We have developed a new method and workflow for localizing radio tags in modern automated radio tracking systems. Radio tracking is one of the oldest methods for tracking animals in the wild. While this method originally involved manually tracking tagged individuals using a small number of mobile receivers, modern radio tracking methods typically employ many automated receivers that continuously detect radio tags. Typically, these receivers detect the signal strength of a tag, which can be used to estimate the distance from the receiver to the tag. Conventional analytical methods to locate the position of the tagged animal, however, are not optimized to make full use of the information provided by modern automated radio tracking systems in which many receivers may simultaneously detect a tag.

Radio fingerprinting, a method commonly used for indoor positioning applications, is an alternative localization method that can use information from the large number of receivers within automated radio tracking systems. We conducted a large-scale field test of radio fingerprinting as a localization method for wildlife tracking and compared the results to the traditional localization method, multilateration. We show that while both methods yield similar localization accuracy under ideal circumstances, radio fingerprinting is more robust to a variety of situations that commonly occur in real-world tracking studies, such as missed detections and equipment failure. We have created a template analytical workflow that is available for others interested in applying this localization method.

In some cases, however, radio fingerprinting as localization method may be impractical to implement. In this case, multilateration can still provide valuable information, but ideally this method should account for the large location estimate errors that can occur due to the uncertainty in the relationship between signal strength and distance. Using a process of ‘repeated multilateration’ in which the relationship between signal strength and distance is estimated with uncertainty, we have developed a method to derive error ellipses around estimated locations. We have used this method to document space use and social interactions in free-living zebra finches.