(CO)VARIANCE COMPONENT ESTIMATION OF YIELD TRAITS BETWEEN DIFFERENT LACTATIONS USING AN ANIMAL-MODEL

Additive and residual (co)variances between the first four lactations were estimated for milk, fat and protein yields. Results were obtained using restricted maximum likelihood and a multiple trait animal model with different design matrices in three sets of four traits per set. Model included different design matrices in the fixed effects and took into account missing records. An analysis was carried out from records of 18 658 cows (10 583 for protein yield), and the order of the coefficient matrix was 118 192 (93 732 for protein yield). A Monte Carlo procedure to approximate the traces involving the coefficient matrix of the mixed model equations was used in the EM algorithm. Heritabilities for milk yield ranged between 0.22 and 0.32, for fat yield between 0.20 and 0.30, and for protein yield between 0.19 and 0.26. Genetic correlations between adjacent lactations were greater than 0.85, with the exception of the third and fourth lactations for protein yield. A second analysis was carried out including the 12 traits and 156 (co)variance components simultaneously. In this case, the order of the coefficient matrix was 352 116. A result of this second analysis was that the genetic correlations between milk, fat and protein yields do not depend on the age of the cow.