Genetic analysis of live weight of Sardi sheep using random regression and multi-trait animal models

Abstract

Genetic parameters and breeding values for live weight from birth to weaning (85 days) of Sardi sheep were estimated using random regression (RR) and multi-trait (MT) animal models. A longitudinal data set for live weight of 44,509 lamb progeny of 605 sires and 18,569 dams were used to describe variance components along growth curve trajectory using RR model. This data set was collected in 89 flocks between 2001 and 2012, with a total of 45,941 animals in the pedigree. Eleven RR models were compared based on the log likelihood and Akaike information criterion. The model with the best fit was found to include quadratic Legendre polynomials for direct and maternal genetic effects and direct permanent environment effects and a linear polynomial for maternal permanent environment effects, including covariance between direct and maternal genetic effects and a heterogeneous residual variance. Direct and maternal heritabilities were 0.11 and 0.03, respectively, at the first day of age, and 0.37 and 0.10, respectively, at 85 days of age. The increase of heritability and genetic variance indicated that there was a large potential for selection of Sardi sheep and suggested that selection during the study period would be more effective if it was done on weaning weight. The low genetic correlation between early weight and weaning weight suggests that these traits are controlled by different genes. Genetic parameters obtained at the selected ages (i.e. 10, 30, 70 days) using MT and RR models were generally similar, but the two methods produced different ranking of genetic merit. This study has demonstrated the possibility of applying RR model for routine genetic evaluation of Sardi sheep.