Use of robust multivariate linear mixed models for estimation of genetic parameters for carcass traits in beef cattle

Authors

S. O. Peters, Berry CollegeFollow
K. Kizilkaya, Iowa State University, Adnan Menderes University
D. J. Garrick, Iowa State UniversityFollow
R. L. Fernando, Iowa State UniversityFollow
E. J. Pollak, USDA-ARSFollow
R. Mark Enns, Colorado State UniversityFollow
M. De Donato, Cornell University, Universidad de OrienteFollow
O. O. Ajayi, Cornell University
I. G. Imumorin, Cornell UniversityFollow

Document Type

Article

Date of this Version

2014

Citation

J. Anim. Breed. Genet. (2014) 1–9 ISSN 0931-2668

doi:10.1111/jbg.12093

Comments

This article is a U.S. government work, and is not subject to copyright in the United States

Abstract

Assumptions of normality of residuals for carcass evaluation may make inferences vulnerable to the presence of outliers, but heavy-tail densities are viable alternatives to normal distributions and provide robustness against unusual or outlying observations when used to model the densities of residual effects. We compare estimates of genetic parameters by fitting multivariate Normal (MN) or heavy-tail distributions (multivariate Student’s t and multivariate Slash, MSt and MS) for residuals in data of hot carcass weight (HCW), longissimus muscle area (REA) and 12th to 13th rib fat (FAT) traits in beef cattle using 2475 records from 2007 to 2008 from a large commercial operation in Nebraska. Model comparisons using deviance information criteria (DIC) favoured MSt over MS and MN models, respectively. The posterior means (and 95% posterior probability intervals, PPI) of v for the MSt and MS models were 5.89±0.90 (4.35, 7.86) and 2.04±0.18 (1.70, 2.41), respectively. Smaller values of posterior densities of v for MSt and MS models confirm that the assumption of normally distributed residuals is not adequate for the analysis of the data set. Posterior mean (PM) and posterior median (PD) estimates of direct genetic variances were variable with MSt having the highest mean value followed by MS and MN, respectively. Posterior inferences on genetic variance were, however, comparable among the models for FAT. Posterior inference on additive heritabilities for HCW, REA and FAT using MN, MSt and MS models indicated similar and moderate heritability comparable with the literature. Posterior means of genetic correlations for carcass traits were variable but positive except for between REA and FAT, which showed an antagonistic relationship. We have demonstrated that genetic evaluation and selection strategies will be sensitive to the assumed model for residuals.