Abstract

SOUSA, J.E.R. et al. Evaluation of average growth curve of goats using random regression model. Arch. zootec. [online]. 2010, vol.59, n.226, pp.267-276. ISSN 1885-4494.

Polynomial functions for modeling growth curves of 946 Anglo-Nubiano goats, born from 1980 to 2005, and their effect on genetic parameter estimates were evaluated. Growth trajectories were fitted by fixed regression of weights on orthogonal polynomial of age, considering polynomials from second to seventh orders. For fixed regression, polynomial models were compared using the following criteria: mean square errors, absolute average residual error, square percent bias and goodness of fit. Akaike's information criterion, Bayesian information criterion and likelihood ratio test criteria were used to evaluate the effect of different polynomial order of fixed regression on genetic parameters estimates. An expressive improvement was found using higher order polynomials and those of fourth order were sufficient for modeling the average growth curve of goats. No influences on genetic parameter estimates were observed adopting higher than cubic order polynomial functions. The direct heritability showed an increasing pattern until 84^th day of age, while the maternal one had a constant pattern for the whole period. Polynomial functions are efficient tools for modeling growth curve and polynomials of at least fourth order should be used for modeling the average growth curve of goat in random regression models.

Keywords : Anglo-Nubian; Growth curve; Polynomial function; Fixed regression.

        · abstract in Portuguese     · text in Portuguese     · Portuguese ( pdf )