Tracking butterfly flight in the field from an unmanned aerial vehicle (UAV): a methodological proof of principle

Tracking and understanding the movements of animals in the wild is a fast-growing area of research, known as movement ecology . However, tracking small animals such as flying insects, which cannot easily carry an electronic tag, remains challenging as existing field methods are costly either in terms of equipment or tracking effort (e.g. VHF radio-tracking, scanning harmonic radar). Here we attempted to record the movements of free-flying butterflies from an unmanned aerial vehicle (UAV), maintaining a static position in the sky and recording video vertically downwards. With an appropriate flight height and image filtering algorithm, we recorded 166 flight tracks of Pieris butterflies ( P. brassicae and P. rapae ), with a median tracking length of 40 m (median flight duration 13 s), and a high temporal resolution of 30 positions per second. Average flight direction varied significantly over the course of the flying season, from a northward azimuth in June and early July, to a southward azimuth in September, congruent with a trans-generational migratory behaviour that has previously been documented by field observations or experiments in flight cages. In addition, UAV imagery unlocks the possibility to measure high-resolution flight movement patterns (e.g. path tortuosity and transverse oscillations), which will possibly help understand perceptual and locomotor mechanisms underlying spatial behaviour. We explore the technical details associated with UAV tracking methodology, and discuss its limitations, in particular the biases associated with a 2D projection of 3D flight movements, the limited spatial scale, and the difficulty to distinguish between visually similar species, such as P. brassicae and P. rapae .