A genetic analysis of mortality in pigs

Area of Science:

• Animal genetics
• Statistical modeling
• Livestock breeding

Background:

• Mortality analysis is crucial in livestock for breeding programs.
• Understanding genetic influences on pig mortality, particularly stillbirths, is essential for improving herd efficiency.
• Hierarchical Bayesian models offer advanced statistical approaches for analyzing complex biological data.

Purpose of the Study:

• To investigate genetic variation for mortality in Danish Landrace and Yorkshire pigs.
• To evaluate the fit and predictive performance of various hierarchical Bayesian models for mortality data.
• To assess the implications of genetic variation in mortality for pig breeding.

Main Methods:

• Analysis of mortality data in Danish Landrace and Yorkshire pig breeds.
• Fitting zero-inflated and standard hierarchical Poisson, binomial, and negative binomial Bayesian models.
• Utilizing Markov chain Monte Carlo (MCMC) methods for model fitting and parameter estimation.

Main Results:

• The standard binomial hierarchical model provided the best data fit for both breeds.
• The hierarchical zero-inflated negative binomial model demonstrated superior predictive ability for stillbirth distribution.
• Inclusion of genetic variation significantly improved the fit of both the binomial and zero-inflated negative binomial models.
• Estimates of additive genetic variance for stillbirth probability were substantial in both breeds.

Conclusions:

• Genetic variation plays a significant role in pig mortality, specifically stillbirths.
• Different models offer complementary strengths for analyzing mortality data: binomial for fit, zero-inflated negative binomial for prediction.
• These findings have direct implications for genetic selection strategies in pig breeding programs to reduce mortality.