Accuracy and performance of low-feature GPS collars deployed on bison (<i>Bison bison</i>) and caribou (<i>Rangifer tarandus</i>)

20 June 2018

Jung, Thomas; Hegel, Troy; Bentzen, Torsten; Egli, Katherina; Jessup, Lars; Kienzler, Martin; Kuba, Kazuhisa; Kukka, Piia; Russell, Kyle; Suitor, Michael; Tatsumi, Kenji

Recently, a new generation of global positioning system (GPS) collars has become commercially available that provides limited daily location fixes, a relatively long battery life, and are low-cost, compared to full-featured GPS collars. However, their performance is untested, and assessing these biases is needed to inform study designs and data analysis protocols. We used stationary tests of 15 Lotek LifeCycle™ GPS collars—an example of low-feature GPS collars—to measure the accuracy of location fixes. In addition, we deployed these GPS collars on 153 caribou (Rangifer tarandus) and 24 bison (Bison bison) in Yukon and Alaska, to assess their field performance. We examined differences among species, sex, location fix schedule, and latitude, on 4 performance metrics (FSR, 3D-V FSR, DOP, and DSR). Stationary trials indicated that mean precision (4.3 ± 4.0 m [SD]) and accuracy (6.0 ± 4.7 m) of location fixes was excellent, and FSR was good (87.2 %), albeit both were slightly affected by forest canopy cover. Field performance varied by species and sex. Notably, the mean DSR for male bison was dismal (27.4 ± 24.2%) likely because of their behaviour, and the mean FSR, 3DV-FSR, and DOP, for male caribou was poor (FSR = 57.3 ± 2.0%), compared to collars deployed on female caribou (72.2 ± 1.7%) or female bison (77.9 ± 1.4%). We also observed that the VHF transmitters often failed when the collar malfunctioned. Biases in the accuracy and performance of these low-cost GPS collars should be taken into account when designing studies. Researcher’s contemplating investing in low-feature GPS collars require information on their ‘real-world’ performance so that they can decide whether they are appropriate for their intended application. Moreover, researchers need to consider biases in their GPS collar data prior to embarking on field studies and when conducting analyses with the data collected from them.