Different Models for Estimating Genetic Parameters for Growth Traits of Holstein- Friesian under Semitropical Conditions

Estimates of genetic parameters resulting from three different models for birth weight, weaning weight, yearling weight and total gain were compared. Data consisted of 823 Holstein-Friesian animals progenies of 25 sires and 459 dams in a single herd. The MTDFREML programs were used to estimate heritabilities, predicted transmitting abilities (PTA) and genetic correlations. The three models compared were: Model I, a sire model, included year-season of birth and sex of calf as fixed effects and sire as random effect. Model II, a sire-dam model, included, in addition to the effects in model I, dam as a second random effect. Model III, a full animal model, included, in addition to the fixed effects of model I, animal additive direct genetic effect, maternal genetic effect and permanent environmental effect, as three random effects. The likelihood function showed that the full animal model best fit the data. Heritability estimates and genetic correlations obtained from model III were higher than those obtained from models I and II for all traits. Models I and II resulted in changes in ranking for PTA in contrast with model III. In conclusion, full animal model could be most appropriate for estimating genetic parameters for growth traits than the sire or the sire-dam models.