Genetic parameters for test day milk production of Holstein cows using factors and principal components analysis

The objective was to compare a standard multi-trait (MT) analysis model with factor (FA) and principal components (PC) analyses models to estimated genetic parameters for Holstein cows test day milk production (TD). The data file was composed by 4.616 TD at first lactation registers. The TD was grouped into ten monthly classes of lactation, from the 5th and the 305th day of lactation (TD1 to TD10). Analyses were performed considering 11 different models: standard multi-traits (MT), five reduced rank models to genetic covariance matrix adjusting one (PC1), two (PC2), three (PC3), four (PC4) and five (PD5) principal components and five models using factor analyses (F1, F2, F3, F4 and F5). To all the models the effects additive genetic and residual were considered as random and the effects of contemporary group, age of cow at parturition (linear and quadratic) and days in lactation (linear) were considered as fixed. The values of Log L, AIC e BIC improved with the augment of the number of parameters until CP4 and AF4. Comparing CP4 and AF4 is possible to verify that CP4 proportioned better values to Log L, AIC e BIC. The heritabilities and genetic correlations estimated to the ten test day milk production using MC, CP4 and AF4 models were similar ranging from 0.06 (PL6) to 0.65 (PL10) and from 0.05 (PL4xPL10) to 0.94 (PL2xPL3), respectively, indicating that the structure of the genetic covariance between the TD milk productions can be adjusted using a reduced rank model with four principal components or four factors.