22 



FISH AND WILDLIFE TECHNICAL REPORT 30 



Location Error (km) 



Fig. 18. Frequency of location errors ( 1 00-m intervals) for PTT's 

 deployed on captive animals. A. All locations. B. Locations 

 with quality index 3 only. 



served while the uplink receiver simultaneously 

 recorded the sensor data (Pank et al. 1987). The collar of 

 each animal was adjusted to uniform tightness. Data were 

 classified into 31 intervals of 2 s each. 



In all cases, data from behaviors unique to the captive 

 situation were omitted (e.g., pacing back and forth along 

 an enclosure fence, feeding from a trough). Recorded 



Table 13. Overall errors (in meters) from the combined 

 data set of Telonics platform transmitter terminals 

 (PTT's) on ungulates at known locations (n 330). 



Location quality index 



activities occurred spontaneously, except for continuous 

 walking and running, which were usually caused by the 

 observer. In cases where animals had to be immobilized, 

 we waited until there were no discernible effects from 

 drugs before recording activities. 



Experiments on caribou were conducted on tame adult 

 females housed in large enclosures at the University of 

 Alaska, Fairbanks. In all cases the mercury switch was 

 inclined at an angle of +2 relative to the circuit board 

 (Pank et al. 1985). Data were not collected during winter, 

 so the activity of caribou digging feeding craters in snow 

 was not observed. 



Mule deer and elk experiments were conducted at 

 Washington State University, Pullman. A captive adult 

 female deer weighing approximately 70 kg was observed 

 within a large, fenced enclosure (roughly 5 ha) shared by 

 eight additional mule deer and eight mountain goats 

 (Oreamnos americanus). The elk was a 6-year-old female 

 (about 250 kg) kept in a 0.25-ha enclosure. To further 

 calibrate the short-term activity index for elk, visual ob- 

 servations of an elk wearing a PTT in Yellowstone Nation- 

 al Park were compared to sensor count output (D. Vales, 

 University of Idaho); the mature bull spent most of its time 

 near developed areas, allowing visual sampling. Its neck 

 had shrunk following the rut, so the collar had loosened. 

 Sampling occurred between 16 January and 16 February. 

 Activity categories defined included bedded, standing (in- 

 cluding grooming), getting up and laying down within the 

 same minute, moving (walking and running), feeding (in- 

 cluding movement while feeding), and sparring. Sample 

 sizes for these categories were 319, 76, 1 1, 24, 312, and 

 25, respectively. 



Experiments on moose were conducted on a tame adult 

 female (about 500 kg) in a 10-ha fenced area at the Moose 

 Research Center operated by the ADFG at the Kenai Na- 

 tional Wildlife Refuge. Three adult females and one adult 

 male also shared the pen with the tame female at the time 

 of the experiments. Activity sensor readings were re- 

 corded during all activities of the moose, although she 

 occasionally disappeared from view for a few minutes. 

 Natural feeding activities did not occur but were simulated 

 by placing cut branches of willow (Salix spp.), aspen 

 (Populus tremuloides), and paper birch (Betula pa- 

 pyrifera) along a fenceline at heights typical for those 

 species. 



Results. There were 1 , 165 sampling periods of 60 s each 

 obtained on the captive caribou. Activity categories were 

 inactive (bedded with head up or down; sleeping, resting, 

 or ruminating), feeding (browsing, eating grain placed on 

 the ground, ruminating while standing, walking among 

 feeding sites), walking (continuously during 60-s period), 

 and running (continuously during 60-s period). Feeding 



