Separating biological signal from methodological noise in home range estimates

Abstract Space use is commonly estimated in animal ecology and has become a cornerstone of evidence‐based conservation planning, with animal tracking increasingly used to underpin the designation of protected areas with high conservation value. However, tracking technologies and analytical methods may introduce biases in home range size estimates. We assessed these potential biases using simulated tracking data and published home range size estimates from empirical animal tracking studies. We first simulated animal movement data and added published location error estimates for different technologies used for tracking sea turtles. Location data were analysed using common space use estimation methods (minimum convex polygon, fixed and autocorrelated kernel density estimation, biased random bridge and dynamic Brownian bridge movement model). Second, we reviewed home range size estimates obtained using different technologies to track hawksbill ( Eretmochelys imbricata ) and green ( Chelonia mydas ) turtles to assess the relative impacts on home range estimates due to (i) tracking accuracy and (ii) analytical methods. For both simulated data and empirical values of space use from the literature ( n = 90 studies), relatively large home range estimates tended to be generated from lower resolution Argos tracking compared to higher‐resolution Fastloc‐GPS tracking. These findings reflect inaccuracies in location data providing spuriously large movements. For example, Argos and Fastloc‐GPS home range size estimates for adult green turtles averaged 393 and 53 km 2 respectively ( n = 64 and 39 individuals). For simulated data, biases introduced by tracking accuracy had a far greater impact on home range size estimation than the analytical method used, apart from when using autocorrelated kernel density estimation (AKDE) which compensated for positional error very well. Our results suggest that in many cases, hawksbill and green turtles have relatively small home ranges (<10 km 2 and in some cases, <1 km 2 ), with this picture of their limited space use only emerging through high‐accuracy tracking. These general conclusions likely apply broadly across taxa and will impact attempts to assess patterns of home range sizes recorded for individuals across studies in different regions.