Auditory brainstem responses during aerial testing with bottlenose dolphins (Tursiops truncatus): Effects of electrode and jawphone locations

The auditory brainstem response (ABR) recorded at the skin surface consists of voltage deflections that are generated by synchronous firing of neurons in the ascending auditory nervous system following acoustic stimulation (Jewett et al., 1970; Hall, 2007). Due to the hypertrophy of auditory structures in dolphins and porpoises (toothed whales, or odontocetes), the ABR is generally at a high amplitude (on the order of many microvolts) when recorded from non-invasive electrodes at the skin surface (Ridgway et al., 1981; Supin et al., 2001). The ABR does not require active attention on the part of the subject, and its presence is resilient to sedatives (Hall, 2007), therefore the ABR [and the related auditory steady-state response (ASSR)/envelope-following response (EFR); Popov and Supin, 1998] provides an efficient method for testing hearing in odontocetes. These methods have been used to determine frequency-specific hearing thresholds following stranding (e.g., Nachtigall et al., 2005; Finneran et al., 2009; Pacini et al., 2011; Strobel et al., 2017) as well as catch-and-release situations (e.g., Mann et al., 2010; Popov et al., 2007; Nachtigall et al., 2008; Mooney et al., 2018) where individuals are not trained for behavioral psychophysics (although psychophysical methods are considered the gold standard for hearing data, especially at the lower frequencies of a species' hearing range; see Mooney et al., 2012).

ABR testing with odontocetes has mostly been conducted underwater, with the sound stimulus delivered from a sound projector located at some distance from the animal. However, in some situations (e.g., hearing screening or threshold estimation of stranded animals) testing animals while out of water using contact transducers to deliver sound is the most practical option. For these in-air testing arrangements, it is useful to determine the best placements of sound transmitters for measuring ABR. A few studies have examined hearing pathways of odontocetes while in air and have found potential species-specific differences in the most efficient location for sound presentation: Møhl et al. (1999) found best sensitivity along the middle of the mandible in the bottlenose dolphin, while Mooney et al. (2015) found best sensitivity for a 100-kHz stimulus along the anterior midline of the jaw of a Risso's dolphin (Grampus griseus). In an underwater study with the bottlenose dolphin, Sysueva et al. (2017) demonstrated frequency-specific differences in the location of best sensitivity for localized acoustic stimulation. These results mirror similar findings with other odontocetes tested under water (Brill et al., 2001; Mooney et al., 2008; Mooney et al., 2014; Popov et al., 2016), and it is possible that these differences will be preserved during in-air audiometric tests using jawphones.

A study on the influence of in-air jawphone and recording electrode locations on the click-evoked ABR of the harbor porpoise (Phocoena phocoena) highlighted how these methodological choices may influence data (Mulsow et al., 2018). The results showed that while early peaks of the ABR (i.e., the auditory nerve response) were best recorded using electrodes on the same side of the head as the jawphone (ipsilateral; see Supin et al., 1991; Popov et al., 2006), the later peaks corresponding to stimulation of a particular ear were most prominent on the opposite side (contralateral; see Voronov and Stosman, 1982). Recordings of ABRs using an electrode behind the blowhole had the highest amplitude when the jawphone was placed on either the middle of the mandible or the tip of the rostrum. These especially large responses, however, likely reflected the voltage summation of responses from both ears being stimulated following acoustic crosstalk in the head (or jaw) rather than a more efficient pathway to the ear, per se.

With these findings in mind, electrode and jawphone locations—as well as specific response components analyzed (e.g., auditory nerve vs higher brainstem)—should be considered when measuring thresholds during in-air ABR or ASSR tests with dolphins (see Popov et al., 2006). These thresholds are often determined using either the lowest stimulus level for which a response is statistically detectable (e.g., Finneran et al., 2007), or by interpolating a linear regression fit to the ABR vs stimulus level functions [input-output (I/O) functions] and extrapolating to the level of zero response (e.g., Supin and Popov, 2007). It is therefore plausible that the dependence of ABR amplitudes based on electrode and jawphone locations, and the previously observed degrees of acoustic crosstalk, could influence frequency-specific threshold estimates.

To investigate these possibilities, this study recorded frequency-specific ABRs of three bottlenose dolphins (Tursiops truncatus) in air. Recording electrodes were placed behind the blowhole (the vertex of the head), near the external auditory meatus ipsilateral, and contralateral to three mandibular jawphone locations (similar to the methods of Mulsow et al., 2018; see also Popov et al., 2006, for a study using similar methods with underwater stimulation). Regression-based thresholds were determined for tone-burst stimuli with center frequencies of 28 to 160 kHz to determine if patterns of receptivity are frequency-dependent over the range typically examined during aerial ABR/ASSR threshold testing.

The amplitudes of two peak complexes from each ABR were measured. The first complex was P1-N2 recorded at the meatus electrodes (P1 is a vertex-positive peak, N2 is vertex-negative). These earliest peaks of the ABR arise from relatively peripheral responses (e.g., the auditory nerve; Supin et al., 1991; Popov and Supin, 1992) and reflect the responses of a single ear. P1-N2 is largest when recorded at the meatus ipsilateral to the stimulus and is much smaller when recorded at the vertex.

The second set of peaks analyzed was P4-N5 (P4 is vertex-positive, N5 vertex-negative). These peaks are probably generated in the lateral lemniscus and/or inferior colliculus (Popov and Supin, 1990, 1998), and differ fundamentally from P1-N2 in that they likely predominantly reflect activity at the ear contralateral to the side that they are best recorded on (e.g., P4-N5 recorded near the left meatus is dominated by activity originating from the right ear, Voronov and Stosman, 1982; Mulsow et al., 2018). This appears to arise from neural decussation (i.e., left-right crossing) at the level of the olivary complex in the pons, based on results with terrestrial mammals (Buchwald and Huang, 1975; Møller et al., 1995). These peaks are of importance to marine mammal hearing studies as they are often the highest-amplitude peaks of the dolphin ABR, persist the longest as stimuli are attenuated to threshold, and substantively contribute to the ASSR used to determine frequency-specific thresholds (Popov and Supin, 1998; Supin et al., 2001).

Three dolphins were included in the study. One of the dolphins (TRO 24 y/o) had a full range of hearing, defined as having a high-frequency hearing limit at or above 140 kHz. The other two dolphins (OLY 32 y/o, COL 15 y/o) had some hearing loss with high-frequency hearing limits of 80 kHz and 70 kHz, respectively. Testing was conducted at the U.S. Navy Marine Mammal Program (MMP) in San Diego, CA, where the dolphins were housed in floating, netted enclosures in San Diego Bay. For testing, the dolphins voluntarily beached onto foam mats located on floating decks and remained under trainer control for the duration of data collection. All procedures followed a protocol approved by the Institutional Animal Care and Use Committee at the Naval Information Warfare Center (NIWC) Pacific and the Navy Bureau of Medicine and Surgery, as well as all applicable U.S. Department of Defense guidelines.

Jawphone and electrode locations (Fig. 1) were based on a similar study with harbor porpoises (Mulsow et al., 2018). The jawphone comprised a TC4013 piezoelectric transducer (Teledyne Reson, Slangerup, Denmark) embedded in room temperature vulcanizing silicone rubber that was degassed to remove air bubbles. The jawphone had a suction cup diameter of 4.5 cm and was similar to that used by Mulsow et al. (2018). It was placed at one of three locations along the dolphin's mandible: posteriorly near the oral commissure (POS), as close to the tip of the rostrum as the suction cup would allow (TIP), and midway between those two locations (MID). Testing was conducted with the jawphone both on the left and the right sides of the mandible.

The stimuli were generated in EVREST software (Finneran, 2008, 2009), converted to analog at a rate of 1 MHz using a USB-6251 multifunction data acquisition (DAQ) card (National Instruments Corporation, Austin, TX), and presented at a rate of approximately 50/s. Stimuli were low-pass filtered at 200 kHz (3 C module, Krohn-Hite Corporation, Brockton, MA) and amplified with a 7600 M amplifier (Krohn-Hite Corporation) prior to being projected through the jawphone. Stimuli consisted of tone bursts with center frequencies of 28, 40, 56, 80, 113, and 160 kHz (Fig. 2). The tone bursts were five cycles of each frequency, gated with a raised cosine function. Each tone burst had two cycles of rise, a single cycle plateau, and two cycles of fall (i.e., a 2–1-2 tone burst). Four stimulus levels were presented per condition: 140, 130, 120, and 110 dB re 1 μPa, peak equivalent sound pressure level (peSPL). These levels were based on underwater calibrations using a TC4013 hydrophone placed 15 cm from the transmitting axis of the jawphone according to the standard “Procedure for Determining Audiograms in Toothed Whales through Evoked Potential Methods” [American National Standards Institute (ANSI), 2018]. This distance corresponds to the approximate distance between the dolphin inner ear and the jawphone at the POS locations, but actual levels received at the dolphins' inner ear should be considered approximations due to the impedance differences between water and the tissues in the dolphins' heads. Due to the frequency response characteristics of the TC4013, the 160-kHz tone burst resulted in a waveform with most spectral energy centered near 135 kHz (as shown in Fig. 2).

Dolphin ABRs were recorded using 10-mm diameter gold cup electrodes (Viasys Healthcare, Madison, WI) embedded in 40-mm diameter silicone suction cups. The non-inverting electrodes included the vertex and two meatal electrodes placed immediately behind the left and right external auditory meatuses (Fig. 1). These left and right electrodes were further identified as ipsilateral and contralateral depending on their location relative to the jawphone. The inverting electrode was placed approximately 40–50 cm behind the vertex electrode, in a location that is relatively inactive with respect to the ABR. A common ground electrode was placed either immediately in front of the dorsal fin, or in the ocean next to the testing location. Incoming electroencephalographic activity was amplified (94 dB) and bandpass filtered (0.3–10 kHz) using three IP511 biopotential amplifiers (Grass Technologies, Warwick, RI, one amplifier per non-inverting electrode, shared inverting and ground electrodes) prior to analog-to-digital conversion at a rate of 40 kHz with the USB-6251 DAQ card. Signal averaging was conducted in EVREST software before data were saved to hard disk for analysis.

A test series was conducted at each of the jawphone locations before moving to a new location. These series consisted of each stimulus level being presented, in descending order, for all frequency conditions. Using this procedure, two replicates of averaged ABRs were collected for each subject, frequency, level, non-inverting electrode, and jawphone location (except in a few cases for TRO where only one averaged ABR was collected). Each ABR was based on the weighted average of 1024 individual ∼21 ms EEG records following the presentation of a stimulus (Elberling and Wahlgreen, 1985). To further reduce electrical background noise, the two ABRs for a condition were coherently averaged to create a single averaged ABR.

The peak amplitudes presented in the paper were subject to the following criteria. The first criterion was based on the signal-to-noise ratio (SNR) of the coherently averaged ABR (i.e., generated from 2048 stimulus presentations). This was estimated as the root-mean-square (RMS) amplitude of the averaged ABR waveform over 5 ms divided by the RMS noise amplitude calculated from the variance of a single amplitude point in the 2048 epochs (Elberling and Don, 1984). If the SNR for a dolphin's ABR was less than 6 dB, it was not included in further analyses. Second, for frequencies all three dolphins could hear (28–56 kHz), a peak amplitude was only reported if all subjects had a value for that point. This was done to avoid biasing the I/O functions with data from one animal (TRO) at low stimulus levels. Values at 80 kHz and above were based only on testing with TRO and were therefore subject solely to the SNR criterion.

ABR peaks were visually identified in the averaged ABR waveforms, and amplitude measurements made by placing a cursor at the maximum and minimum amplitudes of positive and negative deflections, respectively. The resulting I/O functions (in nV/dB) were analyzed using regression analysis (Originlab, 2018). The first type of analysis used nonlinear regression to simultaneously fit all I/O functions at each frequency with lines having a shared slope but independent intercepts. This assumes a similarity in the slopes of the I/O functions, which is most likely to be true for the P1-N2 functions that do not represent the sums of multiple binaural generators (Mulsow et al., 2014). The use of a shared slope allows for the “threshold” differences (i.e., differences between x-intercept parameters) to be evaluated independently of an arbitrarily chosen threshold level. Conditions where less than two data points existed for an I/O function were excluded from these analyses. For the P4-N5 data, a shared slope among jawphone locations is less likely the case. Differential stimulation of the two ears as the jawphone is moved, and subsequent voltage summation in the ABR will likely result in steeper slopes for conditions where both ears are stimulated. Thus, independent linear regression for each of the averaged I/O functions was performed, excluding I/O functions with fewer than three data points.

All frequencies reliably elicited ABRs in TRO; however, frequencies of 80 kHz and higher did not reliably elicit ABRs in COL and OLY. This was expected, as those two subjects had preexisting high-frequency hearing losses. Aside from this, the ABRs recorded from the vertex electrode had the typical morphology of those reported for this species (Fig. 3). ABRs recorded using the meatus electrodes were of similar duration relative to the vertex-recorded ABR, but with different morphologies. The initial wave, P1-N2, was relatively large at the meatus electrodes when compared to recordings at vertex, while P4-N5 was larger at vertex. Both P1 and P4 were negative in polarity at the meatus electrodes and positive at vertex (N2 and N5 were negative at vertex). Although these waves were identifiable based on their latencies, the other intermediate waves of the vertex-recorded ABR were obscured in the meatus recordings and so they were not included in analyses.

Figure 4 shows the I/O functions for P1-N2 recorded at the meatus electrodes. Although P1-N2 responses were often observed at the vertex electrode, they were of much lower amplitude and are not reported in detail here. The amplitudes shown in Fig. 4 generally increased monotonically with increasing stimulus levels. P1-N2 amplitudes were typically largest at the ipsilateral electrode with the jawphone at the POS location, and slightly smaller at the MID location. Responses on the contralateral electrodes were generally smaller, although in some cases their amplitudes approached or exceeded those at ipsilateral electrodes (i.e., at 113 and 160 kHz for subject TRO).

The I/O functions for P4-N5 at the vertex electrode are shown in Fig. 5. The MID jawphone location uniformly produced the highest amplitudes, and often the amplitudes of P4-N5 for the TIP placements were comparable to or exceeded those for the POS placement. Figure 6 shows the P4-N5 responses at the meatus electrodes to highlight the main difference between data for those peaks and P1-N2; the P4-N5 amplitudes at the meatus electrodes were largest when stimuli were delivered from the POS and MID locations that were contralateral to the recording side.

The results of the regression analyses (Table I) show that, in general, most of the I/O functions had slopes of 40–60 nV/dB for P1-N2, and 60–200 nV/dB for P4-N5. The R² values were between 0.96 and 1.0 for all datasets, indicating good fits in all cases. It is worth noting that the slopes for the P4-N5 I/O functions are similar to those for vertex-recorded waves reported by Popov et al. (2006), but the P1-N2 I/O slopes reported in that previous study are approximately 3 times smaller. This is likely due to differences in the physical media of the in-air vertex and underwater meatus recording electrodes used by Popov et al. (2006), and the resulting current loss that occurs while recording the ABR in saltwater.

Regression-based thresholds are plotted in Fig. 7 for the parameters of jawphone and electrode locations. The curves all show a lowest threshold at 40 kHz, with increased thresholds at 28 kHz and at 56 kHz and above. The thresholds for the POS and MID locations were similar for both P1-N2 and P4-N5 (average threshold differences were 3 dB or less), with POS thresholds slightly lower for P1-N2 and MID thresholds lower for P4-N5. The TIP thresholds shown in Fig. 7 were approximately 14 dB higher than the POS and MID locations for the shared-slope P1-N2 analyses, and about 4 dB higher than the POS and MID locations for the P4-N5 independent-slope analyses.

Finally, the apparent differences in stimulation between the ears ipsilateral and contralateral to the jawphone were computed based on the regression fits of the P1-N2 data shown in Fig. 4. These threshold differences are given in Table II. The main trend is that higher levels of interaural attenuation were typically found with the jawphone at the POS location. Some notable exceptions occurred at the highest two frequencies of 113 and 160 kHz, where data were only available for TRO.

The goal of this study was to determine the optimal locations for jawphones and recording electrodes during in-air threshold measurements with odontocetes, and to determine if these locations were dependent on frequency. The results suggested that the POS location provides a slightly better option than the MID location for stimulating the ipsilateral ear, while minimizing contributions from the ear contralateral to the jawphone. The TIP location seemed to provide the worst conditions for transmission to the ipsilateral ear. This is not surprising given that the TIP position is the furthest from the inner ear and has the least direct contact with acoustic fats in the jaw (Norris, 1968).

The MID and POS jawphone locations provided nearly equivalent stimulation at the ipsilateral ear, but there was a greater degree of stimulation of the contralateral ear with the jawphone at MID. This acoustic crosstalk probably led to a larger amplitude for the P4-N5 response as recorded at the vertex (this was also the case for the TIP position). The P4-N5 responses therefore produced a somewhat misleading picture regarding sensitivity as the jawphone was moved rostrally toward the tip of the mandible; the P4-N5 amplitudes and thresholds gave the appearance that the TIP location was more sensitive than what was suggested using the monaural P1-N2 responses. Larger P4-N5 amplitude for the MID position may provide an advantage in rapid hearing screening cases following stranding, where ear-specific information is not required. These stranding tests may use the ASSR (rather than the ABR), which represents superimposed, repetitive P4-N5 complexes at the stimulus modulation rate (Supin and Popov, 1995; Supin et al., 2001). Larger amplitudes for the P4-N5 response at suprathreshold stimulus levels might facilitate rapid response detection prior to attenuating stimuli toward threshold (see Finneran et al., 2019).

There did not appear to be an interaction of frequency and jawphone location in the derived thresholds. This suggests that the frequency being tested is not a particularly important consideration when choosing jawphone location under these testing conditions. How the observed patterns would behave below 28 kHz is not known, but ABRs elicited by stimuli at lower frequencies become progressively less effective down to approximately 10 kHz [below which ABRs are not a valid method in this species due to the loss of cochlear place specificity (Finneran et al., 2016)].

The patterns observed here are similar to those found using a contact transducer and a click stimulus with a bottlenose dolphin (Møhl et al., 1999) and should hold for most other ABR-based tests with bottlenose dolphins in air. Based on underwater ASSR threshold measurements with a bottlenose dolphin, Sysueva et al. (2017) found that the most acoustically sensitive region of the dolphin head did vary as a function of frequency. These studies were carried out with a transmitter held against the skin on various locations on the heads of dolphins. The differences in medium (in air vs under water), transmitter (suction cup vs hand-held), stimulus (tone burst vs SAM tone), and response (ABR vs ASSR) may have led to some of the differences in the observed results. A similar ASSR study that used a jawphone in air with a Risso's dolphin (Grampus griseus) also found frequency- and location-dependent differences in sensitivity (Mooney et al., 2015). The extent to which these results are indicative of species-specific differences (the head morphology of the Grampus differs greatly from that of Tursiops) or other methodological differences is not known.

One perplexing result of the present study was the degree of stimulation at the contralateral ear when the jawphone was placed at the MID or TIP location. Below 56 kHz, where data were available for all three dolphins, interaural attenuation values were 9–11 dB for the POS location. The attenuation was substantially reduced for these frequencies at the MID location (2 dB at most), and stimulation was greater at the contralateral ear with the jawphone at the TIP location. It was expected that stimulation at the contralateral ear would be lower than what was observed given the intervening structures in the dolphin head (i.e., air spaces, skull, and soft tissue; Cozzi et al., 2017). Studies by Popov et al. (2006) have demonstrated approximately 20–30 dB of interaural attenuation with dolphins underwater in direct-field acoustic conditions. In contrast, McCormick et al. (1980) found lower degrees of interaural attenuation (2–23 dB) using ∼20 kHz localized stimulation from a contact transducer, and Mulsow et al. (2014) reported an attenuation of approximately 20 dB for a click delivered at the POS location. The lower and more variable values found with contact stimulation suggest that the contact excitation method produces a greater degree of acoustic crosstalk than what might be found naturally underwater. This likely arises from a number of factors, including the acoustic impedance differences of the dolphin head relative to the external physical medium (Aroyan, 2001) and the localized nature of stimulation compared to direct-field underwater transmission.

Despite the differences between aerial testing with jawphones and the natural mode of stimulation underwater, the current data could prove useful in testing models of sound reception in the heads of cetaceans (e.g., finite element models; Cranford et al., 2008; Tubelli et al., 2012; Cranford and Krysl, 2015; Tubelli et al., 2018). Simulations of localized tone burst stimulation along the mandible should produce patterns similar to those observed in the P1-N2 data (i.e., indicative of direct transmission from the site of stimulation to each ear). Specifically, changes in received amplitude as the point of stimulation is moved antero-posteriorly along the mandible should be replicable in models of the dolphin head.

The authors wish to thank D. Houser, J. Powell, R. Dear, M. Tormey, and the animal care and training staff at Navy Marine Mammal Program. This manuscript greatly benefited from the comments of one anonymous reviewer and Dr. Alexander Ya. Supin at the Institute of Ecology and Evolution, Russian Academy of Sciences. The study was funded and supported by U.S. Navy Fleet Forces Command. Portions of this work were presented at the 178th meeting of the Acoustical Society of America. This is contribution 261 of the National Marine Mammal Foundation.