Acoustic pressure and particle motion thresholds in six sciaenid fishes

Sound in water is composed of two physically linked components, propagating scalar pressure waves and directional particle motion, which differ in the pathways through which they reach the inner ears of fishes(Fay and Popper, 1975). The otoliths of all fishes are biological accelerometers that directly detect the particle motion components of sound as a result of inertial differences between sensory epithelia and otoliths (Lu and Xu, 2002; Popper and Fay,1999). Additionally, the pressure component of sound may be detected indirectly by some fishes via accessory anatomical structures that transform sound pressure waves into particle displacements(Popper and Fay, 1993).

Fishes are categorized as hearing `specialists' and `generalists' on the basis of anatomy, the ability to detect the pressure component of sound, and the range of detectable bandwidth. Hearing specialist species have evolved projections of the swim bladder or skeletal connections that enable the indirect re-radiation of the pressure component of sound as particle displacement capable of stimulating the inner ear(Fay and Popper, 1974; Popper and Fay, 1999). Thus hearing-specialist fishes, which include groups such as clupeids, otophysans,mormyrids and osphronemids, may use both direct (particle motion) and indirect(pressure transduction) mechanisms to enhance their hearing sensitivity and extend their detectable auditory bandwidth(Mann et al., 1997; Popper and Fay, 1993; Yan, 1998; Yan and Curtsinger, 2000). By contrast, hearing generalist fishes lack such specialized structures coupling pressure-to-displacement transducers to the otic capsule, resulting in attenuation of the signal and reduced stimulation of the ear viasound pressure (Casper and Mann,2006). The unaided organs of the inner ear of hearing generalists are thought to be fairly insensitive to the indirect transduction of sound pressure (Sand and Karlsen,2000; Yan et al.,2000); direct particle motion stimulation of the otoliths is likely more relevant to these fishes (Lu and Xu, 2002; Casper and Mann,2006). However, few studies have examined the hearing thresholds of fishes with respect to both pressure and particle motion sensitivity(Myrberg and Spires, 1980; van den Berg, 1985; Lovell et al., 2005; Casper and Mann, 2006).

Sciaenid fishes are model organisms of teleost bioacoustics(Ramcharitar et al., 2006a; Roundtree et al., 2006), but comparatively little is known about their auditory abilities. Sciaenid saccular otoliths are enlarged relative to most fishes, and their morphology and proximity to the swim bladder vary widely(Chao, 1978; Ramcharitar et al., 2001). Both hearing specialists and generalists have been identified within the family (Ramcharitar et al.,2004; Ramcharitar et al.,2006b). Unfortunately, the pressure detection abilities of less than two percent of the 270 sciaenid species have been described [Atlantic croaker, spot, weakfish, black drum, silver perch(Ramcharitar, 2003)], and the particle motion sensitivity of these fishes has not been examined. Comparative work on sciaenid fishes has great potential to elucidate form-and-function relationships in the teleost auditory system(Ramcharitar, 2003). We therefore performed auditory brainstem response experiments using a hydrophone and geophone to categorize the pressure and particle acceleration detection thresholds of six sciaenid fishes. The simultaneous recording of the pressure and particle motion components of sound stimuli allowed us to express audiograms with respect to both. The former allows us to compare our data to previously published results for sciaenid fishes(Ramcharitar and Popper, 2004; Ramcharitar et al., 2006b);the latter allows comparison to recent studies examining particle motion thresholds in other fishes (Casper and Mann, 2006; Mann et al.,2007).

Weakfish (Cynoscion regalis Bloch and Schneider 1801), spotted seatrout (Cynoscion nebulosus Cuvier 1830), Atlantic croaker(Micropogonias undulatus Linnaeus 1766), red drum (Sciaenops ocellatus Linnaeus 1766), spot (Leiostomus xanthurus Lacepede 1802) and northern kingfish (Menticirrhus saxatilis Bloch and Schneider 1801) were captured in Chesapeake Bay, eastern USA, using hook and line (Table 1). Animals were maintained in recirculating 1855 l aquaria at 20±1°C (winter months) or 25±2°C (summer months) and fed a combination of frozen Atlantic menhaden (Brevoortia tyrannus), squid (Loligo sp.)and commercially prepared food (AquaTox flakes; Zeigler, Gardners, PA,USA).

Experimental and animal care protocols were approved by the College of William and Mary's Institutional Animal Care and Use Committee, protocol no. 0423, and followed all relevant laws of the United States. Auditory brainstem response (ABR) experiments were conducted on six animals of each species. All subjects were sedated with an intramuscular (i.m.) dose of the steroid anesthetic Saffan (Glaxo Vet, Glaxo Vet Ltd, Uxbridge, UK; 10 mg kg^–1) and immobilized with an i.m. injection of the neuromuscular blocking drug gallamine triethiodide (Flaxedil; Sigma, St Louis,MO, USA; 10 mg kg^–1). Recording of vertebrate ABR waveforms in anaesthetized and/or immobile subjects is a common practice to minimize the obscuring effect of muscular noise on ABR recordings(Hall, 1992; Kenyon et al., 1998; Casper et al., 2003). Sedated and immobilized animals were suspended within a rectangular 61×31×16.5 cm Plexiglas tank using foam straps, leaving <1 mm of the top of the head protruding from the water. Subjects were ventilated (1 l min^–1) with filtered, oxygenated, and temperature-controlled seawater (25±2°C). At the conclusion of each experiment, fishes were euthanized with a massive i.m. dose of sodium pentobarbital (∼300 mg kg^–1).

Auditory brainstem response (ABR) is a non-invasive recording of the neural activity in the eighth cranial nerve and brainstem in response to synchronized acoustic stimuli (Corwin et al.,1982; Kenyon et al.,1998). The ABR experimental setup and procedure followed that of others (Kenyon et al., 1998). A speaker (model: 40-1034, 27.5 cm in diameter, Radio Shack, Fort Worth, TX,USA), suspended in the air, was mounted 1.5 m directly above the test subject. Two platinum wire needle electrodes (model: F-E7, 10 mm tip, Grass Technologies, West Warwick, RI, USA) were placed subdermally along the midline of each subject: the active electrode was positioned above the medulla, and the reference electrode in the dorsal musculature above the operculum. The system was grounded to the water of the experimental tank via a 6 cm×26 cm stainless steel plate. An omnidirectional hydrophone (Reson A/S, Slangerup, Denmark; sensitivity: –211 dB re: 1V/μPa) was suspended with rubber straps 25 mm below the water surface (i.e. the depth of a subject's otic capsule) and positioned within 2.5 mm of the right opercle-preopercle margin of each subject to measure the sound pressure level of the stimulus and ambient noise.

In the absence of an anechoic chamber, all experiments were conducted in a concrete laboratory. We produced a stochastic differential white noise signal to characterize the echoes resulting from all reflective surfaces at the hydrophone positioned next to the subject. A custom Fourier/inverse Fourier transform algorithm (MATLAB version 6.5, Mathworks, Inc., Natick, MA, USA) was used to analyze these recordings and add to each frequency's pure tone stimulus the appropriate signals needed to destructively interfere with any recorded echoes. Any alteration to the sound field in the laboratory since the last echo-cancellation (i.e. movements, small changes in the tank water level,etc.) required us to re-echo-cancel before proceeding. Visual examination of stimulus waveforms recorded by the hydrophone during ABR experiments(Fig. 1) confirmed that our echo-cancelled stimuli were very similar to pure tone waveforms used in other fish hearing experiments (Kenyon et al.,1998).

A Tucker-Davis Technologies System II (TDT, Inc., Gainesville, FL, USA) and BioSig software were used to produce sound stimuli (10 ms stimulus tone bursts in 100 Hz steps from 100 Hz to 1.2 kHz) and record ABR waveforms. Sound bursts were gated using a Blackman window to provide a ramped onset/decay, preventing speaker transients. ABR traces were recorded twice each in two opposing polarities at each frequency and attenuation (250 sweeps each, four total recordings). The polarity of ABR response waveforms is independent of sound stimulus polarity (Kenyon et al.,1998) but the polarity of stimulus artifacts is not. ABR traces of opposite polarity were therefore summed to remove stimulus artifacts. Periodic experiments were also conducted with euthanized fish to ensure that identified ABR responses were not stimulus artifacts.

The two ABR responses at each frequency and sound pressure level were overlaid to assess the response. Sound pressure levels were successively attenuated in roughly 5 dB steps until repeatable ABR waveforms were no longer produced; thresholds were defined as the lowest sound pressure level for which a repeatable ABR trace could be identified visually(Kenyon et al., 1998). Visual threshold assignment provides results similar to quantitative threshold-seeking algorithms (Yan,1998) and remains the standard method of threshold determination in fish ABRs (Kenyon et al.,1998; Casper et al.,2003). Visually assigned thresholds for each subject of a study species were pooled to produce mean audiograms.

Sound pressure levels of all experimental stimuli were calculated from hydrophone recordings (Burkhard,1984). Cursors were placed one cycle apart (peak-to-peak) on either side of the largest (i.e. center) cycle of a tone-burst recording of the hydrophone (Kenyon et al.,1998). The Bio-Sig software then calculated the root mean square(RMS) of the waveform between the cursors, and the appropriate gain calibration factors were applied to determine actual sound pressure level(SPL) in dB re: 1 μPa.

Particle velocity was calibrated using an underwater acoustic pressure-velocity probe (Mk. 2, Acoustech Corp, Philadelphia, PA, USA)containing two built-in units: a piezoelectric, omni-directional hydrophone(sensitivity: –200 dB re: 1V/μPa) and a bi-directional moving-coil geophone (sensitivity: 0.112 V cm^–1 s^–1). The outer housing of this probe was secured in place of the fish ∼25 mm below the water surface with rubberized clamps, and the inner unit of the probe, designed to approximate neutral buoyancy, moved freely in response to our sound stimuli. The omnidirectional hydrophone was suspended by rubber straps to within 2 mm of the pressure–velocity probe. This setup enabled the simultaneous recording of the sound pressure and particle velocity components of the entire range of our experimental stimuli. Subsequently and separately, measurements of particle displacements were recorded in three orthogonal orientations (sensuCasper and Mann, 2006). The vertical component (z axis) of particle velocity had substantially greater amplitudes than the x (horizontal: head-to-tail) or y axes (left to right) at each frequency and attenuation(Table 2). This vertical axis was therefore considered most appropriate for expressing thresholds and plotting particle acceleration audiograms.

We fitted models with three candidate covariance structures (unstructured,compound symmetry, and first order autoregressive [AR(1)]) to the pressure and particle acceleration threshold data. In the unstructured model (UN), each covariance between measures was estimated individually, allowing the data to dictate the appropriate covariance structure. The second covariance structure,compound symmetry (CS), assumed equal covariances between all pairs of observations. The final covariance structure, first order autoregressive[AR(1)], assumed that the correlation between observations is a function of their lag in space or time; adjacent observations are more likely to be correlated than those taken further apart(Littell et al., 2006). As a simple example involving the relationship between evoked potentials at 200,300 and 900 Hz, the UN model would calculate the variance–covariance of every pair of observations individually, the AR(1) model would assume that evoked potentials at 200 and 300 Hz are likely more similar than responses at 200 versus 900 Hz, whereas the CS model would assume equal covariance.

The ABR waveforms and audiograms for sound pressure and acceleration were species-specific, but with some commonalities. Auditory evoked potentials of the six sciaenid fishes (Fig. 2) generally began 10–15 ms after stimulus onset and were complete by 30 ms (⩾400 Hz) or 50 ms (100–300 Hz). Waveform latency varied inversely with frequency and sound pressure level. Sound pressure,particle velocity, and acceleration audiograms of all species(Fig. 3 A–C) exhibited lowest thresholds at low frequencies (100–500 Hz). Velocity and acceleration audiograms were notably flatter at low frequencies. AIC_c values supported the selection of the first order autoregressive [AR(1)] covariance model for both pressure and particle acceleration analyses (Table 3), supporting the assumptions of the AR(1) model. Visual inspection of sciaenid audiograms (Fig. 3) confirms inferences based on AIC_c; ABR responses at adjacent frequencies were therefore more similar to each other than responses at distant frequencies.

Two-way repeated measures ANOVAs demonstrated significant differences between species for both pressure (F_5,48.6=3.17, P<0.02) and particle motion (velocity: F_5,51.4=3.85, P<0.005; acceleration: F_5,52.3=3.00, P<0.02) thresholds. Sound pressure thresholds of spot were significantly higher(F_1,357=5.05, P<0.03) than those of other sciaenids from 300–700 Hz. Among species with swim bladders, thresholds of those with anteriorly projecting diverticulae (weakfish, spotted seatrout and Atlantic croaker) did not differ from those species without diverticulae(red drum and spot; pressure: F_1,357=2.35, P=0.13). Surprisingly, thresholds of northern kingfish were among the lowest at higher frequencies (>600 Hz) even though the swim bladder atrophies in the adults we studied. Detection thresholds varied inversely with frequencies for both pressure (F_11,324=53.01, P<0.001) and particle motion (velocity: F_11,317=78.47, P<0.0001 acceleration; F_11,315=129.24, P<0.0001). Interactions of species and frequencies were significant for both pressure(F_55,319=3.31, P<0.0001) and particle motion(velocity: F_55,314=8.48, P<0.0001;acceleration F_55,314=9.77, P<0.0001) and are visually evident in the crossing of species-specific curves within audiograms(Fig. 3A–C).

All fishes are able to directly detect the particle motion components of sound, yet fish auditory thresholds are generally assessed only for sound pressure levels (Popper and Fay,1993). Few studies have examined hearing thresholds of fishes with respect to both pressure and particle motion sensitivity (Myrberg and Spires,1988; van den Berg, 1985; Lovell et al., 2005; Casper and Mann, 2006). Moreover, direct particle motion simulation of the otoliths may be more relevant to hearing generalist fishes than the detection of sound pressure(Fay and Popper, 1975; Popper and Fay, 1993). In this study, we measured thresholds and expressed audiograms of six sciaenid fishes in terms of both sound pressure and acceleration using an omnidirectional hydrophone and a bi-directional geophone. Our experiments are the first to assess particle motion thresholds in sciaenid fishes and include first reports of pressure audiograms for spotted seatrout, red drum, and northern kingfish.

Sound stimuli during fish audition experiments contain both pressure and particle motion (Parvulescu,1967; Lu et al.,1996; Casper and Mann,2006). Small experimental tanks can have complex particle motion and sound pressure fields, potentially compromising laboratory investigations unless both components of sound stimuli are measured(Kalmijn, 1988; Popper and Fay, 1993). Placing stimulus-generating speakers in air rather than water purportedly reduces the particle motion (Kenyon et al.,1998). Our results, however, demonstrate that speakers in air can produce notable particle motion fields(Table 2). Similar conclusions were reached by others (Casper and Mann,2006). Particle displacements in small tanks are complex, and for an equal sound pressure level they may be greater in tanks than in an unbounded body of water (Parvulescu,1967; Rogers and Cox,1988). General comparisons across studies may be complicated by differences in the location of the sound source in air versus water,the proximity of subjects to the sound source and air-water interfaces(Fay and Edds-Walton, 1997). Such concerns demonstrate the utility of routine particle motion assessment of experimental sound stimuli. Submersible units capable of generating and measuring particle motion are available(Casper and Mann, 2007a; Casper and Mann, 2007b). Future fish audition experiments should attempt to measure and report both the pressure and particle motion components of their experimental stimuli if possible (Popper and Fay,1993; Casper and Mann,2006).

The frequency range detected by the six sciaenids we studied was similar to those of other hearing generalist fishes (100 to <2000 Hz)(Popper and Fay, 1993; Kenyon et al., 1998; Ramcharitar, 2003; Ramcharitar and Popper, 2004; Ramcharitar et al., 2006b). Pressure detection thresholds of sciaenid fishes were significantly lower at low frequencies from 100–300 Hz. Our mean pressure thresholds for spot,weakfish and Atlantic croaker, obtained with a speaker in air, averaged about 6 dB higher than those obtained by others using a speaker in water(Ramcharitar and Popper, 2004; Ramcharitar et al., 2006b). Whether the different results are a consequence of speaker location/type,different levels of background noise, individual variation due to the use of larger animals in our study, or a combination of these factors, is unclear. Overall, our results generally support the conclusion of Ramcharitar et al.(Ramcharitar et al., 2006b)that enhanced swim bladder–otolith relationships within the Sciaenidae can improve auditory sensitivity. Among sciaenids bearing swim bladders, those possessing diverticulae (weakfish, spotted seatrout and Atlantic croaker) had generally but not significantly lower pressure thresholds than species lacking diverticulae (spot and red drum). Swim bladders lacking mechanical coupling to the otic capsule may not enhance sound pressure detection(Yan et al., 2000). Surprisingly, however, we found the lowest sound pressure thresholds at higher frequencies (800–1100 Hz) in northern kingfish, a species with low hair cell densities and swim bladder atrophy in adults(Chao, 1978; Ramcharitar et al., 2001). Since species lacking swim bladders are unlikely to detect sound pressure(Casper and Mann, 2006; Mann et al., 2007), lower`pressure' thresholds of kingfish at higher frequencies are most likely a response to particle motion during the simultaneous presentation of pressure and particle motion stimuli.

Otoliths are biological accelerometers most sensitive to particle motion on their longitudinal axis (Lu and Xu,2002), and the larger otoliths of sciaenid fishes may confer higher sensitivity to the particle motion components of low frequency sounds(Lychakov and Rebane, 1993; Ramcharitar et al., 2006b). Our particle acceleration audiograms demonstrate significantly greater sensitivity at low frequencies (Fig. 3C) and are comparable to results obtained with elasmobranchs(Casper and Mann, 2006). Sciaenid species with enhanced connections between the swim bladder and otic capsule (Atlantic croaker, spotted seatrout, weakfish) may be able to obtain different information from the acoustic particle motion and sound pressure fields (van den Berg, 1985; Ramcharitar et al., 2001). By contrast, sciaenid fishes lacking connections between these organ systems(spot, red drum) are more likely to be responsive solely to particle motion fields (Ramcharitar, 2003). Similar conclusions have been reached for elasmobranch and teleost fishes lacking swim bladders (Mann et al.,2007; Casper and Mann,2006). Adult kingfish (lacking swim bladders) used in our study probably detect acoustic particle motion rather than pressure. The situation is less clear for juvenile kingfish, which do have swim bladders that are distant from the otic capsule (Chao,1978; Ramcharitar,2003). Unfortunately, little is known about ontogenetic differences in pressure and particle motion discrimination in most fishes,including sciaenids.

A better understanding of particle motion thresholds in fishes is required,particularly with respect to hearing relative to the direction of stimulus(sensuFay and Edds-Walton,1997). In our study, maximum particle displacement occurred along the vertical axis (Table 2). But are sciaenids most sensitive to particle motion on this axis? Spawning aggregations, which involve chorusing fish juxtaposed in close proximity(Mok and Gilmore, 1983; Ramcharitar et al., 2006a; Gilmore, 2003), more likely stimulate otoliths in a horizontal direction. Although density and orientation of hair cell bundles in sciaenid fishes differ among species(Ramcharitar, 2003),behavioral sensitivity of oscars (Cichlidae: Astronotus ocellatus) to particle motion did not differ among orthogonal axes(Lu et al., 1996). The individual presentation of particle motion stimuli in various orthogonal Cartesian planes to sciaenids would shed light on this question(Lovell et al., 2005; Casper and Mann, 2007a; Casper and Mann, 2007b).

Dominant frequencies of most sciaenid reproductive and disturbance vocalizations [100–500 Hz(Ramcharitar et al., 2006a)]lie well within the frequency bandwidths of the six species we measured. Therefore, if they are within range, sciaenids should be able to detect each others' species-specific vocalizations, which differ in their dominant frequency, pulse duration, repetition rate, number of pulses per call and sound pressure level (Ramcharitar et al.,2006a). The extent to which these sciaenids use auditory cues to discriminate among species or between individuals in generally noisy estuarine environments remains unknown. This ability has, however, been demonstrated in other soniferous fishes (Ladich,2000; Ripley et al.,2002; Wysocki and Ladich,2003).

Sound pressure and particle motion detection thresholds in sciaenids were lowest at the lower frequencies at which they communicate, but whether these species primarily detect conspecific and congeneric vocalizations viatheir sound pressure, particle motion, or both components of these sounds remains unknown. Communication in sound-producing fishes occurs over relatively short distances and typically in fairly shallow water, where the acoustic near field is dominated by particle motion(Myrberg, 2001; Bass and Clark, 2002; Weeg et al., 2002). Although the characteristics of sciaenid spawning aggregations differ among species,most occur in waters from 3–50 m depth(Saucier and Baltz, 1993).

Sciaenids and other soniferous fishes communicate in shallow coastal and estuarine waters despite high levels of background noise and the theoretical short-distance propagation of low frequency sounds in shallow water(Lugli et al., 2003; Ramcharitar et al., 2006a). Under idealized conditions, we estimate that sciaenid calls may propagate 8–128 m from the source, based on their amplitudes, simple spherical spreading (a loss of 6 dB for every distance doubled) and auditory thresholds(Table 4). Further, our calculations assumed that background noise was below the auditory thresholds,which is unlikely. For example, background ambient noise levels measured in a North Carolina estuary ranged from 110 to 125 dB re: 1 μPa(Sprague and Luczkovich,2004). There is evidence for frequency selectivity amidst background masking within the Sciaenidae, suggesting that some species may still detect certain sounds amidst the masking din of background noise in coastal environments (Ramcharitar and Popper, 2004). Therefore, the distances at which these vocalizations can be heard depend on the source's sound pressure level, the pressure sensitivity and masked hearing ability of the listener, and environmental variables such as background noise, depth, bottom type and habitat complexity (Mann,2006). Unfortunately, masked auditory thresholds are known for only two sciaenids [Atlantic croaker and black drum(Ramcharitar and Popper,2004)]. Additionally, the propagation of pressure and particle motion fields and actual attraction distances of sound sources in shallow,complex, high-scattering, high-background estuarine habitats, are not well understood at present (Mann,2006; Casper and Mann,2006; Lugli and Fine,2007).

In this study, we presented the pressure and particle motion thresholds of six sciaenid fishes, including the first reports of particle acceleration thresholds in this teleost family and first reports of pressure thresholds for three species. Together, emerging data on sciaenid auditory abilities and sonifery support growing efforts to identify and manage their spawning habitats in environments with ever-increasing anthropogenic noise(Wahlberg and Westerberg,2005; Ramcharitar et al.,2006a; Vasconcelos et al.,2007). Sciaenid bioacoustics therefore remains a fruitful research avenue and critical link between sensory physiology and behavioral ecology(Popper et al., 2005; Ramcharitar et al., 2006a; Roundtree et al., 2006). Such research promotes multidisciplinary syntheses that can mechanistically link processes from the cellular to the individual to the population level in support of fisheries management.

We thank B. Deffenbaugh for designing the echo-cancellation software used for these experiments, and M. Luckenbach, R. Bonniwell, and S. Fate for their logistical assistance, patience and extreme flexibility in support of these experiments. We also thank M. Fabrizio for advice regarding statistical methods and J. Luczkovich for providing red drum vocalization sound pressure levels. J. Lucy, J. Smith, Capt. S. Wray, and the vessels Bada Bingand Sea Beaver helped collect study animals. Assistance with animal husbandry was graciously provided by C. Magel, P. Lynch, A. Buchheister, P. Bushnell, J. Woodward. and K. McNamee. This research was partially funded by the National Marine Fisheries Service and by a small grant to A.Z.H. from the International Women's Fishing Association. This is VIMS contribution number 2907.