Comparative Analysis of Climatic Indices Using Generalized Additive Models: Assessing the Impact of the Most Suitable Indices on Circadian Behaviors of Grazing Cattle in Summer Conditions

Objectives
The primary aim of this study was to evaluate several climatic indices using Generalized Additive Models, identify the most suitable predictor for cattle behavior, and determine critical temperature thresholds impacting grazing, rumination, standing, and inactivity, with an hourly resolution.
Material and Methods
Behavioral data were collected from 69 dairy cattle over two summer seasons (2021–2022) using accelerometers that recorded activities at 5-minute intervals. The cows grazed rotationally across eight plots bordered by hedges providing shade and had unrestricted access to these plots outside of milking and feeding times. Milking occurred twice daily at 6:30 AM and 4:30 PM, with a prepared ration provided in the barn following the morning milking. Hourly climatic data—including temperature, humidity, solar radiation, and wind speed—were recorded at a weather station located 6 km from the farm. To identify the most suitable predictor of cattle behavior, several climatic indices—such as black globe temperature, Temperature Humidity Index, Heat Load Index, Comprehensive Climate Index, Wet Bulb Globe Temperature, Equivalent Temperature Index of Cattle, adjusted Temperature Humidity Index, and Respiration Rate Index—were evaluated using the Akaike Information Criterion.
Generalized Additive Models were used to analyze non-linear relationships and interactions between the time of day, climatic indices, and cattle behaviors. The incorporation of splines in the models, along with their ability to account for individual variability, enabled robust modeling of circadian patterns. From an initial dataset of 2,463,528 behavioral observations, harmonization with hourly meteorological data resulted in a final dataset of 205,294 observations.
Results
The black globe temperature emerged as the most suitable index, achieving the lowest Akaike Information Criterion values across all behavioral categories. This index outperformed alternatives and it appears that Wet Bulb Globe Temperature and Comprehensive Climate Index were respectively the second and third indices. It’s highlighting the superior ability of Black Globe Temperature to predict behavioral variations.
The hourly analysis revealed distinct circadian patterns aligned with milking and morning feeding routines. During these management periods, behavioral variability among cattle was low, indicating a collective synchronization driven by human interventions. However, between these periods, the behavioral cycles exhibited higher variability, reflecting individual differences in responses to environmental conditions and physiological needs or conditions.
Several critical temperature, as inferred from the Black Globe Temperature, thresholds were identified, marking notable fluctuations in animal behaviors. These behaviors transitioned from a positive temperature-related impact to a negative impact. Between these thresholds, behavioral variability remained moderate, with grazing and rumination displaying more consistent patterns. In contrast, standing and inactivity exhibited greater inter-individual variability, highlighting a stronger influence of individual differences in these behaviors.
Conclusions
This study identifies black globe temperature as a reliable predictor of pasture cattle behavior under diverse climatic conditions. By utilizing generalized additive models, the research offers detailed insights into circadian rhythms and temperature-induced behavioral adaptations. These findings inform practical strategies, such as optimizing grazing schedules and enhancing shaded area availability, to alleviate heat stress and promote cattle welfare in extensive farming systems.