Introducing a new tool to derive animal behaviour from GPS data without ancillary data: The red-footed falcon in Italy as a case study

Abstract

Animal monitoring is a key stage in the process of understanding individual behaviour and social dynamics that characterize populations and species. Recent technological advances allow to collect large data sets on animal movement. Using such data to compute activity patterns is very challenging without expensive field surveys, video analyses and/or accelerometer sensors.We introduce here a new tool (UABE: Unsupervised Animal Behaviour Examiner) that makes use of a set of methodologies (if-then-else rules, sensitivity analyses, Markov chains) to interpret GPS data in behavioural terms under worst case conditions, i.e. in absence of ancillary data. The UABE is halfway between unsupervised inductive (clustering or segmentation) and deductive (expert-based) methods. As a case study, we applied the UABE to the investigation of the red-footed falcons' activity at nest at the two largest breeding colonies in Italy during the chick-rearing period.By using accurate GPS data-loggers, we tracked four adults in June–July 2019 and 2020, and collected 5840 GPS points. Activity at nest occupied one fourth of the red-footed falcons' day, with substantial differences between months and between daytime and night-time. On an hourly basis, clear temporal patterns of activity at nest emerged. Through the combined use of frequency and continuity of the studied behaviour, the UABE also allowed to raise plausible hypotheses for the reasons behind birds' activity at nest at specific times of day.The UABE supports the creation of unsupervised ethograms of both stationary (e.g., standing at nest or roost, hovering/perching) and nonstationary (e.g., flight, walking) activity patterns. It can be applied to any species regardless of the tracking instrument used.