Use of sensors for the detection and genetic evaluation of heat stress in dairy cattle.

Reaction of milk production traits is generally used to evaluate heat stress (HS) in dairy cattle. However, milk recording data are in most cases only obtained once per month which drastically limits the number of records obtained during hot days. A solution could be to add sensor information that is obtained daily. In addition, sensors are already used for estrus detection in heifers and cows, but also for health and welfare of animals. The objectives of this study were thus to evaluate the usability of sensor data to detect HS and to assess the possible gain for HS genetic evaluation. Daily SenseHub collar records for activity, rumination and eating time were obtained from 2019 to 2022 for 453 Walloon Holstein cows. Meteorological data, and fat- and protein-corrected milk (FPCM) were also obtained from 2015 to 2022 for 1,740 cows from the same herds. The thresholds at which the different traits start to be affected by HS were estimated at a temperature-humidity index (THI) of 37, 58, 62, and 66, respectively, for FPCM, eating, activity and rumination time. The thresholds were clear for sensors traits while FPCM decreased all along the THI scale. Variation of the sensors traits with the THI were higher in comparison with FPCM. A genetic random regression reaction norm model using extracted pedigree information was performed with these different thresholds to fit at the maximum the trait reaction to THI. Regarding the heat tolerance genetic reaction norm effect, sensor traits showed positive or slightly negative genetic correlations with FPCM (activity time: 0.32 ± 0.54, rumination time: 0.05 ± 0.41, eating time: −0.15 ± 0.36). In addition, these traits respectively presented heritability values of 0.14 ± 0.06, 0.18 ± 0.05, and 0.14 ± 0.05 at the thresholds and these values were similar at high THI. In conclusion, sensors data could be valuable tools to detect HS because they react well and their THI thresholds were easy to detect. They could also be useful in genetic evaluation for HS because their daily records provides information for all HS events. Sensor traits were heritable including at high THI and the genetic correlation of the effects on activity and on FPCM was positive. On this basis, activity time seems to be the most interesting sensor trait for genetic evaluation of HS.