Abstract
Dairy cows typically respond to stressors by altering their behavior, such as reducing eating time (ET) and rumination time (RT). Although declines in milk yield (MY) have been extensively studied, models to quantify perturbations in ET and RT are still lacking. This study adopts a smoothing approach to identify and characterize perturbations in MY, ET, and RT in response to the main primary stressors, heat stress (HS), lameness (L), and mastitis (M), while evaluating the influences of parity and stage of lactation. A total of 350 Italian Simmental cows were monitored in farms equipped with automatic milking systems and accelerometers. Within this population, cows with a lactation period of at least 150 days were selected. A double-curve smoothing model (λ = 100 and λ = 10,000) was applied to calculate response and recovery times and to quantify production and feeding behavior losses. The results indicate that L causes the longest (30.6 d and 28.8 d, respectively) perturbations for both MY and ET. While L caused the greatest loss in milk production (14.7 kg), HS resulted in the greatest losses regarding feeding behavior (ET: 175.2 min and RT: 210.3 min). In general, M had a lower impact, likely due to the timeliness of treatments. Primiparous cows showed faster responses to stress but slower recovery times compared to multiparous ones. However, multiparous cows exhibited greater total MY losses. The method proved effective for quantifying resilience and opens new perspectives in health monitoring, allowing for the identification of both economic loss and each animal's capacity to cope with pathological and environmental events, improving the overall sustainability of the dairy farm.