Breeding values were predicted for test day milk yields of Caracu cows, using univariate model (TDMO) and random regression model (RRM). The ranks were based on traditional 305-day milk yield breeding values (PTA305). For RRM, breeding values were predicted for individual test-day milk yield and for sum of all test-days (RRM305). Spearman correlation were estimated among predicted breeding values from different methodos (TDMO and RRM) and the coincidence of rank of 10% best sires were verified. Genetic lactation curves of five best sires were compared. Spearman correlation between breeding values predicted for test-day milk yield by TDMO were lower than 0.80. When predicted by RRM, the correlations between breeding values were higher than by TDMO, ranging from 0.41 to 1.00. This correlation was 0.87 between PTA305 and MRA305. It was also observed that the best sires for PTA305 did not present the best genetic lactation curves, estimated by RRM. The RRM shows more adequacy than TDMO to substitute traditional 305-day milk yield in genetic evaluation of dairy cattle.
genetic evaluation; Caracu breed; covariance function; random regression; milk yield
