Gone with the wind: Inferring bird migration with light‐level geolocation, wind and activity measurements. Issue 6 (16th March 2022)

• Record Type: Journal Article
• Authors: Werfeli, Mike; Ranacher, Peter; Liechti, Felix
• Is Part Of: Methods in ecology and evolution. Volume 13:Issue 6(2022)
• Page Start: 1265
• Page End: 1274
• Publication Date: 2022-03-16
• Subjects: bird migration -- migration route inference -- Bayesian inference -- light‐level geolocation -- activity measurements -- wind measurements -- air pressure measurements -- GIS; Ecology -- Periodicals; Evolution -- Periodicals; 577
• DOI: 10.1111/2041-210X.13837
• Languages: English
• ISSNs: 2041-210X

Abstract

Abstract: To investigate the complex phenomenon of bird migration, researchers rely on sophisticated methods for tracking long‐distant migrants. While large birds can be equipped with satellite tags, these are too heavy for many species. Instead, researchers often use light‐level geolocation for tracking individual small migratory birds. Unfortunately, light‐level geolocation is often coarse and unreliable, with positioning errors of anything up to hundreds of kilometres. Recent Bayesian models try to constrain the route to plausible corridors: they couple light‐level measurements with information about the bird's likely movement. While these models improve inference, they still lack information on weather conditions, specifically the impact of wind. For example, birds might encounter tailwinds—considerably increasing their (ground) speed and making longer routes more likely, or headwinds—having the opposite effect. Miniaturised multi‐sensor tags allow monitoring not only light but also acceleration and air pressure. These measurements provide essential additional information about the exact timing of flight activity and the corresponding flight altitudes. This article proposes a Bayesian model for inferring bird migration. The model integrates air pressure to estimate flight altitudes and considers wind data to calculate the most likely flight trajectory. The model constrains the migratory routes to those likely given by the winds en route and the observed timing of flight activity. We apply the model to infer the migration of European Hoopoes Upupa epops . Adding wind data for route inference excludes flight trajectories with unrealistic high airspeeds, decreases the uncertainty of the position estimates and returns more plausible migratory routes. Faithful reconstruction of migratory routes helps unravel the influence of physiological and environmental factors on bird migration. This is crucial for habitat protection where limited resources need to be allocated to relevant areas.