Genetic evaluation of growth traits in Nellore cattle through multi-trait and random regression models

• Published in: Czech Journal of Animal Science Vol. 63; no. 6; pp. 212 - 221
• Main Authors: Teixeira, Bruno Bastos, Mota, Rodrigo Reis, Lôbo, Raysildo Barbosa, da Silva, Luciano Pinheiro, Souza Carneiro, Antônio Policarpo, da Silva, Felipe Gomes, Caetano, Giovani da Costa, Silva, Fabyano Fonseca e
• Format: Journal Article Web Resource
• Language: English; Czech; Slovak
• Published: Prague Czech Academy of Agricultural Sciences (CAAS) 01.01.2018; Czech Academy of Agricultural Sciences
• Subjects: Animal production & animal husbandry; Bayesian analysis; body weight; Cattle; Environment; Estimates; genetic parameters; Growth curves; Heritability; Life sciences; Mathematical models; Parameter estimation; Polynomials; Productions animales & zootechnie; Regression analysis; Regression models; Sciences du vivant; Variance
• ISSN: 1212-1819; 1805-9309; 1805-9309
• DOI: 10.17221/21/2017-CJAS

We aimed to evaluate different orders of fixed and random effects in random regression models (RRM) based on Legendre orthogonal polynomials as well as to verify the feasibility of these models to describe growth curves in Nellore cattle. The proposed RRM were also compared to multi-trait models (MTM). Variance components and genetic parameters estimates were performed via REML for all models. Twelve RRM were compared through Akaike (AIC) and Bayesian (BIC) information criteria. The model of order three for the fixed curve and four for all random effects (direct genetic, maternal genetic, permanent environment, and maternal permanent environment) fits best. Estimates of direct genetic, maternal genetic, maternal permanent environment, permanent environment, phenotypic and residual variances were similar between MTM and RRM. Heritability estimates were higher via RRM. We presented perspectives for the use of RRM for genetic evaluation of growth traits in Brazilian Nellore cattle. In general, moderate heritability estimates were obtained for the majority of studied traits when using RRM. Additionally, the precision of these estimates was higher when using RRM instead of MTM. However, concerns about the variance components estimates in advanced ages via Legendre polynomial must be taken into account in future studies.