Measurement of Hearing in the Atlantic salmon (Salmo salar)

Part 1 Audiometry, using Auditory Evoked Potentials (AEP) to Determine Hearing Thresholds of a Number of Cohorts of Salmon

Stephen D Simpson and Rick Bruintjes

Methodology

Study Fish and Holding Conditions

Three groups of fish were tested:

1) Wild Post-Smolt: Ten post-smolt collected in traps in the River Tay as seaward migrating smolt and held for one year in a 1.5 m diameter circular tank indoors (LT1 Tank 3). Standard Length ( SL) 273.5 ± 11.3 mm (mean ± SE) and Fork Length ( FL) 288.7 ± 12.1 mm.

2) Captive Post-Smolt: Ten post-smolt reared since hatching in captivity from wild stock eggs and held outdoors in a 1.5 m diameter circular tank outdoors (Outdoor Tank 16). SL 294.0 ± 11.4 mm, FL 312.0 ± 11.4 mm.

3) Captive Adults: Ten adult salmon reared since hatching in captivity from wild stock eggs and held in a 15 m long tank (Dumbell). SL 379.0 ± 15.0 mm, FL 401.5 ± 15.7 mm.

All holding tanks had running seawater and aeration via airstones, and temperature was a constant 10°C. The indoor tanks had a 9:15 hours Light:Dark lighting regime with low intensity green lamps, while the outside tank received ambient mid-March daylight conditions of 8:16 L:D.

Acoustic conditions in the holding tanks (water 1 m depth, recording 10 cm above bottom where salmon usually resided) were measured using a calibrated omnidirectional HTI-96-MIN hydrophone (frequency response = 2 Hz - 30 kHz, voltage sensitivity = -165 dB re 1 V/μPa; High Tech, Inc., Gulfport, Mississippi) and a Sony PCM-M10 24-Bit recorder (96 kHz sampling rate; Sony Corporation, Tokyo, Japan). Root Mean Square ( RMS) noise levels, analysed using SASLab Pro v4.5.2 (Avisoft Bioacoustics, Berlin), were 127.8 (LT1 Tank 3), 127.0 (Outdoor Tank 16) and 134.5 (Dumbell) dB re 1 μPa (1 sec averaging), with levels only ~1 dB lower when analysed within the range 0-1 kHz indicating most of the noise was low frequency.

Hearing Sensitivity Tests

To minimise stress, fish were transferred from the holding tanks in nets submerged in buckets prior to handling during preparation of the fish in the AEP apparatus. The captive adult fish were immersed in a bin with a nonlethal dose of buffered tricaine methanesulfonate ( MS222) to induce temporary mild anaesthesia (70 mg/l MS222, fish immersed for around one minute until balance was lost) and revived (>five minutes) in fresh seawater (aerated between each testing period) prior to taking AEP measurements. Testing took 30-45 minutes per fish, and no mortality or obvious injury resulted from the experimental procedures; after testing fish were sacrificed by a Home Office Schedule 1 method (fatal blows to the head). All experiments took place in a quiet room in the MS Science Laboratory in the daytime (0800-1800) during 11-15 March 2013.

Acoustic Stimulation and Data Acquisition

To assess hearing sensitivity of fish from the three test cohorts, we used the Auditory-Evoked Potentials ( AEP) technique, which is a non-invasive electrophysiological measure of the synchronized brain response to auditory stimuli ( e.g., Corwin et al., . 1982; Kenyon et al., . 1998). Hearing assessments were conducted on each fish only once. During testing, fish were wrapped in soft knotless fine mesh and secured in a rigid harness suspended 10 cm below the surface of the water in the centre of a 1 m diameter circular fibreglass tank (water depth 1 m). Stainless steel recording electrodes (0.4 mm diameter; Spes Medica S.r.l., Battipaglia, Italy) were insulated except at the tips with vinyl paint, and one was inserted subcutaneously at the front of the dorsal fin (reference electrode) while the recording electrode was inserted subcutaneously at the dorsal midline 2 mm posterior to the cranium. A third non-insulated grounding electrode was placed in the water 5 cm from the fish.

Sound Stimulus

The sound stimulus presented to fish was a repetitive 20 ms pure tone, with a 2 ms Hanning window gate at the beginning and end to avoid loss of quality due to the rapid firing of the speaker. The signal file was developed in SigGenRZ (Tucker-Davis Technologies, Alachua, Florida), with test frequencies of 100, 200, 300, 400, 500, 600, 700 and 800 Hz. Signals were sent from BioSigRZ to a TDT RZ6 Multi-I/O processor, amplified with a Phonic Max 500 amplifier (120W; Phonic Corporation, Taipei, Taiwan) and played through an underwater speaker ( UW30; Lubell Laboratories, Inc., Columbus, Ohio) suspended 57 cm below the fish and facing upwards from the centre of the tank.

To calibrate the received sound pressure levels of stimuli at each frequency and test level, multiple recordings were made in the test tank with the calibrated hydrophone fixed in the same position as the head of the fish would be located during testing, and RMS received levels (re 1 μPa, averaged over the 10ms in the centre of each pulse) were derived in SASLab Pro. Peaks of the focal frequencies were all at the frequencies as intended, and first harmonics of the test frequencies were all >20 dB below the focal frequency.

Electrophysiology

The AEPs at each frequency and signal level were collected from the fish via a TDT Medusa RA4LI headstage and RA4PA preamplifier, based on averages of 200 repeats of 40 ms recordings including the 20 ms tones repeating at a rate of 10 times per second. The brain signal was amplified with a gain of 20 and was low-pass filtered at 3 kHz, high-pass filtered at 10 Hz, and notch filtered at 50 Hz using BioSigRZ software. Averaged AEPs were digitised and stored.

The sound pressure levels presented at each frequency were attenuated in 6 dB steps until the stereotypical AEP responses were lost into background myogenic noise. This measure was based on visual assessment of the AEP signal, which is most commonly used in AEP studies (Hall 1992; Kenyon et al., 1998) and which gives results similar to those of statistical methods (Mann et al., 2001; Brittan-Powell et al., 2002). For each fish, the AEP threshold was defined at each frequency as the lowest sound level that gave a defined repeatable response.

Due to the limitations of our experimental set up and in line with recent AEP studies ( e.g., Halvorsen et al., 2009), we measured the acoustic pressure component of the presented sounds, and thus give hearing thresholds in terms of sound pressure (dB re 1 μPa RMS, 10 ms measured in the centre of the pulse). Due to morphological limitations, it is likely that Atlantic salmon respond predominantly to the particle motion component of the presented sounds. Thus, we would caution the interpretation of thresholds in terms of absolute hearing abilities. Nevertheless, the use of this approach for comparative studies of hearing between the three study groups is valid, and with further experimentation to investigate behavioural thresholds and/or open water AEP measurements with the system calibrated using an in situ accelerometer it would be possible to relate these initial thresholds to hearing thresholds based on particle motion and provide a likely spatial scale of detection of anthropogenic noise ( e.g., pile driving).

Results

Auditory thresholds based on acoustic pressure were recorded at 100, 200, 300, 400, 500, 600, 700 and 800 Hz for each group of salmon to provide a complete AEP audiogram ( Figure 1). The mean AEP threshold value at each frequency and size distributions of fish in each group of individuals for each experimental group is presented in Table 1.

Figure 1: Audiograms based on mean (± SE) minimum received levels (dB re 1μPa) that elicited a characteristic auditory evoked potential. Three groups of Atlantic salmon ( Salmo salar) were tested (n = 10 in each case): Wild post-smolt, Captive post-smolt, Captive adults. See Table 1 for mean values and size distributions of fish.

Discussion

This study generally concurs with the previous findings of Hawkins & Johnstone (1978) that Atlantic salmon ( Salmo salar) does not appear to have sensitive hearing relative to many other marine fish species, including gadoids ( e.g., Atlantic cod, Gadus morhua) and clupeids ( e.g., herring, Clupea harengus). This is likely due to a lack of secondary hearing modifications linking the swimbladder to the auditory system. We found evidence of a response to sounds at higher frequencies (400-800 Hz) than had been previously published for this species ( Figure 2), although hearing up to 600 Hz has recently been reported in the Chinook salmon ( Oncorhynchus tshawytscha; Halvorsen et al., 2009) using the same method as this current study.

Table 1

Mean (± SE) auditory evoked potentials thresholds, measured in the pressure domain (dB re 1μPa), for Atlantic salmon ( SL = standard length; FL = fork length).

Group:	Wild Post Smolt	Captive Post Smolt	Captive Adults
n:	10	10	10
100 Hz	102.3 ± 1.3	106.5 ± 1.5	104.7 ± 1.3
200 Hz	92.9 ± 1.4	93.5 ± 1.3	93.5 ± 2.5
300 Hz	101 ± 1.8	99.2 ± 1.3	101.6 ± 1.4
400 Hz	103.6 ± 1.8	103 ± 1	103 ± 1
500 Hz	113.2 ± 1.7	114.4 ± 1.6	116.2 ± 1.8
600 Hz	116 ± 1.3	116 ± 1.8	119 ± 1.8
700 Hz	120 ± 1.3	123 ± 1.6	122.4 ± 1.6
800 Hz	126.4 ± 1.4	126.4 ± 1.4	128.8 ± 1.3
SL (mm)	273.5 ± 11.3	294.0 ± 11.4	379.0 ± 15.0
FL (mm)	288.7 ± 12.1	312.0 ± 11.4	401.5 ± 15.7

Exploring Discrepancies between Previously Published Audiograms for Atlantic and Chinook Salmon

We found salmon had slightly less sensitive hearing at 100 Hz than reported in Hawkins & Johnstone (1978), with the earlier study suggesting threshold values 10 dB below those measured here, but found more sensitive hearing than the earlier study at frequencies >200 Hz. Since the five fish in the Hawkins & Johnstone study (320-360 mm) were larger than our two post-smolt groups, but smaller than our captive adults (see Table 1), it does not seem likely that size or ontogeny explains the difference in audiograms. Further, since we have used our AEP apparatus to measure goldfish hearing (hearing specialists), giving thresholds at low frequency that concur with many published audiograms, we do not attribute these differences to poor performance of the equipment used here.

Thus we suggest that there are three possible reasons for the differences between our results and those of Hawkins & Johnstone:

1. Since AEP measurements can be ± 20 dB different to measured behavioural and physiological thresholds, the two sets of audiograms may be within the margin of error in the 100-300 Hz range since the two studies use very different approaches. Further studies that combine AEP measurements with behavioural and/or physiological thresholds should explore whether the differences are experimental or are biologically meaningful based on different measures of response.

2. The logistically simpler modern AEP set up uses smaller speakers ( UW30 in this case) compared to the Dyna-Empire J9 and J11 speakers used in the field in the Hawkins & Johnstone study. This will significantly reduce the particle motion component of the broadcast signal, meaning that the current AEP measurements may be conservative at low frequencies. While such large speakers could not be employed in a tank environment, we have discussed the future use of larger speakers in the 10 m diameter tank at the MS Science Laboratory, and ultimately also in open water to measure in situ AEPs, where we would combine measurements of received levels using hydrophones (pressure) and accelerometers (particle motion).

3. The fish in the Hawkins & Johnstone study were reared at the Invergarry Hatchery before being transferred as smolts to an open water sea pen in Loch Ailort where they were grown out. Thus the ambient noise levels in the open water holding pen prior to testing would have been very much lower than those experienced by the three groups in the current study which were held in large tanks at the Marine Scotland Science Laboratory with high levels of low-frequency noise (Root Mean Square ( RMS) noise levels: 127.0-134.5 dB re 1 µPa; 1 sec averaging). The fish tested in the present study had either been kept in captivity from hatching (Captive post-smolts, Captive adults) or else in captivity for at least one year (Wild post-smolts) following capture in the wild. It is possible that prolonged culture of fish in noisy conditions may have brought about a temporary or permanent threshold shift, particularly as much of the noise in the holding tanks was in the frequency range of greatest sensitivity (<300 Hz). To address this issue we suggest the use of fish collected recently from the wild, such as in Halvorsen et al., (2009) which found higher sensitivity in Chinook salmon ( Figure 2), should be used, with fish kept in open water environments or quietened tanks after capture prior to testing.

Figure 2: Figure 2 redrawn with previously published hearing thresholds transformed into dB re 1μPa by Nedwell et al., (2004) from Hawkins & Johnstone (1978), and with AEPs for Chinook salmon measured using similar approach to the current study (Halvorsen et al., 2009). The earlier field-based approach identified hearing thresholds at low frequencies (<200 Hz) slightly below those in the present study, but underestimated hearing sensitivity in higher frequencies (>200 Hz). Present study shows mean (± SE) for ten individuals, while the Hawkins & Johnstone data shows a mean from five fish and Halvorsen et al., study shows a mean from ten wild fish.

General Conclusions

The cohorts of Atlantic salmon held at the MS Science Laboratory, which have been reared in captivity since hatching or since collection as smolt from the River Tay, appear to have hearing abilities that generally concur with previous studies. Thus they, and similar fish, will provide a valuable model system for (a) testing impacts of noise on hearing ( e.g., temporary threshold shifts dues to pile driving noise), (b) exploring relationships between electrophysiological, physiological and behavioural thresholds of hearing, and (c) studying impacts of noise on physiological ( e.g., metabolic rate, opercular beat rate) and behavioural ( e.g., foraging, anti-predator, movement) performance.