Reynolds.

100-123 6/10/05 4:54 PM Page 100

Reynolds.100-123 6/10/05 4:54 PM Page 101

JOHN HILDEBRAND

Impacts of
Anthropogenic Sound

T
HERE IS GROWING CONCERN THAT sound introduced into the sea
by human activities has detrimental effects on marine mammals. For example,
mounting evidence suggests that high-intensity anthropogenic sound from
sonar and airguns leads to strandings and subsequent mortality of beaked whales. Al-
though the mechanisms of injury in these events are unclear, the species affected and
the implicated sound levels follow a consistent pattern. A more pervasive, yet more
subtle, problem may be the effects of increases in background noise levels from com-
mercial shipping. Higher levels of background noise may interfere with marine mam-
mals ability to detect sounds, whether calls from their own species, echoes from prey,
or natural sounds that aid in navigation or foraging. Noise may affect developmen-
tal, reproductive, or immune functions and cause more generalized stress. The effects
of other pollutants (e.g., chemicals) may be additive or synergistic with those of
noise. As Read (this volume) and Plagnyi and Butterworth (this volume) suggest,
human activities may have both direct and indirect consequences. For instance, noise
may have ecosystem-scale effects, including impacts on[AQ1] species that are marine
mammal prey.
Sources of anthropogenic sound are becoming both more pervasive and more
powerful, increasing both oceanic background noise levels and peak sound intensity
levels. Anthropogenic activities in the ocean have increased over the past 50 years, re-
sulting in more low-frequency (<1,000 Hz) and mid-frequency (120 kHz) noise.
Sources of anthropogenic noise include commercial shipping, defense-related activ-
ities, hydrocarbon exploration and development, research, and recreation.

101
Reynolds.100-123 6/10/05 4:54 PM Page 102

102

Anthropogenic sound is created in the ocean both pur- flict between human and marine mammal use of the sea is
posefully and unintentionally. The result is noise pollution fundamentally a consequence of the fact that we do not in-
that is high intensity and acute, as well as lower level and habit the same sound environment. Marine mammals live
chronic. Many sources of noise are located along well- with their ears in the water, and we live, even at sea, mainly
traveled paths in the sea, particularly in coastal and conti- with our ears in the air.
nental shelf waters, areas that often include important A notable exception is the military use of submarines,
marine mammal habitats. where stealth is required, so minimization of sound pro-
There is sufficient evidence to conclude that some high- duction then becomes crucial to human survival. Reduction
intensity sounds are harmful and, on occasion, fatal to ma- in radiated sound has been achieved by placing rotating ma-
rine mammals. Given the opportunity, the animals may chinery on isolating mounts and by designing efficient pro-
avoid high-intensity sound, but in some extreme cases there pellers that thrust without unnecessary vibration and cavi-
has been documentation of injury from anthropogenic tation. Thus, when it has been important to keep the sea
sound exposure. Multiple mass strandings of beaked whales quiet, the necessary technology has been developed and
following high-intensity sound exposure demonstrate a made available.
repeating pattern of events. Following exposure to high-
intensity sonar or airguns, beaked whales have been known
to strand on the shore, and if human intervention does not NOISE LEVELS IN THE OCEAN
return them to the sea they die. Understanding the causes
and consequences of beaked whale mass stranding should Sound is a vibration or acoustic wave that travels through
be a high research priority. What, then, are the mechanisms some medium, such as air or water. Acoustic waves can be
for damage or disturbance? described either by the speed and direction at which a small
A major impediment to assessing the biological effects of piece of the medium vibrates, called the particle velocity, or
ocean noise is the lack of knowledge concerning marine by the corresponding pressure associated with the vibra-
mammal responses to sound. Behavioral data from the wild tion. Frequency is the rate of vibration, given in hertz (Hz)
are needed to examine those responses so that effects can be or cycles per second; we perceive frequency as the pitch of
assessed. Significant effects may prove to be confined to a the sound. A tone is a sound of a constant frequency that
few individuals exposed at high sound pressure levels or continues for a substantial time. A pulse is a sound of short
they may be occurring at a population level as a result of duration and may include a broad range of frequencies.
widespread exposure. Discerning population-level effects is In water, the pressure of sound waves is typically mea-
challenging as observations must be conducted over long sured with a device called a hydrophone. When discussing
distances and extended time periods. background noise, the implicit assumption is that sound
Sound is an extremely efficient way of propagating en- pressure fluctuations are being described, although it is not
ergy through the ocean, and marine mammals have evolved clear whether a particular marine organism is affected by
to exploit its potential. Many marine mammals use sound particle velocity or pressure. Sound pressure is measured
as a primary means for underwater communication and in pascals (Pa) in the international system of units (SI), al-
sensing. Toothed whales have developed sophisticated though it is expressed in bars by the geophysical community
echolocation systems to sense and track prey and engage in (1 Pa = 105 bar). Because mammalian hearing and sound
complex exchanges of vocalizations with members of their production cover a wide range of pressure values, the sound
own species. Baleen whales have developed long-range pressure level (SPL) is usually measured on a logarithmic
acoustic communication systems to facilitate mating and scale called the decibel (dB), and compared against a 1 Pa
social interaction. Some baleen whales produce intricately reference (Po) for underwater sound as follows:
patterned songs that continue for hours or days. Marine
SPL dB re 1 Pa = 10log10 (P/Po )2 = 20log10 (P/Po )
mammals may use sound from natural sources as a guide
for navigation, prey detection, and avoidance of predation. Pressure amplitude can be measured either as a root-
The sound environment of the ocean is an important aspect mean-squared (RMS) or peak value. (Note that in this chap-
of marine mammal habitat and we can expect marine mam- ter I use RMS values unless otherwise noted). Pressure is
mals to choose their locations and modify their behavior squared in the above expression as a proxy for acoustic in-
based, in part, on natural and anthropogenic sounds. tensity, that is, the power flow per unit area in the sound
Human presence at sea is normally on the surface, and wave, with units of watts/m2. Sound intensity is the prod-
the sounds that we produce within the water are rarely uct of pressure (P) and particle velocity (v). Acousticians
given much consideration. The air-sea interface constitutes working in one medium (water or air) use the fact that for
a substantial sound barrier. Sound waves in the water are re- simple plane waves the pressure and particle velocity are re-
duced in intensity by a factor of more than a thousand when lated by the characteristic impedance (Z) of the medium:
they cross the air-sea boundary, which means that we are ef-
Z = P/v
fectively insulated from the noise produced by the rotating
propellers that drive our ships or by the high-intensity sonar This allows the acoustic intensity (I) to be related to the
used to measure depths or probe the sea bottom. The con- pressure squared divided by the impedance:
Reynolds.100-123 6/10/05 4:54 PM Page 103

Impacts of Anthropogenic Sound 103

I = 10log10 [P2/(Z Io )] waves breaking directly above the listener. Although ambi-
ent noise is always present, the individual sources that con-
Acoustic power is obtained by integrating intensity over tribute to it do not necessarily create sound continuously.
some area, and acoustic energy is the power integrated over Acoustic pressures are analyzed into their frequency
some time period. The same acoustic energy can be ob- components or spectrum using a Fourier transform (Brace-
tained from a high-intensity source lasting a short time (im- well 2000). One way to express the result is as a power spec-
pulse) or a low-intensity source lasting a long time (contin- tral density with units of Pa2/Hz. Note that the bandwidth
uous wave). of the power spectral density is explicitly part of the unit,
When sound propagates from water into air, there is a and by convention noise measurements are presented in 1-
30-dB (1,000) decrease in acoustic intensity because the Hz-wide frequency bins. Hearing and other auditory meas-
characteristic impedance of water (1,500,000 kg/s-m2) is ures are often presented in one-octave or one-third-octave
much greater than that of air (415 kg/s-m2). This means frequency bins as an approximation for the filtering charac-
that sounds made by a high-intensity underwater source teristics involved in hearing. Source measurements are typ-
(such as a sonar) are not transmitted into the air with the ically given for varying bandwidths with the following equa-
same intensity. In essence, sailors and seafaring passengers tion allowing for conversion to 1-Hz frequency bins:
are protected from the sounds produced in the sea. Without
dB = 10log10 (bandwidth in Hz)
the air-sea boundary for protection, there would be a strong
incentive to protect human hearing from the noise of sonars The sound level received from a source depends on the
and cavitating ships propellers. (Note that for sound in air distance between the source and receiver, as well as on the
a different reference level is used, Po = 20 Pa, and this propagation characteristics of the environment. Therefore,
may be a source of confusion when comparing sound un- the distance at which a source measurement was made must
der water and in air.) be specified, and the convention is to normalize the pressure
Underwater sounds are classified according to whether to an approximation of what would be received at a range
they are transient or continuous. Transient sounds are of of 1 m from the source (dB re Pa at 1 m). When arrays of
short duration and occur singly, irregularly, or as part of a sources are used, this convention overestimates actual
repeating pattern. For instance, an explosion represents a source levels in the near field (e.g., at a 1-m range) but pro-
single transient, whereas the periodic pulses from a ships vides a good way to predict source levels in the far field (e.g.,
sonar are patterned transients. Broadband short-duration at a 1-km range).
transients are called pulses and sound like clicks or bangs.
Continuous sounds, which occur without pauses, are fur-
ther classified as periodic, such as the sound from rotating NATURAL OCEAN
machinery or pumps, or aperiodic, such as the sound of a ACOUSTIC ENVIRONMENT
ship breaking ice.
Pulsed sounds often are measured in terms of their peak- The ambient background noise of the ocean is highly vari-
to-peak pressure, whereas continuous sounds are measured able. At a given time and place, a broad range of sources
in terms of their RMS pressure. The method of converting may be combined. In addition, conditions at a particular lo-
between RMS and peak-to-peak pressures is well defined for cation may affect how well ambient sounds are received
continuous-wave signals (add 9 dB to the RMS pressure to (e.g., sound propagation, water depth, and bathymetry).
get the peak-to-peak pressure). However, for pulsed sounds, Natural phenomena known to contribute to oceanic ambi-
the conversion is problematic because the duration of the ent noise include: (a) wind, waves, and swell patterns, (b)
signal included dramatically alters the result. For a brief bubbles, (c) currents and turbulence, (d) earthquakes, (e)
pulse, peak-to-peak pressure is measured from the highest precipitation, (f ) ice cover and activity, and (g) marine life.
and lowest portions of the waveform, whereas RMS pressure
is difficult to interpret because it depends on the duration
Wind, Waves, and Ice
over which the signal is measured. An alternative for pulsed
signals is to estimate the total energy, rather than the peak- Ocean surface motions that are due to wind, sea state, and
to-peak pressure or intensity. Energy is proportional to the swell patterns are the dominant physical mechanisms for
time integral of the squared pressure, described in the units natural background noise in the ocean. Noise is primarily
Pa2-s. For brief pulses, energy in dB re 1 Pa2-s is associated with wind acting on the surface, causing wave ac-
less than peak-squared pressure values in the same units. As tivity. In the absence of anthropogenic and biological
others have warned, better standardization of measurement sound, ambient noise is wind dependent over an extremely
methods for pulsed underwater sounds is urgently needed to broad frequency band from below 1 Hz to at least 100 kHz.
permit meaningful comparisons (Green and Moore 1995). At frequencies below 10 Hz, interactions of surface waves
Ambient noise in the ocean is the background sound that are the dominant mechanisms for sound generation. Across
incorporates the broad range of individual sources, some the remainder of the band from 10 Hz to100 kHz, oscillat-
identified and others not. Ocean noise may come both from ing bubbles in the water column are the primary noise
distant sources, such as ships, or from nearby, such as the source, both as individual bubbles and as bubble clouds.
Reynolds.100-123 6/10/05 4:54 PM Page 104

104

In early descriptions, ocean noise was related to sea state Explosions

(Knudsen et al. 1948). By this theory, noise levels increase
with increasing sea state to the same degree across the en- Two classes of man-made explosions create high sound lev-
tire frequency band from 1 to 100 kHz. More recent work els in the ocean: nuclear and chemical. Nuclear devices have
suggests that noise is better correlated with wind speed than been tested underwater in the ocean, in the atmosphere
with sea state or wave height. The correlation between above the ocean, and on oceanic islands. In 1963 all nuclear
noise and wind speed allows for more accurate prediction, states signed the Limited Test Ban Treaty, pledging to stop
as sea states are more difficult to estimate than wind speeds. testing nuclear weapons underwater. The Comprehensive
In the open ocean, the noise of breaking waves is correlated Test Ban Treaty was adopted in 1996, whereby the major
with wind speed. Spilling and plunging breakers raise un- nuclear powers pledged to discontinue all nuclear testing.
derwater sound levels by more than 20 dB across the band The most recent oceanic tests were conducted by France in
from 10 Hz to 10 kHz (Wilson et al. 1985). Precipitation is 19951996 on the islands of Fangataufa and Mururoa in the
another factor that can increase ambient noise levels by up South Pacific. There is currently a low probability of con-
to 35 dB across a broad band of frequencies from 100 Hz to tinued ocean testing of nuclear devices, although the situa-
more than 20 kHz (Nystuen and Farmer 1987). tion could change with geopolitical developments over the
Ice cover alters the ocean noise field depending on its type coming years or decades.
and degreefor instance, whether it is shore-fast pack ice, Nuclear explosions are extremely strong sources of under-
moving pack ice, or at the marginal ice zone (Milne 1967). water sound. Their source levels are expressed as an equiv-
Shore-fast pack ice isolates the water column from the ef- alent weight of chemical explosives with fission devices
fects of wind and results in a decrease in ambient noise of yielding the equivalent of tens to hundreds of kilotons and
1020 dB. Sounds from ice cracking, however, may increase fusion devices yielding the equivalent of tens of megatons.
noise levels by as much as 30 dB. Ice cracking can generate Past tests likely had significant impacts on marine mammals
broadband pulses up to 1 kHz lasting for 1 s or longer. In- in the vicinity of the test sites, although no marine mammal
teraction of ocean waves with the marginal ice zone may monitoring or stranding data are available. To ensure com-
raise noise levels by 412 dB (Diachok and Winikur 1974). pliance with the Comprehensive Test Ban Treaty, an inter-
national monitoring system is being implemented, includ-
ing a series of marine hydrophones and terrestrial (island)
Earthquakes and Thunder/Lightning
seismic sensors to detect high-intensity sounds. This infor-
Earthquakes and thunder/lightning are transient natural mation is transmitted, in real time, to the International Data
sound sources. Underwater recordings of thunder/lightning Centre, where analysts evaluate the data for indications of
from storms 510 km distant show peak energy between 50 nuclear explosions. The physical character of the oceans al-
and 250 Hz, up to 15 dB above background levels (Dubrovsky lows the sounds of such explosions to travel for extremely
and Kosterin 1993). Seismic energy from undersea earth- long distances with little energy loss, and monitoring is con-
quakes couples into the ocean and is called T-phase (terti- ducted over a large fraction of the worlds oceans with a
ary) in addition to the usual P-phase (primary) and S-phase small number of stations. The network designed for ocean
(secondary) seismic waves that are observed on land. At monitoring currently includes eleven stations, located pri-
ranges of less than 100 km, T-phase energy can have fre- marily in the Southern Hemisphere.
quencies greater than 100 Hz, with peak energy at 520 Hz. Chemical explosions are more portable and more easily
It can be as much as 3040 dB above background noise, with conducted in an ocean setting and have been used for
a sharp onset, and can last from a few seconds to several oceanic research, for construction, and for military testing.
minutes (Schreiner et al. 1995). A surprisingly large number (3004,000 per month) of un-
derwater explosions were reported in the North Pacific
during the 1960s (Spiess et al. 1968). At one time chemical
ANTHROPOGENIC SOUND explosions were commonly used for marine seismic explo-
ration, but they have been replaced by airgun arrays, which
Human activity in the marine environment is now an im- provide a more reliable source signature. Chemical explo-
portant component of oceanic background noise. Sound sions continue to be used in the construction and removal
is used both as a tool for probing the ocean and as a by- of undersea structures, primarily by the oil industry, but the
product of other activities. Anthropogenic sound sources frequency of detonations presumably has decreased over
vary in space and time but may be grouped into general cat- the past few decades.
egories: (a) explosions, (b) large commercial ships, (c) air- New classes of military vessels undergo tests, called ship-
guns and other seismic exploration devices, (d) military shock trials, to determine their ability to withstand explo-
sonars, (e) navigation and depth-finding sonars, (f ) research sions (Commander Naval Air Warfare Center 1994). During
sound sources, (g) acoustic harassment devices (AHDs) and a ship-shock trial, a large chemical explosion (e.g., 10,000 lb)
pingers, (h) polar icebreakers, (i) offshore drilling and other is detonated near the vessels hull and measurements of hull
industrial activity, and (j) small ships, boats, and personal stress are taken. Other Navy activities that involve under-
watercraft. water explosions include Sinkex, in which torpedoes or
Reynolds.100-123 6/10/05 4:54 PM Page 105

Impacts of Anthropogenic Sound 105

other chemical explosives are used to sink retired ships; isolated environments, the spectral characteristics of indi-
weapons being tested during development; and operational vidual ships can be discerned. At distant ranges in the open
stores being test fired to monitor their military readiness. ocean, multiple ships contribute to the background noise,
During the recent war in Iraq, Navy SEALS disposed of a and the sum of many distant sources creates broad spectral
dozen 500-lb sea mines confiscated from the Iraqi navy by peaks of noise in the 5- to 500-Hz band.
detonating them simultaneously in the Persian Gulf, creat- Models for representative sound spectra for different
ing a blast that could be heard 50 miles away in Kuwait (Dao classes of ships have been developed by the U.S. Navy. The
2003). research ambient noise directionality (RANDI) model
The spectral and amplitude characteristics of chemical (Wagstaff 1973, Schreiner 1990, Breeding 1993) uses ship
explosions vary with the weight of the charge and the depth length and speed as well as an empirically derived power law
of the detonation. The RMS source level of the initial shock to determine the broadband (5500 Hz) spectral level for
wave, a large component of the energy, is given by various classes of vessels. Peak spectral densities for indi-
vidual commercial ships range from 195 dB re Pa2/Hz at 1
SPL dB re 1Pa at 1 m = 269 dB + 7.53 log10 (w)
m for fast-moving (20 knots) supertankers to 140 dB re
where w is the charge weight in pounds (Urick 1975). For in- Pa2/Hz at 1 m for small fishing vessels. Source-level mod-
stance, 100 lb of TNT produces a shock wave SPL of 284 dB els also have been developed for the propeller tonal blade
re 1Pa at 1 m with an almost constant frequency content lines, occurring at 610 Hz and their harmonics for most of
from 10 to 1,000 Hz. The energy from the bubble pulse os- the worlds large merchant fleet (Gray and Greeley 1980).
cillations contribute approximately an additional 5 dB of Shipping vessel traffic is not uniformly distributed. The
source level, yielding a total SPL of 289 dB re 1Pa at 1 m. major commercial shipping lanes follow great circle routes
The signal duration can be obtained from the first few os- or coastlines to minimize the distance traveled. Dozens of
cillations of the bubble pulse, in this case lasting one-third major ports and megaports handle the majority of the
to one-half a second. traffic, but hundreds of small harbors and ports host smaller
volumes of traffic. The U.S. Navy lists 521 ports and 3,762
traffic lanes in its catalog of commercial and transportation
Commercial Shipping
marine traffic (Emery et al. 2001). Vessels found in areas out-
Commercial shipping is the principal source of low- side major shipping lanes include fishing vessels, military
frequency (5500 Hz) background noise in the worlds ships, scientific research ships, and recreational craftthe
oceans. Ships contribute to the noise level over large geo- last typically found nearshore.
graphic areas, and the sounds of individual vessels are often Lloyds Register of the worlds commercial fleet for the
spatially and temporally indistinguishable in distant vessel year 2001 listed 92,817 vessels National Research Council
traffic. Noise from vessel traffic at high latitudes propagates 2003b. The principal types (their numbers in parentheses)
particularly well over long distances because the oceanic are cargo/passenger transport (34,704), fishing (23,841),
sound channel (zone of most efficient sound propagation) towing/dredging (13,835), oil tankers (10,941), bulk dry
in those regions reaches the surface. transport (6,357), and offshore supply (3,139), but gross ton-
Ships noise is generated primarily from (a) propeller ac- nage may be a more important index of sound production
tion, (b) propulsion machinery, and (c) hydraulic flow over than vessel numbers. From that perspective, oil tankers and
the hull. Propeller noise is associated with cavitation (Ross bulk dry transport vessels represent nearly 50% of the total
1987, 1993), the creation of voids from zones of pressure be- tonnage but less than 19% of the total number of vessels.
low the ambient water pressure. The collapse of these voids Vessel operation statistics indicate steady growth in ves-
generates sound. Cavitation creates both broadband and sel traffic over the past few decades (Mazzuca 2001). There
tonal sounds, as it may be modulated by blade-passage fre- has been an increase in both the number of vessels and in
quencies and their harmonics, which are called the blade the tonnage of goods shipped. For example, there has been
lines in a spectrum. The broadband and tonal components a 30% increase in the volume of goods shipped by the U.S.
produced by cavitation account for 8085% of ship-radiated fleet (by flag and ownership) over the past 20 years (U.S.
noise power (Ross 1987). Propeller noise also may be gen- Maritime Administration 2003). Oceanic shipping is an effi-
erated by unsteady propeller blade-passage forces, and there cient means of transporting large quantities of goods and
is additional ship noise from propulsion machinery. materials globally. The globalization of economic infra-
Particular vessels produce unique acoustic signatures de- structure means that more raw materials, as well as finished
scribed by their source levels and frequency bands. Sharp goods, require long-distance transport. The economic in-
tonal peaks produced by rotating and reciprocating ma- centives for oceanic shipping are strong and, in the near
chinery such as diesel engines, diesel generators, pumps, term, there is no viable alternative for transporting large-
fans, blowers, hydraulic power plants, and other auxiliaries tonnage materials to distant locations.
often are seen in these acoustic signatures. Hydrodynamic The bulk of U.S. waterborne trade is conducted through
flow over the ships hull and hull appendages is an impor- a few ports (Table 7.1). For instance, the combined Califor-
tant broadband sound-generating mechanism, especially nia ports of Los Angeles and Long Beach carry 37% of the
with increased ship speed. At relatively short ranges and in total trade as measured by 20-ft-equivalent container traffic
Reynolds.100-123 6/10/05 4:54 PM Page 106

106

Table 7.1 U.S. foreign waterborne trade, hind a vessel, which operate at pressures of 2,000 psi and are
calendar year 2002 distributed over a region that measures 20 20 m.
The pressure output of an airgun array is proportional to
Rank U.S. port Total Export Import its operating pressure, the number of airguns, and the cube
root of the total gun volume. For consistency with the un-
1 Los Angeles 4,060 866 3,194 derwater acoustic literature, airgun-array source levels are
2 Long Beach 3,184 717 2,467 back-calculated to an equivalent source concentrated into a
3 New York 2,627 747 1,879
1-m-radius volume, yielding source levels as high as 259 dB
4 Charleston 1,197 521 676
5 Savannah 1,014 453 561 peak re 1 Pa at 1 m output pressure (Greene and Moore
6 Norfolk 982 431 551 1995). This effective source level predicts pressures in the far
7 Oakland 979 469 482 field of the array, but in the near field the maximum pres-
8 Houston 851 430 420 sure levels encountered are limited to 220230 dB peak re 1
9 Seattle 850 338 512
Pa. The far-field pressure from an airgun array is focused
10 Tacoma 769 278 491
Total 19,729 6,814 12,916 vertically, being about 6 dB stronger in the vertical direction
than in the horizontal direction for typical arrays. The peak
Source: U.S. Maritime Administration (2003). pressure levels for industry arrays are in the 5- to 300-Hz
Note: Units are thousands of twenty-foot-equivalent (TEU) containers. range.
Airguns are towed at speeds of about 5 knots and are typ-
ically fired every 10 s. A seagoing seismic-reflection opera-
tion includes a series of parallel passes through an area by a
(U.S. Maritime Administration 2003). Within the U.S. Ex- vessel towing an airgun array as well as six to ten seismic re-
clusive Economic Zone, this concentrates the noise from ceiving streamers (hydrophone arrays). A recent practice is
shipping into the regions adjacent to these ports and their the use of repeated seismic reflection surveys for time-
approaches. Significant concentrations of shipping traffic lapse monitoring of producing oil fields, called 4-D sur-
also occur in New York (13%) and the Puget Sound (Wash- veys. More than ninety seismic vessels are available world-
ington) area (8%). wide (Schmidt 2004), and about 20% of them are conducting
Short sea shipping is commercial waterborne trans- field operations at any given time (Tolstoy et al. 2004).
portation that does not transit an ocean. It is an alternative Offshore oil and gas exploration and construction activi-
form of commercial transportation that uses inland and ties occur along continental margins. Areas of currently ac-
coastal waterways to move commercial freight from major tivity include northern Alaska and northwestern Canada,
domestic ports to its destination. The U.S. Maritime Ad- eastern Canada, the U.S. and Mexican Gulf of Mexico,
ministration and the European Commission are playing ac- Venezuela, Brazil, West Africa, South Africa, North Sea,
tive roles in the development of short sea shipping to help Middle East, northwestern Australia, New Zealand, south-
reduce freight congestion on national rail and highway sys- ern China, Vietnam, Malaysia, Indonesia, and the Sea of
tems. Short sea shipping already accounts for 41% of the to- Okhotsk. New areas of exploration include the deepwater
tal goods transport market within Europe, compared to U.S. Gulf of Mexico and deepwater West Africa, both of
44% by road and 8% by rail (European Commission 2001). which have seen increasing activity in the past 510 years.
Short sea shipping is of particular concern with respect to A recent study of ambient noise in the North Atlantic
shipping noise and marine mammals because of its coastal suggests that airgun activity along the continental margins
setting. propagates into the deep ocean and is a significant compo-
nent of low-frequency noise (Nieukirk et al. 2004). Sounds
from airguns were recorded almost continuously during the
Seismic Exploration
summer, originating at locations over 3,000 km from the re-
Seismic reflection profiling uses high-intensity sound to im- ceiving hydrophones.
age the Earths crust. It is the primary technique for finding
and monitoring reserves of oil and natural gas and is used
Sonar
extensively by the fossil-fuel extraction industries. It is also
used by academic and government researchers to gather in- Sonar systems intentionally create acoustic energy to probe
formation for studies on Earths origin and tectonic history. the ocean. They seek information about objects within the
Arrays of airguns are the sound-producing elements in water column, at the sea bottom, or within the sediment.
seismic reflection profiling (Dragoset 1984, 2000). Airguns re- Active sonar emits high-intensity acoustic energy and re-
lease a specified volume of air under high pressure, creating ceives reflected and/or scattered energy. A wide range of
a sound pressure wave from the expansion and contraction sonar systems is in use for both civilian and military appli-
of the released air bubble. To yield high intensities, multiple cations. For purposes of discussion, sonar systems can be
airguns are fired with precise timing to produce a coherent categorized as low-frequency (<1 kHz), mid-frequency
pulse of sound. Oil industry airgun arrays typically involve (120 kHz), and high-frequency (>20 kHz).
twelve to forty-eight individual guns, towed about 200 m be- Military sonars are used for target detection, localization,
Reynolds.100-123 6/10/05 4:54 PM Page 107

Impacts of Anthropogenic Sound 107

and classification. They generally cover a broader frequency Other mid-frequency sonars are used by the Navy for
range with higher source levels than civilian sonars and are depth sounding, communication between platforms, and
operated during both training exercises and combat. Be- device activation. High-frequency sonars are incorporated
cause far more time is spent in training than in combat, train- either into weapons (torpedoes and mines) or weapon
ing exercises may be the primary context in which military countermeasures (mine countermeasures or antitorpedo
sonar is used. Low-frequency active (LFA) sonars are used for devices). They are designed to operate over ranges of a few
broadscale surveillance; they are designed to allow subma- hundred meters to a few kilometers. Mine-hunting sonars
rine tracking over scales of many hundreds to thousands of operate at tens of kilohertz for mine detection and above
kilometers. Specialized support ships are used to deploy LFA 100 kHz for mine localization. These sonars are highly di-
sonars, which consist of arrays of source elements sus- rectional and use pulsed signals. Other high-frequency mil-
pended vertically below the ship. The U.S. Navys surveil- itary sonars include side-scan sonar for seafloor mapping,
lance towed array sensor system (SURTASS) LFA sonar uses generally operated at frequencies near 100 kHz.
an array of eighteen projectors operating in the frequency Over the past decade, there has been a trend in the U.S.
range from 100 to 500 Hz, with a 215 dB re 1 Pa at 1 m Navy to emphasize training operations in coastal and shal-
source level for each projector ( Johnson 2002). These sys- low-water settings. There are now plans to construct shal-
tems are designed to project beams of energy in a horizon- low-water training ranges on both the West and East coasts
tal direction, and the effective source level of an LFA array, of the United States.
when viewed horizontally, can be 235 dB re 1 Pa at 1 m or Commercial sonars are designed for fish finding, depth
higher. The signal includes both constant-frequency (CF) sounding, and sub-bottom profiling. They typically gener-
and frequency-modulated (FM) components with a band- ate sound at frequencies of 3200 kHz, with only a narrow
width of approximately 30 Hz. A ping sequence can last from frequency band generated by an individual sonar system.
6 to 100 s, with a time between pings of 615 min and a typ- Source levels range from 150235 dB re 1 Pa at 1 m. Com-
ical duty cycle of 1015%. Signal transmissions are emitted mercial depth sounders and fish finders are typically de-
in patterned sequences that may last for days or weeks. signed to focus sound into a downward beam. Depth
Mid-frequency tactical antisubmarine warfare (ASW) sounders and sub-bottom profilers are designed, respec-
sonars are designed to detect submarines over several tens tively, to locate the sea bottom and to probe beneath it.
of kilometers. They are incorporated into the hulls of sub- They are operated primarily in nearshore and shallow en-
marine-hunting surface vessels such as destroyers, cruisers, vironments. Fish finders are used in both deep and shallow
and frigates (see Table 7.2). There are 117 of these sonars on areas.
U.S. Navy ships currently in active service and equivalent The acoustic characteristics of small-scale commercial
systems in foreign navies (e.g., British, Canadian, French) sonars are unlikely to change significantly in the future since
bring the worldwide count to about 300 (Watts 2003). The they are limited by several key physical properties. At the
AN/SQS-53C is the most advanced surface ship ASW sonar low-frequency end (about 3 kHz), they are limited by the
in use by the U.S. Navy, and it generates FM pulses of 12 s physical dimensions of the transducers. At the high-
duration in the 1- to 5-kHz band, at source levels of 235 dB frequency end (200 kHz), they are limited by severe attenu-
re 1 Pa at 1 m or higher (Evans and England 2001). This ation of sound. Likewise, the maximum power level that
sonar has a nominal 40 vertical beam width (dependent on can be emitted by a single transducer (200 dB re 1 Pa at 1
frequency), directed 3 down from the horizontal. The m) is limited by cavitation at shallow depths of operation.
AN/SQS-53C is designed to perform direct-path ASW Higher power levels can be achieved by constructing arrays
search, detection, localization, and tracking from a hull- of sensors on the hull of the vessel. For example, multibeam
mounted transducer array of 576 elements housed in a bul- echo-sounding systems (e.g., SeaBEAM or Hydrosweep)
bous dome located below the waterline of the ships bow. form narrow directional beams (e.g., 1 beam width) of
These systems are used to track both surface and sub- sound and are used for precise depth sounding. Using hull-
merged vessels, often detecting surface ships at greater mounted arrays of transducers, these systems can achieve
range than many radar systems. 235 dB re 1 Pa at 1 m source levels; they are typically oper-
ated at 1215 kHz in deep water and at higher frequencies
(up to 100 kHz) in shallow water. They may ensonify a swath
Table 7.2 U.S. Navy surface ships with of a few tens of kilometers along the track of the ship.
mid-frequency antisubmarine warfare sonars Sonar is an extremely efficient means for fish finding and
depth sounding/sub-bottom profiling. Nearly all of the
Type of ship Class Type of sonar Number in use 80,000 vessels in the worlds commercial fleet and many of
the 17 million small boats owned in the United States are
Cruiser Ticonderoga SQS-53 27 equipped with some form of commercial sonar, and new
Destroyer Spruance SQS-53 11 applications may lead to even greater proliferation of these
Arleigh Burke SQS-53 49
systems. It is possible that the impact of the large number
Frigate Oliver Hazard Perry SQS-56 30
Total 117 of these systems in use may be offset to some degree by
their limited range.
Reynolds.100-123 6/10/05 4:54 PM Page 108

108

Research Sound Sources pingers and AHDs have frequencies in the 5- to 160-kHz
band, and generate pulses lasting from 2 to 2,000 ms. To re-
Research in underwater acoustic propagation and acousti-
duce habituation, a single device may transmit with a vari-
cal oceanography often involves the use of sound. Almost
ety of waveforms and time intervals.
all of the programs in the United States are sponsored by
Pingers have been shown to be effective in reducing by-
the Office of Naval Research, and the information obtained
catch, at least for some marine mammal species in some set-
is of value for improving military sonar systems. The sound
tings (Kraus et al. 1997, Culik et al. 2001, Bordino et al.
sources used for these studies are either commercially avail-
2002). A trial of pinger use in the California drift gillnet fish-
able transducers or systems specially designed to meet spe-
ery for swordfish and sharks showed that for both cetaceans
cific research requirements. A wide variety of signals, band-
and pinnipeds, the entanglement rate in nets with pingers
widths, source levels, and duty cycles are transmitted during
was only one-third of what it was in nets without these de-
these projects. The spatial extent of most experiments is
vices (Barlow and Cameron 2003). Likewise, a large-scale
tens of kilometers, but basin-scale projects such as the
trial of pingers in Danish gillnet fisheries showed a reduc-
Acoustic Thermometry of Ocean Climate (ATOC) pro-
tion in bycatch of harbor porpoises (Larsen 1997, Vinther
gram have also been undertaken.
1999).
The ATOC (later the North Pacific Acoustic Laboratory
Concerns have arisen that the use of AHDs in aquacul-
[NPAL]) project was initiated in the early 1990s to study
ture facilities leads to unintended displacement of marine
ocean warming and received much attention from regula-
mammals, for example, killer whales (Morton and Symonds
tory agencies, the public, and the scientific community be-
2002) and harbor porpoises (Olesiuk et al. 2002) in the vicin-
cause of concerns regarding the potential impact of its
ity of salmon farms off British Columbia. Likewise, there
sound source on marine mammals (Baggeroer et al. 1998).
are concerns that widespread use of AHDs may lead to the
This program was extensively discussed in two National Re-
exclusion of porpoises from important feeding habitat
search Council (NRC) reports (National Research Council
( Johnston 2002). AHDs have sufficiently high source levels
1994, 2000a. The ATOC source has a 195 dB re 1 Pa at 1 m
that they could result in hearing damage to marine mam-
level and is deployed at 939 m, near the axis of the deep
mals exposed at close range.
sound channel (Howe 1996). It is designed to study the en-
tire North Pacific basin, with the sounds being received by
the U.S. Navys fixed hydrophone arrays. The transmitted Polar Icebreakers
signal is centered at 75 Hz with a bandwidth of 37.5 Hz. It
Ice-breaking ships are a source of noise in the polar regions
broadcasts at 4-h intervals with a ramp-up period of 5 min
(Erbe and Farmer 2000). Two types of noise have been iden-
and a full-power signal duration of 20 min. The long time
tified in association with ice breaking: bubbler system noise
frame for operation of this experiment was a key aspect that
and propeller cavitation noise. Some ice-breaking ships are
led to questions regarding its potential impact on marine
equipped with a bubbler system that blows high-pressure air
mammals (Potter 1994).
into the water around the ship to push floating ice away.
Another basin-scale sonar research project uses drifting
While the bubbler system is operating, the noise is continu-
sources (Rossby et al. 1986), called SOFAR floats. These de-
ous, with a broadband spectrum below 5 kHz. A source level
vices drift at depth and periodically emit a high-intensity
of 192 dB re 1 Pa at 1 m in one-twelfth-octave bands has
tone (195 dB re 1 Pa at 1 m) that is frequency swept at
been reported for bubbler system noise. Icebreaker pro-
200300 Hz or a continuous signal at 185310 Hz with a du-
peller cavitation noise is associated with the ships ramming
ration of 120 s or more. The sounds are detected by distant
the ice with its propeller turning at high speed. The spec-
receivers and their timing is used to determine the float lo-
trum of propeller cavitation noise is broadband up to at least
cation and therefore its drift, as a proxy for deep currents.
20 kHz, and has a source level of 197 dB re 1 Pa at 1 m.

Acoustic Deterrent Devices and Pingers Industrial Activities, Offshore Drilling,

and Construction
Acoustic deterrent devices (ADD) use sound in an effort to
repel marine mammals from fishing activities. The idea be- Industrial activities and construction both in the ocean and
hind these devices is that they keep the animals away by in- along the shoreline can contribute to underwater noise. Ex-
troducing a local acoustic annoyance or alerting signal. amples include coastal power plants, pile driving, dredging,
Pingers are used in some fisheries to reduce the bycatch of tunnel boring, power-generating wind mills, and canal lock
marine mammals by alerting them to the presence of, or operations (Greene and Moore 1995). The coupling of these
driving them away from, a net or other entangling object. sounds into the marine environment is poorly understood,
These are typically low-power ADDs with source levels of but it is generally more efficient at low frequencies.
130150 dB re 1 Pa at 1 m. Acoustic harassment devices Marine dredging is commonly conducted in coastal
(AHDs) are used to reduce depredation by marine mam- waters to deepen channels and harbors, reclaim land, and
mals on caught or cultured fish. These are high-powered de- mine seabed resources. Reported source levels for dredging
vices with source levels of 185195 dB re 1 Pa at 1 m. Both operations range from 160 to 180 dB re 1 Pa at 1 m for one-
Reynolds.100-123 6/10/05 4:54 PM Page 109

Impacts of Anthropogenic Sound 109

third-octave bands with peak intensity between 50 and 500 boaters (U.S. Fish and Wildlife Service 2001), and the num-
Hz (Greene and Moore 1995). ber of boats in operation is raised seasonally by an influx of
Oil and gas production activities that generate marine boats from out of state.
noise include drilling, offshore structure emplacement and
removal, and related transportation. Sound pressure levels
Comparison of Anthropogenic Sound Sources
associated with drilling are the highest with maximum
broadband (10 Hz10 kHz) energy of about 190 dB re 1 Pa The anthropogenic sound sources discussed previously are
at 1 m. Drill-ship noise comes from both the drilling ma- summarized by source level and other parameters in Table
chinery and the propellers and thrusters used for station 7.3, ordered by their intensity. For sources constructed from
keeping. Jack-up rigs are the most commonly used offshore arrays of elements (e.g., military sonars and airguns), the in-
drilling devices, followed by platform drill rigs. Drilling gen- dividual source elements can be widely distributed. In this
erates ancillary noise from the movements of supply boats case, the source level is given for a range of 1 m to stan-
and support helicopters. Emplacement of offshore struc- dardize the calculation, but in practice the actual levels ex-
tures creates localized noise for brief time periods, and perienced near the source never reach the stated levels. In-
powerful support vessels are used to transport these large stead, these levels are used to calculate accurately what
structures from the point of fabrication to the point of em- the source level is at longer ranges, where the distance to
placement. This activity may last for a few weeks and may the source is much greater than the source dimensions.
occur eight to ten times a year worldwide. Additional noise Table 7.3 is designed to approximate the potential for these
is generated during oil production activities, which include sources to impact acutely or injure marine mammals. In
borehole casing, cementing, perforating, pumping, pipe lay- practice, the sensitivity of marine mammals to various
ing, pile driving, and ship and helicopter support. Produc- kinds of sound is another important consideration, as dis-
tion activities can generate received levels as high as 135 dB cussed later in the chapter.
re 1 Pa at 1 km from the source (Greene and Moore 1995), Underwater nuclear tests and ship-shock trials produce
which suggests source levels as high as 195 dB re 1 Pa at 1 the highest overall sound pressure levels, yet these are rare
m with peak frequencies at 40100 Hz. events and so may be assumed to have limited impact over-
Oil and gas production is moving from shallow-water set- all. Military SURTASS-LFA sonars and large-volume airgun
tings into water depths of up to 3,000 m. Deepwater drilling arrays both have high SPLs. The long ping lengths and high
and production have the potential to generate higher levels duty cycles of LFA sonars increase their total energy levels;
of noise than shallow-water production, owing to the use of both the SURTASS-LFA and airgun arrays have dominant
drill ships and floating production facilities. In addition, energy at low frequencies, where long-range propagation is
noise generated in deep water may be more easily coupled likely. Mid-frequency military sonars (such as the SQS-53C)
into the deep sound channel for long-range propagation. have shorter ping durations and more moderate duty cycles;
The worldwide count of offshore mobile drill rigs in use because they operate at middle frequencies, propagation ef-
fluctuates with business conditions, but there is a growing fects also limit their range. Concern for the impact of these
number of drill rigs available, with an increase of approxi- sonars is for local settings (as discussed later in this chapter).
mately 10% over the past 5 years. Commercial supertankers are arguably the most nearly
ubiquitous sources of high-intensity, with more than 10,000
vessels operating worldwide. Concern with these noise
Small Ships, Boats, and Personal Watercraft
sources is concentrated near major ports and along the most
Small vessels do not contribute significantly to the global heavily traveled shipping lanes. The moored research sound
ocean sound environment, but may be important local source for the ATOC project is a source level equivalent to
sound sources, particularly in coastal settings. Examples of a supertanker, although it operates on a low duty cycle.
sound levels for whale-watching boats range from 115127 AHDs have high source levels, whereas ADDs have rela-
dB re 1 Pa at 1 m for one-third-octave bands (Au and Green tively moderate source levels. Multibeam hull-mounted
2000) and 145169 dB re 1 Pa at 1 m for one-twelfth-octave echo sounders have high source levels, but their narrow
bands, with increased sound levels for high-speed operation beam widths and medium frequencies limit their range and
(Erbe 2002). A recent study of noise levels from small the area that they ensonify. Research acoustic floats (RAFOS)
powerboats suggests peak spectral levels in the 350- to 1200- produce a moderately high source level but are operated at
Hz band of 145150 dB re 1 Pa2/Hz at 1 m (Bartlett and a very low duty cycle. Fishing vessels have moderate source
Wilson 2002). The total number of recreational craft in op- levels and may represent at least local acoustic annoyances.
eration is poorly documented although about 17 million
small boats are owned in the United States (National Marine
Anthropogenic Noise Energy Budget per Year
Manufacturers Association 2003). The vessel categories are
outboard (8.4 million), inboard (1.7 million), stern drive (1.8 An annual energy budget is one approach to comparing the
million), personal watercraft (1.4 million), sailboats (1.6 mil- contribution of each anthropogenic noise source. The ap-
lion), and miscellaneous (2.5 million). In the inshore waters proach taken here is to consider the acoustic energy output
of Florida, there are nearly 1 million registered recreational at the source itself, rather than the sum of many sources af-
Reynolds.100-123 6/10/05 4:54 PM Page 110

110

Table 7.3 Comparison of anthropogenic underwater sound sources ordered by their short-term ( 1 s)
energy output, approximating their potential for acute or injurious effects

Ping
energy Peak
SPL (dB re (dB re 1 frequency
Sound source 1 Pa at1m) Pa2-s) Ping duration Duty cycle (%) (Hz) Bandwidth (Hz) Directionality

Underwater nuclear 328 337 8s Intermittent Low Broad Omnidirectional

device
(30 kiloton)
Ship shock trial 299 302 2s Intermittent Low Broad Omnidirectional
(10,000 lb TNT)
Military sonar 235 243 6100 s 10 250 30 Horizontal
(SURTASS/LFA)
Airgun array 256 241 30 ms 0.3 50 150 Vertical
(2000 psi, 8000 in.3)
Military sonar 235 232 0.52 s 6 2,6003,300 Narrow Horizontal
mid-frequency
(SQS-53C)
Supertanker 185 [AQ4] Continuous 100 23 5100 Omnidirectional
(337 m length,
18 knots)
Research sonar 195 226 1200 s 8 75 37.5 Omnidirectional
(ATOC source)
Acoustic harassment 185 185 0.52 s 50 10,000 600 Omnidirectional
device
Multibeam 235 218 20 ms 0.4 12,000 Narrow Vertical
echosounder
(hull-mounted)
Research sonar 195 216 120 s Small 250 100 Omnidirectional
(RAFOS float)
Fishing Vessel 151 [AQ5] Continuous 100 300 250-1000 Omnidirectional
(12 m length,
7 knots)
Acoustic deterrent 132 127 300 ms 8 10,000 2000 Omnidirectional
device
(AquaMark300)

ter propagation within the ocean, as would be experienced tion, rate of usage, and total number of sources. The first
by a receiver at a particular location. Ambient noise distri- step is to convert sound pressure level (p) to acoustic inten-
butions at a given location result from a complex distribu- sity (I), obtained by dividing the squared pressure by the
tion of worldwide sources and variable acoustic propaga- acoustic impedance (Z):
tion. The question considered here is a simpler one: what is
I = |pv| = |p|2/Z [watts/m2]
the total energy output from each source type at the loca-
tion of the source. That is, all sources are assumed to be at The next step is to account for the directionality of the
a compact location, at a range of 1 m, and the total annual source. For omnidirectional sources, the acoustic power (P)
energy output of each source type is estimated. This is is given by the solid angle (A) emitted by the source (for an
clearly not the most desirable form of energy budget but is omnidirectional source this is 4, the area of a sphere of 1
amenable to a manageable tabulation. Table 7.3 shows the m radius) multiplied by the acoustic intensity (I):
approximate potential of these sources to produce chronic
P = AI [joules/sec]
effects on marine mammal populations. However, many
other factors have to be considered before these data can be The energy per source transmission or ping (Eping ) is given
used to help understand the impacts of sound on marine by the acoustic power multiplied by the duration of the
mammals, including the distribution of sources in space and transmission:
time and the varying sensitivities of marine mammals to
Eping = PTping [joules]
sound.
Starting with the source pressure levels given in Table 7.3, The number of source pings per year per source and the to-
the additional information needed to go from sound pres- tal number of sources in operation yield the annual energy
sure to total energy includes the source directionality, dura- output for each source type:
Reynolds.100-123 6/10/05 4:54 PM Page 111

Impacts of Anthropogenic Sound 111

Etotal = EpingNpings/yearNsources [joules] portance, the anthropogenic sources most likely to have
contributed to increased noise are commercial shipping, off-
For continuous sources, the energy of 1 s of transmission is
shore oil and gas exploration and drilling, and naval and
used for Eping, and the number of seconds the source is in
other uses of sonar.
operation per year is used for Npings/year.
Waters surrounding Australia, which are remote from
A proposed annual anthropogenic energy budget is pre-
most commercial shipping, allow the effects of anthropo-
sented in Table 7.4, starting with the sources and sound
genic and natural noise to be separated. At low frequency
pressure levels from Table 7.3. Underwater nuclear explo-
(100 Hz), Australian data suggest that ocean noise levels
sions, even assuming a 20-year recurrence rate, top the an-
may be as low as 50 dB re 1 Pa2 /Hz, which is about 3040
nual energy budget with 2.1 1015 J. This is comparable to
dB below levels in North American and European waters
a small power plant of 100 MW with an annual energy out-
(Cato and McCauley 2002). These data further suggest that
put of 3.2 1015 J. The highest-energy regularly operated
wind/wave noise increases at low frequencies, in contrast to
sound sources are the airgun arrays from 90 vessels operat-
the predictions of the deepwater curves developed from
ing for 80 days/year to produce 3.9 1013 J. Military sonars
Northern Hemisphere data (Wenz 1962). see The National
for antisubmarine warfare (SSQ-53C) used on 300 vessels for
Research Council 2003b) pointed to the difficulty of sepa-
30 days/year produce 2.6 1013 J. The contribution from
rating wind/wave-generated noise from shipping noise in
shipping comes mostly from the largest vessel classes, with
North American datasets.
11,000 supertankers, operating 300 days/year, to yield 3.7
Trends in background noise and anthropogenic activity
1012 J. Lesser contributions are made by other vessel classes
levels suggest that ocean noise levels increased by 10 dB or
(e.g., merchant and fishing) and by navigation and research
more between 1950 and 1975 (Ross 1987, 1993). These
sonars. For comparison at the low-energy end, a symphony
trends are most apparent in the eastern Pacific and eastern
orchestra produces about 10 W of sound energy, and would
and western Atlantic, where they are attributed to increases
emit 3.2 108 J over the course of a year.
in commercial shipping. A doubling of the number of ships
explains 35 dB, and greater average ship speeds, propul-
LONG-TERM TRENDS sion power, and propeller tip speeds may explain an addi-
IN OCEAN NOISE tional 6 dB.
Other data on long-term noise trends come from a com-
Overall trends for the level of sound in the sea can be bro- parison of historical U.S. Navy acoustic array data (Wenz
ken down into anthropogenic and nonanthropogenic com- 1969) with modern recordings along the west coast of North
ponents. For instance, there is evidence that global climate America (Andrew et al. 2002). A low-frequency noise in-
change may have resulted in higher sea states (Bacon and crease of 10 dB over 33 years was observed at a site off the
Carter 1993, Graham and Diaz 2001), which would have the central California coast. The explanation for a noise increase
effect of increasing background noise levels. Over the past in this band is the growth in commercial shipping in terms
few decades, however, it is likely that increases in anthro- of both the number of ships and the gross tonnage. From
pogenic noise have been more prominent. In order of im- 1972 to 1999 the total number of ships in the worlds fleet in-

Table 7.4 Comparison of anthropogenic underwater sound sources ordered by their total annual
energy output

Intensity Directionality Power Number Operate Repetition

Sound source (dB re 1 W/m2) (angle) (dB re 1 W) of sources (days/year) (pings/day) Total energy ( J)

Underwater nuclear 146 4 157 1 0.05 1 2.1 1015

explosions
Airgun arrays 61 66 90 80 4320 3.9 1013
Military sonar 53 /2 55 300 30 4,320 2.6 1013
(mid-frequency)
Supertankers 3.2 2 11 11,000 300 86,400 3.7 1012
Ship-shock trials 117 4 128 1 0.5 1 3.3 1012
Military sonar 53 58 1 30 175 1.7 1011
(SURTASS/LFA)
Merchant vessels 17 2 8.8 40,000 300 86,400 1.4 1011
Navigation sonar 1.8 3.2 100,000 100 86,400 3.6 1010
Fishing vessels 31 2 23 25,000 150 86,400 1.7 109
Research sonar 13 4 24 10 4 86,400 9.1 108

Note: Although this table is designed to approximate the potential of these sources to produce chronic effects, many other factors must be considered, including the distribution
of sources in space and time and the sensitivities of marine mammals to sound.
Reynolds.100-123 6/10/05 4:54 PM Page 112

112

creased from approximately 57,000 to 87,000, and the total velop a global model of ocean noise (National Research
gross tonnage increased from 268 to 543 million gross tons. Council 2003b:92), that incorporates both transient and
Mazzuca (2001) compared the results of Wenz (1969), continuous noise sources. The development of an accurate
Ross (1987), and Andrew et al. (2002) to derive an overall in- global model depends on access to ocean noise data and an-
crease of 16 dB in low-frequency noise from 1950 to 2000. thropogenic activity data collected by long-term monitor-
This corresponds to a doubling of noise power (3 dB) every ing, as described previously.
decade for the past 50 years, equivalent to a 7% annual in-
crease in noise. During this period the number of ships in Priority 4: Report signal characteristics for anthropogenic noise
the world fleet tripled (from 30,000 to 87,000) and the gross sources. An important component of model development is
tonnage increased by a factor of 6.5 (from 85 to 550 million better understanding of the signal characteristics for repre-
gross tons) (National Research Council 2003b; from Mc- sentative anthropogenic sound sources. The description of
Carthy and Miller 2002). each source should include enough information to allow re-
construction of its character (e.g., frequency content, pres-
sure and/or particle-velocity time series, duration, repeti-
OCEAN NOISE tion rate).
RESEARCH PRIORITIES

Ocean noise is an important component of the marine en- Priority 5: Determine the relationship between anthropogenic
vironment. Data on ocean noise trends are scarce, despite activity level and noise level. Research should be conducted re-
substantial investments by the U.S. Government in the col- lating the overall levels of anthropogenic activity (such as
lection of underwater sound data for military purposes the types and numbers of vessels) with the resulting noise
(e.g., SOSUS and other ASW monitoring systems). Expand- (National Research Council 2003b:90). These correlations
ing use of the sea for commercial shipping, oil and gas de- will help to extend noise modeling to areas without direct
velopment, and advanced warfare has resulted in noise lev- long-term monitoring, but where anthropogenic noise
els that are at least ten times higher today than they were a sources are present.
few decades ago. Without some effort to monitor, reduce,
or at least cap them, these noise levels are likely to increase
and further degrade the marine acoustic environment. Rec- HOW SOUND AFFECTS
ommendations for tracking and improving our under- MARINE MAMMALS
standing of ocean noise sources are presented below.
The responses of marine mammals to sound depend on a
Priority 1: Initiate long-term global ocean noise monitoring. A range of factors, including (a) the sound pressure level and
long-term monitoring program is needed to track future other properties, for example, frequency, duration, novelty,
changes in ocean noise (National Research Council and habituation; (b) the physical and behavioral state of the
2003b:90). Acoustic data should be included in global ocean animals; and (c) the ambient acoustic and ecological fea-
observing systems now being developed by U.S. and inter- tures of the environment. Richardson et al. (1995) reviewed
national research agencies. Data from these monitoring sys- marine mammal responses to specific sound sources, but
tems should be openly available and presented in a manner our present understanding of how marine mammals re-
accessible to decision makers in industry, the military, and spond to sound is insufficient to allow reliable predictions of
regulatory agencies. behavioral responses either to intense sounds or to long-
term increases in ambient background noise.
Priority 2: Analyze historical marine anthropogenic noise data. In humans, the perceived loudness of a sound involves
In tandem with the effort to monitor present-day ocean not only hearing sensitivity but also psychological and phys-
noise, a program should be developed to collect, organize, iological factors (Beranek and Ver 1992). A loudness-level
and standardize data on ocean noise and related anthro- scale has been developed from detailed testing, where the
pogenic activities (National Research Council 2003b:89). In- human subject judges the relative loudness of two sounds;
frastructure appropriate for maintaining an archive of these for instance, the phon (in dB) compares the loudness level
data already exists (i.e., the National Oceanographic Data of tones of varying frequency to a 1-kHz reference tone. In
Center, www.nodc.noaa.gov). Currently, data regarding practice, the annoyance level of a sound depends on a range
shipping, seismic exploration, oil and gas production, and of factors apart from loudness, such as the sounds fluctua-
other marine activities are either not collected or are diffi- tion; intermittent sounds are more annoying than continu-
cult to obtain and analyze because they are maintained by ous ones. The degree to which human and terrestrial ani-
separate organizations. International cooperation in this ef- mal studies can be reliably extrapolated to marine mammals
fort should be encouraged. is uncertain because there are vast differences in the role of
sound in sensing the marine and terrestrial environments
Priority 3: Develop global models for ocean noise. Marine and the ambient and biologically significant sounds, such as
noise measurements and source data should be used to de- those of predators, differ in each setting.
Reynolds.100-123 6/10/05 4:54 PM Page 113

Impacts of Anthropogenic Sound 113

MARINE MAMMAL (Pseudorca crassidens; Thomas and Turl 1990) and for bot-
SOUND PRODUCTION tlenose dolphins echolocating in the presence of noise (Au
1993), and 232 dB re 1 Pa at 1 m for male sperm whales
The frequency band of sounds that are important to marine (Mohl et al. 2000). The short duration of such echolocation
mammals matches, or extends beyond, the range of the clicks (50200 s) means that their total energy is low (197
sounds that they produce. Marine mammal call frequencies dB re 1 Pa2-s) although their source levels are high. Odon-
generally show an inverse correlation with body size tocete whistles have lower source levels than their clicks,
(Watkins and Wartzok 1985), with mysticetes having larger ranging from less than 110 dB re 1 Pa at 1 m for the spin-
bodies and lower frequency calls than odontocetes. ner dolphin (Stenella longirostris; Watkins and Schevill 1974)
For mysticetes, most sound production is in the low- to 169 dB re 1 Pa at 1 m for bottlenose dolphins ( Janik
frequency range of 102,000 Hz (Edds-Walton 1997). Mys- 2000). The detection range for odontocete clicks and whis-
ticete sounds can be broadly characterized as (a) tonal calls, tles is about 5 km, although greater detection ranges also
(b) FM sweeps, (c) pulsed tonals, and (d) broadband grunt- have been reported (Leaper et al. 1992, Barlow and Taylor
like sounds and are generated either as individual calls or 1998, Gordon et al. 2000).
combined into patterned sequences or songs. For odonto- Mysticete calls can be detected over long ranges (Payne
cetes, most sound production is in the mid-frequency and and Webb 1971). For instance, blue whales (Balaenoptera
high-frequency range of 1200 kHz (Matthews et al. 1999). musculus) produce low-frequency (10100 Hz) calls with es-
Odontocetes produce (a) broadband clicks with peak en- timated source levels of 185 dB re 1 Pa at 1 m (McDonald
ergy between 5 and 150 kHz, varying by species, (b) burst- et al. 2001), which are detectable at ranges of 100 km or
pulse click trains, and (c) tonal or FM whistles that range more, depending on the acoustic propagation. Most large
from 1 to 25 kHz. Pinnipeds that breed on land produce a mysticetes (blue, fin, bowhead, right, humpback, Brydes,
limited array of barks and clicks ranging from less than 14 minke, and gray whales) are known to vocalize at frequen-
kHz. Those that mate in the water produce complex vocal- cies below 1 kHz, with estimated source levels as high as 185
izations during the breeding season. All pinnipeds, the sea dB re 1 Pa at 1 m (Richardson et al. 1995).
otter, and manatees use sound to establish and maintain the Source levels and frequencies have been estimated for the
mother-young bond, especially when attempting to reunite underwater calls of several species of pinnipeds. Examples
after separation (Sandegren et al. 1973, Hartman 1979). are the Weddell seal (Leptonychotes weddellii), which pro-
The ability to use self-generated sounds to obtain infor- duces calls from 148 to 193 dB re 1 Pa at 1 m at frequencies
mation about objects and features of the environment, of 0.212.8 kHz (Thomas and Kuechle 1982), and the Ross
called echolocation, has been demonstrated for at least thir- seal (Ommatophoca rossii), which produces calls at 14 kHz
teen odontocete species (Richardson et al. 1995). No odon- (Watkins and Ray 1985). These calls can be detected at
tocete has been shown to be incapable of echolocation, and ranges of several kilometers both in the open ocean and un-
echolocation clicks have been observed in all recorded der ice (Wartzok et al. 1982).
species. These echolocation sounds are produced in for-
ward-directed beams, using specialized fats in the forehead
(melon) as acoustic lenses. Some species of odontocetes MARINE MAMMAL HEARING
produce few or no whistles and very-high-frequency clicks
with peak spectra above 100 kHz. Examples include the Sound propagates efficiently underwater, and one reflection
Amazon river dolphin (Inia geoffrensis; Norris et al. 1972), of its importance to marine mammals is their development
and the harbor porpoise (Phocoena phocoena; Kamminga of broader hearing frequency ranges than is typical for ter-
1988). restrial mammals. Audiograms have been produced for
Other odontocetes produce clicks with peak spectra be- eleven species of odontocetes and nine species of pinnipeds,
low 80 kHz and use whistles regularly. Examples include out of a total of approximately 127 marine mammal species
bottlenose dolphins (Tursiops spp.), which are coastal, and (Wartzok and Ketten 1999, Nachtigall et al. 2000). All hear-
the pantropical spotted dolphin (Stenella attenuata), which ing data are from species that are small enough to be held in
often occurs in offshore waters. Deep-diving odontocetes, captivity. Direct hearing data are not available for species
such as sperm whales (Physeter macrocephalus) and beaked that are not readily tested by conventional audiometric
whales (Ziphiidae), are only known to produce clicks (Mohl methods. For the latter, audiograms must be estimated
et al. 2000, Hooker and Whitehead 2002, Johnson et al. from mathematical models based on ear anatomy or in-
2004). Some odontocete whistles have been described as ferred from the sounds they produce and field-exposure ex-
signature calls that identify individuals (Caldwell and periments (Wartzok and Ketten 1999).
Caldwell 1965). Sounds produced by killer whales are Most delphinids are thought to have functional hearing
known to be group specific (Ford 1991, Tyack 2000), and the from 200 Hz to 100 kHz, and some smaller species may
patterned coda click sequences made by sperm whales hear frequencies as high as 200 kHz. Delphinid audiograms
show geographic variation (Rendell and Whitehead 2003). measured to date show peak sensitivity between 20 and 80
Source levels for odontocete clicks have been reported to kHz, along with moderate sensitivity at 120 kHz. Since
be as high as 228 dB re 1 Pa at 1 m for false killer whales ambient noise decreases at high frequencies, odontocetes
Reynolds.100-123 6/10/05 4:54 PM Page 114

114

may find it advantageous to hear and echolocate at high fre- and total energy fluxes of 186 dB re 1 Pa2-s produces TTS
quencies to avoid low-frequency background noise. An- in beluga whales (Delphinapterus leucas; Finneran et al. 2002).
other factor favoring high frequencies for some odonto- One hypothesis is that animals are most vulnerable to
cetes is the acoustic background noise of inshore and TTS at or near the frequencies of their greatest hearing
riverine environments. Higher frequencies also give better sensitivity. For baleen whales, this suggests low-frequency
spatial resolution, but at the expense of diminished propa- sensitivity and for smaller cetaceans, mid-frequency and
gation distance. high-frequency sensitivity. It also raises the question of
Based on modeling but with no measured audiograms, why marine mammals (apparently) do not damage their
mysticete hearing probably ranges between 20 Hz and hearing by their own sound production, as both the tonal
2030 kHz. Several of the larger species, such as blue and fin and impulsive sounds that they produce can be comparable
whales, are thought to hear at infrasonic (down to 10 Hz) in sound level to those found to induce TTS in the con-
frequencies. Pinniped audiogram data suggest that their trolled experiments mentioned previously. It is thought that
best hearing is between 1 and 20 kHz. True seals (phocids) internal mechanisms may protect an animal from its own
tend to hear higher frequencies underwater than fur seals vocalizations.
and sea lions (otariids). Some pinnipeds hear moderately
well in both water and air, whereas others are better adapted
Masking
for underwater than for in-air hearing. Sea lions have the
most terrestrially adapted hearing, and elephant seals have Acoustic signals are detected against the ambient back-
the most marine-adapted hearing, with good sensitivity be- ground noise. When background noise increases, it may re-
low 1 kHz (Kastak and Schusterman 1998). duce an animals ability to detect relevant sound; this is
Marine mammal hearing is adapted to an aquatic envi- called masking. Noise is effective for masking when it is
ronment, yet their hearing anatomy evolved from terrestrial within a critical band (CB) of frequency around the desired
ancestors. The divergence in hearing physiology between signal. The amount by which a pure tone must exceed the
land and marine mammals is most pronounced in the ex- noise spectral level to be audible is called the critical ratio
ternal ears, which are absent in most marine mammal (CR). The CR is related to the bandwidth (CB) within which
species, and in the middle ears, which are extensively mod- background noise affects the animals ability to detect a
ified in marine mammals. Because levels of ambient noise sound. Estimates of marine mammal CBs and CRs come
in the sea can vary by many orders of magnitude as a result from captive odontocetes and pinnipeds (Richardson et al.
of storms and other natural phenomena, marine mammals 1995). For all species, the CB expressed as a percentage is
may have developed resilient mechanisms to guard against broader at low frequencies (2575% at 100 Hz), and nar-
hearing loss. Existing studies do not allow for prediction of rower (110 %) at middle and high frequencies (1100 kHz).
the impacts of high-intensity sounds on marine mammal This suggests that band-limited noise is more effective at
hearing except in general terms. masking low frequencies than middle and high frequencies.
An animals directional hearing capabilities may help it
avoid masking by resolving the different directions of prop-
Hearing Losses
agation between the signal and the noise. A directivity index
Hearing thresholds may be degraded by exposure to high- of as much as 20 dB has been measured for bottlenose dol-
intensity sound. Hearing losses are classified as either tem- phins (Au and Moore 1984). Directional hearing is less acute
porary threshold shifts (TTS) or permanent threshold shifts in pinnipeds.
(PTS), where threshold shift refers to the raising of the min- Erbe and associates have studied masking of beluga
imum sound level needed for audibility. Repeated TTS is whale sounds by icebreaker noise (Erbe and Farmer 1998,
thought to lead to PTS. The extent of hearing loss is related 2000, Erbe 2000), including construction of software to
to the sound power spectrum, the hearing sensitivity, and model this process (Erbe et al. 1999). Icebreaker noise from
the duration of exposure. High-intensity, impulsive blasts ramming, ice cracking, and bubbler systems produced
can damage cetacean ears (Ketten et al. 1993). Hearing masking at noise-to-signal ratios of 1529 dB. The predicted
losses reduce the range for communication, interfere with zone of masking for beluga calls from ramming noise was
foraging capacity, increase vulnerability to predators, and 40 km (Erbe and Farmer 2000). Beluga whales vocal output
may cause erratic behavior with respect to migration, mat- changes when they are moved to locations with higher back-
ing, and stranding. For cetaceans, which are highly depend- ground noise (Au et al. 1985). With noise at low frequencies,
ent on their acoustic sense, both TTS and PTS should be an animal increases both the sound pressure level and the
considered serious cause for concern. frequency of its vocalizations, perhaps in an attempt to
Relatively few data are available on hearing loss in marine avoid or overcome masking. Beluga whales also have been
mammals. Experiments on captive bottlenose dolphins observed to increase call rates and shift to higher call fre-
suggest that TTS are observed at levels of 193196 dB re 1 quencies in response to boat noise (Lesage et al. 1999). Like-
Pa for exposure to 1-s tones at 20 kHz (Ridgway et al. 1997). wise, it has been suggested that killer whales shift their call
Work with impulsive sources (seismic waterguns) suggests frequencies in response to the presence of whale-watching
that exposure to sound pressure levels of 217 dB re 1 Pa boats (Foote et al. 2004).
Reynolds.100-123 6/10/05 4:54 PM Page 115

Impacts of Anthropogenic Sound 115

Hearing Development human divers exposed to intense underwater sound (160

180 dB re 1 Pa for 15 min). Symptoms during exposure in-
Does increased noise in the oceans cause developmental
cluded head vibrations, lightheadedness, somnolence, and
problems for young animals? High-noise environments af-
an inability to concentrate. These divers reported recurrent
fect auditory development in very young rats (Chang and
symptoms days to weeks after exposure, including, in one
Merzenich 2003). Brain circuits that receive and interpret
case, a partial seizure 16 months after the initial exposure
sound did not develop at the same rate in animals living in
(Stevens et al. 1999). Effects of this type have yet to be stud-
an environment of high continuous background noise as in
ied in marine mammals.
animals that were raised in a quiet environment. It took
Sound exposure in humans may elicit a vestibular re-
three to four times as long for rats raised in a noisy envi-
sponse or dizziness (called the Tullio phenomenon) at
ronment to reach the basic benchmarks of auditory devel-
thresholds as low as 101136 dB re 1 Pa (Erlich and Law-
opment. For marine mammals, comparable data may be dif-
son 1980). When human diver vestibular function was as-
ficult to obtain, but the potential for developmental
sessed before and after underwater sound exposure, tran-
impairment should be an important consideration when as-
sient effects were detected immediately postexposure to 160
sessing the impacts of ocean noise.
dB re 1 Pa for 15 min (Clark et al. 1996). Likewise, rats ex-
posed to 180 dB re 1 Pa for 5 min exhibited mild transient
impairment in vestibulomotor function (Laurer et al. 2002),
NONAUDITORY SOUND IMPACTS and vestibular effects have been detected in guinea pigs im-
mediately following underwater sound exposure of 160 dB
Nonauditory effects involve the interaction of sound with re 1 Pa for 5 min ( Jackson and Kopke 1998).
marine mammal physiology. Sound is known to have direct Acoustic resonance can lead to an amplification of pres-
and indirect physiological effects on mammals apart from sure within mammalian air cavities in response to sound.
its effects on hearing discussed previously. The symptoms of Lung and other air cavity resonance is important for estab-
these physiological effects range from subtle disturbance, to lishing thresholds for injury because at any given level of ex-
stress, to injury, and to death. citation, the vibration amplitude is greatest at resonance. In
The physiology of marine mammals is uniquely adapted vivo and theoretical studies related to tissue damage sup-
to life underwater. For example, deep-diving species have port a damage threshold of the order of 180190 dB re 1 Pa
specialized cardiovascular and pulmonary systems that al- (Cudahy et al. 1999, Cudahy and Ellison n.d.[AQ2]). These
low breath holding and accommodation to changes in pres- studies also provide a relationship between resonance and
sure. The same physiology that allows marine mammals to body mass, based on underwater measurements of terres-
spend extended periods underwater and make deep dives trial mammals, including humans, as well as extrapolation
may also create vulnerabilities to sound exposure, and their from in-air results. Finneran (2003) measured resonance
physiological responses to such exposure may differ from frequencies for beluga whale and bottlenose dolphin lungs
those of humans and other terrestrial mammals. directly and found them to be at low frequencies (30 and 36
Research on human divers, laboratory terrestrial animals, Hz, respectively). An important issue for resonance effects
and captive marine mammals suggests that exposure to is the tuning or amplification effect of the resonance. The
underwater sound can produce nonauditory physiological degree of tuning (defined as Q, with high Q indicating
effects. The range of potential impacts may include physio- sharper tuning) that has been measured in vivo in the lungs
logical stress, neurosensory effects, effects on balance (ves- of pigs and humans is from 3 to 5 (Martin et al. 2000),
tibular response), tissue damage from acoustic resonance, and for beluga whales and bottlenose dolphins is 2.5 and 3.1,
gas bubble formation and/or growth in tissues and blood, respectively (Finneran 2003). This suggests that a moderate
and blast-trauma injury. level of amplification (a factor of 3) occurs at resonance
The term stress is used to describe physiological changes frequencies.
that occur in the immune and neuroendocrine systems fol- Sound can increase gas bubble presence in mammalian
lowing exposure to a stressor. Physiological stress responses tissues, especially when dissolved gases are abundant as a re-
are not fully understood; however, indicators of stress sult of repeated dives. Human divers are obliged to decom-
that is due to noise have been measured in marine mam- press slowly following dives to prevent bubble formation,
mals. For instance, dolphins experience heart rate changes whereas deep-diving marine mammals have evolved a
in response to sound exposure (Miksis et al. 2001). A beluga means to avoid decompression sickness during their routine
whale showed increased stress hormone levels (norepi- diving activity. Intense sound generates bubblesin vivo
nephrine, epinephrine, and dopamine) with increased cavitation (ter Harr et al. 1982); it also leads to bubble
sound exposure level (Romano et al. 2004). Prolonged growthrectified diffusion (Crum and Mao 1996, Houser
noise-induced stress can lead to debilitation such as infertil- et al. 2001). The growth of bubbles increases the potential
ity, pathological changes in digestive and reproductive or- for blocked arteries.
gans, and reduced growth, as documented for some fish and Intense pressures from sources such as explosions can
invertebrates (Banner and Hyatt 1973, Lagardere 1982). damage air-filled cavities, such as lungs, sinuses, ears, and in-
Cases of neurologic disturbance have been described for testines (Cudahy et al. 1999). A dramatic pressure drop, such
Reynolds.100-123 6/10/05 4:54 PM Page 116

116

as occurs from blast waves, may cause air-filled organs to (Madsen and Mohl 2000). A fin whale (Balaenoptera physalus)
rupture. Research on blast damage in animals suggests that continued to call with no change in rate, level, or frequency
the mechanical impact of a short-duration pressure pulse in the presence of noise from a container ship (Edds 1988).
(positive acoustic impulse) is best correlated with organ Age and sex are important factors in noise sensitivity. For
damage (Greene and Moore 1995). Peak pressures of 222 dB instance, juvenile and pregnant Steller sea lions (Eumetopias
re 1 Pa result in perforation and hemorrhage of air-filled jubatus) are more likely to leave a haul-out site in response
intestines in rats (Bauman et al. 1997). Lethal peak pressures to aircraft overflights than are territory-holding males and
of 237 dB re 1 Pa cause pulmonary contusion, hemor- females with young (Calkins 1979). Walruses (Odobenus ros-
rhage, barotraumas, and arterial gas embolisms in sheep marus) may stampede and crush calves (Loughrey 1959) or
(Fletcher et al. 1976). Two humpback whales were found temporarily abandon them (Fay et al. 1984) when exposed
dead following a nearby 5,000-kg explosion, and examina- to sounds from aircraft or vessels. In gray whales, cow-calf
tion of the temporal bones in their ears revealed significant pairs are considered more sensitive than other age or sex
blast trauma (Ketten et al. 1993). classes to disturbance by whale-watching boats (Tilt 1985),
and humpback groups containing at least one calf appear to
be more sensitive to vessel traffic than are groups without
EFFECTS OF NOISE ON calves (Bauer et al. 1993).
MARINE MAMMAL BEHAVIOR Marine mammal responses also appear to be affected by
the context of the exposure, for example, by the location of
The behavioral responses of marine mammals to noise are the source relative to that of the animal, by the motion of
complex and poorly understood (Richardson et al. 1995). the source, and by the onset of the source and its repetition
Responses may depend on hearing sensitivity, behavioral (random versus periodic and predictable). Fin whales are
state, habituation or desensitization, age, sex, presence of more tolerant of a stationary than a moving source
offspring, location of exposure, and proximity to a shore- (Watkins 1986). Humpback whales are less likely to react to
line. They may range from subtle changes in surfacing and a continuous source than to one with a sudden onset
breathing patterns to cessation of vocalization to active (Malme et al. 1985). California sea lions (Zalophus californi-
avoidance or escape from the region of highest sound lev- anus) and harbor seals (Phoca vitulina) react at greater range
els. For instance, several studies suggest that bowhead from a ship when they are hauled out, and this is also true
whales follow a pattern of shorter surfacings, shorter dives, of walruses (Fay et al. 1984). Bowheads are more responsive
fewer blows per surfacing, and longer intervals between to overflights of aircraft when they are in shallow water
blows when exposed to anthropogenic noise (Richardson et (Richardson and Malme 1993). In the St. Lawrence River,
al. 1995), even at moderate received levels (114 dB re 1 Pa). beluga whales are less likely to change their swimming and
Another common response pattern is a reduction or cessa- diving patterns in the presence of vessels moving at low
tion of vocalization, such as for right whales in response to speed than in the presence of fast-moving boats (Blane and
boat noise (Watkins 1986), bowheads in response to play- Jaakson 1994). In Alaska, beluga whales feeding on river
back of industrial sounds (Wartzok et al. 1989), sperm salmon may stop and move downstream in response to
whales in response to pulses from acoustic pingers (Watkins noise from small boats, whereas they are relatively unre-
and Schevill 1975) and in the presence of military sonar sponsive to noise from fishing boats (Stewart et al. 1982). In
(Watkins et al. 1985), and sperm and pilot whales (Globi- Bristol Bay, beluga whales continue to feed when sur-
cephala spp.) in response to an acoustic source for oceano- rounded by fishing vessels, and they may resist dispersal
graphic research (Bowles et al. 1994). Moreover, humpback even when deliberately harassed (Fish and Vania 1971). In
whales lengthen their song cycles when exposed to the LFA Sarasota Bay, bottlenose dolphins had longer interbreath in-
source (Miller et al. 2000), move away from mid-frequency tervals during approaches by small boats (Nowacek et al.
sonar (Maybaum 1993), and tend to cease vocalizations 2001). In Kings Bay, Florida, manatees use of boat-free sanc-
when near boats (Watkins 1986). Beluga whales adjust their tuaries increased as the number of boats in the bay in-
echolocation clicks to higher frequencies and higher source creased (Buckingham et al. 1999).
levels in the presence of increased background noise (Au Only a few studies document long-term marine mammal
et al. 1985). Gray whales (Eschrichtius robustus) exhibited an responses to anthropogenic sound, suggesting abandon-
avoidance response when exposed to airgun noise, and their ment of habitat in some cases. At Guerrero Negro Lagoon
response became stronger as the source level increased from in Baja California, shipping and dredging associated with a
164 to 180 dB re 1 Pa (Malme et al. 1984). They also pref- salt works may have induced gray whales to abandon the
erentially avoided LFA transmissions conducted in a land- area through most of the 1960s (Bryant et al. 1984). After
ward direction (Tyack and Clark 1998). these activities stopped, the lagoon was reoccupied, first by
Marine mammals have been observed to have little or no single whales and later by cow-calf pairs. Killer whales (Or-
reaction to some anthropogenic sounds. For example, cinus orca) in the British Columbia region were displaced
sperm whales continued calling when they encountered from Broughton Archipelago in 19931999, a period when
echosounders (Watkins 1977) and when they were exposed acoustic harassment devices were in use on salmon farms
to received sound levels of 180 dB re 1Pa from a detonator (Morton and Symonds 2002).
Reynolds.100-123 6/10/05 4:54 PM Page 117

Impacts of Anthropogenic Sound 117

HABITUATION AND larly Cuviers beaked whales (Ziphius cavirostris). In many of

TOLERANCE OF NOISE the areas where such events occurred, Cuviers beaked
whale was not thought to be the most abundant cetacean
Habituation is the loss of responsiveness to noise over time. species present.
A diminution of responsiveness over time may be due to the Odontocetes are known to mass strand, that is, to come
animals becoming accustomed to, and no longer threat- ashore in groups of two or more animals (Walsh et al. 2001).
ened by, the signal. Alternatively, animals may return to the Mass strandings of beaked whales, however, are relatively
noisy area because of its importance, despite the annoying rare. The National Museum of Natural History, Smithson-
nature of the sound. The best evidence for habituation of ian Institution ( J. Mead pers. comm.) has compiled a global
marine mammals to intense sound comes from attempts to list of Cuviers beaked whale strandings involving two or
use acoustic harassment devices (AHDs) to keep marine more animals (Table 7.5). Except for a stranding of two in-
mammals away from aquaculture facilities or fishing equip- dividuals in 1914, there are no records of multiple-animal
ment ( Jefferson and Curry 1994). For instance, there is evi- strandings until 1963. Between 1963 and 2004, three to
dence that harbor seals habituate to AHDs, partly because ten mass strandings of Cuviers beaked whales were re-
they modify their swimming behavior to keep their heads ported per decade (Fig. 1) although improved reporting may
out of the water when they are in high-intensity sound fields be a factor in explaining the increasing number of mass-
(Mate and Harvey 1987). Likewise, harbor porpoises have stranding events detected in recent decades.
been shown to habituate to gillnet pingers over a span of 10 The first published suggestion of a connection between
or 11 days (Cox et al. 2002). beaked whale strandings and naval activity was by Sim-
Observations of responses to whale watching and other monds and Lopez-Jurado (1991). They described a set of
vessels also suggest some level of habituation to noise. Near three multianimal strandings associated with naval activity
Cape Cod, common minke whales (Balaenoptera acutoros- in the Canary Islands in 1985, 1988, and 1989, and additional
trata) changed from being attracted to vessels to appearing incidents of beaked whale mass strandings in the Canary
generally uninterested, fin whales from flight reactions to Islands were noted in 1986 and 1987. These authors did not
disinterest; and humpback whales from mixed, but usually posit a connection between beaked whale mass strandings
strongly negative, to strongly positive reactions (Watkins and the use of ASW sonar, but rather related them to the
1986). At San Ignacio Lagoon, Baja California, gray whales nearby presence of naval operations.
become less likely to flee from whale-watching boats as the The increased incidence of multianimal beaked whale
season progresses ( Jones and Swartz 1984). stranding events can be correlated with the advent of mid-
Habituation does not signify that hearing loss or injury frequency ASW sonar. Prototypes of hull-mounted ASW
from high-intensity sounds has not occurred. Humpback sonars (e.g., SQS-23 and 26) were first tested in 1957 (Gerken
whales in Newfoundland remained in a feeding area near 1986) and were deployed on a broad range of naval ships
where there was seafloor blasting (Todd et al. 1996). Re- (frigates, cruisers, and destroyers) belonging to the United
ceived sound pressure levels at 1 mi from the explosions States and other nations beginning in the early 1960s. This
were typically 145150 dB re 1 Pa at 240450 Hz, with pre- timing coincides with increased reports of mass strandings
sumed source levels of 295300 dB re 1 Pa at 1 m based on of Cuviers beaked whales (Fig. 7.1). Eleven out of thirty-
the size of the explosive charges. The whales showed no two of the documented mass strandings of these whales
clear reaction to the blasting in terms of behavior, move- have been associated with concurrent naval activities. Ef-
ment, or residence time. However, increased incidental en- forts to record marine mammal strandings worldwide have
trapment in nets followed the blast exposure (Todd et al. been intensified during the past few decades, so, again,
1996). In addition, two whales were found dead after a greater efficiency of reporting may be a factor in the in-
5,000-kg explosion, and examination of the temporal bones creased numbers recorded.
of their inner ears revealed significant blast trauma Ketten An examination of the circumstances surrounding these
et al. 1993). This incident highlights the difficulty of using mass strandings may help to define the association with the
overt behavioral reactions to monitor the effects of noise or occurrence of high-intensity sound. Two such strandings
high-intensity sound on marine mammals. have been documented by detailed investigative reports: the
Kyparissiakos Gulf, Greece, incident of May 1996 (DAmico
and Verboom 1998) and the Bahamas incident of March
INCIDENTS OF MASS 2000 (Evans and England 2001). Examination of other
STRANDING ASSOCIATED WITH beaked whale mass strandings provides additional perspec-
HIGH-INTENSITY SOUND tive on the diversity of sound sources, environment, and
conditions associated with these events.
Multiple-animal strandings (mass strandings) have been
associated with the use of high-intensity sonar during naval
Kyparissiakos Gulf, Greece, May 1996
operations and with the use of airguns during seismic re-
flection profiling. A key characteristic of these incidents is Frantzis (1998, 2004) first drew attention to a mass strand-
that they predominantly involved beaked whales, particu- ing of Cuviers beaked whales in the Ionian Sea that coin-
Reynolds.100-123 6/10/05 4:54 PM Page 118

118

Table 7.5 Strandings of at least two Cuviers beaked whales, from Smithsonian records

Year Location Species (numbers) Correlated activity

1914 United States (New York) Zc (2)

1963 Italy Zc (15+)
1965 Puerto Rico Zc (5)
1968 Bahamas Zc (4)
1974 Corsica Zc (3), Stenella coeruleoalba (1) Naval patrol
1974 Lesser Antilles Zc (4) Naval explosion
1975 Lesser Antilles Zc (3)
1980 Bahamas Zc (3)
1981 Bermuda Zc (4)
1981 United States (Alaska) Zc (2)
1983 Galapagos Zc (6)
1985 Canary Islands Zc (12+), Me (1) Naval maneuvers
1986 Canary Islands Zc (5), Me (1)
1987 Canary Islands Zc (group), Me (2)
1987 Italy Zc (2)
1988 Canary Islands Zc (3), Me (1), Hyperoodon ampullatus (1), Kogia breviceps (2) Naval maneuvers
1989 Canary Islands Zc (19+), Me (2), Md (3) Naval maneuvers
1991 Canary Islands Zc (2) Naval maneuvers
1991 Lesser Antilles Zc (4)
1993 Taiwan Zc (2)
1994 Taiwan Zc (2)
1996 Greece Zc (14) Navy LFAS trials
1997 Greece Zc (3)
1997 Greece Zc (8)
1998 Puerto Rico Zc (5)
2000 Bahamas Zc (9), Md (3), unid.[AQ6] ziphiids (2), Balaenoptera
acutorostrata (2), Stenella frontalis (1) Naval maneuvers
2000 Galpagos Zc (3) Seismic airgun
2000 Madeira Zc (3) Naval maneuvers
2001 Solomon Islands Zc (2)
2002 Canary Islands Zc (7), Me (2), Md (1), unid. ziphiids (9) Naval maneuvers
2002 Baja California Zc (2) Seismic airgun
2004 Canary Islands Zc (2) Naval maneuvers

Source: J. Mead (pers. comm.), with updates by the author.

Note: Zc, Cuviers beaked whale. Other beaked whales that stranded during these events included Gervais beaked whale, Mesoplodon europaeus (Me), and
Blainvilles beaked whale, Mesoplodon densirostris (Md).

cided with tests of ASW sonar by the North Atlantic Treaty beamwidths of about 20. Three source tows of about 2 h
Organization (NATO). Twelve of these animals stranded duration were conducted each day, and the beaked whale
along 38 km of coastline on 1213 May 1996; another strandings occurred nearest in time to the first two source
stranded along the same coastline about 20 km to the north runs of 12 May (runs 9 and 10; DAmico and Verboom 1998)
on 16 May and was driven back out to sea. Two weeks later, and the last two source runs of 13 May (runs 13 and 14;
one more animal was found decomposing on a remote DAmico and Verboom 1998). Sound propagation modeling
beach on the neighboring Zkinthos Island, located north- suggests that sound pressure levels only exceeded 190 dB re
east of the strandings on the mainland. Twelve of these 1 Pa at ranges of less than 100 m. The sound levels present
fourteen animals stranded alive, with no apparent disease or broadly throughout the Kyparissiakos Gulf are thought to
pathogenic cause. These strandings coincided with a 4-day have been in the range of 140160 dB re 1 Pa (DAmico and
period (1216 May) when the vessel NRV Alliance was tow- Verboom 1998).
ing an acoustic source in the vicinity, primarily at depths The association in space and time between stranding lo-
between 70 and 85 m. The source generated both low- cations and acoustic source tracks suggests that the animals
and mid-frequency sound at source levels of 226 dB re 1 Pa were affected by the ASW sonar (DAmico and Verboom
at 1m. The transmitted low-frequency signal included a 2-s 1998). Figure 7.2 shows the acoustic source tracks and
upsweep at 450650 Hz, and a 2-s continuous tone at 700 stranding locations for 12 and 13 May. There is a general cor-
Hz. The mid-frequency signal included a 2-s upsweep at relation between the offshore source track locations and the
2.83.2 kHz and a 2-s tone at 3.3 kHz. Both frequencies inshore stranding locations. The 13 May source track is
were projected as horizontally directed beams with vertical shifted northward from the 12 May track, and likewise some
Reynolds.100-123 6/10/05 4:54 PM Page 119

Impacts of Anthropogenic Sound 119

Bahamas, 1516 March 2000

Sixteen cetaceans were found stranded along the Provi-
dence Channel in the Bahama Islands during a 2-day period
in March 2000, and the episode was correlated with a U.S.
Navy training exercise using ASW sonar (Evans and Eng-
land 2001). The stranded animals were predominantly
beaked whales (seven Ziphius cavirostris, three Mesoplodon
densirostris, and two ziphiid sp.), although two minke whales
were among the animals that live-stranded. One Atlantic
spotted dolphin (Stenella frontalis) stranded at a somewhat
distant location and is thought to have died of unrelated
causes. Eight of the beaked whales died and the remaining
animals were refloated, their fates remaining unknown
(Balcomb and Claridge 2001). Tissue samples were col-
lected from five of the dead beaked whales. Gross necropsy
results suggested that all five were in good body condition;
Fig. 7.1. Reported Cuviers beaked whale mass stranding events tabulated
by decade. Total numbers of reported events per decade (gray) are none showed evidence of debilitating disease. Hemor-
shown, along with the events documented in association with naval rhages were found in the acoustic fats of the head, the inner
activities or airgun usage (black). Incidence of mass-stranding events ears, and spaces around the brain, with no evidence of ex-
increased following the deployment of mid-frequency ASW sonar
beginning in the 1960s (dashed line), although the level of effort ternal blunt-force trauma. The pattern of injury in the fresh-
dedicated to reporting has also increased during this period. est specimens suggested that the ears were structurally in-
tact and that the animals were alive at the time of injury
(Ketten et al. 2004).
Five Navy ships were operating hull-mounted ASW
sonars in the area, four of which were described in a pre-
of the 13 May stranding locations are located farther north. liminary report (Evans and England 2001). Of the four ships
Correlation of stranding times and source track locations described, two operated SQS-53C hull-mounted sonars and
for 12 May suggests that at least three of the six animals with two used SQS-56 hull-mounted sonars (Watts 2003). The
known stranding times were affected by the 06000800 h former were operated at 2.6 and 3.3 kHz with a source level
source tow as their stranding times precede the 11001300 of 235 dB re 1 Pa at 1 m or higher and ping lengths of 0.52
h source tow. Assuming that they were near the source s alternating between tones and FM sweeps. The latter were
when they were exposed to the sound (and therefore ex- operated at 6.8, 7.5, and 8.2 kHz at 223 dB re 1 Pa at 1 m.
posed at levels above 190 dB re 1 Pa), their swimming dis- Integrated sound exposures of 160165 dB re 1 Pa for
tances to reach the shore would be approximately 30 nmi, 50150 s would have been experienced in near-surface wa-
covered at speeds of approximately 10 knots. Alternatively, ters (15 m depth) throughout much of the Providence
they might have been exposed at locations inshore relative Channel on 15 March 2000 (Evans and England 2001). Peak
to the source, which would suggest lower sound exposure sound pressure levels above 185 dB re 1 Pa would have
levels (140160 dB re 1 Pa) but shorter swimming distances been experienced only within a few hundred meters of the
and speeds. The two 12 May afternoon stranding locations ship tracklines along the central portion of the channel.
with known times likewise occurred at swimming distances The association in space and time between the stranding
of 2030 nmi from the source track locations. locations and acoustic source tracks shown in Figure 7.3 is

38 38
Latitude

Latitude

May 12 May 13

190 06000 Fig. 7.2. Kyparissiakos Gulf, Greece, Cuviers

0600

0 1100 beaked whale (Ziphius cavirostris) mass

37.5 37.5 1630 stranding locations and acoustic source tracks
110 160 for 12 May and 13 May 1996. Times for the
0 0 source tracks and strandings are indicated in
GMT. The stranding locations for each day are
130 indicated by black dots on the respective maps.
0
0800

For 13 May, strandings occurring on the

previous day are indicated by white dots;
strandings occurring after 13 May are shown
21 21.5 21 21.5 by gray dots (DAmico and Verboom 1998,
Longitude Longitude Frantzis 2004).
Reynolds.100-123 6/10/05 4:54 PM Page 120

120

The highest sound exposures would have been experi-

March 15
enced by animals distributed at locations along the source
tracks, suggesting that following exposure they might have

0700
swum toward the stranding sites 1030 nmi distant. How-
nd
a Isla ever, data on beaked whale distribution (K. Balcomb and D.

Abaco
27 ham
d Ba
Gran Claridge pers. comm.) suggest that Mesoplodon densirostris is
afternoon 00930
1200
1430 found predominantly along the margins of the channel, in
1230
Latitude

Island
0
waters about 500 m deep, and that Z. cavirostris is also found
along the channel margins, but in deeper waters. Combin-
081
1100 0835

73
0815
0730
ing the source modeling and the animal distribution data
1100
C D suggests that sound at moderate exposure levels (e.g.,
150160 dB re 1 Pa for 50150 s) would have been received

B 1730
0
26
A 1200
0 at the most likely locations of beaked whales in Providence
Channel (Evans and England 2001).

Longitude 78 77
Madeira, May 2000
Fig. 7.3. Bahamas stranding locations and acoustic source tracks for 15
March 2000. Stranding locations (asterisks) are shown, along with known Three Cuviers beaked whales stranded in May 2000 on the
stranding times (all times are given as local). Tracks for four U.S. Navy Madeira Archipelago in the northeastern Atlantic (Freitas
ships are shown moving from east to west through the Providence
Channel. Ship tracks are shown as broken lines (ships are given
2004). The area south of Madeira Island, specifically the
designations A, B, C, and D following Evans and England 2001). Arrows deep channel between Madeira and Porto Santo islands, is a
along each ship track designates 0730 h, the time of the first recorded known location for Ziphius cavirostris sightings. The animals
stranding, located at the south end of Abaco Island. Ships B and C entered
the channel near Abaco before midnight on 15 March, whereas ships A
stranded on 9, 13, and 14 May: two subadults (one male, one
and D entered the channel at about 0400 h (Evans and England 2001). female) and a female of unknown age. The two subadults
were examined and found to have eye hemorrhages, pleural
hemorrhages, and lesions of the lungs (Freitas 2004). It was
concluded that they had stranded while still alive. The third
compelling evidence that these animals were affected by the animal was found in an advanced state of decomposition
high-intensity sound sources (Evans and England 2001). A and was not examined in detail. A NATO exercise was sig-
few hours before midnight on 15 March, two source ships naled by the presence of naval vessels and aircraft in the
(designated B and C) entered the Providence Channel off deepwater channel between the islands during the period
the southwest end of Abaco Island and moved through the 914 May. The exercise was reported to have involved one
channel toward the west. Both ships were using active sonar aircraft carrier, three submarines, and more than forty sur-
as they entered the channel (ship B 53C sonar, and ship C 56 face vessels. Details of the acoustic sources in use during the
sonar) with pulse repetition rates of about 24 s. The second exercise have not been made available.
pair of source ships (A with 53C sonar, and D with 56 sonar)
entered the channel at about 0400 h on 15 March, again
Canary Islands, 24 September 2002
from the east and moving west.
A cluster of strandings occurred at the south end of A mass stranding of fourteen to nineteen beaked whales oc-
Abaco Island during the morning of 15 March, with the first curred on the Canary Islands of Fuerteventura and Lan-
recorded stranding at 0730 h. These strandings appear to zarote, associated with naval maneuvers by Spain and other
have occurred at the same time or soon after the second pair NATO countries on 2425 September 2002. The stranded
of source ships passed through the channel south of Abaco, animals included seven Ziphius cavirostris, two Mesoplodon
and about 8 h after the first pair of source ships passed this europaeus, and one M. densirostris. On 24 September a total
point in the channel. These standings occurred at minimum of fourteen animals were found stranded; five were dead,
ranges of 1030 nmi from the ships closest points of ap- three were alive and subsequently died, and six were pushed
proach. During the late morning, the source ships moved back to sea. Five more animals were found dead and in a de-
northwestward, approaching Grand Bahama Island, and a composed state between 25 and 28 September. It is possible
cluster of noon and afternoon strandings occurred on the that these included animals that had been pushed out to sea
south coast of Grand Bahama Island, again with minimum and then had stranded again. Preliminary necropsy results
source-to-shore ranges of 2030 nmi. Balcomb and Claridge for six of the beaked whales suggest that they were healthy.
(2001) noted that individual Cuviers beaked whales that The strandings occurred at dawn or in the early morning,
had been identified photographically in this region previ- and the animals that were found alive all appeared disori-
ously have not been sighted subsequent to the stranding ented. Those that were found dead had been feeding shortly
event, suggesting that the beaked whale mortality was prior to stranding (Martn et al. 2004).
higher than simply the number of whales that were known Necropsies and dissections (Fernndez 2004) revealed no
to have mass stranded. visible signs of traumatic lesions physically caused by ship
Reynolds.100-123 6/10/05 4:54 PM Page 121

Impacts of Anthropogenic Sound 121

strikes, fishing activities, or blunt trauma generally. The Doherty Earth Observatory (Malakoff 2002, Taylor et al.
stomach contents, and their freshness and digestive status, 2004). On 24 September at about 14001600 h local time
indicated that there was only a short period between the on- (21002300 h GMT), fishermen discovered two live-
set of illness and death. stranded whales and unsuccessfully attempted to push
Examination of these animals heads and bodies revealed them back out to sea ( J. Urbn-Ramrez pers. comm.). A
hemorrhages and bubbles (Fernndez 2004). Hemorrhages group of marine biologists found the whales dead on 25
were observed along acoustic paths in the head and in the September (B. Taylor pers. comm.). By 27 September, when
brain and spinal cord. The hemorrhagic areas observed one carcass was necropsied, the advanced state of decom-
macroscopically in the acoustic fat were also demonstrated position precluded determination of the cause of death.
histologically. All of the animals were bleeding profusely On 24 September the R/V Ewing had been firing an array
from the eyes and there was evidence of multifocal pe- of twenty airguns with a total volume of 8,500 in.3. Such an
techial (pinpoint) hemorrhages. Fat embolisms were ob- array is expected to have an effective broadband source level
served, which could have been responsible for hemorrhages of 256 dB peak re 1 Pa at 1m, with maximum energy at fre-
in the macrovascular system. Focal hemorrhages were quencies of 4090 Hz. A later attempt to directly measure
found in the dura mater membrane, and there was a large the array source level (Tolstoy et al. 2004) was unsuccessful
quantity of blood in the subarachnoid space around the owing to equipment malfunction and ambiguities in con-
cranial spinal cord. A generalized congestion of the blood verting the pulsed signals into RMS pressure values. Source
vessels in the brain was seen in all the fresh animals, and levels of airgun arrays at middle frequencies (15 kHz) are
multifocal subarachnoid hemorrhages were detected. Addi- thought to be diminished from levels at low frequencies by
tionally, in the tissues that were fresh, empty spaces and 2040 dB (Goold and Fish 1998, Tolstoy et al. 2004). The
bubbles were seen inside the blood vessels. In sections of Ewing airguns were fired with a repetition rate of approxi-
the brain, multifocal petechial hemorrhages were located mately 20 s (50 m distance between shots). Figure 7.4 shows
mainly in the white matter. All the lungs presented general the Ewing track for 2425 September; the ship was on a tran-
diffuse congestion, some subpleural hemorrhages, and alve- sect line directed toward the stranding site and reached the
olar edema. The kidneys were enlarged, with marked vas- closest point of approach (18 nmi) at 1430 h local time (2130
cular congestion and hemorrhages in the capsular and in- h GMT). These animals would have received the highest
terstitial areas. Degeneration (in vivo) of vestibucochlear sound pressure levels if they were exposed at locations near
portions of the ear was noted. Although the exact physical the source tracks. Then, following exposure, they might
mechanism for these injuries is not known, several hy- have swum toward the stranding site 2030 nmi distant. Al-
potheses currently under investigation are focused on ternatively, they might have been exposed at lower source
nonauditory acoustic effects ( Jepson et al. 2003). levels at locations nearer the stranding site.
The strandings occurred along the southeastern coast of
the islands of Fuerteventura and Lanzarote. At the time of
Summary of Beaked Whale Stranding Events
the 2425 September strandings, ten NATO countries (Ger-
many, Belgium, Canada, France, Greece, Norway, Portugal, The mass strandings of beaked whales following exposure
Turkey, the United Kingdom, and the United States) to sound from sonar or airguns present a consistent pattern
were conducting a multinational naval exercise (known as of events. Cuviers beaked whales are, by far, the most com-
NEOTAPON[AQ3] 2002); however, the acoustic sources monly involved species,; making up 81% of the total num-
employed during the exercise are not known at this time. ber of stranded animals. Other beaked whales (including
The participating countries include ASW sonar in their ca- Mesoplodon europaeus, M. densirostris, and Hyperoodon ampul-
pabilities although the details of their systems vary (Watts latus) account for 14% of the total, and other cetacean
2003). Common features of the sonars that may have been species (Stenella coeruleoalba, Kogia breviceps, and Balaenoptera
used in this exercise include high-amplitude source levels acutorostrata) are sparsely represented. It is not clear
(SPL > 223 dB RMS re 1 Pa at 1 m), periodic (1560 s) rep- whether: (a) Ziphius cavirostris are more prone to injury
etition, pulses (up to 4 s), with significant energy at mid- from high-intensity sound than other species, (b) their be-
dle frequencies (310 kHz), and formed into horizontally di- havioral response to sound makes them more likely to
rective beams (Watts 2003). Eight mass strandings of Z. strand, or (c) they are substantially more abundant than the
cavirostris have been documented in the Canary Islands other affected species in the areas and times of the expo-
since 1985, and naval exercises have been recorded as asso- sures leading to the mass strandings. One, two, or three of
ciated with five of them (Table 7.5; Martn et al. 2004). these possibilities could apply. In any event, Z. cavirostris has
proven to be the miners canary for high-intensity sound
impacts. The deployment of naval ASW sonars in the 1960s
Gulf of California, 24 September 2002
and the coincident increase in Z. cavirostris mass strandings
A stranding of two Ziphius cavirostris occurred on 24 Sep- suggest that lethal impacts of anthropogenic sound on
tember 2002 on Isla San Jose in the Gulf of California, Mex- cetaceans have been occurring for at least several decades.
ico, coincident with seismic reflection profiling by the R/V The settings for these strandings are strikingly consistent:
Maurice Ewing operated by Columbia University, Lamont- an island or archipelago with deep water nearby, appropri-
Reynolds.100-123 6/10/05 4:54 PM Page 122

122

Isla San Jose For instance, in the Bahamas, the most likely exposure lev-
els appear to have been 150160 dB re 1 Pa for 50150 s, or

Se

04
00
pt
25 less, well below the level expected to create hearing loss in

25
,0
odontocetes. Given that damage to hearing appears un-

00
0
2200 likely, other mechanisms are needed to explain the connec-
tion between sound exposure and stranding in beaked

00
20
whales.
Latitude

00
16
RESEARCH PRIORITIES FOR SOUND
Ba

EFFECTS ON MARINE MAMMALS

00
ja

12
Ca
lifo

A decade has passed since the National Research Council

rn

(1994) outlined a set of research priorities for understanding

ia

24 the effects of noise on marine mammals. In most of the areas

outlined for study, a basic understanding is still lacking. Many
of the same research priorities were reiterated by two sub-
249 Longitude 250 sequent National Research Council (2000a, 2003b) reports.
The need to study the impacts of noise in the field rather
Fig. 7.4. Gulf of California Ziphius cavirostris stranding location (black
dot) and best estimate for the time of stranding (2200 h GMT), along
than in captive settings means that a clear understanding
with the R/V Ewing track for 2425 September 2002. Times along the may not become available for many years. There is also the
ships track are indicated in GMT. The ships closest point of approach to need to differentiate the effects that are significant for indi-
the stranding site was 18 nmi.
vidual animals from those that are significant on a popula-
tion level. Addressing population-level impacts requires ob-
servations that are distributed in space and time and large
ate for beaked whale foraging habitat. The conditions for numbers of observations to provide statistical power.
mass stranding may be optimized when the sound source
transits a deep channel between two islands, such as in the Priority 1: Understand, in detail, the causes of mass strandings
Bahamas, and apparently in the Madeira incident. When ex- of beaked whales. When exposed to high-intensity sound,
posed to these sounds, some beaked whales swim to the some beaked whales strand and die. Understanding the
nearest beach. The animals appear on the beach not as a causes and consequences of beaked whale mass stranding
tight cluster of individuals but rather distributed over miles represents the highest research priority for marine mam-
of coastline. Such scatter in the distribution of stranding lo- malogists studying the conservation implications of expo-
cations is an important characteristic, which has resulted in sure to sound. The sound levels implicated in these events is
these events being called atypical mass strandings probably not sufficient to cause permanent or temporary
(Frantzis 1998, 2004, Brownell et al. 2004). The stranded an- hearing threshold shifts. What is the mechanism for damage
imals die if they are not returned to the sea by human in- or disturbance? The behavioral reaction is swift and vigor-
tervention, and the fate of the animals that are returned to ous on an individual level. The lack of close animal cluster-
the sea is unknown. Necropsies of stranded animals suggest ing on the beach suggests little or no social component to
internal bleeding in the eyes, ears, and brain, as well as fat these strandings, yet the potential for large numbers of an-
embolisms. imals to strand suggests significance at a population level.
The implicated sounds involve pulses with high-intensity What are the source characteristics that lead to damage? Is
source levels (235 dB re 1 Pa at 1 m) from sonar or airgun low-frequency sound (the primary energy component of
arrays. Middle frequencies (16 kHz) are clearly implicated airguns) as damaging as mid-frequency sound (used by SQS-
in the sonar-induced stranding incidents. It is unclear 53 ASW tactical sonars)? What sound pressure exposure
whether low-frequency sound also has the potential of caus- level creates damage or disturbance? Beaked whale mass
ing injury to beaked whales. Although airguns create pre- stranding events make it clear that high-intensity anthro-
dominantly low-frequency energy, they may also have am- pogenic sound is a threat to at least some marine mammals,
ple mid-frequency energy. The actual sound exposure levels yet key parameters about beaked whale strandings must
received by animals that later strand are unknown although be understood before we can predict the impacts of high-
in the best-documented events these levels may be bounded intensity sound on other species in other settings.
by careful sound propagation modeling and by knowledge
of where the animals are most likely to be found. Source Priority 2: Determine behavioral responses to anthropogenic
levels high enough to create permanent or temporary hear- sound. A key impediment to assessing the biological effects
ing loss would be experienced only at ranges close to the of ocean noise is the paucity of knowledge about marine
source (< 1 km). The sound exposures calculated for sites of mammal behavior and specifically the lack of understanding
most likely animal presence appear to be significantly lower. of their behavioral responses to anthropogenic sound. Be-
Reynolds.100-123 6/10/05 4:54 PM Page 123

Impacts of Anthropogenic Sound 123

havioral data must be collected in the wild to provide a basis bers of individuals will probably never be maintained in cap-
for understanding potential effects. Significant effects of tivity for study of their physiology, so tools are needed to
ocean noise may be confined to a few individuals exposed to study marine mammal physiology in the wild. For instance,
high sound pressure levels, or they may extend to entire pop- indicators of stress may be used to assess the impact of an-
ulations as a result of widespread exposure. Controlled ex- thropogenic noise. If stress factors can be measured from
posure experiments might be helpful in defining obvious or blubber or blood samples, perhaps biopsy or other tissue
short-term effects on individuals but may not reveal long- samples collected in the wild could reveal regional or popu-
term impacts. Discerning population-level effects is chal- lation-wide stress levels associated with noise. Moreover,
lenging as the observations must be conducted over long dis- without a field method that can be rapidly deployed to de-
tances and extended periods of time, and there are many termine hearing capabilities, it is difficult to collect audio-
confounding influences. Relating migration and movement metric data on all marine mammal species and under the full
to noise level is one potential approach. Do marine mam- range of conditions where chronic noise may have degraded
mals systematically avoid habitat areas with high noise lev- hearing capabilities. The ability to collect audiometric data
els? More subtle behavioral changes may be associated with on a beached or net-entangled animal is a first step and will
exposure to high ambient ocean noise. Is natural sound (e.g., be especially useful when high-intensity sound is suspected
snapping shrimp) useful for prey location? A better under- of having played a role in the animals stranding.
standing of how and why marine mammals make and use
sound would greatly aid our ability to predict how ocean Priority 5: Characterize and monitor marine mammal popula-
noise might be disruptive to marine mammal behavior. The tions in areas of high-intensity sound generation. High-intensity
sound avoidance response has been exploited to exclude ma- anthropogenic ocean sound sources are primarily concen-
rine mammals from fish pens and fishing operation areas us- trated in well-defined zones: (a) at major commercial ports
ing acoustic harassment devices. However, does their be- and along shipping lanes, (b) within military test and training
havioral response take place before some hearing loss sites, and (c) within regions of oil exploration and develop-
occurs? Our ignorance about marine mammal behavioral re- ment. The marine mammal populations that inhabit or move
sponses to sound is abysmal, and knowledge of this subject through those zones should be characterized and monitored.
must be improved in the face of rising ocean noise levels. A combination of visual and acoustic monitoring may be
necessary for efficient assessment of marine mammal distri-
Priority 3: Improve tools for assessing and measuring the be- butions, ambient noise, and anthropogenic sound. Such
havior of marine mammals. Better research tools are needed monitoring data will help determine whether noise is a fac-
to observe marine mammal behavior in the wild. Such tools tor in discouraging habitat use by marine mammals.
are needed both to characterize normal behavior and to de-
tect changes in behavior associated with anthropogenic
sound. Acoustic recording tags are important for detailed
behavioral studies in the presence of sound. Technical im- ACKNOWLEDGMENTS
provements are needed over current tags to increase their
duration of attachment, to expand the volume of stored I thank Randy Reeves, Suzanne Montgomery, Tim Ragen,
data, and to enhance the suite of available sensors. Im- David Cottingham, John Reynolds, and others at the Marine
provements to the tag attachment system are particularly Mammal Commission for encouraging me to write this
needed for large cetaceans, which cannot be captured for chapter and for their tireless editorial efforts. I also thank
tagging. Current noninvasive attachments have limited du- Dave Mellinger, Doug Nowacek, Bill Perrin, Ted Cranford,
ration, whereas invasive attachments may involve both dis- Sue Moore, Mark McDonald, and Sean Wiggins for helpful
turbance and injury to the animal. Technology for passive reviews and comments. I thank Gerald DSpain for help
acoustic tracking is another important component of be- with the ocean noise budget calculation. My understanding
havioral study with the potential for improvement. of this topic was aided by long discussions with Jack Cald-
well, Tara Cox, Gerald DSpain, Roger Gentry, Bob Gisiner,
Priority 4: Develop tools to study marine mammal physiology in Mardi Hastings, Lee Langstaff, Rodger Melton, Paul Nachti-
the wild. For many species of marine mammal, large num- gall, Naomi Rose, Peter Tyack, and Erin Vos.

[AQ1] OK?
[AQ2] Update?
[AQ3] OK?
[AQ4] OK?
[AQ5] OK?
[AQ6] Please expand abbreviation here and below.