Input compensation of dolphin and sea lion auditory brainstem responses using frequency-modulated up-chirps

Auditory brainstem response (ABR) measurements have greatly expanded the number of marine mammal species for which hearing data are available over the past 40 years (Supin , 2001; Mooney , 2012; Jäckel , 2022). Despite their advantages in facilitating hearing threshold measurements with untrained subjects, they are limited in many regards relative to behavioral psychophysical methods that are considered the “gold standard” (Houser , 2017). Some of the limitations of ABR methods arise from the specific nature of the response being measured: ABRs measured near the surface of a marine mammal's skin represent the summed voltage generated by the ascending auditory brainstem in response to acoustic stimulation following volume conduction and attenuation through tissues in the head. ABR amplitudes [and signal-to-noise ratio (SNR) in recordings] therefore depend on factors that may be largely independent of auditory detection. For example, the low brain-to-body mass ratios in very large marine mammals [e.g., baleen whales, killer whales (Orcinus orca), northern elephant seals (Mirounga angustirostris)] leads to smaller summed potentials and less chance of detection by experimenters (Szymanski , 1999; Ridgway and Carder, 2001; Supin , 2001; Houser , 2008; Lucke , 2016; Finneran , 2017; Houser , 2019). Methods that might provide an increase in SNR in these often-challenging testing scenarios would aid the development of ABR methods for larger, more exotic species for which hearing data are either scarce or not available.

Interest in increasing the SNR of human ABR recordings led to the development of frequency-modulated (FM) “chirp” stimuli that increase in frequency over time (Shore and Nuttall, 1985; Dau , 2000). Conceptually, these stimuli act as a form of “input compensation” for the features of the traveling wave on the basilar membrane. Cochlear regions that are stimulated first (closest to the oval window/stapes footplate) are tuned to high frequencies, while apical regions with low-frequency tuning are stimulated later in time. By “chirping” a stimulus such that it increases in frequency over time at a rate matched to the traveling wave speed (TWS) on the basilar membrane, all frequency regions mapped by the basilar membrane could be stimulated simultaneously. This should lead to synchronous firing of auditory neurons across the entire bandwidth of the basilar membrane and produce higher-amplitude ABRs through voltage summation of frequency-specific responses. If the frequency sweep rate perfectly compensated for basilar membrane traveling wave characteristics, it would theoretically provide the maximum amplitude ABR.

Two previous studies with delphinid species have demonstrated that broadband chirp stimuli increase the amplitude of ABRs relative to non-FM broadband stimuli [bottlenose dolphin (Tursiops truncatus) (Finneran , 2017); killer whale (Houser , 2019)]. These studies generated chirps using a simple exponential function based on the simplifying approximation of a constant TWS (in ms/octave) in the inner ear of these odontocete species. Both studies found an increase in ABR peak amplitudes for the FM chirp stimuli relative to more traditional broadband clicks.

A more recent study (Finneran , 2022) used a high-pass subtractive masking technique (Teas , 1962; Don and Eggermont, 1978; Parker and Thornton, 1978; Finneran , 2016) to derive frequency-specific derived band ABRs in bottlenose dolphins and California sea lions (Zalophus californianus). The ABR peak latencies as a function of derived-band center frequency were fit with a power function to more accurately estimate TWS. By summing these derived-band ABRs after accounting for the cochlear dispersion described by the power function [the “stacked ABR” method, Don and Elberling (1994) and Don and Kwong (2009)], the authors created an ABR approximating in-phase firing of all frequency-specific components of the full-bandwidth ABR. An “optimal” ABR—reflecting this alignment of frequency-specific responses—could hypothetically be recorded using an “optimal” chirp based on the estimated TWS as a function of frequency [although the optimal values may depend on stimulus SPL, as slower chirp rates tend to be more effective as stimulus level decreases (Elberling , 2010; Finneran , 2017)].

Gains in ABR peak amplitudes resulting from the stacked ABR method were found by Finneran (2022) to be larger for sea lions than for dolphins. The authors concluded that this resulted from the relatively lower TWS in the sea lions. Even in the absence of stimulus FM, the rapid TWS in the bottlenose dolphin results in an ABR that features a large degree of in-phase firing across frequency. Thus, the largest gains for FM chirp stimuli might be observed for species with lower-frequency hearing specialization [this is consistent with stacked-ABR gains in humans (Don , 1997; Don and Kwong, 2009), which also have lower frequency hearing capabilities than dolphins].

The current study was designed to follow up upon the conclusions of Finneran (2022) regarding the potential advantages of FM chirp stimuli in marine mammals with various frequency ranges of hearing. Measurements were made with bottlenose dolphins and California sea lions, two species that are broadly representative of larger marine mammal hearing groups [high-frequency cetaceans and otariid carnivores, respectively (Southall , 2019)]. The first aim of the study was to design an “optimal” broadband chirp for the dolphins and sea lions based on the power function fits of derived-band ABR data from Finneran (2022). A high-pass subtractive masking technique was then used with the dolphins to visualize the alignment of ABR peak latencies across frequency when using the optimal chirp [see de Boer (2022) for a comparable technique in humans]. The second part of the study compared input/output functions (i.e., ABR peak amplitude vs stimulus level) for broadband chirps and noisebursts to examine the gains afforded by chirps relative to non-FM stimuli.

The subjects of the study were three Atlantic bottlenose dolphins (IDs: SHA, SPO, and TRO) and two California sea lions (IDs: MTY and TUT). All of the subjects were considered to have a normal hearing range characteristic for their species based on intermittently conducted hearing tests in the dolphins [see Table I and Strahan (2020)] and the comparisons of the amplitudes and morphologies of the sea lions' ABRs with those previously obtained with normal hearing individuals (Mulsow and Reichmuth, 2013; Mulsow , 2014). All experiments followed a protocol approved by the Institutional Animal Care and Use Committee at the Naval Information Warfare Center Pacific and the Navy Bureau of Medicine.

Dolphin testing was conducted underwater in 9 m × 9 m floating netted enclosures at the U.S. Navy Marine Mammal Program in San Diego Bay, California. The dolphins were trained to position themselves on an underwater “biteplate” station located 1 m underwater and attached to the floating enclosures with an aluminum frame. Two piezoelectric transducers (LL916, Lubell Labs, Whitehall, OH and ITC 1042, International Transducer Corp, Santa Barbara, CA, used simultaneously to increase stimulus bandwidth) were fixed approximately 1 m in front of the dolphins and projected all stimuli. The ITC 1042 was placed against and in the center of the LL916 to minimize time of arrival differences for the stimuli. Underwater ambient noise was dominated by snapping shrimp, shipping traffic, and the dolphins themselves. Median pressure spectral density levels during testing were ∼67 dB re 1 μPa²/Hz at 20 kHz and decreased linearly with the logarithm of frequency to ∼55 dB re 1 μPa²/Hz at 150 kHz.

Sea lions were tested in air inside the veterinary facility at the U.S. Navy Marine Mammal Program. The subjects were in sternal recumbency and under general anesthesia with sevoflurane gas (1.8%–2%) following sedation via intramuscular injection of midazolam (0.3 mg/kg) and meperidine (1 mg/kg). Ephedrine (0.05 mg/kg), dobutamine (0.1–0.2 μg/kg/min), and lithium chloride (0.01 mmol/kg) were administered intermittently as directed by the attending veterinarian. Stimuli were delivered into both ears using HDA300 headphones (Sennheiser Electronic, Wedemark, Germany). Typical ambient noise conditions inside the veterinary facility (with a comparable equipment arrangement) were reported by Mulsow (2011). Power spectral densities decreased approximately linearly, with levels [in dB re (20 μPa)²/Hz] at 10 dB near 1 kHz, 0 dB at 3 kHz, −10 dB at 10 kHz, and −18 dB at 20 kHz.

“Pink-equalized” chirps and noisebursts were used over the course of the study. Spectral densities decreased at 3 dB per octave to equate noise energy across regions of the cochlea, i.e., equal sound pressure levels (SPLs) in each one-third octave band. Pink-equalized stimuli have also been shown to provide the best separation of ABR amplitude and latency values from normal-hearing and hearing-impaired dolphins [relative to “white-equalized” stimuli (Mulsow , 2016)]. Hereafter, SPL will refer to the average one-third octave levels over the stimulus or noise bandwidth: in dB re 1 μPa for dolphins underwater and in dB re 20 μPa for sea lions in air. The general aspects of stimulus production that were common to all stimuli are given in this section while the details specific to a particular segment of the study will be given in the corresponding segment. Short-duration noisebursts with equal bandwidth and amplitude were included for comparison to chirp ABRs. This comparison ideally demonstrates the gain in ABR amplitude provided by the chirps' compensation of frequency-specific acoustic delays in the cochlea.

All stimuli with the dolphins were generated using a National Instruments (NI) USB-6251 data acquisition (DAQ) device (0.5–1 MHz, 16-bit, National Instruments Corporation, Austin, TX) and presented at rates of ∼25 s⁻¹. The stimuli were anti-alias filtered (3C, Krohn-Hite Corporation, Brockton, MA) and attenuated [PA5, Tucker-Davis Technologies (TDT), Alachua, FL] before being sent to a TDT SM5 signal mixer. Stimuli were amplified using a CC4000 amplifier (Crest Electronics Corporation, Meridian, MS), and the signal was sent in parallel to the LL916 and ITC 1042.

Chirps used for the dolphins were generated using Eq. (1), with pink-equalized frequency content from 4 to 152 kHz (Fig. 1). The sweep rate parameter k varied from 3 to 12 and included the optimal k = 6.4 (Finneran , 2022). Noisebursts were likewise generated with pink-equalized spectral content from 4 to 152 kHz, with durations of 314 μs (32-μs rise/fall, 250-μs plateau). Each noiseburst presentation used a unique waveform. Stimuli for the dolphins were calibrated underwater (without the dolphin present) using a TC4013 hydrophone (Teledyne Reson, Slangerup, Denmark) placed midway between the posterior portion of the mandibles. The hydrophone signal was sent to a VP1000 voltage preamplifier (Teledyne Reson) and digitized by the NI USB data acquisition device (1-MHz sample rate, 16 bit). The broadband chirps and noisebursts were finally equalized to provide spectrally pink conditions after accounting for the frequency response of the projectors and the underwater environment.

For the sea lions, stimuli were generated using a NI USB-6259 DAQ (0.5–1 MHz, 16 bit) and presented at rates of ∼21 s⁻¹. Stimuli were anti-alias filtered with a TDT FT5 filter (120-kHz lowpass), attenuated with a TDT PA4, and sent to a TDT SM5 signal mixer (to provide 20 dB of gain) and a TDT HB5 headphone buffer amplifier. The stimuli were projected diotically through a set of Sennheiser HDA300 headphones (Sennheiser Corporation, Wedemark, Germany). Like those used for the dolphins, the sea lion chirps were generated using Eq. (1) and featured pink-equalized frequency content from 0.5 to 32 kHz and sweep rate parameter k varying from 1 to 8, including the optimal chirp of k = 3.4 (Finneran , 2022). Noiseburst stimuli featured pink-equalized spectra from 0.5 to 32 kHz, durations of 1.256 ms (128-μs linear rise/fall times, 1-ms plateau), with each noise burst representing a unique waveform.

Acoustic calibrations for the sea lions were conducted under the HDA300 headphones while not on the sea lion. Signals were measured using an ER7C microphone (Etymotic Research, Elk Grove Village, IL, corrected for frequency response), filtered using a TDT FT5 filter module (150-kHz lowpass), and digitized with the NI USB-6259 DAQ (1-MHz sample rate, 16 bit). Stimuli for the sea lions were calibrated in terms of average SPL (dB re 20 μPa).

The average SPL across the stimulus bandwidth was used as the level metric during data collection with all stimuli. However, since the chirps and the noisebursts have different durations, the energy content was not equal for stimuli equated in terms of SPL. As there is evidence that, in some cases, ABR amplitudes depend on the stimulus energy within a “temporal window” (Jones , 2019; Finneran , 2020), comparisons between chirps and noisebursts were additionally made in terms of sound exposure level [SEL, dB re 1 μPa² s for dolphins, dB re (20) μPa² s for sea lions], which represents the integral of squared acoustic pressure with respect to time.

A set of measurements examined the contributions to the dolphin ABR from specific frequency regions of cochlear activation (this experiment was not conducted in the sea lions due to time limitations associated with the anesthetic procedures in that species). Here, the highpass subtractive masking technique was used (Teas , 1962; Don and Eggermont, 1978; Laukli and Mair, 1985; Finneran , 2016; De Boer , 2022). The masking noise was generated as a 3-s sample that was continuously looped using a NI USB-6259 (500-kHz, 16 bit) and projected through the same signal chain described above for the other stimuli at a level of 115 dB re 1 μPa (with a TDT FT5 anti-alias filter instead of the Krohn-Hite 3C module). Multiple bandpass conditions were used: each had a lowpass cutoff of 152 kHz while highpass cutoffs varied from 4 to 128 kHz in 1/2-octave increments. As 152 kHz exceeds the upper-frequency hearing limit in bottlenose dolphins (Johnson, 1966; Strahan , 2020), the noise can effectively be considered a highpass masker. All conditions were pink equalized in a manner similar to that of the chirps and noisebursts.

Gold cup surface electrodes, embedded in silicone suction cups and placed on the skin of the dolphins, were used to measure ABRs. The non-inverting electrode was placed along the dorsal midline approximately 5 cm behind the blowhole. The inverting electrode was placed next to the right external auditory meatus and a common ground was placed in the seawater next to the aluminum frame. Subcutaneous needle electrodes (14 × 0.38 mm, Natus Neurology, Pleasanton, CA) were used for the sea lions. The non-inverting electrode was placed between the ears at the vertex of the head, the inverting electrode at the base of the neck between the scapula, and the common electrode on the back near the left scapula. The incoming signals from the electrodes were filtered (dolphin: 300–3000 Hz, sea lion: 30–3000 Hz) and amplified (94 dB) with a ICP511 biopotential amplifier (Grass Technologies, West Warwick, RI). The differential voltage between the inverting and non-inverting electrodes was digitized at 100 kHz using either a USB-6251, USB-6259, or USB-6363 DAQ. Average ABRs, synchronized to the onset of the stimuli, were generated from 1024 individual stimulus presentations for each species. One exception was made during the derived-band procedure with the dolphins as the subtractive process of generating derived-band ABRs increases background noise. For these highpass masking noise measurements, four 1024-epoch records were coherently averaged together to create a final, lower-noise ABR (a “grand average”). The ABRs for chirps were adjusted by setting “time zero” to the chirp offset (Fobel and Dau, 2004; Elberling and Don, 2010), while the onset of the stimulus was used for noisebursts (all cases included the additional acoustic delay).

Individual ABR waveforms were digitally low-pass filtered at 3 kHz using a zero-phase implementation of a Butterworth filter (Virtanen , 2020). Derived-band ABRs were created for each dolphin through point-by-point subtraction of grand average ABRs obtained under high-pass masking noise conditions separated by one-half octave. For labeling, the average and derived-band ABRs were visualized along with two “subaverages,” each of which represented the average of half of the total sweeps for each condition. The ABR peaks comprising the auditory nerve response, P1 and N2 (dolphins only; P1 tends to be comparatively low-amplitude in sea lions) and the later peaks, P4 and N5 (both dolphins and sea lions) were identified visually (when possible) as not all peaks were identifiable in all animals and conditions (Fig. 2). Measurements were reported as the peak-to-peak amplitudes of the P1-N2 and P4-N5 complexes. Relevant details regarding the generators of these ABR peaks have previously been described for odontocete cetaceans [e.g., Popov and Supin (1990), Supin (2001), Mulsow (2018), and Mulsow (2020)] and sea lions [Mulsow and Reichmuth (2013), although P4-N5 was previously labeled as P6-N7].

Examples of ABRs for chirps with various sweep rates with the dolphin SHA are shown in the left panel of Fig. 2 (chirp SPL of 100 dB re 1 μPa). At the fastest chirp sweep rates (low k values), the ABRs were essentially identical to those typically reported for this species for non-chirp stimuli (e.g., clicks and tonebursts). With decreasing chirp sweep rate (high k values), the ABRs became more distorted and the amplitudes of waves became progressively smaller.

Figure 3 shows mean P1-N2 and P4-N5 amplitudes across chirp sweep rate at four SPLs for the dolphins. At 100 dB re 1 μPa, there was a modest increase in wave amplitudes up to a value of approximately k = 6–7, beyond which amplitudes decreased. In a few cases, the mean amplitudes of ABR peaks to the optimal chirp (k = 6.4) were slightly smaller than those to adjacent values of k. This was mostly a result of lower amplitudes at this k value for the subject SHA as opposed to a consistent pattern across all dolphins. At lower chirp SPLs, the functions became more compressed and there was less difference across all k values for both wave complexes (although the largest amplitudes remained at the lower k values).

ABRs as a function of chirp sweep rate for the sea lion TUT are shown in the right panel of Fig. 2 (for a chirp SPL of 40 dB re 20 μPa). Like the dolphins, as chirp sweep rates became progressively slower (higher k value), the ABRs became less defined and featured broader P4-N5 complexes. The amplitudes of the sea lion ABR peaks across k and SPL are shown in Fig. 4. As the SPLs differed for each sea lion, the amplitudes were not averaged across individuals. Amplitudes were larger for TUT than for MTY at nearly all k values, despite lower chirp SPLs (e.g., 40 vs 45 dB re 20 μPa, respectively). Nonetheless, the patterns for the two subjects were similar, and comparable to those found with the dolphins. Namely, ABR amplitudes appeared to decrease with increasing k at values above approximately k = 2 for the 40 and 45 dB re 20 μPa stimuli, but this trend was not apparent in the lower SPLs of 30 and 35 dB re 20 μPa. Like the trends seen with the dolphins, k values less than that of the optimal chirp resulted in the largest ABRs at the highest tested SPLs.

Figure 5 shows derived-band ABRs obtained with optimal chirp stimuli (95 dB re 1 μPa) in the dolphin SPO. Mean derived-band ABR latencies and peak amplitudes for all dolphins are shown in Fig. 6. Latencies for both peaks changed little over the included frequency bands; less than ∼220 μs, with slightly more change in P4 and N5 as compared to P1. For comparison, the top panel of Fig. 6 shows the adjusted derived-band latencies for ABRs to noisebursts (i.e., non-FM stimuli) from Finneran (2022). For these data, a constant time was subtracted from the data for each peak such that the latencies at the 128-kHz band matched those for the chirp-evoked ABR peaks from the current study. The changes in latency over the range of derived bands for the data of Finneran (2022) were notably greater than those for the chirp-evoked ABRs.

The amplitudes of the derived-band ABR peaks in the current study generally increased with increasing frequency; however, there was a consistent decrease (i.e., in all single subject data) in the P4-N5 amplitudes around 32 kHz. Variability in amplitude was higher for both P1-N2 and P4-N5 in the higher-frequency derived bands, a pattern not apparent in the latency data.

Waveforms for noiseburst and optimal-chirp ABRs at matched SPLs for dolphin SHA are shown in Fig. 7 and average peak amplitudes from comparable data for all three dolphins are plotted in Fig. 8. When equated in terms of SPL, the average input-output functions shown in Fig. 8 demonstrate the persistence of the chirp ABR at lower SPLs than those for the noiseburst (an approximately 20 dB difference). For P1-N2, the optimal chirp evoked larger ABRs at all tested SPLs. While the P4-N5 amplitudes were larger for the optimal chirp at SPLs of 100 dB re 1 μPa and below, they decreased with increasing SPL above 100 dB re 1 μPa and were lower in amplitude than those for noisebursts at the highest tested SPLs of 120 and 125 dB re 1 μPa. This result can be seen in the Fig. 7 waveforms for SPLs above 100 dB re 1 μPa, where the morphology of the later waves in the ABR differed from those observed for noiseburst and chirp ABRs elicited by lower-level stimuli. These general trends were the same when stimuli were instead equated in terms of SEL; however, the functions were shifted closer together (a 5–10 dB difference) as threshold was approached, and the crossover point at higher levels shifted accordingly.

The amplitude gain of the chirp ABR peaks relative to those for the noisebursts are given for the dolphins in Table II. The gains provided by the chirps were largest at the lowest stimulus levels and decreased with increasing level. Gains became negative at the highest levels for the mean data (with the single exception P1-N2 equated in terms of SPL). In contrast to the dolphins, the sea lion ABRs to the optimal chirps were larger at all tested levels when equated in terms of SPL (waveforms in Fig. 9, peak amplitudes in Fig. 10, amplitude gain in Table III). Nonetheless, there was a similar decrease in the chirp ABR amplitudes at the highest levels, and the amplitudes were approximately equal at the highest levels for MTY when—in order to account for stimulus energy differences—levels were equated in terms of SEL (TUT was not tested at these stimulus levels). Using SPL, there was a 15–20 dB difference between the point at which the peaks for each condition were no longer discernable from background electrical noise. When equated in terms of SEL, this difference was closer to 5–10 dB.

Chirps designed to take into account changes in cochlear traveling wave speed in both species (Finneran , 2022) produced similar patterns to those from an earlier dolphin study using a chirp based on a simple exponential function (Finneran , 2017). First, at the highest chirp SPLs, there was an initial increase in peak amplitudes from the lowest k values up to a local maximum, followed by a more dramatic decrease with further increases in k. Second, decreasing chirp SPL produced progressively compressed functions with more similar ABR peak amplitudes values across chirp sweep rate. The current study did not, however, show that an increase in k values (i.e., slower sweep rate, longer chirp) produced the largest ABR with decreasing stimulus SPL as was reported for dolphins (Finneran , 2017) and humans (Elberling and Don, 2010). Nonetheless, the compression that was observed in the current data at the lower stimulus SPLs did result in more comparable ABR peak amplitudes across the tested k values.

Contrary to expectations, the empirically derived, optimal k parameter Finneran (2022) did not produce the largest ABR peaks in either species, despite being based on an improved fit of derived-band latencies relative to that used earlier by Finneran (2017). The underlying cause for this can be seen in the latencies for the dolphin derived-band ABRs (i.e., Fig. 6), where latencies were not precisely aligned [although they appear relatively well-aligned when compared with derived-band latencies obtained with chirps in humans (De Boer , 2022)]. Better alignment of the higher-frequency components of the ABR, which contribute disproportionately to the ABR amplitude, would in this case be achieved through a chirp with a faster sweep rate relative to the optimal sweep rate obtained when k = 6.4. This is in fact seen for the slightly larger ABR peak amplitudes at k values just below the optimum in Fig. 3. It is counterintuitive that the alignment was better for the dolphins' P1 than it was for P4 and N5, as the latter two peaks were used to obtain the optimal k (Finneran , 2022). While there is not a clear explanation for this result, it is worth noting that the ranges over which these frequency specific P4 and P5 latencies lie (less than 220 μs) are still relatively small compared to the dominant period of the ABR (approximately 1 ms inter-peak interval). From a practical point of view, the reduction in destructive interference that would be afforded by perfect alignment of these responses is certainly very small.

It is worth noting the methodological specifics used by Finneran (2022) in determining the optimal chirp parameters. That study used noiseburst SPLs of 115 and 125 dB re 1 μPa for dolphins and 65 dB re 20 μPa for sea lions, which are above the range of SPLs used to derive the amplitude-vs-k functions in this study. The general pattern of increasing optimal k with decreasing stimulus SPL [observed in dolphins by Finneran (2017) and humans by Elberling (2010)] would, however, predict that the largest ABR amplitudes exist at k values higher than the optimal values used in the current study. In fact, lower k values tended to produce larger ABR peaks. The optimal values for k from Finneran (2022) based on curve fits to single-subject derived-band ABR data were 6.6, 5.7, and 6.3 for SHA, SPO, and TRO (mean = 6.2). Two dolphins (SPA and WHP) tested by Finneran (2022) but not included in the present study had optimal k values of 6.4 and 8.0, respectively. The “true” optimal chirp for dolphins appears to lie at a k value lower than 6.4, but this does not appear to be a result of differences between individually estimated k values and the mean value used here. Neither of the sea lions tested by Finneran (2022) were included in this study. Their individual k values of 3.5 and 2.6 were comparable or lower than the optimal k = 3.4 used here, but the small sample size and differences between subjects and methods for both species preclude a definitive comparison.

The comparison of the noiseburst and chirp stimuli was conducted to examine the gains in ABR peak amplitude afforded by the phase alignment of ABR components when using a chirp. The large gains in peak amplitudes at the low to moderate SPLs in both dolphins and sea lions were unexpected based on the gains that were seen using the stacked ABR “output compensation” based on derived-band ABRs (Finneran , 2022). The gains reported for the stacked ABR from Finneran (2022) should represent what would be expected from an ideal chirp relative to a comparable stimulus without FM. These gains ranged from 20% to 30% for P1-N2 for the dolphins to 119% for P1-N2 in the sea lions. The reason that the gains for the chirps over the noisebursts in the current study were—in nearly all cases—larger than those observed for what should be the “ideal” situation in the stacked ABR is not known. It may to some degree be due to the differences in SPL between the current study (the largest gains are below 100 dB re 1 μPa) and those used by Finneran (2022) (115 and 125 dB re 1 μPa). A reduction in the gains afforded by chirps at high stimulus levels has been previously observed in humans and attributed to the upward spread of cochlear excitation (Dau , 2000; Elberling , 2010). It is worth noting that the magnitude of the gains observed in the present study are largely in line with some of those reported for stacked ABR procedures with humans [100%–300% (Don , 1997; Don and Kwong, 2009)].

Based on their results from stacked ABR procedures, Finneran (2022) predicted that increases in chirp ABR peak amplitudes for sea lions would be greater than those for dolphins. There did not, however, appear to be a large difference in the gains afforded by the chirp in the sea lions vs the dolphins. Mean gains were larger for the dolphins when stimuli were equated in terms of SPL but larger for the sea lions when equated in terms of SEL, while the pattern of response amplitude decline into noise was not markedly different for the two species (although there was a modest difference when equating in terms of SEL with the sea lions). The current results therefore do not provide any clear evidence of a notable advantage when using chirp stimuli in marine mammals with lower-frequency hearing than odontocete cetaceans (e.g., sea lions, or to a potentially greater degree, baleen whales).

As has been the case for previous studies using chirp stimuli, the metric chosen to represent stimulus amplitude was a difficult consideration. Studies with humans and marine mammals have used various metrics to quantify the levels of chirps, including sensation level (i.e., level above threshold), peak-equivalent SPL, spectral level, energy spectral density level, and total energy (Dau , 2000; Elberling and Don, 2010; Finneran , 2017; Houser , 2019; De Boer , 2022). The difficulty in defining stimulus levels for chirps that are longer than the temporal window of ∼250 μs that defines dolphin ABR amplitudes and latencies is an additional complicating factor (Jones , 2019; Finneran , 2020). In the practical application of chirps to marine mammal audiometry (i.e., obtaining an ABR in subjects that are difficult to test due to their large body size), this calibration issue might, however, be of reduced importance relative to the larger goal of estimating a species frequency-specific hearing curve (see Sec. V).

The crossover points in the input/output functions at the highest tested chirp levels beyond which ABR amplitudes were lower than those for equivalent noiseburst levels was an interesting finding. That this effect was more prominent in the P4-N5 data than the P1-N2 data were reminiscent of data from Finneran (2020) where offset responses for noisebursts >256 μs in duration destructively interfered with and reduced P4-N5 amplitudes [but not with P1-N2 amplitudes; see also Mulsow (2021)]. How an offset response might appear for a chirp stimulus that sweeps across cochlear places over time is, however, difficult to envision, i.e., what time within a chirp should be used for the offset of a particular frequency region in the cochlea? What is the effective time of stimulation within a particular cochlear place when using chirps of varying duration? It is difficult to say if a common mechanism related to offset responses underlies the pattern observed in the current study using FM stimuli.

The overarching goal of recent marine mammal studies using chirps has been to provide a means of maximizing ABR amplitudes, specifically to increase SNR in subjects that do not provide optimal testing conditions (e.g., baleen whales that have relatively small brain-to-body mass ratios). The current study has further demonstrated that broadband chirps near (and below) the optimal k values, through synchronization of the timing of ABR peaks across cochlear place, generally provide larger ABR peak amplitudes for stimuli equated in terms of SPL or SEL. This did not prove to be a greater advantage for species with better low-frequency hearing as originally hypothesized. Considerations such as the reduction in peak amplitudes for chirps at the highest stimulus levels and the difficulties in equating levels across chirps and other more traditional stimuli further complicate a simple application to marine mammal hearing studies. Chirp stimuli might therefore be best considered as an additional tool for hearing tests with large exotic species. For example, if recordings with, e.g., broadband clicks at the peak output limit of the testing hardware do not provide an ABR with sufficient SNR, a pre-designed short-duration chirp (i.e., less than the estimated optimal k) may provide a means of visualizing the ABR in challenging test conditions.

The authors thank K. Donohoe, E. McGarvey, R. Dear, M. Wilson, H. Bateman, C. Espinoza, G. Goya, M. Graves, D. Ram, T. Wu, and the animal care staff, training staff, and interns at the U.S. Navy Marine Mammal Program. Madilyn Pardini and two anonymous reviewers provided feedback that greatly improved the quality of this manuscript. This study was supported through funding by the United States Living Marine Resources (LMR) Program. The authors have no conflicts of interest to disclose. The data that support the findings of this study are available from the corresponding author upon reasonable request. This is scientific contribution 361 of the National Marine Mammal Foundation.