Parâmetros genéticos para seleção de linhas puras da raça Rhode Island Red para produção de ovos

• Ano de defesa: 2018
• Autor(a) principal: Metz, Magda
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Zootecnia; UFSM; Programa de Pós-Graduação em Zootecnia; Centro de Ciências Rurais
• Palavras-chave em Português: Análise multivariada; Componentes principais; Regressão aleatória; Multivariate analysis; Principal components; Random regression; CNPQ::CIENCIAS AGRARIAS::ZOOTECNIA
• Link de acesso: http://repositorio.ufsm.br/handle/1/15738

This study aimed at estimating genetic parameters for the characteristic egg production by means of different models. The data used, referring to the production of eggs of pure-bred Rhode Island Red (RIR) line GG, collected in the period from 1992 to 2013, provided by CNPSA EMBRAPA. In Article 1, the objective was to estimate the coefficients of heritability and correlation for partial and total egg production, as well as to verify the possibility of using a reduced rank model. Egg production was evaluated through the egg production rate, composed of 7,087 records grouped in 13 four-week periods (PO1 to PO13) between the 19th to the 70th week of life of the birds, and another 4 accumulated periods (P1 to Ptotal). For the analysis of main components, the periods from PO2 to PO11 were considered. Heritability estimates were 0.39, 0.11, 0.15 and 0.07 (P1, P2, P3 and Ptotal). Genetic correlations between the periods P1xPtotal, P2xPtotal and P3xPtotal were respectively 0.18, 0.53 and 0.99. The reduced post model adjusted for the first two main components provided the best fit. Heritability estimates for this model ranged from 0.10 to 0.26. It was concluded that the selection of birds based on the production of eggs up to the 54th week (P3) has a high genetic correlation with the total egg production, and the genetic covariance structure for posture rate can be adjusted using a reduced rank model containing two main components. The objective of this study was to compare different models of random regression (MRA) with different orders of the Legendre polynomial, in order to identify the best fit model for the RIR bird posture rate and to estimate the covariance and parameter components for this characteristic. The work file consisted of 83,687 records, and weekly productions were grouped into 13 composite periods of 4 weeks (PO1 to PO13). The proposed model included the fixed generation effect, the average production trajectory, the additive random and genetic effect, the direct and residual permanent effect. In order to model the fixed trajectory we used a polynomial of order 4 and for the random trajectories, Legendre orthogonal polynomials were used, varying the orders for the additive and permanent environment genetic effect, with homogeneous or heterogeneous residual variance. Differences between the models were compared using the AIC, BIC, LogL and LRT statistics. The model composed of polynomials of order four for fixed effect, order two for additive effect, eight for permanent environment, containing seven classes of residual variance, was the most adequate for the adjustment of egg production rate data. Heritability estimates ranged from 0.01 to 0.24 and genetic correlations between the periods ranged from -0.27 to 1.00. The selection based on the partial periods PO7 to PO10 have good estimates of heritability, are well correlated and are therefore indicated for the selection of RIR birds.