Bayesian analysis of random regression models to model test-day somatic cell score of primiparous holstein cattle in Iran

Abstract

Several functions (Ali-Schaeffer curve, Legendre polynomials, and the combination of Legendre polynomial and the Wilmink function) were used to adjust the lactation curve of somatic cell score (SCS) of primiparous Holstein cows in Iran. Data included 204,374 test-day records collected on 29,215 animals over 10 freshening years, from 2002 to 2011. The model included herd-year of calving (HY), additive genetic (AG), permanent environmental (PE) and residual effects as random and age-season of calving (A-S) as fixed regression effects. In addition, contemporary group effect (herd-year-month of test-day) was included as fixed effect. Residual variances were modelled considering 1, 4, 7, 10 or 20 classes of days in milk (DIM). Estimates of residual variance were quite similar, rating from 1.30 to 1.56. Estimates of daily heritability were practically constant until DIM 250 of lactation (around 0.03) and increased toward the end of lactation (0.07), thereby indicating the presence of low AG variation for SCS in this population of Holstein cattle. Regarding low heritabilities, a relatively low genetic progress will be expected following selection. The chosen model (according to the deviance- and Bayesianinformation criterion) had Legendre polynomials with six coefficients for A-S effect, four coefficients for HY effect, five coefficients for additive effect, and six coefficients for PE effect with a homogeneous error structure.