Foraging behaviour and activity of a marine benthivorous fish estimated using tri-axial accelerometer biologgers

Fine-scale behaviour such as foraging is difficult to quantify in free-swimming wild fish yet has important basic and applied implications. Here, we used tri-axial accelerometer biologgers to determine accelerometric predictors of bonefish Albula vulpes behaviours (resting, swimming, bursting, coasting, and foraging) in a wetland mesocosm in Eleuthera, The Bahamas. We also used a swim flume to estimate the relationship between acceleration and swimming speed for bonefish (n = 9). In the wetland study, 5 bonefish were tagged externally with accelero meter loggers and monitored for a 5 d period during which visual behavioural observations were conducted for 4 h. Classification tree models were used to identify accelerometric criteria for bonefish behaviours, and a classification algorithm was applied to estimate behavioural frequencies for bonefish in the wetland for the 5 d period. Bonefish spent the majority of time resting (57%), followed by swimming (26%) and coasting (17%), and foraged an average of 11 times h(-1). Bonefish exhibited primarily slow swimming speeds (average 0.18 m s(-1)) while in the wetland, with occasional burst swimming events (14 h(-1)) to maximum swimming speeds ranging from 4.3 to 6.4 m s(-1) across individuals. Swimming and foraging behaviours varied among individuals and over the diel cycle. Bonefish generally swam most at dawn and foraged most at night. Temperature and tide were also significant predictors of swimming behaviour, and fish were generally most active at lower temperatures (similar to 24 degrees C) and during ebbing tidal periods despite the fact that the wetland was disconnected from the ocean. The ability to estimate fine-scale behaviours such as foraging and activity levels has important implications for understanding energy dynamics, which is fundamental to the fitness of wild fish.