A framework to estimate the likelihood of species interactions and behavioural responses using animal-borne acoustic telemetry transceivers and accelerometers.

Interactions between animals structure food webs and regulate ecosystem function and productivity. Quantifying subsurface behavioural interactions among marine organisms is challenging, but technological advances are promoting novel opportunities. Here, we present a framework to estimate when there is a high likelihood that aquatic animal subsurface interactions occur and test for a movement-related behavioural response to those interactions over short temporal scales (days) using a novel multi-sensor biologging package on a large marine predator, the Greenland shark (Somniosus microcephalus). We deployed a recoverable biologging package combining a VEMCO Mobile Transceiver (VMT), accelerometer and a temperature-depth tag to quantitatively assess fine-scale behaviour during detection events, that is when sharks carrying the novel VMT package (animalR , n = 3) detected sharks independently tagged with transmitters in the system (animalT , n = 29). Concurrently, we developed simulations to estimate the distances between animalR and animalT by accounting for their swim speed, the estimated detection efficiency of the VMT and the number of consecutive transmissions recorded. Accelerometer-derived activity indices were then used as a means to test for response to potential interactions when animals are expected to be in close proximity. Based on this approach, the three VMT-equipped Greenland sharks exhibited higher body acceleration and greater depth changes during detections, suggesting a potential behavioural response to the presence of other sharks. A generalized additive model indicated a moderate increasing relationship in activity associated with a greater number of animalT detections. Through the proposed framework, detection events with varying probabilities of interaction likelihoods can be derived and those data isolated and explicitly tested using acceleration data to quantify behavioural interactions. Through inputting known parameters for a species of interest, the framework presented is applicable for all aquatic taxa and can guide future study design.