Challenges of Animal Activity and Behavior Recognition Using Tri-Axial Accelerometers

Thursday, August 21, 2014: 11:10 AM
301B (Centre des congrès de Québec // Québec City Convention Centre)
Franziska Broell , Oceanography, Dalhousie University, Halifax, NS, Canada
Andre Bezanson , Biomedical Engineering, Dalhousie University
Christopher Taggart , Oceanography, Dalhousie University
Andrew Taylor , Biology, Mount Allison University
Matt Litvak , Biology, Mount Allison University
Micro-accelerometer tags are a novel technology used to remotely monitor (aquatic) animals in the wild providing quantitative information on behaviour and locomotion. One of the challenges with this technology is the development of automated methods that relate the accelerometer signal pattern to a range of activities, behaviours, energy expenditure, and environmental parameters (temperature, depth, light, etc.). We developed a novel high-frequency (550 Hz) micro-accelerometer tag to demonstrate how simple analytical techniques can be used to extract information on behavior (e.g. feeding) and activity (swimming, migrating) in the time and frequency domain (spectral analysis, filtering). We use data from deployments of these accelerometers in conjunction with Pop-up Satellite tags (pressure, temperature and light sensor) on Shortnose Sturgeon (Acipenser brevirostrum) in the Kennebecasis River, Canada to demonstrate how activity estimates from accelerometer technology can be combined with environmental variables. This provides insights into the tagged animal’s behavioural pattern in response to its environment (temperature, depth, light). For example, data show that Shortnose Sturgeon change their behavior in response to light level or tidal velocity. In the future, such simple algorithms have the potential of being incorporated into a microprocessor environment (on-board processing) delivering in situ information on animal behavior and movement.