Temporary threshold shifts from mid-frequency airborne noise exposures in seals^a) Open Access

Amphibious marine mammals—including true seals, walruses, sea lions, fur seals, sea otters, and polar bears—present challenges for understanding and mitigating the harmful effects of human-generated noise. These carnivores rely on acoustic information to support essential behaviors both out of water in terrestrial habitats and when submerged at sea. Thus, intense noise that reaches individuals through air- or water-borne pathways could result in behavioral disturbance, auditory masking, temporary hearing loss [i.e., temporary threshold shift (TTS)], or permanent hearing loss [i.e., permanent threshold shift (PTS)]. Assessments of these noise impacts have been centrally important in the development of marine mammal noise exposure criteria (Southall , 2007; Southall , 2019) and their application in regulatory guidelines for managing anthropogenic noise (National Marine Fisheries Service, 2018).

Studies of TTS caused by exposure to controlled noise have been conducted with both phocids (true seals) and otariids (sea lions and fur seals). Most data have been collected with underwater sound exposures [see Finneran (2015) for review], including broadband noise (Kastak , 1999; Kastak , 2005b; Kastak , 2007; Kastelein , 2012; Kastelein , 2013; Kastelein , 2019a; Kastelein , 2020a; Kastelein , 2020b; Kastelein , 2020c; Kastelein , 2021; Kastelein , 2022a; Kastelein , 2022b; Kastelein , 2024), tonal noise (Kastelein , 2019b; Reichmuth , 2019), and impulsive sounds (Finneran , 2003; Kastelein , 2018; Reichmuth , 2016; Sills , 2020). When tested in water, phocids and otariids appear to be similarly susceptible to the effects of noise—TTS onset is comparable for at least three pinniped species when noise exposures are referenced to subjects' hearing thresholds in the same frequency range (Kastak , 1999; Kastak , 2005b).

A single study evaluated TTS induced by airborne noise in an otariid: an adult female California sea lion (Zalophus californianus) routinely exposed to mid-frequency, octave-band noise of different levels and durations in a hemi-anechoic acoustic chamber (Kastak , 2007). These data were compared to measurements of TTS obtained for the same sea lion in water (Kastak , 2005b). Results indicated similar noise effects above and below the water's surface when exposure levels were referenced to this subject's frequency- and medium-specific hearing thresholds. Whether this pattern holds for phocids is unknown. Differences in peripheral auditory anatomy between phocids and otariids may result in functional differences in air but not water: while bone or tissue-conducted hearing presumably occurs in water in both taxa [Nummela (2008); but see Lipatov (1992)], hearing in air is thought to involve the traditional sound conduction pathway used by terrestrial mammals (Møhl, 1968). This ancestral pathway, with airborne sounds received from the environment and transmitted through air-filled passages to the fluid-filled cochlea, differs anatomically between the phocid and otariid lineages [see Nummela (2008)]. Furthermore, phocid seals have additional modifications in auditory structures [e.g., the meatal openings, see Ruscher (2021)]. In light of possible differences in the effects of airborne versus waterborne noise as a result of these structure-function relationships, empirical measurements are needed to describe patterns of TTS for phocids exposed to sounds in air and enable direct comparisons to available underwater datasets.

In the present study, TTS was evaluated following controlled airborne noise exposures in two phocid species—the harbor seal (Phoca vitulina) and the northern elephant seal (Mirounga angustirostris). The two seals tested were also the subjects of prior experiments demonstrating TTS onset, growth, and recovery arising from mid-frequency, octave-band underwater noise (Kastak , 1999; Kastak , 2005b). In terms of sound exposure level (SEL), TTS onset in water was estimated at 183 dB re 1 μPa² s for the harbor seal and 204 dB re 1 μPa² s for the elephant seal, based on threshold shifts observed at the center frequency of the noise exposure and one half-octave higher (Kastak , 1999; Kastak , 2005b). When referenced to hearing capabilities, these SELs are approximately 132 dB above each subject's corresponding auditory threshold. The results of these prior studies indicated that the exposure duration of octave-band noise contributed more to observed patterns of TTS than equal-energy noise of greater amplitude but shorter duration. In contrast, more recent hearing data for otariids from Kastelein and colleagues (2021, 2022a) demonstrated similar amounts of TTS for narrower band noise exposures of comparable SEL despite systematic variations in exposure parameters of noise duration and level.

The present study was conducted to further clarify the exposure conditions leading to TTS onset in seals and the relationship between SEL and TTS, including any tradeoffs between noise amplitude and duration. Noise exposure levels were measured in absolute terms (SPL, dB re 20 μPa) and also defined relative to sensation level (SL, dB re hearing threshold). The specific approach matched that previously reported by Kastak (2007) for the single California sea lion voluntarily exposed to airborne noise. Collecting these data with controlled airborne noise provided information necessary for estimating TTS from anthropogenic noise in air [e.g., Southall (2019)], while avoiding acknowledged difficulties associated with continuous wideband noise exposures for the same subjects voluntarily diving to position within underwater noise fields [as described in Kastak (2005b) and Kastak (2007)].

This research was led by Dr. David Kastak and collaborators at the University of California Santa Cruz and is being published posthumously to honor the value of this work.¹ While these data were obtained more than 20 years ago, they provide valuable measurements of TTS for phocid carnivores exposed to controlled airborne noise. Although these experiments were conceived at a time when fewer TTS studies on marine mammals were available, results are shared in light of the current body of empirical knowledge.

A psychoacoustic procedure was used to measure reliable changes in hearing threshold induced by controlled exposures to octave-band noise centered at the same frequency. The details and timing of this study with two seals are the same as those described by Kastak (2007) for a sea lion. Phase 1 testing occurred from January 2002 to December 2003 and Phase 2 from January 2005 to January 2006.

The seals had prior experience in behavioral audiometry, including studies of noise-induced TTS in water (Kastak , 1999; Kastak , 2005a) and in air (Kastak and Schusterman, 1996). The adult male harbor seal (Sprouts, NOA0001707) was 14–17 years old. He had apparently normal hearing (Reichmuth , 2013), although subsequently published data for conspecific individuals in another laboratory suggest his thresholds below 2 kHz may have been elevated (Kastelein , 2009). The adult female elephant seal (Burnyce, NOA0004829) was 9–12 years old. She had a chronic, unilateral (right) otitis externa (Kastak and Schusterman, 1998). Given that there are no comparative data for another individual of this species, it is unknown whether her hearing was representative (Kastak and Schusterman, 1998, 1999; Reichmuth , 2013). More recent reports describing hearing in Hawaiian monk seals (Neomonachus schauinslandi) suggest her auditory thresholds were reasonable in reference to related (monachid) seals (Ruscher , 2021; Ruscher , 2025; Sills , 2021).

Both seals lived at Long Marine Laboratory in Santa Cruz, CA in seawater-filled pools surrounded by haul-out space. The seals were trained using operant conditioning and positive (fish) reinforcement and all aspects of the study were conducted under voluntary behavioral control. Individual diets were established to maintain optimal health and were not constrained for research purposes. Animal protocols were approved by the Institutional Animal Care and Use Committee at the University of California Santa Cruz. Research was authorized by the United States National Marine Fisheries Service under marine mammal permits 259–1481 and 1072–1771.

Procedures took place in a sound attenuating, hemi-anechoic chamber (Eckel Industries, Cambridge, Massachusetts, USA) located near the seals' living areas. It included a 3.3 m × 2.3 m × 2.2 m test room and adjacent control room. The sides and ceiling of the test room were double walled for sound isolation and covered with fiberglass-filled wedges. Ambient noise levels within the frequency region of interest (0.5–5 kHz) were below −15 dB re (20 μPa)²/Hz [see Mulsow and Reichmuth (2010) and Sills (2014)]. The test room was cool and dim to ensure animal comfort.

The chamber was configured to accommodate behavioral audiometry and noise exposure. A chin cup that served as a “listening station” was mounted at a comfortable height for each seal. A response target was positioned on the same plane to the left of this station. A JBL 2123H speaker (JBL Incorporated, Northridge, CA, USA) was mounted 70 cm in front of the listening station, on axis with the center of the seal's head. This speaker projected tonal signals during audiometry. A small light to delineate trial duration was mounted under this speaker. A second speaker in the chamber played an acoustic cue to mark correct responses. A PVC tube to deliver primary (fish) reinforcement to the seal was built into the wall connecting the test room to the control room.

A separate chin-cup station was located 45° to the right of the listening station. This “exposure station” was used to establish the position of the seal's head during noise exposures. Noise was projected from a Fender Princeton Chorus Guitar Amplifier (Fender Musical Instruments Corporation, Scottsdale, AZ, USA) or a Community EM280 compression driver coupled to a P100 horn projector (Community Professional Loudspeakers, Chester, PA, USA). Another indicator light was mounted above the noise speaker to denote correct stationing behavior to the seals.

Signals used for audiometry were 1 kHz (elephant seal) or 2.5 kHz (harbor seal) pure tones of 500 ms duration with 40 ms linear onset/offset ramps. Signals were generated with custom labview software and an NI PXI-6070 multifunction DAQ device housed in a PXI 8176 controller (NI, Austin, Texas, USA). The tones were filtered (Krohn-Hite 3530, Brockton, MA, USA), amplified (Hafler P7000, Hafler Professional, Tempe, AZ, USA), and played from the JBL speaker into the test room. Projected signals were calibrated at the center position of the seal's head on the listening station with a C550H microphone (Josephson Engineering, Santa Cruz, CA, USA). The tones were calibrated and inspected daily in time and frequency domains using either a PC-based signal analysis package (Spectra Plus, Pioneer Hill) or a combination of custom labview-based virtual instruments. The linearity of attenuation was confirmed through direct measurement.

Noise exposure stimuli comprised octave-band white noise centered on the frequency of the test tone. Noise exposure levels for both seals were normalized in terms of SL, defined as the difference in dB between the noise level (unweighted sound pressure level, SPL, referenced to 20 μPa) and the subject's hearing threshold at the same frequency. Noise was generated with the NI hardware, digitally filtered, and analog bandpass filtered using the Krohn-Hite 3530 to obtain a flattened frequency spectrum at the exposure station. Noise was amplified with a Hafler P9000 power amplifier and projected from the Fender amplifier during Phase 1 of the experiment and the Community horn projector during Phase 2. Fatiguing noise was calibrated daily at the center position of the seal's head on the exposure station. Received noise waveforms and spectra were evaluated for stimulus integrity.

Audiometry signals and fatiguing noise were spatially mapped in the sound fields that encompassed the subject's head on the listening and exposure stations, respectively. Received levels for both signals and noise in these areas were within 2 dB of those at the calibration position in each case.

Testing was conducted in two phases (see Table I). Phase 1 included noise exposures for both subjects with levels of 65, 80, and 95 dB SL and exposure durations of 1.5, 12, 25, and 50 min. These conditions generated maximum unweighted sound exposure levels (SELs) of 133 and 155 dB re (20 μPa)² s for the harbor seal and elephant seal, respectively.

Based on the TTS values determined in Phase 1, higher exposure levels were selected for Phase 2. For the elephant seal, noise levels were 98, 101, and 104 dB SL for continuous exposure durations of 6.25, 12.5, 25, and 50 min. For the harbor seal, this initial range of noise levels (starting at 98 dB SL) did not generate any threshold shifts. Therefore, higher relative noise exposures of 119, 122, and 125 dB SL were used for this subject with exposure durations of 25, 37, and 50 min. These conditions generated maximum unweighted SELs of 163 and 164 dB re (20 μPa)² s for the harbor seal and elephant seal, respectively.

The absolute noise levels presented during the full study ranged from 68 to 128 dB SPL for the harbor seal and from 90 to 129 dB SPL for the elephant seal. The exposure conditions produced by these noise levels (incremented in 3-dB steps) and durations (incremented by doubling) increased in a systematic manner, and were established to evaluate the differential contributions of exposure level and duration to the observed TTS for equal cumulative SEL conditions.

Eight replicate sequences were conducted for every noise exposure condition. Control sessions were also conducted, in which the subject was positioned at the exposure station for the specified duration, but no noise was presented. During control sequences, pre- and post-exposure hearing thresholds were measured in an identical manner as for noise exposure sequences. Three or four control sequences were conducted for each exposure duration.

Prior to data collection, seals were trained to voluntarily move from their home enclosure to the acoustic chamber. They learned to perform the go/no-go hearing procedure and to position calmly at the exposure station for extended periods of continuous, consistent noise or control exposures. Testing occurred in an ordered routine each day and included measurement of a pre-exposure hearing threshold, a rest period, an exposure interval of up to 50 min, and immediate measurement of a post-exposure hearing threshold. This sequence was completed over a period of 2–3 h.

Subjects completed 40–60 signal detection trials to estimate the pre-exposure hearing threshold at the test frequency. Performance was observed and recorded on closed-circuit video monitors. A trial began when the seal placed its head firmly on the listening station, on axis with the projector. A trial light was illuminated to delineate the 4-s listening interval. A tone was presented during 50% of trials (“signal-present” trials), and the seal indicated detection by moving from the station to touch the response target. The remainder of the session consisted of interspersed “signal-absent” trials (0-V signal), and the seal remained still to indicate that no signal was detected. Correct responses to the presence or absence of the test tone were marked by a conditioned reinforcer (bell sound) followed by primary (fish) reinforcement passed to the seal. False alarms and misses were not marked, reinforcement was not provided, and the seal proceeded to the next trial. Stimulus level on signal-present trials followed an adaptive descending staircase sequence (Cornsweet, 1962), with SPL lowered by 2 dB following each correct detection and raised by 2 dB following each miss. The session ended after nine reversals (transitions from a correct detection to a miss or vice versa) and several trials including easily detectable signals. Reversal data were used to calculate the hearing threshold according to the method of Dixon and Mood (1948).

After pre-exposure threshold measurement, the seal returned to a nearby pool for a resting period of at least 30 min. Provided that the pre-exposure session fell within an acceptable range (+/− 3 dB of the baseline threshold determined prior to the study) and response bias was acceptable (false alarms < 25%), the seal returned to the chamber. The noise (or control condition) was established prior to entry. The duration and level of the exposure condition for each session were selected from the shuffled, predetermined exposure matrix shown in Table I. The indicator light was illuminated while the seal positioned calmly at the exposure station. The light was periodically extinguished to denote the delivery of reinforcement and was turned back on when the seal resumed the exposure position a few seconds later.

After the specified exposure duration of 1.5 to 50 min the noise was ramped off. The seal was prompted to position at the listening station, and the hearing threshold measurement was immediately repeated. “Post-exposure” threshold testing followed the same procedure used to determine the “pre-exposure” threshold. The same session was conducted the following day to evaluate hearing 24 h after the exposure event. If this “recovery” threshold was similar to (within 3 dB of) the baseline threshold, it was used as the pre-exposure threshold for the subsequent sequence.

Temporary threshold shifts were measured as the difference between a seal's pre-exposure hearing threshold and the subsequent post-exposure hearing threshold at the frequency of the noise exposure.

For all level/duration exposure conditions, threshold shifts for each seal were evaluated using a repeated-measures ANOVA with pre-exposure, post-exposure, and recovery thresholds grouped by exposure sequence. When the ANOVA results were significant at the 0.05 level, comparisons between each session type were made using a Newman-Keuls multiple comparisons test.

To evaluate how sound level and sound duration contributed to observed TTS under the equal-energy hypothesis, we compared TTS generated by different level/duration combinations resulting in the same SEL. Results were evaluated to determine whether level or duration had a greater apparent effect on observed TTS.

The threshold shifts generated in response to airborne noise were compared to those previously obtained for the same individuals in the same exposure conditions (relative to SL) during testing in water. Comparisons were made with a t-test between the 25 and 50 min noise exposures at 95 dB SL and the corresponding in-water results reported by Kastak (2005b). Using a common SL as the metric of comparison allowed for the exposures in air and in water to be equated for each seal relative to hearing threshold in each medium.

Despite the difficulty and extended duration of this behavioral task and the intensity of the noise exposure conditions presented, both seals maintained a high level of behavioral tolerance and control throughout the study. The seals voluntarily entered the sound field of the acoustic chamber and remained cooperative for every session (193 exposure and control sequences for the harbor seal, 220 for the elephant seal). The mean threshold shift measured for each exposure condition for each seal is shown in Table II for both phases of testing. All mean shifts were below 10 dB.

For the harbor seal, threshold shifts were measured over the interval from 4 to 11 min following the cessation of noise exposure, on average. This roughly equates to TTS₇, as compared to the TTS₄ metric most commonly reported in marine mammal studies (Finneran, 2015). In Phase 1, significant threshold shifts were uncommon when exposure durations were 25 min or less. At these durations, threshold shifts were observed only at the highest exposure level of 95 dB SL. For the longest exposure duration of 50 min, threshold shifts occurred at both 80 and 95 dB SL. The increased amplitudes used in Phase 2 for exposure durations of 25, 37, and 50 min yielded significant threshold shifts in every case. The maximum individual shift of 23 dB was observed at an SL of 125 dB and duration of 37 min. This condition also produced the highest mean threshold shift of 9.5 dB.

Threshold shifts for the elephant seal were measured over the interval from 5 to 14 min following the cessation of noise exposure, roughly corresponding to TTS₁₀. The elephant seal showed measurable TTS in Phase 1 for three noise conditions: 25 min at the highest exposure level of 95 dB SL and 50 min at the two highest amplitudes of 80 and 95 dB SL. The level/duration exposure combinations used in Phase 2 generated threshold shifts in almost every condition. The maximum individual shift of 12 dB was observed at 98 dB SL for the longest duration of 50 min. The greatest mean shift exhibited by the elephant seal was 7.4 dB at the highest exposure level (104 dB SL) and the longest duration (50 min).

Thresholds obtained 24 h after exposures did not differ from pre-exposure thresholds for either subject, indicating complete recovery from any hearing loss and confirming that threshold shifts were temporary. There were no significant threshold shifts following control exposures at any duration.

Threshold shifts arising from every noise exposure sequence are shown for each seal in Fig. 1. TTS values are grouped by SEL with the relative contributions of noise level (in SPL) and noise duration (in min) depicted by symbol hue and symbol size, respectively. The growth of TTS with increasing exposure level covers a span from no effect to modest but consistent threshold shift. The same data are represented in Fig. 2, grouped by common SEL across exposure conditions with the curve fit from Eq. (1) displaying the onset and growth of TTS. The elephant seal showed a predictable response to increasing noise exposure level (a = 4.5; b = 148; R² = 0.94) with TTS onset of 161 dB SEL, defined at 6 dB of effect. The harbor seal demonstrated two distinct responses to noise, with the sound exposure level required to induce TTS increasing by 24 dB in the second experimental phase. The 6-dB onset level for the harbor seal in Phase 1 was 134 dB SEL (a = 2.8; b = 113; R² = 0.83), representing a small extrapolation above the maximum Phase 1 SEL of 133 dB. The 6-dB onset level for TTS in Phase 2 was 158 dB SEL (a = 5.5; b = 147; R² = 0.54), similar to that observed for the elephant seal.

The TTS onset values for each seal were derived from SEL values combining various level/duration conditions. However, when sound exposures with comparable SEL were isolated and compared (Table III), the elephant seal always showed greater threshold shifts when exposed to noise of relatively longer duration and lower sound pressure level. This pattern is consistent with results obtained from a sea lion subject tested in the same equal-energy paradigm (Kastak , 2007). The harbor seal's data showed the same trend in Phase 1, although the relatively greater effect of duration was not apparent in Phase 2. Overall, an equal-energy relationship between noise level and duration did not explain the observed patterns of TTS for the elephant seal in either phase or for the harbor seal in Phase 1 of this study.

Comparisons were possible between the threshold shifts generated in air in this study and in water for the same subjects [Fig. 3 (Kastak , 2005b)]. There were two common noise exposure conditions at similar frequencies for each seal in each medium. The exposure level was 95 dB above the corresponding hearing threshold and durations were either 25 or 50 min. For the 25-min noise exposures, TTS magnitude was similar in air and water. For the longer exposures of 50 min, TTS differed marginally for each individual by medium. The harbor seal showed more TTS in water than in air [two-tailed t-test; t(14) = 2.2, p = 0.04], while the elephant seal showed more TTS in air relative to in water [two-tailed t-test; t(14) = 2.3, p = 0.03].

The reported data provide the first measurements of TTS induced by controlled airborne noise in phocid seals. Results demonstrate recoverable hearing loss from various combinations of exposure level and duration presented in mid-frequency, octave-band noise conditions. The study design—using behavioral audiometric testing performed at the center frequency of the noise exposure—presented two primary advantages. Testing the seals with airborne rather than underwater broadband noise addressed key limitations of typical approaches [e.g., Kastak (1999), Kastak (2005b), Kastak (2007), Kastelein (2012), Kastelein (2013), Kastelein (2019a), Kastelein (2020a), Kastelein (2020b), Kastelein (2020c), Kastelein (2021), Kastelein (2022a), Kastelein (2022b), and Kastelein (2024)], which may be associated with intermittency during diving in underwater noise fields or increased variability in stimuli received by free-swimming subjects [see Tougaard (2022)]. Additionally, the methods applied here allowed for a systematic evaluation of the contributions of noise level, duration, and SEL to the onset of TTS.

It was anticipated that noise exposures could be better controlled in a non-reverberant, quiet testing environment (an acoustic chamber) compared to measurements obtained outdoors in saltwater pools. Subject positioning for continuous noise exposure could also be more easily maintained in the acoustic chamber rather than in water, where the seals would periodically surface during longer-duration exposures to breathe and receive fish reinforcement, temporarily removing them from the calibrated noise field. Because previous results in water indicated that noise levels referenced to absolute detection thresholds (i.e., sensation levels) provided an appropriate measure to compare noise effects between individuals (Kastak , 1999; Kastak , 2005b), testing in air would presumably confirm whether this relation would hold when comparing noise effects between media. Finally, testing the effects of noise in air allowed for simple monitoring of corresponding physiological variables that could be associated with noise-induced stress (i.e., respiratory rate, heart rate).²

We were able to directly compare TTS arising from noise exposures in air to TTS measured previously in water for the same subjects. Noise levels set 95 dB above the seals' thresholds generated similar amounts of TTS for exposures of 25 min in both media, despite the fact that exposures were continuous for each seal in air and intermittent in water. This is consistent with prior findings for a California sea lion (Kastak , 2007). The much longer 50-min exposures at 95 dB SL generated the highest amount of TTS in both media. In water, the measured threshold shifts were more variable, at least in part due to the intermittency of the exposure as the seals surfaced to breathe [see Kastak (2005b)]. Perhaps as a result, there were small to moderate (4 to 7 dB) differences in threshold shifts measured across media for these seals at the higher exposure condition. In general, however, the data support the use of SL as a relevant metric for predicting TTS magnitudes in pinnipeds, independent of exposure medium.

The use of the SL metric to consider noise levels relative to hearing threshold provided a way to equate noise exposures for species with different underlying hearing abilities, with the assumption that the same offset of fatiguing noise level relative to hearing sensitivity would produce a similar amount of TTS [e.g., Kastak (1999), Kastak (2005b), and Kastak (2007)]. Here, the harbor seal showed remarkably similar TTS onset in air and in water when the noise exposure level was referenced to underlying hearing sensitivity (i.e., the difference between noise SEL and hearing threshold SPL). From this relative perspective, the harbor seal's TTS onset SEL was 131 dB above hearing threshold in air and 129 dB above threshold in water (Kastak , 2005b). For the elephant seal, these relative values were 136 dB in air and 138 dB in water (Kastak , 2005b). Both sets of values are similar to those reported for a California sea lion, 130 and 131 dB in air and in water, respectively (Kastak , 2005b; Kastak , 2007).

Based on these findings and consistent with Kastak (2007), the SL metric can be used to equate noise exposures of equal duration in different media, and to normalize exposure levels across individuals and species, at least under some conditions. As TTS data will not become available for all marine mammal taxa, this concept may prove particularly useful in predicting the effects of human-generated noise for semi-aquatic species on land as well as at sea. This finding also has important implications for future auditory research since testing in air is preferable from a procedural point of view—the subject's behavior can be placed under better control, the stimuli can be better characterized, and testing can be accomplished more rapidly.

The auditory effects of noise exposure were readily modeled as a function of increasing SEL for both seals using established equations, although the relative contributions of noise level and duration differed. For the harbor seal, TTS onset occurred at an SEL of 134 dB re (20 μPa)² s and for the elephant seal, at an SEL of 161 dB re (20 μPa)² s. This difference aligns with the far greater hearing sensitivity of the harbor seal to airborne sounds in the frequency range of testing (>20 dB more sensitive, with the harbor seal's baseline hearing threshold of 3 dB re 20 μPa at 2.5 kHz). Below the TTS onset value determined for each seal, auditory effects were rarely observed for noise levels below 85 dB SL or durations less than 12 min. Above the onset value, TTS increased similarly for both seals with increasing SEL.

It is important to note that results from this study apply to hearing effects at the center frequency of the noise exposure; an even greater effect may occur a half-octave higher, particularly at higher exposure levels (Finneran, 2015, but see Kastak , 2005b). Further, given that the behavioral methods used allowed for measurement of TTS 7 to 10 min following the cessation of noise [similar to these subjects performing the same task in water (Kastak , 2005b)], the reported TTS onset values may be somewhat conservative relative to the TTS₄ metric used in other studies [see Finneran (2015)].

Interestingly, the harbor seal showed two distinct responses to noise, corresponding to the two phases of this study. Considerably higher noise levels were used in Phase 2 than originally planned, as exposures at and above 119 dB SL (122 dB SPL) were required to reliably induce TTS for this subject. Accordingly, onset levels for TTS increased from 134 to 158 dB SEL, a difference of about 24 dB between these two experimental phases. The reasons for this difference remain unresolved but suggest some potential protective mechanism that is not involved in aquatic hearing.

The TTS magnitudes measured across both study phases for both seals are somewhat lower than expected relative to those obtained previously from a California sea lion under nearly identical testing conditions (Kastak , 2007), especially for the harbor seal. In the present case, the mean threshold shifts observed were always < 10 dB regardless of noise exposure conditions. In contrast, exposures of similar and smaller magnitude (referenced to SL) generated threshold shifts up to 30 dB in a California sea lion (Kastak , 2007).

Phocid and otariid carnivores have several similarities and differences in auditory anatomy and function that may be relevant to their responses to noise. The phocid seals are partly characterized by the absence of an external ear pinna; in contrast, the otariid sea lions have pinnae that are well-defined, though reduced to hollow, rolled flaps with a small opening at the tip (Lipatov, 1992). The functional role of the pinnae in pinniped hearing, if any, is unknown. The phocids retain a homologue of the pinna in the presence of a helical plate that surrounds the opening to the meatus and is situated in a shallow depression. This cartilaginous structure appears to be important in regulating the opening of the ear canal (meatus). The structure of the auditory meatus has been described in detail (Ramprashad , 1971) and is grossly similar to that of the otariids, except for the pinna (Lipatov, 1992). Both phocids and otariids have cartilaginous sheaths and fibrous connective tissue lining the meatus, although the meatus may be closed by different mechanisms. Some phocids appear able to actively control the opening of the meatus with at least two sets of muscles attached to the meatal wall (Ramprashad , 1971). While it seems unlikely, it is not known whether the otariids can control the opening of the meatus or the position of the pinna.

The elephant seal represents what appears to be an extreme modification of the external ear. Although the auditory meatus is similar to that of other phocids in shape, the opening is no more than a pinhole on the side of the head and is often occluded by a waxy plug. The conduction pathway for airborne sound is likely to be through the meatus, but sound is probably significantly attenuated and bone conduction may play a greater role in hearing in this species than in other terrestrial or amphibious mammals. It is not clear whether this species actively controls the opening and closing of the meatal opening. It has been suggested that all seals from the monachinae lineage share this passively closed auditory canal (Ruscher , 2021).

The observation of a 24-dB increase in TTS onset level for the harbor seal in Phase 2 of the present study is intriguing and suggests the emergence of a self-mitigating (protective) response to noise exposure. This finding of noise “resistance” in air lies in the ability of phocids to close the meatal opening. It is possible that ear closure, under voluntary control, was initiated by the harbor seal as a learned response when exposed to higher-amplitude noise in Phase 2 [see Kastak (2005a)]. Video footage obtained with an endoscopic camera placed near the meatal opening revealed that this subject reliably closed his external meatal openings during noise exposure intervals, and relaxed and opened them during the listening intervals associated with subsequent audiometric testing.³ The reduction in apparent sensitivity during active closure of the meatal canals is notably comparable to the 25–45 dB reduction in hearing threshold observed for human subjects when occluding the meatal opening (Holland, 1967). Furthermore, the active closure of the harbor seal's meatal openings seems to have rendered an attenuating effect similar to that of the elephant seals' passively closed (pinhole) auditory canal. This is reflected by a TTS onset level for the harbor seal in Phase 2 that aligns almost perfectly with that of the elephant seal (note that the TTS curves for each seal shown in Fig. 2 could be easily overlaid). This intriguing observation merits further study to resolve both the extent to which this response may occur in other individuals and the specific neuromuscular mechanisms involved.

TTS onset is defined in terms of the SEL metric—first proposed in marine mammal noise criteria (Southall , 2007) and subsequently established as the primary guidance for allowable marine mammal noise exposures in the United States (National Marine Fisheries Service, 2018; Southall , 2019). Use of this combined metric relies on the assumption that different combinations of noise level and duration yielding comparable SELs will result in similar TTS. However, in the present study, longer-duration exposures generally resulted in greater mean threshold shifts than shorter-duration exposures of the same sound energy. This was not always the case–in Phase 2, results for the harbor seal were more variable. Regardless, the observed TTS patterns did not fully support an equal-energy trading rule, consistent with findings by Kastak (2007) and Kastelein (2012) for sea lions and harbor seals, respectively. Conversely, more recent work by Kastelein and colleagues (2021, 2022a) using trained sea lions in equal-energy scenarios with narrow-band noise failed to find a greater relative effect of noise duration. The exposure conditions under which equal-energy assumptions apply remain equivocal.

Although unequal tradeoffs between level and duration may bias estimates of the effects of anthropogenic noise on marine mammals, the equal-energy rule remains a useful construct. At present, SEL is generally considered the best available predictor of TTS onset in marine mammals [see Finneran (2015) and Tougaard (2025)]. The TTS-onset levels reported in the present study are expressed in terms of SEL to allow for consideration and implementation within existing management frameworks. However, the premise that noise level and duration contribute equally to TTS across exposure scenarios should continue to be evaluated against new data to identify the conditions under which the equal-energy rule applies or requires modification.

These auditory data collected from highly trained marine mammal subjects 20 years ago remain relevant. Such measurements of the effects of airborne noise are important to understand the potential adverse impacts of high-intensity noise in the environment, such as sounds associated with rocket launches near seal breeding rookeries [e.g., Thorson (2000) and Holst (2011)]. These data also provide an empirical basis for comparing the fatiguing effects of noise in air and water, as has been done previously in the context of auditory masking [e.g., Southall (2003) and Erbe (2016)]. It is rather unusual to evaluate TTS in air for mammals that are considered to be marine. However, when noise exposure levels are considered relative to absolute thresholds, it becomes clear that equivalent sensation levels between air and water induce similar amounts of TTS. While testing with additional exposure conditions may be needed to confirm that this relationship holds for all noise types, the results indicate that datasets obtained in both media can be combined to increase predictive power.

When combined with existing reports of noise-induced hearing loss in pinnipeds, these data support the continued derivation of noise exposure criteria for marine mammals, which inspired this work and the many studies to follow.

This project was designed and conducted by Dr. David Kastak with Dr. Ronald Schusterman at the University of California Santa Cruz. This paper is published by the students and collaborators in their team, who assisted with this research and are carrying on the legacy of their work and spirit. We thank the dedicated research and animal care teams at Long Marine Laboratory for their long-term assistance in conducting these experiments and caring for resident marine mammals. Kirstin Jensen provided essential support with physiological monitoring and Dr. Martin Haulena conducted endoscopic videography. We thank Dr. James Finneran, Dr. Dorian Houser, Dr. Mardi Hastings, and Dr. Robert Gisiner for assistance with data interpretation and for helpful early feedback with this research. We thank Dr. Brian Branstetter and an anonymous reviewer for their comments on this manuscript. Data collection for this study was supported by the Office of Naval Research, Marine Mammals and Oceanography program under Awards Nos. N00014-02-1‐0159 and N00014-04-0284. The U.S. Navy's Living Marine Resources Program provided support for data synthesis and manuscript preparation.

The authors have no conflicts to disclose.

Research was conducted cooperatively without harm to animals. Authorization for this work was granted by the United States National Marine Fisheries Service under marine mammal permits 259–1481-00 and 1072–1771-00. Animal research protocols were approved by the Department of Defense Animal Care and Use Program and by the Institutional Animal Care and Use Committee at the University of California Santa Cruz.

Data that support the findings of this study are openly available in DRYAD at http://doi.org/10.5061/dryad.f4qrfj759.