Application of random regression models for assessment of direct and maternal genetic parameters of growth characters in large white Yorkshire crossbred pigs.

Abstract

Body weight is one of the significant profitable traits in swine production. Data on body weight was recorded at monthly intervals from Large White Yorkshire crossbred pigs (75 % LWY & 25% Local desi) kept up at All India Coordinated Research Project on Pig, Sri Venkateswara Veterinary University, Tirupati, Andhra Pradesh. In this study body weight data from 655 animals belonging to 22 sires and 45 dams was subjected to random regression analysis using Legendre polynomials of various orders of fit. Wombat software was used to estimate the covariance components and genetic parameters using the derivative free restricted maximum likelihood method. Best order of fit was identified based on information criterion. The Legendre polynomial with orders of fit 3 for direct genetic and maternal genetic effects and homogeneous residual error variance (11 classes) was considered to be the best fit based on Log l, Akaikes information and Bayesian information criterion. The quadratic Legendre polynomial revealed the highest Log l and the lowest Akaikes information and Bayesian information values (3,3,0,0). The variance for intercept (L0) was found to be large (0.51) for direct additive effects and low (0.04) for maternal genetic effects. The trajectories for 1st and 2nd Eigen functions together accounting for > 99% of genetic variation.