Northern Prairie Wildlife Research Center
|Fig. 2. Movement data (327 locations) collected by GPS collar on a breeding female wolf from February 20 to March 8, 1997 at Camp Ripley, Minnesota. GPS location-attempt rate was set at 1/hour, and success rate was 76% during this period. Dots represent actual locations; lines merely connect consecutive points.|
Approximations of GPS collar life varied from 12.5-50.0 days, while actual GPS life lasted 11-38 days (Table 1) including 2 collars tested in Denali that recorded reliably for 16-17 days and were then removed for convenience while still recording data. Actual longevity of these 2 collars went undetermined.
For successful locations, the mean time to determine the locations was 63 seconds in the Camp Ripley area, 83 seconds for the 2 deer in the Superior National Forest before leafout, and 91 seconds for the 2 wolves during summer, and 57 sec in Denali. Positional Dilution of Position values can range from 0.00 to 9.00, with lower values theoretically representing more accurate estimates of location; in this study mean PDOP for all locations was 3.75 (range = 1.10 - 9.00). However, although PDOP is theoretically a good index of locational accuracy, the GPS collar manufacturer has verified that selective availability error is usually large enough to nullify the effectiveness of PDOP as an accuracy index. Nevertheless, PDOP is probably still a good index for differentially corrected data.
The collars we tested were prototypes, and we uncovered various flaws during their application. Their VHF signal strength was variable, with strength of some signals being about 1/3 to 1/2 that of conventional collars in the forested areas, although not in Denali. At Camp Ripley, 2 of 4 collars stopped emitting a VHF signal about a week after they were deployed and stopped taking fixes about 900 location attempts sooner than the batteries would have allowed. This problem was later attributed to moisture on the microprocessor boards, because of faulty sealing of the housing. Condensation also caused the "squib" used to release the collars (Mech and Gese 1992) to fail on 2 Camp Ripley collars and 1 Denali collar. We did not find evidence that these collars failed from behavior of the wolves (e.g., chewing or swimming). We retrieved these collars by recapturing the wolves through helicopter darting and foothold trapping.
Errors in collar software produced unrecognizable beep patterns in 2 collars. This problem made it difficult to determine whether location attempts had been successful until the collars were retrieved.
As noted above, on 2 collars the GPS antenna leads were too short, so once the collars were placed on deer, the leads disconnected, resulting in zero locations. The third collar was placed on a larger deer; the antenna lead held, but the GPS antenna pointed off to the side. Thus the battery expended too much power per attempt and died prematurely, although the collar still produced 677 locations.