simulated data. Calo's method is based on correlations between sire's predicted transmitting abilities (PTA) in two countries adjusted for reliabilities. Unselected sire's data analysis gave nearly unbiased r sub(G) in all cases, but selected sire's data analysis gave underestimates. Bias was from -0.34 to -0.05 for the 0.25 heritability trait (milk yield), and from -0.42 to -0.17 for the 0.05 heritability trait (functional). Underestimation of r sub(G) decreased with increased effective number of daughters (PTA's reliability), but was quite insensitive to number of sires. The SE of genetic correlations estimates decreased with increased PTA's reliability and sires number, and was higher for selected sires. Approximately 50 sires with PTA's reliabilities greater than or equal to 0.97 on each country are required to obtain accurate (SE less than or equal to 0.02) and unbiased (bias less than or equal to 0.05 degree ') r sub(G) with Calo's method using the best 50% selected sires. Many genetic correlation estimates between countries, already published using the Calo's method, may be underestimates, particularly for low heritability traits, and with low number of effective daughters in the importing country. Therefore, caution is required before interpreting the published r sub(G)