The advantages and challenges of non-invasive towed PILOT tags for free-ranging deep-diving megafauna

Abstract

Background Biologging technologies have yielded new insights into the ecology and behaviour of elasmobranchs, but to date, most studies involve animal capture and restraint to attach tags. Capturing animals usually results in a period of atypical behaviour after release and is undesirable or simply not possible for large and vulnerable elasmobranchs such as mobulas and whale sharks. To avoid animal capture and restraint, we developed and tested two non-invasive multisensor towed tags. The use of towed packages creates additional data analytical challenges relative to fixed packages because towed devices wobble independently of animal movements. We present five examples, two mobulas (reef manta and sicklefin devil ray) and three sharks (blue, tiger and whale shark), to illustrate the advantages and challenges of this approach. We used animal-borne video to validate behavioural data derived from accelerometers and conducted an experiment to compare accelerometer data from attached and towed tags simultaneously deployed on a shark. Results We used fluid dynamic models to calculate the added drag of towed devices on target species. We found that drag impact is acceptable for short-term tagging of large mobulas, but the drag penalty associated with the current camera tag design is greater than 5% for most mature blue sharks. Despite wobble effects, swimming behaviour (tail-beat and wing-stroke frequency) captured by towed accelerometers was consistent with those attached directly to the animal and with data from animal-borne video. Global Positioning System (GPS) sensors recorded up to 28 and 9 geolocations per hour of surface swimming by sicklefin devil ray and blue sharks, respectively. Conclusions Towed tags with non-invasive attachments provide an effective alternative for acquiring high-resolution behaviour and environmental data without capturing and handling animals. This tool yields great potential to advance current knowledge of mobula ecology and behaviour without capture or invasive tagging.

First insights into the shortfin mako shark ( Isurus oxyrinchus ) fine-scale swimming behaviour

“…Tag deployments were conducted in July 2020 near Faial Island (38.60° N, 28.50° W), in the Azores archipelago, Portugal. Three mako sharks were tagged with G-Pilot tags [ 21 ] using a non-invasive harness-like towing system [ 22 ]. Different deployment methods were used for each shark: shark #01, a 143 cm female, was captured using longlines (3 h of soaking time); shark #02, a 155 cm male, was caught with a fishing rod after being attracted with chum ( ca 15 min of handling); shark #03, a 160 cm female, was attracted with chum and tagged while freely swimming following [ 22 ].…”

“…In addition, each package also included a SPLASH10-F tag (ARGOS-link with FastLockGPS, Wildlife Computers, Inc. WA, USA) and a VHF radio transmitter (F1835B, Advanced Telemetry Systems, Inc. MN, USA). After release from the sharks, the tags were recovered using a combination of ARGOS locations and radio telemetry [ 21 ], and their data were offloaded for analysis. No tagging apparel was left on the animals.…”

“…Using the surging acceleration for sharks #02 and #03 and the sway acceleration for shark #01, time-series of dominant tail-beat frequencies (TBF) and amplitudes at 1 Hz resolution were extracted from spectrum analysis based on a continuous wavelet transformation (CWT). Although it has been shown that surging acceleration provides a clearer tail-beat signal for the G-Pilot towed tag system [ 21 ], we used the swaying acceleration for shark #01 due to a failure of the accelerometer recording longitudinal acceleration. Visual comparisons of surge and sway accelerations from sharks #02 and #03 showed similar patterns (electronic supplementary material, figure S1 and figure S2).…”

“…In short, to calibrate the rotation frequency, the tags were vertically dropped multiple times with varying weights. The vertical velocities from each drop were then linearly regressed against the respective average paddle wheel's rotation frequency [ 21 ]. The resulting regression models explained more than 99% of the vertical velocity's variance (electronic supplementary material, figure S3).…”

“…The resulting regression models explained more than 99% of the vertical velocity's variance (electronic supplementary material, figure S3). A sampling rate of 100 Hz was chosen to allow the magnetometer sensor to read the paddle wheel magnets passing at high speed [ 21 ].…”