Fish locomotor dynamics quantified using multiple inertial measurement units

Animals exhibit a diversity of locomotor behaviours that vary with body shape, ontogeny, motivation and environment. Documenting these behaviours often relies on recording videos of animals in controlled conditions displaying stereotyped behaviours. By contrast, animal-borne motion sensing data loggers can document diverse behaviours of animals in their natural environment. Traditionally, one data logger is used to infer body dynamics; however, a single location presents an incomplete picture of body movement. Here, we deploy multiple independent inertial measurement units on individual fish to study within-body dynamics and assess the accuracy of this approach compared to traditional video methods. Using a combination of robotic and animal experiments, we accurately measure the timing, direction and magnitude of movement of multiple locations along the body and can recreate accurate three-dimensional midline kinematics without video data. Furthermore, we show that with as few as two data loggers, parameters such as propulsive wavelength can be calculated, even during high-intensity behaviours. Using multiple data loggers has the potential to provide unbiased spatio-temporal estimates of fish kinematics, allowing kinematic studies to be extended to new species, new habitats and a diversity of natural fish locomotor behaviours beyond those traditionally studied in laboratory settings.