The aim of the present study was estimation of (co)variance components and genetic parameters for body weight of Moghani sheep, using random regression models based on B-Splines functions. The data set included 9165 body weight records from 60 to 360 days of age from 2811 Moghani sheep, collected between 1994 to 2013 from Jafar-Abad Animal Research and Breeding Institute, Ardabil province, Iran. Random regression models were employed to analyze the data. Contemporary groups (year – season of birth – sex – birth type – dam age at the birth) and fixed regression of body weight on age were considered as fixed parts of the models. Random regressions of direct additive genetic, permanent environment, maternal additive genetic and maternal environment were random parts of the models. Linear and quadratic B-Spline functions with two or three coefficients were fitted for fixed and random regressions of the models. A heterogeneous structure of residual variance was considered in five age classes. Variance components were estimated by Average Information algorithm of Restricted Maximum Likelihood (AI-REML). Different models were compared based on Akaike Information criterion (AIC) and Schwarz Bayesian Information criterion (BIC). According to both criteria, the best model was a model with quadratic B-Spline functions with 3, 3, 3, 2 and 2 coefficients for fixed regression and random regressions of direct additive genetic, permanent environmental, maternal additive genetic and maternal environmental effects, respectively. According to this model, low to moderate estimates of direct heritability (0.135 to 0.330) and moderate to high estimates of coefficient of permanent environmental effects (0.229 to 0.613) were obtained, while estimates of maternal heritability (0.05 to 0.14) and coefficient of maternal environment (less than 0.01) were low or negligible in all ages.
