Access to this full-text is provided by PLOS.
Content available from PLOS One
This content is subject to copyright.
RESEARCH ARTICLE
Temperate freshwater soundscapes: A
cacophony of undescribed biological sounds
now threatened by anthropogenic noise
Rodney A. RountreeID
1,2
*, Francis Juanes
2
, Marta Bolgan
3
1The Fish Listener, East Falmouth, Massachusetts, United States of America, 2Biology Department,
University of Victoria, Victoria, British Columbia, Canada, 3Laboratoire de Morphologie Fonctionnelle et
Evolutive, Institut de Chimie, Universite
´de Liège, Liège, Belgium
*rrountree@fishecology.org
Abstract
The soundscape composition of temperate freshwater habitats is poorly understood. Our
goal was to document the occurrence of biological and anthropogenic sounds in freshwa-
ter habitats over a large (46,000 km
2
) area along the geographic corridors of five major
river systems in North America (Connecticut, Kennebec, Merrimack, Presumpscot, and
Saco). The underwater soundscape was sampled in 19 lakes, 17 ponds, 20 rivers and 20
streams, brooks and creeks that were grouped into broad categories (brook/creek, pond/
lake, and river). Over 7,000 sounds were measured from 2,750 minutes of recording in
173 locations over a five-week period in the spring of 2008. Sounds were classified into
major anthropophony (airplane, boat, traffic, train and other noise) and biophony (fish air
movement, also known as air passage, other fish, insect-like, bird, and other biological)
categories. The three most significant findings in this study are: 1) freshwater habitats in
the New England region of North America contain a diverse array of unidentified biological
sounds; 2) fish air movement sounds constitute a previously unrecognized important
component of the freshwater soundscape, occurring at more locations (39%) and in equal
abundance than other fish sounds; and 3) anthropogenic noises dominate the sounds-
cape accounting for 92% of the soundscape by relative percent time. The high potential
for negative impacts of the anthropophony on freshwater soundscapes is suggested by
the spectral and temporal overlap of the anthropophony with the biophony, the higher
received sound levels of the anthropophony relative to the biophony, and observations of
a significant decline in the occurrence, number, percent time, and diversity of the bioph-
ony among locations with higher ambient received levels. Our poor understanding of the
biophony of freshwater ecosystems, together with an apparent high temporal exposure to
anthropogenic noise across all habitats, suggest a critical need for studies aimed at identi-
fication of biophonic sound sources and assessment of potential threats from anthropo-
genic noises.
PLOS ONE
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 1 / 26
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Rountree RA, Juanes F, Bolgan M (2020)
Temperate freshwater soundscapes: A cacophony
of undescribed biological sounds now threatened
by anthropogenic noise. PLoS ONE 15(3):
e0221842. https://doi.org/10.1371/journal.
pone.0221842
Editor: Dennis M. Higgs, University of Windsor,
CANADA
Received: August 16, 2019
Accepted: February 14, 2020
Published: March 18, 2020
Copyright: ©2020 Rountree et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the manuscript and its Supporting
Information files.
Funding: This work was funded by The Eppley
Foundation for Research to RR. The funder had no
role in study design, data collection and analysis,
decision to publish, or preparation of the
manuscript.
Competing interests: The authors have declared
that no competing interests exist.
Introduction
Fish sounds were first scientifically studied in North America by Abbot [1] who in 1877
lamented that “the little fishes of our inland brooks and more pretentious denizens of our
rivers are looked upon as voiceless creatures. . .” and concluded that “certain sounds made
by these fishes are really vocal efforts. . ..”. Almost 100 years later Stober [2] was the first to
describe the underwater noise spectra in relation to freshwater fish sounds in any freshwater
habitat in an effort to determine if the sound of streams entering a lake could be used as a
homing cue for cutthroat trout. He also pointed out the critical lack of data on the ambient
noise and fish sounds in freshwater systems. Fortunately, after long neglect there has been a
recent surge in interest in the potential impacts of anthropogenic noise on freshwater ecosys-
tems [see reviews in 3–9], but efforts to understand such impacts are hampered by the pau-
city of data on the natural soundscape composition [9]. The need for research on the sound
production of freshwater fishes was highlighted in our recent review of the literature, which
found that sounds have been reported in only 87 species in North America and Europe, but
detailed descriptions of sound characteristics are known for only 30 species [9].
Acoustic investigations of freshwater soundscapes can be grouped (S1 Table and citations
therein) into those that seek to characterize the ambient noise [2,10–20], those that focus
on the biophony (without reporting on anthropophony) [21–27], those that focus on noise
impacts on fishes and other organisms (see reviews in [8,28–30]), and those that include
some information on both the biophony and anthropophony [31–43]. Most of the latter stud-
ies focus on sound levels and do not provide information on the relative contribution of both
the biophony and anthropophony to the soundscape in terms of percent occurrence, number
of sounds, percent of time occupied, or diversity (except [31,32,35,36,40]S1 Table). No fresh-
water study provides data on the rate of temporal overlap between anthropomorphic and bio-
phonic sounds (i.e., how often do fish sounds overlap with anthropogenic sounds?).
Descriptive studies that characterize the soundscape of freshwater habitat including the rela-
tive contribution of both anthropogenic and biophonic sounds to the soundscape composition
are needed. Furthermore, only a few studies incorporate some level of real-time acoustic moni-
toring into the sampling design (S1 Table, [2,23,27,31]).
Previous studies of the biophonic component of freshwater habitats in the New England
region of North America include species-specific studies [23,44,45] and a soundscape survey
of the Hudson River [31]. Martin and Popper [42] conducted a survey of the noise levels in the
vicinity of the Tappan Zee Bridge in the Hudson River, but did not quantify the occurrence of
biophonic sounds.
The present study is part of a series of studies seeking to document the soundscape compo-
sition of habitats in New England. In a pilot study focusing on fish sounds in the Hudson
River a high diversity of known and unknown biological sounds was discovered [31]. At a loca-
tion in New York City, arguably one of the world’s most impacted locations, a high diversity
of mostly unknown biological sounds was heard at night likely resulting from a combination
of increasing nocturnal sound production and decreasing masking from boat noise [31]. The
value of reporting information on unknown biological sound occurrence was emphasized
after the discovery that one of the unknown sounds in the Hudson River was produced by
the invasive freshwater drum, Aplodinotus grunniens [44].
Following the Hudson River study, this study was conducted to document the soundscape
of a wide range of habitats within the New England region. Because real-time monitoring was
conducted during all recording, it was possible to notice that some of the most common
sounds appeared to be of the little understood air passage type. It was, therefore, necessary to
conduct a series of surveys over the next ten years to identify some of the air passage sounds
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 2 / 26
and to confirm that they could confidently be attributed to fish [45]. Air passage sounds,
which hereafter are referred to as “air movement” sounds, are produced by internal move-
ments of gas to and from the gas bladder, or release of gas through the mouth, gill, or anal vent
in physostomous fishes [45]. The best-known air movement sound is the “fast repetitive tick”
(FRT) type which consists of a broadband high-frequency burst followed by a train of repeat-
ing ticks [45]. This new understanding of air movement sounds occurring in freshwater fish
allowed us to reprocess recordings collected in the 2008 study in order to classify sounds into
broad biological sound types.
The primary goals of this study were to document the occurrence of biological sounds in a
large variety of freshwater habitats over a large geographic area and to determine the relative
contribution of broad categories of both biological and anthropogenic sounds to the overall
aquatic soundscape composition.
Materials and methods
Ethics statement
No ethics statement or permit were required for this non-invasive observational study.
Study area
The underwater soundscape of freshwater habitats was sampled in a roving survey within a
46,000 km
2
region along the corridors of five major rivers in the New England region of North
America over a five-week period from 30 April to 29 May 2008 (Fig 1, all location and meta-
data are provided in S1 Data set). River corridors surveyed included the 653 km Connecticut
River (30 April to 3 May, N = 32), 188 km Merrimack River (12–16 May, N = 43), 270 km
Kennebec River (17–23 May, N = 53), 219 km Saco River (26–28 May, N = 31) and 42 km
Presumpscot River (28–29 May, N = 14). Within these five major rivers, sound recordings
were made from shore within the main stems of each major river from its origin in the moun-
tains or major lakes to its outlet to the sea, except for the Connecticut River where only the
lower 200 km were surveyed (N = 20, 18, 20, 16, and 5 locations for the Connecticut, Merri-
mack, Kennebec, Saco, and Presumpscot, respectively). In addition, sound recordings were
made from a large variety of other habitats including 15 river tributaries (N = 20 locations), 19
lakes (N = 32), 17 ponds (N = 18), and 20 streams, brooks and creeks (N = 24). Access to the
water was often difficult and usually required sampling within 500 m of a bridge (45%) or
dam (12%), and sometimes involved scrambling down steep ravines, over rocks, or short hikes
through woodlands (S1 Data set). All locations were photo-documented. Although both lotic
and lentic habitats sampled represent a continuum, for comparison purposes, small brooks
and streams were combined with small sluggish rivers or creeks into a “brook/creek” habitat
category, small ponds and large lakes were combined into a “pond/lake” category, and tribu-
tary and main stem river sites were combined into a “river” category (S1 Fig,S1 Data set).
A total of 173 sites were successfully sampled during this survey (S2 Table). Sampling was
avoided during poor weather (high winds and/or rain). Sampling occurred during the day
between 0930 and 1900 (N = 148), and at night between 1900 and 2300 h (N = 25). Recording
durations were from 1 to 49 minutes (mean = 12.5 min) during day and 3 to 119 minutes
(mean = 35.5 min) during night sampling (S2 Table). Night recordings were longer on average
than day recordings because we expected increased biological activity around sunset based on
previous experience [13,31–33,44,45].
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 3 / 26
Fig 1. Study area. Locations where sounds were recorded between 30 April and 30 May 2008 (Location coordinates and other
metadata are provided in S1 Data set).
https://doi.org/10.1371/journal.pone.0221842.g001
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 4 / 26
Acoustic data
Acoustic data were captured from a low noise (nominal sensitivity -165 dB, re 1 μPa), broad-
band (flat response deviating less than ±1 dB from 16–44,000 Hz), cylindrical hydrophone
with 30 m of cable (model C54XRS, Cetacean Research Technology (CRT), Seattle, Washing-
ton) at 48kHz (24 bit) with a MOTU Ultralite, bus powered firewire audio interface to a laptop
computer using SpectraPro332 Professional Sound Analysis software (Sound Technology,
Inc.). The entire system including the hydrophone, audio interface, laptop, and SpectraPro
322 software was provided by CRT with preconfigured calibration settings for two fixed
gains (zero and half full-scale on the audio-interface). All recordings were made with the same
equipment and the same two fixed calibrated gain settings. Data were automatically converted
to sound pressure level (SPL dB re 1 μPa) by SpectroPro322 in real-time during field recording.
Sound from either a dubbing microphone or a second hydrophone was captured to a second
channel in the recordings. When used, the second hydrophone was an uncalibrated, variable
(dial) gain Aquarian model (Aquarian Audio Products, Anacortes, WA, USA). All recordings
were monitored aurally and visually in real-time with Spectropro322 displaying the spectro-
gram and waveform in SPL dB re 1 μPa. Written and oral notes on sounds together with
potential environmental, anthropogenic and biological sound sources were also recorded
throughout each recording. Additional post-processing of acoustic signals was conducted by
listening to all recordings in their entirety while simultaneously viewing the spectrogram
(1024 FFT, Hanning window, 50% overlap) and waveform with Raven Pro 1.5 acoustic soft-
ware (Bioacoustics Research Program 2014).
Most anthropogenic sounds (anthropophony) were positively identified in the field, includ-
ing airplane, boat, traffic (bridge crossing or road), train and other noise. Boat sounds were
divided into sounds of running boats, boats at idle, and other boat sounds (engine cranking,
pumps, etc.). Fishing sounds were those made by recreational fly and spin-cast fishers and
included lines hitting the water surface and sinking, lure retrieval, weights dragging on the
bottom, etc. Other noise included miscellaneous construction sounds (e.g., hammer tapping)
lawn mowers, fish/depth finders, etc.
Biological sounds were classified into broad categories on the basis of previous knowledge of
freshwater biophony [13,31–33,44,45] and the real-time field observations. Biological sounds
were classified as fish, insect-like, surface, bird and other biological. Fish sounds were subdivided
into air movement sounds and other fish sounds. Air movement sounds were further divided
into FRT and other air movement sounds. An example of an air movement sound associated
with a fish jump is provided in S1 Audio (including voice notes), more detailed descriptions and
examples of these types of sounds are presented in our companion paper [45]. Other fish sounds
included more conventionally recognized fish sounds such as drumming and stridulation sounds
which can be pulsed or tonal. Bird sounds were recorded underwater from sounds arising above
water (as determined during real-time monitoring), we include them in the underwater bioph-
ony despite their external origin because they are in fact part of the aquatic soundscape in the
same way that many anthropogenic sounds that arise from terrestrial and aerial sources are con-
sidered part of underwater soundscapes. Because we have previously observed that the sounds
made by fishes when they jump or gulp air at the su face can be useful identification markers,
and may potentially have inter- or intra-biological functions [45], we include them in a “Surface”
sound category. Other biological sounds included beaver (tail slaps) and seal sounds, as well as
unknown snapping sounds. Unknown sounds deemed to be biological based on their frequency
and temporal pattern but which could not confidently be placed within one of the biological cate-
gories were also placed in the other biological sound category. The unclassified sound category
included bubble-like and gurgle sounds because a type of air movement sound labeled “gurgle”
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 5 / 26
described for salmonids cannot yet be reliable distinguished from natural sounds of gas release
from the sediment [45].
Aggregate sound categories included air movement (FRT + other air movement), fish (air
movement + other fish), biophony (fish + insect-like + bird + surface + other biological), boat
(running boat + idle boat + other boat), anthropophony (airplane + boat + fishing + traffic
+ train + other noise), and unclassified (gurgle + bubble-like + unknown).
Within each recording, unique biological sound types were annotated and subjectively
labeled (e.g., FRT-A, FRT-B) and the number of unique biological sound types counted (i.e.,
biological sound diversity per recording). However, while biological sound categories were
consistent across all recordings, individual sound types within each biological category were
not, due to the high variability of sound characteristics, and diversity of habitats sampled, so
total biological sound diversity could not be determined. Many sounds were unique to specific
locations. For example, a sound labeled a “bark” and placed in the other fish category in one
recording might have different acoustic characteristics from a “bark” in another recording,
making it difficult to determine if the observed differences were due to variability or different
sources. However, in that case the sound could still be confidently placed in the “other fish”
category. In addition, two 5 s clips from each recording were annotated to represent the back-
ground ambient noise, herein defined as the background sound when no individually recog-
nizable biological or anthropogenic sounds were observed [sensu 12]. Ambient noise includes
relatively constant acoustic energy, but no individually recognizable sound, from the geoph-
ony, biophony and anthropophony.
Acoustic measurements of all sounds were made in Raven of selected parameters includ-
ing duration, peak frequency and frequency bandwidth [46], and data pooled within the
appropriate sound type category. The “percent frequency of occurrence” of sounds among
recordings was determined as the number of recordings containing a sound type, divided by
the total number of recordings and should not be confused with acoustic frequency parame-
ters. To partially account for bias due to recording length, accumulation curves were calcu-
lated for each location as the number of sound types plotted against time elapsed from the
beginning of each recording. Because the accumulation rates were highly variable among
locations and did not exhibit an asymptote, it was necessary to use only portions of record-
ings that were comparable among locations to compute diversity. Thus, a standardized
diversity was calculated as the number of sound types observed up to 348 s after dropping
recordings with less than 174 s total duration (N = 165). Percent occurrence calculated from
all recordings were similarly compared against a standardized recording length, but because
trends and statistical tests were the virtually the same, we report percent of occurrence based
on all recordings herein.
Sound rate was calculated as the number of sounds per minute in each recording for each
sound category. Percent recording time for each category was calculated by summing the dura-
tions of all individual sounds in the category in a recording and dividing by the recording
duration. In order to examine time-of-day trends, the data were reconfigured to count the
number of sounds of selected types in portions of recordings that corresponded to specific
hours of the day. The hourly counts were then pooled over all recordings to obtain time-of-day
curves. Interpretation of the resulting data is cautioned because estimates for some hours are
based on a small number of observations recorded from highly variable locations and dates.
Temporal overlap among biological and anthropogenic sounds was determined by count-
ing the number of sounds that overlapped completely, or partially, in time (i.e., that occurred
at the same time in the recording).
Received sound pressure levels (RSPL) as root mean square (RMS [47,48]) were automati-
cally calculated in SpectraPro332 over the full 24 kHz bandwidth for each ambient clip and
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 6 / 26
used to assign each recording to an ambient noise level category ranging from low-to-high in 5
dB steps (90–95, 96–100, 101–105, and >105 dB RMS RSPL re 1 μPA). Although these sound
level categories should not be construed to be estimates of the true location noise levels, which
likely vary widely over time and space, they do represent the received background noise levels
affecting the soundscape during the recording time. Sound levels were not estimated in some
recordings due to mechanical noise on the calibrated hydrophone, strong flow, cable strum-
ming, or other factors. Many locations (28%) apparently exhibited electromagnetic field
(EMF) levels high enough to introduce EMF noise into the recording and were excluded.
Power spectral density (PSD, Hanning, FFT 4096, 50% overlap, frequency resolution 11.7,
PSD dB re 1 μPa
2
Hz
-1
) data were automatically calculated and exported from Spectrapro322
for each identified sound over its frequency bandwidth, and for each ambient noise clip over
the full recorded bandwidth (Nyquist frequency = 24 kHz). Average PSD spectra of each
sound category, including ambient, were then calculated from a subsample (S3 Table) of
repr sentative clips.
Because PSD spectra of sound types contain energy from the background ambient noise,
and because sound categories occurred in various sized subsets of the recordings (i.e., some
categories were relatively rare, while others were ubiquitous), a comparison of the mean PSD
spectra of individual sound categories with mean ambient noise PSD can be misleading. To
address this bias, the sound levels of the ambient noise were subtracted from the sound levels
of selected sound categories based only on the locations in which they actually occurred. For
example, after linearizing, the spectra of the ambient noise were subtracted from each FRT
subsample from each of the ten locations where the FRT subsamples were selected, and then
all the resulting spectra were averaged over all 23 FRT subsamples. The resulting PSD spectra
provide an indication of the frequency bands over which the biophony is substantially higher
than the ambient, and comparisons with the PSD spectra of the biophony with that of the
anthropophony after subtraction of the ambient gives an indication of the potential for
masking.
Data analyses
Due to the wide variety of habitats sampled and lack of temporal control (data were not synop-
tic and of variable duration), many environmental factors that potentially influence sounds-
cape composition such as habitat category, diel period, river position and river system were
statistically confounded preventing detailed statistical comparison of their effects. However,
we cautiously compared selected factors based on various subsamples of the data pooled within
broad treatment categories in an exploratory examination of their potential to influence the
soundscape composition. The exploratory nature of these comparisons is emphasized given
the “snap-shot” nature of the data collection and lack of control of environmental variables
such as time-of-day, temperature, season (although all data were collected over a five-week
period), amount of human development (ranging from remote wilderness to heavily populated
urban areas), turbidity, water depth, and propagation distance. Because strong diel differences
were expected, and night sampling was limited, all soundscape measurements (percent fre-
quency, sound rate, and percent time) are reported separately for day and night recordings.
Factors examined for possible influence on the soundscape composition included diel period,
habitat type, position along the river gradient (examined separately for each river), river region
(non-tidal and tidal reaches pooled over all rivers for day samples) and ambient noise level
category. Frequency of occurrence of sound types among diel, habitat type and ambient noise
level category groups were tested with a Chi-Square test in one-way contingency tables. A one-
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 7 / 26
way analysis of variance was used to test for potential single-factor effects (e.g., habitat cate-
gory) sound number and time after transforming to normalize.
Results
A total of 4,825 biological, 1623 anthropogenic noise, and 834 unclassified sounds (Table 1)
were measured from 2,750 minutes of recording in 173 locations (S2 Table). Examples of some
of the most common air movement sounds are described elsewhere [45]. Examples of traffic,
train, running boat and other boat noises are provided in S2–S5 Figs together with their corre-
sponding sound files (S2–S5 Audios). An example demonstrating likely masking of other fish,
other air movement, and FRT sounds by traffic noise can be viewed in S2 Fig and heard in S2
Audio. These examples are provided to demonstrate the temporal variation in frequency con-
tent and relative sound level change within the sound type.
In general, anthropogenic noises were 1 to 2 orders of magnitude longer in duration than
biological sounds and exhibited consistent spectral overlap with them (Fig 2,Table 1,S6 Fig).
FRTs, insect and bird sounds had the longest durations of the biophony averaging 2–3 s, while
boat, plane and train sounds had the longest durations of the anthropophony averaging 28–
277 s. Other fish sounds had the lowest peak frequency of the biophony, while insects and
birds had the highest (Table 1). Traffic sounds had the lowest mean peak frequency and fishing
and other boat sounds had the highest peak frequency of the anthropophony. Train sound
peak frequency was inflated by whistle sounds, and otherwise would have the lowest peak fre-
quency (S6 Fig). Peak frequency of ambient noise was below that of most biophonic sounds
but overlapped strongly with traffic and other anthropogenic noises (Fig 2,Table 1,S6 Fig).
Biophony occurred in 57% and anthropophony occurred in 63% of the locations (Table 2).
Other air movement sounds were the most frequently occurring (39% of locations) component
of the biophony followed by other fish (30% of locations). Air movement sounds were domi-
nated by high-frequency sounds similar to those previously described [45] for salmonids
(46%), alewife (Alosa pseudoharengus, Clupeidae) and alewife-like sounds (27%), and white
sucker-like (Catastomus commersonii, Catostomidae) sounds (8%). Twenty-three percent of
FRT sounds were attributed to alewife while the rest were unknown. Most of the other fish
sounds were unknown, but 13% were catfish-like sounds most likely produced by brown bull-
head (Ameiurus nebulosus, Ictaluridae. [31]).
Biophony was observed in significantly more locations at night than during the day (84% vs
53%, Table 2). All components of the biophony except insect-like and bird sounds occurred in
significantly more night locations than day locations. Traffic sounds were the most common
component of the anthropophony occurring in 40% of the locations. No significant diel differ-
ences in occurrence among locations were observed for the anthropophony.
Other air movement (5%), FRT (7%), other fish (10%), insect-like (9%), bird (11%) and
other biological (4%) sounds overlapped in time with traffic sounds. The only other noises to
overlap in time with biological sounds were: airplane (0.1% and 0.4%, with other fish and
insect-like), other boat (0.1% and 0.3%, with other air movement and other fish), and other
noise (0.1% with other fish).
Diel patterns
Biophony sound type accumulation curves were highly variable among locations and did not
reach an asymptote (Fig 3). Total biological types per location tended to be more diverse at
night averaging 9.6 types per recording (standard error (SE) = 1.8, maximum = 37) compared
to 3.3 types per recording (SE = 0.3, maximum = 19) types during the day. The standardized
biodiversity measure exhibited a similar diel trend, but neither the total accumulated within
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 8 / 26
the first 348 s or the rate of accumulation were significantly different between day (mean = 1.8,
range = 0–12, SE ±0.2) and night (mean = 3.5, range 0–18, SE ±0.9).
The biophony accounted for 66% and 67% (mean = 1.7 and 2.7 sounds/min) of the total
number of sounds during the day (2.6 sounds/min) and night (4.0 sounds/min), respectively
(S7 and S11 Figs, Table 3). However, the anthropophony dominated the soundscape in terms
of relative percent time accounting for 92% and 88% of all sounds during the day and night,
respectively (Fig 4,S7 Fig,Table 3). Unclassified sound accounted for just 9% of the total
sounds by number and less than 5% of the sounds by percent time (S7 and S11 Figs, Table 3).
Although the relative contribution of biophony and anthropophony by both number and
percent time were similar between day and night, the composition of the sounds changed (Fig
4,S7 Fig,Table 3). Insect sounds dominated the biophony by day, while total fish sounds dom-
inated by night. Air movement fish sounds and other fish sounds contributed about equally to
total fish sounds, though other fish sounds were more numerous during the day, while the lon-
ger duration air movement sounds accounted for more of the percent time at night. Traffic
sound was the most numerous component of the anthropophony during both day and night
but the long-duration boat sounds contributed more to the percent time during the day.
Table 1. Sound duration and peak frequency.
Duration (s) Frequency (Hz)
Sound category Samples Mean SE Min Max Samples Mean peak SE peak Max peak Mean IQR-BW SE IQR-BW
Biophony
FRT 135 3.03 0.34 0.16 31.53 125 1717 102 4969 586 46
Other air movement 1636 0.19 0.01 0.02 13.29 1506 1742 28 10359 607 14
Other fish 1807 0.32 0.03 0.01 32.90 1612 726 9 2625 303 5
Insect-like 631 2.21 0.10 0.02 29.62 238 2820 121 12188 776 94
Bird 57 2.29 0.39 0.11 16.02 54 2912 195 4781 752 79
Surface 197 0.84 0.04 0.06 4.60 188 906 35 3797 414 23
Other biological 362 0.20 0.04 0.01 13.05 325 1811 87 11438 933 56
Subtotal 4825 4048
Anthropophony
Airplane 11 27.58 3.66 14.76 51.18 9 307 130 984 120 24
Fishing 74 2.51 0.57 0.16 32.20 67 1483 230 8344 1133 244
Running boat 57 174.64 18.73 15.37 733.44 45 875 126 4266 1094 333
Idle boat 14 276.66 68.96 14.19 788.62 11 435 162 1406 464 138
Other boat 127 6.97 1.30 0.06 100.62 75 1159 90 3609 523 53
Traffic 1237 7.90 0.26 0.05 132.00 1070 225 19 4594 151 6
Train 22 44.53 29.81 0.06 654.99 5 459 103 563 84 50
Other noise 81 15.70 2.58 0.01 182.91 63 847 99 2906 373 48
Subtotal 1623 1345
Unclassified
Gurgle 313 1.24 0.11 0.09 18.13 296 1108 20 3797 229 10
Bubble-like 105 2.39 0.24 0.06 17.44 88 948 29 2109 222 21
Unknown 416 1.02 0.10 0.01 18.24 334 1616 69 10172 580 44
Subtotal 834 718
Ambient 159 297 34 1875 481 84
Total 7282 6270
Summary statistics for sounds pooled over all samples within sound categories. SE = Standard error of the mean, Min = minimum value, Max = maximum value,
peak = frequency of greatest energy, IQR-BW = interquartile bandwidth.
https://doi.org/10.1371/journal.pone.0221842.t001
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 9 / 26
Traffic sound accounted for more than half of the recorded sound during the night based on
mean percent time per location. Hourly trends for sound rates of the biophony reflect a similar
diel pattern (S8 Fig). Insect-like sounds peaked at 87 sound/h in the early afternoon, air move-
ment sounds peaked at 43 sound/h in the early evening, while other fish sounds peaked at 18
sounds/h in the afternoon and 32 sounds/h in the evening (S8 Fig).
Habitat patterns
There was no significant difference in the frequency of occurrence of biophonic categories
among habitat types during the day (S4 Table). At night, insect sounds occurred more fre-
quently in the brook/creek habitat locations (however, the brook/creek sample size was low
N = 2), while other biological and bird sounds were absent from the pond/lake habitat loca-
tions. Traffic sounds were the most widespread noise and were significantly more frequently
occurring in brook/creek habitat locations during the day. In contrast, boat sounds were
absent from brook/creek locations (S4 Table). No significant differences in the frequency of
occurrence of anthropophonic sounds were observed among the habitat categories at night (S4
Table).
Insect and other fish sounds dominated the biophony by percent time in all habitats during
the day, but air movement sounds dominated pond/lake and river habitats at night (S9 Fig,S5
Fig 2. Sound duration and peak frequency. Comparison of acoustic characteristics among sound types and ambient noise. Square
symbols mark the mean peak frequency, blue hats mark one standard error of the mean. The lower stem marks the mean first
quartile frequency and the upper stem the mean 3rd quartile frequency (see Table 1).
https://doi.org/10.1371/journal.pone.0221842.g002
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 10 / 26
Table). Other fish dominated brook/creek habitat at night but the sample size was low (N = 2
locations). Traffic sounds dominated soundscape percent time during both day and night in
brook/creek habitat, while boat sounds dominated other habitats (S9 Fig,S5 Table).
River gradient pattern
There were no consistent trends among rivers in biological or noise sounds in main-stem river
habitats moving along the river gradient from headwaters to mouth, although the highest ele-
vation locations tended to have little or no biological sounds (S1 Data set online). When day
data from all rivers were pooled after grouping locations into non-tidal (N = 46) and tidal
regions (N = 20), all boat noise categories were significantly more frequent in tidal regions (S6
Table). Similar trends were observed for sound rate and percent time (S10 Fig,S6 Table). Boat
noise averaged 31% of the time in tidal regions, but only 2% in non-tidal areas. Similarly, boat
sounds (all types combined) averaged 0.02 sounds/min and 0.21 sounds/min in non-tidal and
tidal regions, respectively (P <0.0001). Average soundscape percent time of traffic sounds and
Table 2. Comparison of sound occurrence between diel periods.
Sound category Day samples Night samples Total Chisq P
Biophony
FRT 22% 40% 24%
Other air movement 39% 76% 45%
Air movement 42% 80% 47%
Oher fish 30% 64% 35%
Fish 48% 84% 53%
Insect-like 14% 24% 15% ns
Other biological 20% 44% 23%
Surface 14% 56% 20%
Bird 13% 20% 14% ns
All biophony 53% 84% 57%
Anthropophony
Airplane 4% 8% 5% ns
Running boat 13% 24% 14% ns
Idle boat 7% 0% 6% ns
Other boat 10% 8% 10% ns
Boat 17% 24% 18% ns
Fishing 5% 12% 6% ns
Traffic 37% 56% 40% ns
Train 2% 0% 2% ns
Other noise 10% 8% 10% ns
All anthropophony 60% 80% 63% ns
Unclassified
Bubble-like 18% 32% 20% ns
Gurgle 21% 48% 25%
Unknown 39% 72% 43%
All unclassified 46% 76% 50%
All Sounds 76% 92% 79% ns
Sample size 148 25 173
Percent of the stations where sound types were observed. P = probability of a significant difference in the frequency of sounds between day and night based on a Chi
Square (Chisq) test in a one-way contingency table on frequency counts (ns = not significant, =<0.05, =<0.01, =<0.001).
https://doi.org/10.1371/journal.pone.0221842.t002
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 11 / 26
Fig 3. Rate of accumulation of new biophonic sound types at each location. Comparison of the number of unique
types of biophony sounds for day (top) and night (bottom) plotted against the time elapsed from the beginning of the
recording for each location illustrating the high variability in type number and rate of accumulation. Note the
difference in the y-axis scale between day and night.
https://doi.org/10.1371/journal.pone.0221842.g003
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 12 / 26
all anthropophonic sounds combined were also significantly different between river regions
(traffic = 8.3% and 0.8%, noise = 12.1% and 33.2%, for non-tidal and tidal regions, respectively,
both P <0.05). Surface sounds were significantly more frequent, numerous, and occupied
more time in tidal regions than non-tidal. No other biophonic sounds were significantly differ-
ent among regions (S10 Fig,S6 Table).
Ambient noise level
Overall received ambient SPL values ranged from 90 to 133 dB re 1 μPA rms. There was no sig-
nificant difference in received ambient SPL among habitat types during the day with averages
(±SE) of 99.4 (2.2), 98.7 (1.2) and 101.1 (1.4) dB re 1 μPA rms, for brook/creek, pond/lake and
river habitats, respectively (based on a one-way ANOVA on log transformed data; N = 77).
However, average power spectral density curves of the ambient noise suggest differences in the
Table 3. Comparison of sounds between diel periods.
Number per minute Percent time
Day Night Day Night
Sound category Mean SE Max Mean SE Max Mean SE Max Mean SE Max
Biophony
FRT 0.028 0.006 0.6 0.054 0.024 0.5 0.13 0.04 4.10 0.27 0.11 1.90
Other air movement 0.239 0.074 10.2 0.940 0.350 7.9 0.06 0.01 1.30 0.34 0.12 2.10
Air movement 0.267 0.075 10.3 0.994 0.370 8.4 0.18 0.04 4.17 0.58 0.20 4.01
Other fish 0.585 0.201 22.3 1.316 0.790 18.9 0.24 0.08 7.50 0.48 0.22 4.90
Fish 0.852 0.221 22.3 2.310 0.913 20.2 0.44 0.10 7.50 1.09 0.33 6.30
Insect-like 0.759 0.597 88.0 0.032 0.020 0.5 1.12 0.59 76.80 0.13 0.10 2.30
Other biological 0.076 0.019 1.5 0.198 0.064 1.1 0.02 0.01 0.40 0.14 0.09 2.10
Surface 0.029 0.010 0.9 0.111 0.042 1.0 0.04 0.01 1.38 0.14 0.08 2.04
Birds 0.015 0.004 0.4 0.017 0.008 0.1 0.04 0.02 1.76 0.07 0.05 1.08
All biophony 1.730 0.647 89.6 2.668 0.940 20.5 1.57 0.57 77.02 1.51 0.47 9.39
Anthropophony
Airplane 0.003 0.001 0.1 0.005 0.003 0.1 0.17 0.09 9.80 0.22 0.16 3.20
Running boat 0.020 0.005 0.3 0.009 0.004 0.1 5.30 1.37 92.40 2.06 0.92 16.50
Idle boat 0.005 0.002 0.1 0.000 0.000 0.0 2.52 0.99 100.00 0.00 0.00 0.00
Other boat 0.092 0.057 8.2 0.017 0.016 0.4 0.57 0.34 46.60 0.36 0.35 8.50
Boat 0.118 0.058 8.2 0.026 0.018 0.5 8.40 1.95 100.00 2.42 1.07 16.50
Fishing 0.009 0.004 0.4 0.045 0.025 0.5 0.04 0.02 2.10 0.11 0.07 1.70
Traffic 0.453 0.087 7.7 0.824 0.482 12.5 5.77 1.05 69.80 8.22 2.89 54.80
Train 0.008 0.005 0.6 0.000 0.000 0.0 0.56 0.41 54.90 0.00 0.00 0.00
Other noise 0.025 0.009 0.9 0.033 0.026 0.6 0.21 0.18 26.00 1.44 1.14 26.50
All anthropophony 0.644 0.106 9.2 0.964 0.490 12.4 15.23 2.17 100.00 12.55 3.54 65.50
Unclassified
Bubbles 0.034 0.008 0.6 0.031 0.014 0.3 0.12 0.03 2.13 0.16 0.09 1.93
Gurgle 0.045 0.009 0.7 0.199 0.104 2.5 0.10 0.03 2.42 0.34 0.18 4.10
Unknown 0.163 0.031 2.7 0.142 0.033 0.6 0.26 0.06 5.03 0.24 0.06 1.10
All unclassified 0.242 0.035 2.8 0.371 0.127 3.0 0.47 0.08 5.03 0.71 0.22 4.92
All sounds 2.616 0.660 89.6 4.002 1.166 25.4 16.55 2.12 100.00 14.26 3.58 66.02
Sample size 148 25 141 24
Mean number per minute and percent of recording time per minute for all locations of selected sound categories for day and night sampling. SE = standard error of the
mean, Max = maximum value.
https://doi.org/10.1371/journal.pone.0221842.t003
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 13 / 26
frequency structure among habitat types (S6 Fig). Brook/creek habitats tend to have the high-
est levels and pond/lake habitats the lowest levels at frequencies below 500 Hz, while river hab-
itats have the highest levels at all higher frequency bands.
There were significant differences in the biophony among the four ambient noise categories
(Fig 5,S7 and S8 Tables). Air movement sounds significantly declined from a high of 72% of
locations to a low of 6% of locations from the lowest to the highest ambient noise level catego-
ries. Similar, but non-significant, trends in occurrence were observed for FRT and other fish
sounds (Fig 5). Rate and percent time of biological sounds followed similar trends with signifi-
cant declines in air movement, fish and total biophony with increasing ambient SPL (S8 Table).
There was a highly significant decline in the standardized diversity of biophonic sound types
from 3.2 (±0.6) to 0.1 (±0.1) sounds from the lowest ambient noise to the highest ambient
noise category (S12 Fig).
A comparison of the average power spectral density curves of major biophony categories
with major anthropophony categories indicates that other fish sounds are above average
anthropogenic sounds, except for running and other boat sounds (S6 Fig). There is also con-
siderable overlap with the spectra of traffic sounds. Air movement and insect sounds produce
low amplitude sounds largely below anthropogenic sound levels and ambient spectra averaged
across all recordings (S6 Fig). Air movement sounds nevertheless exhibit significant energy
above ambient noise from the locations they occurred at, while insect-like sounds tended to be
close to ambient levels (Fig 6).
Fig 4. Soundscape composition. Relative contribution of the anthropophony and biophony andtheir major components to the aquatic soundscape
during the day and night based on mean percent time of each sound type (data provided in Table 3). The composition of the biophony is shown in the
expanded plots.
https://doi.org/10.1371/journal.pone.0221842.g004
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 14 / 26
Discussion
The three most significant findings in this study are: 1) freshwater habitats in the New England
region of North America contain a diverse array of unidentified biological sounds; 2) air
movement sounds constitute an important component of freshwater soundscapes; and 3)
anthropogenic noises dominate the soundscape and their overlap in time and frequency with
the biophony, together with their high sound levels, suggests a high potential for negative
impacts. The impact of anthropogenic noise on the natural freshwater soundscape is most
strongly illustrated in Fig 4 which demonstrates the prevalence of the anthropophony account-
ing for 90% of the soundscape by relative mean percent time. This finding alone signals how
dramatically freshwater soundscapes have been modified by human-produced noise. The
strong spectral overlap between the biophony and anthropophony further hints at potential
impacts (Fig 2). However, masking only occurs when individual biophonic sounds overlap in
time and frequency with individual anthropophonic sounds. Assuming no bias in detection,
high rates of temporal overlap with the biophony were only found with traffic noise (ranging
for 4% to 11% depending on the biophonic category), while if detection was biased, then the
overlap is underestimated. It should be pointed out that any sound level of anthropophonic
noise in these habitats alters the soundscape from that in which aquatic organisms have
evolved. The potential for impact is further supported by the observed higher sound levels of
anthropophony sources compared to biophony sources (Fig 6 and S6 Fig). Finally, the highly
significant decline in the frequency of occurrence (Fig 5,S7 Table), number and percent time
of sounds (S8 Table), and diversity (S12 Fig) of the biophony with increasing ambient received
sound level suggests that high noise levels, whether natural or anthropogenic, can affect sound
production and soundscape composition (although bias from masking of sound detection
must be considered as discussed below).
Fig 5. Ambient noise level effect. Comparison of the frequency of occurrence of selected biophony sound categories among locations grouped into
four ambient noise categories: 90–95 dB (N = 29), 96–100 dB (N = 17) 101–105 dB (N = 14) and >105 dB (N = 17). A Chi Square (Chisq) test in a one-
way contingency table of frequencies tested for differences in the observed group frequency from the expected group frequency (P 0.05, ns = not
significant). See S7 Table for details.
https://doi.org/10.1371/journal.pone.0221842.g005
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 15 / 26
Fig 6. Selected biophony and anthrophony spectra above local ambient noise. Mean power spectral density (PSD)
curves above the ambient for the biophony (A) and anthropophony (B and C) were calculated as the received level
difference between individual sounds and the ambient spectra from the same recording and then averaging among all
subsampled sounds within each biophonic category (Hanning, FFT 4096, 50% overlap, frequency resolution 11.7, PSD
normalized).
https://doi.org/10.1371/journal.pone.0221842.g006
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 16 / 26
Our observations of the widespread occurrence of air movement sounds in many habitats
across a large geographic area, together with their large contribution to the biological sounds-
cape based on sound rate and sound percent time, suggest for the first time that air movement
sounds are an important phenomenon in multiple types of freshwater habitats. We emphasize
that even if air movement sounds are largely incidental, if sounds are species specific [45], they
can be used by scientists and resource managers as an aid in documenting the spatial and tem-
poral distribution of fishes and their associated soniferous behavior [49]. In addition, we point
out that even incidental sounds can expose an organism to predation by organisms that can
hear that sound, and hence would be subject to natural selection pressures.
Despite the lack of significant habitat differences in ambient noise levels, sound level
appeared to have a strong negative influence on biological diversity and soundscape composi-
tion (Fig 5,S7 and S8 Tables), suggesting possible masking, suppression of sound production,
or avoidance of locations with high ambient noise levels regardless of their source. A negative
effect of ambient noise level on biological sound production or detection supports previous
work on potential impacts of noise levels on freshwater fishes [e.g. 10–11,13,20,25,37 and
review in 7], however we caution that the results do not necessarily suggest the observed trend
was due to purely anthropogenic noise effects since no biological sounds were negatively cor-
related with any noise rate or percent time variable. The lack of such correlations may simply
be due to the high diversity of sounds observed while sampling over a wide variety of habitats
and geographies with many different faunal assemblages. The overlapping frequency structure
of anthropogenic noises with the biophony, especially with the other fish category (Fig 2,S6
Fig,Table 1) also suggests the potential for masking (an example of which can be seen in S2
Fig and the corresponding S2 Audio). In contrast, the lack of overlap between peak frequencies
of biological sounds and peak ambient frequency (Fig 2,S6 Fig) lends further support to previ-
ous studies suggesting that fish take advantage of acoustic niches [e.g. 25]. Our data suggest
that a high-frequency “quiet noise window” may also occur in other freshwater habitats, and
that freshwater organisms may have evolved to exploit different acoustic windows in different
habitats. It is also possible that multiple acoustic windows may occur in some habitats which
can be exploited by different organisms. The impact of anthropogenic noise on the natural
noise windows in various habitats remains poorly understood.
Running boats generate long-lasting noise starting with low-amplitude and low-fre-
quency noise while in the distance and progressing to high-amplitude noise that saturates
the recordings when within a few hundred meters (depending on the boat type and speed),
before gradually fading as the boat moves off in the distance again (an example of the sound
of an approaching boat which anchors within 100 m of the recording site is provided in S4
Fig and corresponding S4 Audio). Boat noise in excess of 90% time were observed at some
locations (S1 Data set), highlighting the potential of boat noise as a masking source in fresh-
water. Similar observations have been reported in freshwater [32] and marine habitats [e.g.
50–52]. Smott et al [51] reported three types of boat sounds: burst broadband, variable
broadband, and low frequency, which likely correspond with our other boat, running boat,
and idle boat categories. Locations near important navigation routes, marinas, or boat land-
ings experience chronic boat noise exposure. In some cases, boat noise is so prevalent in the
hours around sunset (when boaters are returning to shore) that recording of biological
sound is nearly impossible. Our sampling was conducted during the early season when boat-
ing activity was sparse and many boat ramps had not yet opened for the season. We would
expect, therefore, that boat noise would be significantly more prevalent during the summer
months than that observed in our study, similar to reports from marine systems [e.g. 50–52].
Boat noise was undoubtedly an important chronic noise source in navigable waters where it
dominated the soundscape and unquestionably masks many biological sounds (Fig 6,S4 and
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 17 / 26
S6 Figs). Boat noise is particularly problematic in enclosed water bodies such as small lakes,
and in linear rivers as the sound travels great distances. We have often detected motor boat
sounds before sighting the vessel in the distance. On the other hand, serpentine waterways
may be less impacted because sounds of an approaching vessel are blocked by land until the
vessel moves around the bend and into the line of sight.
Our observations suggest that boats running idle while docked, anchored, or drifting are a
major component of freshwater soundscapes (Fig 4), and have the potential to mask some
biophony such as insect sounds, but we are not aware of studies that examine its potential
impact in freshwater. However, at least one study in a small estuarine river reported boat
sounds attributed to idling to be important masking sources for sciaenid fishes [51]. The ten-
dency for boaters, and especially ferries and other large vessels, to idle for long periods, and
the lower frequency structure of idling boats (Figs 2and 6,S6 Fig) suggests that boat idling
may be an important chronic noise source in navigable waters. Power density spectra of boat
sounds in this study were remarkably similar to those reported in other freshwater [e.g. 10,42]
and marine studies [e.g. 51].
Traffic noise exhibits similar though much less extreme wax-and-wane patterns as can
clearly be seen in S2 Fig and heard in the corresponding S2 Audio. Our observations support
those of previous researchers that traffic sounds can be important in some freshwater habitats
[34–38,41–42]. Holt and Johnston [37] measured traffic noise propagation distances in
streams and concluded that traffic sounds may have negative impacts in freshwater habitats.
In contrast, ice road traffic (with similar signatures to our traffic sounds) were not thought to
impact burbot (Lota lota, Gadidae) under the ice in Great Slave Lake, Canada [41]. Hopson
[38] found evidence of traffic impacts on both the above and below water biophony in fresh-
water wetlands of Ohio.
Although not as extreme as boat noise, traffic sounds are far more ubiquitous in freshwater
habitats and can be chronic in urban areas and during rush hours. In fact, the increase in traf-
fic contribution to the soundscape at night was due in part to our sampling around sunset
when traffic tends to be heavy. The relatively high temporal overlap between biological sounds
and traffic sounds suggests a high likelihood for impacts, especially for other fish sounds
which had both high temporal (10%) and acoustic frequency overlap (Figs 2and 6,S6 Fig).
A comparison of the power density spectra of traffic sounds with the biophony (Fig 6,S6 Fig)
suggests they would likely mask all but the other fish category. Although we did not examine
propagation distances of traffic sounds, traffic crossing bridges was detected in locations as far
as 270 m from the nearest bridge (S1 Data set), agreeing well with previous studies [37,42].
Differences in the relative contribution of traffic and boat noises to the soundscape among
habitats and river regions demonstrate variations in potential noise impacts among habitats
and river zones (S9 and S10 Figs, S5 and S6 Tables). It should be recognized that traffic sounds
were likely over-represented in our sampling due to the frequent necessity of accessing the
water near bridges. However, as has previously been pointed out [37], in many areas bridge
crossings are common and smaller streams and rivers may be crossed many times over short
stretches in urban and suburban locations.
The inconsistent trends in observations moving along the river main stem locations likely
result from the fact that river systems form a coenocline where gradients in abiotic and biotic
conditions regulate community structure and habitat function [53]. Thus, comparison among
river systems of different lengths and elevation gradients requires a gradient approach. To our
knowledge this is the first study to examine river order changes in freshwater soundscapes,
although differences in ambient noise levels have been compared among rivers [17,19] and
along short reaches of rivers [18]. Major differences in the soundscape were previously found
between two locations in the Hudson River [31]. The only other study to examine the
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 18 / 26
biophony of multiple locations within the same freshwater river, found significant differences
among tributaries classified by degree of connectivity [22].
It is interesting that while the biotic community changes considerably from high elevation
reaches to estuarine reaches, the changes in the biophony type contribution to the soundscape
are minimal, suggesting that although soniferous species may change, the broad sound catego-
ries are more consistent. Classification of sounds to more specific sound types would likely
have resulted in significant differences in the soundscape along the river coenocline due to
species assemblage changes as it has been shown in a series of studies in a small estuarine river
[51,54–56]. A gradient in impacts from different types of anthropogenic noise is expected as
we observed a striking transition from remote wilderness to increasingly developed urban
areas while traveling from the river headwaters to the sea. Some of this transition was captured
in the comparison between non-tidal and tidal reaches of the rivers where there was a shift
between dominance of the soundscape by traffic noise in non-tidal reaches to a dominance by
boat noise in tidal reaches (S10 Fig,S6 Table), highlighting different potential for impacts in
different habitats.
Studies on the biophony of temperate freshwater habitats have used a variety of methodolo-
gies from point sampling to long-term sampling (S1 Table). Studies in rivers have reported 21
[31], 25 [26], and 128 [22] biophonic sound types. Similar numbers have been reported for
ponds (48 [4]) and lakes (26 [43]). Except in single species studies, in most cases biophonic
sound sources are unknown. As we pointed out in our review of sound production in temper-
ate freshwater fishes [9], there is a critical need for studies to describe fish sound sources in
freshwater habitats. Our use of real-time monitoring resulted in the detection of a high diver-
sity of air movement sounds that would likely have been overlooked if the recordings had only
been analyzed in post-processing (see S1 Audio for an example). In addition, real-time moni-
toring allowed us to identify false biophonic sounds such as creaking logs, twig rattling, and
floating dock squeaking that appear very similar to biophonic sounds in their acoustic
characteristics.
Freshwater habitats are highly impacted by terrestrial and aerial sound sources. Bird sounds
are one component of the soundscape that may have important effects on fish behavior. For
example, herring gull sounds are clearly audible underwater in herring runs, and are likely
audible to alewife [45]. Other terrestrial sounds have similar potential ecological interactions.
We advocate for studies that simultaneously record above and below water sounds [38–40,57]
to distinguish between above and below water sound sources, but ultimately to describe what
we refer to as the “holo-soundscape” of freshwater systems. Kuehne et al [40] attempted to cor-
relate above and below water sound recordings, but unfortunately defined the biophony as
sounds from 3–8 kHz and the anthropophony as sounds of 1–2 kHz which confounds the two
sound sources, given that many fish sounds would fall in the lower frequency bandwidth. Hop-
son [38] reported higher sound intensity, more anthropophonic sound occurrences, and lower
acoustic diversity in both above and below water wetland soundscapes in disturbed versus
non-disturbed areas.
Recent developments in acoustic sensor technology (e.g. long performance dataloggers)
and of automated processing methods (e.g. automatic recognition methods and acoustic indi-
ces) now allow researchers to relatively inexpensively collect long-term recordings and to pro-
cess them in timeframes which are unattainable by more traditional recording and analysis
approaches (e.g. manual and aural quantification of sound occurrences) [4]. These develop-
ments are rapidly expanding the field of passive acoustic research and have the potential to
quickly identify biophonic patterns and anthropophonic impacts on them [4]. However, as
the biophonic component of most aquatic habitats is still poorly understood, the application of
automated processing methods alone, without a prior knowledge of the type of signals present
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 19 / 26
in a specific site, might result in interpretations that do not accurately reflect the biodiversity
and the ecological status of the area. Efforts to identify specific sound sources in freshwater
soundscapes are, therefore, critically needed. Attempts to correlate unknown sounds with the
presence of macroinvertebrates (e.g. [13,22,35–36]) or fishes [2,23–24,27,32] are especially
helpful. Follow-up studies that attempt to identify the unknown sound sources are also impor-
tant (e.g. [23,31,45]). Unfortunately, few studies incorporate real-time monitoring of sounds
(S1 Table, [2,23,27,31]). A series of studies in the estuarine May River of South Carolina [51,
54–56] demonstrate the advantage of having well documented sound sources when attempting
to examine anthropophonic impacts and ecological influences on the underwater soundscape.
Whenever possible when beginning a research program in previously unstudied freshwater
habitats, researchers should attempt to conduct preliminary studies utilizing real-time sound
monitoring and visual observations of the holo-soundscape, together with faunal sampling
and field-auditioning of aquatic organisms before, or concurrently with, the collection of long
soundscape sound series.
Supporting information
S1 Fig. Representative sampling sites. A) Brook wp 42, B) Creek wp 60, C) Pond wp 68, D)
Lake wp 83, E) Tributary wp 117, F) River wp 101. The waypoint (wp) number can be used to
look up the location details in the S1 Data set. The individual pictured in S1Fig E has provided
written informed consent (as outlined in PLOS consent form) to publish their image alongside
the manuscript.
(TIF)
S2 Fig. Example of traffic sound. Relative amplitude (top) and spectrogram (bottom) of traf-
fic and fish sounds recorded at night on 14 May 2008 in Sucker Creek, Griffin, New Hamp-
shire (N43˚ 00.257’ W71˚ 20.933’). As a car passes over a nearby bridge (Traffic), catfish
sounds (Other Fish) are partially masked. An unmasked catfish sound occurs later in the clip
and indicates a true peak frequency well within the traffic noise. Examples of a fish splashing at
the surface (Surface), and subsequent air movement (Other Air movement) and FRT sounds
can also be seen relative to the traffic sound. An amplified audio file corresponding to the fig-
ure can be heard in S2 Audio online. Spectrogram parameters: unfiltered, 1,024-point Hann
windowed FFTs with 50% overlap.
(TIF)
S3 Fig. Example of train noise. Recorded on 3 May 2008 in Deep River, a tributary of the
Connecticut River in Deep River, Connecticut (N41˚ 22.978’ W72˚ 25.566’). Top: photograph
of the train passing by at the time of the recording. Bottom: relative amplitude waveform and
spectrogram of the train sound which can be heard in the corresponding S3 Audio online.
Spectrogram parameters: unfiltered, 1,024-point Hann windowed FFTs with 50% overlap.
(TIF)
S4 Fig. Example of running boat noise. The sound of an outboard motor boat as it
approaches from the distance and stops to anchor nearby, recorded on 3 May 2008 in the
mainstem of the Connecticut River in Old Saybrook, Connecticut (N41˚ 19.143’ W72˚
21.028’). Top: Photograph of the boat as it passes, Bottom: relative amplitude waveform and
spectrogram of the noise generated by the passing boat, which can be heard in the correspond-
ing S4 Audio online. Spectrogram parameters: unfiltered, 1,024-point Hann windowed FFTs
with 50% overlap.
(TIF)
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 20 / 26
S5 Fig. Example of other boat noise. Sound produced by the power trim of a nearby outboard
boat, recorded on 1 May 2008 in the mainstem of the Connecticut River in Northampton,
Massachusetts (N42˚ 20.114’ W72˚ 37.211’). The corresponding sound can be heard in the S5
Audio online. Spectrogram parameters: unfiltered, 1,024-point Hann windowed FFTs with
50% overlap.
(TIF)
S6 Fig. Average power spectral density of major sound categories. Power spectral density
(PSD) averaged over a subsample of sounds from each major sound category (samples sizes
are shown in S3 Table). A) selected biophonic sounds, B and C) selected anthropophonic
sounds, D) ambient noise from each habitat category. Spectrogram parameters: Hanning,
FFT 4096, 50% overlap, frequency resolution 11.7, PSD normalized.
(TIF)
S7 Fig. Comparison between day and night soundscapes. Major components of the sounds-
cape: A and C) during the day and night, respectively, based on mean number, B and D) dur-
ing day and night, respectively, based on mean percent time. The relative contribution of the
biophony compared to the anthropophony is shown in the main pie, while the composition of
the biophony slice is shown in the expanded pie. The size of the wedge represents the relative
proportion of the sound out of all sounds based on data found in Table 2 of mean number per
minute (A and C) and mean percent time (B and D).
(TIF)
S8 Fig. Hourly trend in biophony abundance. Mean number of sounds of major biophony
categories by hour of the day.
(TIF)
S9 Fig. Comparison of habitat soundscapes. Comparison of soundscapes among habitat cate-
gories and diel period based on mean percent time. Day: A) creek/brook habitat during the
day (N = 21), B) pond/lake habitat (N = 41), C) river habitat (N = 79). Night: D) creek/brook
habitat (N = 2), E) pond/lake habitat (N = 7), F) river habitat (N = 15). See S5 Table for sum-
mary statistics.
(TIF)
S10 Fig. Comparison of soundscapes between tidal and non-tidal river regions. Relative
contribution of anthropophony and biophony to the aquatic soundscape of non-tidal (top)
and tidal (bottom) main-stem river regions during the day based on mean number of sounds
per minute (A and B) mean percent time (C and D) of each sound type. Data and statistics are
provided in S6 Table, while the size of the wedge represents the relative proportion of the
sound out of all sounds.
(TIF)
S11 Fig. Average day-time soundscape composition. Venn diagram illustrating the relative
composition of the anthropophony and biophony and their major constituents to the day-
time soundscape. The diameter of each circle is proportional to the mean percent of recording
time for the indicated sound category. Circles within large circles represent subcomponents of
the larger category. For example, the fish category contains two nearly equal subcomponents
(other fish and air movement sound).
(TIF)
S12 Fig. Decline of biophony diversity with increasing ambient noise level effect. Compari-
son of the number of biophony sound types among locations grouped into four ambient noise
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 21 / 26
levels based on received sound pressure level (RSPL). The decline in diversity form low to high
noise level is highly significant (P 0.001) during the day (gray bars, N = 74), but nonsignifi-
cant during the night (black bars, N = 16).
(TIF)
S1 Table. Selected review of passive acoustic studies in temperate freshwater habitats with
a few comparative marine studies. Brief description of soundscape studies in temperate fresh-
water systems. Type = is the general type of study. Effort = brief description of sampling
regime. Lake, Pond, River, Stream, Other = the number of unique habitats sampled (i.e., spe-
cific river or lake). The total number of specific sites within the habitat type is given in paren-
thesis. N = total sample size when known. Anthropophony = indicates some attention was
given to describing anthropogenic sound sources. Biophony = when some quantification or
description of the biophonic composition is provided. Real-time = sampling included some
real-time listening/observations of soundscape. Other data = other types of data collected to
correlate to soundscape data.
(XLSX)
S2 Table. Sampling effort for all locations.
(XLSX)
S3 Table. Subsample sizes used for spectral analysis by sound type.
(XLSX)
S4 Table. Percent occurrence of sounds at locations by habitat category. (P = significance
level for a Chi Square (Chisq) test in a one-way contingency table on differences among habi-
tats within a diel period, =<0.05, =<0.01, =<0.001, n/a = no test; N = 148 for day
and 25 for night).
(XLSX)
S5 Table. Mean percent time by habitat category. Comparison of the mean percent time
(total duration of sounds in the category divided by recording duration) of sound types among
habitat types. SE = standard error of the mean. P = probability of a significant difference
among habitats based on a one-way ANOVA on transformed variables, performed separately
by diel period. (=<0.05, =<0.01, =<0.001, n/a = no test).
(XLSX)
S6 Table. Comparison of summary statistics between tidal and non-tidal river regions.
Mean number per minute, percent of recording time per minute, and percent occurrence of
sound categories for day sampling within tidal and non-tidal river regions (SE = standard
error of the mean, P = significance level from a one-way analysis of variance, Chisq
P = significance level from a Chi Square test of expected frequencies in a one-way contingency
table, 0.05, 0.01, 0.001).
(XLSX)
S7 Table. Frequency of occurrence of biophony among ambient noise level categories.
Comparison of the percent frequency of occurrence among day-time recordings grouped into
received ambient noise level categories (RMS = root mean square). (P = significance level from
a Chi Square test of difference among SPL categories from the expected frequency in a one-
way contingency table. ns = not significant, 0.5, 0.01, 0.001).
(XLSX)
S8 Table. Comparison of the mean number and mean percent time among daytime ambi-
ent noise categories. (P = results of a one-way ANOVA on differences among SPL levels of
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 22 / 26
transformed variables. ns = not significant, 0.5, 0.01, 0.001).
(XLSX)
S1 Audio. Example of fish jump and air movement sounds with voice notes. Recorded
observation of a fish jump followed by air movement sounds, together with voice notes on the
event. Recorded at dusk on 30 April 2008 in Deerfield River, West Deerfield, Massachusetts
(N42˚ 31.591’ W72˚ 37.952’). The sound has been amplified for optimal listening online.
(MP3)
S2 Audio. Traffic sound with biophony. Sound file corresponding to S2 Fig containing the
sound of traffic noise and examples of other fish, other air movement and FRT sounds
recorded at night on 14 May 2008 in Sucker Creek, Griffin, New Hampshire (N43˚ 00.257’
W71˚ 20.933’). The sound has been amplified for optimal listening online.
(MP3)
S3 Audio. Train sound. Example of train noise corresponding to S3 Fig and recorded on 3
May 2008 in a tributary of the Connecticut River in Deep River, Connecticut (N41˚ 22.978’
W72˚ 25.566’). Note that several very faint other air movement sounds (chirps) can be heard
in the recording but are high frequency and not shown in the S3 Fig. The sound has been
amplified for optimal listening online.
(MP3)
S4 Audio. Running boat sound. Example of the sound of an outboard motor boat, corre-
sponding to S4 Fig, as it approaches from the distance and stops to anchor nearby. Recorded
on 3 May 2008 in the mainstem of the Connecticut River in Old Saybrook, Connecticut (N41˚
19.143’ W72˚ 21.028’). The sound has been amplified for optimal listening online.
(MP3)
S5 Audio. Other boat noise. Example of noise produced by the power trim of a nearby out-
board boat, corresponding to S5 Fig. Recorded on 1 May 2008 in the mainstem of the Con-
necticut River in Northampton, Massachusetts (N42˚ 20.114’ W72˚ 37.211’). The sound has
been amplified for optimal listening online.
(MP3)
S1 Data set. Meta-data of sounds observed in all recordings. Recording location meta-data
with sound rates and percent time by sound category. The data set is in Excel format with two
worksheets: 1) data, 2) field definitions.
(XLS)
Acknowledgments
Joe Olson provided technical assistance. Megan O’Connor provided the map used in Fig 1.
Author Contributions
Conceptualization: Rodney A. Rountree.
Data curation: Rodney A. Rountree.
Formal analysis: Rodney A. Rountree.
Funding acquisition: Rodney A. Rountree.
Investigation: Rodney A. Rountree, Francis Juanes.
Methodology: Rodney A. Rountree, Francis Juanes.
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 23 / 26
Project administration: Rodney A. Rountree.
Resources: Rodney A. Rountree, Francis Juanes.
Software: Rodney A. Rountree, Francis Juanes.
Supervision: Rodney A. Rountree.
Validation: Rodney A. Rountree, Francis Juanes.
Visualization: Rodney A. Rountree, Francis Juanes.
Writing – original draft: Rodney A. Rountree, Francis Juanes, Marta Bolgan.
Writing – review & editing: Rodney A. Rountree, Francis Juanes, Marta Bolgan.
References
1. Abbott CC. Traces of a voice in fishes. The American Naturalist. 1877; 11:147–156.
2. Stober QJ. Underwater noise spectra, fish sounds and response to low frequencies of Cutthroat trout
(Salmo clarki) with reference to orientation and homing in Yellowstone Lake. Trans. Am. Fish. Soc.
1969; 98(4):652–663.
3. Deichmann JL, Acevedo-Charry O, Barclay L, Burivalova Z, Campos-Cerqueira M, d’Horta F, et al. It’s
time to listen: there is much to be learned from the sounds of tropical ecosystems. Biotropica. 2018
Sep; 50(5):713–8.
4. Desjonquères C, Gifford T, Linke S. Passive acoustic monitoring as a potential tool to survey animal
and ecosystem processes in freshwater environments. Freshwater Biology. 2020; Jan; 65(1):7–19.
5. Gammell MP, O’Brien JM. Acoustic communication in aquatic animals: all quiet on the freshwater front?
Aquatic Conservation: Marine and Freshwater Ecosystems. 2013 Jun; 23(3):363–5.
6. Linke S, Gifford T, Desjonquères C, Tonolla D, Aubin T, Barclay L, et al. Freshwater ecoacoustics as a
tool for continuous ecosystem monitoring. Frontiers in Ecology and the Environment. 2018 May; 16
(4):231–8.
7. Mickle MF, Higgs DM. Integrating techniques: a review of the effects of anthropogenic noise on fresh-
water fish. Canadian Journal of Fisheries and Aquatic Sciences. 2017 Nov 7; 75(9):1534–41.
8. Risch D, Parks SE. Biodiversity assessment and environmental monitoring in freshwater and marine
biomes using ecoacoustics. Ecoacoustics. The Ecological Role of Sounds, edited byFarina A. and
Gage SH Oxford, UK: Wiley. 2017 Jul 24:145–68.
9. Rountree RA, Bolgan M, Juanes F. How Can We Understand Freshwater Soundscapes Without Fish
Sound Descriptions? Fisheries. 2019 Mar; 44(3):137–43. https://doi.org/10.1002/fsh.10190
10. Amoser S, Wysocki LE, Ladich F. Noise emission during the first powerboat race in an Alpine lake and
potential impact on fish communities. The Journal of the Acoustical Society of America. 2004 Dec; 116
(6):3789–97. https://doi.org/10.1121/1.1808219 PMID: 15658729
11. Amoser S, Ladich F. Are hearing sensitivities of freshwater fish adapted to the ambient noise in their
habitats? Journal of Experimental Biology. 2005 Sep 15; 208(18):3533–42.
12. Amoser S, Ladich F. Year-round variability of ambient noise in temperate freshwater habitats and its
implications for fishes. Aquatic Sciences. 2010 Jun 1; 72(3):371–8. https://doi.org/10.1007/s00027-
010-0136-9 PMID: 20922061
13. Bolgan M, Chorazyczewska E, Winfield IJ, Codarin A, O’Brien J, Gammell M. First observations of
anthropogenic underwater noise in a large multi-use lake. Journal of Limnology. 2016 Jun 14; 75(3):
644–651.
14. Lomask MR, Saenger RA. Ambient noise in a deep inland lake. The Journal of the Acoustical Society of
America. 1960 Jul; 32(7):878–83.
15. Roh HS, Sutin A, Bunin B. Determination of acoustic attenuation in the Hudson River Estuary by means
of ship noise observations. The Journal of the Acoustical Society of America. 2008 Jun 14; 123(6):
EL139–43. https://doi.org/10.1121/1.2908404 PMID: 18537300
16. Seppa¨nen J, Nieminen M. Measurements and descriptions of underwater noise in Finland. Geophysica.
2004; 40(1–2):23–38.
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 24 / 26
17. Tonolla D, Acuña V, Lorang MS, Heutschi K, Tockner K. A field-based investigation to examine under-
water soundscapes of five common river habitats. Hydrological Processes. 2010 Oct 30; 24(22):3146–
56.
18. Tonolla D, Lorang MS, Heutschi K, Gotschalk CC, Tockner K. Characterization of spatial heterogeneity
in underwater soundscapes at the river segment scale. Limnology and Oceanography. 2011 Nov; 56
(6):2319–33.
19. Vračar MS, MijićM. Ambient noise in large rivers (L). The Journal of the Acoustical Society of America.
2011 Oct; 130(4):1787–91. https://doi.org/10.1121/1.3628666 PMID: 21973331
20. Wysocki LE, Amoser S, Ladich F. Diversity in ambient noise in European freshwater habitats: Noise lev-
els, spectral profiles, and impact on fishes. The Journal of the Acoustical Society of America. 2007 May;
121(5):2559–66.
21. Desjonqueres C, Rybak F, Depraetere M, Gasc A, Le Viol I, Pavoine S, et al. First description of under-
water acoustic diversity in three temperate ponds. PeerJ. 2015 Nov 5; 3:e1393. https://doi.org/10.7717/
peerj.1393 PMID: 26587351
22. Desjonquères C, Rybak F, Castella E, Llusia D, Sueur J. Acoustic communities reflects lateral hydrolog-
ical connectivity in riverine floodplain similarly to macroinvertebrate communities. Scientific Reports.
2018 Sep 26; 8(1):14387. https://doi.org/10.1038/s41598-018-31798-4 PMID: 30258085
23. Johnson NS, Higgs D, Binder TR, Marsden JE, Buchinger T, Brege L, et al. Evidence of sound produc-
tion by spawning lake trout (Salvelinus namaycush) in lakes Huron and Champlain. Canadian Journal
of Fisheries and Aquatic Sciences. 2017 May 5; 75(3):429–38.
24. Linke S, Decker E, Gifford T, Desjonquères C. Diurnal variation in freshwater ecoacoustics: Implications
for site-level sampling design. Freshwater Biology. 2020 Jan; 65(1):86–95.
25. Lugli M. Sounds of shallow water fishes pitch within the quiet window of the habitat ambient noise. Jour-
nal of Comparative Physiology A. 2010 Jun 1; 196(6):439–51.
26. Morgan L. A Passive Acoustic and Experimental Study of Juvenile Blue Catfish, Ictalurus furcatus,
Sound Production and Agnostic Behavior in the Tidal Freshwater James River. M.Sc. Thesis, Virginia
Commonwealth University. 2014. https://scholarscompass.vcu.edu/etd/3459/
27. Straight CA, Freeman BJ, Freeman MC. Passive acoustic monitoring to detect spawning in large-bod-
ied catostomids. Transactions of the American Fisheries Society. 2014 May 4; 143(3):595–605.
28. Cox K, Brennan LP, Gerwing TG, Dudas SE, Juanes F. Sound the alarm: A meta-analysis on the effect
of aquatic noise on fish behavior and physiology. Global Change Biology. 2018 Jul; 24(7):3105–16.
https://doi.org/10.1111/gcb.14106 PMID: 29476641
29. Whitfield AK, Becker A. Impacts of recreational motorboats on fishes: a review. Marine Pollution Bulle-
tin. 2014 Jun 15; 83(1):24–31. https://doi.org/10.1016/j.marpolbul.2014.03.055 PMID: 24759512
30. Proulx R, Waldinger J, Koper N. Anthropogenic landscape changes and their impacts on terrestrial and
freshwater soundscapes. Current Landscape Ecology Reports. 2019 Sep 15; 4(3):41–50.
31. Anderson KA, Rountree RA, Juanes F. Soniferous fishes in the Hudson River. Transactions of the
American Fisheries Society. 2008; 137(2):616–626.
32. Bolgan M, O’Brien J, Winfield IJ, Gammell M. An investigation of inland water soundscapes: which
sonic sources influence acoustic levels? In: Proceedings of Meetings on Acoustics 2016 Jul 10 ( Vol.
27, No. 1, p. 070004). ASA.
33. Bolgan M, O’Brien J, Chorazyczewska E, Winfield IJ, McCullough P, Gammell M. The soundscape of
Arctic Charr spawning grounds in lotic and lentic environments: can passive acoustic monitoring be
used to detect spawning activities? Bioacoustics. 2018 Jan 2; 27(1): 57–85. http://dx.doi.org/10.1080/
09524622.2017.1286262
34. Cott PA, Mann DA, Higgs DM, Johnston TA, Gunn JM. Assessing disturbance from under-ice noise on
fishes in boreal lakes. In: The Effects of Noise on Aquatic Life 2012 (pp. 363–366). Springer, New
York, NY.
35. Giles J. The underwater acoustic repertoire of the long-necked, freshwater turtle Chelodina oblonga.
Ph.D. Dissertation, Murdoch University. 2005. https://researchrepository.murdoch.edu.au/id/eprint/39/
36. Giles JC, Davis JA, McCauley R, Kuchling G. The ambient sound field in three freshwater environ-
ments. In: Acoustics in a Changing Environment: Proceedings of the Annual Conference of the Austra-
lian Acoustical Society, Australian Acoustical Society, Castlemaine, pp. 383–389.
37. Holt DE, Johnston CE. Traffic noise masks acoustic signals of freshwater stream fish. Biological Con-
servation 2015; 187:27–33.
38. Hopson AM. Impacts of Anthropogenic Noise on Aquatic Invertebrates in Wetland Habitats. Ph.D. dis-
sertation, Kent State University. 2019. http://rave.ohiolink.edu/etdc/view?acc_num=
kent15645874111271
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 25 / 26
39. Johnson NS, Higgs D, Binder TR, Marsden JE, Buchinger T, Brege L, et al. Evidence of sound produc-
tion by spawning lake trout (Salvelinus namaycush) in lakes Huron and Champlain. Canadian Journal
of Fisheries and Aquatic Sciences. 2017 May 5; 75(3):429–38. https://doi.org/10.1139/cjfas-2016-0511
40. Kuehne LM, Padgham BL, Olden JD. The soundscapes of lakes across an urbanization gradient. PLoS
One. 2013 Feb 12; 8(2):e55661. https://doi.org/10.1371/journal.pone.0055661 PMID: 23424636
41. Martin SB, Cott PA. The under-ice soundscape in Great Slave Lake near the city of Yellowknife, North-
west Territories, Canada. Journal of Great Lakes Research. 2016 Apr 1; 42(2):248–55.
42. Martin SB, Popper AN. Short- and long-term monitoring of underwater sound levels in the Hudson River
(New York, USA). The Journal of the Acoustical Society of America. 2016; 139(4):1886–1897. https://
doi.org/10.1121/1.4944876 PMID: 27106335
43. Putland RL, Mensinger AF. Exploring the soundscape of small freshwater lakes. Ecological Informatics.
2020 Jan 1; 55:101018.
44. Rountree RA, Juanes F. Potential of passive acoustic recording for monitoring invasive species: fresh-
water drum invasion of the Hudson River via the New York canal system. Biological Invasions. 2017 Jul
1; 19(7):2075–88.
45. Rountree RA, Juanes F, Bolgan M. Air movement sound production by alewife, white sucker, and four
salmonid fishes suggests the phenomenon is widespread among freshwater fishes. PloS one. 2018
Sep 20; 13(9):e0204247. https://doi.org/10.1371/journal.pone.0204247 PMID: 30235287
46. Charif RA, Waack AM, Strickman LM. Raven Pro 1.4 user’s manual. Cornell Lab of Ornithology, Ithaca,
NY. 2010 May 22;25506974.
47. Hawkins AD, Popper AN. Assessing the impacts of underwater sounds on fishes and other forms of
marine life. Acoustics Today 2014; 10(2):30–41.
48. Merchant ND, Fristrup KM, Johnson MP, Tyack PL, Witt MJ, Blondel P, et al. Measuring acoustic habi-
tats. Methods in Ecology and Evolution. 2015 Mar; 6(3):257–65. https://doi.org/10.1111/2041-210X.
12330 PMID: 25954500
49. Rountree RA, Gilmore RG, Goudey CA, Hawkins AD, Luczkovich JJ, Mann DA. Listeningto fish: appli-
cations of passive acoustics to fisheries science. Fisheries. 2006 Sep 1; 31(9):433–46.
50. Correa JM, Sempere JT, Juanes F, Rountree R, Ruı´z JF, Ramis J. Recreational boat traffic effects on
fish assemblages: First evidence of detrimental consequences at regulated mooring zones in sensitive
marine areas detected by passive acoustics. Ocean & Coastal Management. 2019 Feb 1; 168:22–34.
https://doi.org/10.1016/j.ocecoaman.2018.10.027
51. Smott S, Monczak A, Miller ME, Montie EW. Boat noise in an estuarine soundscape–A potential risk on
the acoustic communication and reproduction of soniferous fish in the May River, South Carolina.
Marine Pollution Bulletin. 2018 Aug 1; 133:246–60. https://doi.org/10.1016/j.marpolbul.2018.05.016
PMID: 30041312
52. Haviland-Howell G, Frankel AS, Powell CM, Bocconcelli A, Herman RL, Sayigh LS. Recreational boat-
ing traffic: a chronic source of anthropogenic noise in the Wilmington, North Carolina Intracoastal
Waterway. The Journal of the Acoustical Society of America. 2007 Jul; 122(1):151–60. https://doi.org/
10.1121/1.2717766 PMID: 17614475
53. Rountree RA, Able KW. Spatial and temporal habitat use patterns for salt marsh nekton: implications for
ecological functions. Aquatic Ecology. 2007 Mar 1; 41(1):25–45.
54. Monczak A, Berry A, Kehrer C, Montie EW. Long-term acoustic monitoring of fish calling provides base-
line estimates of reproductive timelines in the May River estuary, southeastern USA. Marine Ecology
Progress Series. 2017 Oct 13; 581:1–9.
55. Monczak A, Ji Y, Soueidan J, Montie EW. Automatic detection, classification, and quantification of
sciaenid fish calls in an estuarine soundscape in the Southeast United States. PloS one. 2019 Jan 16;
14(1):e0209914. https://doi.org/10.1371/journal.pone.0209914 PMID: 30650120
56. Montie EW, Vega S, Powell M. Seasonal and spatial patterns of fish sound production in the May River,
South Carolina. Transactions of the American Fisheries Society. 2015 Jul 4; 144(4):705–16.
57. Gottesman BL, Francomano D, Zhao Z, Bellisario K, Ghadiri M, Broadhead T, et al. Acoustic monitoring
reveals diversity and surprising dynamics in tropical freshwater soundscapes. Freshwater Biology.
2020; 65(1):117–13218.
PLOS ONE
Freshwater soundscapes: A cacophony of biological sounds threatened by noise
PLOS ONE | https://doi.org/10.1371/journal.pone.0221842 March 18, 2020 26 / 26
Available via license: CC BY
Content may be subject to copyright.
