The objective was to compare different random regression models using Legendre polynomial functions of different orders, to evaluate what best fits the genetic study of the growth curve of meat quails. It was evaluated data from 2136 cut dies quail, of which 1026 belonged to genetic group UFV1 and 1110 the group UFV2. Quail were weighed at 10, 70, 140, 210, 280, 350, 420, 770, 1120 and 1470 days of age, and weights used for the analysis. It was tested two possible modeling of heterogeneous residual variance, which was grouped into three five age classes. After, it was studied the random regression mode, which is better applied to the growth curve quail. The comparison between models was made by the Akaike Information Criterion (AIC), Bayesian Information Criterion of Schwarz (BIC), logarithm of the likelihood function (Loge L) and the likelihood ratio test (LRT), at 1%. The model which considered the heterogeneity of residual variance CL3 was adequate to UFV1 lineage and the lineage model CL5 UFV2. A Legendre polynomial with order 5, for direct genetic effect and 5 for the permanent animal to UFV1 bloodline and order 3 for direct genetic effect and 5 for the permanent animal for UFV2 bloodline should be used in genetic evaluation of the growth curve of meat quails.
Coturnix coturnix; heritability; heterogeneous variance; Legendre polynomial
