Consideration of heat stress in multiple lactation test–day models for dairy production traits

Abstract

In the previous study applying a pedigree based random regression test–day models using multiple lactation records, variance components estimated by Gibbs sampling did not converge for fat and protein (kg) when additive genetic (AG) and permanent environment (PE) effects of heat tolerance (HT) were assigned to each lactation which was treated as different traits; therefore, a common effect was assumed across lactation. In the follow–up analysis, we found many positive estimates of AG effects of HT for milk, fat, and protein (kg), and vice versa (negative estimates) for somatic cell score (SCS). This was expected given the nature of random regression effects applied; however, may need to study further in light of the heat stress definition (phenotypic changes per unit increase in the temperature–humidity index (THI) when THI increases above a threshold). As a simple solution, we pre–adjusted phenotypes by adding / subtracting the expected decrease / increase of phenotypes using the results of the previous study. Gibbs samplings converged for all traits with the model assigning AG / PE effects of HT to each lactation. Positive genetic trends in protein (kg) observed in the previous study disappeared and it was difficult to find trends for all traits. The finding may be reasonable as animals have never been selected for HT directly. The new methodology would provide more reasonable estimates than the previous study.