Estimates of Variance Components and Heritability Using Random Regression Models for Semen Traits in Boars

Abstract

It has been proven that the random regression model has a great advantage over the repeatability model in longitudinal data analysis. At present, the random regression model has been used as a standard analysis method in longitudinal data analysis. The aim of this study was to estimate the variance components and heritability of semen traits over the reproductive lifetime of boars. The study data, including 124,941 records from 3,366 boars, were collected from seven boar AI centers in South China between 2010 and 2019. To evaluate alternative models, we compared different polynomial orders of fixed, additive, and permanent environment effects in total 216 models using Bayesian Information Criterions. The result indicated that the best model always has higher-order polynomials of permanent environment effect and lower-order polynomials of fixed effect and additive effect regression. In Landrace boars, the heritabilities ranged from 0.18 to 0.28, 0.06 to 0.43, 0.03 to 0.14, and 0.05 to 0.24 for semen volume, sperm motility, sperm concentration, and abnormal sperm percentage, respectively. In Large White boars, the heritabilities ranged from 0.20 to 0.26, 0.07 to 0.15, 0.10 to 0.23, and 0.06 to 0.34 for semen volume, sperm motility, sperm concentration, and abnormal sperm percentage, respectively.