Performance of a GPS Animal Location System under Boreal Forest Canopy

Abstract

An automated animal location system, based on Global Positioning System (GPS) technology, is being used for wildlife research. The GPS is a divergent technology, and positional accuracies vary between millimeters and tens of meters, depending on the system used and operating conditions. Before GPS-based tracking data can be used for habitat analyses, the influence of habitat on GPS-collar performance must be evaluated under various canopy conditions, including the optimal condition of no canopy. We evaluated performance of nondifferentially corrected GPS collars in an experimental forest with mature, evenly spaced trees and on wild free-ranging moose (Alces alces) to determine the influence of canopy on positional accuracy and observation rate. In an experimental forest with mature, evenly spaced trees (henceforth called spacing trial), canopy characteristics of tree species, spacing, height, basal diameter, and canopy closure had no influence on positional accuracy (P > 0.05), but had an influence on GPS observation rate (P < 0.001). Location error was greater if positions were based on 2-dimensional rather than 3-dimensional mode of operation (P < 0.001), with location errors of 65.5 and 45.5 m, respectively. Location error in 3-dimensional mode did not differ from the expected error of 40 m (P = 0.43). As tree density increased, observation rate decreased and the probability of the GPS receiver operating in 2-dimensional mode increased (P < 0.001), resulting in increased location error. With future development of differentially corrected GPS collars, location errors of <10 m are expected.