Modeling heat stress in dairy cattle using high spatio-temporal resolution observational and grid-interpolated weather data
Thermal stress affects milk production and quality in dairy cattle. High spatio-temporal resolution weather data is necessary for evaluation of thermal stress effects on milk production traits for management purposes to mitigate production losses. The first objective of this study was to analyze the effects of heat stress on milk production in a tropical environment. The aim was to detect heat stress thresholds, milk yield loss and individual animal variations using random regression models. Grid-interpolated daily average temperature and temperature humidity index (THI) were suitable indicators of heat stress. In the second objective, high spatio-temporal resolution weather data were applied to analyze thermal stress effects on herd average milk traits in a temperate environment. A validation analysis was performed to compare the correlation between weather station data (WSD) and gridded numerical weather prediction (NWP) model data. Results indicated high correlation coefficients (r) of 0.97 for daily average temperature and THI and 0.84 for relative humidity in southern region of Bavaria, Germany. This implied that missing WSD can be adequately filled-up by NWP data. However, care should be taken in mountainous regions to account for both vertical and horizontal topographical profiles. Both WSD and NWP daily average temperature and THI were suitable indicators of heat stress in cattle. Herd average reaction norms for milk production traits and the thresholds for heat stress were estimated for Bavarian conditions. Cooling THI-day, defined as the average hourly THI above a set threshold, was formulated and applied as a heat stress indicator. Models combining THI and cooling THI-days indicated that cooling THI-day provided additional significant (p<0.001) information to THI models. Results show that cooling THI-day would be a suitable heat stress indicator when cold stress is not significant and the threshold for heat stress for the given trait can be estimated.