Using milk tanker pickup and weather data to quantify the impacts of heat stress on milk production in Australia

Meyer, R. S., Graham, A. -M., Hepworth, G. & Eckard, R. J. (2021). Using milk tanker pickup and weather data to quantify the impacts of heat stress on milk production in Australia. 24th International Congress on Modelling and Simulation., pp.148-154. Modelling and Simulation Society of Australia and New Zealand. https://doi.org/10.36334/modsim.2021.b5.meyer.

Access Status

Open Access

URI

https://hdl.handle.net/11343/304772

DOI

10.36334/modsim.2021.b5.meyer

Abstract

Due to high metabolic heat production, dairy cows are more sensitive to heat stress than other livestock. This has consequences on the animal’s welfare and productivity. Reductions in milk production can occur from relatively mild temperatures with the degree of impact increasing with increasing severity of heat exposure. The degree to which heat stress impacts milk production, farm income, and milk supply is becoming increasingly important as the duration and frequency of heat waves increases. This analysis aims to provide estimates of the sensitivity of milk production on Australian dairy farms under heat stress situations. The milk tanker pickup data from three dairy companies was used to investigate the on-farm impacts of heat stress on milk production in three regions of Australia. Milk production and weather data were matched based on the postcode of the farm and Bureau of Meteorology weather station. Weather data were used to calculate the temperature-humidity index (THI) using a formula typically used in Australian studies. The THI is commonly used to quantify the effects of heat stress by combining the effects of temperature and relative humidity. Data that did not meet pre-defined quality criteria were eliminated from the analysis. For instance, records that occurred three or more days since the previous milk pickup, data from farms with less than one year of observations, and years in which more than 10 records were missing between October and April, inclusive, were excluded. Over 960,000 records from 1,286 farms are included in the regional analysis. Linear mixed effect models were fitted to the data from each of three regions using R. The estimated response of milk volume (L) and milk solids (kg) and associated 95% confidence intervals (CIs) are summarised in Table 1. Models using THI values averaged over 7-days, consistently performed better than those averaged over shorter periods. THI values based on minimum and average temperatures performed slightly, but consistently, better than those based on maximum temperature.