Repeated narwhal interactions with moorings challenge safety assumptions of passive acoustic monitoring in the Arctic

Abstract

Passive-acoustic monitoring is known as a non-intrusive and transformative tool for ecology and has been increasingly used for conservation and biodiversity monitoring. This study, however, identifies a high level of curiosity in narwhals (Monodon monoceros) with respect to scientific moorings and partially explains recent cases of narwhal entanglements. Using acoustic data from different locations and years together with stomach content analysis, it is shown that foraging narwhals engaged in repeated hits on seafloor moorings (11 times per day), presumably out of curiosity or due to confusion with food items. It is a behavior previously unknown for odontocetes. These results imply that oceanographic monitoring might alter the behavior of whales and poses a risk to their well-being, which should be investigated and accounted for in design. Our findings reveal the intrusive nature of a key scientific method, with implications for the management and conservation of vulnerable marine mammals.

Passive acoustic monitoring may inadvertently affect narwhal behavior. Field observations of narwhals in Greenland show they repeatedly interact with seafloor moorings, possibly mistaking them for food, raising concerns for cetacean conservation.

Introduction

Listening in the ocean with Passive Acoustic Monitoring (PAM) allows scientists to detect underwater sounds produced by marine animals, natural sources, and human activities¹. Autonomous moored PAM systems enable almost continuous monitoring and study of marine animals in their natural environment from a distance, likely without disturbing their behavior². Using PAM to detect acoustically active animals helps to census biodiversity, understand animal behavior and habitat use, and reduce the negative impacts of human-made noise^1,2. For these reasons, scientists increasingly rely on PAM to answer fundamental ecological questions and manage conservation³. They consider that PAM does no harm while yielding an understanding of undisturbed animal behavior^1,2. This paper presents reproducible evidence from Greenland (Kalaallit Nunaat) that calls for a re-evaluation of this hypothesis. Each summer, thousands of narwhals arrive at their key summering ground in Inglefield Bredning Fjord (Kangerlussuaq), northwest Greenland^4,5, where they presumably raise calves and forage for food. Narwhals are known for their high sensitivity to anthropogenic disturbance and extreme cardiovascular reaction to entanglement⁶. As an Arctic-endemic marine mammal that is strongly affected by climate change and overharvesting, its conservation urgently requires the management of anthropogenic activities at the scale of sub-populations⁷. In August 2022 and 2023, underwater sound recorders were deployed in Inglefield Bredning Fjord for a study of narwhals. In addition, to verify if narwhals were diving for foraging and generally could reach the depths of deployments, the contents of narwhal stomachs were collected from Inuit subsistence hunts and analyzed.

The deepest recorders (~ 260–400 m) were repeatedly physically hit by narwhals foraging in the area. Understanding this behavior is scientifically and ethically important because inhabitants of the area and the Government of Greenland are well aware of deadly narwhal entanglements in scientific and fishing gear⁸, and scientists might rely on lethal methods without recognizing it. There are multiple examples of other human activities in the ocean with unwanted consequences. Instances of baleen and toothed whales entangled in submarine cables have been known since 1878 (e.g., sperm whales Physeter macrocephalus, humpback whales Megaptera novaeangliae;⁹), and a growing body of evidence suggests that entanglement in fishing gear is a significant cause of death in whales and dolphins^10,11, up to 300,000 individuals per year¹².

Results and Discussion

Interaction sounds

In August–November of 2022 and 2023, the acoustic presence of narwhals was detected in 0.7%, 3.2%, and 9.4% of data at Bowdoin, Tracy, and Heilprin fjords, respectively (Supplementary Fig. 1). Amongst the many ultrasonic narwhal sounds collected in glacial fjords at Tracy and Heilprin stations (water depths of ~400 m and  ~259 m), the loudest sounds were produced mechanically when narwhals struck and presumably rubbed the recorders (Fig. 1; Supplementary Movie 1). Approximately 5–11% of data with narwhal acoustic signals contained these hits. Such sounds could exceed 600 Pa in amplitude (or 175.6 dB re 1 μPa), and occurred repeatedly on stations  ~25 km apart, on different years, different days, and up to five times per minute (Supplementary Fig. 2a). The resulting waveforms start with a foraging buzz of increasing amplitude, followed by a low-frequency hit (Supplementary Fig. 3), broadband “rubbing" of the recorder, low-frequency wobbling of the set-up, and finally rarefied echolocation clicks. The terminal buzz is typical of odontocetes and bats and precedes the terminal strike, often called an “attack” ¹³. The inter-click interval of the buzz (ICI; Fig. 1d) decreased to  ~2 ms just before each hit, as expected for a foraging narwhal^14–17. Converting this ICI, Δt, into the maximum range-to-target yields 0.7–1.5 m, assuming the click interval is twice/once the two-way travel time like in false-killer whale and beluga or dolphins (r ≤ Δt × v/(2 × c), where v is the speed of sound in seawater, 1500 m s⁻¹ and c is 2 or 1^17,18). This distance range is less than the tusk or body length of a narwhal. Moreover, assuming that ICIs are twice as long as the two-way travel time, the reduction of the ICI from  ~20 to 2 ms (equivalent to a range-to-target of  ~7.5 to 0.7 m) in 2 s corresponds to a maximum movement speed of 3.4 m s⁻¹, which is in general agreement with the mean and maximum vertical speed of a narwhal recorded by time-depth sensors, i.e., 2–4 m s^−1 19,20. Assuming that ICIs are as long as the two-way travel time (i.e., no time for the narwhal to process the received echo) corresponds to a less realistic maximum speed of 6.75 m s⁻¹.

Fig. 1. Records of narwhal hits on a hydroacoustic mooring near the seafloor in a glacial fjord at Inglefield Bredning, northwest Greenland.

a Sketch of the setup at the Tracy study site. b Sketch of the hydroacoustic mooring on the seafloor. c Typical data produced by narwhals hitting the recorder: unfiltered acoustic data converted to pressure units (Pa) or the sound pressure level (SPL) on the decibel scale (dB re. 1 μPa) with the corresponding spectrogram, created using a 1024-point FFT window size with a 0.016 s Hamming window and 50% overlap (dB re. 1 μPa). d The same data as in c, but high-pass filtered ( >5 kHz) with the corresponding inter-click interval (ICI). Red circles indicate automatically detected peaks.

Temporal variations and total numbers of hits

The high-frequency acoustic data from three moorings yielded 247 aurally confirmed incidents of narwhal hits (Fig. 2). All detections were made at two stations in the inner parts of Inglefield Bredning (138 in front of Tracy Glacier, between 12 August and 7 October 2022; and 109 in front of Heilprin Glacier, between 9 August and 28 September 2023), until the narwhals left the freezing fjord²¹. The shallower (192 m) Bowdoin station only recorded the acoustic presence of narwhals but no hits (Supplementary Fig. 1). As a result of the duty cycle of the recorders (collecting 4.5 minutes of sound every 20 minutes), the data are discontinuous and cover only 22.5% of the total deployment time. Given that 77.5% of the sound data are missing and assuming that narwhal hits occurred relatively consistently over time, it is possible that the recorders near Tracy and Heilprin glacier received up to 613 and 484 narwhal hits during the  ~2-month detection period (i.e., n_t = n_r × 100/22.5, where n_t is the total number of hits and n_r is the number of recorded hits), corresponding to an average of 11 and 10 hits per day, respectively.

Fig. 2. Temporal variation of detected events in Inglefield Bredning in different years and locations.

Occurrence (red) and cumulative number (black) of detected and aurally confirmed narwhal hits (n = 138 and 109) on the hydroacoustic moorings at Tracy (2022) and Heilprin (2023) fjords. Blue dotted vertical lines indicate the beginning and end of the recording periods (the end line for Heilprin is in May 2024 and thus not visible). The insets show the probability of hits by time of day (local time, LT; day and night are defined as 06:00–18:00 and 18:00–06:00, respectively).

Diel rhythm

Most hit detections (65%, Tracy, 2022; 89%, Heilprin, 2023) occurred during daytime between 06:00 and 18:00 local time (Fig. 2). The presence of the diel pattern was ascertained by Rayleigh and Hodges-Ajne tests, allowing rejection of the null hypothesis that the events were uniformly distributed throughout a day (i.e., 24 hours; 2022 data yielded p − values of 0.0014 and  < 0.0001, respectively; 2023 data yielded p − values of  <0.0001).

In comparison, narwhal acoustic presence was observed throughout the day and night, but was also more likely during daytime (Supplementary Fig. 1). Rayleigh and Hodges-Ajne tests confirm the presence of a diel cycle at Tracy and Heilprin data (p − values of  <0.00001), while Bowdoin data is less clear (p − values of 0.0045 and 0.006, respectively).

Foraging and primary prey

All 16 narwhals (13 male and 3 female with an average length of 4.4 m) harvested in the study area had stomach contents (remains of fish, shrimp, squid, and stones). By biomass, 94.2% of the narwhal diet corresponded to cod (Gadidae; Arctogadus glacialis and Boreogadus saida; Fig. 3a, b; Table 1). In the remaining 5.8% of wet mass, decapods and cephalopods were the most significant prey (2.6% and 1.4% of wet mass, respectively). Other fish species, including halibut (Reinhardtius hippoglossoides), were a rare type of prey.

Fig. 3. Narwhal diet.

a, b Otoliths from narwhal stomachs indicate a diet of mainly a polar (Boreogadus saida) and b arctic cod (Arctogadus glacialis). (c) A stone retrieved from the stomach of a narwhal. Scale bars are 1 cm (photographs: M. Ogawa).

Table 1.

Narwhal primary diet (Inglefield Bredning, 2022–2023)

	Taxa	Wet mass, %
Fish	Cod (Gadidae)	94.2
	Boreogadus saida	26.1
	Arctogadus glacialis	68.1
	Other fish	1.8
Invertebrate	Cephalopod	1.4
	Shrimp (Decapoda)	2.6
	Zooplankton	<0.1

Confusion with prey or scrubbing?

Our data reveal animal interaction with deep-sea moorings, a higher likelihood of this interaction during daytime, a higher likelihood of narwhal acoustic presence during daytime near these moorings, and that narwhals were foraging in the same area of Inglefield Bredning Fjord. Earlier described properties of the signal (i.e., sequence, buzz features, and termination) cannot be explained by the small prey hitting the mooring or other animals. Specifically, a review of the biophony in Greenlandic waters suggests that even if polar cod (Boreogadus saida) makes sounds, only beluga (Delphinapterus leucas; narwhal’s closest relative) can make similar ultrasonic sounds²². However, belugas are not seen or hunted in the inner parts of Inglefield Bredning Fjord. Therefore, our results suggest that narwhals repeatedly dived to visit the moorings out of playful curiosity or, more likely, due to confusion with potential prey, as discussed below.

In a report on sperm whale entanglement in submarine cables in the Pacific, Heezen⁹ speculated that whales attacked tangled masses of slack submarine cable, mistaking them for prey. The area around Tracy station comprises a vast area of flat seafloor with depths of  ~400 m; the area around Heilprin station corresponds to a rise in the middle of the fjord surrounded by depths of  ~350 m²³. In both settings, the moorings are exposed and might act as easy-to-spot acoustic reflectors. For echolocating narwhals, the recorder with batteries (a titanium cylinder containing  ~0.2–1.3 L of air) would appear to be an elongate (≤0.5 m) deep-water reflector located several meters above the seafloor. Arctic cod (Arctogadus glacialis) and squid (Gonatus fabricii) found in narwhal stomachs can be up to 30 and 50 cm in size, respectively²⁴, and the halibut (Reinhardtius hippoglossoides) that narwhals prefer to eat has an average length of 39 ± 8 cm (s.d.)²⁵. Considering that cod has a swim bladder, which contributes  >90% of the acoustic backscattering by fish²⁶, the narwhals might confuse the recorder with cod. Whether the echoes from the recorder look like those from a cod might be an interesting question to study. Narwhals can probably discriminate fine differences in texture or density of echolocation targets like other odontocetes²⁷. For example, laboratory studies suggested that bottlenose dolphins (Tursiops truncatus) and false killer whales (Pseudorca crassidens) were capable of subtle cylinder-wall-thickness discrimination, especially with practice^28,29. On the one hand, even if the hearing abilities of narwhals and object perception by narwhals remain unknown, there is no reason to claim they would not have developed the abilities described in other odontocetes to classify underwater targets. On the other hand, advanced sonar abilities do not guarantee safe interaction with anthropogenic structures; for example, some of the sharpest eyes on Earth do not stop griffon vultures and bald eagles from crashing into wind turbines³⁰. The chain of four buoys that keep the mooring suspended would appear to be a larger and stronger reflector than the recorder and might be detected from a longer range, but is too large to be swallowed (Fig. 1b). Similarly, the acoustic release located several meters below the recorder is also too large to swallow. The acoustic data show that many hits occurred directly on the recorder. However, an interaction with the buoys, CTD, or release mechanism cannot be excluded (see also below). Without video, it is difficult to understand the details of the interaction. If the target was prey with a microphone, after the buzz and hit, the prolonged “rubbing" sound could correspond to entering the narwhal’s mouth. This seems unlikely for a recorder tightly attached to a line. Instead, the friction between the recorder and the skin during the sliding of the animal a few meters long seems a reasonable explanation. In fact, other arctic whales, like beluga or bowheads (Balaena mysticetus), rub their bodies on the rocky seabeds in the Canadian Arctic – likely to facilitate molting³¹. Though little is known about molting in narwhals, mooring rubbing could be the associated behavior. Even if the causes of this interaction remain speculative, our findings have the following implications of conservational and scientific relevance.

Temporal variation, incident sequences, and benthic foraging

Our results imply a higher likelihood of interaction with moorings in the daytime. It is reasonable to assume that having such interaction is difficult without more dives to the seafloor during the daytime. This argument suggests that the deep dives necessary to reach the seafloor were also more common during the daytime. The latter assumption is consistent with the diel pattern of dive-depth data for a narwhal tagged in East Greenland²⁰, which was diving irregularly but tended to make the deepest foraging dives around noon and dived to shallower depths (<200 m) at night. Furthermore, such narwhal primary prey as cod and squid are known for diel vertical migration to shallower water at night and, therefore, could pace the diurnal diving behavior of narwhal^20,32.

A sequence of up to four or five hits per minute (Supplementary Fig. 2) is interpreted as bumps by a few narwhals diving in a close group. This inference is difficult to verify from the available data. Nevertheless, after discovering that the recorder is not food, it would be unlikely that the same individual would make additional foraging attempts. Moreover, the largest pods reported in Inglefield Bredning Fjord comprise four to six narwhals^4,33. Immediately after the foraging buzz, one to three secondary low-frequency hit events (i.e., repeating within  <1 s) were sometimes recorded (Supplementary Fig. 2b). Presumably, this corresponds to an interaction between a narwhal and the mooring during the continued upward (or downward) movement of the narwhal.

The frequent occurrence of narwhals approaching close to the seafloor in Inglefield Bredning Fjord during August–October suggests they were foraging. This interpretation is consistent with previous stomach content analyses of narwhals harvested in the area by Inuit hunters, acoustic studies^14,17,34, and the authors’ own experience. In particular, the stomachs of harvested narwhals inspected in 2022 and 2023 contained food in each case and some contained stones (Fig. 3c). However, we cannot explain why the whales interacted with the moorings at Tracy and Heilprin fjords but not near Bowdoin Glacier. No interaction was detected in previously archived acoustic data from a solid sound recorder (SoundTrap ST300 STD) attached to an ocean-bottom seismometer and deployed in July-August 2019 at the seafloor of Bowdoin Fjord³⁵. The number of narwhals is generally higher in the eastern parts of Inglefield Bredning Fjord than in Bowdoin Fjord^4,5,14. Moreover, many hunters stay overnight in Kangerluarsuk, right at the mouth of Bowdoin Fjord, and may strongly affect narwhal’s natural behavior¹⁴. However, little is known about the local differences in habitat use, the presence of prey species, other human activities, or some environmental factors that might affect narwhals’ behavior (such as high turbidity due to subglacial discharge plumes³⁶ or iceberg density and noise near the calving front³⁵).

Entanglement risk, instrumentation and observation bias

Considering that the narwhal has the most directional biosonar beam reported for any species to date³⁷ and some of the longest tooth in the animal kingdom (2–3 m long), moving in complete darkness among lines, shackles, and swindles may present a high risk of entanglement. Since 2017 (i.e., in 2017, 2022–2024), at least six passive oceanographic moorings used by scientists from different countries had major troubles in Inglefield Bredning Fjord. Four losses might have resulted from collisions with icebergs or galvanic corrosion; alternatively, narwhal activity could have affected the integrity of the moorings. For example, physical shocks to some commercially available acoustic releases can result in higher pressure on the case and water may enter the release³⁸. Moreover, sections of two passive acoustic moorings deployed by other teams were found at the sea surface of Inglefield Bredning with two corpses of narwhals entangled in lines. According to local hunters, both narwhals were found with lines tangled around their tails (Supplementary Fig. 4), exactly as shown on the cover of the Government of Greenland report on fishing gear⁸. One narwhal was eaten to the bone, presumably by amphipods. Another incident reminded some of the earliest reports on whale entanglement and breakage of submarine telegraph cables in other regions⁹. Specifically, in 1878, Scientific American published a description by a witness onboard the telegraph streamer, Amberwitch, claiming that the carcass of a large whale was brought to the surface and found entangled by the tail in the cable and devoured by sharks and other fish³⁹. In fact, it is known that the Greenland shark (Somniosus microcephalus) scavenges the corpses of narwhals and can easily cut lines⁴⁰.

It is possible that recorders without air may not attract the animals’ attention in the same way as air-filled recorders, while active devices, such as acoustic pingers, might deter and protect narwhals from potential entanglement⁴¹, but more research is needed in this area⁴². Whale entanglements in deep-sea cables ceased in the second half of the twentieth century, presumably thanks to such technological advances as the replacement of submarine cables with coaxial cables, the introduction of torsionally balanced cables (with a reduced tendency to self-coil on the sea floor), proper cable tension, and burial⁴³. What may be the main contributing factor to entanglement is the length of the mooring (which depends on multiple reasons, like a need to sample a water column with a chain of sensors or to separate instruments from extreme sedimentation rates on the sea floor³⁵). The role of the tusk and the tail in entanglement is unclear. However, we suggest that the tusk may increase the vulnerability of narwhals compared to other Arctic cetaceans, as its shape and rigidity could facilitate entanglement in mooring loops, for example, when the animal moves along a line, pushes through a loop, and becomes caught near the tail. Moreover, the flukes of a mature male narwhal have a concave leading edge with no sweepback⁴⁴, which logically should be a disadvantage to escaping a loop⁸. We need to investigate what technological solutions (e.g., short lines or no lines, avoiding prey-size or air-filled sensors, etc.) could minimize the risks of narwhal entanglement in oceanographic moorings. In any case, at this stage, short lines and avoiding loops seem to be the simplest precautions.

If it is part of the narwhal’s behavior to be curious, a narwhal would be expected to naturally interact with man-made items on the seafloor. Irrespective of our definition of “natural", passive acoustic monitoring in the ocean is currently considered to be non-invasive and to document undisturbed (i.e., unaffected by man-made items) animal behavior^1,2. However, the present results are inconsistent with this assumption. For example, the acoustic presence of a species at a particular location or the occurrence of foraging buzzes might result from the attraction of animals to artificial systems. Even the act of passive observation by the animal, at least in some circumstances, may represent modified behavior. For example, it was presumed that a narwhal used echolocation to purposefully scan a recorder suspended from sea ice at their wintering ground in Baffin Bay¹⁶. It is important to understand how widespread such curiosity behavior might be and the risks posed to the narwhal. To our knowledge, recently, scientists from Canada, Denmark, Greenland, and Japan have been conducting oceanographic observations in the fjord, but hydroacoustic deployments in inner parts of glacial fjords remain rare. Still, there have been at least 26 deployments in the Arctic waters around Greenland, which overlap with narwhal habitats⁴⁵, and more deployments can be expected.

In the present study, we describe previously unknown cetacean behavior. Intriguing by itself, it questions the fundamental assumptions of the key method of applied ecology, currently underlying hundreds of annually published papers. The benefits of hydroacoustic and oceanographic studies cannot be overstated and are indispensable for science and conservation^1,2,46. However, the management of the major socially important whale habitats requires monitoring methods that do not harm, do not dismiss local-people voices, and do not misinform. Our results (1) suggest that entanglement risk must be recognized by scientists, policymakers, and managers, and (2) dispel the myth that PAM scares narwhals by showing that it attracts them instead⁴⁷. As a new ocean is opening up in the Arctic, and more human interaction with naïve endemic species is inevitable, careful re-consideration of broadly used techniques is necessary. Understanding animals’ interaction with industrial and scientific infrastructure can help reduce impacts on wild animals and improve our ability to implement and interpret autonomous field observations.

Methods

Study site and hydroacoustic data collection

Since 2022, for the purpose of studying animals and ambient and anthropogenic noise, several oceanographic moorings have been deployed in Inglefield Bredning Fjord, northwest Greenland. As part of this research, three hydroacoustic moorings were deployed to the seafloor. One station was located in a fjord of Tracy Glacier (Qeqertaarsuusarsuup Sermia),  ~20 km from the calving front (77.63°N, 67.08°W; water depth = 402 m (Fig. 1a). The second station was deployed in a fjord of Heilprin Glacier (Qaqujaarsuup Sermia),  ~9 km from its calving front (77.47°N, 66.36°W; water depth = 259 m). In general, the inner parts of Inglefield Bredning Fjord are a summering ground for narwhals^4,5. The third site was located in Bowdoin Fjord,  ~0.6 km from the terminus of Bowdoin Glacier (Kangerluarsuup Sermia; 77.67°N, 68.63°W; water depth = 192 m). This fjord is a historic Inuit hunting ground (for narwhal and seals) that has been the site of ice–ocean interaction studies for several years, and was previously confirmed to be occupied by narwhals^14,35. Underwater sound recorders were moored  ~15 m above the seafloor. Three Soundtrap ST600 recorders (Ocean Instruments, Auckland, New Zealand) with a sensitivity of –176.2, –177.4 and –177.4 dB relative to 1 V/μPa and flat (±3 dB) frequency response from 20 Hz to 60 kHz were used in this study. Sound recorders were powered by 12 lithium-ion rechargeable batteries (Samsung 18650 30Q 3000 mAh) and were upgraded to a 500 m depth rating by installing Prevco 01104-001 pressure relief valves (Prevco, Fountain Hills, USA). The recorders were programmed to store 4.5 minutes of continuous sound every 20 minutes at a sampling rate of 64 kHz. The total length of each mooring was 22 m (Fig. 1b). Four plastic buoys were connected to an anchor with segments of Taston ropes (12 mm diameter) using stainless steel shackles/swivels and titanium rings. Between the buoys and anchor were located a sound recorder (and a conductivity, temperature, and pressure [CTD] recorder in the case of the shallower moorings; SBE37SM MicroCAT, Sea-bird Scientific, Bellevue, USA) and an acoustic release (TMR-6005B; Kaiyo Denshi, Tsurugashima, Japan). For the anchor, bags filled with 150 kg of sand and rocks were used. The moorings were manually deployed from a small boat in August 2022 or 2023 and retrieved in July/August 2023 or 2024. The Tracy station recorder acquired data from 6 August to 22 November 2022 (586 hours); in April 2024, the manufacturer inspected the recorder and found damage to electronics due to seawater ingress, possibly due to a failure of the pressure relief valve. The Bowdoin station recorder acquired data from 2 August 2022 to 31 July 2023 (1959 hours). The Heilprin station recorder acquired data from 7 August 2023 to 6 May 2024 (1500 hours). Total: 4045 hours (1864 Gb).

Acoustic analysis

After listening tens of hours of data from different stations and months while referring to long-term spectrograms, we confirmed that broadband sounds were produced primarily by icebergs and narwhals as expected in the fjord^14,35. We could not manually find any knocking sounds in the absence of narwhal sounds, but to make sure and avoid confusion with icebergs, we used the following hybrid approach. To find sounds of mechanical interactions of narwhals with moorings, a two-step procedure was executed. First, to detect the predominantly ultrasonic narwhal sounds as well as knocking sounds in the hydroacoustic data, the acoustic complexity index (ACI)⁴⁸ was computed using the high-frequency band (20–32 kHz) and a 512 fast Fourier transform (FFT) window size with Kaleidoscope Pro 5.6.2 software (Wildlife Acoustics, Maynard, USA). This method was chosen because icebergs generate a lot of broadband noise³⁵, preventing us from relying on simple computation of sound levels in the ultrasonic band. Nevertheless, this approach could still yield a high ACI due to iceberg cracking events. For this reason, in our second step, each wav file with an elevated ACI was then opened and manually inspected, including aural and visual analysis using Raven Lite 2.0 (Cornell Laboratory of Ornithology, Ithaca, USA). The presence of a diel signal in the acoustic presence of narwhal and the number of detected events was ascertained by Rayleight and Hodges-Ajne statistical tests for circular uniformity. To represent the acoustic data in terms of pressure, P (in Pa), the data were converted using calibration constants provided by the manufacturer. To express the sound pressure level (SPL) on the decibel scale (dB re. 1 μPa), we used the following equation SPL = 20log₁₀(P/P_ref), where P_ref is the reference pressure of 1 μPa.

Stomach content analysis

To identify the dominant type of narwhal prey, we analyzed stomach contents from mature narwhals (n = 16) harvested in Inglefield Bredning during open-water seasons of 2022 (2 males) and 2023 (11 males and 3 females). Two narwhals were killed on 31 July and 5 August 2022, and another 14 were killed between 9 and 21 August 2023 during the Inuit hunt following quota regulations applicable in the region. Narwhals were hunted primarily in the inner parts of the fjord close to Qeqertaq (77.49°N, 66.69°W), within 5–39 km from Tracy mooring (mean ± s.d.: 21 ± 12 km); or within 8–53 km from Heilprin mooring (mean ± s.d.: 27 ± 16 km) for 14 whales with known location of harpooning. The length of the narwhal (from the notch of the tail to the tip of the upper jaw) was measured for six individuals (mean ± s.d.: 4.4 ± 0.9 m; tusk excluded). Prey items were extracted from the stomachs using water and a series of sieves and identified to the lowest possible taxa under a stereo-microscope. Specifically, whole stomach samples were cut open and washed in water over sieves to collect all items larger than 0.5 mm. The contents were examined by identifying and counting sagittal otoliths, cephalopod beaks, and other remains (i.e., skeletal parts). The examination was made under a stereo-microscope (Micronet, YS03C, Kawaguchi, Japan) with a micrometer. The fish otolith length and cephalopod lower beak rostral length were measured with ImageJ software and used as a proxy for whole-animal weights. To estimate the proportional contribution of each identified species, the wet mass was computed from the measured length of otoliths and beaks following regressions from the previous study conducted in the region (ref. ^25,34 and references within); for species with no regressions available (e.g., decapods), an average weight of intact specimens from the literature or this study was used²⁴.

Ethics

We have complied with all relevant ethical regulations for animal use. This study was observational in nature and complied with the current laws of Greenland and Japan. Biological sampling was supported by local Inuit hunters. The research was overseen by Hokkaido University and received permits from the governments of Greenland (22GL2089638, 23GL2089731) and Japan (Hokkaido University, No. 2021-011; Ministry of Economy, Trade and Industry; No. SBIT-WGL-19-S10001).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

E.A.P., M.Og., M.Ot., K.H., and S.S. conducted fieldwork; E.A.P. designed the study, managed and analyzed data, wrote the original draft, prepared illustrations, and revised the manuscript; M.Og. analyzed narwhal stomach contents; S.S. and E.A.P. secured funding; All authors discussed the results and commented on the manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Michele Repetto and George Inglis. A peer review file is available.

Data availability

The data reported in this paper are available in the Dryad repository^49,50.

Code availability

We did not generate any new code in this study. Data processing and visualizations were made with freely available Sound-Trap Host (Release 4.0.17) and commercially available Matlab R2022b, Kaleidoscope Pro 5.6.2, ImageJ, and Raven (Lite 2.0 is openly available).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-09106-4.