Northern Prairie Wildlife Research Center
Stewart W. Breck, Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, CO 80523 USA, and Dean E. Biggins, U.S. Geological Survey, Biological Resources Division, Midcontinent Ecological Science Center, 4512 McMurray Avenue, Fort Collins, CO 80525 USA
The accuracy of a radio-telemetry data set is reliant upon a variety of factors including skill of the technicians, accuracy of the telemetry system, vagaries of radio signal transmission, and errors due to recording and transcribing data. Studies have measured and analyzed the effects of signal error, bias and precision of telemetry systems, and observer performance, with each describing specific methods to improve a telemetry data set. One aspect of radio-telemetry data that has received less attention is the screening of telemetry data for errors caused from recording, transcription, and poor fixes. Some papers state that error checks were made and a few papers mention methods used to identify outliers, but often the methodology is not detailed. We outline a strategy used to systematically and unbiasedly screen radio-telemetry data for spatial and temporal errors generated during data collection and computer entry, and present examples of how this screening process affected the results.
Three detection criteria (speed, outlier, and error quadrangle) focused on identifying erroneous locational fixes and one detection criterion (time) focused on finding temporal errors. Computer programs were written to detect potential errors, and the data were checked manually to confirm and eliminate errors. Variances of screened data sets tended to be smaller than variances of unscreened data sets; in some comparisons involving black-footed ferret (Mustela nigripes) data, this affected the statistical interpretation during hypothesis testing and model selection. The process outlined is applicable to data from ground and satellite-based tracking systems, but some screening criteria must be objectively determined by a priori knowledge of the species under investigation.