Comparação de modelos mistos generalizados bayesianos para avaliação genética da idade ao primeiro parto e número total de crias em cabras da raça Saanen criadas no Brasil

Abstract

The production of goat milk in Brazil is in search of stability and increased productivity, so the Brazilian Agricultural Research Corporation together with the Association of Goat and Sheep Breeders of Minas Gerais, Agronomic Institute of Pernambuco - IPA , Paraíba Research, Rural Extension and Land Regularization Company – EMPAER and the Rio Grande do Norte Agricultural Research Company – EMPARN develop actions in Dairy Goats Breeding Program (Capragene®) in the states of the southeast and northeast of Brazil. One of the actions of the program consists of genetically evaluating young bucks through the progeny test, in which the breeder needs to have offspring in production to be evaluated. The increase in offspring left by breeders also depends on the reproductive capacity of females, in this sense, the objective of this research was to carry out a genetic study of the characteristics age at first kidding (IPP) and total number of offspring during the doe's reproductive life (NTC) and compare different models to perform the genetic evaluation. The data came from dairy goat herds participating of Capragene®. Two-trait analyzes were performed to adjust IPP and NTC together, assuming three possible combinations of distributions: Normal x Normal (NN), Normal x Poisson (NP) and Normal x Ordinal (NO). To assess the goodness of fit of the models, the Deviance Information Criterion (DIC) was used, the correlation between the observed and adjusted values and the mean square of the estimation error (MSE). The IPP was just over two years and the NTC was over three. Different heritability values for NTC were obtained, ranging from 0.03 to 0.17, according to the different models. The heritability estimated for NTC estimated with the NN model was lower than the models that assumed non-normal distribution for this trait. The estimated heritability for IPP was 0.04, regardless of the distribution assumed for NTC. The non-linear models presented a better fit for NTC, being recommended for the genetic evaluation of this trait.