Cross-validation analysis for genetic evaluation models for ranking in endurance horses

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Equine Science

Background:

• Traditional linear and threshold models are commonly used for estimating breeding values in competition horses.
• Recent research suggests Thurstonian models may offer improved prediction accuracy by accounting for race effects.
• Accurate genetic evaluation of ranking traits is crucial for selecting elite competition horses.

Purpose of the Study:

• To compare the predictive ability of linear, threshold, and Thurstonian approaches for the genetic evaluation of ranking traits in endurance horses.
• To identify the optimal genetic models incorporating various random effects (rider, rider-horse interaction, environmental effects) within each approach.
• To assess the heritability and prediction accuracy of different genetic models for endurance horse ranking.

Main Methods:

• Utilized a dataset of 4065 ranking records from 966 endurance horses and a pedigree of 8733 animals.
• Applied eight genetic models for each of the linear, threshold, and Thurstonian approaches, including systematic effects (gender, age, race) and random effects.
• Evaluated prediction ability using a cross-validation approach, calculating correlations between real and predicted racing performances.

Main Results:

• Thurstonian approaches achieved the highest average correlation (0.60) between real and predicted performances, followed by linear (0.58) and threshold (0.25).
• The best models included specific random effects: rider and rider-horse for threshold, rider and permanent environment for linear, and all random effects for Thurstonian.
• High correlations were observed between Thurstonian and threshold models' predicted breeding values (0.85-0.91), while linear and threshold showed lower correlations (0.51-0.65).

Conclusions:

• The Thurstonian approach demonstrates superior predictability for genetic evaluation of ranking traits in endurance horses.
• This method effectively accounts for race-specific competitive levels, leading to more accurate breeding value estimations.
• The Thurstonian approach is recommended for routine genetic evaluations in endurance horse breeding programs.