Temporary hearing threshold shift in harbor seals (Phoca vitulina) due to one-sixth-octave noise bands centered at 0.5, 1, and 2 kHz

Marine mammals that are exposed to high-amplitude sounds generated by anthropogenic offshore activities, such as percussion pile driving, seismic surveys, sonar, and detonations, may suffer temporary threshold shift (TTS) or permanent hearing threshold shift (PTS) (Melnick, 1991; Yost, 2007). For the regulation and management of such activities, it is important to know at what sound exposure level (SEL), a combination of received sound pressure level (SPL) and exposure duration, hearing may be reduced in marine mammals (cetaceans, sirenians, otariids, and phocids, see overview by Finneran, 2015). Safety criteria for underwater sound to protect marine mammal hearing were proposed by Southall et al. (2007, 2019), but were based on the limited TTS data available for each of the “marine mammal hearing groups” into which they divided the marine mammals.

One such group, the “phocid carnivores” (Southall et al., 2019), includes the harbor seal (Phoca vitulina). The harbor seal occurs in temperate and Arctic coastal areas of the northern hemisphere (Burns, 2009), where high levels of human activity can produce underwater sound with high enough SELs to cause TTS or PTS (Ainslie et al., 2009; Mannerla et al., 2013; Merchant et al., 2016).

Until 2013, only four studies on TTS in harbor seals had been published (Kastak et al., 1999; Kastak et al., 2005; Kastelein et al., 2012a, 2013), but the results were difficult to compare due to differences in fatiguing sound types, exposure duration, hearing frequencies tested relative to the center frequencies of the fatiguing sounds, and differences in methodology of measuring TTS. Therefore, a research project was started on the frequency-dependent susceptibility of harbor seals to TTS in which all fatiguing sounds were continuous (100% duty cycle) and narrow-band [a sinusoidal wave (CW) and one-sixth-octave noise bands (NBs)], the exposure duration was 1 h (except in the present study when longer periods were also tested), and, in most cases, three hearing frequencies were tested (center frequency of the fatiguing sound, half an octave above the center frequency, and one octave above the center frequency). So far, fatiguing sounds with center frequencies of 6.5, 16, 32, and 40 kHz have been tested (Kastelein et al., 2019a, 2019b, 2020a, 2020b). The present study is the final part of the large research project. The goal was to expose harbor seals to continuous one-sixth-octave NBs centered at 0.5, 1, and 2 kHz, and quantify the hearing frequencies affected, the recovery of hearing after the exposure stopped, and the TTS in relation to the SEL of the fatiguing sound.

The goals of the larger project are to present and compare information on TTS after exposure to sound at various frequencies, to establish equal-TTS curves for harbor seals, and to develop a research-based weighting curve for harbor seals (as described by Houser et al., 2017). The data can be used by government regulators to estimate the numbers of individual seals that will be affected by human activities. Regulators can then set safe underwater sound-level criteria for harbor seals, which can probably be extended to other phocids (Southall et al., 2019).

The study animals were two healthy adult female harbor seals, identified as F01 and F02. They were 12–13 yr old during the period of data collection. The two seals had very similar girths, and body weights which fluctuated seasonally between ∼45 kg in summer and ∼62 kg in winter. Details on their husbandry and food rations are provided by Kastelein et al. (2019b).

The study was conducted at the SEAMARCO Research Institute in an outdoor pool measuring 8 × 7 m and 2 m deep (see Kastelein et al., 2019a for details) with haul-out areas. During the sound exposure sessions, these haul-out areas were barred so that the seals could not leave the water. During the hearing tests, the seal not being tested was kept in the water next to the main haul-out area, and was asked to perform quiet husbandry behaviors.

Acoustical terminology follows ISO 18 405 (2017). The ambient noise was measured, and the fatiguing sound and hearing test signals were calibrated once every two months during the study period by acoustic consultant TNO (The Hague, the Netherlands; see Kastelein et al., 2019a, 2019b). Under test conditions (i.e., only researchers involved in the study allowed within 15 m of the pool, water circulation system off, no rain, and wind force Beaufort 4 or below), the ambient noise in the pool was very low and fairly constant in amplitude (Fig. 1; see Kastelein et al., 2019a, 2019b). The critical ratios of harbor seals for 0.5, 1, and 2 kHz are 15, 21, and 25 dB respectively (Terhune 1991; Turnbull and Terhune, 1990; Southall et al., 2000), indicating that no masking occurred.

Continuous (100% duty cycle for 1–6 h) one-sixth-octave NBs centered at 0.5, 1, and 2 kHz, without harmonics, were selected as fatiguing sounds. The lowest frequency for which a sufficiently high SPL could be generated to elicit TTS was 0.5 kHz; pilot studies with 125 and 250 Hz showed no TTS_1–4, as hearing sensitivity in harbor seals is reduced below 1 kHz (Kastelein et al., 2009a, 2009b). For details on the emitting and receiving equipment, see Kastelein et al. (2019a, 2019b). The NBs at 0.5 and 1 kHz were played by a low-frequency transducer (Model 350 Hydrosounder, Data Physics Corporation, San Jose, CA) placed at 1 m depth on one side of the pool. The NB at 2 kHz was played by a transducer (Lubell LL 1424HP, Lubell Labs Inc., Columbus, OH) in the same location at 1.5 m depth. During the first sessions with NB at 0.5 kHz, it was clear that the seals preferred certain locations in the pool and it was suspected, and later verified, that these were low SPL locations due to standing waves in the pool. Results from these sessions were removed from the data set. To prevent standing waves (causing varying SPLs in the pool) during the following sessions, a linear actuator was used to move the low-frequency transducer (Hydrosounder 350) continuously back and forth in a 270° arc in the horizontal plane every 30 s during the projection of the 0.5 kHz fatiguing sound (it was not needed for the 1 kHz fatiguing sound). Before each sound exposure test, the voltage output of the emitting system to the transducer and the voltage output of the sound-receiving system were checked (Kastelein et al., 2019a, 2019b). The voltage meter and the underwater sound were monitored by the operator throughout the exposure periods.

To determine the distribution of each fatiguing sound in the pool, the SPL was measured on a horizontal grid of 1.2 × 1.3 m, at three depths per location on the grid (0.5, 1.0, and 1.5 m below the surface) per sound. The SPL of the NB at 0.5 kHz could not be measured below the platform, because the floating platform was rigged tighter at the time of the calibration measurements for this frequency. No gradient existed in the SPL in relation to the distance to the transducer, causing a fairly homogeneous sound field (apart from in the area within 1 m of the transducer, where the SPL was higher). Figure 2 shows examples of SPL distributions for each fatiguing sound frequency. The variation in mean SPLs between sessions was <1 dB. The highest SPLs used per fatiguing sound were the highest that were attainable without harmonics, as harmonics could potentially affect the results. The fatiguing sounds were projected at the following mean SPLs in the pool: for the NB at 0.5 kHz at 168 dB re 1 μPa; for the NB at 1 kHz at 155, 161, and 164 dB re 1 μPa; and for the NB at 2 kHz at 144, 150, 156, 162, 165, and 173 dB re 1 μPa.

The seals generally swam throughout the entire pool during each exposure, so the mean SPL of all measurement locations could be used to calculate the mean SPL to which they were exposed. Most (61%) of the 388 exposures to fatiguing sound lasted for 1 h. However, in order to reach higher SELs, the NBs were also presented at longer exposure durations: 4 and 6 h for the NB at 0.5 kHz (mean SPL 168 dB re 1 μPa); 2, 4, and 6 h for the NB at 1 kHz (mean SPL 164 dB re 1 μPa); and 2 and 4 h for the NB at 2 kHz (mean SPL 173 dB re 1 μPa). To allow for direct comparison, SELs were calculated from mean received SPLs and exposure durations. The exposure durations were not all identical, as they were in previous studies in this project (1 h; Kastelein et al., 2019a, 2019b, 2020a, 2020b).

The hearing test signals that the seals had to detect in the hearing tests before and after the fatiguing sound exposure were generated digitally with the software Adobe® Audition 3.0 (Adobe®, Sunrise, FL). The hearing thresholds were tested at the center frequency of the three fatiguing sounds (0.5, 1, and 2 kHz, respectively), at half an octave above the center frequency of the fatiguing sounds (0.71, 1.4, and 2.8 kHz, respectively), and at one octave above the center frequency of the fatiguing sounds (1, 2, and 4 kHz, respectively).

The hearing test signals were 1-s linear upsweeps starting and ending at ±2.5% of the center frequency, and including a linear rise and fall in amplitude of 50 ms (Kastelein et al., 2019a, 2019b). All hearing test signals were transmitted with a balanced tonpilz piezoelectric acoustic transducer (Lubell LL916, Lubell Labs Inc., Columbus, OH) that was 1.5 m away from the listening station (an L-shaped poly-vinyl-chloride 3-cm diameter water-filled tube with an end cap on which the seal positioned itself during hearing tests) at 1 m depth (same depth as the listening station). The underwater SPL at the location of a harbor seal's head at the listening station was varied by the operator in 2-dB increments (for details of calibration, see Kastelein et al., 2019a, 2019b). During free-field calibrations before the hearing tests, two hydrophones, one where each of the seal's ears would be when the seals were positioned at the listening station, were used to measure the received SPL during the hearing tests. The SPL at the two locations differed by 0–2 dB, depending on the test frequency; the mean SPL of the two measurement locations per hearing test frequency was used to calculate the stimulus SPL during hearing threshold tests.

A maximum of one total sound exposure or control test was conducted per day, consisting of: (1) pre-exposure hearing test session in which the baseline hearing threshold of each seal for one hearing frequency was quantified; (2) fatiguing sound exposure for 1, 2, 4, or 6 h (depending on the NB center frequency and intended SEL) or ambient noise exposure for 1 h (control); and (3) post-sound exposure (PSE) hearing test session(s) in which the hearing threshold was quantified for comparison to the baseline threshold (using the same test frequency as in the pre-exposure hearing test). To avoid a startle response, the SPL of the fatiguing sound was increased slowly during the first 60 s of each exposure period. Although the exposure periods are reported here in hours, they were carefully timed up to the minute using stopwatches.

Each pre- and post-exposure hearing test session consisted of ∼25 trials and lasted for up to 12 min per seal. For each seal, the first PSE session was divided into three 4-min periods: 1–4 (PSE_1–4), 4–8 (PSE_4–8), and 8–12 (PSE_8–12) min for seal F02; and 12–16 (PSE_12–16), 16–20 (PSE_16–20), and 20–24 (PSE_20–24) min for seal F01. Sessions were comprised of two-thirds signal-present trials and one-third signal-absent trials, offered in quasi-random order (Kastelein et al., 2019a, 2019b). Good control over the animals' behavior allowed the PSE hearing test to commence within 1 min after the fatiguing sound had stopped for seal F02, and at 12 min after the sound had stopped for seal F01. The two seals were always tested in this order to ensure a consistently quick and efficient start after the sound exposure stopped. The audiometric method used (go/no-go, behavioral method, operant conditioning) is described in detail by Kastelein et al. (2019a, 2019b).

Besides the magnitude of the initial TTS, subsequent changes in hearing were recorded over time. The hearing sensitivity of seal F02 was always tested during PSE_1–4, PSE_4–8, and PSE_8–12. If F02's hearing had not recovered during PSE_8–12, it was also tested 60 (PSE₆₀) min and, if not recovered during PSE₆₀, 120 (PSE₁₂₀) min after the fatiguing sound exposure ended. The hearing sensitivity of seal F01 was always tested during PSE_12–16, PSE_16–20, and PSE_20–24. If F01's hearing had not recovered during PSE_20–24, it was also tested 72 (PSE₇₂) min after the fatiguing sound exposure ended.

Sample sizes were chosen to maximize the study time available for testing SELs in which TTS seemed to occur and avoiding repeated testing of SELs for which TTS obviously did not occur. To protect their hearing, the seals were exposed to a fatiguing sound for a maximum of once per day. Randomizing the order in which the seals were tested while maintaining equal sample sizes was considered, but would have doubled the length of the study period.

Control tests, conducted the same way as sound exposure tests but without fatiguing sound exposure, were randomly dispersed among the fatiguing sound exposure tests during the study period. Data were collected over three overlapping periods (due to the availability of the transducers): between June and October 2019 (NB at 0.5 kHz), July 2019 and January 2020 (NB at 1.0 kHz), and July 2018 and March 2020 (NB at 2.0 kHz).

The mean incidence of pre-stimulus responses (“pre-stimuli”) by the seals for both signal-present and signal-absent trials (in the latter, the stimulus for the correct lack of response was a whistle; see Kastelein et al., 2019a, 2019b) was calculated as the number of pre-stimuli as a percentage of all trials in each hearing test period.

To investigate behavioral responses and determine the mean received SELs during fatiguing sound exposure, the seals were monitored by a researcher who was hidden in a research cabin next to the pool (see Fig. 1 in Kastelein et al., 2012a). Additionally, recordings were made by a video camera mounted on a pole to provide a complete top view of the pool. From the video recordings, the seals' locations in the pool were monitored, as was the position of their heads, which could be (1) completely submerged; (2) at the water surface with their nose in the air, and mandible and lower part of the skull (containing the inner, middle, and most of the outer ear) under water; or (3) completely out of the water. In addition, the swimming speed was categorized as higher, the same, or lower than during non-exposure periods (control periods). For details of the method used for the analysis of video recordings, see Kastelein et al. (2019). Due to poor lighting, the recordings of 7% of the sessions could not be analyzed.

In the hearing tests, a switch from a test signal level to which the harbor seal responded (a “hit”) to a level to which she did not respond (a “miss”), and vice versa, was called a “reversal”. The pre-exposure mean 50% hearing threshold (PE_50%) for each hearing test session was determined by calculating the mean SPL of all reversals in the pre-exposure hearing session. Only pre-exposure sessions with at least ten reversals were included in the analysis (the maximum was 12 reversals per session). The TTS in seal F02 during PSE_1–4 (TTS_1–4) for each hearing test frequency was calculated by subtracting the PE_50% from the mean 50% hearing threshold measured during PSE_1–4 (using only sessions with at least four reversals during each 4-min period). The same method was used to calculate the other TTSs in seal F02 (TTS_4–8, TTS_8–12, etc.) and the TTSs in seal F01.

Before the study started, recovery of hearing had to be defined, so that hearing tests could stop once recovery had taken place. Based on the threshold fluctuations observed in previous TTS studies with the same harbor seals (Kastelein et al., 2012a, 2018b, 2019a, 2019b, 2020a, 2020b), recovery was defined as a return to within 2 dB of the pre-exposure hearing threshold (TTS ≤ 2 dB).

After the study was conducted, we defined the onset of biological TTS in the present study as occurring at the lowest SEL at which a significant difference could be detected between the hearing threshold shift due to the fatiguing sound exposures (i.e., a TTS) and the hearing threshold shift as measured after the control exposures (this shift was close to zero). This biological TTS onset differs from the TTS onset defined as the SEL causing 6 dB TTS by Southall et al. (2007, 2019), and used in Sec. C of the Discussion. When statistical analysis of significance was not possible due to a sample size of ≤ 3 TTS measurements, the data were not included in the analysis. For larger sample sizes, the level of significance was established by conducting a one-way repeated measures analysis of variance (ANOVA) on the TTS, separately for each seal, each fatiguing sound, and each hearing test frequency, with the factor SEL (including the control). When the ANOVA produced a significant value overall, the levels were compared to the control by means of Dunnett multiple comparisons. All analyses were conducted in Minitab 18 (Minitab LLC, State College, PA), and data conformed to the assumptions of the tests used (Zar, 1999).

During the 1 h control sessions, both seals swam throughout the pool, at all depths, and spent ∼20% of their time with their heads at the water surface, and ∼80% of their time with their heads completely submerged. During exposure to the continuous one-sixth-octave NB centered at 0.5, 1, and 2 kHz, they spent 17%–73% of their time with their heads at the water surface, and the rest of the time with their heads completely submerged (Table I). No change was observed in this pattern over time. During both control and exposure periods, the seals raised their head completely out of the water a few times, for only a few seconds each time.

During almost all sessions, both seals swam throughout the pool, at all depths, confirming that their mean received SPL had been calculated appropriately as the energetic mean of the SPL at all 96 individual measurement locations (Fig. 2). The only exception was when seal F01 stayed in one location at the water surface for a relatively long time during exposure to the NB at 2 kHz. In that case, the SPL measured at that particular location at 0.5 m depth was exactly the same as the mean SPL of the pool, thus the SEL was not affected.

The pre-stimulus response rates of both seals for all trials in the pre-exposure, post-exposure, and control hearing tests were low and of the same order of magnitude (Table II). The control sessions for both seals showed that the hearing thresholds for all three hearing test signals before and after 1-h exposure to low ambient noise were very similar (Tables III, IV, and V). After sound exposure periods, the seals were always willing to participate in the hearing tests, and no change in susceptibility to TTS was observed over the duration of the study.

After the harbor seals were exposed to an NB centered at 0.5 kHz, the ANOVA showed that TTS_1–4 (seal F02) was significantly affected by the fatiguing sound's SEL at the hearing frequency 0.71 kHz, and TTS_12–16 (seal F01) was significantly affected by the fatiguing sound's SEL at the hearing frequencies 0.71 and 1 kHz (i.e., half an octave and one octave above the exposure frequency). No TTS occurred in either seal at the hearing frequency 0.5 kHz. In seal F02, at the hearing frequency 0.71 kHz, TTS_1–4 (biological TTS onset) occurred at an SEL of 210 dB re 1 μPa²s and in seal F01, TTS_12–16 occurred at an SEL of 211 dB re 1 μPa²s (Table III and Fig. 3). In seal F01, small but significant TTS_12–16 occurred at the hearing frequency 1 kHz at an SEL of 211 dB re 1 μPa²s. Hearing always recovered during the next 4–8-min period in seal F02 and during the 16–20 min period in seal F01 (Figs. 4 and 5).

After the harbor seals were exposed to the NB centered at 1 kHz, the ANOVAs showed that TTS_1–4 (seal F02) and TTS_12–16 (seal F01) were significantly affected by the fatiguing sound's SEL at the hearing frequencies 1.4 and 2 kHz. No TTS occurred at the hearing frequency 1 kHz. In seal F02, at the hearing frequency 1.4 kHz, TTS_1–4 (biological TTS onset) occurred at an SEL of 200 dB re 1 μPa²s and at the hearing frequency 2 kHz, biological TTS onset occurred at an SEL of 207 dB re 1 μPa²s. In seal F01, at both the hearing frequencies 1.4 and 2 kHz, TTS_12–16 occurred at an SEL of 207 dB re 1 μPa²s (Table IV and Fig. 6).

In F02, hearing at 1.4 kHz recovered within 12 min for exposure to sounds up to an SEL of 206 dB re 1 μPa²s, and >60 min after exposure to an SEL of 207 dB re 1μPa²s (Fig. 7). Hearing was not measured later than 60 min after the exposure, as it was the end of the working day; hearing had recovered at 0830 h the next day. For a hearing test signal of 2 kHz and an SEL of 207 dB re 1 μPa²s, hearing recovered within 60 min (Fig. 7). In seal F01, hearing at 1.4 kHz always recovered within 20 min, and hearing at 2 kHz recovered within 72 min after exposure to an SEL of 207 dB re 1 μPa²s (Fig. 8).

After the harbor seals were exposed to the NB centered at 2 kHz, the ANOVA showed that TTS_1–4 (seal F02) and TTS_12–16 (seal F01) were significantly affected by the fatiguing sound's SEL at all three hearing frequencies (2, 2.8, and 4 kHz). In seal F02, at all three hearing frequencies, TTS_1–4 (biological TTS onset) occurred at an SEL of 192 dB re 1 μPa²s. In seal F01, TTS_12–16 occurred at an SEL of 215 dB re 1 μPa²s at the hearing frequency 2 kHz, at an SEL of 198 dB re 1 μPa²s at the hearing frequency 2.8 kHz, and at an SEL of 212 dB re 1 μPa²s at the hearing frequency 4 kHz (Table V and Fig. 9).

In seal F02, hearing at 2 kHz recovered within 8–12 min, except after exposure to an SEL of 215 dB re 1 μPa²s, when it recovered within 60 min (Fig. 10). Hearing at 2.8 kHz recovered within 8–12 min after exposure to an SEL of 192 dB re 1 μPa²s; otherwise, it recovered within 60 min. Hearing at 4 kHz recovered within 60 min, except after exposure to an SEL of 198 dB re 1 μPa²s, when it recovered within 120 min (Fig. 10). In seal F01, hearing at 2 kHz recovered within 20 min, hearing at 2.8 kHz recovered within 16–20 min (after SEL ≤198 dB re 1 μPa²s) or within 20–24 min (after SEL 201, 215 dB re 1 μPa²s), and hearing at 4 kHz recovered within 20 min (Fig. 11).

Although the hearing of seal F02 was always tested before the hearing of seal F01, their TTSs may be compared. The closest possible comparison is of TTS_8–12 in seal F02 and TTS_12–16 in seal F01, and only the TTSs at hearing frequencies half an octave above the center frequency of the fatiguing sounds are directly compared (Table VI). Allowing for the recovery in the minute between the two measurement periods and the potential differences in sound exposure due to small differences in swimming tracks of the two seals, the TTSs are of the same order of magnitude. Due to recovery of hearing, one would expect the TTS_12–16 in seal F01 to be slightly lower than the TTS_8–12 in seal F02; this was only the case for the higher TTSs (Table VI).

When the low-frequency transducer was stationary and producing the 0.5 kHz fatiguing sound, each seal stayed much longer in particular locations and did not swim randomly throughout the pool. It was suspected that the animals found locations with lower than average SPLs and, thus, the SEL the seals were exposed to could not be determined, so the results from the stationary transducer were discarded. When the linear actuator was used to move the transducer, the seals went back to regular swimming as they had done during all previous TTS studies with other frequencies. This measure did not increase their TTS, but it made certain that we knew the SEL to which the animals were exposed.

The pre-exposure hearing thresholds found in the present study for the hearing test frequencies were within a few dB (−1 to +4 dB, depending on the seal and frequency) of the hearing thresholds measured in these same harbor seals for tonal signals approximately 10 yr before the present study (Kastelein et al., 2009a, 2009b, 2010).

The similar pre-stimulus response rates in hearing tests before and after the sound exposures in both seals show that their decision-making process in the post-exposure hearing tests was not influenced by the sound exposure.

In the present study, not all exposures were of the same duration. In order to increase the elicited TTS to statistically significant levels, the exposure period of the maximum SPL that could be produced without harmonics was increased. Even though we calculated the SEL the seals were exposed to, it remains a question whether the SELs composed of 1 h exposures can be compared to SELs based on other exposure durations. The equal energy hypothesis assumes that continuous (100% duty cycle) sound exposures with the same energy (expressed in SEL) lead to the same TTS (Southall et al., 2007). However, Kastelein et al. (2012a) showed that this assumption is not met for TTS resulting from low-SPL, long-duration (1 h) noise exposures in harbor seals; different levels of TTS resulted from exposure to sounds with identical SELs, but consisted of different duration/level combinations. The assumption also fails for bottlenose dolphins (Tursiops truncatus; Mooney et al., 2009; Finneran and Schlundt, 2010), and for harbor porpoises (Phocoena phocoena; Lucke et al., 2009; Kastelein et al., 2012b), in which an increase in SEL due to an increase in exposure duration has a different effect on the induced TTS than the same increase in SEL due to an increase in SPL. However, a study in which harbor porpoises were exposed to continuous 1–2 kHz sweeps showed that the same cumulative SELs (composed of various combinations of SPL, exposure duration, and duty cycle) did elicit the same magnitude of TTS_1–4, as predicted by the equal energy hypothesis, as long as the duty cycle between the SEL/TTS comparisons was the same (Kastelein et al., 2014). Thus, within a certain range, SEL can be a predictor of the induced initial TTS, but it must be kept in mind that SEL criteria obtained from only high-SPL, short-duration exposures might underestimate the TTS induced as a function of the exposure duration, particularly for low-SPL, long-duration exposures.

Both seals spent more time swimming at the water surface with their nose in the air, and mandible and lower part of the skull under water as the 1 and 2 kHz SELs increased. However, they did not keep their heads completely out of the water for long periods. This suggests that they were somewhat disturbed by the noise levels, but not to the point that they showed stress, demonstrated by, e.g., increased swimming speed. Also, they were always motivated to participate in the hearing threshold measurements immediately after the cessation of the fatiguing noise. It is possible that the fatiguing noise levels were slightly lower near the water surface, and that SELs were, thus, overestimated. However, this would not have had a major effect with respect to the frequency and SEL combinations that were or were not causing TTS. Had the seals raised their heads completely clear of the water, then the SEL levels that the seals were exposed to would definitely have been overestimated.

When the seals used in the present study were exposed to one-sixth-octave NBs centered at 16, 32, and 40 kHz (Kastelein et al., 2019b, 2020a, 2020b), they spent approximately the same amount of time swimming at the water surface during exposure to the fatiguing sound and during control periods. The seals lifted their entire heads out of the water only a few times per session, and each time only for a few seconds, generally to quickly scan their above-water surroundings. They could have used the same tactic in the present study to reduce their exposure to the fatiguing sound for longer time periods, but apparently did not feel the need to do so. However, in some previous studies they spent more time swimming at the water surface during exposure periods (pile-driving sounds: Kastelein et al., 2018b; 6.5-kHz, CW: Kastelein et al., 2019a), possibly because those sounds (an impulsive sound and a tone) were experienced as more annoying than the narrow NBs used in the present project. Because the seals did not lift their heads fully out of the water in the studies with pile-driving sounds and a 6.5-kHz tone, we assume this behavioral response had no effect on the received SEL (Kastelein et al., 2018a).

Although TTS may have an ecological effect on an animal, that animal is probably not aware of the fact that it has TTS and, thus, TTS does not cause a form of stress. By reducing hearing, TTS reduces the perceived level of a sound and thereby reduces distraction from other potentially more relevant sounds at frequencies differing from that of the fatiguing sound.

It is not yet clear whether noise-induced hearing loss in harbor seals can be predicted based on the equal energy hypothesis (that all combinations of SPL and exposure durations that result in the same SEL elicit the same initial TTS). Therefore, we only compare TTS after the exposure for 1 h to continuous fatiguing sounds, with the same methodology; the data from the present study on TTS after longer exposure durations are not included in the comparison (Fig. 12). Once the equal energy hypothesis has been tested in harbor seals, the results of the >1-h exposures, depending on the outcome, may be included. Southall et al. (2007, 2019) proposed 6 dB TTS_1–4 as a TTS onset marker. In the present study, 1-h exposures to 0.5 and 1 kHz fatiguing sound at the maximum SPL that could be generated without producing high-SPL harmonics (168 dB re 1 μPa for 0.5 kHz, and 164 dB re 1 μPa for 1 kHz) did not elicit ≥6 dB TTS_1–4. Only exposure for 1 h to 2 kHz fatiguing sound at ∼194 dB re 1 μPa²s resulted in 6 dB TTS_1–4 at 2.8 kHz in seal F02. The susceptibility of harbor seals to TTS caused by underwater sound, as indicated by the SEL required to elicit 6 dB TTS_1–4 in the fatiguing sound frequency range tested in previous studies (4–40 kHz; Kastelein et al., 2012a, 2018b, 2019a, 2019b, 2020a, 2020b), varied little with hearing frequency (Fig. 12). However, a higher SEL was required to cause 6 dB TTS by the 2 kHz fatiguing sound in the present study.

The present study clearly shows that harbor seal hearing is less susceptible to TTS after exposure to lower frequencies than after exposure to higher frequencies. Most anthropogenic sounds have most of their energy at low frequencies, so when defining criteria for acceptable broadband SELs to protect seal hearing, levels should be weighted to assess impacts on harbor seal hearing, in order to avoid being overly restrictive for human offshore activities.

We thank students Pepijn Degger, Femke Kuiphof, Roos de Lepper, Luna Korsuize, Kyra Robbemont, Bette Bluijs, Eline Theuws, Amber Verhoef, Anna van Neerven, Stefan van der Graaf, Danu Hoftijzer, Lorena Dhondt, Femke Bucx, Emmy Post, Joshi van Berlo, Jimme Bruining, Carmen van Duijn, Bram Carree, and Lin Hopmans; and volunteers Stacey van der Linden, Stephanie de Ruijter, Kimberly Biemond, Naomi Claeys, and Brigitte Slingerland for their help in collecting the data. We thank Arie Smink for the design, construction, and maintenance of the electronic equipment. We thank MARIN in Wageningen, the Netherlands for making the hydrosounder transducer available for the project. We thank Bert Meijering (Topsy Baits) for providing space for the SEAMARCO Research Institute. Erwin Jansen (TNO) conducted the acoustic calibration measurements. We also thank Nancy Jennings (Dotmoth.co.uk) for her valuable constructive comments on this manuscript, and for conducting the statistical analysis. Funding for this study was obtained from the U.S. Navy's Living Marine Resources (LMR) program (Contract No. N39430-18-C-2041). We thank Mandy Shoemaker and Anu Kumar for their guidance on behalf of the LMR program. The seals were made available for the research by Ecomare. The training and testing of the harbor seals were conducted under authorization of the Netherlands Ministry of Economic Affairs, Department of Nature Management, with Endangered Species Permit no. FF/75A/2016/031.