Random regressions to model growth in Nilagiri sheep of South India

Abstract

Random regression models (RRM) have been recognised as suitable for analysis of longitudinal data like growth. Growth in Nilagiri sheep was modeled using RRM to estimate genetic parameters and results obtained were compared with values already estimated using conventional methods A mixed RRM with contemporary group of year and season of birth, sex, litter size and age of dam at lambing, as fixed effects and direct additive, maternal genetic, maternal permanent environmental and individual permanent environmental factors, as random effect was used in different orders of fit. Age at recording of weight was included as the control variable. The RRM with fourth order for all random effects and heterogenous error variance (9 classes) was found to have the best fit. The pattern of variation over ages was similar for direct and maternal genetic variance, while maternal permanent environmental variance showed contrasting trend. Direct additive and maternal genetic variance showed a parabolic trend up to 36 months of age with maximum variance between 12 and 18 months. Phenotypic and individual permanent environmental variance showed similar increasing trend with age. The heritability estimates obtained through RRM for 3 month (3W), 6 month (6W), 9 month (9W), 12 month (12W), 18 month (18W) and 24 month (24W) body weights were 0.131, 0.240, 0.275, 0.254, 0.131 and 0.068, respectively. The values for 3W and 18W were similar to those obtained through univariate analysis, while other estimates were on the higher side. The trend of estimates over ages was similar between RRM and conventional methods Values of direct and maternal genetic correlations were positive and high. This finding of high positive genetic correlation among body weight at different ages was also evident from the trend of Eigen functions estimated from genetic (co)variance matrix. The correlation between breeding values estimated from RRM and conventional REML methods: were 0.904, 0.331, 0.774 and 0.771 for 3W, 6W, 9W and 12W, respectively. The estimates of heritability obtained were precise for ages up to 24 months and values indicate good scope for improvement through selection.