Modelos de equações estruturais generalizados mistos aplicados à avaliação genética de codornas de corte

• Ano de defesa: 2014
• Autor(a) principal: Pacheco, Rodrigo de Oliveira
• Orientador(a): Torres, Robledo de Almeida
• Banca de defesa: Silva, Fabyano Fonseca e , Teixeira, Rafael Bastos , Nascimento, Moysés , Carneiro, Antônio Policarpo Souza
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Viçosa
• Programa de Pós-Graduação: Doutorado em Zootecnia
• Departamento: Genética e Melhoramento de Animais Domésticos; Nutrição e Alimentação Animal; Pastagens e Forragicul
• País: BR
• Palavras-chave em Português: Codornas; Genética animal; Modelos lineares (Estatística); Modelo recursivo; Estrutura causal; Acurácia de predição
• Palavras-chave em Inglês: Quail; Animal genetics; Linear models (Statistics); Recursive model; Causal structure; Accuracy of Prediction
• Área do conhecimento CNPq: CNPQ::CIENCIAS AGRARIAS::ZOOTECNIA::GENETICA E MELHORAMENTO DOS ANIMAIS DOMESTICOS
• Link de acesso: http://locus.ufv.br/handle/123456789/1883

Structural Equation Models (SEM) are able to account the cause effect relationship among traits. The aims of this study were propose an extension for structural equation models in the generalized mixed linear models context, to consider traits with a Poisson distribution and evaluate the Generalized SEM under the predictive point of view by the accuracy of prediction estimation. Five meat-type quails traits were evaluated: Birth Weight (BW), Weight at 35 days of age (W35), Age at the first egg (AFE), Average Eggs Weight from 42 to 182 days of age (AEW), and number of eggs produced from 42 to 182 days of age (NE). I was determined a hypothetical structure representing the causal relationship among traits, and the direct genetic effects over each trait. In this case, the genetic correlations between these genetic effects were not considered. The equations of the recursive model were estimated following Generalized Linear Mixed Models formulation. The genetic effects per se were not sufficient for phenotype determination, for this reason, these effects should be associated with the whole information that the causal structure can offer in relation to traits. Accuracy of prediction values were considered high (0.87 0.99) for all traits but number of eggs (0.36). According to the AIC and BIC values, the model considering Number of eggs with Poisson distribution can be considered more indicated than the one considering number of eggs with a Gaussian distribution. Generalized Mixed Structural Equation model is efficient for phenotypic description and the evaluation considering NE with Poisson distribution was better than the one considering NE as a Gaussian distributed trait.