Comparison of Random Regression Models with Different Order Legendre Polynomials for Genetic Parameter Estimation on Race Completion Speed of Arabian Horses.
Authors: Önder Hasan, Şen Uğur, Piwczyński Dariusz, Kolenda Magdalena, Drewka Magdalena, Abacı Samet Hasan, Takma Çiğdem
Journal: Animals : an open access journal from MDPI
Summary
# Editorial Summary Researchers analysed over 233,000 race speed records from Arabian horses competing in Turkey between 2005 and 2016, comparing four random regression models using different orders of Legendre polynomials to estimate heritability and genetic parameters for racing speed (measured in metres per second). The L(2,2) model—using second-order polynomials for both additive genetic and permanent environmental effects—proved most statistically reliable based on Akaike and Bayesian information criteria, with a mean heritability estimate of 0.294 (±0.0746), suggesting that approximately 29% of variation in race speed is genetically determined. Heritability remained stable across the racing lifespan under the L(2,2) model because permanent environmental variances showed a consistent downward trend, whereas the more complex L(3,3) model produced substantially greater variability in heritability estimates (±0.1018), reducing confidence in its predictions. For breeding programmes and performance prediction in Arabian racing, the L(2,2) model offers a practical, statistically robust approach to evaluating genetic merit for speed across repeated race observations, avoiding unnecessary model complexity that compromises reliability without improving fit. These findings have direct application for stud selection and pedigree analysis in endurance and racing contexts, where longitudinal performance data inform genetic decisions.
Read the full abstract on PubMed
Practical Takeaways
- •For breeders selecting Arabian horses for racing speed, genetic merit accounts for approximately 29% of performance variation, suggesting substantial opportunity for selective breeding improvement
- •The L(2,2) model provides a reliable, parsimonious tool for evaluating genetic parameters in repeated racing performance measurements across the population
- •Understanding heritability estimates helps breeders set realistic expectations for performance gains and make informed breeding decisions based on genetic potential rather than environmental factors alone
Key Findings
- •The L(2,2) Legendre polynomial model provided the best fit for estimating heritability of racing speed in Arabian horses according to AIC and BIC criteria
- •Heritability of race completion speed ranged from 0.285 to 0.302 across different polynomial models tested
- •The L(2,2) model showed a steady decreasing trend in permanent environmental variances, providing stable heritability estimates
- •Analysis of 233,491 race speed records from Arabian horses demonstrated that lower-order polynomial models can adequately capture genetic variation in racing performance