• Version
• Download 8
• File Size 457.34 KB
• Create Date 1 January 2001
• Download

Incorporation of activity sensors in radiotags provides a mechanism for acquiring activity data on free-ranging bears (Ursus spp.). Tests of sensor sensitivity and correlation of sensor data with known levels of activity are rarely done. We used a surrogate test animal and generated data at 3 levels of activity (rest, walk, and run) from radiocollars with activity sensors constructed by 2 manufacturers (Types A and B). For the test, activity-sensor data differed between types, and variance among collars from the same manufacturer was substantial. Signals from Type A collars differed when the animal was resting and walking, and resting and running, but the signals were indistinguishable when the animal was walking or running. Type B collars were less sensitive to movement and gave inconsistent predictions of energy expended: a resting animal produced a measure of activity statistically > 0, a walking animal's signal was not different from zero, but a running animal's activity was >> 0. Because of the large variance both between and within collar types, we suggest that individual collars be calibrated to known levels of activity prior to attaching collars to bears. We collected 24-hr activity data on female black bears (Ursus americanus) in south central Louisiana during 2 winters and 2 falls. We found strong evidence that activity differed among 5 reproductive classes and seasons. However, high variability in activity among individual bears requires that large sample sizes be obtained to accurately depict circadian activity patterns. Because of high variability among activity sensors, among bears, and for activity bout duration, measuring activity levels requires greater care than distinguishing among patterns. We compared our data with subsamples of our data modified to mimic discrete tip and reset switch data. Some analyses of measurements based on the simulated data agreed with results from our original analysis, but discrete data, especially reset switch data, overestimated activity level and poorly paralleled diel pattern.