Context. GPS tags have revolutionised the field of wildlife spatial ecology by providing a large number of animal location datapoints at a very fine spatial scale. Although GPS tags have been used on large animals for several decades, it is only recently that technological advances have allowed lightweight GPS tags to be deployed on small animals with limited travel distances. Importantly, factors such as canopy cover, topography, and tag orientation can affect the ability of GPS tags to obtain satellite fixes, which can subsequently affect the accuracy of recorded locations.

Aims. If not corrected for, biases in location data obtained from GPS tags could lead to erroneous inferences regarding animals’ habitat use, home-range sizes, and movement paths, which could reduce the effectiveness of conservation efforts based on such inferences.

Methods. Here, we used a double-sampling method (i.e. GPS tags and ground-truthing with radio-telemetry) and quantified the effects of habitat characteristics and data screening on the accuracy of location data obtained from GPS tags deployed both in stationary tests and on wild eastern box turtles and spotted turtles, which co-occur at our study sites but use different habitat types.

Key results. We found that canopy cover reduced both the number and accuracy of locations obtained from GPS tags, that dense ground vegetation decreased the fix success rate of GPS tags, and that GPS tags were ineffective when submerged underwater. We further showed that using a simple method to screen data and exclude low-accuracy locations is essential if locations obtained from GPS tags are used to make inferences about a species’ habitat use or spatial ecology.

Conclusions. Screening data to reduce location error is particularly important for animals with small home-range sizes and short travel distances because a small number of erroneous locations can introduce substantial bias in inferences regarding a species’ space use.

Implications. We encourage researchers to report measures of error (i.e. location error, horizontal dilution of precision, number of satellites received) for GPS location data and to employ data-screening methods to exclude low-accuracy locations and improve the reliability of published animal location data, and the inference drawn there from.