Genetic evaluation of Murrah buffalo by fitting random regression models using B-spline function

Abstract

In the present study, random regression models with both random and fixed effect regressions fitted by B-spline functions were used to estimate genetic parameters with 5 knots. The common effects for all models were month of recording, year of recording as fixed regressions on daily milk yield records and random regressions for additive genetic and permanent environmental effects. Among different model studied with B-spline functions considering homogeneity and heterogeneity of residual variances, the model that best fit the data was BSQ5H1 (quadratic B-spline model of polynomial order 6 with homogenous residual variance) having knot at 5^th, 80^th ,155^th, 230^th and 305^th DIM for the first lactation daily milk yield records of Murrah buffaloes. For BSQ5H1, the R² value with estimated arithmetic mean of daily milk yield for first lactation was 93.7%. The highest values of additive genetic (1.22 kg²) and permanent environment variance (5.27 kg²) were observed in the initial (5^th) and last (305^th) DIM of lactation. Heritability estimates ranged from 0.07±0.05 to 0.21±0.07. It was observed that the heritability estimates were higher in early and late lactation while lower in mid of lactation (DIM 110 to 154). The estimated value of genetic correlation ranged from -0.50 (DIM 5 with DIM 174 to DIM 187) to 1.00. The DIM 5 had negative genetic correlations with peak yield DIM 65 to DIM 243. The rank correlation between sires with 6^th order of Legendre polynomial function (RLP6) with BSQ5H1 was more than 0.99 and highly significant (P<0.001). The high rank correlation between two models, indicated that both are equally efficient for genetic evaluation of Murrah buffaloes.