Combining different functions to describe milk, fat and protein yield in goats using bayesian multiple- trait random regression models

Detalhes bibliográficos
Ano de defesa: 2015
Autor(a) principal: Oliveira, Hinayah Rojas de
Orientador(a): Não Informado pela instituição
Banca de defesa: Não Informado pela instituição
Tipo de documento: Dissertação
Tipo de acesso: Acesso aberto
Idioma: eng
Instituição de defesa: Universidade Federal de Viçosa
Programa de Pós-Graduação: Não Informado pela instituição
Departamento: Não Informado pela instituição
País: Não Informado pela instituição
Palavras-chave em Português:
Link de acesso: http://www.locus.ufv.br/handle/123456789/6954
Resumo: The present study aimed to propose multiple-trait random regression models (multiple-trait RRM) combining different functions to describe milk yield, fat and protein percent in a dairy goats genetic evaluation by using MCMC (Markov Chain Monte Carlo) Bayesian inference. Were analyzed 3,856 milk yield (MY), fat (FP) and protein (PP) percent test-day records from 535 first lactation of Saanen and Alpine goats (including crosses). The initial analyses were performed using single- trait RRM, in which for all effects (average curve, additive genetic and permanent environmental) the following models were considered: third and fifth order Legendre polynomials, linear B-splines with three (at 1, 20 and 40 weeks) and five (at 1, 8, 15, 20 and 40 weeks) knots, Ali and Schaeffer function (Ali and Schaeffer, 1987) and Wilmink function (Wilmink, 1987). Residual variances were modeled by a step function with three classes: 1 to 3, 4 to 8, and 9 to 40 weeks of lactation. After definition of the best single-trait RRM to describe each trait (MY, FP, PP) based on the Deviance Information Criterion (DIC), the functions were combined to compose the multiple-trait RRM. The model based on Ali and Schaffer function fitted better for MY and PP, while the model based on fifth order Legendre polynomials (Leg5) was the best one for FP. All tested RRM considering the combinations of functions presented lower DIC values, showing the superiority of these models when compared to other multiple-trait RRM based only on one function. Among the combined RRM, those considering Ali and Schaeffer function to describe the MY and PP, and Leg5 to describe the FP, presented the best fit. Estimates of heritability for MY and FP were close until 20 weeks, ranging from 0.25 at 0.54. The estimates of heritability for PP were, in general, higher than the estimates for MY and FP, ranging from 0.35 until 0.51. The genetic correlation between MY and FP and between MY and PP throughout the lactation period were negative, except for the period immediately after lactation peak. The genetic correlation between FP and PP was positive and approximately constant throughout the lactation (about 0.54). We concluded that combining different functions in a unique multiple-trait RRM can be an plausible alternative for joint genetic evaluation of different longitudinal traits.