Karin Elisif Schütz

Heat stress in dairy cows poses a significant challenge, particularly on pasture where cows are exposed to variable climates. The study objective was to use weather variables and established heat load indices to predict grazing dairy cow respiration rate (RR) and drooling. Data from 483 Friesian × Jersey crossbred cows across 7 New Zealand farms was collected over 2 consecutive 5-d periods per farm. RR and drooling were observed up to 6 times per cow daily across 3 time periods (0900-1100h, 1200-1400h, 1700-1900h) and matched to 15-min weather data. Although farmers were asked to minimize heat mitigation, the observational nature of the study meant shade and/or sprinklers were available periodically on 6 farms as part of their standard management. Weather variables (temperature, humidity, wind speed, solar radiation), Temperature Humidity Index (THI) and 2 Grazing Heat Load Indices (GHLI 2023; GHLI 2025), were used to model RR and drooling. Data subsets (1200-1500h and 0900-1800h) were analyzed separately, with the former explaining more RR variance (65.4 vs. 61.3%). GHLI (2025), GHLI (2023) and THI explained 59.8%, 59.7%, and 58.7% of RR variance, respectively. Nonlinear relationships between RR and individual weather variables were detected (P < 0.001), and temperature, solar radiation, and wind speed together explained approximately 9% more RR variation than indices. RR was stable to about 19°C, with sharper increase above 23°C and 0.99 MJ/m² solar radiation; wind speed had a mitigating effect. Despite both behaviors being linked to heat stress, RR explained just 30% of drooling variance. Drooling probability reached 50% at a GHLI (2025) of 82; however, < 20% drooling variance was explained, likely due to between-farm variability. These limitations aside, results illustrate the nonlinear effects of weather variables on heat load responses of grazing dairy cows; given large between-farm variation, individual weather variables may be more accurate predictors than linear indices. However, farmers require clear guidelines for decision making, and indices are straightforward; thus, continued research should refine nonlinear predictive models that could be used in conjunction with on-cow devices. Models should be validated across different climatic regions, farming systems, breeds, and individual animal characteristics, as doing so will support sustainability and resilience in dairy farming in the face of climate change.

Dairy calves reared for beef are commonly fed restricted milk allowances pre-weaning, which may negatively influence their welfare and growth. We investigated the effects of 2 milk allowances on the behavior and liveweight of 22 Holstein-Friesian bull calves in a pasture-based system. Calves were managed indoors in one group and individually offered 5 or 10 L of milk/d via an automatic milk feeder until 5 wk of age (n = 11 calves/treatment) when they were transferred onto pasture. Behavior (at meal, milk, hay and bentonite feeders, ruminating, exploring, drinking, lying, grooming, play running and cross-suckling) was recorded for 48.5 h at 4 wk of age using video cameras and leg-based accelerometers. Milk feeder data were automatically recorded over 2 wk (number of visits, milk consumption and drinking speed). Liveweight was recorded weekly until 18 wk of age and monthly thereafter until 1 yr of age. At 4 wk of age, calves offered 10 L had fewer unrewarded visits to the milk feeder (3.1 vs 38.2 no/d), fewer visits to the meal (0.7 vs 2.8% of observations) and hay feeders (0.4 vs 0.8% of observations) and drank the milk slower (473 vs 388 mL/min) than calves offered 5 L. Calves offered 10 L/d also spent more time (78 min/24 h) lying, but took more steps (303 vs 245 no/24 h), and engaged more in self- and allo-grooming behavior (self-grooming: 3.2 vs 1.8%, receiving: 1.0 vs 0.5% of observations) compared with calves offered 5 L. Calves offered 10 L/d were, on average, 4.3 kg heavier at 5 wk of age compared with the calves offered 5 L/d even though they did not always drink their full allowance (average milk intake: 6.6 vs 4.9 L/d). Liveweights of the 2 treatment groups were similar at 1 yr of age. Calves offered more milk grew faster pre-weaning and showed indications of being less hungry and more content, thus confirming previous research that feeding calves higher milk allowances is associated with enhanced welfare and early growth.

This study aimed to evaluate a tri‐axial accelerometer attached to a collar to determine the ability to accurately predict rumination, lying, grazing, and head shaking behaviour of ewes on pasture. Collars with accelerometers recording at 1 Hz were attached to 30 Romney type ewes managed in groups of 5. The behaviour of each group was observed for 1 day between 0830 and 1630 h using 30‐s instantaneous scan observations. Head shaking behaviour was stimulated in the yards using a toy water pistol and analysed as present or not present. Grazing activity was well classified by a Random Forest algorithm using the out‐of‐bag error estimation with an accuracy of 83.4% (95% CI: 82.9–83.9%). The accuracy for ruminating activity was 82.6% (95% CI: 82.1–83.0%), and for lying activity 85.5% (95% CI: 85.2–86.0%). Grazing, ruminating, and lying activities were predicted with sensitivity values of 80.0%, 68.0%, and 79.4%, respectively. The specificity values for grazing, ruminating and lying were 86.6%, 87.6%, and 89.3%, respectively. The precision values for grazing, ruminating, and lying were 85.0%, 65.1%, and 81.7%, respectively. However, the accuracy to classify head shaking behaviour was only 52.2% with poor performance metrics potentially due to the low sampling frequency of the accelerometers in combination with a behaviour of short duration. The results show that accelerometers can monitor the behaviour of ewes on pasture; however, the sampling frequency likely needs to be higher to detect behaviours of short durations.

The study of farm animal behaviour often relies on labour‐intensive observations, which can be impractical in extensive systems. Advances in technology (e.g. accelerometers and global positioning satellite (GPS) devices) can be used to record behaviours frequently, long‐term and remotely. Monitoring behaviour of livestock, in combination with other key information, such as environmental factors (e.g. weather, spore counts), feed quality and quantity, and animal performance, can be used to gain a deeper understanding of animal health and welfare. Changes in behaviour may be used as early warning indicators for important welfare and productivity issues, such as facial eczema and gastrointestinal parasites. This article describes current research where accelerometer sensors have demonstrated high accuracies (>80%) in tracking behaviours like grazing, ruminating and lying, and have successfully identified behavioural changes associated with gastrointestinal parasites. Research currently being undertaken in New Zealand to evaluate the ability of accelerometers and other technologies, such as acoustic sensors and GPS, to record measures associated with facial eczema, heat stress (e.g. respiration rate) and maternal behaviour of sheep, will be discussed. In summary, technologies including the use of accelerometers have the potential to guide management decisions and ultimately improve animal health, welfare and production efficiency on‐farm.

Limited on‐animal sensor technology is available to monitor sheep behaviour. We investigated the accuracy and suitability of Hobo Pendant G and IceQube loggers for recording lying behaviour, and steps (IceQube) in pasture‐managed sheep. Twenty mixed‐breed ewes wore both sensors on their hind legs for 20 consecutive days. Lying behaviour was recorded for 12 h daily (8:00–20:00 h) using video cameras. For step count, four 15‐min intervals per sheep ( n = 20) were analysed. Bland–Altman plots revealed that the percentage of time recorded as lying per hour by both loggers closely matched video observations. For Hobo/Video and IceQube/Video comparisons, 93% and 99% of data respectively were within ±2% of video‐observed lying time. The two loggers were also closely matched, with 93% of data within ±2%. Sheep spent on average 13.7 ± 1.24 h (SD) lying per day (range: 7.4–19.2 h). IceQube underestimated steps compared to manual video counts (mean difference; 7 steps, limits of agreement −21.5 to 35.5). After 5 days, one sheep had minor rubbing (IceQube) and after 20 days, 5 sheep had minor or major rubbing from the loggers (3 IceQubes, 4 Hobos). Both loggers accurately captured lying time, but improved attachment materials are needed to prevent rubbing.

This systematic scoping review examines the past and present focus of welfare research in pastoral dairy production systems and uses these insights to suggest future research priorities. We included original research articles published between 2000 and 2024 that studied dairy cows, weaned youngstock or preweaning calves in pastoral dairy systems and reporting on animal welfare outcomes. Review articles were excluded. Three thousand peer-reviewed manuscripts written in English were screened for inclusion. The 678 eligible articles extracted were categorized by welfare domain using the Five Domains framework (i.e., nutrition, health, environment, behavior, and mental state) and inductively assigned up to 5 descriptive key words to reflect the main welfare issue(s) addressed. Welfare research output increased from 19 to 60 publications per year over the past (2000-2019) to present periods (2020-2024), respectively. Most studies (74.0%) focused on cows, with fewer publications on the calf and weaned youngstock. Across all cattle classes and years, health was the most represented welfare domain, addressed by 69% of calf, 54% of youngstock, and 53% of cow studies, whereas the mental state domain was addressed in less than 1% of research. The article key words were used to assess changes in welfare focus over time. From 2000 to 2019, 34 calf welfare topics were identified, with research primarily focusing on colostrum management and passive immunity (21%). Presently (2020-2024), of the 42 topics published on the calf, extended suckling systems (19%) dominate (i.e., systems providing extended cow-calf contact). For weaned youngstock, 33 topics were published between 2000 and 2019, and parasitism was a key welfare theme. The present youngstock research continues to focus on parasite control (19% of studies covering 31 topics). Past research on cow welfare (94 topics) concentrated on pasture access and lameness. In recent years, 106 different topics have been covered by the cow literature. Of these, emerging technologies, such as virtual fencing and automated health monitoring, represent the most studied topics (14%), closely followed by pasture access (13%). Future research should increase focus on calves and youngstock, particularly in relation to environment, nutrition, behavior, and their effects on mental state. Extended suckling systems warrant continued attention. Longitudinal studies exploring how early-life environments shape heifer resilience and adaptability are encouraged. Addressing climate-related challenges while preserving the welfare benefits of pasture access is critical. Progress will require ongoing collaboration between science and industry, balancing immediate needs with investment in longer-term, transformative welfare solutions.

Understanding heat stress in dairy cows is important for ensuring their welfare and productivity. While the Temperature Humidity Index is commonly used to determine heat stress conditions it does not consider outdoor factors (e.g. wind speed and solar radiation). To address this, the Grazing Heat Load Index (GHLI) which predicts respiration rate, an indicator of heat stress, was previously developed using data from the Waikato region in New Zealand. This study updated the GHLI using a weather data and animal heat stress indicators (e.g. respiration rate and drooling), collected from dairy herds in the Northland, Waikato, Canterbury, and Southland regions. A mixed‐effects linear model, which included air temperature, solar radiation and wind speed as inputs was developed using data from these diverse locations. The updated model slightly outperformed the previous model on a validation dataset when predicting respiration rate. While air temperature remained the primary driver of respiration rate, solar radiation was an exacerbating factor, whereas wind exerted a mitigating effect. Challenges persist in characterising heat stress in conditions such as high wind speed. Non‐linear modelling techniques could enhance the predictive capability of the GHLI model, thereby refining heat stress assessment in pasture‐based dairy systems nationwide.