Threshold-linear versus linear-linear analysis of birth weight and calving ease using an animal model: I. Variance...

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Statistical Genetics

Background:

• Accurate estimation of genetic parameters for traits like birth weight and calving difficulty is crucial for genetic improvement in cattle.
• Traditional linear models may not fully capture the discrete nature of calving difficulty, necessitating advanced modeling approaches.

Purpose of the Study:

• To analyze birth weight and calving difficulty in Gelbvieh cattle using various Bayesian statistical models.
• To compare the performance of univariate, bivariate, linear, and threshold models in estimating genetic parameters.
• To assess the suitability of simulation studies for validating model estimates.

Main Methods:

• Bayesian methodology with Gibbs sampling was employed.
• Univariate linear, bivariate linear, univariate threshold, and joint threshold-linear models were utilized.
• Analysis included 26,006 field records from Gelbvieh cattle, with simulated populations for validation.

Main Results:

• Direct heritability for calving difficulty ranged from 0.18 to 0.23 (univariate) and 0.18 to 0.21 (bivariate).
• Direct heritability for birth weight was consistently around 0.25-0.26 across models.
• Genetic correlations between direct effects for birth weight and calving difficulty were high (0.79-0.81).

Conclusions:

• Bayesian threshold and joint threshold-linear models provide robust estimates for calving difficulty heritability.
• The high genetic correlation suggests selection for lighter birth weight could reduce calving difficulty.
• Simulation studies confirm the utility of posterior means for parameter estimation in complex models.