Stroke rate detection from accelerometer data on marine mammals.

Overall dynamic body acceleration (ODBA) is widely used as a proxy for movement activity. However, for large, slow-moving animals such as cetaceans, ODBA is dominated by orientational shifts rather than dynamic movements. We argue for using stroke rate frequency as an alternative to ODBA and present a method for estimating stroke rate from acceleration data, using narwhals as a case study. The resulting stroke rates aligned with diving phases and depths, where gliding is expected to be prevalent. The results were also competitive with state-of-the-art magnetometer-assisted approaches. The method also successfully detected individual strokes during a 30 s video of narwhal surface swimming. Using combined surge and heave acceleration produced better stroke estimates than using either in isolation. Our findings demonstrate that accelerometer-based tail stroke frequency is a valid proxy for activity level in marine mammals and highlight the importance of correcting for orientational effects.