Effect of heat stress on production traits of Holstein cattle in Japan: parameter estimation using test-day records of first parity and genome wide markers

Abstract

Variance components were estimated for test-day milk (kg), fat (kg), protein (kg), and somatic cell score (SCS) using Gibbs sampling applying single step genomic BLUP (ssGBLUP). The phenotypes were 820,573 (752,514 for SCS) test-day records of 233 dairy farms in Japan with 1,170 randomly selected genotyped cows; the sample excluded Hokkaido (an island in the northern part of Japan where heat stress was minimal). Matrix H, which combined additive and genomic relationships, included 93,725 (86,435 for SCS) phenotyped cows and 8,401 genotyped bulls and cows. Dairy farms were linked to meteorological offices based on their covering areas for the announcement of weather forecasts, advisories, and warnings that reflect local climates, and then each phenotype was linked to the average temperature–humidity index (THI) for up to four days before each test day. Heat stress was defined as changes in phenotypes per unit increase in THI when THI increases were above the threshold of 60, and additive genetic (AG) and permanent environmental (PE) effects of the heat tolerance of each cow were added to the Japanese national genetic evaluation model. PE variances of heat tolerance were larger than AG variances of heat tolerance in all four traits. These results suggest that accumulation of various non-AG factors may affect the heat tolerance of individual cows. Average AG correlations between general effect and heat tolerance were negative, except for SCS. Therefore, antagonistic characteristics of the two AG effects should be carefully considered. With appropriate determination of THI, the use of total AG effects could be a feasible option. Inclusion of later parity is required for further study, as they are more affected by heat stress than the first parity. However, genetic evaluation of heat tolerance would be feasible.