Efeito da autocorrelação na modelagem da curva de lactação de cabras
Year: 2009
Type: dissertation
Abstract: In this study, the Wood model was adjusted to1069 lactations curve from dairy goats, for three different situations of residual structure. At first, the errors were considered as independent for all lactations, in the second as autoregressives of first order also for all lactations and in the third as independent for lactations that showed no residual autocorrelation and autoregressives of first order for lactations that showed, respectively, according the Durbin- Watson test. These three situations were compared by the percentage of convergence and the average and coefficients of variation of the mean square errors (MSE) and coefficients of determination (R2) adjusted. Subsequently, estimates of the variance components and genetic parameters for total milk yield (MY), for coefficients of the Wood model (a, b and c ) and for milk yield at peak (ymax), time to reach the peak (Tmax) and persistency factor (pers) of each lactation were obtained using the procedure REMLF90 (MISZTAL, 2002), with the objective to determine the influence of residual autocorrelation on heritability (h2) and repeatability (r) estimates and of animals rank according to their breeding values. Two genetic evaluations were made, one for MY with coefficients of the Wood model and other for MY with ymax, tmax e pers. The statistical model included the random effects of animal, permanent environmental and error, the fixed effects of contemporary group (year - season of kidding), number of kids per birth, genetic groups and kidding order, and the days in milk (linear) and age at kidding (linear and quadratic) covariates. In the study of individual fitting of the Wood model to lactation curves differences between the three cases of residual structure were very small, then it was not possible to give a residual structure that results in better fit of the model of Wood. The low values for h2 and r estimates for a, b, c coefficients in this three situations indicated that the gains obtained by selection for the shape of the curve was small. The different residual structures used influence estimates of the parameters h2, r and the correlations between traits. In addition, caused differences in animals rank, based on their breeding values of all traits. Therefore, in applying the Wood model, the correction of errors that are not independent must be done.
Source: Americanae (AECID Library)
Author: Ana Lúcia Puerro de Melo
Language: Portuguese
Subfield: Genetics
Domain: Life Sciences
SDG: Zero hunger
Open Access status: green
OpenAlex ID: W2947017993
