Comparative evaluation of Gibbs sampler for milk yield and composition traits in Sahiwal and Tharparkar cattle

Abstract A total number of 802 and 300 records of Sahiwal and Tharparkar cows respectively were collected from history-cum-pedigree sheet for milk production and composition, animals born and reared at Livestock farm unit of ICAR-NDRI Karnal, Haryana, India. The period of study were divided into 6 periods of 5 years each (1990 to 2019). Presented analyzed data were received from selected herd for first lactation milk yield (FLMY), first lactation SNF yield (FLSNFY) and first lactation fat yield (FLFY) in both cattle breeds. Multivariate/Trivariate analyses were carried out by Bayesian approach using Gibbs sampler Animal model. Total heritability estimates for FLMY, FLSNFY and FLFY by Bayesian modeling were valued as 0.20±0.0125, 0.51±0.0114 and 0.45±0.0112 for Sahiwal and 0.46±0.0375, 0.54±0.0195 and 0.52±0.0204 for Tharparkar respectively along with its Monte Carlo Error. Direct genetic and environmental covariance between production traits were ranging from 94553 and 58851 for FLMY-FLFY to 180380 and 185780 for FLSNFY-FLFY in Sahiwal and from 116650 and 95344 for FLSNFY-FLFY to 245820 and 227400 for FLMY-FLFY in Tharparkar respectively. Comparable estimates of heritability and breeding values for ranking of sires were obtained by BLUP and Bayesian analyses from the presented data. Estimates of genetic and phenotypic (co)variance components and heritability estimates obtained using LSML which were positive but very medium to high for Sahiwal while genetic correlations were positive and high in Tharparkar. However, Gibbs Sampling (GS) is an advance and comparatively correct method of choice for many applications because of its greater robustness with comparable but slightly higher estimates of genetic parameters than LSML. The Gibbs Sampling method is flexible and dependable practice for the genetic evaluation and it can be useful in the breeding programs for highly heritable traits in Tharparkar as compared to low estimates in Sahiwal from present work.