Predicting mountain lion activity using radiocollars equipped with mercurytip-sensors

Radiotelemetry collars with tip-sensors have long been used to monitor wild life activity. However, comparatively few researchers have tested the relia bility of the technique on the species being studied. To evaluate the effic acy of using tip-sensors to assess mountain lion (Puma concolor) activity w e radiocollared 2 hand-reared mountain lions and simultaneously recorded th eir behavior and the associated telemetry signal characteristics. We noted both the number of pulse-rate changes and the percentage of time the transm itter emitted a fast pulse rate (i.e., head up) within sampling intervals r anging from 1-5 minutes. Based on 27 hours of observations, we were able to correctly distinguish between active and inactive behaviors >93% of the ti me using a logistic regression model. We present several models to predict activity of mountain lions; the selection of which to use would depend on s tudy objectives and logistics. Our results indicate that field protocols th at use only pulse-rate changes to indicate activity can lead to significant classification errors.