Estimation & Prediction

Comparison of survival analysis and a linear mixed model for predicting genetic parameters of tail-biting incidence in pigs

---

Abstract

Genetic parameter estimates for the incidence of being a tail-biting victim have been primarily analysed using linear mixed models. These models usually do not handle censored records (common in longitudinal tail-biting data) and the non-linearity of binary traits. Consequently, the genetic variance is not adequately captured when using linear models, resulting in low heritability estimates. However, survival analysis can better address both issues of censoring and non-linearity in longitudinal data and would improve the magnitudes of estimates of genetic parameters for tail-biting in pigs. This study aimed to compare estimated genetic parameters for the risk of being a tail-biting victim using survival time and linear mixed models in three pig lines. A longitudinal dataset (n = 33,758 pigs) collected between August 2022 and December 2023 from an Australian pig breeding company was used. Tail biting was assessed weekly for an average of 130 days, from weaning to 19 weeks, when a performance weight was taken. The event was defined as tail biting, being absent or present (0/1) from weaning until the performance recording date. Survival time was defined as the time interval in days from weaning until the onset of the first tail-biting event recorded. A survival analysis was combined with a linear animal model to estimate the variance components and heritability. The likelihood of surviving a tail-biting incident (i.e., remaining unbitten) differed significantly between lines (p < 0.001). The heritability (± standard errors) of the survival time combining the three lines was 0.15 ± 0.02, which was higher than that of the estimated heritability (0.06 ± 0.01) obtained using the classic linear model. The estimated within-line heritability for survival time ranged from 0.15 ± 0.02 to 0.19 ± 0.04. These findings suggest that more of the genetic variance is captured using survival analysis than conventional linear models. This implies that there is more potential for reducing the risk and incidence of tail-biting through breeding than previously thought, allowing breeders to make more informed selection decisions and optimise breeding programs globally.

Keywords: 2026

How to Cite:

Reda, A., Gurman, P., Harper, J., Hermesch, S. & Aldridge, M., (2026) “Comparison of survival analysis and a linear mixed model for predicting genetic parameters of tail-biting incidence in pigs”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2286251. doi: https://doi.org/10.31274/wcgalp.23894

Rights: 1