Increasing agency in grazing dairy cows: Voluntary use of a treed area

Pasture-based dairy systems are valued by consumers for allowing the expression of natural behaviors; however, evolving stakeholder expectations include the ability for cows to exercise agency and choice. This study examined whether dairy cows would choose to voluntarily exit their grazing paddock and walk to an independent treed area, and quantified visit patterns and their relationship to weather, paddock distance, behavior, and rumen temperature. Sixty mixed-aged Friesian-cross cows were randomly allocated to Treed (n = 2) and Reference (n = 2) herds from November 2024 to March 2025. On-animal measurement devices included collars (latitude, longitude, grazing, ruminating), leg-worn accelerometers (lying/standing), and rumen boluses (rumen temperature, drinking). Weather data included temperature, relative humidity, wind speed, rainfall, solar radiation, and a calculated grazing heat load index (GHLI). After familiarization to the treed areas, cows were monitored over 4 observation periods (the first was 23 d; remaining were 15 d) that varied in milking frequency (once/twice daily), supplementation (present/not present), and herd size (separate, 15 cows/herd; or combined, 30 cows/herd). Latitude-longitude data were used to define treed area visits (threshold of ≥ 80% of the herd present at any time > 10 min). Bayesian logistic regression models (Treed area herds only) examined whether weather variables (temperature, GHLI, solar radiation, humidity; each categorized at the 75th percentile into levels: 'below-threshold', lower 75%, and 'above-threshold', upper 25%) and distance predicted treed area visit probability. Bayesian generalized additive mixed models (GAMMs; all herds) examined hourly behaviors and rumen temperature, with fixed effects for 'visit status' (Treed area herd - No Visit, Treed area herd - Visit, Reference herd), GHLI level, and their interaction. Splines captured temporal patterns. Daily behavioral and rumen temperature summaries were derived from hourly predictions. Results were reported as odds ratios (OR), and median with 95% credible intervals (CrI). Treed area herds visited treed areas 3-22 times per observation period (1.1-2.1 visits on visit days), spending 0.9-5.5 h/d there. Use peaked in the afternoons, varied by herd and observation period, and occurred in both above- and below-threshold thermal conditions. Comparing visit and non-visit days within the Treed area herds showed that paddocks ≥ 250m from the treed area reduced visit odds by 56% compared with Paddocks < 250 m from the treed area (OR = 0.44, 95% CrI: 0.27-0.69). Further, above-threshold temperature (≥22°C) doubled visit odds compared with below-threshold (OR = 2.03, 95% CrI: 1.13-3.44). Comparison of Treed and Reference herds showed that during treed area visits when the GHLI was above-threshold (≥54) daily ruminating-while-standing doubled (4.7 vs 2.8 h/d, respectively), while hourly grazing (OR = 0.77, 95% CrI: 0.72-0.81), drinking (OR = 0.29, 95% CrI: 0.21-0.39), and rumen temperature (-0.28°C, 95% CrI: -0.30 to -0.27) decreased relative to Reference herds. However, daily time budgets for lying, ruminating, and grazing remained similar across herds. While lower in magnitude, rumen temperature and hourly behavioral changes were also noted during non-visit hours, suggesting carry-over benefits beyond discrete visits. Despite minimal effects on daily behavioral time budgets, treed area access reduced rumen temperature, and provided opportunities for choice expression. This proof-of-concept study demonstrates that voluntary access to treed areas is feasible which could be implemented with remanent tree stands; such access is valued by cows for both thermal relief and non-thermal benefits, offering potential for enhancing agency in pastoral dairy systems.