Fitting random regression models with Legendre polynomial and B-spline to model the lactation curve for Indian dairy goat of semi-arid tropic.

Abstract

The present investigation aimed at genetic evaluation of tropical Indian dairy Jamunapari goat using random regression models (RRM) for the estimation of genetic parameters in the first three lactations across test days (TD) and also to come out with a pragmatic breeding plan in the nucleus. Variations in the lactation curves were modelled using 67,172 TD milk yield (TDMY) records. To obtain adequate and parsimonious models for the estimation of genetic parameters, orthogonal Legendre Polynomials (LP) and B-splines (BS) were compared. The analysis was carried out using a single-trait RRM approach. Average TDMY was 0.72, 0.81 and 0.79 kg in 1st to 3rd parities that also had 4th TD peak yield in common. BS function resulted in robust genetic parameters and a smoother curve for lactation as compared to LP. Maternal effects were evaluated and then dropped from the final model, owing to no significant contribution to the genetic variance. The best RRM was a quadratic BS function with six knots for the mean trend, curves of additive genetic, animal permanent environmental (c2 ) and 22 classes of residual variance. Additive variances and heritability (h2 ) estimates were higher in the early lactation. For first parity, the estimates of h2 varied between 0.19 to 0.35 across TD. Moderate h2 estimate suggests further scope for selection using desirable combinations of TD over the lactation. We observed a very high variance due to c2 across TD in three lactations. Genetic correlations were positive and larger between adjacent TDMY and weakened for distant TDMY. Looking into the robust estimates of genetic parameters and better fitting of lactation curve, we suggest the use of B-spline function for regular genetic evaluation of Jamunapari goat.