Importance of evaluating GPS telemetry collar performance in monitoring reintroduced populations

Abstract

ABSTRACTGlobal Positioning System (GPS) tracking collar technology has been widely employed in wildlife research and is valuable for understanding movement patterns of reintroduced populations, yet the associated fix‐rate bias and positional errors are a serious concern. During 2013, we deployed GPS collars at fixed sites varying by habitat and terrain conditions in the mountains of southwestern Virginia, USA, where elk (Cervus elaphus) were recently reintroduced, to test 2 hypotheses: 1) collar precision would decrease and fix rate bias would increase with increasing terrain and vegetation obstructions; and 2) variability in fixes and associated errors can affect evaluations of habitat use by reintroduced elk populations. Our model predicted a 1.67‐m decrease in root mean square error (RMSE) for every 100‐m increase in elevation, and a 1.14‐m increase in RMSE for every 10° increase in slope. We found that collars under deciduous, mixed, and coniferous cover were predicted to decrease the fix rate by 14.5%, 25.6%, and 47.4%, respectively. We computed an uncertainty index based upon Monte Carlo simulations to show the effects of variability in fixes on habitat assignment of elk locations. Based upon the uncertainty index, we predicted that the relatively small and patchily distributed herbaceous cover type had the most potential for misassignment of use. Failure to account for these sources of error could improperly indicate that elk almost exclusively use open ridge tops and erroneously affect manager's decisions regarding potential human–elk conflict and disease transmission. Our results illustrate a potential danger in interpreting raw point data from a GPS tracking study of reintroduced populations. Errors caused by local habitat conditions in patchy and structurally complex landscapes should be accounted for in future models and management decisions. © 2017 The Wildlife Society.