Noise pollution in aquatic environments can cause hearing loss in noise-exposed animals. We investigated whether exposure to continuous underwater white noise (50–1000 Hz) affects the auditory sensitivity of an aquatic turtle Trachemys scripta elegans (red-eared slider) across 16 noise conditions of differing durations and amplitudes. Sound exposure levels (SELs) ranged between 155 and 193 dB re 1 μPa2 s, and auditory sensitivity was measured at 400 Hz using auditory evoked potential methods. Comparing control and post-exposure thresholds revealed temporary threshold shifts (TTS) in all three individuals, with at least two of the three turtles experiencing TTS at all but the two lowest SELs tested, and shifts up to 40 dB. There were significant positive relationships between shift magnitude and exposure duration, amplitude, and SEL. The mean predicted TTS onset was 160 dB re 1 μPa2 s. There was individual variation in susceptibility to TTS, threshold shift magnitude, and recovery rate, which was non-monotonic and occurred on time scales ranging from < 1 h to > 2 days post-exposure. Recovery rates were generally greater after higher magnitude shifts. Sound levels inducing hearing loss were comparatively low, suggesting aquatic turtles may be more sensitive to underwater noise than previously considered.
I. INTRODUCTION
The detection and use of sound by aquatic animals evolved in pristine soundscapes, yet these species are increasingly pressured by a host of anthropogenic activities that increase noise levels incidentally (e.g., pile driving, vessel noise) or deliberately (e.g., sonar, seismic surveys) across marine and freshwater habitats (Gedamke , 2016; Rountree , 2020; Duarte , 2021). Noise pollution fundamentally alters the acoustic environment experienced by wildlife and can induce stress and injury, in addition to disrupting critical needs like communication and predator avoidance (Tyack, 2008; Slabbekoorn , 2010; Kunc and Schmidt, 2019). Sound is a particularly valuable sensory mechanism in aquatic habitats where factors like high turbidity, low light, and rapid diffusion compromise other senses like vision and olfaction (Giles , 2009). Therefore, many aquatic species have evolved to take advantage of hearing as a key sensory modality. However, these species may in turn be more vulnerable to noise pollution given the relationship between auditory sensitivity and an increased susceptibility to noise-induced hearing loss (Smith , 2004a). Underwater hearing sensitivity and noise susceptibility is currently unknown for many aquatic taxa, compromising an accurate assessment of the potential impacts of anthropogenic sound for many species.
Noise-induced temporary threshold shifts (TTS) occurs when an individual is exposed to a fatiguing sound that results in a temporary but recoverable loss of sensitivity, as measured by an increase in auditory thresholds. Threshold shifts have been measured across a variety of taxa, including mammals, birds, and fishes (Clark, 1991). The widespread presence of TTS indicates a resiliency of auditory systems to maintain baseline hearing sensitivity by repairing or recovering from noise-induced effects. However, the prevalence, amplitude, and duration of anthropogenic sounds in today's waters are vastly different from the historical natural soundscapes (Hildebrand, 2009; Rountree , 2020) under which auditory repair and recovery mechanisms evolved. Even temporary hearing loss impairs an individual's ability to perceive its acoustic environment and can be associated with progressive damage to the inner ear under certain conditions (Kujawa and Liberman, 2009; Lin , 2011). Measures of TTS are often used to predict permanent hearing loss (PTS), and levels of TTS and PTS are used in analyses of noise impacts by regulatory agencies (Department of the Navy, 2017).
Reptiles are taxa for which we know relatively little about hearing sensitivity and noise-induced hearing loss (aerial or underwater) or the effects of anthropogenic sound more generally (Kunc and Schmidt, 2019; Jerem and Mathews, 2021). This limits predictions of how anthropogenic noise impairs sound-mediated behaviors, such as acoustic communication to attract mates or for parental care (Giles , 2009; Ferrara , 2013) or navigation using acoustic cues (Lohmann , 2008). Available evidence supports that reptiles experience auditory damage and reduced auditory sensitivity in response to aerial noise [Dipsosaurus dorsalis (Bondello, 1977); Uma scoparia (Brattstrom and Bondello, 1983); Gerrhonotus multicarinatus (Tilney , 1982; Mulroy , 1990; Henry and Mulroy, 1995)]. Electrophysiological studies have shown reptilian species (e.g., Alligator mississipiensis, Hydrophis stokesii, Caretta caretta, Chelonia mydas) have greatest underwater auditory sensitivity to low-frequency sounds [< 1 kHz (Ridgway , 1969; Higgs , 2002; Lavender , 2014; Piniak , 2016; Chapuis , 2019)], but there are no data addressing if high intensity underwater sound affects reptile hearing.
Given this lack of data, surrogates are often used to predict hearing loss in aquatic turtles. For example, in a regulatory context, freshwater fishes are currently used as the underwater surrogate for predicting noise-induced hearing loss in sea turtles (Department of the Navy, 2017). This is motivated by general similarities in the frequencies to which these taxa have greatest auditory sensitivity, but does not account for vast differences in parameters such as ear morphology, mechanisms of sound conduction to auditory hair cells, and potentially higher-order processing (Webster , 1992). These differences make it challenging to extrapolate from fishes to aquatic turtles the degree of TTS susceptibility upon exposure to high intensity underwater sounds. To better manage and conserve turtle and tortoise species there is a clear need to examine whether this taxon experiences noise-induced hearing loss and the noise levels and durations inducing TTS. Testudines (turtles, tortoises, terrapins) face an array of anthropogenic stressors (Gibbons , 2000) that have led to more than a third of species being listed as Endangered and over half as Threatened [2018 IUCN Red List (Rhodin , 2018)], with ecosystem-level consequences (Lovich , 2018). Understanding how high intensity underwater noise affects the hearing of aquatic turtles is critical to successfully mitigating the impacts of noise pollution on these species.
The amphibious nature of many Testudine species makes them interesting candidates for noise exposure studies. Trachemys scripta elegans (red-eared slider; Wied-Neuwied, 1838) has greater sensitivity to underwater sound compared to aerial, attributed in part to the large resonant air-filled middle-ear cavity (Christensen-Dalsgaard , 2012; Willis , 2013). It has been hypothesized that the middle ear cavity evolved in the most recent common ancestor of extant Testudines by enhancing the detection of conspecific vocalizations and auditory scene analysis in aquatic habitats (Willis , 2013). This highlights the importance of underwater sound to these species and the potential impact of anthropogenic noise on individual survival and fitness.
We evaluated whether T. scripta elegans is susceptible to noise-induced hearing loss. We tested the hypothesis that underwater auditory thresholds would increase after exposure to intense continuous white noise. Further, we tested the hypothesis that the magnitude of shift would increase with increasing exposure duration and amplitude, and consequently sound exposure level (SEL). We addressed these hypotheses by testing underwater auditory sensitivity, measured using auditory evoked potential methods (AEP), of three individuals in response to exposure to 16 combinations of noise amplitudes and durations. These data provide the first evidence of TTS in a turtle species and support their vulnerability to noise-induced hearing loss.
II. METHODS
A. Sedation and monitoring
We used three adult female T. scripta elegans acquired from local sources approved by the Massachusetts Division of Fisheries and Wildlife (Permit No. 075.20LP) and the Institutional Animal Care and Use Committee (WHOI ID Nos. 25252.01 and 25999). We conducted testing from September 2020 through May 2021. For additional details on animals and husbandry, refer to the supplementary material.1
We obtained AEP data using a combination of sedation and mechanical restraint (Harms , 2014; Piniak , 2016). The sedation plan insured lack of movement underwater to minimize muscle artifact and allowed turtles to maintain the ability for spontaneous respiration. We used a combination of intramuscular (IM) midazolam (1 mg/kg) and dexmedetomidine (3–90 μg/kg), and adjusted dosages of dexmedetomidine for individual turtles throughout the study period to provide a consistent plane of sedation for each trial (RES01: 20–90 μg/kg; RES04: 25–50 μg/kg; RES05: 3–30 μg/kg). A single turtle (RES01) required the adjustment of midazolam up to 3 mg/kg and the addition of ketamine (2–3 mg/kg IM) for the final 13 sessions to obtain adequate sedation; for more information and effects on data collection, refer to the supplementary material.1 We monitored turtle alertness, reactivity, and respiratory rate during testing, and brought the turtles to the surface to breathe every 60–90 s, or when showing intention signs to surface (Harms , 2014). Onset and duration of sedation varied among individuals, but typically allowed for 60–90 min of AEP data collection, which was initiated 45–50 min post-administration of sedatives. Atipamezole (1:1 volume with dexmedetomidine; 10:1 mg:mg dose) was administered IM in cases with prolonged recovery (> 3–4 h) and was needed in approximately one-third of the sessions. Turtle health was assessed by weekly physical exams performed by a veterinarian and evaluation of blood parameters [lactate concentration, packed cell volume (PCV), and total solids (TS)], which supported the experimental procedures did not have a significant effect on aneraerobic metabolism. For blood parameter methods and results, refer to the supplementary material.1
B. AEP test setup and methodology
We conducted auditory measurements in a rectangular PVC test tank [0.84 m L × 0.38 m H × 0.53 m W; Fig. 1(A)] with water maintained between 24 and 26 °C and housed inside a custom 5-sided plywood box lined with sound-reducing pyramid foam [as in Stanley (2020)]. This setup was in a designated room to which electricity was turned off during testing to reduce potential electromagnetic interference. Similarly, the speaker (wrapped in aluminum foil), tank water, and all AEP equipment were grounded. After reaching the desired plane of sedation, the turtle was wrapped with an elastic bandage to minimize movement and secured (using the same elastic bandage) to a firm plastic mesh platform made negatively buoyant with a flexible dive weight attached to the underside. We applied povidone iodine prior to inserting a 27-gauge, 6-mm stainless steel subdermal recording electrode (Rochester Electromedical Inc., FL) 2–3 mm dorsal to the dorsal margin of the right tympanic membrane and < 1 mm under the epidermal skin layer in a rostral-to-caudal direction. Electrodes were coated in clear nail varnish (except for the terminal 3 to 4 mm) for insulation and to prevent water intrusion. As needed, we used a small amount of cyanoacrylate adhesive to secure the electrode in place. We inserted the reference electrode into the subcutaneous tissue lateral to the tail. The ground electrode hung in the water.
After electrode placement, we lowered the turtle platform into the tank on a custom water-filled PVC pipe frame on which we could raise and lower the platform for the turtle to breathe and ensured the turtles were consistently in the same location in the acoustic field for the AEP tests [Fig. 1(A)]. Water depth at the tympanum was approximately 13 cm. The front of the platform, at which we aligned the front of the cranial margin of the plastron, was 45 cm from the test speaker (UW-30; Lubell Labs Inc., OH) located on the opposite side of the tank [Fig 1(A)].
We first evaluated auditory sensitivity by measuring thresholds to 175, 400, 600, 800, and 1000 Hz, as informed by Piniak (2016), using an AEP equipment setup described in Stanley (2020). We used custom labview software (National Instruments, TX) to monitor and record the turtle's auditory responses and to control the test stimulus used to elicit these responses by generating 30-ms tone bursts at the desired frequency with alternating polarities at a rate of 1/10.1 s. For AEP setup details, refer to the supplementary material.1
We started the acoustic stimulus at zero attenuation and reduced the amplitude of the stimulus in 5 dB steps to find the amplitude at which the turtle no longer showed an AEP response. We continued to decrease the stimulus amplitude a minimum of two steps beyond the level at which there was no evoked response in either the waveform [e.g., Fig 2(A)] or the average frequency spectrum in labview. If a tone was audible, a fast Fourier transform (FFT) of the AEP response produced a peak at twice the test frequency [Fig. 3 (Piniak , 2016)], similar to many fishes and invertebrates (Egner and Mann, 2005; Mooney , 2010; Colleye , 2016; Dinh and Radford, 2021). This peak decreased with decreasing stimulus amplitude until reaching levels comparable to the ambient electrophysiological background noise, indicating no detection (Mooney , 2010; Stanley , 2020). The number of responses collected ranged from 250–500 depending on how quickly the average frequency spectrum stabilized.
We imported the averaged AEP waveforms into matlab (vR2019a, MathWorks, MA) to determine auditory thresholds post hoc. Using a custom matlab script, we calculated frequency spectra using 1024-point FFTs of the AEP waveform of the response (e.g., Fig. 3). The peak AEP amplitude (in μV) at twice the test frequency was then plotted relative to the stimulus sound pressure level (SPL) [root-mean squared (RMS); dB re 1 μPa] that elicited that response [Fig. 2(B)], and a regression line was fitted to those decreasing response values. The SPL at which the regression crossed zero (μV) was taken as the theoretical no response and the animal's threshold (Supin and Popov, 2007). Due to frequency binning in the FFT calculation, we chose the frequency that provided the regression line with the highest r-squared value (e.g., for the 400 Hz test frequency, the best line of fit was typically from using either 781 or 796 Hz). In some cases, this frequency was not the maximum peak response across all attenuation levels, as shown in Fig. 3. Regardless, regression lines typically had r-squared values > 0.9, indicating a clear linear decrease in response with decreasing sound level, as noted in other taxa (Supin and Popov, 2007; Mooney , 2010; Pacini , 2010; Mooney , 2020; Dinh and Radford, 2021). We routinely removed suprathreshold and no response points from the regression analysis (Fig. 2; refer to the supplementary material1). Following the final noise exposure trial that had the highest tested SEL, the threshold determination for each of the three turtles required modified approaches because these data presented differently compared to previous exposure sessions; for details, refer to the supplementary material.1
The non-attenuated SPLs of the AEP test tones and the background noise level in the tank were measured at the beginning and end of each test day using an HTI-96-MIN hydrophone (High-Tech Inc., MS; sensitivity: –165.2 dB re 1 V/μPa) placed at the position of the turtle's head on the platform with no turtle present. The hydrophone was attached to an autonomous recorder (Ocean Instruments SoundTrap 4300; –4 dB gain; Ocean Instruments, Auckland, New Zealand). The mean of the two calibrations for each test day was used to calculate the thresholds measured that day (refer to the supplementary material1). Frequency spectra of representative examples of a 400 Hz tone and the AEP-tank ambient noise are shown in Fig. 1(B).
C. TTS experimental design
In general, we conducted sound exposure and control sessions on alternating test days. This allowed us to measure post-exposure and control thresholds while the turtles were under a similar plane of sedation, which would not have been possible if we had conducted AEP testing before and after exposure on a single day. We conducted sound exposures in a circular fiberglass tank [1 m diameter, 0.6 m depth; Fig. 1(C)] with freshwater maintained at 24–26 °C. A UW-30 speaker was centered on the tank bottom with the front face of the speaker pointed upwards; for sound system details, refer to the supplementary material.1 We constructed a plastic mesh (1 cm grid size) cylinder to encase the speaker and turtle to ensure a consistent position of the turtle relative to the speaker [20 cm distance; Fig. 1(C)]. A consistent position of the turtles' heads was maintained during and across sessions by using a small red zip-tie affixed to the base cylinder on which we aligned the turtle's nose. In an initial behavioral survey, we observed unsedated turtles exposed to low noise levels (120 dB re 1 μPa SPL for approximately 5 min), and they displayed no horizontal or vertical movement. In occasional instances when the turtle showed a propensity to move in the apparatus before the trial began, we further secured the turtle in a basket of the same plastic mesh to restrict movement.
To evaluate the conditions that may induce TTS onset (i.e., the lowest SEL at which TTS is predicted to occur) and growth (i.e., how shift magnitude increases with increasing noise exposure), we tested a combination of SPLs and exposure durations comprising a 4 × 4-exposure matrix (16 sessions per turtle). Durations were 5, 10, 20, and 30 min and goal RMS SPLs were 128, 138, 148, and 158 dB re 1 μPa (Table I). This provided a range of SELs between 155 and 193 dB re 1 μPa2 s. The fatiguing sound was white noise bandpass-filtered 50–1000 Hz. Turtles were sedated prior to the exposure to allow AEP measurements directly following the exposure duration. As examples, turtles were sedated approximately 25 min prior to a 20 min exposure and 40 min prior to a 5 min exposure to achieve relatively the same plane of sedation for AEP testing. For exposure durations > 5 min, we brought the turtle to the surface every 5 min to breathe and assess its sedation level. If the turtle indicated the need to breathe by raising its head, we manually raised it to the surface (Harms , 2014). These surface intervals lasted approximately 1–3 s and we considered the noise exposures to be continuous given the brevity of this surface time relative to the length of even the shortest exposure duration.
Goal sound pressure level (dB re 1 μPa) . | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Exposure duration (min) . | 128 . | 138 . | 148 . | 158 . | ||||||||
RES01 . | RES04 . | RES05 . | RES01 . | RES04 . | RES05 . | RES01 . | RES04 . | RES05 . | RES01 . | RES04 . | RES05 . | |
30 | 166 | 162 | 162 | 169 | 168 | 169 | 181 | 182 | 181 | 193 | 193 | 192 |
20 | 160 | 162 | 162 | 165 | 165 | 166 | 178 | 179 | 178 | 190 | 190 | 190 |
10 | 159 | 161 | 160 | 166 | 166 | 166 | 176 | 177 | 176 | 187 | 187 | 187 |
5 | 155 | 155 | 155 | 163 | 161 | 161 | 175 | 176 | 175 | 184 | 184 | 184 |
Goal sound pressure level (dB re 1 μPa) . | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Exposure duration (min) . | 128 . | 138 . | 148 . | 158 . | ||||||||
RES01 . | RES04 . | RES05 . | RES01 . | RES04 . | RES05 . | RES01 . | RES04 . | RES05 . | RES01 . | RES04 . | RES05 . | |
30 | 166 | 162 | 162 | 169 | 168 | 169 | 181 | 182 | 181 | 193 | 193 | 192 |
20 | 160 | 162 | 162 | 165 | 165 | 166 | 178 | 179 | 178 | 190 | 190 | 190 |
10 | 159 | 161 | 160 | 166 | 166 | 166 | 176 | 177 | 176 | 187 | 187 | 187 |
5 | 155 | 155 | 155 | 163 | 161 | 161 | 175 | 176 | 175 | 184 | 184 | 184 |
After the noise exposure, we quickly prepared the turtle for AEP testing, as described above. The time elapsed between the end of the exposure trial and the beginning of auditory testing ranged between 3 and 6 min (mean = 3.8 min ± 0.91 s.d.). All goal SPLs were calibrated before the experimental day, but to account for daily variation we determined the received level of the fatiguing noise to which each turtle was exposed for each noise session. Immediately after turtle removal, we recorded the white noise (for approximately 30 s) for that trial by placing a hydrophone at the red marker (location of the turtles' head and ears). The aforementioned HTI hydrophone and recorder were used for the three lowest goal SPL noise exposures, and an Ocean Instruments ST300 (sensitivity: −176.7 dB re 1 V/μPa) was used to measure and record the highest goal SPL exposures. We calculated the RMS SPL (in dB re 1 μPa) using the time domain of the recorded noise after using a 50–1000 Hz bandpass filter. Including additional frequencies did not appreciably change the calculated SPL. We calculated the noise SEL for each turtle for each exposure trial using the equation provided in Au and Banks (1998) and concatenated repeats of the recorded sound. Refer to the supplementary material1 for additional details. Example frequency spectra of noise conditions and the ambient sound in the noise exposure tank are shown in Fig. 1(D).
To detect TTS and track short-term recovery, we measured auditory thresholds to a 400 Hz tone five times in succession post-noise exposure, taking approximately 1 h. Due to individual differences in breathing patterns during AEP testing, the time elapsed between the end of the exposure and these five measurements varied somewhat between individuals and sessions. We chose 400 Hz because it was the turtles' frequency of best auditory sensitivity (Fig. S11). We compared these post-exposure thresholds to 400-Hz thresholds measured during control sessions.
In control sessions, we used the same protocol as described for the noise exposure sessions, but no sound was played in the exposure tank. Instead, the turtle experienced only the ambient sound [Fig 1(D)]. We conducted 17 control sessions with RES01 and 18 with RES04 and RES05 using the same durations as used in the exposure sessions. For all three individuals we conducted control sessions using 5, 10, 20, and 30 min durations, and 60 min for RES01 and RES05. After removal from the exposure tank we measured five successive 400-Hz control thresholds, as in the noise exposure sessions. The time elapsed between the end of the control duration and the beginning of AEP testing ranged between 1 and 11 min (mean = 3.9 min ± 1.4 s.d.). In addition to the thresholds collected during these control trials, we also included 2 (RES01) and 4 (RES04 and RES05) 400-Hz thresholds collected during audiogram sessions (i.e., no fatiguing noise) during a week when no exposures took place. In total, we collected 76 (RES01) and 94 (RES04 and RES05) control thresholds during the experiment.
Each turtle was tested in one exposure session and one control session per week. In general, the sessions occurred so that exposure and control sessions alternated, with control sessions typically occurring 2 days after noise exposures (in isolated cases, 5 days). There were two occasions when scheduling for all three turtles necessitated conducting two exposure sessions without a control session in between, but these were separated by 5 days.
D. Data analyses
To construct the individual-level, 5-frequency audiograms, we calculated the mean threshold for each frequency for each turtle. We collected 2 to 8 thresholds for frequencies other than 400 Hz, and for 400 Hz we calculated the mean of all control thresholds for each turtle. To construct the species-level audiogram, we calculated the mean and standard deviation of all the thresholds collected across the three individuals for each test frequency.
We compared the mean of all post-exposure thresholds to the mean of all control thresholds for each turtle using a Wilcoxon rank sum test, after testing for normality using a Shapiro-Wilk test (matlab; 0.05 significance level). To evaluate potential TTS, we compared each post-exposure threshold to the mean of the 20 thresholds measured over four adjacent control trials. In general, we used the data from the closest two control trials preceding and the closest two control trials succeeding an exposure trial. In cases where this was not possible, we used the four control trials closest to the exposure trial. We termed these groups of 20 thresholds “control groups,” and there were 9 (RES01) and 13 (RES04 and RES05) control groups in total used to test for significant TTS in the 16 exposure sessions. In one control session for RES05, the thresholds were elevated from an exposure 2 days prior, and the data from this session were excluded from analyses. Following previously used methods (Southall , 2007; Finneran, 2015; Southall , 2019), we determined that TTS occurred when a post-exposure threshold exceeded the mean of the control group (control mean) by > 6 dB. We determined an exposure session to demonstrate TTS when at least one of the five post-exposure thresholds was > 6 dB above the control mean. We calculated threshold shift magnitude by calculating the difference between the five post-exposure thresholds in each session and the control mean used for that exposure session. The highest of these five values was considered the TTS shift magnitude for that exposure session. We compared the mean of these maximum shifts (pooled across all exposures and all turtles) to a similar metric calculated for control sessions. For each turtle, we calculated the difference between the highest threshold in each control session and the overall control mean and compared the mean of these values (pooled across individuals) to the mean TTS shift magnitude using a Wilcoxon rank sum test after testing for normality (Shapiro-Wilk test). To address TTS growth, we tested for linear and exponential relationships between shift magnitude and SEL in r (v. 1.4.1106) for each individual, and also considered the mean response of the three turtles to each exposure level by calculating the mean TTS shift magnitude and mean SEL and similarly tested for a significant relationship. TTS onset was calculated using the best fit mean relationship and interpolating the SEL inducing a shift of 6 dB [e.g., Southall (2007) and Finneran (2015)].
We defined recovery to have occurred when a threshold returned to within 6 dB of the control mean and there was no TTS in any of the following thresholds in that session. If recovery occurred, we calculated time to recovery as the time elapsed between the end of the sound exposure and the temporal middle of the first threshold measurement that was within 6 dB of the control mean. To consider rate of recovery, we plotted threshold shift by time elapsed post-exposure for each individual for only the exposure sessions in which TTS had occurred. For each turtle, the recovery rate appeared to vary depending on if the maximum shift magnitude for an exposure was < 20 dB or > 20 dB. We then combined the data from all individuals and tested for a significant linear and logarithmic regression (p < 0.05) for all pooled data and pooled data from exposure sessions with shifts > 20 dB and < 20 dB, in addition to at the individual level (shifts < 20 dB). For each individual, we also considered the observed fluctuation of thresholds during recovery by creating boxplots of the differences between each of the five post-exposure threshold measurements (i.e., four difference values). Theoretically, one might expect that after removal from the noise exposure the first threshold would be highest, followed by decreasing thresholds with recovery. These changes from threshold to threshold would, in this example, be negative, showing recovery back to the lower baseline auditory threshold. However, positive values in the boxplots demonstrate a deviation from this pattern (non-monotonic recovery). We calculated a net recovery rate by summing the differences between the five post-exposure thresholds for each exposure session where TTS had occurred for each individual and dividing these sums by the times elapsed from removal from the noise until recovery, if it occurred, or the fifth threshold measurement (regardless of if this threshold showed recovery).
III. RESULTS
A. Audiogram and control trials
Across the three turtles, the auditory thresholds were lowest at 175, 400, and 600 Hz (Fig. S11). Highest sensitivity (lowest thresholds) occurred at 400 Hz, with a mean (± s.d.) threshold across the three turtles of 70.3 (5.6) dB re 1 μPa (Fig. S11). The shape of the audiogram and mean thresholds at 200, 600, and 800 Hz were very similar across the three turtles, with greater individual variation in the mean threshold at 400 Hz (range 64.6–74.9 dB re 1 μPa) and 1000 Hz (range 97–107.3 dB re 1 μPa; Fig. S11).
There was relatively low variation in the 400-Hz thresholds collected across all control trials for each turtle. The mean auditory threshold (± s.d.) in dB re 1 μPa was 64.9 (2.9) for RES04, 69.6 (3.4) for RES05, and 74.2 (3.7) for RES01. The mean thresholds across the control groups had a range of (mean ± s.d.): 72.7–78.1 (73.8 ± 1.7) for RES01, 62.9–66.3 (65.2 ± 1.1) for RES04, and 67.4–71.4 (69.3 ± 1.1) for RES05.
Within single control trials, the mean range (± s.d.; in dB re 1 μPa) of the five sequentially measured 400-Hz thresholds was 5.7 (2.2), 6.0 (2.6), and 5.0 (1.9) for RES01, RES04, and RES05, respectively, also indicating consistency of auditory thresholds and limited impact of ambient noise or other factors during testing. The minimum and maximum range of thresholds (in dB) within a single control session was 2.6 and 9.7 for RES01, 2.1 and 10.3 for RES04, and 2.6 and 9.2 for RES05.
B. Noise exposure trials
Temporary hearing loss was observed in all turtles tested. At 14/16 exposure SELs, at least two out of three individuals exhibited TTS (Table I). No TTS occurred for any individual at the lowest SEL (approximately 155 dB re 1 μPa2 s), and TTS was observed in a single individual (RES04) at the second lowest SEL (approximately 160 dB re 1 μPa2 s; Table I). Threshold shifts were observed for all three turtles in 6/16 exposure trials (Table I; at SELs of 166 dB and > approximately 179 dB re 1 μPa2 s). Despite inclusion of post-exposure thresholds representing recovery and noise exposures with no TTS, there was a significant difference for RES04 between the mean post-exposure threshold (74.4 ± 6.4 dB re 1 μPa; range: 62.6–96.2 dB) and control threshold [as above; T = 10 318, z = 10.0, p = 1.3 × 10−23; Fig. 4(B)]. There was also a significant difference for RES05 between the mean post-exposure threshold (75.3 ± 6.1 dB re 1 μPa; range: 65.8–97.9 dB) and control threshold [as above; T = 9267, z = 6.85, p = 7.6 × 10−12; Fig 4(B)]. There was no significant difference for RES01 between the mean post-exposure threshold (77 ± 9.4 dB re 1 μPa; range: 64.9–113.2 dB) and control threshold (as above; T = 6370, z = 1.17, p = 0.24), although the highest post-exposure thresholds exceeded those seen in control sessions [Fig 4(B)]. Within single exposure sessions, the mean range (± s.d.; in dB) of the five sequentially measured 400-Hz thresholds exceeded that observed in control trials, and was 11.5 (5.1), 9.2 (5.1), and 8.2 (6.5) for RES01, RES04, and RES05, respectively. The minimum and maximum range of thresholds within a single noise exposure session was (in dB) 3.2 and 20.6 for RES01, 3.5 and 25.3 for RES04, and 3.2 and 28.6 for RES05.
Maximum shift magnitudes across exposure sessions with TTS ranged from 6.9–40.5 dB (7.4–40.5 for RES01, 8.0–31.0 for RES04, and 6.9–30.5 for RES05). Combining data across the three individuals, there were significant positive linear relationships between shift magnitude and exposure duration (adj. r2 = 0.21, p = 0.0006) and shift magnitude and exposure SPL (adj. r2 = 0.27, p = 9.45 × 10−5; Fig. 5). There was also a significant exponential relationship between mean threshold shift and mean SEL (adj. r2 = 0.64, p = 0.0001; Fig. 6), indicating a predicted TTS onset at 400 Hz of 160 dB re 1 μPa2 s for T. scripta elegans. This relationship between shift magnitude and SEL also held at the individual level. Both linear and exponential relationships were significant for RES01 and RES04, and for each of these individuals the linear relationship explained more variance (RES01: y = 0.69x – 108.4, adj. r2 = 0.64, p = 0.007; RES04: y = 0.32x – 40.8, adj. r2 = 0.23, p = 0.035). For RES05, there was a significant linear relationship between shift magnitude and SEL (y = 0.043x – 64.6, adj. r2 = 0.33, p = 0.012). There was a significant difference (T = 3407, z = 6.5, p = 7.3 × 10−11) between mean pooled TTS magnitude (11.5 dB ± 8.5) and the mean pooled difference between the highest control thresholds and control means [2.6 dB ± 3.0; Fig. 4(A)].
While the three turtles showed generally consistent TTS patterns, there was individual variation (Table I and Figs. 4 and 6). RES04 showed the greatest TTS susceptibility, showing TTS after all exposures except the lowest SEL and in 58 of the 80 thresholds measured after exposure. This turtle also showed the greatest overall sensitivity at 400 Hz. The lowest SEL at which TTS was observed in RES04 was 161 dB re 1 μPa2 s. TTS occurred in RES05 in 11 of the 16 exposure trials, and the lowest SEL that led to TTS was 161 dB re 1 μPa2 s. TTS was observed in 34 of 80 post-exposure thresholds for RES05. RES01 showed the lowest susceptibility to TTS, exhibiting a significant shift in 9 of the 16 sound exposure trials; the lowest SEL where TTS occurred was 166 dB re 1 μPa2 s. TTS was observed in 19 of 78 of RES01's post-exposure thresholds. Interestingly, the order of auditory sensitivity at 400 Hz for each turtle (Fig. S11) reflected the order of the observed TTS sensitivity.
C. Recovery
Auditory thresholds typically recovered to within 6 dB of the mean control thresholds during the approximate hour post-exposure, or by the next test session 2 or 5 days later. However, there was one instance where the testing schedule led to a control session for RES01 and RES05 2 days following the 30 min exposure at 148 dB re 1 μPa. While RES01 recovered, RES05 continued to show elevated thresholds, 8.5–10.6 dB greater than the control mean. Thresholds were no longer significantly elevated 5 days later.
Using only the exposure sessions in which TTS was observed, the pattern of change in shift magnitude over time elapsed post-exposure appeared to separate for each individual into two groups, TTS > 20 dB and TTS < 20 dB [Figs. 7(A)–7(C)]. There was a significant linear relationship between shift magnitude < 20 dB and time elapsed for RES01 (y = –0.17x + 10.0; adj. r2 = 0.28; p = 0.0003) and RES04 (y = –0.092x + 11.1; adj. r2 = 0.19; p = 0.0003), but not for RES05 (y = 0.016x + 5.9; adj. r2 = –0.014; p = 0.53). Thus, using the slopes of these regressions, RES01 recovered the most quickly, followed by RES04, and last RES05. Combining data from all three individuals, there was a significant negative linear relationship between all shift magnitudes and time elapsed data (adj. r2 = 0.08; p = 0.0001), the points with TTS < 20 dB (adj. r2 = 0.10; p = 7.37 × 10−5), and the points with TTS > 20 dB [adj. r2 = 0.46; p = 2.9 × 10−5; Fig 7(D)]. There was also a significant logarithmic relationship between shift magnitude and time elapsed using all data (y = 19.5 – 3.23 × log(x); adj. r2 = 0.08; p = 8.6 × 10−5), TTS < 20 dB (y = 14.5 – 2.4 × log(x); adj. r2 = 0.11; p = 1.9 × 10−5), and TTS > 20 dB [y = 49.0 – 8.6 × log(x); adj. r2 = 0.40; p = 0.00015]. Thus, both relationships fit the data very similarly, but the linear regression is shown in Fig. 7(D) since this model explains slightly more variance than the logarithmic model for TTS > 20 dB.
There were instances where the thresholds measured post-exposure showed a non-monotonic pattern of recovery (Figs. 8 and 9). Specific exposure trial examples that share control groups are shown in Fig. 8 for each individual. Additional examples from RES05 that highlight this observation include a sound exposure of 175 re 1 μPa2 s, after which the second- and fourth-measured thresholds exhibited TTS, yet the first, third, and fifth thresholds did not, and in a sound exposure trial of 166 dB only the third threshold showed TTS. Indeed, RES05 showed the most variable pattern of recovery [Fig. 9(A)], with positive changes in threshold (indicating an increase in a threshold compared to the previous measurement) similar to the negative changes that indicated recovery. This result aligns with the non-significant linear regression for RES05 for recovery from TTS < 20 dB (as above; Fig 7). RES01 showed the least variable pattern of recovery, with mostly negative changes in threshold, and RES04 was moderate between RES01 and RES05 [Fig. 9(A)]. Net recovery rate from exposures with lower shift magnitudes varied, but, in general, there was a significant trend for a faster recovery (in dB/min) with increasing TTS magnitude [r2 = 0.24; p = 0.002; Fig. 9(B)]. This result matched the different recovery slopes observed for TTS < 20 dB and TTS > 20 dB in Fig. 7(D).
There was individual variation in the occurrence and rate of recovery within the approximate hour of AEP testing following sound exposure (Figs. 7–9). This variation reflected the relative susceptibility of the turtles to TTS. RES04 (who showed the greatest TTS susceptibility) did not show recovery during the post-exposure AEP testing in 9 of the 15 exposure trials where TTS occurred, thus recovery required longer than 49–76 min in those sessions. In the 6 trials where TTS recovery was observed during the post-exposure testing, it occurred within 60–70 min. RES01 (who showed the least noise susceptibility) recovered during the post-exposure testing session (within 20–60 min) in 8/9 trials where TTS occurred. Immediate recovery was not observed only after the highest SEL exposure, and the sedation resistance of this turtle prevented a control session to confirm recovery following that final exposure. RES05 was intermediate to these cases. Six of the TTS incidences recovered within 27–62 min and 5 additional cases did not show full recovery by the end of the post-exposure testing, thus recovery required longer than 41–65 min.
IV. DISCUSSION
Here, we provide evidence that a freshwater turtle, T. scripta elegans, experiences a significant temporary reduction in underwater auditory sensitivity after exposure to broadband noise. We saw the occurrence of TTS at all SELs tested except the lowest level (155 dB re 1 μPa2 s), and in at least two, and often all three, individuals at all SELs tested except the lowest two levels. This prevalence of temporary hearing loss underscores concern that similar noise impacts may occur in natural environments. Evidence of underwater acoustic communication (Giles , 2009; Ferrara , 2013) and the prediction that aquatic turtles are most sensitive to sound underwater reflects the importance of audition in their aquatic habitats for enhanced environmental awareness (Christensen-Dalsgaard , 2012; Willis , 2013). Key sound-mediated behaviors and the detection of acoustic cues and signals may be impaired if hearing loss occurs through exposure to anthropogenic noise.
These first hearing loss results for turtles suggest Testudines may be more sensitive to noise exposure than previously understood. The predicted TTS onset of 160 dB re 1 μPa2 s was lower than the scientifically recommended underwater TTS onset threshold for non-impulsive or continuous noise for marine low- and high-frequency cetaceans, phocids, otariids, and sirenians, and freshwater fishes (Smith , 2004b; Popper , 2014; Navy, 2017; Southall , 2019). This emphasizes the limitations of accurately predicting TTS in taxa for which we have limited hearing and noise exposure data or extrapolating across taxa with anatomical and mechanistic differences in the auditory system. Importantly, the SELs at which we observed TTS to occur in this freshwater turtle (Fig. 6 and Table I) are also considerably lower than the predicted threshold of 200 dB re 1 μPa2 s for endangered sea turtles. Those predictions are based on surrogate TTS data from freshwater fishes (Department of the Navy, 2017). However, sea turtles appear to be slightly less sensitive (e.g., higher thresholds at some frequencies) than our results for T. scripta elegans (Piniak , 2016). Sea turtle species also have anatomical differences compared to freshwater turtles [e.g., larger adult body size and layer of subtympanal fat (Willis , 2013)] that challenge our ability to directly predict TTS in marine turtles using these results. These findings suggest our poor understanding of TTS in Testudines may be limiting the effectiveness of noise pollution management across aquatic systems.
The use of a small tank (relative to the 400-Hz wavelength) needs to be considered when extrapolating our results to potential TTS in wild populations undergoing noise exposures in natural water bodies. Here, it is important to take into account the relevant acoustic stimulus generating auditory responses in turtles, which is considered to be sound pressure (Webster , 1992; Popper and Fay, 1997; Saunders , 2000; Hetherington, 2008; Christensen-Dalsgaard , 2012; Willis , 2013; Piniak , 2016). Many Testudine species rely on both aerial and underwater hearing, and a hearing mechanism relying on particle motion would presumably not have a selective advantage for hearing in air where particle motion provides a weak auditory stimulus. Indeed, Willis (2013) support an aquatic origin for Testudines, yet further predict that the middle ear cavity does not impede aerial pressure hearing. The outer and middle ear of the turtle functions similarly to that of other reptiles and birds and as a simplified version of the mammalian middle ear. The tympanum abuts the columella, which is housed in an air-filled middle ear cavity. When an environmental sound wave acts upon this compressible air-filled middle ear, the tympanum vibrates. The columella conducts this movement to the inner ear by pressing upon the oval window, ultimately generating auditory evoked responses. This pressure-detection mechanism is entirely different from animals that detect only particle motion (Popper and Lu, 2000; Mooney , 2010; Popper and Fay, 2011).
However, while not yet directly tested, Hetherington (2008) offer potential mechanisms that may enable particle motion to contribute to audition in turtles to not exclude that possibility, yet ultimately support the likelihood of pressure-sensitive hearing. If evidence emerges that particle motion plays an important role in Testudine hearing, the application of our results to noise exposed animals at distance from noise sources in free field conditions must be made with additional cautions. The turtles in our study were within the predicted region of the speaker where incompressible flow dynamics dominate, as produced by movement of the sound source. If sufficiently large, the vibration of the fluid particles, put into motion by the transducer, could potentially have vibrated the turtles' heads due to their proximity to the speaker (in both the exposure and AEP tanks) (Kalmijn, 1988; Coombs , 1992). If Testudines detect particle motion, and this bulk flow influenced their responses to the noise exposure and/or hearing test tones (although the auditory mechanism is unclear), then our results are less transferrable to natural scenarios where turtles are unlikely to encounter this acoustic condition. However, we quantified the acoustic conditions (pressure) that are most likely responsible for both auditory responses and TTSs given the predominance of evidence supporting the role of pressure in Testudine hearing. Like most TTS studies, the real-world constraints of our system precluded working in free field conditions, yet our tank-based TTS results are broadly transferrable to Testudine underwater hearing given our routine pressure calibrations, well-established TTS and AEP methodologies, and integration of our current understanding of reptilian hearing. In regards to the comparison of our absolute auditory thresholds to hearing studies for turtles and other taxa, study-specific factors such as background sound conditions, equipment, and methodologies should be considered in these threshold comparisons (Ladich and Fay, 2013; Sisneros , 2016; Salas , 2023).
We can compare our results in T. scripta elegans to another reptile using a series of studies on aerial hearing in the alligator lizard (Gerrhonotus multicarinatus) (Tilney , 1982), acknowledging limitations in aerial and underwater audition comparisons. This comparison can be aided by using sensation level, which is the difference (in dB) between the SPL of an exposure noise and the tested animal's auditory threshold (Kastak , 2007). Tilney (1982) measured electrophysiological audiograms both before and after a 24-h sound exposure of 105 dB re 20 μPa. This exposure produced a mean aerial shift of 33 dB in the frequencies with highest auditory sensitivity. Using the approximate auditory threshold across the same frequencies, a 50 dB sensation level produced this 33 dB threshold shift. In our study, the mean shift of 34 dB (range: 31–41 dB) following the 30-min exposure at the goal SPL of 158 dB re 1 μPa most closely matched the shift observed in Tilney (1982). Here, the mean sensation level inducing this shift was 91 dB (range: 86–96 dB). Thus, while not directly comparable, a similar underwater shift was produced in T. scripta elegans as observed in an aerial exposure in G. multicarinatus at a much shorter duration (30 min versus 24 h) but a considerably greater sensation level (91 dB versus 50 dB). Of note, we potentially underestimated the maximum shifts in T. scripta elegans. It required 6 to 14 min to measure the first post-exposure thresholds, and initiation of recovery in G. multicarinatus was observed within 5 min after removal from the sound exposure (Mulroy , 1990; Henry and Mulroy, 1995). While T. scripta elegans showed recovery from underwater TTS in the single frequency tested, G. multicarinatus showed incomplete recovery from aerial TTS 11 days post-exposure (i.e., not all tested frequencies retuned to baseline) (Tilney , 1982). Combined with aerial TTS studies in reptiles, the evidence that turtles are susceptible to TTS from underwater sounds supports the vulnerability of hearing sensitivity in reptiles to both aerial and underwater noise. Our results are for 400 Hz, the frequency with highest auditory sensitivity, similar to other studies (Christensen-Dalsgaard , 2012; Lavender , 2014; Piniak , 2016), and a key next step is to test TTS at other frequencies that may show different patterns.
Investigating the occurrence of TTS in other Testudines, both aquatic and terrestrial, is necessary to understand the generality of our results and under what acoustic conditions hearing loss may occur. We observed increasing threshold shift magnitude with increasing exposure duration and SPL (Fig. 5), supporting our hypotheses. The slopes of the regressions for these relationships were both 0.4, suggesting a similar influence of duration and SPL on TTS magnitude. Shift magnitude for a given range of SELs was highly variable (Fig. 5). For example, for exposure sessions with SELs of approximately 185–193 dB re 1 μPa2 s, we observed the highest TTS (30–40 dB) after exposures with durations of 30 min, but shifts < 10 dB were observed after lower duration exposures in this SEL range. This effect of exposure duration also explains the wide range of shift magnitudes observed across similar SELs in Fig. 5(B). The effect of increasing TTS magnitude with increasing exposure duration in opposition to the equal-energy hypothesis in this reptile joins similar findings in marine mammals and fishes (Kastak , 2005; Kastak , 2007; Mooney , 2009; Finneran , 2010; Popper and Hawkins, 2019). The adjusted r2 values for the relationships between TTS magnitude and duration and TTS magnitude and SPL were also similar at approximately 0.2, indicating that a large amount of variation is not explained by either of these variables. Day-to-day variability in the turtles' physiology and in the experimental setup, even with maintaining consistency as much as possible in sedation, temperature, and equipment, may have contributed to this variation in TTS growth and that observed in control thresholds. The exponential relationship between mean TTS magnitude and SEL explained more variance than either duration or SPL (adj. r2 = 0.64), yet considering Figs. 5 and 6 together, it is clear the combination of exposure SEL, duration, and SPL is necessary to interpret the acoustic conditions that factor into TTS growth [as noted in Finneran (2010)].
The three turtles in our study varied in their susceptibility to TTS, the magnitude of shifts, and time to recovery (Figs. 4, 6, 7, and 9), showing the importance of conducting TTS studies with several individuals, if feasible. Our observations reflect the findings in G. multicarinatus (Mulroy , 1990; Dew , 1993; Henry and Mulroy, 1995), which showed considerable individual variation in the fatiguing noise duration required to produce similar TTS across individuals [e.g., 1 versus 24 min (Mulroy , 1990)]. Also, like G. multicarinatus, the three T. scripta elegans showed individual variation in recovery times and similar time scales of recovery (Mulroy , 1990; Dew , 1993; Mulroy , 1998). Recovery in T. scripta elegans could occur relatively quickly (i.e., < 30 min), but in some instances required longer (in one instance, more than 2 days) depending on exposure and individual. While for marine mammals the relationship between TTS magnitude and recovery time best fits a linear-log equation (Finneran, 2015), here we found a linear-linear equation to provide a slightly better model. Rates of recovery were best described by grouping the data into shifts > 20 dB and shifts < 20 dB, supporting observations of increasing recovery rate with higher initial TTS [as in Finneran (2015)], which is also seen in Fig. 9(B). Yet, recovery from higher shift magnitudes still typically requires more time compared to the more gradual recovery from lower shift magnitudes (Finneran, 2015). This is supported by the equations in Fig. 7(D): when the variable of time is solved for a shift of 6 dB (the threshold of significant TTS), the recovery time for shifts < 20 dB is 41 min compared to 81 min for shifts > 20 dB. The faster recovery for shifts > 20 dB does not counterbalance the higher initial shift. We observed that recovery was not completed by the end of the post-exposure testing session across a range of shift magnitudes [Fig. 9(B)], likely resulting from more complex recovery patterns.
All individuals showed complex recovery patterns [Figs. 6 and 8 (Finneran, 2015)] that contributed to the observed mean recovery rates, and the degree to which thresholds measured during recovery varied depended on the individual. RES01 showed recovery within the post-exposure testing session from every shift except the largest (40 dB), and this aligns with the observation that threshold changes during recovery in this turtle were mostly negative [Fig. 9(A)]. The pattern of a non-monotonic recovery was more prevalent in RES04 and RES05 [Figs. 7 and 9(A)], which likely led to more instances where full recovery required longer than the time span of the post-exposure testing. RES05 was the only individual to show no significant relationship between shift magnitude and time elapsed post-exposure for shifts < 20 dB (Fig. 7). These variable patterns of recovery could be a consequence of the interacting, varied effects of noise to the inner ear. For G. multicarinatus, these effects included damage to hair cells such as microlesions, depolymerization of the cytoskeleton, and displacement and detachment of the hair cell from the cuticular plate, in addition to reduced afferent synapses and synaptic vesicles (Tilney , 1982; Mulroy , 1990; Dew , 1993; Henry and Mulroy, 1995; Mulroy , 1998). If these sources of damage co-occur, parallel recovery mechanisms may be operating at different rates that vary by individual and could lead to variable patterns of recovery at the level of threshold measurements. These findings for G. mulitcarinatus indicate that mechanical damage can occur in reptiles in even temporary hearing loss, potentially with lasting effects in noise-exposed animals (Kujawa and Liberman, 2009). We saw that the variable of time elapsed post-exposure significantly explained 8% to 46% of the variation in recovery, and we would expect there to be additional variables at play in a study extending over nine months in these ectothermic animals with seasonal patterns in metabolism. The repair mechanisms for potential sources of damage may vary by slight variations in temperature, metabolic state, or season.
It is interesting to consider the implications of these results for T. scripta elegans and other closely related species in their freshwater habitats. The turtles in our study showed no perceivable agitation or behavioral response to the presence of the fatiguing noise in the initial testing period or during the exposures that differed from their behavior in the control sessions. The sedation level was not sufficient to prohibit reactions, especially in the longer exposures where the turtles were exhibiting no or only minimal sedation effects at the initiation of the noise. This is similar to observations of Malaclemys terrapin (diamondback terrapin) showing no behavioral responses to playbacks of boat engine noise (Lester , 2013). While difficult to compare sound sources measured in different environments, the sound levels tested in this study were similar to those recorded from small boat engines at some distances (Lester , 2013; McCormick , 2019). Taken together, the susceptibility to TTS at relatively low SELs [with respect to current noise exposure criteria (Department of the Navy, 2017)] and seemingly low avoidance of high amplitude noises suggests freshwater turtles could potentially experience TTS in noise-polluted natural habitats even if they do not demonstrate an obvious behavioral response. Thus, it may not be safe to assume that turtles will avoid or move away from high amplitude sounds that might induce hearing loss.
Our findings warrant caution and awareness that threatened and endangered freshwater, estuarine, and sea turtles may incur hearing loss at lower sound levels than currently estimated. Diamondback terrapins and sea turtles inhabit coastal and offshore marine habitats, with spatiotemporal overlap with a variety of noise-producing anthropogenic activities. These impacts range from chronic daily inundation of small boat noise to high-amplitude construction activities such as pile driving for pier or wind farm construction. Increasing anthropogenic influence in our oceans heightens both noise pollution and the predicted auditory impacts to aquatic species. The data presented here suggest we need further initiatives to address the effects of varied noise sources and exposure regimes to inform noise mitigation strategies for these diverse Testudine species with great conservation needs.
ACKNOWLEDGMENTS
We thank Maggie Aydlett and Seth Cones (turtle husbandry assistance), Rick Galat, Edward Doherty, and the WHOI Facilities team (general support), and the Rainforest Reptile Refuge and Animal Sanctuary (Beverly, MA; providing two of the study turtles). Work was permitted under the Massachusetts Division of Fisheries and Wildlife, and we thank them, specifically Michael T. Jones (Massachusetts State Biologist), for their support and guidance. Research was supported by the U.S. Navy's Living Marine Resources Program and the National Oceanographic and Atmospheric Administration's National Marine Fisheries Service.
See supplementary material at https://doi.org/10.1121/10.0020588 for additional details regarding turtle husbandry, sedation, blood parameter methods and results, AEP methodology, and sound level measurements.