Modelos estatísticos para análise genética de escores visuais em bovinos de corte

• Ano de defesa: 2015
• Autor(a) principal: Petterson Souza Sima
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Limiar; Nelore; Avaliação visual; Características de crescimento; EPMURAS; Zootecnia
• Link de acesso: http://hdl.handle.net/1843/BUBD-A52GCT

Had as objective to evaluate genetic parameters for performance traits and visual scores features as well as the classification of Nellore, obtained in multi-traits Bayesian analysis with linear model and threshold. The evaluated young bulls were participants from 538 performance tests (PGP), conducted between 2003 and 2012, official by the Brazilian Association of Zebu Breeders (ABCZ), totaling 24,910 animals. The evaluated characteristics were weight corrected for 550 days (P550), average daily gain (GMD), and visual scores of body structure (E), precocity (P), muscling (M), navel (U), breed standard (R), feet and legs (A) and sexual characteristics (S). In statistical models were considered fixed effects of PGP and age at end of the test as nested covariate in PGP, and random effects additive animal and residual. The models presented themselves suitable for obtaining the variance estimates (additive genetic, residual and phenotypic), heritability and correlations (genetic and phenotypic), obtaining means close, but high density ranges to 95% overlapping, not characterizing difference there between. The estimates suggest that the evaluated traits have potential for individual selection for presenting genetic variability and heritability favorable. Scores E, P and M were highly favorable correlated with each other and with growth traits, indicating genetic possibilities of progress on a trait from the selection in another. Moreover, the joint selection among performance traits and visual scores can provide selection of animals with good performance and presents desirable biotypes. Regarding the classification of young bulls by breeding values, there was no significant difference between the models, with Spearman correlations above 0.99 when considered all animals. There was difference only on the classification of young bulls considered top 1% for each trait. These differences in the classifications of young bulls top 1%, together with the difference in the adjustment of the models according to Deviance Information Criterion, where the threshold model showed better adjustment, allow us to say that there is difference in genetic evaluation from Bayesian analyzes with linear and threshold model. These results indicate that threshold model is most appropriate for visual scores in breeding beef cattle programs.