The subject was an adult female bottlenose dolphin Tursiops truncatus, 260 cm body length, provisionally 28 years old. Before being used for experiments, the animal was kept in a commercial dolphinarium for 25 years. After two months of experimentation, she was returned back to the dolphinarium.
The study was performed at the Utrish Marine Station of the Russian Academy of Sciences on the Black Sea coast (Krasnodar Province, Russia). The animal was kept in a round plastic tank 6 m in diameter, 1.7 m deep, filled with sea water. During the experiments, the water level in the tank was lowered down to 0.4 m. The animal was laid on a stretcher in such a manner that its dorsal head surface was above the water surface and the rostrum tip was in the center of the tank (Fig 1). The animal was kept in such a position during the experiment, which regularly lasted 2 to 3 h. After the experiment, the animal was released, and the water depth in the pool was restored to its normal level of 1.7 m.
The study design was approved by The Ethics Committee of the Institute of Ecology and Evolution of the Russian Academy of Sciences (№ 31, 30.04.2019).
The sound stimuli were tone pips of 64 kHz carriers. The pip envelope was one cycle of a raised cosine function containing 6 cycles of the carrier (Fig 2a, 1). The stimuli were digitally generated by a standard computer at an update rate of 500 kHz using a custom-made program based on the LabVIEW software (National Instruments, Austin, TX, USA). The program generated two pips in separate channels with independent control of the pip amplitudes and interpip delay. The generated signals were digital-to-analog converted by a 16-bit data acquisition board NI USB-6251 (National Instruments), amplified and attenuated by a custom-made two-channel power amplifier-attenuator with a passband of 200 kHz and output impedance of 50 Ohm.
The signals generated in two channels fed two B&K 8104 transducers (Bruel & Kjaer, Naerum, Denmark). The transducers were positioned symmetrically with respect to the animal head axis, at a distance of 1 m from the melon tip, at a depth of 20 cm, i.e., at the mid-depth of the water. The azimuthal position of the transducer relative to the head axis varied within a range of ±90°.
To minimize sound reflections from the tank walls, the area around the animal’s head and transducers was surrounded by a circular fence of sound-absorbing material (rubber with closed air cavities) (see Fig 1). The fence covered the whole water depth. The radius of the fence was 1.05 m, so the transducers were positioned inside it.
Stimuli were monitored by a B&K 8103 hydrophone (Bruel & Kjear, Naerum, Denmark) coupled with a Nexus 2690 amplifier (Bruel & Kjaer) and an NI-5132 digital oscilloscope (National Instruments). The monitoring hydrophone was positioned next to the lower jaw of the animal. The acoustic measurements were performed between, not during the AEP-recording sessions.
The waveform of the acoustic signals that were picked up by the monitoring hydrophone did not exactly reproduce the electronic signal that activated the sound-emitting transducers (Fig 2). The acoustic signal was longer than the electronic signal. An initial component of a peak level of 135 dB re. 1 µPa (5.6 Pa acoustic pressure) followed by a few smaller (–6 to –10 dB re. the initial peak) components. During 0.5 ms after the peak, the acoustic pressure fell to approximately –20 dB re. the signal peak.
For noninvasive AEP recording, gold-plated electrodes imbedded in silicon suction cups (Cetacean Research Technology, Seattle, WA, USA) were used. Two manners of AEP recording were used: the vertex and lateral.
For vertex AEP recording, one of the electrodes was fastened at the dorsal head surface, 6- 9 cm behind the blowhole. This point has been found to feature the highest amplitude of a noninvasively recorded fast AEP known as the auditory brainstem response, ABR (Popov and Supin, 1990). This electrode was considered active for ABR recording. The other electrode was fastened at the dorsal fin where the ABR amplitude was negligible. This electrode was considered reference. Both the electrodes were above the water surface.
For lateral AEP recording, the active electrode was fastened underwater at the lateral head surface, next to the auditory meatus. For this electrode, the embedding suction cup served both as electrode fastening and as insulation from low-impedance sea water. The reference electrode was positioned at the dorsal fin above the water surface.
The electrodes were connected to the input of a brain-potential amplifier LP511 (Grass Technologies) that provided amplification by 80 dB within a frequency range of 0.3 to 3 kHz. This frequency range was chosen as covering the frequency band of ABR but minimally transferring noise outside this band.
Amplified EEG signals were analog-to-digital converted and processed by the same acquisition board NI USB-6251 that was used for signal generation. The sampling rate for conversion was 40 kHz. To process the brain potential signals, a custom-made program based on LabVIEW (National Instrument) software was used.
Experiment design and data processing
The forward-masking processes were investigated by recording AEPs to two successively presented stimuli, the first of which was considered a masker (conditioning stimulus), and the second was considered a test. The peak levels of both the masker and test stimuli were equal, specifically, 135 dB re. 1 µPa.
To separate the ABRs to the two stimuli, a subtraction procedure was used: the response to the masker alone was subtracted from the response to the pair of stimuli. This procedure requires the responses to the masker to be equal in records obtained with the pairs of stimuli (masker + test) and masker alone. To make this equality as precise as possible, the masker + test pairs and maskers only were presented in an interleaving manner (Fig 3): the masker + test, (stimuli 1 and 2 in Fig 3), the masker only (stimulus 3 in Fig 3), masker + test again, etc. With this manner of presentation, responses to the masker in pairs and maskers only were equally subjected to any long-term variation in hearing sensitivity that might occur from long-term hearing adaptation.
The stimuli were presented at a rate of 10/s. For extraction of AEPs from brain-potential noise, 1000 responses to masker + test pairs and 1000 responses to maskers were collected. Thus, each recording trial lasted 200 s.
AEPs were extracted from brain-potential noise by coherent averaging. AEPs to maskers + test pairs and AEPs to maskers were collected and averaged separately (epochs 4 and 5 in Fig 3, respectively). The averaged AEP to the test stimulus was extracted by point-to-point subtracting the averaged response to the masker from the averaged response to the masker + test. Thus, each recording trial provided three AEP waveforms: (i) the masker, (ii) test, and (iii) masker + test combination.
AEP amplitude evaluation
The AEP amplitude was assessed as the largest positive-to-negative span within a 2.5-ms window where the response was expected, specifically, from 2.5 to 5 ms after stimulus. A record was considered response-present when this span was 3.5 times as high as the root-mean-square (RMS) of the background record noise. The background noise was evaluated in records obtained using the same procedure as for AEP records but with no stimulus. The criterion of 3.5 RMS was used because for Gaussian noise, RMS is quantitatively equal to the standard deviation (SD) of the amplitude distribution. Therefore, 99.95% of the noise samples are within a range of ±3.5 SD. At a sampling rate of 40 kHz, the 2.5-ms long window contained 100 samples. The probability of background fluctuations exceeding 3.5 SD is 0.9995100= 0.95. Therefore, according to the commonly adopted 95% criterion, voltage spans exceeding 3.5 RMS can be considered not random background noise fluctuation but statistically significant indications of responses.