Avaliação de modelos estatísticos considerando a incerteza da paternidade
Carregando...
Data
2011-02-25
Autores
Shiotsuki, Luciana [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O sistema de acasalamento mais empregado na pecuária de corte extensiva é a monta a campo com o uso de reprodutores múltiplos. Este sistema traz diversas facilidades ao produtor, entretanto não permite a identificação da paternidade das crias, impossibilitando a comparação do desempenho reprodutivo e produtivo dos diferentes touros, o que afeta negativamente as avaliações genéticas e o progresso genético pela seleção. Objetivou-se com este trabalho comparar metodologias estatísticas que permitam considerar rotineiramente o uso da informação de acasalamentos com touros múltiplos na avaliação genética. Para tanto, foram analisados registros de ganhos de pesos pós-desmama e peso ao sobreano, mensurados em machos e fêmeas de animais da raça Nelore, nascidos no período de 1984 a 2006. Foram aplicados diferentes procedimentos estatísticos que consideram dados de animais com incerteza de paternidade na estimação de parâmetros e valores genéticos. As estimativas dos componentes de (co)variâncias genéticas e dos valores genéticos para os animais presentes no pedigree foram comparadas entre os diferentes procedimentos estatísticos. No capítulo 2, foram utilizados o modelo de paternidade desconhecida e o modelo de grupos genéticos para comparação. Verificou-se que o modelo de grupo genético, definido pelo intervalo de gerações dos machos deve ser usado para classificar e predizer o mérito genético dos filhos de reprodutores múltiplos. No capítulo 3, comparou-se o modelo com base na matriz de parentesco médio e o modelo hierárquico bayesiano (HIER) para incerteza de paternidade. Concluiu-se que o HIER é o modelo que melhor ajustou os dados para estimar os parâmetros genéticos de animais que possuem paternidade incerta. No capítulo 4, propôs-se uma aproximação do modelo hierárquico Bayesiano, usando procedimentos Bayesianos empíricos e máxima...
Multiple-sire (MS) mating is the most used mating system in extensive beef cattle systems. This system provides several conveniences to the farmers, but does not allow for paternity identification of the offspring, making impossible to compare the reproductive and productive performance of different bulls, negatively affecting the genetic evaluations and therefore the genetic progress. The aim of this work was to compare statistical approaches that allow for considering information from mating with multiple sires in genetic evaluation. Records from postweaning gain and long yearling weight, measured in males and females of Nellore animals, born from 1984 to 2006 were used. We applied different statistical approaches, which consider data from animals with uncertain paternity in the estimation of genetic parameters and values. The estimated values of the (co) variance components and breeding values for the animals in the pedigree were compared from the different statistical procedures. In chapter 2, we compared unknown parentage and genetic groups models. It was found that the genetic group model defined by the generation interval of males was more appropriate to predict the genetic merit of animals with unknown paternity. In chapter 3, we compared the model based on the use of an average numerator relationship matrix and a hierarchical Bayes model (HIER) accounting for uncertain paternity. We concluded that HIER was the best data fitting model to estimate genetic parameters of animals with uncertain paternity. In chapter 4, we proposed a Bayesian hierarchical model approximation, using empirical Bayesian procedures and maximum likelihood. The proposed model represented a computationally feasible alternative to calculate the probabilities of candidate sires and animal genetic effects on genetic evaluations of large datasets, when there are uncertainty paternity assignments for some animals
Multiple-sire (MS) mating is the most used mating system in extensive beef cattle systems. This system provides several conveniences to the farmers, but does not allow for paternity identification of the offspring, making impossible to compare the reproductive and productive performance of different bulls, negatively affecting the genetic evaluations and therefore the genetic progress. The aim of this work was to compare statistical approaches that allow for considering information from mating with multiple sires in genetic evaluation. Records from postweaning gain and long yearling weight, measured in males and females of Nellore animals, born from 1984 to 2006 were used. We applied different statistical approaches, which consider data from animals with uncertain paternity in the estimation of genetic parameters and values. The estimated values of the (co) variance components and breeding values for the animals in the pedigree were compared from the different statistical procedures. In chapter 2, we compared unknown parentage and genetic groups models. It was found that the genetic group model defined by the generation interval of males was more appropriate to predict the genetic merit of animals with unknown paternity. In chapter 3, we compared the model based on the use of an average numerator relationship matrix and a hierarchical Bayes model (HIER) accounting for uncertain paternity. We concluded that HIER was the best data fitting model to estimate genetic parameters of animals with uncertain paternity. In chapter 4, we proposed a Bayesian hierarchical model approximation, using empirical Bayesian procedures and maximum likelihood. The proposed model represented a computationally feasible alternative to calculate the probabilities of candidate sires and animal genetic effects on genetic evaluations of large datasets, when there are uncertainty paternity assignments for some animals
Descrição
Palavras-chave
Bovino de corte, Paternidade, Reprodutor múltiplo, Bayesian inference, Genetic group model, Uncertain paternity
Como citar
SHIOTSUKI, Luciana. Avaliação de modelos estatísticos considerando a incerteza da paternidade. 2011. vi, 85 f. Tese (Doutorado) - Universidade Estadual Paulista, Faculdade de Ciências Agrárias e Veterinárias, 2011.