ArticlePDF Available

Air movement sound production by alewife, white sucker, and four salmonid fishes suggests the phenomenon is widespread among freshwater fishes


Abstract and Figures

We sought to describe sounds of some of the common fishes suspected of producing unidentified air movement sounds in soundscape surveys of freshwater habitats in the New England region of North America. Soniferous behavior of target fishes was monitored in real time in the field in both natural and semi-natural environments by coupling Passive Acoustic Monitoring (PAM) with direct visual observation from shore and underwater video recording. Sounds produced by five species including, alewife (Alosa pseudoharengus, Clupeidae), white sucker (Catastomus commersonii, Catostomidae), brook trout (Salvelinus fontinalis, Salmonidae), brown trout (Salmo trutta, Salmonidae), and rainbow trout (Oncorhynchus mykiss, Salmonidae) were validated and described in detail for the first time. In addition, field recordings of sounds produced by an unidentified salmonid were provisionally attributed to Atlantic salmon (Salmo salar, Salmonidae). Sounds produced by all species are of the air movement type and appear to be species specific. Our data based on fishes in three distinct orders suggest the phenomenon may be more ecologically important than previously thought. Even if entirely incidental, air movement sounds appear to be uniquely identifiable to species and, hence, hold promise for PAM applications in freshwater and marine habitats.
Content may be subject to copyright.
Air movement sound production by alewife,
white sucker, and four salmonid fishes
suggests the phenomenon is widespread
among freshwater fishes
Rodney A. RountreeID
*, Francis Juanes
, Marta Bolgan
1The Fish Listener, East Falmouth, Massachusetts, United States of America, 2Biology Department,
University of Victoria, Victoria, BC, Canada, 3Laboratoire de Morphologie Fonctionnelle et Evolutive,
´de Liège, Liège, Belgium
We sought to describe sounds of some of the common fishes suspected of producing
unidentified air movement sounds in soundscape surveys of freshwater habitats in the New
England region of North America. Soniferous behavior of target fishes was monitored in real
time in the field in both natural and semi-natural environments by coupling Passive Acoustic
Monitoring (PAM) with direct visual observation from shore and underwater video recording.
Sounds produced by five species including, alewife (Alosa pseudoharengus, Clupeidae),
white sucker (Catastomus commersonii, Catostomidae), brook trout (Salvelinus fontinalis,
Salmonidae), brown trout (Salmo trutta, Salmonidae), and rainbow trout (Oncorhynchus
mykiss, Salmonidae) were validated and described in detail for the first time. In addition,
field recordings of sounds produced by an unidentified salmonid were provisionally attrib-
uted to Atlantic salmon (Salmo salar, Salmonidae). Sounds produced by all species are of
the air movement type and appear to be species specific. Our data based on fishes in three
distinct orders suggest the phenomenon may be more ecologically important than previ-
ously thought. Even if entirely incidental, air movement sounds appear to be uniquely identi-
fiable to species and, hence, hold promise for PAM applications in freshwater and marine
Passive acoustic monitoring (PAM) has become an important tool for spatial and temporal
location of marine fishes based on their incidental and/or purposeful sound production [12].
However, little is known about sound production by freshwater fishes in the New England
region of North America [3]. While conducting a pilot survey of the soundscapes of freshwater
habitats in five regions of New England in the spring of 2008, we recorded possible air move-
ment sounds likely produced by alewife (Alosa pseudoharengus, Clupeidae) and various species
of unidentified salmonids [4]. We also recorded a wide variety of Fast Repetitive Tick (FRT)
PLOS ONE | September 20, 2018 1 / 32
Citation: Rountree RA, Juanes F, Bolgan M (2018)
Air movement sound production by alewife, white
sucker, and four salmonid fishes suggests the
phenomenon is widespread among freshwater
fishes. PLoS ONE 13(9): e0204247.
Editor: Bernd Sokolowski, University of South
Received: March 29, 2018
Accepted: September 4, 2018
Published: September 20, 2018
Copyright: ©2018 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: The authors confirm
that all data underlying the findings are fully
available without restriction. All relevant data are
within the Supporting Information files. Derived
data have been provided as an excel file in the
supplementary material.
Funding: The authors received no specific funding
for this work.
Competing interests: The authors have declared
that no competing interests exist.
sounds [5] from unknown sources with different tick decay patterns (rate of increase in the
interval between ticks), durations and frequency ranges. However, we could not validate the
identity of the sound sources based on the data collected at that time. We use the term “air
movement” sound to include a diverse array of sounds sometimes referred to as “air passage”
or “pneumatic” sounds that arise from a variety of mechanisms involving internal air move-
ment, and sometimes external air release, in physostomous fishes (e.g., [69]).
The fundamental goal of this paper was to confirm that the widespread occurrence of air
movement-like sounds previously recorded in the 2008 field survey [4] were indeed produced
by fishes, and to describe specific sounds produced by species suspected of producing them.
The specific aims of this study were to: i) describe the acoustic characteristics and the behav-
ioral association of air movement sounds in the alewife, the white sucker (Catastomus commer-
sonii, Catostomidae), and the brook, brown and rainbow trout (Salvelinus fontinalis,Salmo
trutta, and Oncorhynchus mykiss, Salmonidae, respectively) and ii) compare the acoustic fea-
tures of air movement sounds emitted by these species to determine if they are species specific,
and thus potentially useful in ecological studies based on PAM.
Materials and methods
Sampling locations and species identification
Fish behavior was observed in real time in the field in natural and semi-natural environments
by coupling PAM with direct visual observation from shore and with underwater video
recording. The species-specific identity of the sound emitter was validated through direct
observation or because observations were carried out where only one species was present.
Fishes were observed in private locations with permission from the owners and public loca-
tions that did not require permissions. Animals were observed remotely and not disturbed in
any way. No permitting or ethics oversight approval was necessary. Field observations were
made on wild fish in seven locations within Massachusetts and Maine, and on captive fish held
in semi-natural conditions at the Blue Stream Aquaculture facility in Barnstable, Massachu-
setts (detailed description of sampling locations and methods are provided in S1 Appendix).
Sounds were recorded with either an uncalibrated HTI-96-MIN (High Tech Industries, Gulf-
port, MS; sensitivity = -165 dB re: 1 V/μPa, frequency response: 2 Hz to 30 kHz) or SQ26-08
(Sensitivity = -169.00 re. 1V/μPa rms, Cetacean Research Technology, Seattle, WA) hydro-
phone. Ambient aerial sounds and voice notes were recorded to a second channel from a
microphone to verify if unknown sounds originated underwater. Underwater video was
observed with one or more underwater cameras mounted in a fixed location on the bottom or
suspended within the water column. Post-processing of acoustic signals was conducted by lis-
tening to all recordings in their entirety while simultaneously viewing the sound’s spectrogram
(1,024 FFT, Hanning window, 50% overlap) and waveform with Raven Pro 1.5 acoustic soft-
ware [10].
Sounds were categorized into three types: 1) surface event, 2) fish sounds, and 3) air bubble
sounds. Surface events were identified when a fish rose to the surface to gulp air often creating
a splash or jump sound. Although such sounds are not generally considered as “fish sounds”
we consider them a critical component of air movement sound production behavior and
potentially useful in fish identification and informative of behavior and physiological differ-
ences among species. Sounds produced by the fish following the surface event were classified
as fish sounds, except for air bubble sounds which were difficult to detect and appeared to be
incidental to sound production. Acoustic measurements of selected parameters of all sound
types were made in Raven following Charif et al. [11]. Parameter definitions are provided in S1
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 2 / 32
It was sometimes possible to attribute a continuous train of sounds as a “fish sound series”
produced by a single fish, either through direct observation, or temporal isolation from other
sounds. The fish sound series is analogous to a fish call series, but we refrain from labeling it as
such because that implies a communication function which has not been established. In some
cases, we were able to identify the surface event preceding the fish sound series and recognized
a “surface event sound series” which includes the fish sound series and is also attributable to a
single fish. For each fish sound series, the following metrics were measured (Fig 1): i) number
of individual sounds (excluding bubble sounds); ii) duration (the time between the start of the
first sound to the end of the last sound; iii) sound rate (the number of sounds in a fish sound
series divided by the series duration), and iv) sound period (series duration divided by the
number of sounds). For the surface event sound series, the following additional metrics were
measured; i) duration (the time between the start of the surface event to the end of the last fish
sound); ii) latency (the time elapsed between the start of the surface event and the start of the
first fish sound).
To provide additional insight into the sound production mechanisms, the pulse structure of
known bubble sounds was compared with that of individual ticks of FRT sounds recorded dur-
ing this study, and with those of Pacific herring (Clupea pallasii, Clupeidae) recorded with the
same type of hydrophone and recording system (sound samples provided by Amalis Riera,
University of Victoria, Canada).
Statistical analysis
To provide the best descriptions of individual sound types, data were pooled over all measure-
ments for each species because many more sounds were observed than could be reliably attrib-
uted to a single sound series. However, the determination of the most common sounds
observed for each species was based only on the subset of sounds that could be attributed to
Fig 1. Sound series measurement definitions. The surface event sound series includes the surface event, latency, fish
sound series, and incidental bubble sounds. In this example from a white sucker, the fish sound series consists of 11
sounds (2 snitches, 3 other, and 5 snorts). The surface event series duration is measured from thestart of the surface
event to the end of the last fish sound. Latency is measured as the duration from the start of the surface event to the start
of the first fish sound. In this example, the latency is measured from the start of the surface event to the start of the
“snitch” sound. The fish sound series duration is measured between the start of the first fish sound to the end of the last
fish sound. Note, that when the surface event could not be detected, only the fish sound series measurements could be
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 3 / 32
individual fish sound series, thus it represents the types of sounds most likely to be produced
by the species rather than the most numerous sound type.
In order to explore the potential of PAM for surveying specific species based on air move-
ment sounds, selected acoustic parameters were statistically compared among species. We
tested for univariate among-species differences in sound series parameters (surface event series
duration, latency, fish sound series duration, number of fish sounds, fish sound period and
rate) with a one-way analysis of variance (ANOVA) for each variable using SAS/STAT soft-
ware, Version 12.1 [12]. Similarly, we tested for among-species differences in selected sound
parameters for the most common sound types. Different sets of species were included in the
analyses depending on the sound type, as many sounds were not shared among species. For
both sets of analyses variables were first normalized with the most appropriate transformation
based on a maximum likelihood test. In a few cases statistical outliers were omitted.
Canonical Discrimination Analysis (CDA) and Multivariate Analysis of Variance (MAN-
OVA) were used to test for multivariate differences in acoustic parameters among species for
the most common sound types after first conducting a stepwise discriminant analysis to select
a subset of the variables for group discrimination [1215]. The CDA tests whether species can
be distinguished by a sound type, and if so, which acoustic parameters contribute the most to
the observed differences (for example, a CDA on the surface event sound compares the acous-
tic parameters of the surface event sound among all species, since all exhibited that type of
sound, however, the CDA on bubble sounds only compares the three species that exhibited
them). A second CDA analysis compared all species by including all sounds they exhibited,
and thus takes into account differences in sound types among species. Pearson correlations of
the original (but transformed to normalize) variables with the derived canonical variables
from the CDA analyses were calculated to determine which sound parameter contributed
most to the group discrimination [1415].
A total of 56.5 h (3,388 min) of observations were made, in which we measured the acoustic
parameters for 1,195 sounds attributed to the study species (S1 Appendix). A total of 160 fish
sound series were identified of which 117 had associated surface events (so there were 117 sur-
face event series, Table 1).
Over 30 h of observations (1,857 min) were made of alewife behavior (S1 Appendix), however
to reduce the possibility of inclusion of sounds from blueback herring (Alosa aestivalis, Clupei-
dae), we excluded sound data collected after May 2 from the measurement data. Alewife obser-
vations were usually done during the afternoon daylight hours through the early evening.
Typically, from a few to a dozen or more individuals occupied the observation chambers at
any given time in both the Mill Creek and Webber Pond raceways as they rested prior to mov-
ing into the adjacent ponds. The mill pool typically held from dozens to hundreds of alewives.
In contrast, Sevenmile Brook below the Webber Pond dam held thousands of fish. Fish
appeared to segregate into small groups, or “pods”, of up to a dozen individuals.
Alewife were frequently observed to make a series of sounds after rising to the surface to
gulp air (Fig 2A,S1 Audio,S1 and S2 Videos). Individuals would make a rapid dash to the sur-
face and create a small splash, occasionally jumping, as they gulped air. They then rapidly dove
to resume swimming at the same depth they started from. Sounds were primarily produced
after resuming their previous swimming depth. Of 54 surfacing events recorded in field notes,
34 (63%) were followed by sound production. In many cases silent bubbles escaped from the
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 4 / 32
mouth and gills during the initial descent from the surface. However, once the fish sounds
began, a faint “pop” sound could sometimes be detected as one or more bubbles escaped from
the gills after each sound. On a few occasions, a sound series concluded with a loud pop as a
single large bubble was released from the mouth (Fig 2A,S1 Video). The bubble sounds were
difficult to detect and were distinct from high frequency burst sounds produced by the alewife
which were labeled “coughs” and “snitches”. Although the cough and snitch sounds were asso-
ciated with air bubble release, air bubble release alone did not produce them. For example,
Table 1. Sound series statistics.
Species Statistic Surface event series duration Latency Fish sound series duration Number of fish sounds Fish sound rate Fish sound period
Alewife N 21 21 33 33 33 33
min 1.15 0.35 0.08 1.00 0.50 0.08
max 11.66 4.14 8.45 16.00 12.05 2.01
SE 0.54 0.21 0.33 0.61 0.38 0.06
mean 5.39 1.19 3.29 5.94 2.47 0.57
White sucker N 11 11 12 12 12 12
min 8.02 1.47 1.73 3.00 0.28 0.58
max 28.62 4.46 24.89 11.00 1.73 3.56
SE 1.90 0.24 2.07 0.66 0.12 0.26
mean 20.52 2.58 16.75 7.08 0.56 2.33
Brook trout N 4 4 6 6 6 6
min 0.83 0.45 0.14 1.00 0.66 0.14
max 12.35 12.08 4.57 3.00 7.35 1.52
SE 2.88 2.98 0.70 0.33 0.90 0.21
mean 6.62 6.24 1.12 1.67 3.36 0.50
Brown trout N 4 4 26 26 26 26
min 0.36 0.29 0.07 1.00 0.57 0.07
max 9.36 8.38 5.23 3.00 13.33 1.74
SE 1.88 1.78 0.23 0.11 0.71 0.09
mean 4.14 3.31 1.31 2.19 3.85 0.56
Rainbow trout N 42 42 42 42 42 42
min 0.75 0.32 0.06 1.00 0.26 0.06
max 7.20 6.50 3.81 7.00 16.67 3.81
SE 0.23 0.20 0.15 0.20 0.55 0.09
mean 2.77 1.68 1.09 1.93 3.87 0.56
N 35 35 41 41 41 41
min 0.81 0.45 0.11 1.00 0.48 0.11
max 19.25 4.66 17.30 10.00 9.43 2.10
SE 0.51 0.19 0.45 0.30 0.23 0.08
mean 4.22 1.63 2.88 2.93 1.56 0.95
MANOVA      
Means comparison WS >all WS >all WS >all WS = A >all WS >all WS >all
Sample size 117 117 160 160 160 160
Comparison of sound series parameters among species. MANOVA = results of a Multivariate Analysis of Variance
= p <0.05
 = p <0.01
= p <0.0001.
WS = white sucker, A = alewife, all = remainder species. N = sample size, min = minimum, max = maximum, SE = standard error of the mean
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 5 / 32
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 6 / 32
four of 19 field observations of air bubble release resulted in no detectable sound production
(21%). We therefore considered these faint bubble sounds to be incidental to the production of
the cough and snitch sounds and excluded them from the fish sound series. Although readily
visible to an observer from shore, these events were very difficult to capture on video. How-
ever, we were successful in capturing video of alewife sound production (S1 Audio,S1 and S2
Videos), as well as air bubble release without detectable sound production (S3 and S4 Videos)
on several occasions. No obvious reaction of conspecifics to alewife sound production was
Alewife sound series had an average of 5.9 fish sounds (range 1–16) and a mean fish sound
rate of 2.47 sounds/s (Table 1). Alewife surface event series duration averaged 5.39 s while the
latency before sound production ranged from 0.35 to 4.14 s and averaged 1.19 s. The most fre-
quently occurring sounds in an alewife sound series (Table 2) were the “cough” (91%) and
“snitch” (21%). Air bubbles were not included as part of the sound series but were acoustically
detected in 20 series (61%) and occurred an average of 0.36 s after each cough or snitch (range
0.026–0.986 s, standard error = 0.034 s). Coughs had a mean peak frequency of 1,258 Hz,
bandwidth of 3,609 Hz and duration of 0.043 s (Fig 2A, 2B and 2D,Table 3).
Means (and standard error of the mean) of selected acoustic parameters measured for the
most common sound types for each species (1094 sounds out of 1197). Characteristics of the
air gulp event (surface splash or jump) and individual bubble release sounds are also provided.
FRT = fast repetitive tick, VFRT = very fast repetitive tick.
The much less frequently observed snitches (Fig 2C and 2E) had a similar bimodal fre-
quency structure, but were weaker and shorter in duration. FRT-like sounds (examples in Fig
3A and 3B) occurred in only 9% of the sound series (Table 2) and were characterized by a lon-
ger duration (mean = 1.3 s) and higher frequency (mean peak frequency = 6,254 Hz, Table 3).
Bubble sounds (example Fig 4B) had a higher peak frequency (mean = 1,716 Hz), and shorter
duration (mean = 0.009 s), than the coughs and snitches (Fig 2A,Table 3).
White sucker
Behavior of white sucker was monitored over about 16 h (958 min) across 11 dates at the
Stony Brook Herring run, and for an additional 7.5 h (450 min) during spawning events at the
Stony Brook Herring run and Webber Pond sites (S1 Appendix). Mill pool held only a few
white sucker at any one time. At dusk, sedentary individuals that had been resting on the bot-
tom became more active, and then would rise to the surface and gulp air in a loud splash or
jump. They could then be observed to trail bubbles from the mouth and gills as they
descended. As with alewife, sound production occurred after the individual had returned to
the bottom, or to a level swimming location above the bottom. We did not observe air release
during sound production but cannot rule out the possibility due to the difficulty of observing
individuals in the limited visibility at dusk and the limited video field of view. White sucker
sounds were only heard after a surface event just prior to or after sunset.
During observations of actively spawning white sucker in both the Stony Brook and Web-
ber Pond sites, two to four groups of three to five males and a single large female were observed
to repeatedly spawn throughout the late afternoon and well after dark. Thrashing and rattling
sounds could clearly be heard as females deposited eggs in the stream or pond bed. At Webber
Fig 2. Alewife sounds. Example of a surface event sound series and the two most frequently observed fish sounds. (A) Spectrogram of a surface
event sound series (S1 Audio,S1 and S2 Videos), (B) waveform of a cough, (C) waveform of a snitch, (D) spectrogram of the cough in B, and (E)
spectrogram of the snitch in C. Figure labels: “Surface” = air gulping surface event, “B” = bubble sound, “C” = cough fish sound, “O” = other fish
sound, “M” = mouth bubble sound. Spectrogram parameters: unfiltered, 1,024-point Hann windowed FFTs with 50% overlap. Fish sound
waveforms were filtered around 600 Hz to 4000 Hz.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 7 / 32
Table 2. Most common sounds.
Sound type Frequency Percent
Alewife N = 33
Cough 30 91
Snitch 7 21
Other 5 15
FRT-like 3 9
Snort 2 6
Surface 21 64
Bubble 20 61
Mouth bubble 2 6
White sucker N = 12
Snort 12 100
Other 4 33
Snitch 4 33
FRT-like 1 8
Sneeze 1 8
Surface 11 92
Bubble 7 58
Brook trout N = 6
VFRT 3 50
Other 3 50
Snitch 2 33
FRT long 1 17
Snort 1 17
Surface 4 67
Brown trout N = 26
VFRT 25 96
Other 8 31
Chirp 7 27
Bubble FRT 1 4
200 Hz FRT 1 4
Splash 4 15
Bubble 1 4
Rainbow trout N = 42
Gurgle 23 55
Other 9 21
VFRT 7 17
Snitch 7 17
Chirp 3 7
Long FRT 3 7
FRT-like 2 5
Surface 42 100
Unknown salmonid N = 41
Moan 14 34
Other 13 32
Snort FRT 6 15
FRT-like bubbles 5 12
FRT-like 4 10
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 8 / 32
pond, spawning occurred within inches of the shore in still water as shallow as 10–15 cm,
while at Stony Brook spawning occurred in fast flowing riffles over gravel. In between spawn-
ing bouts, females would move to quieter water and rest on the bottom for a period prior to
the next spawn. Males would sometimes accompany the resting female, but some continued to
swim actively in the area. Groups of spawning individuals did not exhibit either the air gulping
or sound production behaviors even at sunset in contrast to the non-spawning individuals
located a short distance away (<20 m) within the mill pool.
White sucker produced 3 to 11 loud “snort” sounds (mean = 7) with a latency of 2.6 s
after the surface event, and a mean fish sound series duration of 16.75 s (Fig 5A–5E,
Table 1,S2 Audio). Snorts occurred in all white sucker sound series, while snitches occurred
in 33% (Table 2). Surface event sounds were detected in 92% and bubble sounds in 58% of
the sound series. Snorts had a mean peak frequency of 1,954 Hz, mean bandwidth of 12,285
Hz, and duration of 0.175 s (Fig 5B and 5D;Table 3), while snitches had a mean peak fre-
quency of 2,291 Hz, bandwidth of 10,348 Hz, and duration of 0.083 s (Fig 5C and 5E;
Table 3). Bubbles sounds had a mean peak frequency of 1,548 Hz, bandwidth of 2,794 Hz,
and duration of 0.014 s (Table 4,Fig 4C). FRT-like sounds occurred in 8% of the sound
series (Table 2). No overt reaction of conspecifics was observed to white sucker sounds,
but observations were hampered by low visibility and large separation distances among
Brook trout
Brook trout were observed for 5.4 h (322 min) over 3 dates in December 2014 (S1 Appendix).
Brook trout sounds were infrequent despite the high stocking density in the raceway. Only 6
sound series with one to three fish sounds and a 6 s latency, were positively identified due to
the high fish density and limited camera field of view in the raceway (Tables 1and 2,Fig 6A,
S3 Audio). Fish were observed to surface and expel air from gills as they descended followed
by sounds after the fish had returned to the bottom, or its original swimming position in the
water column. Gas release was not observed in association with sounds but cannot be ruled
out due to the difficulty of following individuals after the air gulp event. The most frequently
observed sound was a short duration FRT-like sound we term a “Very Fast Repetitive Tick”
(VFRT) with a mean peak frequency of 4,993 Hz, bandwidth of 10,407 Hz, and duration of
0.096 s (Tables 2and 3,Fig 6B and 6D). The next most common sound was named a “snitch”
which had similar properties as the VFRT except for a narrower bandwidth of 5,166 Hz and
more burst-like waveform lacking the repetitive ticks (Tables 2and 3,Fig 6C and 6E).
Although we observed bubble release immediately after surfacing, brook trout bubble sounds
were not acoustically detected, and no bubbles were observed in association with any brook
trout sound. No overt reaction of conspecifics to fish sounds was observed.
Table 2. (Continued)
Sound type Frequency Percent
Snort 4 10
Jump 35 85
Bubbles 10 24
Sound types most frequently exhibited by a species, including surface event and bubble types, based on
measurements from fish sound series (N = number of sound series, FRT = fast repetitive tick, VFRT = very fast
repetitive tick)
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 9 / 32
Table 3. Acoustic characteristics of the common sounds.
Frequency statistics
Type Sample
Low 1st
Peak Center 3rd quartile High 90%
bandwidth 90%
Cough 154 603 (36) 1064 (48) 1258 (61) 1489 (60) 1979 (77) 4212 (149) 2183 (106) 3609 (140) 0.03 (0.002) 0.043
Snitch 27 733 (98) 1243 (132) 1434 (172) 1529 (166) 2069 (249) 3682 (396) 1755 (297) 2949 (358) 0.014 (0.004) 0.02 (0.005)
Other 9 817 (106) 1796 (191) 1864 (183) 2197 (228) 2750 (364) 4966 (744) 2614 (581) 4149 (708) 0.062 (0.024) 0.094
FRT-like 7 2884 (704) 4258
7700 (2563) 11641
7031 (3011) 8756 (3660) 1.234 (0.558) 1.373
Snort 7 463 (115) 1118 (212) 1198 (272) 1392 (255) 2156 (473) 5320 (709) 2571 (631) 4857 (690) 0.066 (0.005) 0.096
Unknown 5 1207 (246) 2418 (349) 3253 (901) 3290 (880) 4068 (1120) 9497
5287 (2559) 8289 (3844) 0.618 (0.437) 0.745
Splash 19 747 (164) 1741 (374) 1939 (261) 2407 (471) 3658 (662) 10189
4894 (794) 9441 (2422) 0.204 (0.050) 0.262
Bubble 59 890 (58) 1571 (99) 1716 (140) 1966 (145) 2412 (169) 4410 (285) 1917 (216) 3520 (272) 0.003 (0.001) 0.009
2 1596 (281) 1875 (187) 1992 (304) 2320 (117) 3140 (796) 4253
1664 (1054) 2656 (1844) 0.139 (0.107) 0.169
White sucker
Snort 60 756 (63) 1534 (111) 1954 (134) 2260 (165) 3554 (266) 13042
5829 (613) 12285
0.13 (0.012) 0.175
Snitch 9 929 (219) 1791 (225) 2291 (279) 2385 (264) 3875 (552) 11277
5145 (572) 10348
0.06 (0.020) 0.083
Sneeze 5 840 (121) 3956
26943 (2665) 45906 (118) 0.116 (0.021) 0.131
Other 3 941 (267) 1500 (162) 2218 (125) 2062 (143) 2468 (136) 3250 (198) 1656 (250) 2308 (176) 0.046 (0.005) 0.059
Splash 6 312 (185) 718 (289) 1187 (292) 1218 (322) 2343 (1003) 26436
5828 (2199) 26123
0.248 (0.090) 0.391
Jump 5 442 (192) 1237 (439) 1949 (826) 1725 (645) 2437 (777) 24730
5100 (1215) 24288
0.271 (0.133) 0.438
Bubble 23 808 (133) 1406 (208) 1548 (221) 1581 (205) 2013 (191) 3603 (226) 1288 (243) 2794 (225) 0.007 (0.002) 0.014
Bubble like 15 279 (25) 450 (33) 687 (185) 750 (165) 1156 (193) 2454 (358) 1518 (303) 2174 (370) 0.031 (0.008) 0.038
Brook trout
VFRT 19 2378 (141) 4332 (377) 4993 (400) 5531 (696) 6962 (1072) 12785
6809 (1417) 10407
0.058 (0.005) 0.096
Snitch 12 2654 (288) 4023 (382) 4617 (588) 4773 (498) 5664 (593) 7820 (878) 3742 (610) 5166 (770) 0.074 (0.010) 0.099
Chirp 4 3673
5648 (1313) 6485
2109 (728) 2811 (735) 0.131 (0.040) 0.2 (0.051)
FRT 3 2295 (663) 4312 (614) 3562 (248) 5593
6656 (1647) 11997
6406 (2890) 9702 (3228) 0.297 (0.117) 0.378
Snort 3 1888 (324) 2843 (706) 2937 (773) 3031 (706) 3406 (596) 4383 (894) 1843 (450) 2495 (570) 0.112 (0.036) 0.177
Splash 7 2214 (668) 4888 (771) 5758 (857) 6977
9321 (1786) 26683
15696 (3312) 24469
0.181 (0.034) 0.239
Brown trout
VFRT 216 2568 (60) 4187 (84) 4582 (101) 5056 (104) 6131 (137) 12260
5191 (218) 9692 (464) 0.078 (0.003) 0.111
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 10 / 32
Brown trout
A total of 5.8 h (348 min) of observations of brown trout were made on two dates in December
2014 (S1 Appendix). Additional video observations were recorded on 25 October 2016. Brown
trout were stocked at a similar density to the brook trout but produced many more sounds.
Air gulping and sound production appeared to increase at dusk. Typically, individuals would
rise slowly from a bottom resting position, or midwater swimming position, before accelerat-
ing slightly to gulp air usually producing little splash or sounds. Only 4 surface events were
Table 3. (Continued)
Frequency statistics
Type Sample
Low 1st
Peak Center 3rd quartile High 90%
bandwidth 90%
Chirp 9 2279 (210) 3729 (199) 4760 (436) 5000 (248) 6697 (425) 9333 (714) 5479 (562) 7054 (748) 0.065 (0.010) 0.082
Other 7 2834 (756) 4151 (979) 4566
4781 (970) 5625 (1005) 8138
3455 (724) 5303 (940) 0.12 (0.027) 0.146
Unknown 7 1195 (446) 1794 (480) 2035 (533) 2544 (794) 3656 (1448) 5222
2892 (1178) 4027 (1431) 0.846 (0.527) 0.979
Splash 11 1340 (315) 4926 (906) 7678
19321 (3481) 40970
0.241 (0.071) 0.373
Bubble 3 431 (215) 843 (324) 1031 (378) 1281 (360) 1562 (409) 9861
1750 (307) 9430 (1585) 0.004 (0.002) 0.016
Rainbow trout
Gurgle 54 1442 (51) 2175 (57) 2409 (86) 2505 (74) 2954 (98) 3911 (146) 1779 (107) 2468 (142) 0.278 (0.039) 0.337
VFRT 19 2089 (209) 3651 (378) 4199 (533) 4741 (667) 6024 (949) 12775
6029 (1347) 10686
0.088 (0.016) 0.129
Snitch 14 1969 (133) 3187 (332) 3877 (440) 3850 (451) 5176 (733) 8903
4794 (1068) 6934 (1632) 0.05 (0.009) 0.068
Chirp 10 1875 (94) 2531 (99) 2793 (180) 2953 (168) 3674 (388) 5320 (711) 2737 (638) 3444 (746) 0.195 (0.069) 0.247
Other 8 1309 (147) 2027 (101) 2062 (217) 2601 (214) 3363 (482) 4855 (988) 2660 (894) 3546 (1002) 0.137 (0.030) 0.185
Splash 49 1573 (55) 3587 (170) 4302 (338) 5491 (313) 8274 (364) 37069
15059 (660) 35495
0.268 (0.018) 0.399
Unknown salmonid
Gurgle 58 421 (19) 662 (24) 748 (29) 796 (28) 934 (35) 1694 (62) 745 (41) 1272 (67) 0.313 (0.043) 0.401
Snort 32 495 (12) 729 (10) 829 (26) 820 (13) 949 (25) 1691 (66) 638 (47) 1195 (66) 0.111 (0.011) 0.141
Moan 15 445 (35) 818 (88) 962 (122) 943 (110) 1087 (110) 1983 (319) 756 (127) 1538 (310) 0.338 (0.052) 0.438
Other 13 467 (69) 923 (114) 1031 (129) 1247 (146) 1600 (160) 7273
1838 (319) 6805 (3467) 0.326 (0.162) 0.571
Snort FRT 10 375 (65) 1031 (147) 1218 (203) 1218 (155) 1640 (213) 4111 (520) 1593 (226) 3736 (547) 0.142 (0.025) 0.19 (0.029)
Snitch 6 1004 (204) 1718 (258) 2281 (272) 2296 (187) 2937 (446) 4886 (602) 2531 (625) 3882 (649) 0.052 (0.013) 0.076
FRT-like 5 908 (286) 1293 (291) 1481 (236) 1443 (294) 1837 (429) 3535 (627) 1593 (422) 2626 (473) 0.314 (0.088) 0.389
Jump 72 310 (16) 661 (23) 865 (75) 919 (40) 1388 (80) 10912
2790 (205) 10601
0.551 (0.026) 0.821
Bubbles 13 507 (80) 894 (114) 1168 (167) 1060 (137) 1348 (135) 3719 (648) 980 (143) 3211 (682) 0.194 (0.056) 0.286
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 11 / 32
Fig 3. Comparison of “FRT” sounds. (A) and (B) Alewife, (C) brook trout, (D) brown trout, and (E) rainbow trout. Spectrograms
in each graph are unfiltered and set to the same parameters (1024 Hann windowed FFTs with 50% overlap), time scale (0 to 1.4 s),
and frequency scale (0 to 12 kHz).
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 12 / 32
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 13 / 32
acoustically detected out of 26 fish sound series (Table 1). Diffuse streams of air bubbles of
varying size could sometimes be observed escaping from the mouth and gills as the fish
descended, but such gas release usually did not produce detectable sounds (example S5
Video). However, a unique sound we term a “gill-bubble FRT” was recorded on several occa-
sions simultaneously with bubble release as a brown trout dove from the surface (Fig 7A,S4
Audio,S6 and S7 Videos). Unlike the diffuse bubble streams usually observed during the dive
for brown trout, as well as all other species, the FRT-like sound was produced by uniform
sized bubbles in a single file that rapidly streamed from either the right or left gill cover, but
not both. Shortly after settling, individuals were sometimes observed to simultaneously release
one to a few bubbles from each gill, often with a yawn which may have been coincident with
VFRT sound production. We refer to this event as a “double gill bubble release”. One individ-
ual was observed to surface and produce sounds twice within a 45 s period.
Brown trout sound series averaged 2.2 fish sounds, a 3.3 s latency, and were 1.3 s in dura-
tion (Table 1,Fig 7A,S4 Audio,S6 and S7 Videos). A well-defined VFRT was the most fre-
quently occurring sound occurring in 96% of the sound series (Fig 7B and 7D,Table 2), and
was characterized by a mean peak frequency of 4,582 Hz and bandwidth of 9,692 Hz (n = 216,
Table 3). In several instances, the VFRT appeared to be produced just prior to the double gill
bubble release event, but more data are needed to confirm this. A bird-like chirp (Fig 7C and
7E) was the second most frequent sound (27%) in the brown trout series (Table 2) and had a
peak frequency of 4,760 Hz (n = 9, Table 3). It should be noted that simultaneous recording
with a microphone confirms the bird-like chirp sound is produced underwater and not from
an aerial source. Bubble sounds were acoustically detected in only 4% of the brown trout
sound series (Table 2), and had a mean peak frequency of 1,031 Hz, bandwidth of 9,430 Hz,
and duration of 0.016 s (Table 3). No bubble release was ever observed for any sound other
than uncommon gill-bubble FRTs and double gill bubble release events. No overt reaction
among conspecifics to brown trout sounds were observed.
Rainbow trout
Rainbow trout were observed for 3.2 h (193 min) over two dates in December 2014 (S1 Appen-
dix). Unlike brown and brook trout, surface events were acoustically detected for all sound
series (N = 42, Tables 1and 2). Rainbow trout sound series averaged 1.93 fish sounds, a dura-
tion of 1.1 s, and a latency of 1.7 s (Table 1,Fig 8A,S5 Audio). The most frequently observed
sound was described as a “gurgle” which occurred in 55% of the series (Table 1), and had a
peak frequency of 2,409 Hz, bandwidth of 2,468 Hz, and duration of 0.34 s (Table 3,Fig 8A–
8D,S5 Audio). A VFRT similar to that observed in brook and brown trout occurred in 17% of
the series (Table 2,Fig 8C and 8E), and had a mean peak frequency of 4,199 Hz, mean band-
width of 10,686 Hz, and mean duration of 0.13 s (Table 3). No bubble release was observed for
any sound type. No overt reaction by conspecifics to rainbow trout sounds were observed.
Unidentified salmonid
Sounds of an unidentified salmonid were recorded in the Presumpscot River over a 3.6 h
period on 29 May 2014 (S1 Appendix). Small groups of brook trout were observed among the
Fig 4. Comparison to bubble sounds. Waveforms of individual air bubble sounds, and individual FRT ticks are
compared on the same 500 ms time scale after filtering around 500 Hz to 6000 Hz (amplitudes are relativeand not
directly comparable). (A) Single bubble from bottom sediment gas seep, (B) single bubble from alewife gills, (C) single
bubble from white sucker, (D) two ticks from an alewife FRT, (E) three ticks from a rainbow trout FRT, (F) three ticks
from a brown trout bubble FRT, (G) multiple ticks from a brown trout VFRT, and (H) four ticks from a Pacific herring
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 14 / 32
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 15 / 32
rocks by an observer from a 2 m high vantage point directly overlooking the site and with a
video camera (S1 Appendix). No other species was observed on camera, and several fly fisher-
men in the area indicated they had only caught brook trout. Occasionally brook trout could be
observed from shore as they rose to the surface to gulp air with a small splash and then
returned to the bottom. Air bubbles could sometimes be observed flowing from the fish as it
descended. Sounds followed after the individual returned to the river bed. Unfortunately, we
were not able to capture a surface event on the underwater video camera. At dusk, unidentified
salmonids began to periodically jump creating a loud splash followed by gurgle sounds and an
unusual moan sound. Individuals observed while it was still light appeared to be salmonids but
could not be positively identified. Jumps and associated sounds increased in frequency after
Surface events were exclusively jumps and were acoustically detected in 85% of 41 detected
sound series (Tables 1and 2). Sound event series were characterized by an average of 2.9
sounds after a 1.6 s latency (Table 1,Fig 9A,S6 Audio). Gurgle (63%) and moan (34%) sounds
Fig 5. White sucker sounds. Example of a surface event sound series and the two most frequently observed fish sounds. (A) Spectrogram of a
surface event sound series (see S2 Audio), (B) waveform of a snort, (C) waveform of a snitch, (D) spectrogram of the snort in B, and (E)
spectrogram of the snitch in C. Figure labels: “Surface” = air gulping surface event, “S” = snort, “Gull” = herring gull sound heard underwater.
Spectrogram parameters: unfiltered, 1,024-point Hann windowed FFTs with 50% overlap. Fish sound waveforms were filtered around 700 Hz to
13000 Hz.
Table 4. Comparison of common sounds among species.
Sound type and species included
Bubbles Chirp VFRT Gurgle Snitch Snort Surface event
Low A>WS = T BK>R ns R>TA = WS<BK = R WS>T BK = R<A = WS = BN = T
Q1 A = WS>TBK = BN>R ns R>TA = WS<BK = R WS>TA = WS = T<BK = BN = R
Peak ns BK = BN>R ns R>TA = WS<BK = R WS>TA = WS = T<BK = BN = R
Center ns BK = BN>R ns R>TA<WS = BK = R WS>TWS = T<A<BR = BN = R
Q3 A>WS = T BK = BN>R ns R>TA<WS = BK = R WS>TWS = T<A<BR = BN = R
High ns BN>BK = R ns R>TA<WS = BK = R WS>TA = T<WS = BK = BN = R
90% Bandwidth A>WS = T BN>BK = R ns R>TA<WS = BK = R WS>TA = WS = T<BK = BN = R
Bandwidth ns BN >BK = R ns R>TA<WS = BK = R WS>TA = T<WS = BK = BN = R
90% duration A = WS<TBN>BK = R ns ns A<WS = R<BK ns T>A = WS = BR = BN = R
Duration A<WS<TBN<BK = R ns ns A<WS = BK = R ns T>A = WS = BR = BN = R
MANOVA (p)      
Analysis of variance (ANOVA) comparison of sound parameters among species for selected common sound types. Variables were normalized by appropriate
transformations indicated by a maximum likelihood test. In a few cases statistical outliers were omitted. For variables with significant among species main effects a SNK
means comparison tested for specific differences among species (for example BK = BN>R indicated that the mean for rainbow trout was significantly less than for
brown and brook trout which were not different). Because some species did not exhibit a sound type, or sample sizes were too low, species included in the ANOVA for
each type are indicated under the type. MANOVA indicates the results of a multivariate comparison among species based on all variables. A = alewife, WS = white
sucker, BK = brook trout, BN = brown trout, R = rainbow trout, T = unknown salmonid. Q1 = mean first quartile frequency, Q3 = mean 3rd quartile frequency.
= p <0.05
 = p <0.01.
= p <0.001.
ns = nonsignificant.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 16 / 32
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 17 / 32
were the most frequently observed sound types and had mean peak frequencies of 748 Hz and
962 Hz, respectively (Tables 2and 3,Fig 9). Bubble sounds were detected in 24% of the sound
series (Table 2).
Statistical comparison of sounds among species
There were strong qualitative differences in air movement sounds across species. Some sound
types were unique to particular species: coughs to alewife, and moans to the unidentified sal-
monid. The VFRTs were only produced by brook, brown and rainbow trout, and gurgles only
by rainbow trout and the unknown salmonid. When the entire surface event sound series typi-
cal of a species is considered, the differences are pronounced. We found significant species dif-
ferences in the surface event series and fish sound series attributes (Table 1). White sucker had
significantly higher event series duration, latency, sound series duration, and fish sound
period, and lower fish sound rate, than all other species. The number of fish sounds in a sound
series was higher for white sucker and alewife than all other species. Note that among species
differences for surface event series duration and latency were not tested due to the low sample
size, and fish sound period was omitted because it is statistically redundant with fish sound
A few of the most common sound types, bubbles, chirps, VFRTs, gurgles, snitches, snorts
and surface events (= jumps + splash + surface) were tested for differences in univariate and
multivariate acoustic parameters among the species which exhibited them (in some cases a
species may have been dropped due to a low sample size, Table 4). The MANOVA indicated
significant difference among species for all sound types tested (Table 4). Multivariate CDA
analysis resulted in significant discrimination among species for bubble, chirp, snitch and
surface sound types (Table 5,Fig 10A–10D). Bubble sounds were only weakly discriminant
among species (Fig 10A). Differences in the Q1 frequency, 90% bandwidth, 90% duration
and duration drove a significant discrimination in chirp sounds among the three trout spe-
cies (Table 5,Fig 10B). The 5% and 95% frequency percentiles, IQR duration, and duration
drove a significant discrimination among four species (Fig 10C). Only surface events could
be compared among all species (Table 5,Fig 10D). Brown, brook and rainbow trout tended
to separate from alewife, white sucker and the unknown salmonid, with weaker separation
among unknown salmonid, alewife and white sucker (Table 5,Fig 10D). The statistical dis-
crimination among species in their surface sounds was driven most strongly by the 95% fre-
quency percentile, 90% bandwidth, and duration (Table 5). The unknown salmonid surface
event was significantly longer in duration than all other species (Table 4). No significant
univariate differences among VFRT sounds of the three known trout species were observed
(Table 4).
When data are pooled over all sound types, CDA analysis reveals strong discrimination
among species groups (Table 5,Fig 11). The first canonical explains 71% of the variance while
the second only 22%. Center, peak and first quartile (Q1) frequency (positive) and duration
and 90
percentile bandwidth (negative) drive the discrimination along the first canonical
axis. As might be expected, brook and brown trout overlap but discriminate from rainbow
trout, all three are strongly different from alewife and white sucker. Perhaps unexpectedly, the
unknown salmonid discriminates strongly from the three trout species.
Fig 6. Brook trout sounds. Example of a surface event sound series and the two most frequently observed fish sounds. (A) Spectrogram of a
surface event sound series (see S3 Audio), (B) waveform of a FRT, (C) waveform of a snitch, (D) spectrogram of the FRT in B, and (E)
spectrogram of the snitch in C. Figure labels: “Surface” = air gulping surface event. Spectrogram parameters: unfiltered, 1,024-point Hann
windowed FFTs with 50% overlap. Fish sound waveforms were filtered around 2300 Hz to 13000 Hz.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 18 / 32
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 19 / 32
Bubble comparison
The comparison of the pulse structure of air bubble sounds with individual ticks in the various
FRT-like sounds indicates a strong similarity (Fig 4). Both bubbles and FRT ticks show a simi-
lar damping oscillation pattern. However, alewife cough and snitch sounds, and trout gurgle
and chirp sounds do not appear similar to bubbles in their pulse structure (compare Fig 4 with
Figs 2B and 2C and 7C and 8B). White sucker snort and snitch sounds (Fig 5B and 5C) show
some similarity to bubbles, but the relationship is uncertain.
This study represents the first descriptions of sound production in the alewife, the white
sucker, the brook trout and the brown trout. Limited descriptions of rainbow trout sound pro-
duction have previously been reported [1618]. These five species belong to three different
orders of fishes: Clupeiformes, Cypriniformes, and Salmoniformes, respectively, suggesting
that air movement sound production is widespread within physostomous fishes.
The most frequently recorded alewife sounds in this study were the “cough” and “snitch”. The
cough trains produced by alewife after gulping air (Table 4) are not similar to sounds reported
for other clupeiform fishes which typically primarily produce FRTs [5,78,19]. However, ale-
wife do occasionally (9% of the sound series) produce FRT sounds similar to those reported
for Clupea spp. and Sardinops sagax. Although the mechanism by which the cough and snitch
sounds are produced is uncertain, it appears that each cough is produced within the pharyn-
geal or branchial chambers just prior to emission of one or more gas bubbles from one or both
gill covers. However, it is clear that release of gas through the gills or mouth alone does not
produce sounds similar to either the coughs, snitches or FRTs (compare S1 and S2 Videos
exhibiting strong cough sounds, with S3 and S4 Videos where bubbles are released with almost
no detectable sound). We suggest that the air bolus in the pharyngeal or branchial chambers
amplifies stridulation sounds produced internally within the chambers in a way similar to that
hypothesized for the croaking tetra (Glandulocauda inequalis =Mimagoniates inequalis, Char-
acidae, Characiformes) [20]. Gas release and sound production do not appear to be caused by
changes in pressure with depth, as most sounds are produced after the fish has returned to its
previous swimming depth and in very shallow water.
Although we did not observe overt reactions of conspecifics to alewife sounds, the sounds were
well within the species’ hearing range [2123]. The observations that sounds are only produced
after some air gulp events could be interpreted as possible evidence for voluntary sound produc-
tion, or it could simply be that the fish sometimes has to expend more effort to expel the gas (i.e.,
it is literally an involuntary cough to expel gas trapped within the gill chamber).
White sucker
Our observations of air movement sound and associated behavior for white sucker are similar
to those reported for the Eastern creek chubsucker (Moxostoma oblongus, Catostomidae; [24].
Fig 7. Brown trout sounds. Example of a surface event sound series and the two most frequently observed fish sounds. (A) Spectrogram of a
surface event sound series (see S4 Audio,S6 and S7 Videos), (B) waveform of a FRT, (C) waveform of a chirp, (D) spectrogram of the FRT in B,
and (E) spectrogram of the chirp in C. Figure labels: “Surface” = air gulping surface event, “Gill bubble FRT” = sound of rapid bubble release
from gill covers. Spectrogram parameters: unfiltered, 1,024-point Hann windowed FFTs with 50% overlap. Fish sound waveforms were filtered
around 2300 Hz to 13000 Hz.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 20 / 32
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 21 / 32
Chubsucker were reported to produce four or five sounds in a single bout, whereas white
sucker were found to produce sounds in bouts of 3 to 11 snorts in this study (Table 1). Abbot
[24] described chubsucker sounds in association with air-bubble discharge, however, it is not
clear if they were coincident with bubble emission or occurred after bubble release as in the
white sucker.
The only other sounds known to be produced by catostomids are those of redd cutting dur-
ing spawning [25]. River redhorse (Moxostoma carinatum, Catostomidae) and robust redhorse
(Moxostoma robustum, Catostomidae) spawning activities were recorded in Georgia (USA),
and the authors proposed that passive acoustic surveys could be used to document spatial and
temporal distribution patterns of spawning for these and other endangered or threatened
catostomids based on the redd cutting sounds. Their failure to report air movement, or air
gulping behaviors, for the two redhorse species is consistent with our observations that white
sucker do not exhibit these behaviors during active spawning.
The behavior of the trout and unidentified salmonid observed in this study is similar to that
observed for other salmonids [6,8,9,17]. Stober [6] notes that a visible trail of bubbles was
emitted from the gill of cutthroat trout, Oncorhynchus clarkii, as they dove after rising to the
surface and creating a splash. Squawks (600–800 Hz) usually occurred several seconds after the
fish had returned to its original position, which is very similar to our observations for brook,
brown, rainbow trout, and the unidentified salmonid. Similarly, the behaviors associated with
Arctic charr (Salvelinus alpinus) sound production [9] are similar to those observed for trout
in this study and agree with the general pattern of air movement sound production described
herein (Fig 12).
The unidentified salmonid sounds recorded from the Presumpscot River are provisionally
attributed to landlocked Atlantic salmon (Salmo salar, Salmonidae) which are abundantly
stocked at the sampling location. The attribution to Atlantic salmon is supported by similari-
ties with other salmon species and significant differences with the three trout species. The
qualitative description of salmon sounds (Salmo and Oncorhynchus) provided by Neproshin
and Kulikova [17] were similar to the unidentified salmonid sounds. They reported that
salmon make sounds in the form of splashes when emerging from the surface followed by
croaking, rumbling and whistle (similar to our moan) noises from air passage. More recently
Kuznetsov [8] presents data from chum salmon (Oncorhynchus keta, Salmonidae) and pink
salmon (Oncorhynchus gorbuscha, Salmonidae). He reports a sound duration of chum salmon
of 250–1,750 ms, and peak frequency in two modes at 100–330 Hz and 450–740 Hz. Pink
salmon had shorter sound durations of 420–800 ms, and slightly higher peak frequency modes
of 200–400 Hz and 420–950 Hz. The dominant frequency of Atlantic salmon is similar to that
reported for chum and pink salmon [8], and significantly lower than that of the three captive
trout species we recorded (Tables 3and 4). Further, although brook trout were also abundant
at the Presumpscot River site, the behavior and sounds of captive trout were different from
those observed for the unidentified salmonid. First, the dramatic moan sounds were unique to
fish in the river. Second, the gurgle sounds produced by rainbow trout were of significantly
higher frequency than the unidentified salmonid gurgles, and rainbow trout are not stocked in
Fig 8. Rainbow trout sounds. Example of a surface event sound series and the two most frequently observed fish sounds. (A) Spectrogram of a
surface event sound series (see S5 Audio), (B) waveform of a gurgle, (C) waveform of a FRT, (D) spectrogram of the gurgle in B, and (E)
spectrogram of the FRT in C. Figure labels: “Surface” = air gulping surface event. Spectrogram parameters: unfiltered, 1,024-point Hann
windowed FFTs with 50% overlap. The gurgle waveform was filtered around 1400 Hz and 3900 Hz, while the FRT waveform was filtered around
2000 Hz to 12000 Hz.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 22 / 32
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 23 / 32
the river. Lastly, there was a clear change in behavior of the unidentified salmonid at dusk with
individual fish beginning to jump. Although surface events did appear to increase at dusk for
the captive trout, the increase was not as dramatic as in the river, and jumping was not com-
mon. However, since sound production of Atlantic salmon has not been previously reported,
further research is needed to confirm the river fish sounds recorded in this study were indeed
made by the species.
The potential of PAM for behavioral, ecological, and conservation monitoring studies is
increasingly being recognized, and applied both to purposeful sounds as well as incidental
sounds produced by fishes (see review in [1]). Since air passage sounds appear to be common
among Salmonidae, we reiterate previous calls for the use of PAM in field and laboratory stud-
ies of salmonids (e.g., [6,89,26]). Although the observed qualitative and quantitative differ-
ences among the sound series of brook trout, brown trout, rainbow trout and Atlantic salmon
strongly suggests that each species can be identified in the wild by their sounds, more research
is needed to quantify sound characteristics and to confirm species specific characteristics.
We suggest that measurements of the pulse structure of the VFRT sounds (e.g., number of
ticks, tick rate, period, frequency, etc.) may be particularly useful for PAM. In addition, several
of the observed behavioral and acoustic differences among the salmonid species have potential
Fig 9. Unknown salmonid sounds. Example of a surface event sound series and the two most frequently observed fish sounds. (A) Spectrogram
of a surface event sound series (sound S6 Audio), (B) waveform of a gurgle, (C) waveform of a moan, (D) spectrogram of the gurgle in B, and (E)
spectrogram of the moan in C. Figure labels: “Jump” = air gulping surface event involving a leap from the water. Spectrogram parameters:
unfiltered, 1,024-point Hann windowed FFTs with 50% overlap. The gurgle waveform was filtered around 300 Hz and 3000 Hz, while the moan
waveform was filtered around 200 Hz to 2000 Hz.
Table 5. Results of canonical discrimination analysis.
Bubble sounds Chirp sounds Snitch sounds Surface sounds All sounds
Variable Can1 Can2 Can1 Can2 Can1 Can2 Can1 Can2 Can1 Can2
Low frequency 0.32 0.45
5% frequency 0.80 -0.43 0.42 0.52 0.74 0.35
Q1 frequency 0.62 0.30 0.76 0.21 0.85 0.33
Center frequency 0.91 0.26
Peak frequency 0.43 0.29 0.86 0.32
Q3 frequency 0.80 ns 0.84 0.18
95% frequency 0.65 ns 0.91 0.15 0.84 0.20
High frequency 0.71 0.38 0.81 ns 0.67 0.19
IQR bandwidth 0.74 0.25
90% bandwidth 0.42 0.22 0.63 -0.49 0.87 0.15 0.69 0.11
Bandwidth 0.78 ns 0.57 0.11
IQR duration ns ns 0.50 ns -0.34 0.52
90% duration -0.66 0.33 -0.47 ns -0.41 0.67
Duration -0.72 0.39 -0.50 ns 0.78 ns -0.53 0.44 -0.43 0.73
Canonical R
0.34 0.28 0.78 0.50 0.75 0.41 0.78 0.40 0.65 0.59
Proportion of variance (%) 57 43 78 22 80 18 72 14 71 22
Canonical P<0.0001 0.0001 0.0001 0.0149 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
Species effect P<0.0001 0.0001 0.0001 0.0001 0.0001
Results of Canonical discrimination analyses (CDA) among species for selected sound types based on transformed sound measurements. Blank fields indicate the
variable was not included in the analysis. Species effect = significance level for a multivariate difference among species. ns = variables that were included in the model
but were not significantly correlated with the canonical. Q1 = mean first quartile frequency, Q3 = mean 3rd quartile frequency, IQR = interquartile.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 24 / 32
for use in PAM surveys. All frequency and sound duration parameters exhibited differences
among the four salmonid species (Tables 3and 4,Fig 10C). Multivariate analysis based on data
pooled over all sound types suggests that brook and brown trout sound parameters overlap but
differ strongly from rainbow trout and Atlantic salmon which were different from each other
(Fig 10). Rainbow trout and Atlantic salmon were more likely to make a detectable splash
sound than either brook or brown trout, and their most frequent sound was the gurgle. Brown
trout were the most prolific sound producers among the captive trout but tended to make
quiet air gulps and almost always produced VFRT sounds. Brook trout were the least sound
productive but produced similar VFRT and snitch sounds as those of the brown trout.
Fig 10. Comparison of selected sounds among species. Canonical discrimination analysis of species groups for selected variables (see Table 5
for canonical statistics): (A) differences among species based on bubble sound parameters; (B) species differences based on chirp sound
parameters; (C) species differences based on the snitch sound parameters; (D) species differences based on the surface event sound.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 25 / 32
Generalized behavior cycle
Based on our observations we can construct a generalized air movement sound production
behavior for physostomus fishes that includes a rise to the surface, air gulp event, dive to
depth, and resumption of normal behavior at a presumed acclimation depth (Fig 12). Studies
of air movement sound production should examine the entire cycle to better understand the
behavior. Differences among species in behavior at each step can be informative. For example,
differences in the surface event and latency among the three trout species, suggests differences
in the physiological process of filling the gas bladder or different adaptations to predation risk
at the surface. The presence or absence of gas release during any stage has implications for the
sound production mechanism. For example, we did not see evidence of FRT sound production
arising from gas release due to pressure changes during either the ascent or descent stages. In
fact, most sound production occurred at the end of the dive after the fish had settled to the bot-
tom or assumed a stable swimming depth where it appeared to be neutrally buoyant which we
refer to as the “acclimation” depth.
What is a FRT?
The results of this study, and the qualitative comparisons of previously reported FRT sounds,
reveal a diverse array to the form of FRT-like sounds. In general, FRTs consist of an initial
Fig 11. Comparison among species based on all sounds. Canonical discrimination analysis of species groups based on transformed variates for all
sound types. The mean canonical and 95% confidence boundary for the mean are shown. Variables that loaded highest on the canonicals, and
canonical statistics are provided in Table 5. AW = alewife, BR = brook trout, BN = brown trout, RT = rainbow trout, WS = white sucker, and
SA = unknown salmonid.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 26 / 32
broadband burst pulse where ticks are too close to resolve, followed by a train of ticks with a
decaying tick interval and tick bandwidth. However, the initial broadband burst is not always
present, and the tick train can be either very fast and short in duration (e.g., the VFRT), or
long (as much as 10 s or more). It is also likely that some of the burst pulse sounds we have
observed are FRTs that lack the tick train.
The relationship between FRT sound production and air passage is uncertain; some
researchers have observed FRT sounds coincident with air emission from the anal vent (e.g.,
[5,19,27]) while others report FRT sounds without air emission [89,1718,28]. The qualita-
tive comparison of the waveform of various bubble and FRT sounds suggests that they have a
similar pulse structure (Fig 4). Furthermore, the gill-bubble FRT suggests a relationship
between bubble release and some kinds of FRT-like sounds that do not arise from air passage
from the anal vent. It seems most likely, based on these observations and those of most past
researchers, that bubble release is incidental to FRT sound production, sometimes occurring
and sometimes absent, and that the sounds are often produced by internal movements of gas.
However, more data are needed on the precise timing of the sounds and the air bubble release
events. In addition, most authors fail to examine the relationship between air gulping and
sound production in any detail; the quantification of the full cycle of air gulping behavior (Fig
12) in relation to sound production would provide important insights to further elucidate FRT
Fig 12. Sound production behavior. Schematic illustration of generalized air movement sound production behavior.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 27 / 32
production mechanisms, as well as other behaviors related to air gulping such as buoyancy
compensation. It is also suggested that the temporal patterns of the FRT sounds (e.g., tick
interval decay rate, frequency decay rate) may be useful in species identification, as well as to
elucidate interspecies differences in the mechanism of FRT production. Therefore, studies that
more clearly define what a FRT is, on the basis of its acoustic properties, and a classification of
different FRT types, would be of great aid in studies of fish behavior and contribution to the
freshwater soundscape.
Other types of sounds associated with air gulping
Although the FRT sounds have recently captured the attention of both scientists and the pub-
lic, they are just one of a wide diversity of air movement sounds that were observed in this and
previous studies. All six species observed during this study occasionally produced FRT sounds
similar to herring, however, other types of air movement sounds were far more prevalent
(including the VFRT). In a companion survey of the soundscape of freshwater habitats in the
northeastern United States [4], air movement sounds other than FRTs were an order of magni-
tude more numerous (0.239 sounds/min) and occurred at more recording locations (45%)
compared to FRT sounds (0.028 sounds/min and 24% of locations). Both types of air move-
ment sounds were major components of the biological soundscape comparable to the contri-
bution of insects and other fish sounds (i.e., conventionally recognized stridulation and gas
bladder sounds).
The observations of air bubble release from the gills after alewife cough sounds, and possi-
bly after trout VFRT sounds, suggests that the sound originates inside the gill chamber, per-
haps by vibrating the gill rakers or other internal structures. The weaker alewife snitch sounds,
and frequent observation of a lack of sound production with air bubble release, might result
from an insufficient pressure of the moving air bolus over the internal structures. Acoustic
similarities between the alewife cough and snitch sounds and the trout VFRT sounds support
this hypothesis, as do Nelson’s [20] observations of air movement sound production in the
croaking tetra. Thus, if confirmed by further study, a similar mechanism may have evolved
independently in at least three distinct orders of fishes.
It is possible that the gurgle sound and the gill-bubble FRT sounds may be related but fur-
ther data are necessary to evaluate their relationship. It should be noted that in the companion
study of the freshwater soundscapes of New England [4], gurgle sounds were very common,
but were underestimated due to their similarity to highly variable sounds of gas seeps from the
sediment. Quantification of acoustic differences between gas seeps and air movement sounds
like the gurgle and gill-bubble FRT are necessary before investigators can use passive acoustics
to map spatial and temporal distribution patterns of these types of sounds. Moans of the
unknown salmonid are among the most unique sounds produced by fishes, and thus, if further
research confirms the hypothesis that they are produced by Atlantic salmon, they may be use-
ful in PAM studies of Atlantic salmon behavior.
Although not a type of air movement sound, splash and jump sounds associated with the
air gulping behavior were among the most discriminant among species, suggesting that the
sounds made by fish splashing and jumping, when associated with air gulping, may be identifi-
able to species in certain circumstances. In addition, the hypothesized communicative func-
tion of sturgeon jumping [29] may extend to other species. It is notable that the sounds
produced by the surface gulp are broadband and extend into the lower frequencies and may be
audible to some fishes. Jumps may be particularly effective as a mechanism to communicate
fitness since the airborne duration is likely correlated with swimming speed and power of the
individual. Jumping sounds may be considered under the category of percussive sounds,
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 28 / 32
usually produced by bumping or slapping the bottom sediment or hard structures [7] and
studies that actually examine jumping/splashing sounds should be undertaken to examine
their potential communication function. Unfortunately, few studies have examined jumping
behavior, and only Sulak et al. [29] considers sound function. For example, Soares and Bier-
man [30] recently examined aerial jumping in the Trinidadian guppy (Poecilia reticulata, Poe-
ciliidae) and concluded that it was a deliberate strategy for dispersal and dismissed the
possibility of a sound function, although they apparently did not examine jump sounds
Whether fishes can hear air movement sounds is uncertain due to a lack of data on auditory
threshold studies of most species. However, the assumptions of similar hearing abilities among
species within families are problematic due to high intra- and inter-specific variation, and thus
research to quantify hearing in physostomous fishes is critically needed. For example, although
some clupeids can hear high frequency sounds, others apparently do not [31]. In fact, C.palla-
sii has an upper auditory threshold of about 5,000 Hz indicating that they may hear only the
lower frequency components of their FRT sounds. However, Mann et al. [31] conclude that
this is sufficient for FRT sound detection. Although an overt reaction of conspecifics to alewife
sounds was not observed, the cough and snitch sounds were well within the species’ hearing
range [2123]. The snort sounds of the white sucker were often dramatic against the back-
ground noise and are within the hearing range of cyprinids and catostomids [22].
The low upper frequency auditory thresholds of the few salmonids studied to date [3233]
suggest that a social communication function of FRT sounds is unlikely in the group, but
other air movement sounds and jump sounds may be audible to some species. Stober [6] tried
different conditioning methods on cutthroat trout and found considerable intraspecific varia-
tion with startle responses at frequencies as high as 650 Hz but averaged 443 Hz. Some of the
salmonid sounds reported in this study, particularly for the unidentified salmonid, had signifi-
cant energy at these frequencies (Table 5). In a detailed study of Arctic charr spawning behav-
ior and sound production Bolgan et al. [9] concluded that air movement sounds likely did not
play a role in communication.
However, even if air movement sounds are in fact incidental in the majority of physosto-
mous fishes, this study suggests that air movement sounds can be used to identify species and
thus are a potentially important tool in conservation, resource management and ecological
studies utilizing PAM techniques. One of the most important uses of PAM is in studies of ani-
mal behavior in the wild. For example, we observed different air gulping behavior among the
species ranging from nearly undetectable surface events to vigorous splashing and jumping.
Such different behavior might arise from differences in predation risk, or perhaps energy bud-
gets. Differences in latency times and sound production might provide information on physio-
logical adaptations for gas bladder inflation or deflation. The use of PAM to identify habitat
and locations where fish are present is especially promising at night and in locations where
water depth or clarity render visual surveys ineffective. Studies to determine the frequency of
air gulping behavior in individual fish are needed to determine the potential of PAM of air
movements sounds for fish census. For example, if individual fish only gulp air once at sunrise
and sunset to inflate and deflate their gas bladders, then one fish sound series would represent
one fish.
Supporting information
S1 Appendix. Detailed methodology. Detailed description of sampling locations and meth-
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 29 / 32
S1 Audio. Alewife sounds. Recording of alewife sound series shown in Fig 2A.
S1 Video. Alewife sound production behavior. Movie of alewife behavior while producing
the sound series shown in Fig 2A.
S2 Video. Alewife behavior in slow motion. Movie of alewife behavior while producing the
sound series shown in Fig 2A. The video is slowed to half speed to clarify behavior and the rela-
tionship between sounds and bubble release.
S3 Video. Alewife bubble release. Movie of alewife behavior and bubble release in which only
weak bubble sounds were acoustically detected (compare with S1 Video).
S4 Video. Alewife bubble release in slow motion. Movie of alewife behavior and bubble
release in which only weak bubble sounds were acoustically detected. Video slowed to half
speed to clarify behavior and sounds at the time of bubble release (compare with S3 Video).
S2 Audio. White sucker sounds. Recording of white sucker sound series shown in Fig 5A.
S3 Audio. Brook trout sounds. Recording of brook trout sound series shown in Fig 6A.
S5 Video. Brook trout sound production behavior. Movie of brown trout behavior during
bubble release when fish sounds were not detected.
S4 Audio. Brown trout sounds. Recording of brown trout sound series shown in Fig 7A.
S6 Video. Brown trout sound production behavior. Movie of brown trout behavior during
production of the sound series shown in Fig 7A.
S7 Video. Brown trout behavior in slow motion. Movie of brown trout behavior during pro-
duction of the sound series shown in Fig 7A. Video slowed to half speed to clarify the relation-
ship between bubble release and sound production.
S5 Audio. Rainbow trout sounds. Recording of rainbow trout sound series shown in Fig 8A.
S6 Audio. Attributed atlantic salmon sounds. Recording of unknown salmonid sound series
shown in Fig 9A. Sounds from the unknown salmonid are provisionally attributed to Atlantic
S1 Data. Raw measurement data. Data file containing acoustic measurements for each sound
and sound series. Data compiled from Raven Pro 1.5 acoustic software [11] selection tables
and edited for clarity.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 30 / 32
The authors wish to thank Alexander Kasumyan (Moscow State University) for assistance with
Russian literature. Joe Olson, Cetacean Research Technology, provided acoustic technical sup-
port. Blue Stream Aquaculture generously provided access to their facilities. Michael Fine pro-
vided much needed encouragement and stimulating conversations on the phenomenon of air
movement sounds in fishes.
Author Contributions
Conceptualization: Rodney A. Rountree, Francis Juanes.
Data curation: Rodney A. Rountree.
Formal analysis: Rodney A. Rountree, Marta Bolgan.
Funding acquisition: Rodney A. Rountree, Francis Juanes.
Investigation: Rodney A. Rountree.
Methodology: Rodney A. Rountree, Francis Juanes.
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.
Writing – original draft: Rodney A. Rountree, Francis Juanes, Marta Bolgan.
Writing – review & editing: Rodney A. Rountree.
1. Rountree RA, Gilmore RG, Goudey CA, Hawkins AD, Luczkovich J, & Mann D. Listeningto Fish: appli-
cations of passive acoustics to fisheries science. Fisheries 2006; 31: 433–446.
2. Luczkovich JJ, Mann DA, Rountree RA. Passive acoustics as a tool in fisheries science. Transactions
of the American Fisheries Society. 2008 Feb 1; 137(2):533–41.
3. Anderson KA, Rountree RA, Juanes F. Soniferous fishes in the Hudson River. Transactions of the
American Fisheries Society. 2008 Feb 1; 137(2):616–26.
4. Rountree RA, Juanes F, Bolgan M. Fish farts, burps, moans and groans; a cacophony of murmurs in
danger of being lost before they are recognized by society. Scientific Reports
5. Wilson B, Batty RS, Dill LM. Pacific and Atlantic herring produce burst pulse sounds. Proceedings of the
Royal Society of London B: Biological Sciences. 2004 Feb 7; 271(Suppl 3):S95–7.
1098/rsbl.2003.0107 PMID: 15101430
6. Stober QJ. Underwater noise spectra, fish sounds and response to low frequencies of cutthroat trout
(Salmo clarkii) with reference to orientation and homing in Yellowstone Lake. Transactions of the Ameri-
can Fisheries Society 1969; 98:652–663.
7. Kasumyan AO. Sounds and sound production in fishes. Journal of Ichthyology. 2008 Dec 1; 48
8. Kuznetsov MY. Traits of acoustic signalization and generation of sounds by some schooling physosto-
mous fish. Acoustical Physics. 2009 Nov 1; 55(6):866.
9. Bolgan M, O’Brien J, Rountree RA, Gammell M. Does the Arctic charr Salvelinus alpinus produce
sounds in a captive setting? Journal of Fish Biology. 2016 Sep 1; 89(3):1857–65.
1111/jfb.13067 PMID: 27349486
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 31 / 32
10. Bioacoustics Research Program. Raven Pro: Interactive Sound Analysis Software, version 1.5. Ithaca,
NY: The Cornell Lab of Ornithology. Available from
11. Charif RA, Strickman LM, Waack AM. Raven Pro 1.4 User’s Manual. The Cornell Lab of Ornithology,
Ithaca, NY. 2010.
12. SAS S, Guide SU. Cary, NC: SAS Inst. 2012.
13. Pielou EC. The interpretation of ecological data: a primer on classification and ordination. John Wiley &
Sons; 1984 Sep 6.
14. Harris RJ. A primer of multivariate statistics. Psychology Press; 2001.
15. Friendly M. SAS system for statistical graphics. SAS Publishing; 1991 Nov 1.
16. Neproshin A. Some physical characteristic of sound in Pacific salmons (in Russian). Zoologichesky
Zhurnal 1972; 51:1025–30.
17. Neproshin A, Kulikova W. (1975). Sound-producing organs in salmonids. Journal of Ichthyology 1975;
18. Phillips MJ. The feeding sounds of rainbow trout, Salmo gairdneri Richardson. Journal of Fish Biology.
1989 Oct 1; 35(4):589–92.
19. Wahlberg M, Westerberg H. Sounds produced by herring (Clupea harengus) bubble release. Aquatic
Living Resources. 2003 Jul 1; 16(3):271–5.
20. Nelson K. The evolution of a pattern of sound production associated with courtship in the characid fish,
Glandulocauda inequalis. Evolution. 1964 Dec 1; 18(4):526–40.
21. Mann DA, Lu Z, Hastings MC, Popper AN. Detection of ultrasonic tones and simulated dolphin echolo-
cation clicks by a teleost fish, the American shad (Alosa sapidissima). The Journal of the Acoustical
Society of America. 1998 Jul; 104(1):562–8. PMID: 9670546
22. Mann DA, Cott PA, Hanna BW, Popper AN. Hearing in eight species of northern Canadian freshwater
fishes. Journal of Fish Biology. 2007 Jan 1; 70(1):109–20.
23. Popper AN, Plachta DT, Mann DA, Higgs D. Response of clupeid fish to ultrasound: a review. ICES
Journal of Marine Science. 2004 Jan 1; 61(7):1057–61.
24. Abbott CC. Traces of a voice in fishes. The American Naturalist 1877; 11:147–56.
25. 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 1; 143(3):595–605. https://
26. 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.
27. Mu¨ller J. U
¨ber Fische, welche To
¨ne von sich geben und die Entstehung dieser To
¨ne. Arch. Anat. Phy-
siol. Wiss. Med. 1857:249–79.
28. Schwarz AL, Greer GL. Responses of Pacific herring, Clupea harengus pallasi, to some underwater
sounds. Canadian Journal of Fisheries and Aquatic Sciences. 1984 Aug 1; 41(8):1183–92.
29. Sulak KJ, Edwards RE, Hill GW, Randall MT. Why do sturgeons jump? Insights from acoustic investiga-
tions of the Gulf sturgeon in the Suwannee River, Florida, USA. Journal of Applied Ichthyology. 2002
Dec 1; 18(4-6):617–20.
30. Soares D, Bierman HS. Aerial jumping in the Trinidadian guppy (Poecilia reticulata). PloS one. 2013
Apr 16; 8(4):e61617. PMID: 23613883
31. Mann DA, Popper AN, Wilson B. Pacific herring hearing does not include ultrasound. Biology Letters.
2005 Jun 22; 1(2):158–61. PMID: 17148154
32. Van Derwalker JG. Response of salmonids to low frequency sound. Marine Bio-acoustics. 1967; 2:45–
33. Hawkins AD, Johnstone AD. The hearing of the Atlantic salmon, Salmo salar. Journal of Fish Biology.
1978 Dec 1; 13(6):655–73.
Air movement sounds in three orders of fish
PLOS ONE | September 20, 2018 32 / 32
... Although many of these studies do not include an underwater recording, several could meet our inclusion criteria. Similarly, some fish species identified by Rountree et al. (2018a) and Rountree et al. (2018b) were missed by our search. Nevertheless, the search terms used do capture the most significant studies in the field of freshwater bioacoustics and provide a representative sample of the literature. ...
... well-studied orders of temperate freshwater fish, with qualitatively similar results to those from this systematic review. Specifically, Rountree et al. (2018b) and Rountree et al. (2018a) reported that 68 species of fish in 12 orders have been studied, including 28 species of Cypriniformes in four families, 18 species of Perciformes in five families, 11 species of Salmoniformes in one family, and 11 species of Acipenseriformes in one family. (Aiken, 1982a,b) Arthropod Palmacorixa nana Corixidae (Hemiptera) (Aiken, 1982a,b) Arthropod Micronecta batilla Corixidae (Hemiptera) Bailey (1983) Arthropod Callicorixa praeusta, Sigara striata Corixidae (Hemiptera) (Finke, 1968) Arthropod Sigara striata Corixidae (Hemiptera) (Finke and Prager, 1980) Arthropod Corixa panzeri, C. dentipes, C. ...
... of biological sound production, and (b) studies that investigated anthropogenic and physically generated sound and their effects on aquatic animals.Linke et al. (2018) also noted that passive acoustic monitoring is becoming increasingly popular with freshwater ecologists, but knowledge of sounds produced by freshwater species is relatively scarce.Rountree et al. (2018b) andRountree et al. ...
Full-text available
Globally, freshwater biodiversity has declined by 81% in the last 45 years. Furthermore, wetlands are disappearing three times faster than rainforests. In the UK, 75% of ponds have been lost due to land-use change in the last century. It is therefore imperative to act quickly to conserve freshwater biodiversity. Central to the effort to conserve biodiversity is the effort to monitor biodiversity. Many conventional survey methods have been developed to quantify trends in biodiversity, however, they are often labour-intensive and invasive. Two novel non-invasive survey methods, environmental DNA (eDNA) and ecoacoustics, have been shown to exploit an aquatic medium to effectively survey marine biodiversity due to the wide transport of free floating DNA and the propagation of sound waves underwater. In this thesis, I explore the use of eDNA and ecoacoustics for surveying freshwater biodiversity. I report the use of eDNA-based methods to detect the invasion fronts of an advancing signal crayfish (Pacifastacus leniusculus) population in Yorkshire, UK. In addition, I use eDNA-based methods to map the distributions of invasive and endangered crayfish in Norfolk, UK, and evaluate the trade-off between reactive and proactive strategies to inform crayfish conservation. I conduct a systematic review of the freshwater bioacoustics literature and identify promising areas for future research. In addition, I co-develop the first standardised survey protocol for the collection of acoustic data from small waterbodies, and establish an open-access online repository for freshwater soundscape data. Next, I explore the diel acoustic activity cycles of temperate ponds, relate acoustic complexity to macroinvertebrate composition, and suggest guidelines for survey design. Finally, I explore the use of field recordings as a powerful tool for public engagement and science communication.
... These sounds are generated internally, frequently with specialized sonic organs or structures, and are generally used for communication (Kasumyan 2008). Passive sounds, sometimes referred to as incidental, unspecialized, or mechanical sounds, are generated by interacting with the environment and may not be associated with specialized sound-producing anatomical structures nor with specific behaviors or situations, though they may still serve some signal function (Fish 1954;Moulton 1960;Kasumyan 2008;Rountree et al. 2018). The definitions of active and passive sounds are flexiblepassive sounds can be associated with specific behaviors like moving substrate during spawning activities , and active sounds can be produced without highly specialized structures such as grunts made with pharyngeal jaws (Kasumyan 2008). ...
... In addition to active sounds, likely all fishes are capable of producing passive sounds (Kasumyan 2008), but which ones regularly do, in what contexts, and with what signaling purpose-if any-remains unresolved (Rountree et al. 2018). Our review identified 140 species that make passive feeding sounds, such as chewing, and 141 species that make other passive sounds, such as those produced from swimming, digging, or bumping into objects (Fig. 1b). ...
... The rate of studies reporting passive fish sound production has also largely remained unchanged through time (Fig. 3), in contrast to studies on active sound production and in spite of the overall trend of increasing scientific publications (Bornmann and Mutz 2015). Even with greater numbers of species examinations, difficulties may remain in determining the rate of occurrence of passive sounds because of their situational dependence and challenges faced when distinguishing between passive and active sounds, especially as both may occur in quick succession within behavioral sequences (Rountree et al. 2018). ...
Full-text available
Sound production in fishes is vital to an array of behaviors including territorial defense, reproduction, and competitive feeding. Unfortunately, recent passive acoustic monitoring efforts are revealing the extent to which anthropogenic forces are altering aquatic soundscapes. Despite the importance of fish sounds, extensive endeavors to document them, and the anthropogenic threats they face, the field of fish bioacoustics has been historically constrained by the lack of an easily accessible and comprehensive inventory of known soniferous fishes, as is available for other taxa. To create such an inventory while simultaneously assessing the geographic and taxonomic prevalence of soniferous fish diversity, we extracted information from 834 references from the years 1874–2020 to determine that 989 fish species from 133 families and 33 orders have been shown to produce active (i.e., intentional) sounds. Active fish sound production is geographically and taxonomically widespread—though not homogenous—among fishes, contributing a cacophony of biological sounds to the prevailing soundscape globally. Our inventory supports previous findings on the prevalence of actively soniferous fishes, while allowing novel species-level assessments of their distribution among regions and taxa. Furthermore, we evaluate commercial and management applications with passive acoustic monitoring, highlight the underrepresentation of research on passive (i.e., incidental) fish sounds in the literature, and quantify the limitations of current methodologies employed to examine fishes for sound production. Collectively, our review expands on previous studies while providing the foundation needed to examine the 96% of fish species that still lack published examinations of sound production.
... In addition to active sounds, all fish can produce passive sounds incidentally, which may also carry information (Kasumyan 2008;Rountree et al. 2018). For example, feeding sounds and hydrodynamic sounds can be acoustically distinguished, and they may both have varying characteristics with differing fish sizes and species (Kaparang et al. 1999). ...
... Identification of unobserved sounds based on comparison with known sounds relies on the existence of comprehensive libraries of sounds, which are rarely complete for a given geographic area, giving rise to frequent calls for developing sound libraries (e.g., Lindseth and Lobel 2018;Luczkovich et al. 2008;Rountree et al. 2006). To identify sound sources more directly, wild studies may incorporate unaided visual observation (e.g., Abbott 1877) or underwater video (e.g., Rountree et al. 2018), among other methods. Even with direct observation, however, it may not be possible to provide conclusive proof of sound source identity if obvious sound production behavior is not evident (e.g., Rountree and Juanes 2010). ...
Conference Paper
Even with widespread evidence of the ecological importance of fish sounds for signaling and communication, and an increasing awareness of the negative impacts of anthropogenic sound, research into fish sound production still faces ongoing challenges limiting growth. To acquire quantitative data on research into fish sounds and assess topics related to their study, a systematized review was conducted to survey over 3000 references for published examinations of sound production by fish species. Additional information was further extracted on the study of marine, subtropical species from a subset of the assembled references. The review findings showed that the rate of soniferous fish species discovery remains inefficient for understanding even a sizeable portion of the behaviors of the 34900 extant fish species known to exist. In the literature surveyed, there was also no standardized sound nomenclature, fish species were not frequently tested in multiple environments, and sound visualizations were inconsistently provided. Advancements in these areas would improve the ability to document, describe, and synthesize fish sounds, which would in turn increase understanding of their global scope and the anthropogenic threats that they face.
... Several fish species have been observed to produce audible splashes at the surface of the water when they spawn, including Carp (Cyprinus carpio) and Northern Pike (Esox Lucius, Clark 1950), Bream (Abramis brama, Poncin et al. 1996), Sturgeon (Acipenser oxyrinchus, Sulak et al. 2002, Catostomids and Salmonids (Rountree et al. 2018). In some European fisheries, anadromous species such as Allis Shad (A. alosa) and Twaite Shad (A. fallax) are monitored by counting the number of these spawning splashes produced by male fish (e.g., Lebel et al. 2001, Baglinière et al. 2003, Chanseau et al. 2004). ...
... Other fish species that migrate in the Choptank River during the spring, such as White Perch (Morone americana) and Yellow Perch (Perca flavescens), may also confound our results (Ogburn et al. 2017). Thus, to assess speciesspecific spawning activity, terrestrial acoustic monitoring of spawning splashes must be paired with aquatic (hydrophone) acoustic monitoring to collect species-specific acoustic signals (Rountree et al. 2018). Biological samples (e.g. ...
Objective We used passive acoustic monitoring (PAM) and automatic detection of spawning splashes to examine the timing and environmental drivers of spawning in river herring (Alewife Alosa pseudoharengus and Blueback Herring A. aestivalis ). Methods Acoustic recordings of spawning splashes were collected from March to May 2021 in the Choptank River, Maryland, using an AudioMoth recorder. Recordings were analyzed using a random forest model on the Rainforest Connection ARBIMON platform to determine hourly presence–absence of splashes. Result At a seasonal scale, our results suggested two peaks in spawning activity: early March and mid‐April, corresponding with the known phenologies of Alewife and Blueback Herring. Hourly patterns in spawning activity suggested distinct diel cycles, with spawning most concentrated at dawn. In contrast, sonar fish counts collected for 1 week during the season indicated that migration occurred throughout daylight hours. We also found a potential relationship between spawning activity and the presence of great blue herons Ardea herodias . Conclusion Overall, our results demonstrate that PAM can be an efficient and affordable method for studying the spawning ecology of anadromous fish.
... Underwater soundscapes can influence the composition of a diverse array of aquatic communities and are important for aquatic organisms that rely on hearing for orientation, prey detection, predator avoidance, social interactions, and other behavioral responses (Cotter 2008;Fay 2009;Bruintjes and Radford 2013;Mensinger et al. 2016;Rountree et al. 2018). However, underwater soundscapes are increasingly subjected to anthropogenic sounds Pratchett et al. 2011;Radford et al. 2014;Arthington et al. 2016;Poikane et al. 2017;Popper and Hawkins 2019;Rountree et al. 2020). ...
... Motorized watercraft, nearshore construction, seismic testing, and urbanization have led to an increase in background sound and can have detrimental effects on aquatic organisms (Popper et al. 2005;Kuehne et al. 2013;Putland and Mensinger 2020). Despite previous studies investigating the effects of anthropogenic sound on marine organisms, fewer in situ studies have observed the effects of sound on fish behavior in freshwater lakes exposed to anthropogenic sound or wilderness environments that prohibit the use of motorized watercraft and vehicles (Mickle and Higgs 2017;Rountree et al. 2018;Putland and Mensinger 2020). ...
Full-text available
Freshwater lake soundscapes yield crucial information regarding biological, geological, and anthropogenic activity, yet is a relatively unexplored area of study. These soundscapes are particularly important to aquatic life that may use sound to navigate, find food, avoid predators, and communicate. Further research is required to understand how aquatic species, such as native fishes, are impacted by increased anthropogenic interference. Many wilderness lakes restrict the use of motorized boats and equipment, providing an opportunity to compare fish behavior in the presence and absence of anthropogenic sound. Underwater videos and passive acoustic monitoring were used to evaluate fish behavior under different soundscapes in the upper Midwest United States: John Lake (nonmotorized, Boundary Waters Canoe Area Wilderness, MN), Rush Lake (nonmotorized, Huron Mountain Club, MI), and Caribou Lake (motorized, Duluth, MN). Intermittent short and long anthropogenic sound playback experiments showed behavioral changes in bluegill sunfish (Lepomis macrochirus, centrarchids), bluntnose minnows (Pimephales notatus, cyprinids), mimic shiners (Notropis volucellus, cyprinids), and yellow perch (Perca flavescens, percids). Overall, cyprinids in wilderness lakes were the most responsive to boat sound 36–52 dB above ambient sound levels, with bluegills in the public lake more likely to remain in the area during longer duration sound stimuli. Taken together, these results indicate that behavioral responses are species specific and depend on environmental variables such as anthropogenic exposure and fishing pressure.
... Alongside active sound production for the purported purpose of communication, many aquatic species produce "passive sounds" as a by-product of other life-functions, such as eating, swimming, and crawling (e.g., Fish, 1948;Moulton, 1958Moulton, , 1960Moulton, , 1963Moulton, , 1964Uno and Konagaya, 1960;Mallekh et al., 2003;Radford et al., 2008;Rountree et al., 2018;Ajemian et al., 2021;Tricas and Boyle, 2021; Figure 1). These passive sounds may be less acoustically complex or distinct than active sounds; however, they still provide important contributions to the soundscape and have demonstrated ecological signal potential in select circumstances (Banner, 1972;Connor et al., 2000;Tricas and Boyle, 2014;Rountree et al., 2018). ...
... Alongside active sound production for the purported purpose of communication, many aquatic species produce "passive sounds" as a by-product of other life-functions, such as eating, swimming, and crawling (e.g., Fish, 1948;Moulton, 1958Moulton, , 1960Moulton, , 1963Moulton, , 1964Uno and Konagaya, 1960;Mallekh et al., 2003;Radford et al., 2008;Rountree et al., 2018;Ajemian et al., 2021;Tricas and Boyle, 2021; Figure 1). These passive sounds may be less acoustically complex or distinct than active sounds; however, they still provide important contributions to the soundscape and have demonstrated ecological signal potential in select circumstances (Banner, 1972;Connor et al., 2000;Tricas and Boyle, 2014;Rountree et al., 2018). Thus, while collating global records of known sound production may be feasible to accomplish (e.g., for fishes; Looby et al., 2021), because of the variation in sound within and among species and individuals, the effort required to collect and maintain representative sounds for every species is a continuous and laborious process. ...
Full-text available
Aquatic environments encompass the world’s most extensive habitats, rich with sounds produced by a diversity of animals. Passive acoustic monitoring (PAM) is an increasingly accessible remote sensing technology that uses hydrophones to listen to the underwater world and represents an unprecedented, non-invasive method to monitor underwater environments. This information can assist in the delineation of biologically important areas via detection of sound-producing species or characterization of ecosystem type and condition, inferred from the acoustic properties of the local soundscape. At a time when worldwide biodiversity is in significant decline and underwater soundscapes are being altered as a result of anthropogenic impacts, there is a need to document, quantify, and understand biotic sound sources–potentially before they disappear. A significant step toward these goals is the development of a web-based, open-access platform that provides: (1) a reference library of known and unknown biological sound sources (by integrating and expanding existing libraries around the world); (2) a data repository portal for annotated and unannotated audio recordings of single sources and of soundscapes; (3) a training platform for artificial intelligence algorithms for signal detection and classification; and (4) a citizen science-based application for public users. Although individually, these resources are often met on regional and taxa-specific scales, many are not sustained and, collectively, an enduring global database with an integrated platform has not been realized. We discuss the benefits such a program can provide, previous calls for global data-sharing and reference libraries, and the challenges that need to be overcome to bring together bio- and ecoacousticians, bioinformaticians, propagation experts, web engineers, and signal processing specialists (e.g., artificial intelligence) with the necessary support and funding to build a sustainable and scalable platform that could address the needs of all contributors and stakeholders into the future.
... This study found that the variance in the sound signal within a frequency range previously established as relevant for turbot behavior (6-8 kHz, Lagardere & Mallekh, 2000) was significantly linked with feeding activity and even ration size. Although there have not been similar studies aimed at feeding in salmonids, the sound production capabilities of six freshwater species, including four salmonids, were recently investigated by Rountree et al. (2018). This study concluded that most of the sounds generated by these species was due to air movement, and that it is possible to distinguish the species based on PAM analyses. ...
Pig farming systems face an increasingly diversified challenge to consider simultaneously the economic, environmental, and social pillars of sustain ability. For animal nutrition, this requires the development of smart feeding strategies able to integrate these different dimensions in a dynamic way and to be adapted as much as possible to each individual animal. These developments can be supported by digital technologies including data collection and processing, decision making and automation of applications. Classical traits such as feed intake and growth benefit from new technologies that can be measured more frequently. New sensors can be indicative for other traits related to body composition, physiological status, activity, feed efficiency, or rearing environment. A challenge for data collection is to obtain information on a large number of animals and with sufficient frequency, quality, and precision and use it cost-effectively. Another challenge is to analyse the ever-increasing volume of data and use it in decision-making. Nutritional models for pigs and sows, classically mechanistic, have to evolve to integrate real-time data. With the development of data-driven modelling methods (e.g., machine-learning or deep-learning), a synergy between mechanistic models and data-driven approaches is required in smart pig nutrition. Moreover, the practical application of smart pig nutrition must consider the evolution in pig farming systems towards increased diversity in terms of size, space allowance, and outdoor access, and return on investment. Finally, the transition of pig nutrition in the digital era must consider the social acceptance of an increasing role of digital technologies in animal production systems.KeywordsActivityArtificial intelligenceAutomatonConcept-driven modellingData collectionData-driven modellingData processingDecision support systemFattening pigsFeed efficiencyFeed intakeGestating sowHealth statusLactating sowMineralNutritionNutritional requirementsPerformancePhysiological statusPig farming systemPrecision feedingRearing environmentSensors
... Fish also produce sound from rapid changes in swimming speed in large shoals and during feeding (Amorim 2006;Aalbers and Drawbridge 2008). Some species produce sounds from air movement (air passage or pneumatic sounds) (Rountree et al. 2018). ...
Full-text available
This study implemented Passive Acoustic Monitoring (PAM) to evaluate temporal acoustic patterns in a protected coastal reef area in Tamandaré beach (100 km south of Recife city, Pernambuco State, Brazil). We used an autonomous underwater recorder that allowed continuous recordings from December to March. The sounds detected in the marine protected area indicated the presence of six chorus types, two of them occurring only in winter. We detected choruses occurring in different times of the day and presenting a daily pattern, with differences in the initial time of detection (p < 0.05). Overlapping signals from four choruses occurred mainly after sunset (17:30 p.m.), while two choruses occurred after midnight. Choruses usually lasted for 1 h 30 m. One of the choruses produced a wide frequency band (300 to 4000 Hz) that masked the frequency of the other choruses. Lunar phase changes influenced all choruses, with major differences during the first-quarter moon. Vessel noise occurred primarily in the early morning and at night. Vessels had low dominant frequencies, with higher peaks below 155 Hz and different peaks that can reach 7000 Hz. The vessels produced noises with energy of 90 dB re 1 μPa2 Hz−1, distributed in a wide frequency band. These noises were enough to mask all the choruses, although characterized by short peaks (< 10 min of detection). Fish chorus and vessel noise were detected using passive acoustic monitoring that indicated the need to implement short- and long-term monitoring and management plans.
In this chapter, we discuss how digital tools can be used to achieve more intelligent feeding and nutrition in commercial cage-based farming. Using farmed salmon as a model species, we first outline industrial practices in cage-based farming, and then present the state-of-the-art in how digital technologies are being utilized in aquaculture research and industry. We then discuss how the intelligent feeding methods of the future could be devised based on the current state-of-the-art, and further how these could potentially be important for ongoing industrial developments toward new production concepts for fish farming.Our findings show that many of the digital components required to realize intelligent feeding systems in commercial fish farms are already in place, or under development. It is thus already possible to start combining existing systems into new technological solutions that improve our ability to monitor, adjust, and optimize the feeding process in aquaculture fish production. This is the focus of several ongoing research efforts that aspire to apply the principles of precision fish farming. A similar trend is also present in the industrial sector, manifested through the rapid growth in the portfolio of commercially available products for feeding optimization in aquaculture.KeywordsFish farmingSmart fish nutritionFeeding technologyPrecision fish farmingAtlantic salmonAquacultureIntelligent fish feedingSustainable fish farmingBiosensorsTelemetryOptical sensorsHydroacoustic sensorsMathematical fish modellingSensor fusionUnderwater roboticsIntensive fish farmingMarine fish farmsFeedback controlState estimationFuture methods for fish farming
Continuous data on the condition of fish is necessary to monitor, control and document biological processes in fish farms in real-time, yet acquiring it from a large net-pen environment is challenging. Tools to rapidly detect change in the entire net-pen population are lacking. Automated passive acoustic monitoring is emerging as an effective monitoring tool in wildlife monitoring but has not before been tested in an aquaculture setting. Here, we explore the possibilities for passive acoustic monitoring in an aquaculture perspective. We investigated whether the soundscape of a net-pen could infer information on the condition of the whole net-pen population. In three cases, conducted at two different fish farms, we tested whether Atlantic salmon (Salmo salar) influence the soundscape of the net-pen. We provide evidence that Atlantic salmon alter the acoustic environment when compared to an empty net-pen. We observe from a 24-h recording that the acoustic fingerprint of the net-pen varies over time and mirrors the feeding status of the fish. Our results demonstrate the potential for passive acoustic monitoring in fish farms and provide a new direction for data-driven management in aquaculture to improve fish welfare and operational feeding routines.
Full-text available
full version: The aims of this study were to (i) assess the efficacy of passive acoustic monitoring (PAM) for detecting Arctic Charr at their spawning grounds and (ii) characterize the overall acoustic soundscape of these sites. PAM was carried out over three Arctic Charr spawning grounds in the UK, one lotic and two lentic. 24-h cycles of recordings were collected prior to and during the Arctic Charr spawning season, which was determined from data returns by simultaneous net monitoring. Acoustic analysis consisted of manual quantification of sound sources, Acoustic Complexity Index (ACI) calculation and spectral analysis in 1/3 octave band (SPL; dB re 1 μPa). In the lotic spawning ground, prior to the beginning of Arctic Charr spawning, SPL and ACI showed a restricted range of variation throughout the 24-h, while during spawning the night values of SPL and ACI were found to significantly increase, concurrently with the rate of gravel noise induced by fish spawning activities and fish air passage sounds. Both prior to and during the Arctic Charr run, the lentic soundscape was characterized by diel variation due to the daytime presence of anthropogenic noise and the night-time presence of insect calls, while only a few occurrences of fish air passage sounds and gravel noise were recorded. These findings suggest that PAM over Arctic Charr spawning grounds could provide meaningful information to be used in developing management plans for this threatened species, such as determining the location and time of arrival, diel pattern and length of spawning activities
Full-text available
Arctic charr Salvelinus alpinus did not appear to invest in acoustic communication during courtship and agonistic interactions in captivity. Salvelinus alpinus did, however, produce four different types of sounds which were found to be associated with three different types of air exchange behaviours which probably have a swimbladder regulation function. Since air passage sounds appear to be common among Salmonidae, it is suggested that the potential of passive acoustics techniques for behavioural and ecological monitoring should be further investigated in future field and laboratory investigations.
Full-text available
Passive acoustic technologies are those technologies that enable us to listen to and record ambient underwater sounds. Such technologies have existed for decades; however, a major initiative to develop and promote their use in fisheries applications and as an important new tool for the census and exploration of marine life is now needed. Given the significant advancement in underwater tech- nologies, passive acoustic research promises to be an important new field in fisheries and related areas/disciplines. The ability to listen to fish and other marine life allows scientists to identify, record and study underwater animals, even in the absence of visual information. Coupling passive acoustics with conventional visual monitoring and sampling techniques provides a powerful new approach to undersea research. The Sea Grant College Program has recognized the great potential of passive acoustics for fisheries and related fields, and has taken a leadership role in supporting the development of innovative new research programs using this approach. (note, glossy print copies available from the authors or MIT SeaGrant)
Full-text available
A number of species of clupeid fish, including blueback herring, American shad, and gulf menhaden, can detect and respond to ultrasonic sounds up to at least 180 kHz, whereas other clupeids, including bay anchovies and Spanish sardines, do not appear to detect sounds above about 4 kHz. Although the location for ultrasound detection has not been proven conclusively, there is a growing body of physiological, developmental, and anatomical evidence suggesting that one end organ of the inner ear, the utricle, is likely to be the detector. The utricle is a region of the inner ear that is very similar in all vertebrates studied to date, except for clupeid fish, where it is highly specialized. Behavioural studies of the responses of American shad to ultrasound demonstrate that they show a graded series of responses depending on the sound level and, to a lesser degree, on the frequency of the stimulus. Low-intensity stimuli elicit a non-directional movement of the fish, whereas somewhat higher sound levels elicit a directional movement away from the sound source. Still higher level sounds produce a "wild" chaotic movement of the fish. These responses do not occur until shad have developed the adult utricle that has a three-part sensory epithelium. We speculate that the response of the American shad (and, presumably, other clupeids that can detect ultrasound) to ultrasound evolved to help these species detect and avoid a major predator - echolocating cetaceans. As dolphins echolocate, the fish are able to hear the sound at over 100 m. If the dolphins detect the fish and come closer, the nature of the behavioural response of the fish changes in order to exploit different avoidance strategies and lower the chance of being eaten by the predators.
Documenting timing, locations, and intensity of spawning can provide valuable information for conservation and management of imperiled fishes. However, deep, turbid or turbulent water, or occurrence of spawning at night, can severely limit direct observations. We have developed and tested the use of passive acoustics to detect distinctive acoustic signatures associated with spawning events of two large-bodied catostomid species (River Redhorse Moxostoma carinatum and Robust Redhorse Moxostoma robustum) in river systems in north Georgia. We deployed a hydrophone with a recording unit at four different locations on four different dates when we could both record and observe spawning activity. Recordings captured 494 spawning events that we acoustically characterized using dominant frequency, 95% frequency, relative power, and duration. We similarly characterized 46 randomly selected ambient river noises. Dominant frequency did not differ between redhorse species and ranged from 172.3 to 14,987.1 Hz. Duration of spawning events ranged from 0.65 to 11.07 s, River Redhorse having longer durations than Robust Redhorse. Observed spawning events had significantly higher dominant and 95% frequencies than ambient river noises. We additionally tested software designed to automate acoustic detection. The automated detection configurations correctly identified 80-82% of known spawning events, and falsely indentified spawns 6-7% of the time when none occurred. These rates were combined over all recordings; rates were more variable among individual recordings. Longer spawning events were more likely to be detected. Combined with sufficient visual observations to ascertain species identities and to estimate detection error rates, passive acoustic recording provides a useful tool to study spawning frequency of large-bodied fishes that displace gravel during egg deposition, including several species of imperiled catostomids.
Underwater ambient noise was investigated in the stream-mouths of Clear, Cub and Pelican Creeks. Noise levels were determined from 0.1 to 10 KHz during periods of high stream discharge and wave action. Minimum noise levels could not be determined due to instrument noise interference. Two noise sources contributed to ambient pressure spectrum levels in the stream-mouths: (1) flow noise and/or bubbles, and (2) surf-beats. The former is mainly composed of frequencies below 4 KHz while the latter is above 5 KHz. Four cutthroat trout (Salmo clarki) sounds were recorded and analyzed. The “thump” sound occurred when fish were alarmed and gave a sudden tail-flip. The principal frequency was 150 Hz in the band from 100 to 200 Hz. The “squawk” sound had principal frequencies in the band from 600 to 850 Hz. The “squeak” sound was infrequent and usually of low intensity. A sound with maximum energy above 2 KHz was created when a trout shifted bottom materials while preparing a redd. The response to low frequency sound was tested with 29 cutthroat trout. The conditioned response technique was applied using shock or light as the unconditioned stimulus. A positive response was not readily obtained. A natural response to sound stimuli was found in six fish with an average upper frequency limit of 443 Hz. Underwater sounds are within the range of fish perception but it was not determined if homing fish recognized or utilized stream noises.