Abstract
Advancements in tracking technology allow researchers to understand the spatial ecology of many terrestrial and aquatic species. Argos Doppler is a widely used technology for wildlife telemetry as it suits smaller species and have longer life span than miniaturized GPS. In practice, large Argos location errors often occur due to communication conditions such as transmitters settings, local environment, area of reception, behaviour of tracked individual, etc. Considering the specificity of errors and the lack of benchmark studies in Eastern Europe, our research objectives are (1) to provide empirical evidences of the accuracy of Argos Doppler locations in Romania, (2) investigate the effectiveness of straightforward destructive filters for improving Argos data quality, and (3) to provide guidance for handling Argos wildlife monitoring data to researchers in Eastern Europe. We assessed the errors associated to Argos locations in 4 geographic locations from Romania in static, low speed and high-speed tests and then we evaluated the effectiveness of Douglas Argos distance angle filter algorithm to minimize location errors. Argos locations received in our tests had larger horizontal errors than those indicated by the operator of the Argos system, including when reception conditions are ideal. The errors are highly variable within each location class, however, positions from location class 0 were constantly prone to large errors. The errors were anisotropic, predominantly oriented East and West, a pattern confirmed by the larger longitudinal errors in the vast majority of data. Errors were mostly related to movement speed of Argos transmitter at the time of reception, but other factors such as topographic conditions and position of the device toward the sky at the time of the transmission contribute at receiving low quality data. Douglas-Argos filter successfully excluded largest errors while retained a large amount of data when the threshold was properly defined for local scale (2 km). Thus, filter selection requires previous knowledge about the movement patterns and behavior of the species of interest, and parametrisation of the selected filter must follow a trial and error approach.
