Estimates of covariance functions for growth of Kordi sheep in Iran using random regression models

Abstract

In the current paper, restricted maximum likelihood estimates of covariance functions for growth records from birth to 360 days of age in Kordi sheep population were estimated using random regression models (RRM). Kordi sheep is an important breed known for adaptability and resistance to unfavorable climatic conditions and suitable feed efficiency in mountainous regions of Iran. Data included 23,720 body weight records of 6730 Kordi lambs collected from 1989 to 2012 at Hossein Abad Kordi sheep Breeding Station in Shirvan city, North Khorasan province, Iran. Fixed effects in the model were lamb sex, birth type, and birth year as well as dam age. Legendre polynomials (LP) of order 3 was used for age to illustrate the average growth curve. In the random part, the functions with different orders were included to model variances associated to direct and maternal effects. Results showed that a model with LP of order 4,3,4,3 for direct additive genetic effects, animal permanent environmental effects, maternal additive genetic and maternal permanent environmental effects was sufficient to model changes in (co)variances with age, according to Akaike’s Information Criterion (AIC) and Bayesian information criterion (BIC). Direct additive genetic and animal permanent environment variances increased with age. Direct heritability (h2a) increased with age from 0.133 ± 0.021 at birth to 0.391 ± 0.034 at 360 days of age. The ratio of animal permanent environmental variances to phenotypic variance (pe2) was 0.047 ± 0.083 at birth, increased with age and reached 0.537 ± 0.034 at the end of the growth trajectory. Maternal heritability (h2m) was 0.152 ± 0.025 at birth decreased to 0.067 ± 0.016 until weaning and then fluctuated between 0.06 and 0.10. The ratio of maternal permanent environmental variances to phenotypic variance (mpe2) was 0.082 ± 0.02 at birth; however, it gradually decreased until the end of trajectory. The first eigenvalues of the coefficient matrices explained 69% and 91% of the total direct additive variance and maternal genetic variance, respectively. Direct genetic correlations ranged from low and negative to positively high, and generally decreased as the distance between tests increased. Current study has demonstrated that RRM have potential for modeling growth of Kordi sheep and a greater response to selection would be expected, if selection focuses on body weights from 30 days of age to 120 days of age.