Calibration of remotely collected acceleration data with behavioral observations of roe deer (Capreolus capreolus L.)

Long-term, remote monitoring of animals under natural conditions is essential for obtaining information on animal activity. Advances in biotelemetry have led to the construction of low-power accelerometers placed on Global Positioning System (GPS) collars. Such acceleration data from roe deer have not yet been classified to the various behavioral categories. Here, we determined the threshold values of such data for different behavioral categories. We equipped captive roe deer with Global Positioning System-Global System for Mobile Communications/dual-axis acceleration sensor neck collars and simultaneously measured their movement and observed their behavior. The difference between feeding and slow locomotion was significant on the x- but not the y-axis, and both of these two behavioral categories differed significantly from resting and fast locomotion. Specific thresholds for the behavioral categories-resting, feeding, and slow and fast locomotion-were established by recursive partitioning. We compared the behavior determined by these threshold values with observed behavior and found that 92% of the behavioral categories were correctly assigned. A comparison of our results with those of earlier studies showed that thresholds derived for one species cannot be directly applied to another species. We provide recommendations for the further development of acceleration sensors based on the results obtained in this study.