Northern bottlenose whales in a pristine environment respond strongly to close and distant navy sonar signals

Impact assessments for sonar operations typically use received sound levels to predict behavioural disturbance in marine mammals. However, there are indications that cetaceans may learn to associate exposures from distant sound sources with lower perceived risk. To investigate the roles of source distance and received level in an area without frequent sonar activity, we conducted multi-scale controlled exposure experiments (n = 3) with 12 northern bottlenose whales near Jan Mayen, Norway. Animals were tagged with high-resolution archival tags (n = 1 per experiment) or medium-resolution satellite tags (n = 9 in total) and subsequently exposed to sonar. We also deployed bottom-moored recorders to acoustically monitor for whales in the exposed area. Tagged whales initiated avoidance of the sound source over a wide range of distances (0.8–28 km), with responses characteristic of beaked whales. Both onset and intensity of response were better predicted by received sound pressure level (SPL) than by source distance. Avoidance threshold SPLs estimated for each whale ranged from 117–126 dB re 1 µPa, comparable to those of other tagged beaked whales. In this pristine underwater acoustic environment, we found no indication that the source distances tested in our experiments modulated the behavioural effects of sonar, as has been suggested for locations where whales are frequently exposed to sonar.

The base of operations was the 32-m motorised sailing vessel Donna Wood. The vessel had four VHF antennas mounted on top of its 25-m foremast that were connected to a radio direction finder (DFHorten, ASJ Electronic Design) to enable tracking of the focal animal at the surface. DTAGs were deployed from the deck using a pneumatic launcher (Aerial Rocket Tag System (ARTS), LK-ARTS) or from the bowsprit using a 7-m carbon fibre pole. Satellite tags were deployed from the deck of the sailing vessel or from a dinghy, using a pneumatic launcher (JM Standard, Dan-Inject ApS; or ARTS). Each year, an acoustic recorder (DSG-ST, Loggerhead Instruments Inc.) attached to a mooring was deployed upon arrival in the study area, which recorded continuous sound (144 kHz; -168 dB re V/μPa) before, during and after the experiments. At the end of each field season, this passive acoustic monitoring (PAM) device was redeployed with a reduced duty cycle of 6% (2.5/45 min) to allow for acoustic data collection until the next summer. The recorder was positioned at a depth of~100 m above the sea floor, in a geographical location where bottlenose whale abundance was expected to be high (Fig. 1b). To enable accurate estimations of the sonar received levels experienced by the whales, we collected oceanographic measurements at locations between the source and the tagged whale and near the mooring using a conductivity, temperature, and depth-profiler (Mini-CTD, Valeport Ltd.) or an expendable bathythermograph (MK21/USB, Sippican Inc.).
The tagged whales were subsequently exposed to simulated naval sonar from the drifting sailing vessel. Once a DTAG was attached to a whale, this became the focal whales, which was tracked visually by observers on deck or in a 13-m high crow's nest until the tag released from the animal.
(Visual contact was temporarily lost for the Distant experiment). Surface observations collected by the visual observers at every surfacing period included the distance and bearing to the focal animal, and the size and composition of its group. Locations of the animals with satellite tags were tracked via the ARGOS system. Bottlenose whales are inquisitive animals and often approach boats, which is the best opportunity to place tags upon them. Therefore, we attempted to satellite tag other individuals in the group for a maximum of 1 h after the DTAG was attached. Controlled exposure started after 4 h of baseline DTAG data were collected.
Each focal whale (with DTAG) was subjected to either a Close exposure or a Distant exposure treatment. In both treatments we transmitted a sequence of simulated sonar pulses that was representative of signals from active sonars used by navies for long-range detection of targets. The acoustic source and details of navigation and transmission protocols differed between the experiments (Table 1). During Close exposure, the vessel was positioned to the front and side of the focal animal's travel path at the start of the experiment, at~800 m from the animal. Sonar stimuli were amplified (Z8000, Cadence Sound Systems, Inc.) and projected into the water by a moving-coil transducer (LL9642T, Lubell Labs Inc.) at 8 m depth. During Distant exposure, the sailing vessel moved away from the focal animal to a location that was determined to maximise the SPL in the area where the animal was last sighted based upon in-situ acoustic propagation modelling. To produce the same target received SPLs as during Close exposure, the Distant exposure was performed using an underwater acoustic source system (ULHPAS, Applied Physical Sciences Corp.). This system consisted of a vertical line array of 15 disk transducers, a deck-top control unit, and a laptop. The source array was positioned at a mid-array depth of 17 m during transmission.
Tagging and experiments were conducted under permits from the Norwegian Animal Research Authority (permit no 2011/38782 and 2015/23222) and Icelandic Ministry of Fisheries in compliance with ethical use of animals in experimentation. Experimental procedures were also approved by the Animal Welfare Ethics Committee at the University of St Andrews.

Detailed descriptions of the experiments
The behaviour of the animals during each experiment was described in detail based on the DTAG, PAM and satellite tag data. These detailed descriptions, which can be found below, were partly based on expert-identification of the responses in the DTAG data (sensu Southall et al. 2007 andMiller et al. 2012) in addition to the other types of analyses. We used this method to judge whether a behavioural change had occurred, whether it was likely in response to the sonar, when the response started and when it approximately ended. Responses were first independently identified by two panels of four experts based upon inspection of standardised plots (Appendix I). One panel included authors C.C., S.I., and P.W. and the other panel authors P.K., F.P.L. and R.H. The panels were blind to each other's assessments, but were not blind to the experimental condition or timing of the exposure phase because they were familiar with the experiments. Thereafter, the two panels compared and assimilated their results in the presence of an adjudicator (author P.M.) to reach consensus (Table S6). Prior to consensus, 71% of responses were identified by both panels.

Close experiment 2015-1
When the group of whales was first encountered it consisted of 4 animals that were often getting close to ('seeking') the sailing vessel and 3-4 animals that stayed further away. All whales in the group were judged to be small adults or large juveniles. One animal in the group was tagged with a DTAG (ha15_171a). Group size remained 6-8 throughout the tracking period. The tagged whale started foraging about halfway into the baseline period, making deep dives (>452 m) with regular search clicks and foraging buzzes, and continued foraging for 2 h until the start of exposure (Fig. S1). The experiment started at 20-Jun-2015 15:13:00 and had a duration of 15 min. A few seconds after the first sonar signal was received, the whale broke off a foraging dive, ceased sound production, and made a right turn towards the drifting source vessel (66 dB re 1µPa; 633 m). Next, the whale started moving towards the source on a highly directed course and subsequently kept encircling the source vessel until the end of exposure, reaching a minimum distance of~20 m. The first subsequent foraging dive of the animal started 24 min after the CEE had ended, suggesting that the behavioural disruption due to this low-level sonar exposure was relatively short. Change-points were not identified in the MD metric for avoidance or for change in locomotion (Fig. S1).
Bottlenose whale sounds were not detected in the PAM recording in the last 1.3 h before and during exposure ( Fig. S3a), at 37 km from the source. However, click-present periods were frequently observed before that period and in the next two days, with the first one starting only 7.5 min after exposure ( Fig. S3a), which suggested regular occurrence of northern bottlenose whales in the area.
The sonar signals were likely inaudible to the animals close to the moored acoustic recorder location due to the combination of high transmission loss and low SL (Table 1).

Close experiment 2015-2
The whale was part of a group with 2 other small adults and one juvenile when it was tagged with a DTAG (ha15_179b). Before exposure, the whale made several deep foraging dives (>393 m) that were separated by bouts of one or more shallow dives (100-200 m) (Fig. S2). The sonar transmission sequence started at 29-Jun-2015 02:48:00 UTC and had a duration of 15 min (Table 1). The sonar exposure started when the whale was in a shallow diving bout. Upon reception of the first sonar signal, the whale made a sudden movement (potential startle response) and initiated a high-speed descent (127 dB re 1µPa; 814 m). Sounds from the focal whale were not recorded during the entire 840-m dive, and this lasted substantially beyond the end of exposure. The tag record showed highly elevated swim speeds, low variations in pitch and heading, and strong and consistent fluking motions throughout the exposure period. The whale's horizontal movement was directed away from the source location during and after exposure, for a total duration of 6.5 h, although the animal had started travelling on a course away from the source prior to exposure. Deep foraging dives were recorded again before the end of the tag record. A change-point in the MD metric for avoidance was reached in the beginning of the exposure period (but not for changes in locomotion), and the MD remained elevated until the animal resumed foraging (Fig. S2).
Three satellite tags were deployed prior to this experiment (Fig. 1b). The presence of a period of tortuous horizontal movements and deep dives indicated that one satellite-tagged whale (ID134668) initiated a 9-h foraging bout around the time of exposure (Fig. 3b). Avoidance behaviour was not apparent for this whale (82 dB re 1µPa; 38 km) (Fig. S12). The other two satellite-tagged whales were further from the sound source (>201 km) when the sonar transmissions started, because these two animals had been on a highly directed southern course for several days (Figs S13-S14). Their horizontal movements directly before and during exposure were classified as high-speed directional, hampering our ability to assess avoidance. However, the diving behaviour of one whale (ID134670) was undisturbed (62 dB re 1µPa; 345 km) so this whale did not appear to exhibit a strong avoidance response (Fig. 3b). Dive data were not sufficiently available for the other whale (ID134669) to make the assessment (Figs 3b, 4; Table S2).
Bottlenose whale clicks were not detected at the moored acoustic recorder location (26 km from the source) over a period between 6 h before and 4.8 h after exposure (Fig. S3a), suggesting northern bottlenose whales were never close to the recorder for that period.

Distant experiment 2016-1
One DTAG (ha16_170a) was deployed on a group of 4 animals (1 immature male, 2 immature females, 1 subadult) prior to the experiment. During the 5.2-h baseline period the tagged whale made regular foraging dives to 500-700 m (Fig. 2) within a limited spatial area (Fig. 1c). The sonar transmission sequence, with a duration of 35 min, was initiated at 18-Jun-2016 12:16:00 UTC. This exposure period coincided with a 25-min dive (Fig. 2) that began as a typical shallow dive but then was extended in depth and duration (similar as in experiment 2015-2, Fig S2). The first response was judged to have occurred at the start of the first ascent period, when the animal simultaneously reduced speed and initiated a 360°turn in heading (77 dB re 1 µPa; 16.8 km). After several depth inflections and just before the final ascent, the animal started an avoidance response (117 dB re 1 µPa; 16.6 km) ( Fig. 1d). Foraging buzzes or consistent periods of regular clicking by the focal animal were not detected during or after exposure (Fig. 2). After the unusual dive the animal kept moving away from the exposure site for >7.5 h (Figs 1c,2). The tag was released 36.9 km from the location where the avoidance response had started (in comparison, the horizontal displacement during the preceding 5.5 h had been only 3.5 km).
A change-point in the MD metric for avoidance movement was detected at the beginning of the exposure period, and these MDs stayed elevated until the end of the record (Fig. 2). No change-point was identified in the MD metric for energetic cost of locomotion.
Six satellite tags were deployed (Fig. 1c), which included two tags (ID161592 and ID161593) on members of the same group as the focal animal with DTAG. For these six whales, the source distance at the start of exposure ranged from 12.8-27.1 km and received SPLmax from 120-126 dB re 1 µPa.
All satellite-tagged whales appeared to initiate avoidance responses similar to the focal whale, with animals travelling on directed courses towards southwest for several hours after the exposure (Fig.   1c). Horizontal movements before exposure were classified predominantly as tortuous and thereafter mostly as high-speed directional (Fig. 1c). These state predictions correlated to dive patterns during a 24-h period around the time of exposure, with the high-speed directional state often being associated with repeated diving to intermediate depths (Fig. 3a). Four of the six whales initiated a long (1.2-2.2 h) and deep (992-1552 m) dive during exposure (Fig. 3a). The only deep dive for which higherresolution depth measurements were collected by the satellite tag had a shallow ascent (ID161591).
Echolocation clicks of northern bottlenose whales were detected in the PAM recording during exposure, and these detections were followed by a click-absent period that started when the sonar was still transmitting (Fig. S3b). The duration of this period was 13h55min (13h46min according to the auditor). This observation was a statistical outlier (at 0.05 level) compared with the durations of preexposure click-absent periods (Fig. S3c), suggesting that the experimental exposure caused whale groups near the moored acoustic recorder location (25 km from the source) to stop echolocating and/or move out of the area. The received SPL for these groups at the start of the click-absent period was 95 dB re 1 µPa (90% confidence interval from 80-103 dB re 1 µPa; Table S2).

Long-term averaged spectrograms
To give an indication of the pristineness of the Jan Mayen area, the presence of active sonar was detected using long-term averaged spectrograms (LTSA) of acoustic data that were recorded on the bottom-moored acoustic recorders over a 2-year period (June 2015 -June 2017). LTSAs were constructed by computing for each frequency the maximum short-term PSD averaged over N=8192 samples (corresponding to~0.06 s) within a 2.5-min period. These LTSAs were then manually inspected for presences of naval sonar, which were identified as instances where a significant acoustic energy was observed within a limited frequency band of 1-10 kHz. Sonar activities would have been easily detectable for SPLs exceeding the ambient levels by 10 dB, corresponding to roughly~100 dB re 1 µPa for the average ambient noise conditions on the recorders. During the recording periods outside of June, the recorders were sampled at a duty cycle of 6% (2.5 min every 45 min). We could therefore not exclude that short duration exposures were present during the two-year period monitored. However, since sonar operations tend to continue over longer periods, it is likely that nearby sonars would have been picked up by this method. Figure S1: DTAG data from a northern bottlenose whale (ha15_171a), which underwent controlled exposure to naval sonar during experiment 2015-1. Grey vertical lines indicate the start and end of the exposure period. Figure S2: DTAG data from a northern bottlenose whale (ha15_179b), which underwent controlled exposure to naval sonar during experiment 2015-2. Grey vertical lines indicate the start and end of the exposure period.    * Only one sonar pulse, received at minimum distance, was detected in the audio recording. Therefore, the received levels were estimated by assuming spherical spreading and applying a correction for the difference in source-whale distance (20 * log10(15.2 / 632.5)) with the measured SPL. This correction factor was assumed to be accurate (90% confidence interval) within plus/minus a factor of~3 based upon the typical position uncertainty in this type of movement tracks and the fact that the whale was sighted <1 min before it was at the minimum distance to the source. Response onset was taken as the time when last faint click was detected in the recording. ** Received levels omitted as they were far below the expected ambient noise levels because of the low SL.^R eceived levels are rough estimates based upon spherical spreading with absorption loss at 1 kHz (0.036 * distance(km); Richardson et al., 1995). *** Cessation of sound production that was judged to be potentially associated with avoidance to the sonar exposure. Response onset corresponded with the start of 14-h click-absent period in the PAM recording. Distance-dependent effect of sonar on transitioning from state 1 to states 1, 2, and 3 Table S5. ΔAICs for hidden Markov models fitted to satellite tag data of reduced sets of individuals. Based on visual surface observation and a comparison of the horizontal movement and diving patterns of the whales, we judged that ID161592 and ID161593 were probably associated with each other and ID161588 and ID161590 were possibly associated with each other, during the 8-h period following the sonar exposure.