Seals (phocids) are generally not thought to produce vocalizations having ultrasonic fundamental frequencies (≥20 kHz), although previous studies could have been biased by sampling limitations. This study characterizes common, yet, previously undescribed, ultrasonic Weddell seal (Leptonychotes weddellii) vocalizations. The vocalizations were identified in more than one year (2017–2018) of broadband acoustic data obtained by a continuously recording underwater observatory in McMurdo Sound, Antarctica. Nine recurrent call types were identified that were composed of single or multiple vocal elements whose fundamental frequencies spanned the ultrasonic range to nearly 50 kHz. Eleven vocal elements had ultrasonic center frequencies (≥20 kHz), including chirps, whistles, and trills, with two elements at >30 kHz. Six elements had fundamental frequencies always >21 kHz. The fundamental frequency of one repetitive U-shaped whistle element reached 44.2 kHz and descending chirps (≥3.6 ms duration) commenced at ≤49.8 kHz. The source amplitude of one fully ultrasonic chirp element (29.5 kHz center frequency) was 137 dB re 1 μPa-m. Harmonics of some vocalizations exceeded 200 kHz. Ultrasonic vocalizations occurred throughout the year with the usage of repetitive ultrasonic chirp-based calls appearing to dominate in winter darkness. The functional significance of these high-frequency vocalizations is unknown.

The Weddell seal (Leptonychotes weddellii) is a large and relatively abundant true seal (family Phocidae) with a circumpolar distribution around Antarctica, including the highest-latitude coastal regions (Reeves et al., 2002). In contrast to the other seals of the Antarctic clade, they prefer expanses of heavy pack ice or thick shore-fast sea ice, using their teeth to maintain access holes in the ice. They dive to at least 600 m and for up to 82 min in search of fish and invertebrate prey year round (Thomas and Terhune, 2009). Weddell seals have been extensively studied, owing to their prevalence near several research stations, their aggregation on the sea ice for pupping and breeding in the austral spring (October–December), and their approachability when hauled out on the sea ice surface.

The Weddell seal's extensive and relatively high amplitude (to 193 dB re 1 μPa-m) repertoire of multiple-element frequency- and amplitude-modulated underwater chirps, whistles, buzzes, and chugs, among other sounds, forms a major component of the underwater soundscape in areas where they are abundant (Terhune, 2019; Thomas and Kuechle, 1982). Thomas and Kuechle (1982) provided the first comprehensive quantification of the species' underwater vocalizations. They described 34 sonic call types (<20 kHz, human-audible) plus 9 accessory sounds recorded in McMurdo Sound in the southwestern Ross Sea. Studies have now described repertoires consisting of 14–50 sonic call types from populations around Antarctica with the variation in repertoire size estimations likely due to geographic and temporal differences and inconsistent definitions of call types. Weddell seals have the most diverse vocal repertoire of any phocid (Pahl et al., 1997; Terhune, 2019; Thomas and Kuechle, 1982).

It is likely that the full diversity of Weddell seal underwater vocalizations remains to be described. Indeed, most studies have been limited to short-term recordings (hours to days) from near the surface beneath shore-fast sea ice and, typically, detected only calls at ≤15 kHz (see Fig. 1). Long-duration recordings appear to be limited to those from the multi-year Perennial Acoustic Observatory in the Antarctic Ocean (PALAOA) effort in the Weddell Sea. In that study most analyses were conducted at ≤15 kHz at a coarse subsampling, and the recording site was beneath an ice shelf, 1 km from the edge (Klinck et al., 2016; van Opzeeland et al., 2010).

FIG. 1.

Maximum reported fundamental frequencies of Weddell seal vocalizations. Bars indicate the mean or maximum of the highest-frequency fundamentals reported in the cited studies, and lines show the upper limit of the recording/analysis equipment frequency response (FR). Ultrasonic fundamental frequencies (≥20 kHz) have been presented in two prior studies (in a trill and a sequence of chirps; Russell et al., 2016; Schevill and Watkins, 1971); most others reported sounds to ≤15 kHz despite higher equipment capabilities. Two studies (asterisks) did not report maximum frequencies of vocalizations. Details of each study are available in the supplementary material.1 The present study (not shown) is based on recordings with an upper FR limit of 256 kHz.

FIG. 1.

Maximum reported fundamental frequencies of Weddell seal vocalizations. Bars indicate the mean or maximum of the highest-frequency fundamentals reported in the cited studies, and lines show the upper limit of the recording/analysis equipment frequency response (FR). Ultrasonic fundamental frequencies (≥20 kHz) have been presented in two prior studies (in a trill and a sequence of chirps; Russell et al., 2016; Schevill and Watkins, 1971); most others reported sounds to ≤15 kHz despite higher equipment capabilities. Two studies (asterisks) did not report maximum frequencies of vocalizations. Details of each study are available in the supplementary material.1 The present study (not shown) is based on recordings with an upper FR limit of 256 kHz.

Close modal

Weddell seal sonic underwater vocalizations are thought to be used primarily for mediating social interactions (Russell et al., 2016; Terhune, 2019). Social functions are supported given that the seals respond with specific vocalizations when presented with playbacks of their recorded calls (Thomas et al., 1983; Watkins and Schevill, 1968), by behavioral observations (Evans et al., 2004; Russell et al., 2016), and since most vocalizations appear to occur when the seals are near the surface (Evans et al., 2004; Moors and Terhune, 2005).

Weddell seals are typically not thought to produce vocalizations having ultrasonic fundamental frequencies (F0 ≥ 20 kHz, above the human hearing range; Terhune, 2019; Thomas and Kuechle, 1982), although studies could have been biased by sampling limitations. Thomas and Kuechle (1982) stated they “found no vocalizations above 20 kHz” and, therefore, recorded data at ≤19 kHz. Likewise, the majority of other studies used an effective upper frequency response (FR) of 15–20 kHz (see Fig. 1). However, two studies have presented limited evidence of ultrasonic vocalizations in Weddell seals: Schevill and Watkins (1971) reported a series of short-duration descending chirps with fundamentals to ≤30 kHz, and Russell et al. (2016) recorded a trill-type vocalization reaching to 22 kHz. These findings are not widely recognized, and it remains unknown whether Weddell seals regularly use vocalizations originating at ultrasonic frequencies.

Other than the two recordings from Weddell seals, there exists only scant evidence for pinniped (seals, eared seals, and walrus) vocalizations having fundamental frequencies ≥20 kHz. In one study of a single captive leopard seal (Hydrurga leptonyx), ultrasonic frequency-modulated (FM) sweeps, buzzes, and pulses were recorded underwater (maximum frequency 164 kHz, peak energy was typically from 50 to 60 kHz; Awbrey et al., 2004; Thomas and Awbrey, 1983). However, field studies have only reported leopard seal vocalizations in the sonic range (≤6 kHz; Erbe et al., 2017). Several other seal species may produce broad-bandwidth roars, hisses, moans, and short-duration clicks with some energy ≥20 kHz (reviewed in Southall et al., 2019). Yet, these appear to be based on sonic-range fundamentals (<20 kHz). Vocalizations with ultrasonic fundamental frequencies have not been reported from eared seals (family Otariidae) or walrus (family Odobenidae; reviewed in Southall et al., 2019).

Ultrasonic vocalizations are, however, produced by a number of aquatic and terrestrial animals for communication and other functions (Sales and Pye, 1974). Perhaps best known are those used in the highly evolved echolocation (active biosonar) abilities of toothed whales (odontocetes) and bats (chiropterans). In these, the reflections of their pulsatile ultrasonic vocalizations permit obstacle avoidance and locating prey with high accuracy, given that short durations and increased sound frequency improve precision (Au, 1993). A primary indicator that vocalizations are being used for echolocation is the emission of a series of pulsed sounds (“click trains”) whose interval varies directly as a function of distance to a target in order to avoid overlapping emissions and returns (Au, 1993).

Longer-duration ultrasonic vocalizations are also known from some toothed whales in which the functions are typically attributed to intraspecific communication. Several dolphins produce whistles whose fundamental frequencies extend into the ultrasonic range (e.g., to 25, 27, and 34 kHz for Stenella longirostris, Stenella frontalis, and Lagenorhynchus albirostris, respectively; Lammers et al., 2003; Rasmussen and Miller, 2002). In addition, some killer whales (Orcinus orca) produce high-frequency sweeping whistles with fundamentals to 75 kHz and durations of ten to a few-hundred ms, the functions of which are unknown (e.g., Samarra et al., 2010).

The present study characterizes a variety of previously undescribed, yet, commonly occurring, ultrasonic underwater vocalizations produced by Weddell seals identified in a long-term dataset of high-frequency recordings (to 256 kHz FR) from McMurdo Sound, Antarctica.

Year-round digital recordings of Weddell seal underwater vocalizations were collected by passive acoustic monitoring over two years (November 2017–November 2019) in southeastern McMurdo Sound, Ross Sea, Antarctica (Fig. 2). The recording equipment was integrated into the shore-cabled McMurdo Oceanographic Observatory (MOO) mooring, which also included a self-cleaning pan-tilt-zoom camera (Octopus, View into the Blue, Boulder, CO) and ocean condition sensors (conductivity–temperature–depth, CTD; SBE37-SMP, SeaBird Electronics, Bellevue, WA). The mooring was installed by divers at a bottom depth of 21 m at the base of the seaward terminus of the McMurdo Station seawater intake jetty (S 77.8510°, E 166.6645°). Recordings were collected continuously throughout the deployment (>90% coverage with occasional short gaps from network and power outages and software bugs), yet, the present study focuses on only the first 13 months of the dataset (November 2017–November 2018).

FIG. 2.

Geographic location, bathymetry, and local distribution of seals. (A) The hydrophone was deployed as part of the McMurdo Oceanographic Observatory (MOO) mooring at 21 m deep in southeastern McMurdo Sound, Antarctica. Except in January to early April 2018 when the ship's channel (SC) was open, thick shore-fast sea ice likely precluded most penguins and marine mammals other than Weddell seals from diving within 10–30 km of the recording site (see Secs. II, Methods and III, Results). (B) Details of MOO environs. Weddell seals are common in Erebus Bay, where they aggregate around predictable access holes in the sea ice (stars). Bathymetry (m) is estimated based on relatively few data points (Davey and Nitsche, 2013), although it largely matches field observations (Cziko, 2020). McM, McMurdo Station (USA); SB, Scott Base (New Zealand).

FIG. 2.

Geographic location, bathymetry, and local distribution of seals. (A) The hydrophone was deployed as part of the McMurdo Oceanographic Observatory (MOO) mooring at 21 m deep in southeastern McMurdo Sound, Antarctica. Except in January to early April 2018 when the ship's channel (SC) was open, thick shore-fast sea ice likely precluded most penguins and marine mammals other than Weddell seals from diving within 10–30 km of the recording site (see Secs. II, Methods and III, Results). (B) Details of MOO environs. Weddell seals are common in Erebus Bay, where they aggregate around predictable access holes in the sea ice (stars). Bathymetry (m) is estimated based on relatively few data points (Davey and Nitsche, 2013), although it largely matches field observations (Cziko, 2020). McM, McMurdo Station (USA); SB, Scott Base (New Zealand).

Close modal

The calibrated broadband omnidirectional digital hydrophone (icListen HF-SB2-ETH, Ocean Sonics, Nova Scotia, Canada; ethernet-connected, GeoSpectrum M24–205 transducer; 118 dB dynamic range, sensitivity −170.8 ± 3.4 dBV re 1 μPa for 10 Hz–200 kHz) was mounted vertically on a stainless-steel strut-channel attached to a 150-kg concrete block, holding the transducer 70 cm off the mud/gravel seabed. Data were recorded at 512 kS s−1 (256 kHz Nyquist frequency), 24 bits, and written as 10-min WAV files (about 900 MB each; with UTC-based timestamps) to a storage array in a heated structure on shore, and then losslessly compressed using the Free Lossless Audio Codec (FLAC; Xiph.org Foundation, Somerville, MA). A software pipeline computed three audio spectrograms (upper limits of 2.5, 25, and 256 kHz) per file and combined those into timestamped PNG images (see the example in the supplementary material1).

Erebus Bay in southeastern McMurdo Sound (Fig. 2) is one of the most populous haul-out areas for Weddell seals, annually hosting up to 2000 individuals (Smith, 1965; Testa and Siniff, 1987). The largest concentrations of individuals occur at major sea ice breeding sites in austral spring (October–December), 10–20 km north of the MOO, where over 400 pups are born in most years (Ainley et al., 2015; Cameron et al., 2007). Weddell seals are also common around the southern end of Hut Point Peninsula (Stirling, 1969) in the MOO's immediate vicinity. From October to December in 2017 and 2018 (when project personnel were present), daily maxima of 5–30 Weddell seals were observed hauled out on the sea ice near crack features emanating from Hut Point, <1 km north of the MOO, with smaller aggregations near the tip of Cape Armitage, 1 km to the south. Weddell seals occasionally hauled out at cracks <100 m from the MOO. No other species of marine mammals were noted during these observations. Following the breeding season (October–December), the seals disperse more widely throughout McMurdo Sound and northward into the Ross Sea (Goetz, 2015) with only 250 individuals estimated to remain throughout the austral winter (Smith, 1965).

During the majority of the project's first year, southern McMurdo Sound was covered with 2–3 m of solid shore-fast sea ice and the water column was essentially isothermal (−1.9 °C; slight upward refraction of sound). In 2017–2018, the natural fast-ice edge was from 30 km (November 2017 and November 2018) to 10 km (March 2018) from the MOO [NASA EOSDIS Worldview2; Fig. 2(A)]. In January 2018, an icebreaker created an open water channel from the ice edge to about 0.5 km north of the MOO and near-surface temperatures rose slightly (maximum −0.4 °C, recorded by the MOO at 21 m in late January 2018) before the channel refroze by late March or early April.

Weddell seals are the only mammals that routinely inhabit and dive beneath the thick, shore-fast sea ice of southern McMurdo Sound (see Sec. III, Results). Other potentially soniferous marine mammals and diving birds may transiently visit the area, but typically only when open water exists in the austral summer (January–April; Kim et al., 2018; Thomas et al., 1987; Thomas and Kuechle, 1982). These most commonly include leopard and crabeater seals (Lobodon carcinophaga), killer and Antarctic minke (Balaenoptera bonaerensis) whales, and Adelie (Pygoscelis adeliae), and emperor (Aptenodytes forsteri) penguins (Cziko, 2020). Nevertheless, aside from the sounds attributed to Weddell seals, only those of killer whales (Wellard et al., 2020) were noted in the year-round recordings and on only about five total days throughout February 2018. Some penguin species may produce brief sounds underwater at ≤7 kHz (Thiebault, 2019). However, the nearest rookery, of Adelie penguins at Cape Royds, is 35 km north of the recording site, and no similar vocalizations were noted in the dataset. Various notothenioid fishes (≤30 cm) were continuously present at the recording site but no sounds could be attributed to them.

Natural and anthropogenic interfering sounds were relatively common throughout the dataset. Identifiable sounds included irregular low-intensity, broad-spectrum clicks and cracks from the sea ice cover, occasional wind noise, a 1.5-s gurgle with components to 200 kHz every 90 s from the CTD's pump, a broad-spectrum mechanical sound for 3 min every 4 h from the camera's cleaning system, low-intensity whines (about 18, 58, 83, and 130 kHz) thought to be from the station seawater pumps (>100 m away within the jetty's well casing), and intermittent noises from tracked-vehicles and helicopters (September–February), SCUBA divers (October–December), and ships (January). Given the overlying ice cover, overall background noise levels from sources other than Weddell seals and the observatory itself were generally very low. Aside from a thin layer of diatoms, neither biofouling nor anchor ice were observed on the hydrophone.

Ultrasonic vocalizations of Weddell seals were identified by browsing archived spectrogram images and watching the real-time spectrogram display at McMurdo Station or remotely over the internet. Signals of interest were further investigated using sound analysis software. In this way, a search set of discrete sounds was compiled from a relatively exhaustive review of an estimated 30% of the 13-month dataset. Archived spectrograms covering at least 2500 h (15 000 images) were visually inspected.

Vocalization types that occurred exclusively when the ship's channel was open (January to early April 2018) were excluded from analyses. As such, novel sounds from killer whales or other species in the nearby open water would not be attributed to Weddell seals. All broad-spectrum click sounds were excluded as many evidently originated from sea ice movements and, lacking predictable repetition rates or frequency characteristics, none could be attributed to the seals. Broad-spectrum “jaw claps” (to >200 kHz) produced by Weddell seals (Thomas and Kuechle, 1982) were excluded since they are not vocalizations per se.

Ultrasonic vocalizations from the search set were assigned to call types based on whether they consistently occurred alone or, for multi-element calls, in series with one or more other sounds in recurrent stereotyped patterns (Moors and Terhune, 2004). Archived spectrogram images from select days throughout the 13-month dataset were then visually browsed in order to collect multiple examples of each call type at levels substantially above the background noise. To attempt to reduce bias toward individual seals, calls were typically chosen for analysis only if separated from their previous occurrences by ≥24 h. Call types and their elements were analyzed for frequency, waveform, and time characteristics in Raven Pro 1.5 (Center for Conservation Bioacoustics, 2014). Analysis settings varied depending on call type and are presented in Table I. For multi-element chirp-based calls, inter-chirp intervals were measured between the beginnings of successive chirps. Durations of individual chirp elements were measured for the time containing 90% of the energy in order to avoid misinterpretation of start and stop times due to echoes or multipath transmission.

TABLE I.

Characteristics of the fundamental frequencies of Weddell seal ultrasonic underwater vocalizations recorded by the MOO in McMurdo Sound, Antarctica. Means ± standard deviation are presented with other listed values in brackets.a

Call typeElement typeNumber analyzedMaximum frequency [maximum] (kHz)Minimum frequency [minimum] (kHz)Center frequency [range] (kHz)Peak amplitude frequency [range] (kHz)Duration [range]
C101b Full call 10 — — — — 16.5 ± 3.5 sc [11.2–21.0] 
 10 41.6 ± 4.7 [49.8] 26.7 ± 0.7 [25.7] 28.5 ± 0.7 [27.5–29.8] 28.0 ± 0.8 [26.8–29.0] 6.6 ± 1.7 msd [4.8–9.2] 
 10 37.4 ± 2.1 [47.0] 20.7 ± 0.4 [19.4] 22.6 ± 1.1 [21.0–25.3] 22.3 ± 1.3 [21.0–25.8] 5.8 ± 1.1 msd [4.0–7.2] 
 10 21.9 ± 1.8 [24.5] 4.0 ± 0.2 [3.8] 5.1 ± 0.2 [5.0–5.5] 4.9 ± 0.1 [4.8–5.0] 128.6 ± 21.5 msd [107.1–179.8] 
C102b Full call 15 — — — — 8.7 ± 0.6 sc [7.8–9.5] 
 15 39.7 ± 2.4 [42.8] 25.5 ± 1.8 [21.4] 29.6 ± 1.3 [27.3–31.3] 28.7 ± 1.8 [25.5–31.5] 5.9 ± 1.3 msd [4.0–8.8] 
 15 34.0 ± 2.3 [40.7] 18.0 ± 0.4 [15.8] 20.4 ± 1.4 [17.8–24.3] 19.3 ± 0.9 [17.5–24.0] 4.5 ± 0.4 msd [3.6–4.8] 
 15 21.3 ± 2.9 [30.0] 4.2 ± 0.2 [3.9] 5.9 ± 0.3 [5.5–6.5] 5.6 ± 0.3 [5.0–6.0] 39.6 ± 10.1 msd [29.6–63.7] 
C103b Full call 12 — — — — 3.2 ± 3.5 sc [1.2–10.7] 
 12 37.3 ± 4.1 [44.9] 24.6 ± 1.9 [22.5] 26.8 ± 1.9 [24.5–30.5] 26.4 ± 1.8 [24.3–30.0] 6.6 ± 1.5 msd [4.8–8.8] 
 12 22.7 ± 2.8 [28.7] 5.4 ± 0.4 [4.7] 7.3 ± 0.8 [6.3–9.0] 6.8 ± 0.3 [6.3–7.3] 5.5 ± 3.2 msd [3.6–12.0] 
 12 42.7 ± 3.3 [48.1] 31.8 ± 1.5 [28.5] 34.1 ± 2.2 [29.3–37.5] 33.8 ± 2.2 [29.0–37.8] 5.5 ± 0.9 msd [4.4–6.8] 
 11.7 ± 2.2 [14.0] 0.2 ± 0.2 [0.04] 0.6 ± 0.3 [0.3–1.0] 0.6 ± 0.4 [0.3–1.0] 66.8 ± 49.5 msd [22.4–110.5] 
U101e Full call 20 — — — — 10.5 ± 4.0 sc [5.2–21.4] 
 20 40.8 ± 1.7 [44.2] 34.4 ± 1.4 [32.5] 36.3 ± 1.5 [34.8–41.0] 36.0 ± 1.7 [33.9–41.5] 6.3 ± 3.0 s [2.7–11.9] 
 20 17.7 ± 3.1 [23.8] 5.1 ± 0.3 [4.6] 6.4 ± 0.5 [5.8–7.4] 5.9 ± 0.5 [5.3–7.4] 4.5 ± 2.6 s [1.9–9.9] 
W101f — 32.0 ± 1.5 [34.7] 28.8 ± 0.3 [28.3] 29.6 ± 0.5 [29.2–30.4] 29.4 ± 0.7 [28.4–30.5] 1.9 ± 0.3 s [1.5–2.2] 
W102f — 31 21.3 ± 0.4 [22.8] 20.3 ± 0.3 [19.6] 20.8 ± 0.3 [20.3–21.3] 20.8 ± 0.3 [20.2–21.3] 7.7 ± 1.7 s [4.8–10.7] 
W103f — 24.9 ± 0.3 [25.3] 19.7 ± 0.1 [19.5] 20.0 ± 0.1 [19.9–20.1] 20.1 ± 0.3 [19.8–20.6] 9.7 ± 2.1 s [6.2–11.3] 
T101f Full call 19 — — — — 50.9 ± 13.9 sc [10.6–75.6] 
 19 26.7 ± 1.2 [28.9] 17.3 ± 0.6 [15.6] 21.2 ± 1.2 [19.4–24.1] 21.0 ± 1.9 [17.7–24.7] 14.3 ± 1.9 s [7.3–15.6] 
 19 30.9 ± 4.4 [36.1] 0.1 ± 0.1 [0.0] 0.7 ± 0.2 [0.5–1.4] 0.7 ± 0.1 [0.5–0.8] 28.3 ± 10.7 s [2.4–52.3] 
T102f — 23 24.4 ± 2.2 [29.5] 9.5 ± 0.3 [9.0] 11.3 ± 0.7 [10.4–13.3] 10.8 ± 1.0 [10.0–14.7] 6.7 ± 2.5 s [3.7–10.1] 
Call typeElement typeNumber analyzedMaximum frequency [maximum] (kHz)Minimum frequency [minimum] (kHz)Center frequency [range] (kHz)Peak amplitude frequency [range] (kHz)Duration [range]
C101b Full call 10 — — — — 16.5 ± 3.5 sc [11.2–21.0] 
 10 41.6 ± 4.7 [49.8] 26.7 ± 0.7 [25.7] 28.5 ± 0.7 [27.5–29.8] 28.0 ± 0.8 [26.8–29.0] 6.6 ± 1.7 msd [4.8–9.2] 
 10 37.4 ± 2.1 [47.0] 20.7 ± 0.4 [19.4] 22.6 ± 1.1 [21.0–25.3] 22.3 ± 1.3 [21.0–25.8] 5.8 ± 1.1 msd [4.0–7.2] 
 10 21.9 ± 1.8 [24.5] 4.0 ± 0.2 [3.8] 5.1 ± 0.2 [5.0–5.5] 4.9 ± 0.1 [4.8–5.0] 128.6 ± 21.5 msd [107.1–179.8] 
C102b Full call 15 — — — — 8.7 ± 0.6 sc [7.8–9.5] 
 15 39.7 ± 2.4 [42.8] 25.5 ± 1.8 [21.4] 29.6 ± 1.3 [27.3–31.3] 28.7 ± 1.8 [25.5–31.5] 5.9 ± 1.3 msd [4.0–8.8] 
 15 34.0 ± 2.3 [40.7] 18.0 ± 0.4 [15.8] 20.4 ± 1.4 [17.8–24.3] 19.3 ± 0.9 [17.5–24.0] 4.5 ± 0.4 msd [3.6–4.8] 
 15 21.3 ± 2.9 [30.0] 4.2 ± 0.2 [3.9] 5.9 ± 0.3 [5.5–6.5] 5.6 ± 0.3 [5.0–6.0] 39.6 ± 10.1 msd [29.6–63.7] 
C103b Full call 12 — — — — 3.2 ± 3.5 sc [1.2–10.7] 
 12 37.3 ± 4.1 [44.9] 24.6 ± 1.9 [22.5] 26.8 ± 1.9 [24.5–30.5] 26.4 ± 1.8 [24.3–30.0] 6.6 ± 1.5 msd [4.8–8.8] 
 12 22.7 ± 2.8 [28.7] 5.4 ± 0.4 [4.7] 7.3 ± 0.8 [6.3–9.0] 6.8 ± 0.3 [6.3–7.3] 5.5 ± 3.2 msd [3.6–12.0] 
 12 42.7 ± 3.3 [48.1] 31.8 ± 1.5 [28.5] 34.1 ± 2.2 [29.3–37.5] 33.8 ± 2.2 [29.0–37.8] 5.5 ± 0.9 msd [4.4–6.8] 
 11.7 ± 2.2 [14.0] 0.2 ± 0.2 [0.04] 0.6 ± 0.3 [0.3–1.0] 0.6 ± 0.4 [0.3–1.0] 66.8 ± 49.5 msd [22.4–110.5] 
U101e Full call 20 — — — — 10.5 ± 4.0 sc [5.2–21.4] 
 20 40.8 ± 1.7 [44.2] 34.4 ± 1.4 [32.5] 36.3 ± 1.5 [34.8–41.0] 36.0 ± 1.7 [33.9–41.5] 6.3 ± 3.0 s [2.7–11.9] 
 20 17.7 ± 3.1 [23.8] 5.1 ± 0.3 [4.6] 6.4 ± 0.5 [5.8–7.4] 5.9 ± 0.5 [5.3–7.4] 4.5 ± 2.6 s [1.9–9.9] 
W101f — 32.0 ± 1.5 [34.7] 28.8 ± 0.3 [28.3] 29.6 ± 0.5 [29.2–30.4] 29.4 ± 0.7 [28.4–30.5] 1.9 ± 0.3 s [1.5–2.2] 
W102f — 31 21.3 ± 0.4 [22.8] 20.3 ± 0.3 [19.6] 20.8 ± 0.3 [20.3–21.3] 20.8 ± 0.3 [20.2–21.3] 7.7 ± 1.7 s [4.8–10.7] 
W103f — 24.9 ± 0.3 [25.3] 19.7 ± 0.1 [19.5] 20.0 ± 0.1 [19.9–20.1] 20.1 ± 0.3 [19.8–20.6] 9.7 ± 2.1 s [6.2–11.3] 
T101f Full call 19 — — — — 50.9 ± 13.9 sc [10.6–75.6] 
 19 26.7 ± 1.2 [28.9] 17.3 ± 0.6 [15.6] 21.2 ± 1.2 [19.4–24.1] 21.0 ± 1.9 [17.7–24.7] 14.3 ± 1.9 s [7.3–15.6] 
 19 30.9 ± 4.4 [36.1] 0.1 ± 0.1 [0.0] 0.7 ± 0.2 [0.5–1.4] 0.7 ± 0.1 [0.5–0.8] 28.3 ± 10.7 s [2.4–52.3] 
T102f — 23 24.4 ± 2.2 [29.5] 9.5 ± 0.3 [9.0] 11.3 ± 0.7 [10.4–13.3] 10.8 ± 1.0 [10.0–14.7] 6.7 ± 2.5 s [3.7–10.1] 
a

All files were 512 kS s−1, 24 bit WAV; only the fundamental frequencies of vocalizations were included in analysis selection bounds.

b

Analyzed with a 2048-point Hann window, 90% overlap, 2048-point DFT sample length = 250 Hz filter bandwidth.

c

For multiple-element calls, the full call duration was measured from the beginning of the first element to the end of the last element.

d

For chirp-type elements only, the duration is the interval containing 90% of energy for ten randomly selected individual elements.

e

Analyzed with a 4096-point Hann window, 90% overlap, 4096-point DFT sample length = 125 Hz filter bandwidth.

f

Analyzed with a 8192-point Hann window, 50% overlap, 8192-point DFT sample length = 62.5 Hz filter bandwidth.

For an initial assessment of whether the usage of ultrasonic call types varied throughout the year, their presence or absence was tabulated by calendar month over the 13-month dataset. Beginning at the start of each month, archived spectrograms were visually inspected until at least one instance of each call type was found or until the end of the month was reached.

The proportional usage of ultrasonic calls was investigated by analyzing a single 24-h period in austral spring (November 20, 2017) and one from near the winter solstice (“midwinter,” June 19, 2018). The sampled days were chosen because they maximized differences in solar illumination and breeding status, vocalizations occurred throughout the entire 24-h period, and vocal activity appeared to be broadly representative of their respective seasons. The spring sample was in the height of the breeding season and characterized by 24 hours of continuous sunlight (sun altitudes from 8° to 32°, always above the horizon). Conversely, the midwinter sample was likely prior to the commencement of major breeding-oriented behaviors (Thomas and Terhune, 2009) and characterized by near absolute darkness (sun altitudes from −11° to −36°, always below the horizon; crescent moon ≤1.8° above the horizon for about 5 h).

In each 24-h sample, all archived spectrograms were visually inspected, counting occurrences of ultrasonic call types that were readily distinguishable (see the example labeled spectrogram in the supplementary material1). Sonic-range vocalizations could not be accurately counted due to their high abundance and frequent overlap in the spring sample. Instead, occurrence of a relatively common and easily identified sonic vocalization was used as a proxy for overall sonic-range vocal activity. This narrowband descending-frequency whistle (from 18 to 12 kHz over about 5 s) has been previously attributed to Weddell seals (Thomas and Kuechle, 1982; see the example in the supplementary material1), and is referred to herein as the “sonic standard call.” Results from one study suggest that seasonal variation in the proportional usage of sonic descending whistles is relatively low (32%–38% of total sonic calls in nonbreeding and breeding seasons respectively; Doiron et al., 2012).

A simultaneous video and audio recording of a Weddell seal producing a repetitive ultrasonic chirp-based call (C102, see Sec. III, Results) in close proximity to the MOO permitted estimation of the source sound pressure levels (SPLs) of its elements. The seal-hydrophone distance was estimated using the apparent size of benthic landmarks on video together with their measured dimensions and distances (by divers with tape measure), the known geometry of the mooring, and the estimated length of an adult seal using a range of plausible values (2.5–3.3 m total length; Thomas and Terhune, 2009). Using hydrophone calibration coefficients, the “inbound power” was measured in Raven Pro for the fundamental and prominent harmonics (25–70, 15–65, and 0–70 kHz bands for the C102-a, C102-b, and C102-c elements, respectively) over the duration of the sounds while excluding obvious echoes. Lower and upper estimate bounds for the source SPLs were computed using the sonar equation to account for the transmission loss (source SPL = received level + transmission loss), assuming spherical spreading [transmission loss = 20 × log10 (distanceseal–hydrophone)] over the range of estimated seal-hydrophone distances (Rogers, 2014). With the seal ≤26 m from the hydrophone (see Sec. III, Results), spherical spreading of sound could be assumed and any frequency-dependent absorption was considered negligible (Au, 1993).

All ultrasonic vocalizations described herein were attributed to Weddell seals with high confidence. For the majority of the dataset, the thick, shore-fast sea ice would generally preclude all other marine mammals and penguins from diving within 10–30 km of the recording site [Fig. 2(A); Kim et al., 2018; Thomas et al., 1987; Thomas and Kuechle, 1982]. This is supported by the results of comprehensive surveys of seals in the greater Erebus Bay area, conducted about six times annually in November through mid-December since 1969 (Rotella, 2018). In each survey during the present study (2017 and 2018), about 1000 hauled-out Weddell seals were documented. By comparison, there were only three total sightings of crabeater seals, and no other pinnipeds or whales were observed on or diving beneath the shore-fast sea ice in areas away from the ice edge (Rotella, 2020). Errant Adelie and emperor penguins occasionally wander over the ice throughout southeastern McMurdo Sound, but they do not typically dive through the isolated holes or cracks in the shore-fast sea ice (Cziko, 2020).

With the exception of killer whale vocalizations, present only intermittently in February 2018 when the ship's channel was open (January–March), the underwater vocalizations of Weddell seals were the only identifiable sounds of nonhuman biological origin in the recordings. The novel ultrasonic vocalizations described herein were both comparatively common and nearly always interspersed with the sonic trills, chirps, buzzes, and chugs that have been previously attributed to Weddell seals (see the example spectrogram in the supplementary material1; Thomas and Kuechle, 1982; Pahl et al., 1997). Finally, the MOO's underwater camera provided regular visual confirmation of Weddell seals producing multiple sonic call types and, in one instance, an ultrasonic call (see below). However, most vocalizing individuals were beyond the visual range of the camera (≤300 m).

Nine recurrent call types were identified that were composed of 17 vocal element types whose fundamental frequencies (F0) were partially or entirely ≥20 kHz (Fig. 3, Table I; recordings are available in the supplementary material1). Individual elements of multi-element calls sometimes occurred alone, although the vast majority occurred within the presented stereotyped calls. Call types were named based on their predominant ultrasonic elements, i.e., chirps (C), U-shaped whistles (U), relatively constant-frequency whistles (W), and FM trills (T), with numbers starting at 101 to avoid confusion with other naming systems. Distinct element types identified within multiple-element calls were designated with lowercase letters. No clipping or other acoustic artifacts were found that could have skewed the results.

FIG. 3.

Spectrograms of Weddell seal ultrasonic underwater vocalizations. These recurrent, stereotyped single- and multiple-element call types were based on chirp (C), U-shaped (U), relatively constant-frequency whistle (W), and FM trill (T) elements having ultrasonic fundamental frequencies (≥20 kHz). Distinct element types of multi-element calls are named with lowercase letters. Some details are shown in Fig. 4. Note the different time and frequency scales between panels. The summary statistics are presented in Fig. 5 and Table I. The presented spectrograms were computed from resampled data (128 kS s−1) using an 8192-point Hann window, 90% overlap with 8192-point discrete Fourier transform (DFT) sample length. Recordings are available in the supplementary material.1

FIG. 3.

Spectrograms of Weddell seal ultrasonic underwater vocalizations. These recurrent, stereotyped single- and multiple-element call types were based on chirp (C), U-shaped (U), relatively constant-frequency whistle (W), and FM trill (T) elements having ultrasonic fundamental frequencies (≥20 kHz). Distinct element types of multi-element calls are named with lowercase letters. Some details are shown in Fig. 4. Note the different time and frequency scales between panels. The summary statistics are presented in Fig. 5 and Table I. The presented spectrograms were computed from resampled data (128 kS s−1) using an 8192-point Hann window, 90% overlap with 8192-point discrete Fourier transform (DFT) sample length. Recordings are available in the supplementary material.1

Close modal

The fundamental frequencies of individual vocal elements spanned the ultrasonic spectrum from 20 to 49.8 kHz (see Figs. 4, 5, and Table I). The highest-frequency fundamental was found at the start of a C101-a chirp element (49.8 kHz), and the element type with the highest mean maximum frequency was the C103-c chirp (42.7 ± 3.3 kHz, mean ± standard deviation). As shown by their center frequencies (the frequency that divides the selection into two frequency intervals of equal energy), the most energy in all elements was focused in the lower half of their fundamental's frequency spectrum. Nevertheless, 11 element types had mean fundamental center frequencies ≥20 kHz with 2 element types >30 kHz (C101-c, U101-a). The fundamental frequencies of six elements were entirely >21 kHz. Element U101-a exhibited the highest mean fundamental center frequency at 36.2 kHz.

FIG. 4.

Some details of the ultrasonic vocalizations presented in Fig. 3 are shown. The various element types with the highest fundamental frequencies are presented as spectrograms (top subpanels) and waveforms (bottom). Only a portion of T101-a and the leading whistle for T101-b are shown. Note the different axis scales between the panels. The presented spectrograms were computed from 512 kS s−1 data using a 256-point Hann window, 90% overlap with 4096-point DFT sample length. The amplitude is presented as raw instrument voltage output (at various scales) after bandpass filtering (15–50 kHz) for clarity.

FIG. 4.

Some details of the ultrasonic vocalizations presented in Fig. 3 are shown. The various element types with the highest fundamental frequencies are presented as spectrograms (top subpanels) and waveforms (bottom). Only a portion of T101-a and the leading whistle for T101-b are shown. Note the different axis scales between the panels. The presented spectrograms were computed from 512 kS s−1 data using a 256-point Hann window, 90% overlap with 4096-point DFT sample length. The amplitude is presented as raw instrument voltage output (at various scales) after bandpass filtering (15–50 kHz) for clarity.

Close modal
FIG. 5.

Characteristics of the fundamental frequencies of Weddell seal ultrasonic underwater call types analyzed in this study. Bars indicate the mean maximum and minimum frequencies of the fundamental, lines show the range of fundamental frequencies, white circles are the mean center frequencies. The ultrasonic range (≥20 kHz) is shown with a white background. n = 4–23 for each element type. Values and analysis parameters are presented in Table I.

FIG. 5.

Characteristics of the fundamental frequencies of Weddell seal ultrasonic underwater call types analyzed in this study. Bars indicate the mean maximum and minimum frequencies of the fundamental, lines show the range of fundamental frequencies, white circles are the mean center frequencies. The ultrasonic range (≥20 kHz) is shown with a white background. n = 4–23 for each element type. Values and analysis parameters are presented in Table I.

Close modal

Call type U101 typically presented as a repetitive series of 5–37 discrete ultrasonic U-shaped whistles (U101-a) between 32.5 and 44.2 kHz (minimum and maximum, respectively), followed by a rapid, sonic buzz (U101-b). Occasionally, the U-shaped elements appeared to be merged into a continuous, irregular sinusoid.

Call types T101 and T102 were based on trills that began at ≥20 kHz, i.e., continuous long-duration FM calls with relatively wide envelopes. T101 included two distinct trill elements that frequently occurred sequentially and only in the presented order, although element type T101-b also occurred alone. Element type T101-a maintained relatively constant frequency contours over its duration with most energy ≥20 kHz and reaching 28.9 kHz. A lower-frequency variant of this element (≤22 kHz) was presented by Russell et al. (2016). A low-frequency trill element often occurred between T101-a and T101-b (visible at 21 s in Fig. 3), although its usage was sporadic and it was not characterized. A portion of T101-b (≤12.8 kHz) appears to have been previously described as call type T6 by Thomas and Kuechle (1982). The recordings herein now show that this element begins as a somewhat variable descending narrowband ultrasonic whistle (≤36.1 kHz; Fig. 4) before transitioning to a trill whose frequency envelope descends into the sonic range as the amplitude increases. A similar leading whistle also characterized call type T102, whose single element occurred both independently and in a call similar to C103, where it replaced chirp element C103-b.

Multiple-element chirp-based calls C101, C102, and C103 (Figs. 3 and 4) recurred regularly in the dataset. Ultrasonic chirps initiated with fundamental frequencies ranging from 21.3 to 44.7 kHz (mean maximums; Fig. 5, Table I), followed by rapid downward linear or exponential FM sweeps. Chirp fundamentals descended at 1.2–2.0 kHz/ms (46–192 octaves/s, minimum and maximum, excluding the lower-frequency terminal elements) with 90% of the energy contained within 3.6–9.2 ms (Table I).

Call types C101 and C102 each began with a unique ultrasonic chirp (C101-a, C102-a) at the highest frequencies of the call, followed by a series of 5–29 similar fully or partially ultrasonic chirps (C101-b, C102-b) at predictable intervals and somewhat lower frequency contours, and terminated with the lowest-frequency chirp (C101-c, C102-c). These two call types segregated based on small but consistent differences in the frequency contours of their elements (Figs. 3–5 and Table I) and by the relatively stereotyped progression of their inter-chirp time intervals (ICIs; Fig. 6). Conversely, the ICIs of call type C103 were rather variable, having a typically short first ICI (<1 s), and longer ICIs thereafter (1–10 s). A fourth chirp-based call type occurred infrequently in the dataset and was not analyzed. It was similar to C101 and C102 but with fewer elements and seemingly consistent but much longer ICIs (8–10 s; visible in supplementary Fig. 11). Calls resembling those presented by Schevill and Watkins (1971) were not found. No calls were observed to terminate with rapidly decreasing ICIs akin to the “terminal buzz” commonly referenced in the echolocation literature (e.g., DeRuiter et al., 2009).

FIG. 6.

Repetitive ultrasonic chirp-based call types (C101 and C102) segregated based on the stereotyped progression of their inter-chirp time intervals (ICIs).The ICI was measured as the time interval between the onset of successive chirp elements within a series of chirps within an individual call. Circles mark the time interval between the first and second chirps (the start of the call) with subsequent chirps in each series shown by connected lines. For clarity, given the characteristics of the calls, the ICI number is referenced to the final ICI (0, the end of the call). Some data points are hidden by overlap.

FIG. 6.

Repetitive ultrasonic chirp-based call types (C101 and C102) segregated based on the stereotyped progression of their inter-chirp time intervals (ICIs).The ICI was measured as the time interval between the onset of successive chirp elements within a series of chirps within an individual call. Circles mark the time interval between the first and second chirps (the start of the call) with subsequent chirps in each series shown by connected lines. For clarity, given the characteristics of the calls, the ICI number is referenced to the final ICI (0, the end of the call). Some data points are hidden by overlap.

Close modal

An example of call type C102 was recorded simultaneously with underwater video observation of the source individual (likely an 18-year-old male based on contemporaneous surface sightings, yellow tag number 9410; Rotella, 2018). The vocalizing seal was estimated to be between 18 and 26 m from the hydrophone and facing about 90° off-axis (see the video in the supplementary material1). Movements of the seal's head, throat, and chest area coincided with the emissions of individual chirps, and no air was observed to escape from the mouth or nostrils. Estimated source SPLs were lower for the ultrasonic chirps (from 135 to 152.0 dB re 1 μPa-m for C102-a and C102-b) than for the terminal sonic chirp (154–158 dB re 1 μPa-m; Table II). Equivalent continuous sound level (Leq) values for all elements were essentially equal to inbound power measurements, and background noise levels in the bandwidths used to measure the sounds were <91 dB re 1 μPa.

TABLE II.

Estimated source sound pressure levels (SPLs) of chirps from a single type C102 call, derived from a simultaneous underwater video and audio recording by the MOO (estimated seal-hydrophone distance = 18–26 m).a

Element typeCenter frequency (kHz)Durationb (90%, ms)Source SPLc (dB re 1 μPa-m)
C102-a (initial chirp; n = 1) 29.5 6.0 137 (135–138) 
C102-b (repetitive chirps; n = 26) 19.7 ± 0.9d 5.7 ± 0.7d 144 ± 1d (142–152) 
C102-c (terminal chirp; n = 1) 6.3 37.2 156 (154–158) 
Element typeCenter frequency (kHz)Durationb (90%, ms)Source SPLc (dB re 1 μPa-m)
C102-a (initial chirp; n = 1) 29.5 6.0 137 (135–138) 
C102-b (repetitive chirps; n = 26) 19.7 ± 0.9d 5.7 ± 0.7d 144 ± 1d (142–152) 
C102-c (terminal chirp; n = 1) 6.3 37.2 156 (154–158) 
a

Selection bounds included the fundamental and prominent harmonics, excluding obvious echoes (see Sec. II, Methods). Analyzing filter bandwidth 250 Hz (2048-point DFT length, 512 kS s−1 data).

b

Time containing 90% of the energy for individual elements.

c

At median estimated seal-hydrophone distance; range of source SPL values for individual chirps given full range of distance uncertainty in parentheses; calculated as inbound power plus estimated transmission loss; the seal was facing about 90° off-axis.

d

Means ± standard deviation.

The ultrasonic calls of Weddell seals were common almost year-round. Based on an assessment of presence/absence only, 8 out of the 9 ultrasonic call types were found at least once in ≥11 of the 13 analyzed months [Fig. 7(A)]. None were recorded in February. Overall, the prevalence of ultrasonic and sonic vocalizations appeared to be highly correlated. Both were most common during the austral spring breeding season (October–December), comparatively less frequent at other times, and rare or absent for extended periods in austral summer (January–March; data not shown). A similar pattern has been previously reported for sonic-range vocalizations at other locations (Green and Burton, 1988; Thomas et al., 1988; van Opzeeland et al., 2010). It likely results from seasonal changes in the abundance of seals at the recording site (Goetz, 2015; Smith 1965) and/or their propensity to vocalize. Weddell seals may also reduce their vocal activity in summer to avoid detection by potential predators (e.g., killer whales) in nearby open water (Thomas et al., 1987).

FIG. 7.

Monthly occurrence and seasonal variation in ultrasonic calling. (A) Presence (black squares, ≥1 occurrence) or absence (white circles) of ultrasonic call types in each of 13 calendar months. Gray shading demarcates the breeding seasons. (B) Vocal activity over a single 24-h period in austral spring and one in midwinter. The relative prevalence of the total ultrasonic calls compared to the sonic standard call was approximately constant in the two samples, although the vocal activity for each was about threefold lower in midwinter. C103 was excluded from the calculations because its detection was unreliable in the midwinter sample (asterisks). The sonic standard call (a descending whistle) was used as a proxy for overall sonic vocal activity. (C) Proportional ultrasonic call type usage in the spring and midwinter samples. Bar heights for each call depict their percentage of the total ultrasonic calls in each 24-h period, excluding counts of C103 (hatched bar, asterisks; not counted in midwinter). Four disparate call types occurred at similarly high proportions in spring, whereas the two similar repetitive ultrasonic chirp-based calls dominated in midwinter. The actual call counts are presented above the bars in (B) and (C).

FIG. 7.

Monthly occurrence and seasonal variation in ultrasonic calling. (A) Presence (black squares, ≥1 occurrence) or absence (white circles) of ultrasonic call types in each of 13 calendar months. Gray shading demarcates the breeding seasons. (B) Vocal activity over a single 24-h period in austral spring and one in midwinter. The relative prevalence of the total ultrasonic calls compared to the sonic standard call was approximately constant in the two samples, although the vocal activity for each was about threefold lower in midwinter. C103 was excluded from the calculations because its detection was unreliable in the midwinter sample (asterisks). The sonic standard call (a descending whistle) was used as a proxy for overall sonic vocal activity. (C) Proportional ultrasonic call type usage in the spring and midwinter samples. Bar heights for each call depict their percentage of the total ultrasonic calls in each 24-h period, excluding counts of C103 (hatched bar, asterisks; not counted in midwinter). Four disparate call types occurred at similarly high proportions in spring, whereas the two similar repetitive ultrasonic chirp-based calls dominated in midwinter. The actual call counts are presented above the bars in (B) and (C).

Close modal

Seasonal variation in ultrasonic call activity and proportional call type usage was assessed by counting calls over a single 24-h period in austral spring (November 20, 2017; 24-h sunlight, breeding season) and one near the winter solstice (June 19, 2018, midwinter; 24-h darkness, nonbreeding). Detection of call type C103 was unreliable in the midwinter sample because of its visual similarity to the prevalent cracking sounds from the sea ice. Thus, it was not counted in midwinter and was excluded from comparative analyses. The sonic standard call was taken as a proxy for the total sonic vocal activity in both samples (see Sec. II, Methods).

Using this methodology, the total ultrasonic vocal activity was found to be 2.8-fold lower in midwinter compared to spring [299 and 848 total ultrasonic calls in 24 h, respectively, both excluding counts of C103; Fig. 7(B)]. The midwinter decrease in total ultrasonic calling was approximately matched by the decrease in occurrences of the sonic standard call (3.4-fold). This may signify that the seals' relative use of ultrasonic vs sonic vocalization remains relatively constant year-round.

The proportional usage of the individual ultrasonic call types varied between the two sampled days [Fig. 7(C)]. In the spring sample, four disparate call types (C103, W102, T101, and T102) were most prevalent. Each accounted for between 19% and 27% of the total ultrasonic calls (full range of proportional usage, both including and excluding counts of C103; each call averaging 8.5–9.5 occurrences per hour). Conversely, the two similar repetitive ultrasonic chirp-based calls (C101 and C102) were dominant in the midwinter sample where, together, they accounted for 62% of all ultrasonic calls (averages of 3.1 and 4.6 occurrences per hour, respectively).

Vocalizations with both sonic and ultrasonic fundamentals exhibited harmonics with energy regularly present above background levels to over 100 kHz and occasionally to over 200 kHz, especially when received with high signal-to-noise ratios (SNRs ≥40 dB, as measured in the same 1/3 octave band as the fundamental). Some examples are presented in Fig. 8. No clipping of high-intensity sounds was observed, i.e., the presented harmonics are not artifacts. No emphasis on higher-order harmonics was noted for any vocalizations, rather the fundamental frequency always contained the most energy. When received at these high SNRs, ultrasonic chirps were accompanied by coincident very low intensity sounds at frequencies below the fundamental (e.g., C101-b and C102-b in Fig. 3 and at 15–25 ms in C103-c in Fig. 8).

FIG. 8.

Harmonics of sonic and ultrasonic chirp, whistle, and trill elements extended to >200 kHz when received with high SNRs. To illustrate this, the entire recorded harmonic series for portions of three diverse element types is presented. Power spectra (left panels, 2 kHz resolution), computed for the time segment between the dashed lines in the spectrograms (right panels), are referenced to raw instrument voltage. The fundamental always contained the most energy. Subharmonics below the fundamental were not evident. In these examples, SNRs exceeded 60 dB as measured in the same 1/3 octave band as the fundamental. The presented spectrograms were computed from 512 kS s−1 data using a 1024-point Hann window, 90% overlap with a 2048-point DFT sample length.

FIG. 8.

Harmonics of sonic and ultrasonic chirp, whistle, and trill elements extended to >200 kHz when received with high SNRs. To illustrate this, the entire recorded harmonic series for portions of three diverse element types is presented. Power spectra (left panels, 2 kHz resolution), computed for the time segment between the dashed lines in the spectrograms (right panels), are referenced to raw instrument voltage. The fundamental always contained the most energy. Subharmonics below the fundamental were not evident. In these examples, SNRs exceeded 60 dB as measured in the same 1/3 octave band as the fundamental. The presented spectrograms were computed from 512 kS s−1 data using a 1024-point Hann window, 90% overlap with a 2048-point DFT sample length.

Close modal

Despite years of acoustic studies on Weddell seals throughout the Antarctic, this study is the first documentation of their relatively extensive and diverse ultrasonic repertoire. With fundamental frequencies reaching nearly 50 kHz, Weddell seals now appear to be rivaled only by killer whales (75 kHz; Samarra et al., 2010) and possibly leopard seals (164 kHz; Awbrey et al., 2004; Thomas and Awbrey, 1983; if validated, see the Introduction) for the highest frequencies of tonal vocalizations produced by aquatic mammals. In considering the presented ultrasonic call types, the current findings increase the known Weddell seal vocal repertoire by nine call types. Adding these to the accounting by Terhune (2019) increases the total size of the species' known vocal repertoire to 59 call types of which 17% have elements with ultrasonic center frequencies (10 of 59, including chirps described by Schevill and Watkins, 1971). From the previously reported lowest-frequency fundamentals (32 Hz; Terhune, 2019) to the highest-frequency fundamental reported herein (49.8 kHz), Weddell seal vocalizations span more than ten octaves.

While the Weddell seals' routine use of higher frequencies was unknown, the time-frequency contour shapes of these ultrasonic call types have been previously described for calls at sonic frequencies (Doiron et al., 2012; Pahl et al., 1997; Thomas and Kuechle, 1982). Similarly, the stereotyped repetition of similar elements within calls (Moors and Terhune, 2004) and mixed-element calls (Terhune and Dell'Apa, 2006) also occurs in the sonic range. The mixing of ultrasonic and sonic elements in stereotyped multi-element calls suggests that some sonic elements previously thought to occur individually may have belonged to more complex calls.

It is likely that similar vocalizations were missed in previous recordings from around Antarctica owing primarily to temporal biases and/or limitations of recording equipment (e.g., Fig. 1); however, other possibilities exist. Weddell seals have geographically distinct repertoires on various scales (e.g., Thomas and Stirling, 1983), thus, ultrasonic call usage could be unique to the McMurdo Sound population. This could explain why most other researchers did not note ultrasonic components despite some ability to record at the necessary frequencies. It is also conceivable that other recording sites were more influenced by environmental or biological sounds (e.g., Klinck et al., 2008) that precluded the detection of ultrasonic vocalizations or their attribution to seals. It is implausible that ultrasonic vocalization constitutes a behavior learned by the local population since the earlier recordings in McMurdo Sound (e.g., Thomas and Kuechle, 1982) given that Schevill and Watkins (1971) previously recorded a sequence of ≤30 kHz chirps in the area.

It is relevant to question whether the ultrasonic vocalizations presented herein are the product of a single individual (or a few) with an atypical repertoire or rather represent a more general feature of the species as a whole. The former case is unlikely given the temporal distribution of calls over the lengthy dataset [Fig. 7(A)], the large local population (Ainley et al., 2015), the diving range of the seals (5 km; Thomas and Terhune, 2009), and that overlapping ultrasonic calls were occasionally recorded (data not shown). The present recordings may be biased toward certain individuals over shorter time periods (hours to weeks), and the trill-type vocalizations may be specific to males (Oetelaar et al., 2003; Thomas and Kuechle, 1982). On the other hand, one of the present authors (J.M.T.) recorded trills that appeared to commence above 22 kHz (the upper FR of the equipment) at Davis Station in 1997 (>5000 km from McMurdo Sound). This suggests that ultrasonic vocalizations may be a common feature of the Weddell seal repertoire throughout their distribution.

Exactly how seals produce their vocalizations has been the subject of some speculation. Sonic Weddell seal underwater vocalizations occur with the mouth and nostrils closed such that no air escapes, and they may be accompanied by pulsing or bobbing of the head, neck, or torso (Oetelaar et al., 2003; Schevill and Watkins, 1971). The video evidence (presented in the supplementary material1) indicates that the same is likely true for ultrasonic vocalizations. Seals, including Weddells, are, thus, thought to vocalize by vibrating vocal folds and resonating pressure waves in contained air spaces, as in between the larynx and the trachea (Piérard, 1969). In a response-driven system such as this, the emitted frequency would be at least partially controlled by the properties of the air chambers that the vibrations excite (Bradbury and Vehrencamp, 1998; Moors and Terhune, 2005). That is, higher frequencies should arise from the compression of air spaces with increasing hydrostatic pressure during dives (Falke et al., 1985; Kooyman et al., 1970).

However, for harp (Pagophilus groenlandicus) and Weddell seal audible vocalizations, Moors and Terhune (2005) found no relationship between the vocalization frequency and the depth of emission up to 90 m, suggesting that the characteristics of the air spaces have minimal influence on the frequencies of the emitted sounds. Likewise, the ultrasonic elements presented herein likely do not represent sonic calls shifted to higher frequencies because they were produced at great depth, given especially the presented video evidence,1 the local bathymetry of the recording site [Fig. 2(B)], and previous recordings of ultrasonic chirps produced near the surface (Schevill and Watkins, 1971).

Taken together, the fundamental frequencies of the ultrasonic element types spanned the full range from 20 to 50 kHz (Fig. 5). Weddell seals do not, therefore, appear to be limited to the use of a discrete set of ultrasonic frequencies as might occur in response-driven systems with specific resonances due to the geometry of the vocal tract (Au and Suthers, 2014). The coincident emission of low-intensity sounds below the frequencies of highest intensity (Figs. 3 and 8) could possibly indicate that sonic-range fundamentals (i.e., subharmonics) are selectively filtered in the vocal tract of the seals, allowing predominantly ultrasonic overtones to escape (e.g., Hartley and Suthers, 1988). However, the spectra of the sounds do not support this conclusion (Fig. 8). It is more likely that the low-frequency sounds arise from physical movements of the body or displacement of air internally during vocalization. At this point, the most parsimonious explanation for the production of ultrasonic vocalizations in Weddell seals is that, as for those in the sonic range, they are primarily created by vibrations of the vocal folds themselves, i.e., they are source driven.

It appears that the ultrasonic vocalizations of Weddell seals are produced at lower amplitudes than their sonic vocalizations given the range of estimates for the elements of a single C102 call (135–152 dB re 1 μPa-m for C102-a and C102-b vs 153–193 dB re 1 μPa-m for previously described sonic vocalizations; Table II; Thomas and Kuechle, 1982). For calls that contained both ultrasonic and sonic fundamental frequencies, the ultrasonic components were always received at lower amplitudes than those in the sonic range. However, the presented estimates of source SPLs remain only a minimum bound given that the vocalizing seal was oriented approximately 90° away from the hydrophone and the greatest sound pressure is likely to emanate in a more-or-less wide cone (possibly to 90° wide), angled somewhat downward from the throat area (Schevill and Watkins, 1971). For harp seal sonic vocalizations, source SPLs apparently vary by up to 12 dB around the animal (Rossong and Terhune, 2009), thus, it is possible that on-axis source SPLs for the ultrasonic chirps of Weddell seals could reach to over 164 dB re 1 μPa-m.

It is likely that the seals can perceive at least the fundamental frequencies of all of their ultrasonic vocalizations presented herein. Phocids as a group have an overall best underwater hearing range (+20 dB from the lowest threshold) of about 125 Hz–50 kHz with maximum sensitivity around 12 kHz (Southall et al., 2019). While the upper frequency limit of Weddell seal hearing has not been tested, it is unlikely that the seals would be able to produce stereotyped vocalizations to 50 kHz that they could not hear themselves. Although harmonics of both sonic and ultrasonic elements were detected to over 200 kHz (Fig. 8), the Weddell seals' auditory sensitivity is likely poor at >60 kHz given data for other phocids (Cunningham and Reichmuth, 2016; Kastelein et al., 2009). Thus, the higher-order harmonics are probably undetectable to them.

Most known Weddell seal vocalizations are expected to be produced for intraspecific communication purposes (e.g., Russell et al., 2016; Thomas et al., 1983), and the same may be true for those in the ultrasonic range. Schevill and Watkins (1971) noted that the ultrasonic chirps they recorded were used by seals travelling between access holes, perhaps a warning of their impending arrival to conspecifics at the distant site. Similarly, the video in the supplementary material1 shows a seal producing an ultrasonic chirp-based call C102 at <20 m depth, immediately after leaving a breathing hole and with no other seals in view (visible range of 200–300 m). However, in the preliminary analysis herein, the proportional usage of the ultrasonic call types varied substantially between periods of sunlight/breeding and darkness/nonbreeding [Fig. 7(C)]. This suggests that individual call types may be associated with specific behaviors that change seasonally.

Sound production over a larger frequency range could provide various benefits. Given that higher frequencies attenuate more rapidly with distance compared to lower frequencies (Au, 1993), the use of the ultrasonics could restrict communications to conspecifics at short range while also avoiding detection by distant predators such as killer whales (Rogers, 2014). At present, these suppositions remain poorly supported since most ultrasonic calls included lower-frequency components and were also generally interspersed with sonic vocalizations.

The Weddell seals' use of ultrasonic frequencies could also serve as an additional communication channel in areas where the lower frequencies are cluttered with the vocalizations of other species or conspecifics. Moreover, because ultrasonic emissions typically have a narrower beam than those at lower frequencies (Sales and Pye, 1974), their use could possibly allow communicative signals to be emitted with better directionality. The relative extent to which higher frequencies and overtones are attenuated in received calls could also provide another metric besides intensity for determining the distance or orientation of vocalizing conspecifics (Wartzok et al., 1992).

Previous authors have asserted that pinnipeds do not echolocate, using a definition of the term associated only with food capture and the high-precision biosonar of toothed whales and bats (Schusterman et al., 2000). Weddell seals may, however, possess the characteristics necessary for at least a rudimentary form of echo-based acoustic spatial perception (for which no standardized gradational terminology seems to exist). As with other seal species, they likely have relatively sensitive hearing over a wide frequency range (Southall et al., 2019) and can localize sound sources (Terhune, 1974; Wartzok et al., 1992), and they are now known to regularly produce repetitive, short-duration ultrasonic vocalizations (this study; Schevill and Watkins, 1971). Any communicative functions of ultrasonic calls would not exclude the possibility that echo and reverberation patterns also provide some information about the surroundings. However, there remain substantial differences between these seals and animals with an acute echolocating ability: seals do not possess any specialized structures for directional emission or reception of sounds (Schusterman et al., 2000; Vater and Kössl, 2004) and their target detection range would be limited by the lower amplitudes of their vocalizations (>40 dB lower than the maximum of toothed whale echolocation clicks; Au, 1993). Moreover, the durations of the shortest ultrasonic chirps presented herein are still comparatively long (≥3.6 ms), resulting in a ranging error of ≥5.4 m given the speed of sound in seawater (≅1500 m/s).

Nevertheless, the echoes of the ultrasonic vocalizations emitted by Weddell seals could conceivably provide finer-scale information on obstacles, the sea/ice surface, or the water depth compared to those at lower frequencies. They might, therefore, facilitate orientation and navigation especially in dark or limited-visibility conditions under the sea ice where egress points are limited. Notably, the proportional usage of repetitive ultrasonic chirp-based calls (C101 and C102) appeared to be higher in midwinter darkness compared to in the spring [Fig. 7(C)]. Although only a preliminary finding, this might lend support to their use in acoustic spatial perception. Additional studies are needed to determine to what extent Weddell seals use their own sounds to navigate and find prey in nature.

Given that Weddell seals have long been the subject of acoustic research, the discovery that they routinely use a relatively diverse repertoire of ultrasonic vocalizations reinforces the need for broad-bandwidth, long-term passive acoustic monitoring. At present, it is unclear whether ultrasonic emissions could comprise an important facet of the underwater vocalizations of other seals. As for Weddell seals, many previous studies of other species used recording equipment or analyses with relatively low upper FR. It is also possible that infrequently used or low-intensity ultrasonic vocalizations were simply missed or attributed to other species. Given the evolution of recording and analysis technologies, future researchers might consider replicating previous studies to assess whether other seals also produce ultrasonic vocalizations. Indeed, recording at higher frequencies could contribute to a better understanding of the range of ways that marine mammals employ sounds to enable their survival in a complex underwater environment (e.g., Tyack, 1997).

We thank the personnel at McMurdo Station, including, especially, laboratory, construction and IT staff, and divers R. Robbins and S. Rupp. We thank T. Mendelow (View into The Blue) and Ocean Sonics for technical assistance and J. Hildebrand and S. Wiggins for helpful advice. H. Kaiser, K. Meister, W. Turner, and A. L. DeVries provided assistance in Antarctica. A. M. Wood provided laboratory space, and S. Nash contributed excellent administrative support at the University of Oregon. Figure 2 was created with assistance from the Polar Geospatial Center under U.S. National Science Foundation Award No. OPP 1559691. J.M.T. acknowledges the support of the University of New Brunswick. This work was primarily supported by Award No. OPP 1644196 to P.A.C. and Arthur L. DeVries. The authors thank I. Charrier, the editor, and an anonymous reviewer whose comments helped to improve this manuscript.

1

See supplementary material at https://www.scitation.org/doi/suppl/10.1121/10.0002867 for details of previous Weddell seal recordings cited in Fig. 1, an example of the archived spectrogram images used for assessing call prevalence, spectrograms of the sonic standard call, audio files of the presented calls in full resolution and modified human-audible versions, and an underwater video of a vocalizing seal from which the chirp source SPLs were derived.

2

See https://worldview.earthdata.nasa.gov (Last viewed October 27, 2020).

1.
Ainley
,
D. G.
,
Larue
,
M. A.
,
Stirling
,
I.
,
Stammerjohn
,
S.
, and
Siniff
,
D. B.
(
2015
). “
An apparent population decrease, or change in distribution, of Weddell seals along the Victoria Land coast
,”
Mar. Mammal Sci.
31
(
4
),
1338
1361
.
2.
Au
,
W. W. L.
(
1993
).
The Sonar of Dolphins
(
Springer
,
New York)
, Chaps. 1, 5, 7, and 11.
3.
Au
,
W. W. L.
, and
Suthers
,
R. A.
(
2014
). “
Production of biosonar signals: Structure and form
,” in
Biosonar
, edited by
A.
Surlykke
,
P. E.
Nachtigall
,
R. R.
Fay
, and
A. N.
Popper
(
Springer
,
New York
), pp.
61
105
.
4.
Awbrey
,
F. T.
,
Thomas
,
J. A.
, and
Evans
,
W. E.
(
2004
). “
Ultrasonic underwater sounds from a captive leopard seal (Hydrurga leptonyx)
,” in
Echolocation in Bats and Dolphins
, edited by
C. F.
Moss
and
M.
Vater
(
University of Chicago Press
,
Chicago, IL)
, pp.
535
541
.
5.
Bradbury
,
J. W.
, and
Vehrencamp
,
S. L.
(
1998
).
Principles of Animal Communication
(
Sinauer
,
Sunderland, MA
), pp.
75
112
.
6.
Cameron
,
M. F.
,
Siniff
,
D. B.
,
Proffitt
,
K. M.
, and
Garrott
,
R. A.
(
2007
). “
Site fidelity of Weddell seals: The effects of sex and age
,”
Antarct. Sci.
19
(
2
),
149
155
.
7.
Center for Conservation Bioacoustics
(
2014
). “
Raven Pro: Interactive sound analysis software (version 1.5)
,” available at http://ravensoundsoftware.com/ (Last viewed October 27, 2020).
8.
Cunningham
,
K. A.
, and
Reichmuth
,
C.
(
2016
). “
High-frequency hearing in seals and sea lions
,”
Hear. Res.
331
,
83
91
.
9.
Cziko
,
P. A.
(
2020
). (personal observation).
10.
Davey
,
F.
, and
Nitsche
,
F. O.
(
2013
). “
Ross Sea bathymetry grid (2005) based on Fred Davey's bathymetry map (2004)
,” Interdisciplinary Earth Data Alliance, available at http://get.iedadata.org/doi/100405 (Last viewed October 27, 2020).
11.
DeRuiter
,
S. L.
,
Bahr
,
A.
,
Blanchet
,
M. A.
,
Hansen
,
S. F.
,
Kristensen
,
J. H.
,
Madsen
,
P. T.
,
Tyack
,
P. L.
, and
Wahlberg
,
M.
(
2009
). “
Acoustic behaviour of echolocating porpoises during prey capture
,”
J. Exp. Biol.
212
(
19
),
3100
3107
.
12.
Doiron
,
E. E.
,
Rouget
,
P. A.
, and
Terhune
,
J. M.
(
2012
). “
Proportional underwater call type usage by Weddell seals (Leptonychotes weddellii) in breeding and nonbreeding situations
,”
Can. J. Zool.
90
(
2
),
237
247
.
13.
Erbe
,
C.
,
Dunlop
,
R.
,
Jenner
,
K. C. S.
,
Jenner
,
M.-N.
,
McCauley
,
R. D.
,
Parnum
,
I.
,
Parsons
,
M.
,
Rogers
,
T.
, and
Salgado-Kent
,
C.
(
2017
). “
Review of underwater and in-air sounds emitted by Australian and Antarctic marine mammals
,”
Acoust. Aust.
45
(
2
),
179
241
.
14.
Evans
,
W. E.
,
Thomas
,
J. A.
, and
Davis
,
R. W.
(
2004
). “
Vocalizations from Weddell seals (Leptonychotes weddellii) during diving and foraging
,” in
Echolocation in Bats and Dolphins
, edited by
J. A.
Thomas
,
C. F.
Moss
, and
M.
Vatek
(
University of Chicago Press
,
Chicago, IL
), pp.
541
547
.
15.
Falke
,
K. J.
,
Hill
,
R. D.
,
Qvist
,
J.
,
Schneider
,
R. C.
,
Guppy
,
M.
,
Liggins
,
G. C.
,
Hochachka
,
P. W.
,
Elliott
,
R. E.
, and
Zapol
,
W. M.
(
1985
). “
Seal lungs collapse during free diving: Evidence from arterial nitrogen tensions
,”
Science.
229
(
4713
),
556
558
.
16.
Goetz
,
K. T.
(
2015
). “
Movement, habitat, and foraging behavior of Weddell seals (Leptonychotes weddellii) in the Eestern Ross Sea
,” Antarctica (Ph.D. dissertation,
University of California
, Santa Cruz), available at https://escholarship.org/uc/item/0jx2107r (Last viewed October 27, 2020).
17.
Green
,
K.
, and
Burton
,
H. R.
(
1988
). “
Annual and diurnal variations in the underwater vocalizations of Weddell seals
,”
Polar Biol.
8
,
161
164
.
18.
Hartley
,
D. J.
, and
Suthers
,
R. A.
(
1988
). “
The acoustics of the vocal tract in the horseshoe bat, Rhinolophus hildebrandti
,”
J. Acoust. Soc. Am.
84
(
4
),
1201
1213
.
19.
Kastelein
,
R. A.
,
Wensveen
,
P. J.
,
Hoek
,
L.
,
Verboom
,
W. C.
, and
Terhune
,
J. M.
(
2009
). “
Underwater detection of tonal signals between 0.125 and 100 kHz by harbor seals (Phoca vitulina
),”
J. Acoust. Soc. Am.
125
(
2
),
1222
1229
.
20.
Kim
,
S.
,
Saenz
,
B.
,
Scanniello
,
J.
,
Daly
,
K.
, and
Ainley
,
D.
(
2018
). “
Local climatology of fast ice in McMurdo Sound, Antarctica
,”
Antarct. Sci.
30
(
2
),
125
142
.
21.
Klinck
,
H.
,
Kindermann
,
L.
, and
Boebel
,
O.
(
2008
). “
Detection of leopard seal (Hydrurga leptonyx) vocalizations using the Envelope-Spectrogram Technique (TEST) in combination with a Hidden Markov model
,”
Can. Acoust.
36
(
1
),
118
124
.
22.
Klinck
,
H.
,
Kindermann
,
L.
, and
Boebel
,
O.
(
2016
). “
PALAOA: The perennial acoustic observatory in the Antarctic Ocean—Real-time eavesdropping on the Antarctic underwater soundscape
,” in
Listening in the Ocean
, edited by
W. W. L.
Au
and
M. O.
Lammers
(
Springer
,
New York
), pp.
207
219
.
23.
Kooyman
,
G. L.
,
Hammond
,
D. D.
, and
Schroeder
,
J. P.
(
1970
). “
Bronchograms and tracheograms of seals under pressure
,”
Science.
169
(
3940
),
82
84
.
24.
Lammers
,
M. O.
,
Au
,
W. W. L.
, and
Herzing
,
D. L.
(
2003
). “
The broadband social acoustic signaling behavior of spinner and spotted dolphins
,”
J. Acoust. Soc. Am.
114
(
3
),
1629
1639
.
25.
Moors
,
H. B.
, and
Terhune
,
J. M.
(
2004
). “
Repetition patterns in Weddell seal (Leptonychotes weddellii) underwater multiple element calls
,”
J. Acoust. Soc. Am.
116
(
2
),
1261
1270
.
26.
Moors
,
H. B.
, and
Terhune
,
J. M.
(
2005
). “
Calling depth and time and frequency attributes of harp (Pagophilus groenlandicus) and Weddell (Leptonychotes weddellii) seal underwater vocalizations
,”
Can. J. Zool.
83
(
11
),
1438
1452
.
27.
Oetelaar
,
M. L.
,
Terhune
,
J. M.
, and
Burton
,
H. R.
(
2003
). “
Can the sex of a Weddell seal (Leptonychotes weddellii) be identified by its surface call?
,”
Aquat. Mamm.
29
(
2
),
261
267
.
28.
Pahl
,
B. C.
,
Terhune
,
J. M.
, and
Burton
,
H. R.
(
1997
). “
Repertoire and geographic variation in underwater vocalisations of Weddell seals (Leptonychotes weddellii, Pinnipedia: Phocidae) at the Vestfold Hills, Antarctica
,”
Aust. J. Zool.
45
(
2
),
171
187
.
29.
Piérard
,
J.
(
1969
). “
Le larynx du phoque de Weddell (Leptonychotes weddelli, Lesson, 1826)” (“The larynx of the Weddell seal”)
,
Can. J. Zool.
47
(
1
),
77
87
.
30.
Rasmussen
,
M. H.
, and
Miller
,
L. A.
(
2002
). “
Whistles and clicks from white-beaked dolphins, Lagenorhynchus albirostris, recorded in Faxaflói Bay, Iceland
,”
Aquat. Mamm.
28
(
1
),
78
89
.
31.
Reeves
,
R. R.
,
Stewart
,
B. S.
,
Clapham
,
P. J.
, and
Powell
,
J. A.
(
2002
). “
Weddell seal
,” in
Guide to Marine Mammals of the World
(
Knopf
,
New York)
, pp.
166
169
.
32.
Rogers
,
T. L.
(
2014
). “
Source levels of the underwater calls of a male leopard seal
,”
J. Acoust. Soc. Am.
136
(
4
),
1495
1498
.
33.
Rossong
,
M. A.
, and
Terhune
,
J. M.
(
2009
). “
Source levels and communication-range models for harp seal (Pagophilus groenlandicus) underwater calls in the Gulf of St. Lawrence, Canada
,”
Can. J. Zool.
87
(
6
),
609
617
.
34.
Rotella
,
J. J.
(
2018
). “
Demographic data for Weddell seal colonies in Erebus Bay through the 2017 Antarctic field season
,” U.S. Antarctic Program (USAP) Data Center, available at https://www.usap-dc.org/view/dataset/601125 (Last viewed October 27, 2020).
35.
Rotella
,
J. J.
(
2020
). (personal communication).
36.
Russell
,
L.
,
Purdy
,
J.
, and
Davis
,
R.
(
2016
). “
Social context predicts vocalization use in the courtship behaviors of Weddell seals (Leptonychotes weddellii): A case study
,”
Anim. Behav. Cogn.
3
(
2
),
95
119
.
37.
Sales
,
G.
, and
Pye
,
D.
(
1974
).
Ultrasonic Communication by Animals
(
Chapman and Hall
,
London
), pp.
1
281
.
38.
Samarra
,
F. I. P.
,
Deecke
,
V. B.
,
Vinding
,
K.
,
Rasmussen
,
M. H.
,
Swift
,
R. J.
, and
Miller
,
P. J. O.
(
2010
). “
Killer whales (Orcinus orca) produce ultrasonic whistles
,”
J. Acoust. Soc. Am.
128
(
5
),
EL205
EL210
.
39.
Schevill
,
W. E.
, and
Watkins
,
W. A.
(
1971
). “
Directionality of the sound beam in Leptonychotes weddelli (Mammalia: Pinnipedia)
,” in
Antarctic Pinnipedia, Antarctic Research Series
, edited by
W. H.
Burt
(
American Geophysical Union
,
Washington, DC
), Vol.
18
, pp.
163
168
.
40.
Schusterman
,
R. J.
,
Kastak
,
D.
,
Levenson
,
D. H.
,
Reichmuth
,
C. J.
, and
Southall
,
B. L.
(
2000
). “
Why pinnipeds don't echolocate
,”
J. Acoust. Soc. Am.
107
(
4
),
2256
2264
.
41.
Smith
,
M. S. R.
(
1965
). “
Seasonal movements of the Weddell seal in McMurdo Sound, Antarctica
,”
J. Wildl. Manage.
29
(
3
),
464
470
.
42.
Southall
,
B. L.
,
Finneran
,
J. J.
,
Reichmuth
,
C.
,
Nachtigall
,
P. E.
,
Ketten
,
D. R.
,
Bowles
,
A. E.
,
Ellison
,
W. T.
,
Nowacek
,
D. P.
, and
Tyack
,
P. L.
(
2019
). “
Marine mammal noise exposure criteria: Updated scientific recommendations for residual hearing effects
,”
Aquat. Mamm.
45
(
2
),
125
232
.
43.
Stirling
,
I.
(
1969
). “
Ecology of the Weddell seal in McMurdo Sound, Antarctica
,”
Ecology
50
(
4
),
573
586
.
44.
Terhune
,
J. M.
(
1974
). “
Directional hearing of a harbor seal in air and water
,”
J. Acoust. Soc. Am.
56
(
6
),
1862
1865
.
45.
Terhune
,
J. M.
(
2019
). “
The underwater vocal complexity of seals (Phocidae) is not related to their phylogeny
,”
Can. J. Zool.
97
(
3
),
232
240
.
46.
Terhune
,
J. M.
, and
Dell'Apa
,
A.
(
2006
). “
Stereotyped calling patterns of a male Weddell seal (Leptonychotes weddellii)
,”
Aquat. Mamm.
32
(
2
),
175
181
.
47.
Testa
,
J. W.
, and
Siniff
,
D. B.
(
1987
). “
Population dynamics of Weddell seals (Leptonychotes weddelli) in McMurdo Sound
, Antarctica,”
Ecol. Monogr.
57
(
2
),
149
165
.
48.
Thiebault
,
A.
,
Charrier
,
I.
,
Aubin
,
T.
,
Green
,
D. B.
, and
Pistorius
,
P. A.
(
2019
). “
First evidence of underwater vocalisations in hunting penguins
,”
PeerJ
7
,
e8240
.
49.
Thomas
,
J. A.
, and
Awbrey
,
F. T.
(
1983
). “
Ultrasonic vocalizations of leopard seals (Hydrurga leptonyx)
,”
Antarct. J. U.S.
17
,
186
.
50.
Thomas
,
J. A.
,
Ferm
,
L. M.
, and
Kuechle
,
V. B.
(
1987
). “
Silence as an anti-predation strategy by Weddell seals
,”
Antarct. J. U.S.
22
,
232
234
.
51.
Thomas
,
J. A.
,
Ferm
,
L. M.
, and
Kuechle
,
V. B.
(
1988
). “
Patterns of underwater calls from Weddell seals (Leptonychotes weddellii) during the breeding season at McMurdo Sound
, Antarctica,”
Antarct. J. U.S.
23
,
146
148
.
52.
Thomas
,
J. A.
, and
Kuechle
,
V. B.
(
1982
). “
Quantitative analysis of Weddell seal (Leptonychotes weddelli) underwater vocalizations at McMurdo Sound, Antarctica
,”
J. Acoust. Soc. Am.
72
(
6
),
1730
1738
.
53.
Thomas
,
J. A.
, and
Stirling
,
I.
(
1983
). “
Geographic variation in the underwater vocalizations of Weddell seals (Leptonychotes weddelli) from Palmer Peninsula and McMurdo Sound, Antarctica
,”
Can. J. Zool.
61
(
10
),
2203
2212
.
54.
Thomas
,
J. A.
, and
Terhune
,
J. M.
(
2009
). “
Weddell seal (Leptonychotes weddelli)
,
” in Encyclopedia of Marine Mammals
, 2nd ed., edited by
W. F.
Perrin
,
B.
Wrusig
, and
J. G. M.
Thewisse
(
Academic
,
Amsterdam
), pp.
1217
1220
.
55.
Thomas
,
J. A.
,
Zinnel
,
K. C.
, and
Ferm
,
L. M.
(
1983
). “
Analysis of Weddell seal (Leptonychotes weddelli) vocalizations using underwater playbacks
,”
Can. J. Zool.
61
(
7
),
1448
1456
.
56.
Tyack
,
P. L.
(
1997
). “
Studying how cetaceans use sound to explore their environment,” in Communication
,
Perspectives in Ethology
, edited by
D. H.
Owings
,
M. D.
Beecher
, and
N. S.
Thompson
(
Springer
,
Boston, MA
), Vol.
12
, pp.
251
297
.
57.
van Opzeeland
,
I.
,
Van Parijs
,
S.
,
Bornemann
,
H.
,
Frickenhaus
,
S.
,
Kindermann
,
L.
,
Klinck
,
H.
,
Plötz
,
J.
, and
Boebel
,
O.
(
2010
). “
Acoustic ecology of Antarctic pinnipeds
,”
Mar. Ecol. Prog. Ser.
414
,
267
291
.
58.
Vater
,
M.
, and
Kössl
,
M.
(
2004
). “
The ears of whales and bats
,” in
Echolocation in Bats and Dolphins
, edited by
J. A.
Thomas
,
C. F.
Moss
, and
M.
Vater
(
University of Chicago Press
,
Chicago, IL
), pp.
89
99
.
59.
Wartzok
,
D.
,
Elsner
,
R.
,
Stone
,
H.
,
Kelly
,
B. P.
, and
Davis
,
R. W.
(
1992
). “
Under-ice movements and the sensory basis of hole finding by ringed and Weddell seals
,”
Can. J. Zool.
70
(
9
),
1712
1722
.
60.
Watkins
,
W. A.
, and
Schevill
,
W. E.
(
1968
). “
Underwater playback of their own sounds to Leptonychotes (Weddell seals)
,”
J. Mammal.
49
(
2
),
287
296
.
61.
Wellard
,
R.
,
Pitman
,
R. L.
,
Durban
,
J. D.
, and
Erbe
,
C.
(
2020
). “
Cold call: The acoustic repertoire of Ross Sea killer whales (Orcinus orca, Type C) in McMurdo Sound, Antarctica
,”
R. Soc. Open Sci.
7
,
191228
.

Supplementary Material