The origins of ambient biological sound from coral reef ecosystems in the Line Islands archipelago.

The origins of ambient biological sound from coral reef ecosystems in the Line Islands archipelago Simon E. Freeman Marine Physical Laboratory, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California 92093-0238

Forest L. Rohwer Department of Biology, San Diego State University, San Diego, California 92182

Gerald L. D’Spain Marine Physical Laboratory, Scripps Institution of Oceanography, 291 Rosecrans Street, University of California San Diego, Building 4, San Diego, California 92106

Alan M. Friedlander Department of Biology, University of Hawaii, 2450 Campus Road, Honolulu, Hawaii 96822

Allison K. Gregg Department of Biology, San Diego State University, San Diego, California 92182

Stuart A. Sandin Center for Marine Biodiversity and Conservation, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California 92093-0202

Michael J. Buckingham Marine Physical Laboratory, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California 92093-0238

(Received 29 May 2013; revised 19 January 2014; accepted 4 February 2014) Although ambient biological underwater sound was first characterized more than 60 years ago, attributing specific components of ambient sound to their creators remains a challenge. Noise produced by snapping shrimp typically dominates the ambient spectra near tropical coasts, but significant unexplained spectral variation exists. Here, evidence is presented indicating that a discernible contribution to the ambient sound field over coral reef ecosystems in the Line Islands archipelago originates from the interaction of hard-shelled benthic macro-organisms with the coral substrate. Recordings show a broad spectral peak centered between 14.30 and 14.63 kHz, incoherently added to a noise floor typically associated with relatively “white” snapping shrimp sounds. A 4.6 to 6.2 dB increase of pressure spectral density level in the 11 to 17 kHz band occurs simultaneously with an increase in benthic invertebrate activity at night, quantified through time-lapse underwater photography. Spectral-level-filtered recordings of hermit crabs Clibanarius diugeti in quiet aquarium conditions reveal that transient sounds produced by the interaction between the crustaceans’ carapace, shell, and coral substrate are spectrally consistent with Line Islands recordings. Coral reef ecosystems are highly interconnected and subtle yet important ecological changes may be detected quantitatively through passive monitoring that utilizes the C 2014 Acoustical Society of America. acoustic byproducts of biological activity. V [http://dx.doi.org/10.1121/1.4865922] PACS number(s): 43.30.Nb, 43.30.Sf, 43.80.Ka [MS]

I. INTRODUCTION

The existence of an acoustic cacophony produced by organisms in coastal underwater environments was first identified during the Second World War, and results from several studies were published soon thereafter (Johnston et al., 1947; Everest et al., 1948; Knudsen et al., 1948). These early works and a great deal of subsequent research have identified a broad range of marine taxa that generate sounds, ranging from crustaceans through to fishes and marine mammals. Given that the sound field contains information relating to the diversity and density of marine sound-generating J. Acoust. Soc. Am. 135 (4), April 2014

Pages: 1775–1788

taxa, passive acoustic recording holds great promise for evaluating and monitoring certain aspects of ecosystem state (Payne et al., 2008). The soundscapes of most coastal regions are as yet unsampled and unknown. However, many recordings collected by single-element hydroacoustic recorders in tropical and subtropical regions are dominated by broadband sounds with varying spectral structures that follow a strongly diel pattern in which pressure spectral density levels are typically higher at night and are independent of weather (Everest et al., 1948). Consequently it must be that these sounds are of biological origin. The recorded sound fields are often an

0001-4966/2014/135(4)/1775/14/$30.00

C 2014 Acoustical Society of America V

1775

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 146.189.194.69 On: Wed, 08 Apr 2015 10:03:46

ensemble average of many short duration sounds from a great many sources (Cato and McCauley, 2002). Such fields are not well characterized by recorders with a relatively low sampling frequency, nor are the sources of the sounds easily verified. Low frequency, pulsed sounds recorded over coral reefs are largely known to originate from soniferous fishes (i.e., McCauley and Cato, 2000). Much of the higher frequency sound recorded in tropical waters has been attributed to snapping shrimp (Johnston et al., 1947; Knudsen et al., 1948; Everest et al., 1948; Takemura and Mizue, 1968; Cato, 1993; Readhead, 1997; Au and Banks, 1998; Chitre et al., 2012). Johnston et al. (1947) first identified the shrimp to be of the genera Crangon and Synalpheus. The family Alpheidae, to which they belong, now contains approximately 38 genera and are typified by their ability to create a loud “snap” sound through the implosion of a cavitation bubble produced by a high-velocity jet of water extruded from one large claw (Versluis et al., 2000). In fact, snapping shrimp produce transient acoustic emissions of a sound pressure source level that can be up to 215 dB re 1 lPa @ 1 m (Schmitz, 2002). The dominant portion of the snapping shrimp waveform is highly impulsive (Au and Banks, 1998) and consequently produces a spectrum that varies only very weakly with frequency. Spectral analysis of the entire “snap” waveform reveals that the pressure spectral level of a single snap does not vary by more than 20 dB between 20 and 200 kHz. Individual snaps create distinct spectral peaks below 20 kHz. Such distinct peaks are created by the interaction of a low-level precursor with the rising portion of the main waveform (Au and Banks, 1998). Fourier analysis from averaging over multiple snaps reveals a monotonically decreasing pressure spectral density (Knowlton and Moulton, 1963; Cato and Bell, 1992; Kim et al., 2010). This spectral smoothing is likely reflective of variation in the size and shape of individual snapping claws, both between conspecifics of differing size, and different species. Although a number of studies identify the sonic emissions of snapping shrimp activity as the major contributor to ambient underwater biological noise in tropical and subtropical regions, these same studies and others show a wide variation in the spectral characteristics of ambient underwater sound by region (Everest et al., 1948; Cato, 1980; Zakarauskas, 1986; Bjorno, 1987; Cato and Bell, 1992; Readhead, 1997; Bardyshev, 2007; Radford et al., 2008a; Bardyshev, 2010; Radford et al., 2010). In contrast to the broadband characteristics of snapping shrimp noise, longterm, averaged pressure spectral density estimates can show distinct spectral peaks. In addition, few studies have yet to consider acoustic frequencies above 12 kHz or so in the analysis of ambient noise (Radford et al., 2010; Kim et al., 2011). While snapping shrimp noise is the dominant contributor to the sound field in many locales, it is likely that some contributions to the sound field originate from biological sources other than the cavitation sounds from shrimp. These incoherent contributions may alter the spectral characteristics of the sound field, even in the presence of far louder snapping shrimp sounds. Incoherent addition of spectra may be one cause of the variation in the spectral 1776

J. Acoust. Soc. Am., Vol. 135, No. 4, April 2014

characteristics of ambient sound recorded in environments typically inhabited by snapping shrimp. An analysis is presented in this article of ambient sound recorded in five tropical coral reef environments in the Line Islands archipelago. The five environments represent a variety of oceanographic conditions and ecological state. Acoustic data are compared with time-lapse, in situ underwater photographs recorded concurrently at the same locations. For comparison with the coral reef data, the acoustic signatures of hard-shelled benthic organisms (blue-eyed spotted hermit crabs, Clibanarius diugeti) were recorded in a controlled, salt-water aquarium. The implications of these observations for ecological evaluation and monitoring of tropical ecosystems through passive acoustic recording are then discussed. II. EXPERIMENTAL SETUP AND PROCEDURE A. Field data collection

Acoustic time series were collected on a research expedition to the Line Islands Archipelago [Fig. 1(A)] during October and November 2010. Recordings were obtained from Christmas (Kiritimati), Fanning (Tabuaeran), Washington (Teraina), and Jarvis Islands, as well as Kingman Reef. The first three sites are part of the Republic of Kiribati, while the latter two are no-take marine protected areas and are part of the United States Pacific Remote Islands National Marine Monument [Fig. 1(B)]. Recordings were taken using a portable battery-powered “DSG Ocean” recorder (Loggerhead Instruments, Inc.) equipped with a single hydrophone (High Technology Instruments, Ltd., type 96-MIN, SN# 437094). System sensitivity was 169.9 dB re 1 V/lPa, with a frequency response between 0.1 and 20 kHz that does not vary by more than 0.5 dB of this value (Loggerhead Instruments, 2011). Data sampling was conducted at 16-bit resolution at a rate of either fs ¼ 50 or 80 kHz (corresponding to Nyquist frequencies of 25 and 40 kHz, respectively) depending on available memory capacity and anticipated deployment length. Recording duty cycles of either 5 or 7 min out of every 10 min were utilized to extend the total data collection period. Recording durations at each island ranged from two to five days and were dependent on weather and ship scheduling. Self-contained underwater breathing apparatus (SCUBA) divers deployed the recorder at an approximate depth of 10 m on the outer reefs of the four islands, and at a similar depth within the lagoon at Kingman Reef [Fig. 1(B)]. An anchoring rope secured the recorder to dead coral while a closed-cell foam float enabled the positioning of the hydrophone receiving element at a depth of approximately 2 m above the substrate. The float additionally kept the mooring line in tension to prevent contact between the hydrophone and line [Fig. 2(A)]. Canon D10 underwater cameras were temporarily mounted to the reef near hydrophone deployment sites and positioned to record biological activity over a range of substrate types including sand, live and dead coral, and algae-dominated reef patches. The remote cameras were programmed to take one image every five minutes, with the built-in flash automatically illuminating the scene during Freeman et al.: Origins of coral reef sound


FIG. 1. (A) Chart showing the location of the five field sites in the Line Islands Archipelago relative to the Hawaiian Islands in Mercator projection. (B) Charts of each island or reef where hydrophone and camera deployments took place. A star denotes the approximate location of each hydrophone deployment.

periods of darkness. The cameras were limited to capturing images that covered approximately 1 m2 of seafloor area because of range limitations imposed by the built-in flash and the 12.1 Mpixel resolution of the image sensor. Battery capacity restricted use of these cameras to durations of approximately 6 to 8 h, between 16:00 until midnight local time. Between deployments, the Loggerhead recorder and cameras were cleaned, recharged, and previously recorded files were downloaded to a personal computer. B. Aquarium data collection

In order to characterize the sounds of hard-shelled benthic organisms, ten blue-eyed spotted hermit crabs (Clibanarius diugeti) with a mean weight of 1.7 g (including shell) were obtained from commercial distributors and sounds of their activity were recorded in a quiet laboratory tank environment. Sounds produced by hermit crabs are by-products of locomotion, foraging, and/or feeding behavior. In general, crustaceans are capable of hearing sound J. Acoust. Soc. Am., Vol. 135, No. 4, April 2014

(Budelmann, 1992). However, from the observations presented in this paper, there is no evidence to suggest that hermit crabs produce sonic emissions for the express purpose of social interaction or communication. It may be possible that hermit crabs use these sounds incidentally, that is, to identify and locate similar organisms and thus locations where a desirable resource is plentiful, though investigation of this hypothesis is outside the scope of this work. Although sound generated by only one type of organism was sampled in this experiment, it is speculated that these sounds may be generally representative of acoustic emissions produced by hard-shelled benthic organisms and their interaction with coral reef substrate. The sound generation mechanisms observed are not exclusive to Clibanarius diugeti or hermit crabs in general. The crabs were placed in a rectangular glass-sided saltwater aquarium (500 400 and 300 mm water depth) along with two pieces of “live rock,” a term used by aquarists for pieces of dead coral covered in algae and natural bacterial growth that assist in water conditioning, approximately 3 kg Freeman et al.: Origins of coral reef sound

1777


FIG. 2. (A) Diagram showing how the Loggerhead Instruments hydrophone system was deployed at each field site at a depth of approximately 10 m. The omnidirectional hydrophone element (black object atop the gray cylinder) was placed approximately 2 m above the reef. (B) Schematic showing the experimental setup in which sounds created by the hermit crab species Clibanarius diugeti were recorded.

in weight each. A layer of coralline sand covered the floor of the aquarium to a depth of 30 mm. The aquarium was equipped with a tropical marine filtration and water conditioning system that was turned off during periods of acoustic recording to reduce background noise. Recordings were conducted overnight, when human-generated noise was at a minimum. The volume of water inside the tank was sufficient to limit rapid changes in temperature or the level of oxygenation for the duration of the experiment. No marked changes in organism behavior were observed throughout the experiment and all hermit crabs were observed moving and/or feeding after the completion of the experiment. Acoustic recordings were made using a hydrophone manufactured by the International Transducer Center (ITC) corporation (Type 6080C, SN # 1545) with a sensitivity of 156 dB re 1 V/lPa and a frequency response that remains within 2.5 dB of this value between 1 and 30 kHz (International Transducer Corporation, 1996). Data sampling was performed using an IntelV “High Definition Audio” specification sound card set at 24-bit resolution and a sampling rate of fs ¼ 96 kHz (corresponding to a Nyquist frequency of 48 kHz). The hydrophone was mounted approximately 200 mm from the live rock, mid-water in the aquarium at depth level with the upper portions of the rock. Care was taken to ensure the hydrophone could not contact the bottom or aquarium walls, and that hermit crabs could not come into direct contact with it [Fig. 2(B)]. The aquarium was set up in a quiet room that contained no other powered-on electronic equipment or noisegenerating devices. However, spectrograms revealed tonal noise peaks between 0 and 7 kHz, likely generated by the air conditioning system. Although this noise was a dominant feature in the time-series, it was almost entirely restricted to frequencies below 6 kHz. R

1778


Furthermore, pressure spectral density level estimates of background noise above approximately 6 kHz remained approximately 40 dB lower than the level of sound generated by the hermit crabs. C. Aquarium resonance

The perpendicular walls of glass-sided, rectangular aquaria and the typically flat pressure-release surface of the water contained create a highly reverberant acoustic environment within the tank. Unless recordings are sufficiently time-gated so as only to capture the primary arrival (for which source and receiver location must be precisely known, and the tank large enough to accommodate the duration of the signal), such environments can influence the received waveform considerably (Akamatsu et al., 2002; Okumura et al., 2002). Arrivals of a transmitted signal over multiple paths lead to frequency-dependent interference and tank resonance. Such resonance can alter the received pressure spectra near resonant frequencies from what would be recorded if transmission occurred in an open-water environment, although resonance typically weakens as mode number increases as observed by Akamatsu et al. (2002). The frequencies of the first three resonant modes of each Cartesian direction were calculated for the experimental setup described using the methodology described by Akamatsu et al. (2002) (Table I). The experimental setup used in this study was estimated to create resonant frequencies of between 3.54 kHz (first mode, horizontal) and 10.61 kHz (third mode, vertical). Pressure spectral densities of signals in this band may be overemphasized near the frequencies displayed in Table I, independent of the receiver position. As one tank dimension is longer than the others the aquarium may be approximated as a waveguide, although in Freeman et al.: Origins of coral reef sound


TABLE I. Estimated resonant mode frequencies for the first three modes orthogonal to the horizontal and vertical faces of the test aquarium. Frequencies are indicated in kHz. Horizontal mode number (along tank width, y) Vertical mode 1 (kHz) Horizontal mode number (along tank length, x)

1 2 3

Vertical mode 2 (kHz) Horizontal mode number (along tank length, x)

2

3

3.54

4.84

6.46

4.42 5.52 6.98 5.59 6.50 7.78 Horizontal mode number (along tank width, y) 1

1 2 3

Vertical mode 3 (kHz) Horizontal mode number (along tank length, x)

1

1 2 3

2

3

5.66 6.56 7.83 6.25 7.07 8.26 7.12 7.86 8.94 Horizontal mode number (along tank width, y) 1

2

3

8.03 8.46 9.12

8.69 9.08 9.71

9.68 10.04 10.61

this study the largest tank dimension was only 20% greater than the next. The “attenuation distance” defined by Akamatsu et al. (2002) corresponding to a horizontal waveguide along the longest dimension of the tank is complex above 9.56 kHz for the third mode, meaning that if the spectrum at 9.56 kHz does not appear over-represented, it is unlikely that higher order waveguide modes will substantially alter the ambient spectra. D. Data processing

Acoustic data collected on the Loggerhead hydrophone in the Line Islands were converted to wav file format and then processed using MathworksV MATLABV (R2011b). After low-pass finite impulse response (FIR) filtering (corner frequency 0.9 fs/2), Fourier analysis was conducted using 8192-point fast-Fourier transforms (FFT’s) of 50% overlapped time-series segments windowed with a Kaiser–Bessel (a ¼ 2.5) function. Unilateral frequency domain data were scaled to normalize for window magnitude and pressure spectral density estimates were obtained by scaling from voltage to pressure input. A 200-point moving average of the frequency domain spectra was taken before pressure spectral density estimates were generated. Estimates were then plotted sequentially to obtain a spectrogram of the acoustic time-series. Visually detected organisms from images taken by remote cameras were counted by trained observers and invertebrates identified to the taxonomic assemblies crustacea (crabs and shrimp), gastropoda (snails), ophiuroidea (brittle stars), holothuroidea (sea cucumbers), and annelida (segmented worms). Acoustic data collected with the ITC hydrophone from the enclosed tank experiments were also processed using V MATLAB (R2011b). A low-pass filter (corner frequency fc ¼ 20 kHz) built in to the digitizer excluded analysis R

R


R

of higher frequencies. A high-pass FIR filter was applied (fc ¼ 6 kHz) to remove noise from electronic devices and air conditioning. Due to the temporally sparse and highly transient nature of sounds generated by the hermit crabs in the tank, a band-limited, level-threshold detection filter was implemented in order to estimate the mean spectral properties of the sounds. Limiting the detection filter to a band between 11 and 17 kHz avoided potential contamination of filtered sound by the first three modes of tank resonance. A threshold level of 61 dB re 1 lPa2/Hz provided a compromise between rates of false-alarm detection (of noise pollution from outside the tank) and a sufficient number of true detections to characterize the sounds made by the hermit crabs. Fourier analysis was performed using the same procedure as for the acoustic data from the Line Islands. Pressure spectral density levels were created without averaging, leading to one spectral estimate every 0.043 s. Estimates in which pressure spectral density level in any bin between 11 and 17 kHz exceeded the threshold level were retained and concatenated into spectrograms. III. RESULTS

Comparisons between spectrograms of the sound field at each island or reef over two to four deployment days (averaged by time of day to result in a single spectrogram spanning one night for each site) reveal a common temporalspectral pattern [Fig. 3(A)]. Pressure spectral density level in the 11–17 kHz band increases by 4.6–6.2 dB, depending on location, within 10 min of sunset and decreases by an equivalent amount within 10 min of sunrise. The pressure spectral density at any given time shows a pronounced structure: a broad peak with a maximum level between 14.30 and 14.63 kHz, the exact peak frequency varying with location (Table II). These patterns are ubiquitous over all the days over which acoustic data were collected, at every field site, and were independent of weather conditions and sea state. The spectral peak rises approximately 4 to 8 dB above a noise floor that does not vary by more than 4 dB between 3 and 11 kHz and decreases above 17 kHz. Representative daytime and night-time pressure spectral densities for each field site, averaged over an hour (16:00 to 17:00 for day-time and 19:00 to 20:00 for night-time), are shown in Fig. 3(B). The pressure spectral density level in the 11–17 kHz band remains approximately 3–7 dB higher than in the adjacent 6–11 kHz and 17–20 kHz bands during the day, suggesting that the source of this sound remains present but in a reduced form. Anthropogenic noise from SCUBA divers and small watercraft is visible in the averaged recordings taken between 16:00 to 17:00 at some sites. This noise occurs primarily in the 16–18 kHz and 0–6 kHz bands and so does not affect the band of interest between 14 and 15 kHz. Comparisons with generalized spectral representations of ambient noise sources (National Research Council, 2003, adapted from Wenz, 1962) show that the spectra of ambient noise recorded at these field sites differs markedly from well characterized sources, such as wind- and rain-generated noise (Fig. 4). In particular, noise above 10 kHz is well above the limits of prevailing noise and exhibits a different spectral shape from the monotonically decreasing spectrum Freeman et al.: Origins of coral reef sound

1779


FIG. 3. (Color online) (A) Spectrograms of acoustic pressure spectral density collected at the five field sites in the Line Islands archipelago. Each spectrogram represents acoustic recordings spectrally averaged by day, from 16:00 to 08:00 local time. Deployment durations ranged from two days duration (Christmas Island) to four (Kingman Reef). During the deployments, sunset and sunrise occurred at approximately 18:30 and 06:30 local time, respectively. (B) Estimated pressure spectral density averages for the five field sites during daytime and night-time periods. Solid lines represent averaged pressure spectra between 19:00 and 20:00 local time, while dashed lines represent the same between 16:00 and 17:00 local time. The thin lines above and below indicate one standard deviation about the daytime and night-time pressure spectral density estimates. High frequency noise seen above 16 kHz during day-time recording at some sites is due to the presence of SCUBA divers, and does not affect the analysis of ambient noise in this study.

of the generalized representation of wind noise (Knudsen et al., 1948). The results of time-lapse photo frame analyses are shown in Fig. 5, superimposed upon pressure spectral density levels in the 11 to 17 kHz band from acoustic data

recorded simultaneously at the same field sites. The heterogeneous distribution of substrate type and fauna on reef habitats, and the limited area photographed by each camera, lead to substantial variation in organism counts between sites. Nevertheless, photo frame analysis shows a marked increase

TABLE II. Summary of peak pressure spectral density levels and respective peak frequencies during day-time (16:00–17:00 local time) and night-time (19:00–20:00 local time). Standard deviations are shown in parentheses.

Site location Kingman Reef Washington Island Fanning Island Christmas Island Jarvis Island

1780

Day-time peak pressure spectral density (rþ,r), dB re 1 lPa2/Hz

Night-time peak pressure spectral density (rþ,r), dB re 1 lPa2/Hz

Day-time peak frequency (kHz)

Night-time peak frequency (kHz)

62.28 (0.51, 0.57) 69.73 (0.58, 0.67) 67.56 (0.67, 0.79) 64.44 (0.91, 1.17) 70.14 (0.69, 0.83)

68.35 (0.44, 0.48) 75.48 (0.54, 0.62) 73.71 (0.48, 0.53) 69.01 (0.78, 0.96) 75.59 (0.54, 0.62)

14.63 14.48 14.50 14.62 14.51

14.32 14.32 14.34 14.30 14.35


Freeman et al.: Origins of coral reef sound


FIG. 4. (Color online) Approximated pressure spectral density levels from various noise sources in the ocean (National Research Council, 2003, adapted from Wenz, 1962). The solid line from 100 Hz to 20 kHz shows the approximate averaged pressure spectral density characteristics of ambient noise from Jarvis, Washington, and Fanning Islands between 19:00 and 20:00 local time.

in benthic organism activity at night across all field sites, the onset of which coincides with the increase in pressure spectral density over the 11 to 17 kHz band. At all sites, crustaceans (including hermit crabs, other crabs, and shrimp) were the dominant organism group identified in the photographs, followed by gastropods (snails) and ophiuroids (brittle stars). Aquarium recordings of hermit crabs were typified by periods of silence punctuated by transient sounds. Although quiet, sounds emanating from the aquarium could be clearly heard by the human ear and visually matched to hermit crab activity. These acoustic emissions subjectively sounded similar to the noise produced when fingernails are gently scraped over rough porcelain. Visual observation confirmed that the sounds were a result of the interaction between the body parts of the hermit crab (e.g., exoskeleton and shell) and the rocky substrate. Such interactions occurred when animals were observed moving over and feeding on material growing on the surface of the live rock. A threshold pressure spectral density of 61 dB re 1 lPa2/Hz in the 11–17 kHz band resulted in 211 s of data that exceeded this threshold from 16 h and 46 s of continuous recording in the aquarium. Thresholds of 59 and 63 dB re 1 lPa2/Hz resulted in the collection of 358.4 and 109.1 s of data from the same band, respectively. A concatenated spectrogram of the filtered sound for a threshold of 61 dB re 1 lPa2/Hz is shown in Fig. 6(A). Because of reflections from the pressure release surface, rock, walls, and floor of the tank in addition to absorption by the sand covering the base of the tank, no accurate full-band estimate of source level could be made. J. Acoust. Soc. Am., Vol. 135, No. 4, April 2014

FIG. 5. (Color online) Organism counts from time-lapse photographs taken at each field site, plotted with simultaneously recorded peak pressure spectral density in the 11–17 kHz band (solid line). The dashed line shows the time-series of cumulative organism counts per photograph. Organism types are displayed as solid circles (crustaceans—crabs and shrimp), solid squares (gastropods snails), solid diamonds (ophiuroids—brittle stars), solid downward-pointing triangles (holothuroids—sea cucumbers), and solid right-pointing triangles (annelids—segmented worms). Organism count levels are shown on the vertical axis to the left, pressure spectral density in dB re 1 lPa2/Hz is shown on the right vertical axis. Local sunset at all sites took place at approximately 18:30 local time. Freeman et al.: Origins of coral reef sound

1781


FIG. 7. (Color online) Densities of intervals between detections for a detection threshold of 61 dB re 1 lPa2/Hz between intervals of 0 to 60 s (black dots). A change in slope is observed at an interval of 4.40 s (vertical dashed line). Exponential density model curves are shown for the two regimes. The steeper curve at left corresponds to a rate parameter of 0.39 snaps per second. The shallower curve at right corresponds to a rate parameter of 0.064 snaps per second.

FIG. 6. (Color online) (A) Concatenated spectrogram showing the estimated pressure spectral density of hermit crab sounds that exceeded 61 dB re 1 lPa2/Hz above 11 kHz. The equivalent of 211 s of sound recorded at fs ¼ 96 kHz was collected over 16 h and 46 s. (B) Averaged pressure spectral density curves from 211 s of concatenated hermit crab sounds (solid line) and from 1 h of ambient reef noise recordings at each of the five field sites from 19:00 to 20:00 local time (dashed lines). The narrow-band peak in the solid line at approximately 6.4 kHz was due to machine noise that could not be turned off during hermit crab recordings.

However, when using a threshold pressure spectral density of 61 dB re 1 lPa2/Hz, the average maximum pressure spectral density received over the 211 s of recording was 59 dB re 1 lPa2/Hz for frequencies between 14.17 and 14.57 kHz. Although this sound level is greatly reduced compared to those observed in recordings from the Line Islands [Fig. 3(A)], comparisons of pressure spectral density recordings taken between 19:00 and 20:00 local time from the five field sites and aquarium-based hermit crab recordings show a peak centered over a similar band—between 14.3 to 14.63 kHz [Fig. 6(B)]. The distribution of time intervals between recorded sounds that reached the threshold of the detection filter is shown on a semi-logarithmic axis in Fig. 7. Time intervals were partitioned into 200 histogram bins and truncated to a maximum interval time of 60 s to remove rare outliers. Two regions of data, modeled by two separate rate parameters, were apparent and curves were fitted to the data based on a modified exponential distribution, as follows: yn ¼ An kn ekn xn ;

(1)

where x represents the intervals between detections and y is the output density of occurrence. The subscript n represents 1782


the nth region and the constant An is a scaling factor not equal to unity that is specific to each regime. As such, the relationship between xn and yn is not modeled by a typical exponential probability density function unless the scaling factor An is accounted for during normalization. This parameter is influenced by the probability of detection and the probability of a false alarm. These probabilities in turn affect the measured rate parameters. Correct normalization of time interval data requires further analysis of signal, noise, and detector properties outside the scope of this report. Regardless of normalization, these data suggest two distinct relationships between detection interval and occurrence frequency, differentiated by detection interval. Two least squares fitting algorithms using Eq. (1) were implemented simultaneously on the distribution of time intervals, each restricted to a segment separated by a “changeover” interval. An optimal changeover interval of 4.40 s for a threshold pressure spectral density of 61 dB re 1 lPa2/Hz was determined through minimizing the residuals arising from both least-squares operations. Below an interval of 4.40 s, the data are described by an event rate k1 ¼ 0.38 detections per second and A1 ¼ 0.13, while an event rate of k2 ¼ 0.064 detections per second and A2 ¼ 0.26 was observed for data where the intervals between detections were greater than 4.40 s. IV. DISCUSSION A. Ambient noise recordings

In many tropical and temperate coastal seas, the temporal variation in the sound field around shallow reefs is strongly diurnal. Sounds above approximately 6 kHz are unlikely to be from non-biological sources in calm to moderate seas (Wenz, 1962). Although light rain in calm conditions has been shown to produce a spectral peak at around 15 kHz, the simultaneous occurrence of drizzle and low sea state is not a precisely regular diurnal event and the spectral peak produced by drizzle is broader in frequency than the Freeman et al.: Origins of coral reef sound


peak recorded in the Line Islands (Scrimger et al., 1989; Ma et al., 2005). Thus, the nightly chorusing described here is almost certainly created by animals. The biological origins of some diurnal acoustic patterns have been documented in specific cases, including fishes (Cato, 1980; D’Spain and Batchelor, 2006), echinoderms (Radford et al., 2008b), and the aforementioned snapping shrimp (Everest et al., 1948). However, to our knowledge no previous work offers a valid explanation for the frequency characteristics of the diurnally varying sound field in these recordings. For comparison with the Line Islands spectra shown in Fig. 3, ambient sound measurements were made in a shallow-water environment known to be populated by large numbers of snapping shrimp: the Scripps Institution of Oceanography pier in San Diego, California (Fig. 8). Snapping shrimp show an apparent preference for living inside and among pier structures (Ferguson and Cleary, 2000). The pier pilings are covered in sponge and algae dominated sessile communities. Removal of these layers reveals a relatively monospecific and abundant community of snapping shrimp beneath. Recordings were made with the same equipment used to record hermit crab sounds and using identical signal processing parameters. Figure 8(A) shows an example spectrogram of 60 s duration recorded from 19:50

local time (apparent sunset occurred at 18:14 local time). The spectrogram shows steady machinery noise below 2 kHz (created by water pumps operating on the pier) and the broad-band impulsive sounds created by nearby snapping shrimp. Figure 8(B) shows an averaged pressure spectral density of the same recording, truncated below 2 kHz. The averaged spectrum decreases monotonically by approximately 10 dB between 2 to 18 kHz, likely due to the characteristics of the snapping shrimp noise itself (Au and Banks, 1998) and frequency-dependent attenuation by the surrounding environment. In contrast to the recordings taken in the Line Islands, an environment in which snapping shrimp noise is not augmented by sounds from other organisms is relatively free of broad spectral peaks. The sounds produced by both snapping shrimp and the interaction between benthic invertebrates and the substrate are both transient in nature. Assuming a large population of individuals produce such sounds within the audible range of the hydrophone, it is likely that a number of individual sounds will arrive at the hydrophone within a given sampling period corresponding to one realization of the spectra (equivalent to the length of the FFT). Consider a simplified case in which S2(x) represents the pressure spectral density of an individual snapping shrimp spectrum and I2(x) is the representative spectrum produced by other hard-shelled benthic invertebrates. Assuming that all sounds arriving at the hydrophone are incoherent over time, then the pressure spectral density estimate for a given realization would be XðxÞ ¼ Nshrimp S2 ðxÞ þ Ninvertebrate I2 ðxÞ;

FIG. 8. (Color online) (A) Spectrogram of a 60 s recording of ambient sound from under the pier at the Scripps Institution of Oceanography, San Diego. The recording was made during the evening chorus at 19:50 local time (sunset occurred at 18:14 local time). Snapping shrimp impulses are visible as broad-band transient sounds. The low-frequency tones visible below 2 kHz were caused by pump noise. (B) Averaged pressure spectral density from the same recording from 2 to 20 kHz showing the gently sloping monotonic decrease in spectral density attributed to snapping shrimp sounds. J. Acoust. Soc. Am., Vol. 135, No. 4, April 2014

(2)

where X(x) is the total estimated pressure spectral density, Nshrimp S2(x) represents the snapping shrimp spectrum multiplied by the number of snapping shrimp sounds received over the time period required to form the spectral estimate, and Ninvertebrate I2(x) represents the equivalent for other invertebrates. Given that the energetic cost of a snapping shrimp “snap” is high in comparison to the production of incidental noise by invertebrates, it is possible that invertebrate-produced sounds are recorded more frequently, especially if their numbers greatly exceed those of snapping shrimp. If the number of invertebrate sounds incoherently adding to the spectrum is larger than the number of snapping shrimp sounds received over one realization period, the contribution to X(x) by Ninvertebrate I2(x) may remain discernable in certain bands, despite S2(x) I2(x). As S2(x) is comparatively “white” to the more “colored” spectrum I2(x), the incoherent addition of I2(x) may result in the peaked spectra observed in so many environments dominated by snapping shrimp sounds. Indeed, such spectral peaks are often accompanied by a spectral “floor” that may represent the broad-band contribution to the ambient sound field by snapping shrimp. Beyond the scope of this study are considerations of range-dependent attenuation and the dispersive and/or attenuative characteristics of the benthic material. Spectral modification of distant snapping shrimp cavitation sounds by bottom interaction and frequency dependent dispersion may substantially change their recorded characteristics. However, Freeman et al.: Origins of coral reef sound

1783


such attenuation would typically act as a weak low-pass filter, rather than a band-pass filter of the type that would be needed to produce the spectra observed here. Spectral comparisons between sites are of particular interest given the gradient of human disturbance across the northern Line Islands (DeMartini et al., 2008; Sandin et al., 2008). Recordings at all sites were consistent in that the distribution of the dominant components of pressure spectral density above approximately 1 kHz were broadly similar, as seen in Figs. 3 and 7(B). Such consistency is suggestive of similar biological processes occurring across this gradient of human disturbance, in contrast to visual survey results. However, the mechanisms through which anthropogenic impacts such as pollution and fishing influence the underwater sound field are virtually unknown. Additionally, Jarvis Island is located along latitude 0.37 S, within the region of equatorial upwelling. Jarvis Island experiences distinctly different oceanographic conditions from the other islands discussed in this study, which are in essentially oligotrophic tropical surface waters and experience relatively little nutrient upwelling, with the exception of island wakes (Coutis and Middleton, 1999). The similarity between the spectral components of the sound field recorded at Jarvis and the other islands suggests that the processes that lead to the creation of such sounds may be less dependent on these regionalscale oceanographic conditions as they are to other, larger scale forcing. While the recordings from Jarvis, Washington, and Fanning Islands were taken on the outer reef slope, recordings at Kingman Reef and Christmas Island were taken from the inner lagoon and lee harbor, respectively. As the recordings from the latter two sites were consistently lower in level across much of the sampled spectrum, it is likely that the local environment and/or the subsequent effect of the environment on the local ecology influenced the received levels. However, until acoustic data are collected from regions with similar physical characteristics under similar conditions, conclusive connections between received level and ecological state cannot be made. The relative ratio of spectral levels between day and night may still be compared between sites. This measure provides an indication of the relative change in acoustic emissions between the two time periods, and is compared with ecological metrics estimated by previous studies of the same areas (Fig. 9). While the percentage of coral cover, fish biomass, and percentage of piscivorous fishes are generally correlative, there is no discernable correlation between these data and the ratio of night-time and day-time acoustic spectral density level at each site. However, the limited spatial sample (n ¼ 5) of acoustic recordings results in a low probability of detecting a real correlation. A statistically robust investigation of the hypothetical link between the characteristics of ambient reef noise and ecological state will require a more comprehensive acoustic sampling strategy that, in part, includes a greater number of sampling locations. The biological activity that produces the recorded sounds may be relatively independent of previously measured ecological metrics, or is indicative of other ecological factors that have not yet been quantified by these metrics. As 1784


FIG. 9. (Color online) Plots showing comparisons between linear ratios of night-time over day-time maximum average pressure spectral density at the peak frequency between 14 and 15 kHz and ecological survey data from equivalent field sites. Error bars show 61 standard deviation where available. Asterisks denote data from Sandin et al. (2008), while the plus sign denotes unpublished data from the same study.

recordings of sounds produced by benthic organisms are a direct measurement of activity level, analysis of these sounds may provide information regarding the ecological contribution made by cryptobenthic organisms. This group of creatures, significant in biomass, diversity, and influence on reef ecology, is almost entirely missed by the current methods of reef state assessment such as fish counts and estimations of percentage of coral cover (Plaisance et al., 2011). B. Remote cameras

Time-lapse remote photography provides a unique discrete-time view of animal activity. This imaging technique is especially useful at night, when continuous light sources required for video capture or direct observation would influence organism behavior. Figure 5 shows that the increase in benthic invertebrate activity at night correlates with the increase of pressure spectral density level in the 11–17 kHz band. As each species occupies a unique ecological niche, measured activity levels are dependent on the type of benthic environment within the limited range of the camera and strobe. Combined with territorial behavior observed in some reef-dwelling species and the highly variable composition of the sea floor in reef environments, the limited optical range of each camera creates variance in observed results. Furthermore, the efficiency of underwater acoustic propagation leads to an integration of acoustic information from a much larger area than that photographed by each camera, so that the photographic data only offers information regarding a small subset of the acoustically surveyed area. Crustacea were the group primarily responsible for the increase in organism counts at night. Shelled mollusks Freeman et al.: Origins of coral reef sound


(gastropods) and brittle stars (ophiuroids) were the next most abundant groups after sunset. All of these organism groups are characterized by either a rigid exoskeleton made of chitin and calcium carbonate and/or a hard external shell made of calcium carbonate, or a test (calcite ossicles fused to form armor plating) in ophiuroids. Such rigid external structures suggest that these organisms are capable of interacting with hard substrate in a way that creates sounds similar to those created by the hermit crabs in the aquarium. Indeed, a sound generation mechanism for larger, rigid-bodied benthic organisms (sea urchins) has been reported by Radford et al. (2008b). The general increase in benthic invertebrate activity at night is inversely related to ambient light levels. Sight-hunting predators such as octopi and carnivorous reef-dwelling fishes are known predators of benthic invertebrates (Hiatt and Strasburg, 1960; Hobson, 1972). Thus, the cover of darkness reduces foraging restrictions on benthic invertebrates. The broad spectral peak in ambient sound between 11 and 17 kHz that occurs at night is also present during the day, although the estimated pressure spectral density level is reduced by 4.7–6.9 dB. While remote cameras show relatively low levels of invertebrate activity above the substrate during daylight hours, sound-producing activity may continue beneath the outer surface of coral reefs. A time lag is observed between the increase in pressure spectral density level and organism counts at every location. A coral reef is a highly porous environment in which the surface area inhabited by small benthic creatures is almost entirely below the outer skin of the reef. The uppermost surface of the reef is relatively devoid of such creatures during the day, when predators that rely on vision, such as the two-spot red snapper Lutjanus bohar in the Line Islands, are common. While hidden benthic organisms cannot be seen, propagation of sound from within the reef permits their relative activity levels to be heard. While the increase in the observed number of benthic organisms after sunset is delayed, the increase in activity level while the organisms are still within their subterranean habitat may be tracked more accurately through acoustic recordings. Predator avoidance may be the principal driver of diurnal acoustic patterns on the reef. A limitation of the remote camera imaging technique is that organisms must be large enough (on the order of 20 mm total length) and on the surface of the substrate to be adequately resolved and classified in photographs. The diversity of benthic crustacean species is extremely high. For example, one recent study recovered 525 species from only 6.3 m2 of distributed reef area (Plaisance et al., 2011). The majority of these species are small (total length of less than 5 mm) and found within, rather than on top of, coral structures. In fact, most coral reef diversity is made up of small, cryptic species and species from poorly known groups (Plaisance et al., 2009). Ecological work using settlement structures (artificial tiles that are placed on a reef, then removed some time later and evaluated for resident organism diversity, biomass, and number) has shown that small crustaceans can occupy protected surfaces with densities of up to 153.1 (S.E. ¼ 14.1, n ¼ 6) individuals per 50 cm2 area (Smith et al., 2001). Given that size of these organisms and J. Acoust. Soc. Am., Vol. 135, No. 4, April 2014

that the majority of habitable area in a coral reef is below the visible surface it is likely that the quantity and diversity of undetected small organisms is very high. For the origins of all reef noise components to be truly resolved, the contribution to ambient noise by these smaller organisms must also be determined. If such a contribution exists, even if small individually, the sheer number of these organisms moving about on the substrate is likely to create a discernable contribution to the sound field. C. Aquarium recordings

The sources of sound recorded from inside the test aquarium were either noise from sources outside the tank, such as the air conditioning system, or sound produced by Clibanarius diugeti inside the tank. Any sound propagating inside the tank was subject to frequency dependent amplitude modification through tank resonance. Noise from outside the tank was clearly discernable as constant, tonal noise largely restricted to frequencies below 6 kHz, and considering the biological sounds of interest consisted of higher frequencies, was filtered from the recordings. The first three resonant modes aligned with the three principal dimensions of the tank are given in Table I. Figure 6(B) shows the averaged pressure spectral density from 211 s of hermit crab sounds. The peak at 6.3 kHz may correspond to modes (1, 3, 1); (3, 2, 1); and (2, 1, 2), where each number corresponds to the nth mode in the x, y, and z directions, respectively. No higher resonant modes were evident between 6 and 11 kHz. The lack of modal excitation at higher resonant frequencies is likely due to two factors. The live rock and the sand layer covering the base of the tank act as attenuators, although the degree of attenuation they provide in this band and above is unquantified. Second, modes above 7 kHz or so are of higher order and are correspondingly weaker in amplitude. Lower order modes may have been more obvious in the spectra below 6 kHz, but distinguishing these modes from background noise would require the identification of node locations through a spatial-acoustic survey of the tank. The broad-band nature of the peak between 11 and 17 kHz, combined with the 20 dB increase in level above the noise floor between 6 and 10 kHz, strongly suggest that the spectral characteristics observed here are primarily representative of hermit crab spectra, rather than an artifact of the experimental setup. The intermittent nature of sounds produced by Clibanarius diugeti and the visual connection between the sounds and movement/foraging activity suggests that such sounds are purely incidental and not part of an effort to communicate between individuals, or part of a cyclical process such as respiration. As the sounds originate from mechanical interaction between the hermit crab’s exoskeleton, shell, and the substrate, they are not likely to be species specific. It is speculated that other hard-shelled, reef dwelling animals are capable of similar sound production as they feed and move about the reef. The distribution of intervals between sounds that achieve a received level threshold sufficient for detection does not appear to be comprised of a single random process. Instead, two regimes in the distribution were observed, Freeman et al.: Origins of coral reef sound

1785


with the shift between the “fast” and “slow” regimes occurring at an interval of 4.40 s. Although these distributions could not be fitted precisely to an exponential distribution, it is postulated that the sound creation mechanisms associated with the hermit crabs fall into two categories. Behaviors pertaining to both categories of sound production were observed during the recording period. The first involves transient sounds that occur in clusters, such as those made as the exoskeleton and/or shell makes repeated contact with the rock during steady locomotion or when repeated scraping is required to remove food items that may be anchored to the substrate. In these acoustic events, trains of similar sounds are created over a short time before the organism changes its behavior and the acoustic output similarly changes. These events correspond to intervals of less than 4.40 s as seen in Fig. 7. The second involves single acoustic events produced when the substrate is struck incidentally. Such sounds may be associated with less coordinated movement that is less likely to be associated with continuous locomotion or repetitive feeding behavior. These correspond to intervals greater than 4.40 s in Fig. 7. Although beyond the scope of this work, the sound produced by the interaction between the hermit crabs and the substrate almost certainly contains information pertaining to substrate type. It is also likely that the size and numbers of organisms additionally influence the characteristics of the acoustic emissions. The seismoacoustic properties of the substrate change with density and elastic moduli, while larger organisms possess correspondingly larger and heavier exoskeletons and/or shells. Through incoherent addition a greater number of organisms will produce a higher noise level overall. It is possible that the dominant frequency components of this incidental sound are affected by substrate density, elastic modulus, and attenuation characteristics, along with the mean organism size. Consequently such sounds, discernable even in the presence of snapping shrimp noise, may contain information regarding not just the size and number of organisms, but of the acoustic properties of the substrate. D. Implications for acoustic ecological evaluation

The three types of observations reported here show a link between the production of sound in the 11 to 17 kHz band on coral reefs, sounds made by the hermit crab Clibanarius diugeti and the benthic activity recorded in the time-lapse photographs. Laboratory tests demonstrate that the spectral characteristics of sounds created through mechanical interaction between hard-shelled invertebrates and coral reef substrate are similar to those of ambient sounds recorded on coral reefs in the Line Islands. The increase in acoustic pressure level in the 11–17 kHz band at night is correlated with the general increase in crustacean activity at night in the same locations. These observations suggest that the contribution of benthic invertebrate activity to the ambient sound field is discernable at some frequencies despite the dominance of snapping shrimp noise typically found on coral reefs. The similarity of acoustic spectra recorded over the five Line Islands shows that biological processes responsible for 1786


the recorded acoustic emissions are ubiquitous across a wide range of ecological states; from the relatively untouched Kingman Reef to the heavily degraded coral reef ecosystems adjacent to the mouth of St. Stanislas Bay, Christmas Island. Such similarity suggests that the processes underlying the recorded acoustic emissions are themselves ubiquitous and may occur on a global scale in shallow coral reef ecosystems, although the sounds discussed here may not always be discernable over snapping shrimp noise if the disparity in pressure level is too great and/or the rate of occurrence is insufficient. The sounds produced by invertebrates discussed in this article may also be reduced in level by additional environmental variables. In regions where coral reefs have degraded to an algae-dominated community, algal smothering of the substrate may reduce direct contact between hard-shelled benthic organisms and the underlying rock. The physical barrier imposed by the comparably soft algae may reduce the acoustic output of the activity discussed here, other impacts on their ecological state notwithstanding. Before estimates of relative ecological state can be made with any accuracy through passive listening, further work is necessary to understand the sources of sound recorded on coral reefs. Concurrent acoustic sampling at the same locations as non-acoustic ecological surveys can enable validation of sound sources with other ecological data sets. The spatial heterogeneity of coastal underwater sound fields has been observed before (Radford et al., 2010), but the mechanisms behind this acoustic diversity are still not fully understood. The effects of acoustic propagation and dispersion in shallow reef environments need to be considered not just for local sources but also for sound that may arrive at reefs from more distant regions. It should also be noted that some reef-dwelling species are known to use sound as a means of communication during spawning periods. These acoustic events, while containing information regarding specific species, are less likely to be reflective of overall local ecological condition as the evolutionary benefits to congregational mass spawning events are likely independent of local ecological state. Consequently, they must be accounted for when attempting to obtain a representative recording of ambient noise. With the decreasing cost and increasing capacity of battery and memory storage technology, remote passive acoustic sensing techniques that offer continuous monitoring over year-scale periods of time for relatively little cost are now commonplace (Mann and Tucker, 2000; Wiggins and Hildebrand, 2007; Lammers et al., 2008). If connections between ambient noise characteristics and ecosystem state can be established, remote, continuous passive acoustic monitoring of such ambient noise could serve as a powerful and complimentary ecological survey technique to SCUBAbased visual observations, which are typically poor in estimating the abundance and diversity of cryptobenthic organisms. Furthermore, the interconnected nature of coral reef ecosystems suggests that changes to one ecological aspect of an area will influence all trophic groups. Such cascading changes have been observed in many underwater ecosystems (Jackson et al., 2001; Sandin et al. 2008; Schiel, 2011). The characteristics of sounds produced by hard-shelled benthic Freeman et al.: Origins of coral reef sound


invertebrates may serve as an indicator by proxy of other aspects of ecological state. Changes in sounds produced by hard-shelled benthic invertebrates may serve a similar purpose as the song of a canary in a coalmine, indicating important ecological changes that may otherwise be difficult to detect and quantify. Passive acoustic recording facilitated a substantial improvement in the accuracy of marine mammal monitoring (Winn et al., 1975; Cummings and Holliday, 1985). Through identification and validation of ambient biological noise sources to an ecologically functional level, passive acoustic recording may beget similar advances in marine ecological survey techniques. V. SUMMARY

Acoustic data collected from five islands and reefs in the Line Islands archipelago follow a clear diurnal pattern. Pressure spectral density level increases by 4.6 to 6.2 dB in the 11 to 17 kHz band with the setting of the sun, and the reverse occurs at sunrise. The spectra at all sites are dominated by a peak between 14.3 and 14.63 kHz that is 4 to 8 dB higher than the simultaneously observed spectral plateau between 3 and 11 kHz. This peak persists at a lower level during the day. Although the majority of reef sounds are thought to originate from the cavitation-derived emissions of snapping shrimp, sounds recorded in the Line Islands do not appear consistent with the “flat” spectrum created by the implosion of cavitation bubbles created by snapping shrimp. Incoherent addition of species-specific spectra creates a sound field modified from that produced solely by dominant snapping shrimp noise. Simultaneous time-lapse underwater photographs of the benthic environment reveal an increase in invertebrate activity at night that is consistent with the increase in pressure spectral density level. Crustaceans appear to be the dominant organism group represented in the photographs. Acoustic recordings under controlled aquarium conditions of the hermit crab Clibanarius diugeti show a similar spectral peak to the coral reef in situ recordings. The sounds were observed to originate from mechanical interactions between the hard-shelled organisms and the coral rock substrate. This observation suggests that the component of reef noise that peaks between 14 and 15 kHz in the Line Islands archipelago is created by the physical interaction between a great number of small, hard-shelled organisms and the substrate. Conclusive identification of sound producers to the functional-group level and consideration of the environmental effects on recorded sound may lead to passive acoustics-based, long-term monitoring and evaluation methods that offer quantitative, continuous information regarding ecological state. ACKNOWLEDGMENTS

The authors would like to thank Lauren Freeman and Greg Rouse of Scripps Institution of Oceanography, Laura Coleman of San Diego State University, William Coles of University of California, San Diego, Marc Lammers of Oceanwide Science Institute, and Nancy Knowlton of Smithsonian Institution for valuable advice and assistance. The authors would also like to acknowledge the owner, J. Acoust. Soc. Am., Vol. 135, No. 4, April 2014

skipper, and crew of the MY Hanse Explorer. The National Science Foundation Interdisciplinary Graduate Education and Research Traineeship (IGERT) program and the Office of Naval Research sponsored this work. Akamatsu, T., Okumura, T., Novarini, N., and Yan, H. Y. (2002). “Empirical refinements applicable to the recording of fish sounds in small tanks,” J. Acoust. Soc. Am. 112(6), 3073–3082. Au, W. W. L., and Banks, K. (1998). “The acoustics of the snapping shrimp Synalpheus parneomeris in Kaneohe Bay,” J. Acoust. Soc. Am. 103, 41–47. Bardyshev, V. I. (2007). “Underwater ambient noise in shallow-water areas of the Indian Ocean within the tropical zone,” Acoust. Phys. 53, 167–171. Bardyshev, V. I. (2010). “Underwater sounds of the Pacific Ocean in the shelf zone of Shikotan Island,” Acoust. Phys. 56, 49–58. Bjorno, L. (1987). “Sources of underwater ultrasound,” in Ultrasonics International 87 Conference Proceedings (Butterworth, Guildford, U.K.), pp. 244–251. Budelmann, B. U. (1992). “Hearing in crustacean,” in The Evolutionary Biology of Hearing (Springer, New York), pp. 131–139. Cato, D. H. (1980). “Some unusual sounds of apparently biological origin responsible for sustained background noise in the Timor Sea,” J. Acoust. Soc. Am. 68, 1056–1060. Cato, D. H. (1993). “The biological contribution to ambient noise in waters near Australia,” Acoust. Aust. 20, 76–80. Cato, D. H., and Bell, M. J. (1992). “Ultrasonic ambient noise in Australian shallow waters at frequencies up to 200 kHZ,” Material Research Laboratory technical report No. MRL-TR91-23, Ascot Vale, Victoria. Cato, D. H., and McCauley, R. D. (2002). “Australian research in ambient sea noise,” Acoust. Aust. 30(1), 13–20. Chitre, M., Kuselan, S., and Pallayil, V. (2012). “Ambient noise imaging in warm shallow waters; robust statistical algorithms and range estimation,” J. Acoust. Soc. Am. 132(2), 838–847. Coutis, P. F., and Middleton, J. H. (1999). “Flow-topography interaction in the vicinity of an isolated, deep ocean island,” Deep-Sea Res. Part I: Ocean. Res. Papers 46, 1633–1652. Cummings, W. C., and Holliday, D. V. (1985). “Passive acoustic location of bowhead whales in a population census off Point Barrow, Alaska,” J. Acoust. Soc. Am. 78(4), 1163–1169. DeMartini, E. E., Friedlander, A. M., Sandin, S. A., and Sala, E. (2008). “Differences in fish assemblage structure between fished and unfished atolls in the northern Line Islands, central Pacific,” Mar. Ecol.: Prog. Ser. 365, 199–215. D’Spain, G., and Batchelor, H. (2006). “Observations of biological choruses in the Southern California Bight: A chorus at midfrequencies,” J. Acoust. Soc. Am. 120(4), 1942–1955. Everest, F. A., Young, R. W., and Johnson, M. W. (1948). “Acoustical characteristics of noise produced by snapping shrimp,” J. Acoust. Soc. Am. 20, 137–142. Ferguson, B. G., and Cleary, J. L. (2000). “In situ source level and source position estimates of biological transient signals produced by snapping shrimp in an underwater environment,” J. Acoust. Soc. Am. 109(6), 3031–3037. Hiatt, R. H., and Strasburg, D. W. (1960). “Ecological relationships of the fish fauna on coral reefs of the Marshall Islands,” Ecol. Mon. 30, 65–127. Hobson, E. S. (1972). “Activity of Hawaiian reef fishes during evening and morning transitions between daylight and darkness,” Bull. Natl. Ocean. Atmos. Admin. 70(3), 715–740. International Transducer Corporation, Ltd. (1996). Specifications of ITC 6080-C hydrophone, Ser. No.1545 (International Transducer Corporation, Santa Barbara), pp. 1–10. Jackson, J. C. B., Kirby, M. X., Berger, W. H., Bjorndal, K. A., Botsford, L. W., Bourque, B. J., Bradbury, R. H., Cooke, R., Erlandson, J., Estes, J. A., Hughes, T. P., Kidwell, S., Lange, C. B., Lenihan, H. S., Pandolfi, J. M., Peterson, C. H., Steneck, R. S., Tegner, M. J., and Warner, R. R. (2001). “Historical overfishing and the recent collapse of coastal ecosystems,” Science 293, 629–638. Johnston, M. W., Everest, F. A., and Young, R. W. (1947). “The role of snapping shrimp (Crangon and Synalpheus) in the production of underwater noise in the sea,” Biol. Bull. 93(2), 122–138. Kim, B.-N., Hahn, J., Choi, B. K., and Kim, B.-C. (2010). “Snapping shrimp sound measured under laboratory conditions,” Jpn. J. Appl. Phys. 49(7), 1–4. Kim, B. N., Hahn, J., Choi, B. K., Kim, B. C., Park, Y., Jung, S. K., and Lee, Y. K. (2011). “Acoustic characteristics of the snapping shrimp sound observed in the Coastal Sea of Korea,” Jpn. J. Appl. Phys. 50(7), 1–6. Freeman et al.: Origins of coral reef sound

1787


Knowlton, R. E., and Moulton, J. M. (1963). “Sound production in snapping shrimps Alpheus (Crangon) and Synalpheus,” Bio. Bull. 125, 311–331. Knudsen, V. O., Alford, R. S., and Emling, J. W. (1948). “Underwater ambient noise,” J. Mar. Res. 7, 410–429. Lammers, M. O., Brainard, R. E., Au, W. W. L., Mooney, T. A., and Wong, K. B. (2008). “An ecological acoustic recorder (EAR) for long-term monitoring of biological and anthropogenic sounds on coral reefs and other marine habitats,” J. Acoust. Soc. Am. 123(3), 1720–1728. Loggerhead Instruments, Inc. (2011). DSG Manual: Digital Spectrogram Long-Term Acoustic Recorder, V. 1.7, pp. 1–37. Ma, B. B., Nystuen, J. A., and Lien, Ren-Chieh (2005). “Prediction of underwater sound levels from rain and wind,” J. Acoust. Soc. Am. 117(6), 3555–3565. Mann, D. A., and Tucker, T. J. (2000). “Programmable DSP datalogger for animal bioacoustics monitoring (A),” J. Acoust. Soc. Am. 108, 2584. McCauley, R. D., and Cato, D. H. (2000). “Patterns of fish calling in a near shore environment in the Great Barrier Reef,” Philos. Trans. Biol. Sci. 355(1401), 1289–1293. National Research Council (2003). Ocean Noise and Marine Mammals (National Academy Press, Washington, DC), p. 196. Okumura, T., Akamatsu, T., and Yan, H. Y. (2002). “Analyses of small tank acoustics: Empirical and theoretical approaches,” Bioacoustics 12, 330–332. Payne, A., Cotter, J., and Potter, T. (2008). Advances in Fisheries Science: 50 Years on From Beverton and Holt (Wiley-Blackwell, Oxford), pp. 451–471. Plaisance, L., Caley, M. J., Brainard, R. E., and Knowlton, N. (2011). “The diversity of coral reefs: What are we missing?,” PLoS ONE 6(10), e25026. Plaisance, L., Knowlton, N., Paulay, G., and Meyer, C. (2009). “Reef-associated crustacean fauna: Biodiversity estimates using semi-quantitative sampling and DNA barcoding,” Coral Reefs. 28(4), 977–986. Radford, C. A., Jeffs, A. G., Tindle, C. T., and Montgomery, J. C. (2008a). “Temporal patterns in ambient noise of biological origin from a shallow water temperate reef,” Oecologia 156, 921–929. Radford, C., Jeffs, A., Tindle, C., and Montgomery, J. C. (2008b). “Resonating sea urchin skeletons create coastal choruses,” Mar. Ecol. Prog. Ser. 362, 37–43.

1788


Radford, C. A., Stanley, J. A., Tindle, C. T., Montgomery, J. C., and Jeffs, A. G. (2010). “Localised coastal habitats have distinct underwater sound signatures,” Mar. Ecol. Prog. Ser. 401, 21–29. Readhead, M. L. (1997). “Snapping shrimp noise near Gladstone, Queensland,” J. Acoust. Soc. Am. 101(3), 1718–1722. Sandin, S. A., Smith, J. E., DeMartini, E. E., Dinsdale, E. A., Donner, S. D., Friedlander, A. M., Konotchick, T., Malay, M., Maragos, J. E., Obura, D., Pantos, O., Paulay, G., Richie, M., Rohwer, F., Schroeder, R. E., Walsh, S., Jackson, J. B. C., Knowlton, N., and Sala, E. (2008). “Baselines and degradation of coral reefs in the Northern Line Islands,” PLOS One 3(2), 1–11. Schiel, D. R. (2011). “Biogeographic patterns and long-term changes on New Zealand coastal reefs: Non-trophic cascades from diffuse and local impacts,” J. Exp. Mar. Biol. Ecol. 400(1–2), 33–51. Schmitz, B. (2002). “Sound production in Crustacea with special reference to the Alpheidae,” in The Crustacean Nervous System, edited by K. Wiese (Springer-Verlag, New York), pp. 536–547. Scrimger, J. A., Evans, D. J., and Yee, W. (1989). “Underwater noise due to rain—Open ocean measurements,” J. Acoust. Soc. Am. 85(2), 726–731. Smith, J. E., Smith, C. M., and Hunter, C. L. (2001). “An experimental analysis of the effects of herbivory and nutrient enrichment on benthic community dynamics on a Hawaiian reef,” Coral Reefs. 19, 332–342. Takemura, A., and Mizue, K. (1968). “Studies of underwater sound. I—On the underwater sound of genus Alpheus fabricius in the coastal waters of Japan,” Bull. Faculty Fish., Nagasaki Univ. 26, 37–48. Versluis, M., Schmitz, B., von der Heydt, A., and Lohse, D. (2000). “How snapping shrimp snap: Through cavitating bubbles,” Science 289, 2114–2117. Wenz, G. M. (1962). “Acoustic ambient noise in ocean spectra and sources,” J. Acoust. Soc. Am. 34(12), 1936–1956. Wiggins, S. M., and Hildebrand, J. A. (2007). “High-frequency acoustic recording package (HARP) for broad-band, long-term marine mammal monitoring,” in Symposium on Underwater Technology and Workshop on Scientific Use of Submarine Cables and Related Technologies (IEEE, Piscataway, NJ), pp. 551–557. Winn, H. E., Edel, R. K., and Taruski, A. G. (1975). “Population estimate of the Humpback Whale (Megaptera novaeangliae) in the West Indies by visual and acoustic techniques,” J. Fish. Res. Board Can. 32, 499–506. Zakarauskas, P. (1986). “Ambient noise in shallow water: a literature review,” Can. Acoust. 14, 3–17.

Freeman et al.: Origins of coral reef sound


Coral mucus fuels the sponge loop in warm- and cold-water coral reef ecosystems.

A comprehensive investigation of mesophotic coral ecosystems in the Hawaiian Archipelago.

Length-based assessment of coral reef fish populations in the main and northwestern Hawaiian islands.

Coral reef disturbance and recovery dynamics differ across gradients of localized stressors in the Mariana Islands.

Coral reef disturbance and recovery dynamics differ across gradients of localized stressors in the Mariana Islands.

Coral Reef Health Indices versus the Biological, Ecological and Functional Diversity of Fish and Coral Assemblages in the Caribbean Sea.

Extent of mangrove nursery habitats determines the geographic distribution of a coral reef fish in a South-Pacific archipelago.

Management Strategy Evaluation Applied to Coral Reef Ecosystems in Support of Ecosystem-Based Management.

Sensing coral reef connectivity pathways from space.

Medical hazards of the coral reef.

Coral reef soundscapes may not be detectable far from the reef.

From multiple pathogenicity islands to a unique organized pathogenicity archipelago.

Mercury (Hg) speciation in coral reef systems of remote Oceania: Implications for the artisanal fisheries of Tutuila, Samoa Islands.

Refining reef-coral refugia.

Coral reef community composition in the context of disturbance history on the Great Barrier Reef, Australia.

Diet and diversification in the evolution of coral reef fishes.

Oceanographic and behavioural assumptions in models of the fate of coral and coral reef fish larvae.

A population genetic assessment of coral recovery on highly disturbed reefs of the Keppel Island archipelago in the southern Great Barrier Reef.

Towards a new paleotemperature proxy from reef coral occurrences.

Coral Reef Resilience, Tipping Points and the Strength of Herbivory.

Ecological solutions to reef degradation: optimizing coral reef restoration in the Caribbean and Western Atlantic.

Recruitment Variability of Coral Reef Sessile Communities of the Far North Great Barrier Reef.

The contribution of microbial biotechnology to mitigating coral reef degradation.

The DNA of coral reef biodiversity: predicting and protecting genetic diversity of reef assemblages.