Bayesian analysis of twinning and ovulation rates using a multiple-trait threshold model and Gibbs sampling

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Statistical Genetics

Background:

• Traditional animal models often focus on continuous traits.
• Analyzing ordered categorical traits, such as twinning rates, requires specialized statistical approaches.
• Previous methods may not fully capture the genetic architecture of complex reproductive traits.

Purpose of the Study:

• To extend the Multiple-Trait Gibbs Sampler for Animal Models to incorporate ordered categorical data.
• To implement a Bayesian threshold model for analyzing multiple continuous and categorical traits simultaneously.
• To evaluate the genetic parameters for twinning and ovulation rates in cattle.

Main Methods:

• Utilized a Bayesian threshold model with data augmentation for ordered categorical data.
• Generated liability values from an unobserved underlying normal distribution for each categorical observation.
• Applied the extended Gibbs sampler to analyze twinning and ovulation rates in a cattle herd.

Main Results:

• Heritability estimates were 0.128 for twinning rate and 0.168 for ovulation rate.
• Permanent environmental effects accounted for 0.103 of variance in twinning rate and 0.079 in ovulation rate.
• High genetic correlation (0.808) and permanent environmental correlation (0.517) were observed between the traits.

Conclusions:

• The Bayesian threshold model effectively analyzes ordered categorical traits in animal breeding.
• This approach allows for more accurate genetic evaluation and selection, particularly for traits like twinning rate.
• Improved selection based on higher heritability on the underlying liability scale can accelerate genetic gain in livestock populations.