Estimation of genetic parameters of Wood's lactation curve parameters using Bayesian and REML methods for milk production trait of Holstein dairy cattle

Bayesian and average information restricted maximum likelihood (AI-REML) approaches were used to estimate variance components and genetic parameters of Wood's lactation curve parameters of milk production traits in three lactations of Iranian Holstein dairy cows. Wood's function parameters (a, b, and c) of each lactation were initially estimated for individual cows using the NLIN procedure of the SAS programme, separately. The variance components of the parameters of each lactation were estimated by the AI-REML method using a single-trait animal model inairemlf90 programme. To do this with the Bayesian method by using the Gibbs sampling technique, the same animal model was used in the gibbs3f90 programme. The estimated heritability values by the AI-REML were 0.018, 0.00, and 0.019 for a parameter; 0.015, 0.007, and 0.02 for b parameter; and 0.23, 0.625, and 0.049 for c parameter in the 1st, 2nd, and 3rd lactations, respectively. The corresponding results from the Bayesian procedure (a posterior means) were 0.019, 0.01, and 0.035 for a parameter; 0.018, 0.019, and 0.043 for b parameter; and 0.024, 0.1, and 0.058 for c parameter, respectively. These obtained results reveal that the differences between the Bayesian and AI-REML methods in terms of estimation of heritability and variance components were small.