Impact of Habitat-Specific GPS Positional Error on Detection of Movement Scales by First-Passage Time Analysis

Abstract

Advances in animal tracking technologies have reduced but not eliminated positional error. While aware of such inherent error, scientists often proceed with analyses that assume exact locations. The results of such analyses then represent one realization in a distribution of possible outcomes. Evaluating results within the context of that distribution can strengthen or weaken our confidence in conclusions drawn from the analysis in question. We evaluated the habitat-specific positional error of stationary GPS collars placed under a range of vegetation conditions that produced a gradient of canopy cover. We explored how variation of positional error in different vegetation cover types affects a researcher's ability to discern scales of movement in analyses of first-passage time for white-tailed deer (Odocoileus virginianus). We placed 11 GPS collars in 4 different vegetative canopy cover types classified as the proportion of cover above the collar (0–25%, 26–50%, 51–75%, and 76–100%). We simulated the effect of positional error on individual movement paths using cover-specific error distributions at each location. The different cover classes did not introduce any directional bias in positional observations (1 m≤mean≤6.51 m, 0.24≤p≤0.47), but the standard deviation of positional error of fixes increased significantly with increasing canopy cover class for the 0–25%, 26–50%, 51–75% classes (SD = 2.18 m, 3.07 m, and 4.61 m, respectively) and then leveled off in the 76–100% cover class (SD = 4.43 m). We then added cover-specific positional errors to individual deer movement paths and conducted first-passage time analyses on the noisy and original paths. First-passage time analyses were robust to habitat-specific error in a forest-agriculture landscape. For deer in a fragmented forest-agriculture environment, and species that move across similar geographic extents, we suggest that first-passage time analysis is robust with regard to positional errors.

A cost-effective and informative method of GPS tracking wildlife

“…A GPS antenna is most accurate and most efficient when it is pointed towards the sky with nothing obscuring its view (Belant 2009;Williams et al 2012), so we moved the position of the battery on the collar to counterweight the GPS antenna. We also increased the size of the battery, because larger batteries will allow the device to record for longer periods of time (data storage permitting), and larger batteries are also heavier, creating a better counterbalance.…”

Informing Disease Models with Temporal and Spatial Contact Structure among GPS-Collared Individuals in Wild Populations

“…GPS locations were stored on board the collars that were remotely detached from study animals and retrieved after approximately 1 yr ( = 271 days). Positional error associated with GPS locations was <10m in most cases ( = 5.3 m, SD = 5.3 m) [30] .…”

Fine‐scale spatial genetic structure of deer in a suburban landscape

“…We reclassified land cover at a 30 × 30‐m resolution into 6 cover types: urban developed areas of medium and high intensity (12%), urban areas of low intensity development (i.e., suburban residential; 16%), forest (25%), rangeland (40%), wetlands and open water (7%), and barren (<1%). We classified woody wetlands as forest because deer use lowland forests similarly to upland forest types (Hygnstrom and VerCauteren , Williams et al ). Rangeland cover included agricultural lands and developed open space (parkland, cemeteries, golf courses, yards) because deer use these cover types similarly in urban, suburban, and exurban environments (Grund et al , Storm et al , Potopov et al ).…”