Analyzing geolocator data for birds that roost in cavities year‐round. Issue 1 (6th January 2016)

• Record Type: Journal Article
• Authors: Gow, Elizabeth A.
• Is Part Of: Journal of field ornithology. Volume 87:Issue 1(2016)
• Page Start: 74
• Page End: 83
• Publication Date: 2016-01-06
• Subjects: animal movement -- calibration methods -- geolocator error -- Kalman filter models -- sensitivity analysis -- tracking; Birds -- Periodicals; Ornithology -- Periodicals; Ornithologie -- Périodiques; Oiseaux -- Baguage -- Périodiques; Oiseaux -- Périodiques; 598
• DOI: 10.1111/jofo.12130
• Languages: English
• ISSNs: 0273-8570

Abstract

ABSTRACT: The use of light‐level geolocators for monitoring migration has been limited to non‐cavity roosting species because light transitions for cavity‐roosting species are obscured. Using Northern Flickers ( Colaptes auratus ), nocturnal cavity‐roosting woodpeckers, as a model, I describe a method for analyzing geolocator data that initially adjusts light transitions to account for differences between the time of minimum light threshold and when a bird enters or exits a cavity. Using known locations from the breeding grounds, I assessed the precision of this adjustment method for estimating location by examining the associated error, the repeatability of the length of time individuals roosted in cavities, and by conducting a sensitivity analysis to assess uncertainty. Mean location error decreased from 1417 ± 277 km (SD) to 129 ± 194 km when sunrise and sunset times were adjusted and locations from >25 d were averaged. Sensitivity analysis showed that if an adjusted sunrise or sunset time was "incorrect" by 10 min, the error was 121–137 km from the actual location. This adjustment method significantly improved location estimates at known sites, suggesting that adjusting light transitions based off a calibration is a good initial step for determining location. However, to account for behavioral changes in entrance and emergence times, applying state‐space Kalman filter models can further improve the accuracy of location estimates. The combination of adjusting transitions and applying a state‐space Kalman filter thus allows location estimates to be obtained from cavity‐roosting species using geolocator data.