Sound Production And The Effect Of Noise On Communication In The Blacktail Shiner 
(Cyprinella venusta) 
 
by 
 
Daniel Edward Holt 
 
 
 
 
A dissertation submitted to the Graduate Faculty of 
Auburn University 
in partial fulfillment of the 
requirements for the Degree of 
Doctor of Philosophy 
 
Auburn, Alabama 
May 7, 2012 
 
 
 
 
Keywords: acoustic communication, Cyprinella, environmental noise, 
fish, hearing, propagation  
 
 
Copyright 2012 by Daniel Edward Holt 
 
 
Approved by 
 
Carol E. Johnston, Chair, Professor of Ichthyology 
Dennis M. Higgs, Professor of Sensory Biology 
Geoffrey E. Hill, Professor of Ornithology 
Mary T. Mendon?a, Professor of Endocrinology 
David A. Mann, Professor of Biological Oceanography 
 
ii 
 
Abstract 
  Animal communication is important in the success and survival of species.  In 
order to find a potential mate, individuals must be able to communicate effectively with 
each other.  This communication may involve identification of the proper species, 
individuals which are in the proper physiological and motivational states for 
reproduction, and possibly the coordination of gamete release.  Communication is also 
necessary for many aspects of an animal?s life that relate indirectly to reproduction.  The 
interaction and competition between males for resources and mates, for example, is 
often dependent upon communication.  Although most people are not aware of it, the 
underwater world of freshwater streams is alive with sound, and many fishes in these 
environments communicate acoustically.  The ability to communicate effectively using 
acoustics can be highly affected by levels of ambient noise.  In a natural setting, this 
noise may be a selective pressure in the evolution of acoustic signals.  However, in 
recent years, the breadth and level of anthropogenic noise has increased drastically, and 
little attention has been given to the effect it may have on freshwater fishes, especially 
those inhabiting small to moderately sized streams.  Freshwater fishes in the genus 
Cyprinella have been shown to communicate acoustically during the reproductive 
season.  Several species have been described in detail, but Cyprinella venusta, an 
extremely common species in the southeast United States, has not.  This dissertation 
provides the first detailed description of acoustic signals and associated behaviors in C. 
ii 
 
venusta and makes the first comparison of acoustic signals and behaviors recorded in 
the field and laboratory in the genus Cyprinella.  A description is also made of the 
natural soundscape, anthropogenic noise sources, hearing abilities, and propagation of 
signals and noise sources in the natural environment in order to determine how far 
acoustic signals are utilized, and how anthropogenic noises may impact the ability of C. 
venusta to communicate.  Finally, I investigate the ability of C. venusta to adapt to 
elevated noise levels in the laboratory.  The results of the current study advance our 
understanding of sound production in the genus Cyprinella, and also our knowledge of 
how noise may impact vocal fishes in freshwater systems. 
     
   
 
ii 
 
Acknowledgments 
 Partial funding was provided by: the Auburn University Graduate Research 
Award, the Department of Fisheries and Allied Aquaculture, and the Sigma Xi Grants-in-
Aid of Research award.  
I would first like to give my most sincere thanks to my advisor, Dr. Carol 
Johnston, for all the support, encouragement, and insightful advice she has provided me 
during the portion of my life at Auburn.  I would also like to thank the many colleagues 
who provided valuable input and instruction, especially Drs. David Mann, Dennis Higgs, 
and Frank Jordan. 
 I would also like to thank my parents, brother and sister for their unwavering 
support and love.  They have been a great source of strength in my life, and have always 
been there when the going gets tough, a common occurrence during the course of a 
dissertation.  
To Ivey.  From late nights at buffalos, making check-lists on how to graduate on 
time, and spending an hour under a little covered bridge, to the moment we said ?I do? 
under our Arbor, the time I?ve been blessed with you has been the greatest of my life.  I 
love you! 
  
iii 
 
Table of Contents 
 
 
Abstract ................................................................................................................................ ii 
Acknowledgments..............................................................................................................  iv 
List of Tables .....................................................................................................................  vii  
List of Figures .....................................................................................................................  ix  
Chapter One: Sound Production And Associated Behaviors In Cyprinella venusta;  
             A Comparison Between The Laboratory And The Field.   ....................................... 1 
 Introduction   .......................................................................................................... 3 
 Methods   ................................................................................................................ 6 
 Results   ................................................................................................................. 17 
 Discussion   ............................................................................................................ 26 
 Literature Cited   ................................................................................................... 37 
Chapter Two: The Effect Of Noise On Acoustic Communication  
             In Cyprinella venusta.   .......................................................................................... 64 
 Introduction   ........................................................................................................ 65 
 Methods   .............................................................................................................. 70 
 Results   ................................................................................................................. 78 
 Discussion   ............................................................................................................ 82 
 Literature Cited   ................................................................................................... 92 
 
 
iii 
 
Chapter Three: Hearing Sensitivity, Environmental Soundscape, 
             And Propagation Of Acoustic Signals In Cyprinella venusta.  ............................. 122 
 Introduction   ...................................................................................................... 123 
 Methods   ............................................................................................................ 127 
 Results   ............................................................................................................... 133 
 Discussion   .......................................................................................................... 138 
 Literature Cited   ................................................................................................. 147 
 
ii 
 
List of Tables 
 
 
Table 1-1.  Definitions and terminology of signal parameter.............................................. 48 
Table 1-2.  Percentage of behaviors associated with sound in lab?????????????? 49 
Table 1-3.  Percentage of behaviors associated with sound in field..????????????. 50 
Table 1-4.  Comparison of signal types in lab and field?.??????????????????? 51 
Table 1-5.  Comparison of agonistic and courtship contexts for growl 
 and knock signal parameters during approach behaviors in the lab??????. 52 
Table 1-6.  Comparison of agonistic and courtship contexts for growl 
 and knock signal parameters during lateral display behaviors 
 in the lab???????????????????????????????????????.. 53 
Table 1-7.  Comparison of agonistic and courtship contexts for growl 
 signal parameters during approach behaviors in the field???????????... 54 
Table 1-8.  Comparison of agonistic and courtship contexts for growl 
 signal parameters during chase behaviors in the field??????????????. 55 
Table 1-9.  Comparison of approach, lateral display, and crevice inspection 
 growls and knocks signal parameters in the lab?????????????????.. 56 
Table 1-10.  Comparison of multiple behaviors for growl signal parameters 
 in the field??????????????????????????????????????? 58 
Table 1-11.  Comparison of approach, mm lateral display, and 
 mm circle swim behaviors for knock signal parameters 
 in the field??????????????????????????????????????? 60 
Table 1-12.  Comparison of laboratory and field localities for growl 
 and knock signal parameters during approach behaviors???????????? 61 
Table 1-13.  Comparison of laboratory and field localities for knock 
iii 
 
 signal parameters during lateral display behaviors???????????????? 62 
Table 1-14.  Comparison of laboratory and field localities for growl 
 signal parameters during crevice inspection behaviors?????????????. 63 
Table 2-1.  Comparison of agonistic and courtship contexts for growl 
 signal parameters during approach behaviors under  
 noisy conditions???????????????????????????????????? 116 
Table 2-2.  Comparison of agonistic and courtship contexts for growl 
 and knock signal parameters during lateral display behaviors 
 under noisy conditions???????????????????????????????? 117 
Table 2-3.  Comparison of signal types in lab and field under noisy conditions??????? 118 
Table 2-4.  Comparison of noisy and quiet conditions on growl 
 signal parameters for approach behaviors????????????????????? 119 
Table 2-5.  Comparison of noisy and quiet conditions on knock 
 signal parameters for approach and lateral display behaviors?????????.. 120 
Table 2-6.  Comparison of noisy and quiet conditions on growl 
 signal parameters for lateral display behaviors??????????????????. 121 
Table 3-1.  Estimated active areas of natural and anthropogenic 
 noise sources?????????????????????????????????????. 162 
 
ii 
 
List of Figures 
 
 
Figure 1-1.  Experimental tank setup ................................................................................ 43 
Figure 1-2.  Visual representations of C. venusta acoustic signals ................................... 44 
Figure 1-3.  Scatterplot showing relationship between distance from 
 the hydrophone and burst harmonic frequency level .................................... 45 
Figure 1-4.  Scatterplot showing relationship between distance from the  
 hydrophone and knock fundamental and harmonic frequencies .................. 46 
Figure 1-5.  Correlations between SL and burst harmonic frequency and 
 pulse interval ................................................................................................... 47 
Figure 2-1.  Experimental tank setup for noise trials. ....................................................... 99 
Figure 2-2.  Power spectra of background noise during quiet and noisy trials .............. 100 
Figure 2-3.  Correlations between standard length and pulse interval .......................... 101 
Figure 2-4.  Correlations between standard length and burst harmonic frequency ...... 102 
Figure 2-5.  Correlations between standard length and pulse duration ........................ 103 
Figure 2-6.  Correlations between standard length and burst duration ........................ 104 
Figure 2-7.  Correlations between standard length and burst rate ................................ 105 
Figure 2-8.  Correlations between standard length and knock duration ....................... 106 
Figure 2-9.  Correlations between standard length and 
 burst fundamental frequency ....................................................................... 107
Figure 2-10.  Correlations between standard length and 
 burst fundamental frequency level. ............................................................. 108 
Figure 2-11.  Correlations between standard length and 
iii 
 
 knock fundamental frequency level ............................................................. 109 
Figure 2-12.  Correlations between standard length and  
 knock harmonic frequency ........................................................................... 110 
Figure 2-13.  Scatterplot showing relationship between distance from 
 the hydrophone and burst harmonic frequency level 
 under noisy conditions .................................................................................. 111 
Figure 2-14.  Scatterplot showing relationship between distance from the  
 hydrophone and knock fundamental and harmonic frequencies 
 under noisy conditions .................................................................................. 112 
Figure 2-15.  Error bar plots showing interfish distance during acoustic 
 communication under noisy and quiet conditions. ...................................... 113 
Figure 2-16.  Error bar plots showing % behaviors associated with sound  
 production, and behavior rate during noisy and quiet trials.  ...................... 114 
Figure 2-17.  Error bar plot showing levels of sounds produced during 
 noisy and quiet trials with reference to background noise. ......................... 115 
Figure 3-1.  Audiograms of C. auratus and C. venusta .................................................... 154 
Figure 3-2.  Examples of evoked potentials from C. venusta  
 obtained through the Auditory Brainstem Response ................................... 155 
Figure 3-3.  Attenuation of semi-trailer truck noise ....................................................... 156 
Figure 3-4.  Attenuation of train noise ........................................................................... 157 
Figure 3-5.  Attenuation of C. venusta acoustic signals .................................................. 158 
Figure 3-6.  Attenuation of C. venusta acoustic signals 
 in terms of one-third octave bands .............................................................. 159 
Figure 3-7.  Averaged power spectra of the natural soundscape,  
 anthropogenic noise sources, and growl signals from  
 C. venusta at a bin resolution of 21.5 Hz. ..................................................... 160 
Figure 3-8.  Averaged power spectra of the natural soundscape, 
 Anthropogenic noise sources, and knock signals from 
 C. venusta at a bin resolution of 86.1 Hz. ..................................................... 161 
1 
 
 
 
 
 
Chapter 1 
  
Sound Production And Associated Behaviors In Cyprinella venusta; A Comparison 
Between The Laboratory And Field. 
Although it has been known that C. venusta produced acoustic signals during 
reproductive interactions, the only previous description was basic and provided very 
limited information describing both behaviors and physical characteristics of 
vocalizations.  Since the initial description of sounds in C. venusta, the acoustic signals 
and associated behaviors of a number of other fishes in the genus Cyprinella have been 
investigated and described in detail.  A detailed description of sounds produced by C. 
venusta is a necessary first step in the quest to further understand how this species lives 
in close association with other closely related, acoustically active species and its 
environment.  In the current study, I described the sounds produced by C. venusta, 
along with associated behaviors in both a laboratory, and a field setting.  I found that 
males were the only sex to vocalize and did so during reproductively associated 
behaviors such as courtship, aggression, and spawning.  Sounds were similar in gross 
structure to a few other species of Cyprinella in that they were composed of bursts and 
knocks, but seemed to differ in a number of signal parameters including frequency, 
pulse duration, pulse interval, and pulse rate.  One of the more striking findings was that 
the frequency distributions of both growls and knocks in C. venusta were bi-modal, a 
characteristic not mentioned for any other species in the genus Cyprinella.  I also 
2 
 
observed several differences between field and laboratory recordings with the 
laboratory generally showing a smaller subset of behaviors that were observed in the 
field.  Also, sounds recorded in the laboratory were found to be significantly different 
from those recorded in the field. 
  
3 
 
Introduction 
In all sexual animals, successful reproduction is not possible without some form 
of communication (Bradbury and Vehrencamp 1998).  The use of acoustics as a mode of 
communication has been documented in a wide variety of terrestrial taxa including birds 
(Slater 2003), reptiles (Dooling et al. 2000), frogs (Narins et al. 2007), mammals 
(McComb and Reby 2005), and insects (R?mer 1998).  Acoustic communication has also 
been documented in aquatic organisms including marine mammals (Janik 2005), aquatic 
insects (Aiken 1985), and fishes (Ladich 2004).  Despite the fact that fishes account for 
approximately 51% of all recognized vertebrate biodiversity (Nelson 2006), and inhabit a 
wide variety of habitats ranging from shallow, freshwater streams, to the deepest 
reaches of the oceans, only a small percentage of fish species have had their acoustic 
repertoires thoroughly described.  A more complete understanding of the 
commonalities between the acoustic signals of different fish species may provide 
additional insight into the evolutionary history of acoustic signaling in fishes, as well as 
provide a better framework on which further bioacoustic studies may be based. 
 Acoustic communication is likely important for many fishes due to its association 
with important tasks that directly influence fitness such as territory defense (Lugli 1997; 
Myrberg 1997), aggressive interactions (Ladich and Myrberg 2006), and courtship and 
reproduction (Winn and Stout 1960; Stout 1975; Lugli et al. 1995; Phillips and Johnston 
2008).  Acoustic signals likely act as indicators of male quality, which can be 
advantageous to both males during agonistic interactions, and females during courtship 
interactions.  When competing for a resource, a male will fare better if it is able to 
4 
 
assess the fighting ability of a competitor without escalating directly to a physical 
contest, which typically risks injury or death.  Ladich (1998) found that body size was 
negatively correlated with dominant frequency, and positively correlated with the level 
of acoustic signals produced by the male croaking gourami (Trichopsis vittata), and that 
larger males typically won agonistic contests.  Ladich also showed that traits not 
correlated with male body size (such as number of sounds produced) did not predict 
winning.  A thorough description of acoustic signals and the relationships between 
signal parameters and potential indicators of fitness such as body size are necessary 
before further studies investigating more complex issues such as functional significance 
can be conducted. 
Descriptions of sound production for many fishes have often been conducted in 
small aquaria under laboratory conditions (Drewry 1962; Stout 1963; Stout 1975; Vilinski 
and Rigley 1981; Colson et al. 1998; Ladich and Yan 1998; Pruzsinszky and Ladich 1998; 
Lugli and Torricelli 1999; Johnston and Johnson 2000a, b).  Numerous complications may 
arise when recording sounds in an aquarium due to the effects of reverberation and 
resonance in the tank.  Although empirical solutions have been determined to help 
minimize these effects (Akamatsu et al. 2002), direct comparisons should also be used 
to verify that signal parameters measured in the lab are consistent with sounds 
produced under natural conditions.  A few studies have provided evidence that sounds 
and behaviors observed in the laboratory are representative of those under natural 
conditions (Lugli et al. 1995; Crawford et al. 1986; Kaatz and Lobel 2001), however, the 
majority of studies have not attempted to make this comparison. 
5 
 
In recent years, the acoustic signals of several new species in the genus 
Cyprinella have been described in great detail (C. galactura, Phillips and Johnston 2008; 
C. gibbsi, C. trichroistia, C. callisema, Phillips and Johnston 2009; C. lepida, Phillips et al. 
2010).  However, all of these descriptions, as well as earlier, general descriptions of C. 
lutrensis, C. venusta (Delco 1960), and C. analostana (Winn and Stout 1960) were based 
on signals obtained under laboratory conditions.  Despite the growing number of 
studies, no study to date has attempted to compare the acoustic signals recorded under 
natural field conditions to those recorded in laboratory conditions for fishes in the 
genus Cyprinella. 
Although sound production has been documented in C. venusta by Delco (1960), 
the detail with which the signals were described was minimal, reporting only the 
duration and dominant frequency of the acoustic signals.  Furthermore, Delco only 
reports one sound type from C. venusta, which is suspicious considering most other 
descriptions of sounds from fishes in the genus Cyprinella report two or more sound 
types.  The current study provides a detailed description of the acoustic signal structure 
of C. venusta, and compares acoustic signals recorded under laboratory conditions to 
those recorded under natural conditions in the field.  Acoustic signals produced during 
agonistic and reproductive contexts, as well as specific behaviors are compared in terms 
of signal structure.  Correlations between fish size and signal parameters are discussed, 
and agonistic and reproductive behavioral repertoires and corresponding sounds are 
briefly described. 
 
6 
 
Methods 
Laboratory Recordings 
For the laboratory experiments, fish were captured by seine from tributaries to 
the Chatahoochee River in Lee County, East Alabama, specifically Little Uchee Creek 
(32.508579? N, -85.184215? W), Halawakee Creek (32.697579? N, -85.266951? W, and 
Wacoochee Creek (32.622799? N, -85.132685? W).  Fish were transported back to the 
lab and stored in 76 l aquariums with gravel substrates.  Hanging, external filters were 
used on all tanks.  Fish were kept at approximately 26?C, given a 12L:12D light cycle, and 
fed a diet of commercial flake food. Males that had not been recorded were kept 
isolated from other males, and females were stored with other females and males that 
were not to be, or had already been used. A 76 l experimental tank was setup in an 
acoustic chamber to conduct each experiment (Figure 1-1).  A tile propped up on a brick 
forming a 4 mm crevice served as an artificial nest, and was placed on the wall in the 
center of the tank. The tank also contained an underwater speaker (University Sound 
UW-30, Oklahoma City, OK) suspended behind a thin mesh barrier located on one end 
of the tank. This speaker was used to test the potential effect of elevated noise on 
acoustic communication, the results of which will be included in a forthcoming paper. 
To determine the distance between fishes during trials, a white piece of 
plexiglass with gridlines at 5 cm intervals was placed on the bottom of the tank, and an 
identical grid was placed on the back of the tank. A 60 W blub was hung approximately 
1.5 m above the tank, and a camera was placed in front of the tank in such a way that 
both the bottom and rear grids were visible.  The light bulb above the tank cast shadows 
7 
 
of fish in the tank on the bottom grid, allowing the position of the fish along the X and Z 
axes of the tank to be determined from the video camera. The grid on the back of the 
tank allowed the position of the fish along the Y axis to be determined from the video 
camera. 
Trials were conducted between 2 June ? 6 August 2009, and 6 June ? 12 August 
2010. During each trial, behavioral data was gathered from male-male agonistic 
interactions, and male-female courtship interactions.  Typically for each trial, 1 male and 
1-4 females were placed in the experimental tank and allowed to acclimate. Acclimation 
periods varied by trial, and fish were considered to be acclimated when normal 
behaviors resumed (swimming freely), and the male appeared to be defending the 
artificial nest. After the fish had acclimated, a second male was placed into the tank in 
an attempt to stimulate aggressive behavior.  During a typical successful trial, males 
would perform agonistic behaviors toward each other to establish dominance. After 
dominance had been established, the victorious male would begin courting the females 
while the losing male tried to avoid interaction with the dominant male. If dominance 
had obviously been established, and agonistic behaviors persisted, the losing male was 
removed from the tank and recording continued on the dominant male.  Each trial was 
video recorded on a Sony handycam digital HD video recorder (HDR-SR11) and sounds 
were recorded onto Raven 1.3 (Cornell University, Ithaca, NY, U.S.A) using a Br?el and 
Kjaer 8103 hydrophone and a Br?el and Kjaer 2635 charge amplifier.  Recording began 
at the onset of agonistic or courtship behaviors and ended when the behaviors ceased, 
or after 2 hours had passed. 
8 
 
The percent occurrence of sound production was determined for behaviors from 
all the trials.  Percent occurrence of sound for each behavior was calculated as the 
number of times the behavior was observed with sound accompaniment divided by the 
total number of times the behavior was observed.  For laboratory trials, video of each 
trial was watched, starting from the beginning.  Behaviors that were clearly observed 
and well within acoustic range of the hydrophone (about 15 cm) were documented, 
along with whether or not they were accompanied by sound production, and if so, what 
type of sound was produced (knock or growl).  Because in the laboratory, many trials 
contained too many behaviors to reasonably analyze, a subset was selected by analyzing 
each trial from beginning to end, or until 200 individual behaviors had been observed, 
whichever came first.  In the field, because there were many more fishes present, it was 
often difficult to assign a sound to a particular fish performing a behavior.  For this 
reason, field trials were always analyzed from beginning to end, and only behaviors that 
could be clearly distinguished as being accompanied by sound or not were included in 
the analysis.   
  
9 
 
Field Recordings 
Field behavior recordings were conducted between 18 May ? 28 July 2010, and 
again on 7 - 8 June, 2011 at Moffits Mill (Little Uchee creek, Lee Co., AL, 32.508579? N, -
85.184215? W).  Recordings were also attempted at other sites including Beans Mill 
(Halawakee Creek, Lee Co., AL, 32.697658? N, -85.266788? W) and Meadows Mill (Little 
Uchee Creek, Lee Co., AL, 32.528591? N, -85.254299? W), but a number of factors (such 
as a lack of abundant reproductive aggregations and highly turbid water) hampered my 
efforts at these sites, and useful data was unable to be collected.  On each day of 
sampling, I would begin by walking up and down the study site, looking for spawning 
aggregations of C. venusta.  Care had to be taken not to startle the aggregations, and 
polarized sunglasses were used to help detect them at farther distances.  Reproductive 
males were readily identified by the white tips on their fins, and were typically found at 
the nesting site.  Females and smaller males could be found at the nesting site and in 
surrounding areas, but the majority of them were almost always downstream of the 
active nest.  When an aggregation was found, I carefully placed an underwater camera 
(Aqua-Vu ZT 60, Nature Vision, Inc., Brainerd, MN, U.S.A) and hydrophone (Hi-Tech HTI-
96-MIN, sensitivity -164.4 re 1V/?Pa) as close to the active nest as possible.  Video and 
audio were recorded separately onto an AquaVu Digital Video Recorder, and a Marantz 
PMD 661 digital audio recorder (sampling rate 44100, PLM).  A click was made in front 
of the camera to synchronize the audio and video for later combination. 
When the aggregation was disrupted by placement or adjustment of the 
recording equipment, the fish would swim away.  In some circumstances, the fish would 
10 
 
not return after disruption, and could not be found afterwards.  Other times, the fish 
would begin spawning on another nearby nest after disruption.  In these cases, it was 
typically impossible to film the fish because they would switch nests any time the 
recording equipment was placed near them.  In the rarely successful cases, the fish 
would return to the nesting site containing the recording equipment and resume 
reproductive activity. 
 Because of the nature of the field, I was unable to keep a count of behaviors for 
individual males.  Multiple males that were indistinguishable on video would 
continuously swim in and out of the field of view, making it impossible to keep a count 
on any one specific male.  Because of this, the data could not be classified into distinct 
samples, which made it impossible to statistically compare the occurrence of behaviors, 
and the presence of sound with behaviors between the laboratory and the field. 
 
Signal Analysis 
Cyprinella venusta acoustic signals were composed of two main components, 
knocks and growls (Figure 1-2), and after some experience looking at and listening to 
the sounds, it was fairly easy to identify an uncontaminated sound from the waveform 
and spectrogram.  Knocks were always produced as a single pulse.  Growls, on the other 
hand, were composed of two smaller subunits: bursts, and pulses.  Pulses were the most 
basic component of growl call types and occurred at relatively consistent intervals 
within bursts.  Bursts were composed of groups of pulses that occurred in succession 
without a break that deviated much from the typical interpulse interval.  Bursts were 
11 
 
variable in duration and in length, and growls typically consisted of multiple bursts made 
in succession with variable inter-bust intervals. 
Video of each laboratory trial was watched while simultaneously reviewing the 
audio in Raven.  Any time a behavior occurred in the video that was accompanied by a 
sound that was suitable for analysis (not contaminated by the fish hitting something, or 
mostly buried under the ambient noise), it was labeled in Raven.  A total of 78 males 
were included in the laboratory trials, but only 20 males produced sounds.  Of the 20 
sound producing males, a varying number produced sounds associated with each 
behavior that were of high enough quality for analysis.  A male, for example, may have 
performed a wide variety of behaviors accompanied by sound, but only a small subset of 
those behaviors may have been performed close enough to the hydrophone to obtain 
recordings of a quality that was adequate for analysis.  Because of this, the ability to 
statistically compare signals between agonistic and courtship contexts, and between 
specific behaviors in the laboratory was limited to only the most commonly performed 
behaviors. 
 Methods for describing sounds produced in the field were similar to those for 
the laboratory.  Recording was attempted at a total of 22 sites.  All of the sites that 
produced data (14 sites) were located at Moffits Mill.  Some behaviors (male-female 
chase, male-female circling, and male-female zig zag) whose signals were unable to be 
compared in the laboratory due to low sample sizes were recorded more frequently in 
the field, and were able to have their associated sounds compared statistically. 
12 
 
 Temporal parameters of acoustic signals were measured using the waveform 
view in Raven (Table 1-1).  Pulse duration was measured as the time difference between 
the first deviation from random background noise in the waveform to the point at which 
the pulse pattern in the waveform could no longer be detected.  Pulse interval was 
measured as the time difference between identical peaks in adjacent pulses.  Pulse rate 
was defined as the number of pulses within a single burst divided by the duration of that 
burst.  Burst duration was measured as the time difference between the first deviation 
of the first pulse within the burst, and the point at which equilibrium was regained after 
the last pulse of the burst (the first and last pulses of a burst were subjectively 
determined by the same researcher throughout the study).  Burst interval was 
measured as the time difference between the end of one burst and the beginning of the 
next.  Burst rate was measured as the number of bursts within a call divided by the 
duration of the call (the number of bursts within a call was subjectively determined by 
the researcher, and depended on how long the corresponding behavior lasted).  Knock 
duration was measured in a manner similar to pulse duration. 
Spectral parameters were measured from the power spectrum function of Raven 
(Table 1-1).  Both bursts and knocks showed two peaks in their spectral composition.  
For bursts, the peak at the lower frequency was referred to as the burst fundamental 
frequency, and the peak at the higher frequency was referred to as the burst harmonic 
frequency.  Knock fundamental frequency and knock harmonic frequency also refer to 
the distinct low and high spectral peaks in knocks.  The sound pressure level for each of 
the frequencies was referred to as the level for that frequency.   
13 
 
Due to the fact that in the lab, fish signals were recorded in a small glass tank, 
only signals that occurred within the calculated attenuation distance from the 
hydrophone for frequencies below 1 kHz (15 cm) were included in any type of analysis 
(Akamatsu et al. 2002).  Because I was not certain of the distances fish were from the 
hydrophone in the field, only sounds that were clearly audible and clearly visible on the 
waveform were included in analyses.  All statistical analyses were conducted in PASW 
software package (IBM SPSS Corporation, Chicago, Illinois, U.S.A.).  To obtain the most 
accurate frequency information from the power density spectrum, sample sizes of FFT?s 
were maximized for each knock or burst so that each point on the power density 
spectrum represented the smallest frequency increment possible.  Decibel levels of the 
fundamental frequency and first harmonic of knocks and bursts were also measured 
from the pressure density spectrum, and were adjusted to represent actual levels by 
taking into consideration the gain imposed by each unit in the recording chain.  Decibel 
levels were also corrected to represent spectrum level (1Hz bands; dB re 1?Pa2/Hz) by 
subtracting 10log(bin resolution) of each measured signal. 
Prior to all statistical analyses, an average value for each signal parameter (pulse 
duration, burst duration, dominant frequency, etc) was calculated for each individual 
male.  For sounds recorded in the laboratory, Pearson correlation coefficients were 
calculated using these averaged values to determine if any relationship existed between 
body size measured as standard length and signal parameters.  If a correlation did exist, 
standard length was included as a covariate for that parameter during any analysis that 
was not a repeated measures or paired design.  Correlations were also performed on 
14 
 
sounds recorded in the laboratory (prior to averaging all data for each individual) to 
determine if there was a relationship between level and distance from the hydrophone.  
If levels were correlated with distance from the hydrophone, levels were corrected by 
multiplying the distance at which each sound was produced from the hydrophone by 
the slope of the linear best fit line, and adding this value to the original level of the 
knock.  This eliminated the effect of distnace without affecting variability around the 
best fit line.  Values that had been corrected for distance from the hydrophone were 
used in further analysis.  Correlations between sound parameters and fish standard 
length and distance from the hydrophone were not performed on sounds recorded in 
the field because there was no consistently reliable reference for determining fish size 
or distance from the hydrophone. 
Repeated measures ANOVAs were used, independently for laboratory and field 
conditions, to determine whether knock, burst, and pulse call components differed 
significantly from each other in terms of duration.  Paired T tests, also independent for 
laboratory and field conditions, were used to determine whether knocks and bursts 
differed in fundamental frequency, first harmonic frequency, or level.  Pulses were not 
included in frequency or dB comparisons because those signal parameters were not 
measured for pulses.  Comparison of knock, burst, and pulse call components was 
limited to approach behaviors for both laboratory and field conditions because the 
approach behavior was the only behavior that was consistently accompanied by both 
knock and growl call types, and I believed it was more appropriate to compare knocks, 
15 
 
growls, and pulses within a single behavior than to compare them among multiple 
behaviors. 
To determine whether a difference existed between growls or knocks produced 
during agonistic and courtship contexts, behaviors that were accompanied by either 
knocks or growls under both contexts had to be identified.  A challenge to this was that 
behaviors that were definitely part of the courtship or agonistic sequences were never 
shared.  I therefore had to settle for behaviors that were somewhat ambiguous in 
context, and that were preformed towards both males and females.   In the laboratory, 
knocks and growls were produced for approach and lateral display behaviors during 
both male-male and male-female interactions.    Multivariate analysis of variance 
(MANOVA) with standard length as a covariate was used to determine whether signal 
parameters (dependent variables) differed between agonistic and courtship contexts 
(fixed factors) under laboratory conditions.  This resulted in 4 separate MANOVA?s; 
growls during approaches, knocks during approaches, growls during lateral displays, and 
knocks during lateral displays.  In the field, knocks were not associated with courtship 
contexts and so only growls associated with approaches and chases were compared 
between male-male and male-female interactions.  For sounds produced in the field, 
two separate MANOVA?s were conducted; growls during approaches, and growls during 
chases.  No covariate was used in the field because body sizes were not known.  If the 
MANOVA showed a significant main effect of context for either laboratory or field 
conditions, then univariate ANOVA?s or ANCOVA?s were run on the specific signal 
parameters with fish standard length applied as a covariate where appropriate.  
16 
 
Bonferroni corrections were applied to the alpha values by multiplying 0.05 by the 
number of ANOVA?s being run. 
Methods used to determine whether signal parameters differed between 
specific behaviors (eg. approach vs lateral display behaviors) were similar to those used 
to test for differences in context, but in this case, I used doubly multivariate analysis of 
variance (with body size as a covariate in the laboratory; MANCOVA) with signal 
parameters as dependent variables and specific behaviors as the fixed factors.  If 
behaviors were shared between courtship and agonistic contexts, and signals had been 
shown not to be different between contexts within a behavior, then the contexts were 
combined (eg. if signals produced during male-male approach and male-female 
approach behaviors were not different, they were combined into simply approach 
behaviors).  A wider variety of behaviors were able to be included in the analysis 
because I was not confined by the requirement of the behavior being present in both 
agonistic and courtship contexts.  In the laboratory, I was able to compare growls 
produced during approach, lateral display, and male (m) crevice inspection behaviors.  
In the field, I was able to compare sounds associated with m circling/guarding nest, m 
crevice inspection, approach, chase, mf circling, mf hover, mf spawning, and mf zig zag 
behaviors for growls, and approach, mm circle swim, and mm lateral display behaviors 
for knocks.  If a doubly multivariate MANOVA showed a significant main effect of 
behavior type, then univariate ANOVAs with Bonferroni corrected post hoc tests on 
each dependent variable were conducted separately to see which behaviors differed, 
and in what signal parameters they differed. 
17 
 
Comparisons were made to determine whether sounds recorded in the 
laboratory differed from sounds recorded in the field for given behaviors.  Behaviors 
that were recorded in both the lab and field with sufficient sample sizes for this 
comparison included m crevice inspection and approaches for growl call types, and 
approaches and lateral displays for knock call types.  Four MANOVA?s were performed, 
one for each behavior associated with growls, and one for each behavior associated 
with knocks.  Signal parameters were used as the dependent variables, and the location 
in which they were recorded (laboratory or field) as the fixed factors.  Pulse interval, 
burst first harmonic frequency, and levels were not included as dependent variables.  
Pulse interval and burst first harmonic frequency were not included because they were 
significantly correlated with fish standard length in the laboratory, and I could not test 
for this correlation in the field.  I did not include any levels because they could not be 
tested for correlations with distance from the hydrophone in the field.  If a MANOVA 
showed a significant main effect of laboratory or field locations, then univariate 
ANOVA?s with Bonferroni correction were run on the specific signal parameters. 
 
Results 
Behavioral description 
I observed a number of behaviors from C. venusta in the field and in the 
laboratory.  Males were always the initiators of behaviors.  Agonistic behaviors were 
typically directed at other males and courtship behaviors were typically directed at 
females.  Sounds directed towards females in the laboratory included in order of 
18 
 
increasing escalation; male-female (mf) approach, mf chase, mf lateral display, mf zig 
zag, mf circling, and mf hover.  Behaviors performed towards other males in the 
laboratory included in order of increasing escalation; male-male (mm) approach, mm 
chase, and mm lateral display.  I also observed three other behaviors that were not 
directed at another fish.  The first, I called male circling/guarding nest behavior, which 
did not fall under either agonistic or courtship contexts due to the fact that I did not 
know if the male was guarding the nest, or advertising it to potential females.  The other 
behaviors I called male crevice inspection and male crevice pass. 
 A wider variety of behaviors were observed in the field than in the laboratory.  
Field behaviors directed at females included in order of increasing escalation; mf 
approach, mf chase, mf zig zag, mf circling, mf hover, and mf spawning.  Field behaviors 
directed at males included in order of increasing escalation; mm approach, mm chase, 
mm lateral display, mm circle swim, mm fight, mm lip locking, and mm bite.  Other 
behaviors that were not directed at another fish included male circling/guarding nest, m 
crevice inspection, and m crevice pass, which were the same as under the laboratory 
setting. 
 Approach, lateral display, and chase behaviors were similar when directed 
towards either males or females.  During approaches, the sound producing male would 
slowly approach the other fish, from any direction, and vocalize.  Lateral display 
behaviors were rather variable, and consisted of the sound producing male displaying 
next to another fish with its fins erect.  If the other fish was a male, it would show no 
response, swim away, or respond back with a lateral display.  If the other fish was a 
19 
 
female, it would never respond with another lateral display, and would not respond at 
all, or swim away.  Chases consisted of a male quickly pursuing another fish for a 
distance greater than approximately 10 cm.  During courtship, mf zig zag consisted of 
the male swimming very rapidly between the nest and the female being courted.  The 
zig zag behavior seemed to function as a way of drawing the female closer to the nest, 
and only occurred with females that were already relatively close to the nest.  Male-
female circling behavior would occur closer to the nest than mf zig zag behaviors, and 
consisted of the male rapidly orbiting the female numerous times in rapid succession.  
The next behavior that typically took place was the mf hover, which took place in very 
close proximity to the nest.  During mf hovers, the male would hover just above or 
beside the female, directly in front of the nest.  The next behavior was typically mf 
spawning, during which time the male and female would swim together along the nest 
crevice, vibrating their bodies and presumably releasing gametes. 
 More escalated agonistic behaviors included mm circle swim, which involved two 
males chasing each other?s tails in a circular pattern.  Male-male fights and mm lip 
locking were the two most advanced agonistic behaviors.  Male-male fights consisted of 
the two males circling each other while simultaneously hitting the sides of their bodies 
together, or hitting the each other with their heads.  Male-male fights occurred 
extremely quickly and required slowing the video down in order to decipher what was 
happening.  Male-male lip locking has been described previously for one species (C. 
galactura ; Phillips and Johnston 2008), and consisted of the males locking at the mouth 
for a short period of time.  When mm lip locking did occur in C. venusta, it always 
20 
 
occurred just prior to a mm fight.  Male-male fights, however, did not have to be 
preceded by a mm lip lock. 
 Variation was observed in the type of sound that accompanied behaviors, as well 
as the degree to which behaviors were accompanied by sound in both the laboratory 
(Table 2), and field (Table 3).  In the laboratory, mm approaches were accompanied by 
knocks 29.11% (n=158, total number of mm approaches) of the time, growls 59.49% of 
the time (n=158), and no sound 10.76% (n=158) of the time.  Male-male chases were 
accompanied by knocks 17.17% (n=99) of the time, growls 62.63% (n=99) of the time, 
and no sound 16.16% (n=99) of the time.  Male-male lateral displays were accompanied 
by knocks 42.5% (n=240) of the time, growls 51.25% (n=240) of the time, and no sound 
6.25% (n=240) of the time. 
In the laboratory, mf approaches were accompanied by knocks 19.34% (n=181) 
of the time, growls 64.64% (n=181) of the time, and no sound 16.02% (n=181) of the 
time.  Male-female lateral display behaviors were accompanied by knocks 11.41% 
(n=184) of the time, growls 72.83% (n=184) of the time, and no sound 15.22% (n=184) 
of the time.  Male-female chase behaviors were accompanied by knocks 25.00% (n=44) 
of the time, growls 63.64% (n=44) of the time, and no sound 11.36% (n=44) of the time.  
Male-female zig zag behaviors were accompanied by knocks 0% of the time (n=42), 
growls 92.9% (n=42) of the time, and no sound 7.14% (n=42) of the time.  Male-female 
circling behavior was accompanied by knocks 5.1% (n=118) of the time, growls 84.7% 
(n=118) of the time, and no sound 10.2% (n=118) of the time.  Male-female hovers were 
accompanied by knocks 1.9% (n=53) of the time, growls 92.4% (n=53) of the time, and 
21 
 
no sound 5.7% (n=53) of the time.  Male guarding/circling nest was accompanied by 
knocks 7.1% (n=14) of the time, growls 92.9% (n=14) of the time, and no sound 0% 
(n=14) of the time.  Male crevice inspections were accompanied by knocks 1.7% (n=121) 
of the time, growls 67.8% (n=121) of the time, and no sound 29.8% (n=121) of the time.  
Male crevice passes were accompanied by knocks 0% (n=51) of the time, growls 25.5% 
(n=51) of the time, and no sound 72.5% (n=51) of the time.   
 In the field, mm approaches were accompanied by knocks 46.7% (n=107) of the 
time, growls 36.4% (n=107) of the time, and no sound 16.8% (n=107) of the time.  Male-
male chases were accompanied by knocks 30% (n=40) of the time, growls 55% (n=40) of 
the time, and no sound 15% (n=40) of the time.  Male-male lateral displays were 
accompanied by knocks 50.4% (n=141) of the time, growls 26.2% (n=141) of the time, 
and no sound 23.4% (n=141) of the time.  Male-male circle swims were accompanied by 
knocks 85.4% (n=48) of the time, growls 0% (n=48) of the time, and no sound 14.6% 
(n=48) of the time.  Male-male fighting was accompanied by knocks 89.4% (n=47) of the 
time, growls 0% (n=47) of the time, and no sound 10.6% (n=47) of the time.  Male-male 
lip locking was accompanied by knocks 60% (n=10) of the time, growls 0% (n=10) of the 
time, and no sound 40% (n=10) of the time.  Male-male biting was never accompanied 
by sound (n=10).   
 In the field, mf approaches were accompanied by knocks 0% (n=110) of the time, 
growls 95.5% (n=110) of the time, and no sound 4.5% (n=110) of the time.  Male-female 
chases were accompanied by knocks 0% (n=58) of the time, growls 96.6% (n=58) of the 
time, and no sound 3.4% (n=58) of the time.  Male-female zig zag (n=78), mf circling 
22 
 
(n=112), mf hover (n=83), and mf spawning (n=33) behaviors were all accompanied by 
growls 100% of the time.  Male circling/guarding behaviors were accompanied by 
knocks 0% (n=133) of the time, growls 86.5% (n=133) of the time, and no sound 13.5% 
(n=133) of the time.  Male crevice inspections were accompanied by knocks 0% (n=167) 
of the time, growls 72.5% (n=167) of the time, and no sound 27.5% (n=167) of the time.  
Male crevice passes were accompanied by knocks 0% (n=71) of the time, growls 4.2% 
(n=71) of the time, and no sound 95.8% (n=71) of the time. 
 
Acoustics 
No correlation was found between distance from the hydrophone and level for 
either the fundamental frequency (Pearson correlation = 0.002, P = 0.931, N = 1397; 
Figure 1-3a) or the first harmonic frequency (Pearson correlation = -0.021, P = 0.438, N = 
1397; Figure 1-3b) of bursts recorded in the lab.  A significant negative correlation was 
found, however, between distance from the hydrophone and level for both the 
fundamental frequency (Pearson correlation = -0.384, P < 0.001, N = 143) and level of 
the first harmonic frequency (Pearson correlation = -0.378, P < 0.001, N = 143) of knocks 
recorded in the lab (Figure 1-4a).  After correction, the correlation was eliminated for 
knocks (Figure 1-4b). 
In the laboratory, pulse interval showed a statistically significant positive 
correlation with fish standard length for both approach (Pearson correlation = 0.564, P = 
0.015, N = 18), and lateral display behaviors (Pearson correlation = 0.638, P = 0.019, N = 
13; Figure 1-5b,d).  Burst harmonic frequency showed a statistically significant negative 
23 
 
correlation with fish standard length for both approach (Pearson correlation = -0.665, P 
= 0.002, N = 19), and lateral display behaviors (Pearson correlation = -0.660, P = 0.007, N 
= 15; Figure 1-5a,c).  No other signal parameters showed a significant correlation with 
body size in the laboratory.  No correlations between signal parameters and body size 
could be performed in the field. 
An overall difference was found in duration between the three call types in both 
the laboratory (repeated measures ANOVA: F2,9 = 26.37, P <0.001) and field (repeated 
measures ANOVA: F2,6 = 62.10, P <0.001; Table 1-4).  Under laboratory conditions, burst 
duration (mean ? SD; 0.08928 ? 0.03689 sec, N = 11) was significantly longer than both 
knock duration (0.01576 ? 0.005279 sec, N=11; Bonferroni corrected post hoc: P < 
0.001) and pulse duration (0.007576 ? 0.001006 sec, N=11; Bonferroni corrected post 
hoc: P < 0.001).  Under field conditions, burst duration (0.04527 ? 0.01555 sec, N=8) was 
also significantly longer than both knock duration (0.009314 ? 0.001720 sec, N=8; 
Bonferroni corrected post hoc: P = 0.001) and pulse duration (0.005901 ? 0.000972 sec, 
N=8; Bonferroni corrected post hoc: P < 0.001).  Knocks were significantly longer in 
duration than pulses under both laboratory (Bonferroni corrected post hoc: P = 0.001), 
and field (Bonferroni corrected post hoc: P = 0.010) conditions. 
Paired-samples T tests (Table 1-4) found that, in the laboratory, the fundamental 
frequency of knocks (165.07 ? 58.64 Hz, N=12) was significantly higher than the 
fundamental frequency of bursts (80.10 ? 3.93 Hz, N=12; paired t test: t11=4.905, P < 
0.001), and that the first harmonic frequency of knocks (587.68 ? 101.43 Hz, N=12) was 
also significantly higher than the first harmonic of bursts (301.58 ? 45.11 Hz, N=12; 
24 
 
paired t test: t11=8.806, P < 0.001).  The level of the fundamental frequency of knocks 
(99.32 ? 5.75 dB, N=12) was significantly greater than the level of the fundamental 
frequency of bursts (76.79 ? 5.50 dB, N=12; paired t test: t11=13.544, P < 0.001), and the 
level of the first harmonic of knocks (102.73 ? 7.00 dB, N=12) was significantly greater 
than the level of the first harmonic of bursts (70.70 ? 5.72 dB, N=12; paired t test: 
t11=17.080, P < 0.001).   
Paired-samples T tests (Table 1-4) also found that, in the field, the fundamental 
frequency of knocks (297.03 ? 45.77 Hz, N=8) was significantly higher than the 
fundamental frequency of bursts (114.15 ? 5.77 Hz, N=8; paired t test: t7=10.870, P < 
0.001), and that the first harmonic frequency of knocks (615.11 ? 40.19 Hz, N=8) was 
significantly higher than the first harmonic of bursts (391.89 ? 45.79 Hz, N=8; paired t 
test: t7=14.718, P < 0.001).  The level of the fundamental frequency of knocks (89.84 ? 
11.81 dB, N=8) was significantly greater than the level of the fundamental frequency of 
bursts (68.40 ? 9.42 dB, N=8; paired t test: t7=7.360, P < 0.001), and the level of the first 
harmonic frequency of knocks (95.08 ? 9.04 dB, N=8) was also significantly greater than 
the level of the first harmonic of bursts (73.03 ? 9.37 dB, N=8; paired t test: t7=9.468, P < 
0.001). 
 Signal parameters did not differ between contexts (male-male versus male-
female interactions) for approach behaviors (growls; MANCOVA: F10,10=0.867, P=0.587; 
knocks; MANCOVA: F5,8=0.903, P=0.523; Table 1-5) or lateral display behaviors (growls; 
MANCOVA: F10,4=1.743, P=0.312; knocks; MANCOVA: F5,10=1.166, P=0.389; Table 1-6) in 
the laboratory.  There was also no significant difference between contexts for approach 
25 
 
behaviors (growls; MANOVA: F10,4=0.647, P=0.738; Table 1-7) or chase behaviors 
(growls; MANOVA: F10,2=0.570, P=0.777; Table 1-8) in the field. 
Growl signal parameters did not differ between specific behaviors in the 
laboratory (MANCOVA: F20,48=0.709, P=0.797; Table 1-9), or field (MANOVA: 
F60,204=0.933, P=0.615; Table 1-10).  Knocks were also not significantly different between 
specific behaviors in the laboratory (MANCOVA: F5,17=0.449, P=0.808; Table 9) or in the 
field (MANOVA: F5,8=2.407, P=0.129; Table 1-11). 
An overall main effect of location (laboratory vs field) was found for growls and 
knocks produced during approach behaviors (growls; MANOVA: F6,21=15.072, P < 0.001; 
knocks; MANOVA: F3,17=12.680, P < 0.001).  Individual univariate ANOVA?s with 
Bonferroni correction found that all signal parameters included in the MANOVA?s for 
growls and knocks were significantly different between the laboratory and field (Table 
1-12).  An overall main effect of location was also found for knocks produced during 
lateral display behaviors (MANOVA: F3,17=9.897, P = 0.001).  Univariate ANOVA?s with 
Bonferroni correction found that knock fundamental frequency and harmonic frequency 
differed between the laboratory and field (fundamental frequency; F=23.957, P < 0.001, 
harmonic frequency; F = 9.043, P = 0.007), while knock duration did not (F=3.527, 
P=0.076; Table 1-13).  An overall main effect of location was also found for growls 
produced during crevice inspection behaviors (MANOVA: F6,7=4.541, P=0.034; Table 11).  
Univariate ANOVA?s with Bonferroni correction found that pulse duration (ANOVA: F = 
11.354, P = 0.006), burst interval (ANOVA: F = 15.050, P = 0.002), and burst fundamental 
frequency (ANOVA: F = 20.446, P = 0.001) were significantly different between the 
26 
 
laboratory and the field for growls produced during crevice inspection behaviors (Table 
1-14). 
 
Discussion 
 This descriptive study provides new information that advances our 
understanding of sound in the reproductive behaviors of C. venusta.  An appreciable 
amount of variability exists in the sound production of different species within the 
genus Cyprinella (Delco 1960; Winn and Stout 1960; Stout 1975; Phillips and Johnston 
2008; Phillips and Johnston 2009; Phillips et al. 2010), and my results indicate that 
sounds used by C. venusta are equally distinct.  Phillips and Johnston (2009) found that 
although the tricolor shiner (Cyprinella trichroistia) and the Tallapoosa shiner (Cyprinella 
gibbsi) produced growls similar to that of C. venusta during courtship behaviors, they 
rarely produce sounds during male-male aggressive interactions, and when they did, 
they were growls, not knocks.  Phillips et al. (2010) also found that the Edwards plateau 
shiner (Cyprinella lepida) produces only growls and single pulses during agonistic 
interactions.  Cyprinella trichroistia and C. gibbsi also produce chirps and rattles, two 
types of sounds that are not produced by C. venusta.  Cyprinella galactura and 
Cyprinella callisema both produce knocks and growls under similar contexts as C. 
venusta and do not produce chirps or rattles.  Differences between C. venusta and other 
species existed in dominant frequency of knocks, which varied between 79 - 92 Hz in 
other species of Cyprinella, and was approximately 165 Hz (293 Hz in field) in C. venusta.  
Pulse duration was much shorter in C. venusta (laboratory 7.5 ms; field 5.9 ms) than in 
27 
 
other species of Cyprinella, which varied between 15.8 - 34.3 ms.  Pulse interval was also 
shorter in C. venusta (laboratory 13.3 ms) than in other species, which ranged from 17.3 
to 27.3 ms.  Due to shorter pulses and pulse intervals, the pulse rate of C. venusta 
(approximately 81 pulses/sec) was considerably higher than other species (24.3 ? 32.8 
pulses/sec). 
Although I was unable to statistically compare the acoustic signals of C. venusta 
to other species, the relatively drastic differences in signal parameters such as pulse 
duration and pulse interval suggests that acoustic signals are likely different, and could 
act as an interspecific isolating mechanism (Wells and Henry 1992).  This is especially 
intriguing due to the fact that C. venusta?s range overlaps with so many other sister 
species. 
The behavioral repertoire observed in the laboratory differed slightly from that 
observed in the field for male-female interactions.  Spawning events, for example, were 
only observed in the field.  This is not entirely surprising considering that spawning is the 
end product of courtship behaviors, which have been shown to be negatively affected 
by stressors in the Atlantic cod (Gadus morhua; Morgan et al. 1999).  As with all 
laboratory studies, a large number of differences beyond my control existed between 
the physical conditions in the laboratory and the field, any number of which could have 
played a role in the absence of spawning behaviors in the laboratory.  The absence of 
flow, which was present at all field nest sites, and which is thought to be an important 
factor in nest selection in C. venusta (Baker et al. 1994), I believe, may be the reason 
that spawning was not observed in the laboratory.  Because of the energetic investment 
28 
 
associated eggs, females should be more selective about a nest than males when it 
comes to spawning.  A nest associated with no flow is likely less desirable because there 
is less oxygen availability.  Females were probably more responsible for the lack of 
spawning in the laboratory, a suggestion that is supported by the fact that courtship 
behaviors preceding spawning (mf zig zag, mf circling, and mf hover) were seen in both 
the laboratory and field, suggesting that males were prepared to spawn, but females 
were not. 
Male-female lateral display was the only other behavior (besides spawning) that 
occurred between males and females that was not observed under both laboratory and 
field conditions.  I do not have a clear explanation as to why this discrepancy was 
observed.  One possibility is that the inability of the female to exit the nesting arena, 
combined with the lack of motivation to cooperate with the courting male, provided a 
mixed signal to the male that is probably not normally encountered in the field.  This 
may result in a type of displacement activity (Tinbergen and Vaniersel 1948) by the 
male, in this case, a lateral display.  Lateral displays are typically one of the lowest levels 
of the agonistic behavioral sequence performed between two males challenging each 
other.  The reason I believe this may be a displacement activity is that the male does not 
proceed along the normal escalation of agonistic behaviors with the female, but 
continues to perform courtship behaviors and nest guarding behaviors.  If a mf lateral 
display were part of a normal aggressive interaction between the male and female, 
some normal progression of aggression should have been observed.  However, such a 
29 
 
progression was never observed in the laboratory, and a mf lateral display was never 
observed at all under field conditions. 
Observation of only the first three agonistic behaviors between males (mm 
approach, mm chase, and mm lateral display) in the laboratory was somewhat of a 
surprise to us.  I originally believed that the constraints of the tank in the lab would 
force males into highly escalated states of aggression.  I believe that the most likely 
explanation for the lack of escalated agonistic behaviors between males in the 
laboratory is the same as for the lack of spawning between males and females.  Just as 
spawning is very costly for females, escalating to advanced agonistic behaviors is more 
energetically costly for males, and poses a greater risk of injury.  If the nest is considered 
substandard, and only a limited number of females are present, a male may not 
consider it beneficial to escalate to a higher aggressive state. 
 Based on the accompaniment of growls and knocks with certain behaviors, an 
indication of the function of the sounds may be inferred.  In both the laboratory and 
field, behaviors that are undoubtedly part of courtship such as mf zig zag, mf circling, mf 
hover, and mf spawning (mf spawning observed in the field only) are almost always 
exclusively associated with growls sounds.  Likewise, behaviors that were exclusively 
performed during male-male agonistic interactions such as mm circle swim, mm fight, 
and mm lip lock (all of which were only observed in the field) were associated 
exclusively with knock sounds.  From these associations, it is clear that knocks carry a 
more aggressive meaning, while growls are used more to communicate courtship. 
30 
 
 There are, however, a number of behaviors that are observed during both male-
male and male-female interactions.  These behaviors include approaches, chases, and 
lateral displays.  The association of knocks and growls with these behaviors is not as 
clear as with behaviors that are unique to courtship and agonistic contexts.  During 
male-male and male-female approaches, chases, and lateral displays produced in the 
laboratory, growls and knocks both make up a substantial percentage of the 
accompanying sounds.  Male-male approaches, chases, and lateral displays in the field 
also show accompaniment by a mixture of knocks and growls.  However, mf approaches 
and mf chases in the field are unique in that when they are accompanied by sound 
(approximately 95% of the time), they are always accompanied by growls, never by 
knocks.  This is an interesting finding as it shows a significant discrepancy between 
laboratory and field findings.  Because field conditions are the more natural setting, the 
lack of knocks during mf approach and mf chase behaviors should be considered the 
normal condition.  The association of knocks with mf approach and mf chase behaviors 
in the laboratory may be explained through the hypothesis that under laboratory 
conditions, the inability of a reproductively unmotivated female to leave the nesting 
area could prompt males to increase the level of aggression towards that female, 
whereas under field conditions, an unmotivated female would not likely persist around 
the nesting area. 
 My results also indicated that during highly aggressive behaviors between males, 
the association of sound production with behaviors decreased (as with mm lip lock), or 
ceased altogether (as with mm bite).  These results corroborate, to some extent, those 
31 
 
found for C. galactura (Phillips and Johnston, 2008).  Phillips and Johnston (2008) found 
that during highly aggressive behaviors (mm circle swim and mm lip lock), C. galactura 
did not produce any sounds.  They attributed this to a possible tradeoff between the 
energetic costs of calling and the benefits acquired from the calls.  If sound production 
has not resolved an aggressive encounter at lower levels of aggression, it is unlikely to 
help once behaviors have escalated to physical contests.  I agree with this logic, but 
found that C. venusta differs in the level at which sound production is abandoned.  
Cyprinella venusta continues to produce sounds at a high rate during mm circle swims 
(85.5%) and mm lip locks (60.0%), and only ceases to accompany aggressive behaviors 
with sounds during bite behaviors.  I am unsure whether this relates to differences in 
the physiology of sound production, or whether the payoff of sound production at 
higher levels of aggression differs between C. galactura and C. venusta.  To answer 
these questions, more elaborate studies will be needed comparing the energetic costs 
of sound production, and the benefit of sounds during agonistic behaviors in the two 
species.  Both of these studies will probably require knowledge of the sound producing 
mechanism in these species for purposes of muting or stimulating sound production.  
 I also found that C. venusta continued to produce sounds throughout the entire 
courtship and spawning process, with sounds being produced during 100% of the 
observed spawns.  Sound production during the act of spawning has been found to be 
inconsistent across several closely related species of the family Cyprinidae.  Stout (1975) 
reported that no sounds were made during the act of spawning in the satinfin shiner 
(Cyprinella analostanus).  Phillips and Johnston (2008) reported that although sounds 
32 
 
were observed during the act of spawning in C. galactura, they were produced less 
frequently relative to other courtship behaviors.  Johnston and Vives (2003) reported 
that, like C. venusta, the ornate minnow (Codoma ornata) produced sounds during 
100% of observed spawning acts.  The reason for the inconsistency in sound production 
during spawning is unknown, and especially interesting considering the importance of 
effective communication at the moment of gamete release. 
 The correlations between burst harmonic frequency and pulse interval with fish 
standard length may be used by females to evaluate the fitness of a male.  In the field, I 
observed that the dominant male guards the nest from potential predators such as 
sunfish and other smaller fishes who frequently attempt to eat eggs from the crevice 
nest.  Large sunfishes were not observed attempting to eat from the nest, possibly 
because of difficulties reaching eggs deposited in the small crevice.  On a few occasions, 
the dominant male was observed chasing a sunfish away from a nest at which he had 
frequently spawned.  Since the defending male protects the nest from egg predators 
such as sunfishes or other fishes, a female may prefer to spawn with a larger male that 
is better suited to protect the nest.   
An interesting side note is that the male C. venusta sometimes produced knocks 
during the sunfish chase, indicating that knocks are not reserved specifically for 
interspecific communication, but also function as a general sign of aggression.  The use 
of agonistic acoustic signals is typically found in intraspecific interactions.  However, 
there are records in the literature indicating agonistic acoustic signals being used 
33 
 
interspecifically (Brawn 1961; Gray and Winn 1961; Winn et al. 1964; Salmon 1967; 
Ballantyne and Colgan 1978; Rigley and Muir 1979) 
 I found in the laboratory and field that both knocks and growls were bimodal in 
their frequency distribution, a finding that has not yet been mentioned in the sound 
production literature for fishes in the genus Cyprinella.  The dominant frequency for C. 
venusta was typically the lower (fundamental) frequency for both knocks and growls, 
and closely resembles the dominant frequencies of growls and knocks of other closely 
related species in the genus Cyprinella (Delco 1960; Phillips and Johnston 2008, 2009; 
Phillips et al. 2010).  I am not yet certain of the significance of this bimodal frequency 
distribution, but (D.E. Holt and C.E. Johnston, unpublished data) showed that the 
harmonic frequency of C. venusta growls corresponds nicely to a quiet window in their 
natural spawning habitat.  This, along with the fact that body size is correlated with the 
harmonic frequency and not the fundamental frequency suggests that the harmonic 
frequency may carry a greater biological significance than the dominant, fundamental 
frequency. 
 Although knocks were more typically associated with aggressive contexts, and 
growls were more associated with courtship contexts, the lack of a difference in growl 
signal parameters between male-male and male-female interactions for approach, 
lateral display, and chase behaviors was surprising.  Other studies with closely related 
fishes have found differences in the characteristics of sounds produced under different 
contexts.  Phillips and Johnston (2008), for example, found that in the whitetail shiner 
(C. galactura), growls produced during lateral display behaviors differed in numerous 
34 
 
signal parameters between male-male (agonistic) and male-female (courtship) 
interactions including dominant frequency, pulse duration, pulse interval, pulse rate, 
and burst duration.  The lack of this specialization in C. venusta was further seen by the 
absence of a difference in growl or knock signal parameters between not just agonistic 
and courtship contexts, but between any behaviors able to be compared for C. venusta.  
These results portray a more general function of acoustic communication by C. venusta 
during courtship and agonistic behaviors than I had expected based on other studies of 
closely related species.  According to my results, a growl is a growl regardless of what 
type of behavior it is associated with, and a knock is a knock. 
 I am unsure why growls showed differences between the laboratory and the 
field.  These differences were highly significant for approach behaviors, with sounds 
recorded in the laboratory generally being sluggish, and more drawn out with longer 
pulse durations, burst durations, and burst intervals, and slower pulse and burst rates.  
These same trends were seen for growls produced during crevice inspection behaviors, 
although only pulse duration, burst interval, and burst rate were statistically significant.  
Growl harmonic frequency was also significantly lower under laboratory conditions for 
both approach and crevice inspection behaviors.  A possible explanation for the 
differences in growls between the laboratory and field could be a lack of motivation.  If 
males are less motivated to court females, they may expend less energy in sound 
production.  If muscle contraction on the swim bladder is the source of sound 
production in C. venusta, a slower rate of muscle contraction will result in slower pulse 
repetition rate (pulse rate), which is directly tied to the fundamental frequency of the 
35 
 
growl.  Exposure to stressors associated with capture, transport, and movement 
through an unfamiliar laboratory environment could also decrease motivation and 
energy availability (Bonga 1997), which could have a similar effect on growl 
fundamental frequency, pulse duration, and pulse rate. 
 The slower burst rate under laboratory conditions is naturally explained by the 
longer burst durations and intervals.  However, I do not have an explanation as to why, 
under laboratory conditions, burst duration and interval are longer than under field 
conditions.  There are many parameters that were unable to be controlled between the 
laboratory and the field that may have led to differences in signal parameters.  For 
example, Penteriani (2003) found that the Eurasian Eagle Owl (Bubo bubo) decreased 
call duration in response to a more sparse conspecific density.  Call duration is an honest 
signal of male quality in B. bubo, and when competition is scarce, males do not put as 
much energy into calling.  Conspecific density in the current study was very different 
between laboratory and field settings, and could have played a role in burst duration 
and interval.  To answer this question, further studies will have to be performed that 
control for density. 
 I am also unsure why knock fundamental and harmonic frequencies are 
significantly higher under field conditions than in the laboratory.  This cannot be 
explained through pulse repetition rate because knocks are composed of a single unit.  
The only other explanation is that the difference in knock duration between laboratory 
and field conditions is actually real, even though it is not statistically significant in the 
current study.  A shorter knock duration would result in a higher acoustic frequency for 
36 
 
knocks (Connaughton et al. 2002).  Although knock duration between the laboratory and 
the field for C. venutsta was not significant for lateral display behaviors, it was 
significant for approach behaviors, and showed the same trend of being shorter in the 
field for both behaviors.  I believe that knock duration is the most likely explanation for 
the difference in frequency between laboratory and field conditions. 
 Although it is not the first to describe sound production by C. venusta, the 
current study is the first to provide a detailed description of sounds and associated 
behaviors in C. venusta, and a comparison of sounds and behaviors performed in the 
laboratory to those in the field for the genus Cyprinella.  This is also the first study within 
the genus Cyprinella that provides data concerning the decibel levels of vocalizations.  
Knowing decibel levels of acoustic signals is crucial in understanding how fish 
communicate in their environment.  Such information, in conjunction with 
environmental information such as propagation distances and the environmental 
soundscape of the habitat can be used to answer important questions such as the active 
area of vocalizations and the potential effect of natural and anthropogenic noises.  
Finally, the current study illustrates the importance of making field comparisons to 
substantiate findings in the laboratory. 
 A solid groundwork of detailed acoustic descriptions for the genus Cyprinella has 
been laid by Phillips and Johnston (2008; 2009; 2010), to which the current study adds 
another species.  Future studies concerning the genus Cyprinella should attempt to 
include field comparisons in order to verify laboratory findings, and to see if the 
differences found between the laboratory and the field in C. venusta are consistent 
37 
 
across other species.  The incorporation of decibel levels into acoustic descriptions 
should also be considered. 
 
References: 
Aiken, R.B.  1985.  Sound production by aquatic insects.  Biological Reviews.  60:163-211. 
Akamatsu, T., Okumura, T., Novarini, N. and Yan, H.Y.  2002.  Empirical refinements  
applicable to the recording of fish sounds in small tanks.  Journal of the 
Acoustical Society of America.  112:3073-3082. 
Baker, J.A., Kilgore, J.K., and Foster, S.A.  1994.  Population variation in spawning current  
speed selection in the blacktail shiner, Cyprinella venusta (Pisces: Cyprinidae).  
Environmental Biology of Fishes.  39:357-364. 
Ballantyne, P.K. and Colgan, P.W.  1978.  Sound production during agonistic and  
reproductive behavior in the pumpkinseed (Lepomis gibbosus), the bluegill 
(Lepomis macrochirus), and their hybrid sunfish.  II.  Response.  Biol. Behav.  
3:221-292. 
Bonga, S.E.W.  1997.  The stress response in fish.  Physiological Reviews.  77:591-625. 
Brawn, V.M.  1961.  Sound production by the cod (Gadus callarias L.).  Behavior.  18:239- 
255. 
Bradbury, J.W. and Vehrencamp, S.L.  1998.  Principles of Animal Communication.   
Sinauer, Sunderland, Mass. pp. 5. 
Colson, D.J., Patek, S.N., Brainerd, E.L. and Lewis, S.M.  1998.  Sound production during  
38 
 
feeding in hippocampus seahorses (Syngnathidae).  Environmental Biology of 
Fishes.  51:221-229. 
Connaughton, M.A., Fine, M.L. and Taylor, M.H.  2002.  Weakfish sonic muscle: influence  
of size, temperature and season.  The Journal of Experimental Biology.  
205:2183-2188. 
Crawford, J.D., Hagedorn, M. and Hopkins, C.D.  1986.  Acoustic communication in an  
electric fish, Pollimyrus isidori (Mormyridae).  Journal of Comparative Physiology 
A.  159:297-310. 
Delco, E.A.  1960.  Sound discrimination by males of two cyprinid fishes.  Texas Journal  
of Science.  12:48-54. 
Dooling, R.J., Lohr, B., and Dent, M.L.  2000.  Hearing in birds and reptiles.  In:  
Comparative hearing: Birds and Reptiles.  (eds) Dooling, R.J., Fay, R.R. and 
Popper, A.N.  New York: Springer-Verlag. 
Drewry, C.E.  1962.  Some observations of courtship behavior and sound production in  
five species of Fundulus.  Masters thesis, University of Texas at Austin. 
Gray, G.A. and Winn, H.E.  1961.  Reproductive Ecology and sound production of the  
toadfish, Opsanus tau.  Ecology.  42:247-282. 
Janik, V.M.  2005.  Underwater acoustic communication networks in marine mammals.   
In: Animal Communication Networks.  (ed)  McGregor, P.K.  pp. 390-415.  
Cambridge: University Press. 
Johnston, C.E. and Johnson, D.L.  2000a.  Sound production during the spawning season  
39 
 
in cavity-nesting darters of the subgenus Catonotus (Percidae: Etheostoma).  
Copeia.  2000:475-481. 
Johnston, C.E. and Johnson, D.L.  2000b.  Sound production in Pimephales notatus  
(Rafinesque)(Cyprinidae).  Copeia.  2000:567-571. 
Johnston, C.E. and Vives, S.P.  2003.  Sound production in Codoma ornate (Girard)  
(Cyprinidae).  Environmental Biology of Fishes.  68:81-85. 
Kaatz, I.M. and Lobel, P.S.  2001.  A comparison of sounds recorded from a catfish  
(Orinocodoras eigenmanni, Doradidae) in an aquarium and in the field.  
Biological Bulletin.  201:278-280. 
Ladich, F.  1998.  Sound characteristics and outcome of contests in male croaking  
gouramis (Teleostei).  Ethology.  104:517-529. 
Ladich, F.  2004.  Sound production and acoustic communication.  In: The senses of fish  
adaptations for the reception of natural stimuli.  (eds) Von der Emde, G., 
Mogdans, J. and Kapoor, B.G.  pp. 210-230.  New Delhi: Narosa Publishing House. 
Ladich, F. and Myrberg, A. A. 2006. Agonistic behaviour and acoustic communication. In:  
Communication in Fishes. Vol. 1 (Ed. by F. Ladich, S. P. Collin, P. Moller & B. G. 
Kapoor), pp. 122e148. Enfield, New Hampshire: Science Publishers. 
Ladich, F. and Yan, H.Y.  1998.  Correlation between auditory sensitivity and vocalization  
in anabantoid fishes.  Journal of Comparative Physiology A.  182:737-746. 
Lugli, M.P.  1997.  Response of male goby, Padogobius martensii, to aggressive sound  
playback following pre-experimental visual stimulation.  Behaviour 134:1175-
1188. 
40 
 
Lugli, M.P., Pavan, G., Torricelli, P. and Bobbio, L.  1995.  Spawning vocalizations in male  
freshwater gobiids (Pices, Gobiidae).  Environmental Biology of Fishes.  43:219-
231. 
Lugli, M.P. and Torricelli, P.  1999.  Prespawning sound production in Mediterranean  
sand-gobies.  Journal of Fish Biology.  54:691-694. 
McComb, K. and Reby, D.  2005.  Vocal communication networks in large terrestrial  
mammals.  In: Animal communication Networks (ed) McGregor, P.K.  pp 372-
389.  Cambridge: University Press. 
Morgan, M.J., Wilson, C.E. and Crim, L.W.  1999.  The effect of stress on reproduction in  
the Atlantic cod.  Journal of Fish Biology.  54:477-488. 
Myrberg, Jr., A.A.  1997.  Sound production by a coral reef fish (Pomacentrus partitus):  
Evidence for a vocal, territorial ?keep-out? signal.  Bulletin of Marine Science.  
60:1017-1025. 
Narins, P.M., Feng, A.S., Fay, R.R. and Popper, A.N. (eds).  2007.  Hearing and sound  
communication in Amphibians (Springer handbook of auditory research).  New 
York: Springer. 
Nelson, J.S.  (ed).  2006.  Fishes of the World (4th edition).  pp. 3-4.  New Jersey:  John  
Wiley and Sons Inc. Penteriani, V.  2003.  Breeding density affects the honesty of 
bird vocal displays as possible indicators of male/territory quality.  Ibis.  145:127-
135. 
Penteriani, V.  2003.  Breeding density affects the honesty of bird vocal displays as  
41 
 
possible indicators of male/territory quality.  The International Journal of Avian 
Science.  145:E127-E135. 
Phillips, C.T. and Johnston, C.E.  2008.  Sound production and associated behaviors in  
Cyprinella galactura.  Environmental Biology of Fishes.  82:265-275. 
Phillips, C.T. and Johnston, C.E.  2009.  Evolution of acoustic signals in Cyprinella: degree  
of similarity in sister species.  Journal of Fish Biology.  74:120-132. 
Phillips, C.T., Johnston, C.E. and Henderson, A.R.  2010.  Sound production and spawning  
behavior in Cyprinella lepida, the Edwards Plateau shiner.  The Southwestern 
Naturalist.  55:129-135. 
Pruzsinszky, I. and Ladich, F.  1998.  Sound production and reproductive behavior of the  
armored catfish Corydoras paleatus (Callichthyidae).  Environmental Biology of 
Fishes.  53:183-191. 
Rigley, L. and Muir, J.  1979.  The role of sound production by the brown bullhead  
Ictalurus nebulosus.  Proceedings of the Pennsylvania Academy of Science.  
53:132-134. 
R?mer, H.  1998.  The sensory ecology of acoustic communication in insects (Springer  
handbook of auditory research).  New York: Springer. 
Salmon, M.  1967.  Acoustic behavior of the menpachi, Myripristis berndti, in Hawaii.   
Pac. Sci.  11:364-381. 
Slater, P.J.B.  2003.  Fifty years of bird song research: A case study in animal behavior.   
Animal Behaviour.  65:633-639. 
Stout, J.F.  1963.  The significance of sound production during the reproductive behavior  
42 
 
of Notropis analostanus (Family Cyprinidae).  Animal Behaviour.  11:83-92. 
Stout, J.F.  1975.  Sound communication during the reproductive behavior of Notropis  
analostanus (Pisces: Cyprinidae).  American Midland Naturalist.  94:296-325. 
Tinbergen, N. and Vaniersel, J.J.A.  1948.  Displacement reactions in the 3-spined  
stickleback.  Behavior.  1:56-63. 
Valinski, B.W. and Rigley, L.  1981.  Function of sound production by the skunk loach  
Botia horae (Pisces, Cobitidae).  Z. Tierpsychol.  55:162-172. 
Wells, M.M. and Henry, C.S.  1992.  The role of courtship songs in reproductive isolation  
among populations of green lacewings of the Genus Chrysoperia (Neuroptera: 
Chrysopidae).  Evolution.  46:31-42. 
Winn, H.E. and Stout, J.F.  1960.  Sound production by the satinfin shiner, Notropis  
analostanus and related fishes.  Science.  132:222. 
Winn, H.E., Marshall, J.A., and Hazlett, B.  1964.  Behavior, diel activities, and stimuli that  
elicit sound production and reactions to sound in the longspine squirrelfish.  
Copeia.  2:413-425. 
  
43 
 
 
 
Figure 1-1 ? Experimental tank setup.  Underwater speaker is suspended behind a mesh 
barrier on the right hand side of the tank (according to the viewer?s perspective).  A 
hydrophone is suspended in front of an artificial nest (composed of a tile propped up on 
top of a brick), which is placed against the far wall of the tank.  A light suspended above 
the tank cast shadow?s onto the bottom grid, which together with the grid on the rear 
side of the tank allow the position of the fishes to be determined inside the tank. 
  
44 
 
 
 
Figure 1-2 ? Waveforms (blue lines with white background) and spectrograms (green 
signature with blue background) of C. venusta call repertoire.  Both the (a) growl and (c) 
knock were recorded from the same individual under laboratory conditions.  The growl 
is approximately 9 seconds long and represents an entire call.  (b) A closer look at the 
burst bracketed in the growl.  A total of 8 pulses making up the burst can be clearly 
seen, along with the relatively uniform pulse interval separating them.  The empty space 
on either side of the burst in (b) represents the burst interval between this burst and 
adjacent bursts.  Waveform amplitude values and units are arbitrary.  The 3 knocks are 
produced in succession by the same male.  Plate (d) is an expanded view of the middle 
knock of plate (c). 
  
Time (sec) Time (sec) 
Time (sec) Time (sec) 
45 
 
 
 
Figure 1-3 ? Scatterplot of burst fundamental frequency level (a) and first harmonic level 
(b) with distance of sound producing fish from the hydrophone.  Neither showed a 
significant correlation.  
  
b 
a 
46 
 
 
 
Figure 1-4 ? Scatterplot with best fit line of fundamental frequency and first harmonic 
levels of knocks and distance of fish from hydrophone uncorrected (a,b), and corrected 
(c,d).   First harmonic level of knocks and distance of fish from hydrophone (top left) 
uncorrected, and (bottom left) corrected.  Values were corrected by multiplying the 
distance from the hydrophone of each point by the slope of the best fit line (0.0727), 
and adding the product to the original dB value.   
  
b 
d c 
a 
47 
 
 
 
Figure 1-5 ? Scatterplots showing correlations between fish standard length and burst 
harmonic frequency during approach behaviors (a), and lateral display behaviors (c).  
The correlations between standard length and pulse interval are also shown for 
approach behaviors (b) and lateral display behaviors (d).  Data was combined for both 
agonistic and courtship contexts.  Pearson correlation showed a significant negative 
correlation between SL and frequency of the first harmonic for both approach (Pearson 
correlation = -0.665, P = 0.002, N = 19), and lateral display behaviors (Pearson 
correlation = -0.660, P = 0.007, N = 15).  Pulse interval also showed a significant positive 
correlation with SL for both approach (Pearson correlation = 0.564, P = 0.015, N = 18), 
and lateral display behaviors (Pearson correlation = 0.638, P = 0.019, N = 13). 
  
48 
 
Parameters Description 
Temporal:  
Pulse duration Time from beginning to end of one pulse. 
Pulse interval Time between identical points on adjacent 
pulses. 
Pulse rate Number of pulses within burst/burst 
duration. 
Burst duration Time from beginning to end of burst. 
Burst interval Time between the last pulse and the first 
pulse of adjacent bursts. 
Burst rate Number of bursts within a call/call 
duration. 
Knock duration Time from beginning to end of one knock. 
  
Spectral:  
Burst fundamental  
frequency 
Lower frequency of the two frequency 
peaks measured from a burst. 
Burst harmonic  
frequency 
Higher frequency of the two frequency 
peaks measured from a burst. 
Knock fundamental  
frequency 
Lower frequency of the two frequency 
peaks measured from a knock. 
Knock harmonic  
frequency 
Higher frequency of the two frequency 
peaks measured from a knock. 
  
Acoustic Intensity:  
Burst fundamental 
dB 
Sound pressure level (in dB re 1?Pa2/Hz) of 
burst fundamental frequency. 
Burst harmonic  
dB 
Sound pressure level (in dB re 1?Pa2/Hz) of 
burst harmonic frequency. 
Knock fundamental 
dB 
Sound pressure level (in dB re 1?Pa2/Hz) of 
knock fundamental frequency. 
Knock harmonic  
dB 
Sound pressure level (in dB re 1?Pa2/Hz) of 
knock harmonic frequency. 
 
Table 1-1 ? Definitions and terminology of signal parameters measured for acoustic 
signals of C. venusta.  All temporal parameters were measured from the waveform view 
and all spectral and acoustic intensity parameters were measured from the power 
spectrum view of Raven 1.3. 
  
49 
 
Laboratory Knocks Growls No Sound 
mm approach 29.2 60.0 10.8 
 (n=158) (n=158) (n=158) 
mm chase 18.2 63.9 17.9 
 (n=99) (n=99) (n=99) 
mm lateral display 42.5 51.2 6.3 
 (n=240) (n=240) (n=240) 
mf approach 19.2 64.7 16.1 
 (n=181) (n=181) (n=181) 
mf lateral display 11.5 72.9 15.6 
 (n=184) (n=184) (n=184) 
mf chase 25.0 63.7 11.3 
 (n=44) (n=44) (n=44) 
mf zig zag 0.0 92.9 7.1 
 (n=42) (n=42) (n=42) 
mf circling 5.1 84.8 10.1 
 (n=118) (n=118) (n=118) 
mf hover 1.9 92.5 5.6 
 (n=53) (n=53) (n=53) 
m guarding/circling nest 7.1 92.9 0.0 
 (n=14) (n=14) (n=14) 
m crevice inspection 1.7 68.3 30.0 
 (n=121) (n=121) (n=121) 
m crevice pass 0.0 25.9 74.1 
 (n=51) (n=51) (n=51) 
 
Table 1-2 ? Percentage of behaviors associated with sound in the laboratory. 
  
50 
 
Field Knocks Growls No Sound 
mm approach 46.8 36.5 16.7 
 (n=107) (n=107) (n=107) 
mm chase 30.0 55.0 15.0 
 (n=40) (n=40) (n=40) 
mm lateral display 50.4 26.3 23.3 
 (n=141) (n=141) (n=141) 
mm circle swim 85.5 0.0 14.5 
 (n=48) (n=48) (n=48) 
mm fight 89.4 0.0 10.6 
 (n=47) (n=47) (n=47) 
mm lip lock 60.0 0.0 40.0 
 (n=10) (n=10) (n=10) 
mm bite 0.0 0.0 100.0 
 (n=10) (n=10) (n=10) 
mf approach 0.0 95.5 4.5 
 (n=110) (n=110) (n=110) 
mf chase 0.0 96.6 3.4 
 (n=58) (n=58) (n=58) 
mf zig zag 0.0 100.0 0.0 
 (n=78) (n=78) (n=78) 
mf circling 0.0 100.0 0.0 
 (n=112) (n=112) (n=112) 
mf hover 0.0 100.0 0.0 
 (n=83) (n=83) (n=83) 
mf spawning 0.0 100.0 0.0 
 (n=33) (n=33) (n=33) 
m guarding/circling nest 0.0 86.5 13.5 
 (n=133) (n=133) (n=133) 
m crevice inspection 0.0 72.5 27.5 
 (n=167) (n=167) (n=167) 
m crevice pass 0.0 4.3 95.7 
 (n=71) (n=71) (n=71) 
 
Table 1-3 ? Percentage of behaviors associated with sound in the field. 
  
51 
 
Laboratory Knock Burst Pulse Statistic 
     
Duration (s) 0.01576 ? 0.08928 ? 0.007576 ? ANOVA: 
F=26.37 
 0.005279 0.03689 0.001006 P < 0.001 
 (n=11) (n=11) (n=11)  
Fund. Freq. (Hz) 165.07 ? 80.10 ? - t=4.905 
 58.64 3.93 - P < 0.001 
 (n=12) (n=12) -  
Fund. Freq. (dB) 99.32 ? 76.79 ? - t=13.544 
 1.66 5.50 - P < 0.001 
 (n=12) (n=12) -  
First Harmonic (Hz) 587.69 ? 301.58 ? - t=8.806 
 101.43 45.12 - P < 0.001 
 (n=12) (n=12) -  
First Harmonic (dB) 102.73 ? 70.70 ? - t=17.080 
 7.00 5.72 - P < 0.001 
 (n=12) (n=12) -  
     
Field Knock Burst Pulse Statistic 
Duration (s) 0.009314 ? 0.04527 ? 0.005901 ? ANOVA: 
F=62.10 
 0.001720 0.01555 0.000972 P < 0.001 
 (n=8) (n=8) (n=8)  
Fund. Freq (Hz) 297.03 ? 114.15 ? - t=10.870 
 45.77 5.76 - P < 0.001 
 (n=8) (n=8) -  
Fund. Freq. (dB) 89.84 ? 68.40 ? - t=7.360 
 11.81 9.42 - P < 0.001 
 (n=8) (n=8) -  
First Harmonic (Hz) 615.12 ? 391.89 ? - t=14.72 
 40.19 45.79 - P < 0.001 
 (n=8) (n=8) -  
First Harmonic (dB) 95.08 ? 73.03 ? - t=9.468 
 9.04 9.37 - P < 0.001 
 (n=8) (n=8) -  
 
Table 1-4 ? Means and standard deviations of frequency and intensity properties of the 
three call types for approach behaviors produced in the lab and field.  The far right 
column provides statistics (repeated measures ANOVA for duration, and paired-samples 
t tests for fundamental frequency, first harmonic, and levels of each). 
  
52 
 
Laboratory 
Approach behaviors 
Agonistic  
(male-male 
interactions) 
Courtship  
(male-female 
interactions) 
Pulse duration 0.006860 ? 0.000976 0.007801 ? 0.000955 
 (n=8) (n=14) 
Pulse interval 0.012702 ? 0.002303 0.013625 ? 0.001936 
 (n=8) (n=14) 
Pulse rate 89.805 ? 15.913 77.758 ? 10.090 
 (n=8) (n=14) 
Burst duration 0.068195 ? 0.031114 0.090812 ? 0.034505 
 (n=8) (n=14) 
Burst interval 0.181828 ? 0.111263 0.144097 ? 0.073966 
 (n=8) (n=14) 
Burst rate 5.714 ? 2.666 5.161 ? 1.414 
 (n=8) (n=14) 
Burst fund. freq. 85.105 ? 12.553 78.704 ? 5.233 
 (n=8) (n=14) 
Burst fund. freq. dB 72.558 ? 5.517 77.520 ? 4.529 
 (n=8) (n=14) 
Burst harmonic freq. 321.017 ? 55.544 294.237 ? 34.218 
 (n=8) (n=14) 
Burst harmonic freq. 
dB 
68.291 ? 3.380 71.999 ? 5.094 
 (n=8) (n=14) 
Knock duration 0.013799 ? 0.002953 0.017776 ? 0.006129 
 (n=9) (n=6) 
Knock fund. freq. 177.831 ? 68.453 147.117 ? 75.120 
 (n=9) (n=6) 
Knock fund. freq. dB 97.963 ? 5.975 100.365 ? 4.529 
 (n=9) (n=6) 
Knock harmonic freq. 589.381 ? 103.173 599.350 ? 105.841 
 (n=9) (n=6) 
Knock harmonic freq. 
dB 
101.098 ? 6.754 103.635 ? 6.840 
 (n=9) (n=6) 
 
Table 1-5 ? Means and standard deviations of growl and knock signal parameters 
produced during approach behaviors, for agonistic and courtship contexts.  Multivariate 
analysis of co-variance with fish standard length as the covariate indicated no main 
effect of context on signal structure for either growls (MANCOVA: F10,10=0.867, 
P=0.587), or knocks (MANCOVA: F5,8=0.903, P=0.523). 
  
53 
 
Laboratory 
Lateral display behaviors 
Agonistic 
(male-male 
interactions) 
Courtship 
(male-female 
interactions) 
Pulse duration 0.006739 ? 0.001223 0.007496 ? 0.000998 
 (n=7) (n=9) 
Pulse interval 0.012632 ? 0.001618 0.013382 ? 0.000957 
 (n=7) (n=9) 
Pulse rate 85.208 ? 14.987 81.210 ? 9.107 
 (n=7) (n=9) 
Burst duration 0.083522 ? 0.043875 0.081745 ? 0.032045 
 (n=7) (n=9) 
Burst interval 0.131530 ? 0.075161 0.113314 ? 0.048929 
 (n=7) (n=9) 
Burst rate 5.873 ? 2.42 5.908 ? 1.423 
 (n=7) (n=9) 
Burst fund. freq. 88.414 ? 14.133 78.246 ? 4.843 
 (n=7) (n=9) 
Burst fund. freq. dB 72.548 ? 3.936 75.756 ? 4.295 
 (n=7) (n=9) 
Burst harmonic freq. 314.893 ? 56.623 291.056 ? 27.516 
 (n=7) (n=9) 
Burst harmonic freq. dB 70.364 ? 4.192 71.001 ? 4.511 
 (n=7) (n=9) 
Knock duration 0.013616 ? 0.002958 0.01479 ? 0.004101 
 (n=10) (n=7) 
Knock fund. freq. 177.408 ? 55.438 181.678 ? 90.271 
 (n=10) (n=7) 
Knock fund. freq. dB 96.276 ? 4.725 96.342 ? 6.329 
 (n=10) (n=7) 
Knock harmonic freq. 585.369 ? 68.498 526.240 ? 113.825 
 (n=10) (n=7) 
Knock harmonic freq. dB 98.426 ? 5.764 99.032 ? 9.177 
 (n=10) (n=7) 
 
Table 1-6 ? Means and standard deviations of growl and knock signal parameters 
produced during lateral display behaviors, for agonistic and courtship contexts.  
Multivariate analysis of co-variance with fish standard length as the covariate indicated 
no main effect of context on signal structure for either growls (MANCOVA: F10,4=1.743, 
P=0.312), or knocks (MANCOVA: F5,11=0.674, P=0.652). 
  
54 
 
Field 
Approach behaviors 
Agonistic 
(male-male 
interactions) 
Courtship 
(male-female 
interactions) 
Pulse duration 0.006154 ? 0.001032 0.00584 ? 0.000872 
 (n=6) (n=9) 
Pulse interval 0.011071 ? 0.001415 0.01034 ? 0.000851 
 (n=6) (n=9) 
Pulse rate 93.191 ? 9.669 100.269 ? 10.852 
 (n=6) (n=9) 
Burst duration 0.050268 ? 0.020921 0.042519 ? 0.01406 
 (n=6) (n=9) 
Burst interval 0.090107 ? 0.067009 0.073431 ? 0.046326 
 (n=6) (n=9) 
Burst rate 6.134 ? 2.474 10.263 ? 3.254 
 (n=6) (n=9) 
Burst fund. freq. 116.729 ? 20.548 112.779 ? 5.563 
 (n=6) (n=9) 
Burst fund. freq. dB 67.242 ? 6.574 66.393 ? 8.432 
 (n=6) (n=9) 
Burst harmonic freq. 377.267 ? 54.236 388.427 ? 47.966 
 (n=6) (n=9) 
Burst harmonic freq. 
dB 
72.659 ? 6.428 71.054 ? 8.287 
 (n=6) (n=9) 
 
Table 1-7 ? Means and standard deviations of growl signal parameters produced during 
approach behaviors in the field, for agonistic and courtship contexts.  Multivariate 
analysis of variance indicated no main effect of context on signal structure for growls 
(MANOVA: F10,4=0.647, P=0.738). 
  
55 
 
Field 
Chase behaviors 
Agonistic 
(male-male 
interactions) 
Courtship 
(male-female 
interactions) 
Pulse duration 0.005973 ? 0.001017 0.005324 ? 0.000582 
 (n=5) (n=8) 
Pulse interval 0.010261 ? 0.001354 0.010045 ? 0.001069 
 (n=5) (n=8) 
Pulse rate 95.648 ? 13.575 101.966 ? 8.130 
 (n=5) (n=8) 
Burst duration 0.041221 ? 0.010918 0.042722 ? 0.013421 
 (n=5) (n=8) 
Burst interval 0.117603 ? 0.11188 0.072041 ? 0.026946 
 (n=5) (n=8) 
Burst rate 8.329 ? 3.264 9.277 ? 1.923 
 (n=5) (n=8) 
Burst fund. freq. 119.413 ? 17.391 111.049 ? 9.448 
 (n=5) (n=8) 
Burst fund. freq. dB 72.962 ? 6.149 67.827 ? 4.655 
 (n=5) (n=8) 
Burst harmonic freq. 356.259 ? 48.456 401.912 ? 50.545 
 (n=5) (n=8) 
Burst harmonic freq. 
dB 
77.956 ? 5.197 72.837 ? 6.920 
 (n=5) (n=8) 
 
Table 1-8 ? Means and standard deviations of growl signal parameters produced during 
chase behaviors in the field, for agonistic and courtship contexts.  Multivariate analysis 
of variance indicated no main effect of context on signal structure for growls (MANOVA: 
F10,2=0.570, P=0.777). 
  
56 
 
Laboratory Approach Lateral Display Crevice Inspection 
    
Pulse duration 0.007481 ? 
0.000965 
0.007045 ? 
0.000959 
0.007212 ? 
0.000697 
 (n=18) (n=12) (n=7) 
Pulse interval 0.013348 ? 0.00201 0.012857 ? 0.00138 0.01314 ? 0.001881 
 (n=18) (n=12) (n=7) 
Pulse rate 81.295 ? 12.417 84.282 ? 12.442 82.173 ? 14.583 
 (n=18) (n=12) (n=7) 
Burst duration 0.082731 ? 
0.036821 
0.082752 ? 
0.037888 
0.06076 ?  
0.026764 
 (n=18) (n=12) (n=7) 
Burst interval 0.162006 ? 
0.091154 
0.12366 ?  
0.051364 
0.137435 ? 
0.042685 
 (n=18) (n=12) (n=7) 
Burst rate 5.445 ? 2.086 6.080 ? 2.117 5.905 ? 1.638 
 (n=18) (n=12) (n=7) 
Burst fund. freq. 81.275 ? 9.577 82.243 ? 10.126 84.645 ? 9.139 
 (n=18) (n=12) (n=7) 
Burst fund. freq. dB 75.819 ? 5.361 74.724 ? 4.710 73.945 ? 7.674 
 (n=18) (n=12) (n=7) 
Burst harmonic freq. 307.090 ? 44.338 306.506 ? 43.349 313.556 ? 39.984 
 (n=18) (n=12) (n=7) 
Burst harmonic freq. 
dB 
70.887 ? 4.877 70.489 ? 4.395 70.720 ? 6.254 
 (n=18) (n=12) (n=7) 
Knock duration 0.015812 ? 
0.005037 
0.014425 ? 
0.003546 
- 
 (n=12) (n=12) - 
Knock fund. freq. 165.066 ? 58.637 175.336 ? 52.481 - 
 (n=12) (n=12) - 
Knock fund. freq. dB 99.320 ? 5.752 96.951 ? 4.619 - 
 (n=12) (n=12) - 
Knock harmonic freq. 587.685 ? 101.434 571.439 ? 71.423 - 
 (n=12) (n=12) - 
Knock harmonic freq. 
dB 
102.728 ? 7.002 99.613 ? 5.985 - 
 (n=12) (n=12) - 
 
Table 1-9 ? Means and standard deviations of growl and knock signal parameters 
produced during approach, lateral display, and crevice inspection behaviors.  
Multivariate analysis of co-variance with fish standard length as the covariate indicated  
 
57 
 
Table 1-9 (Continued) - no main effect of behavior type on signal structure for either 
growls (MANCOVA: F20,48=0.709, P=0.797), or knocks (MANCOVA: F5,17=0.449, P=0.808). 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
58 
 
Field Circling/ 
guarding 
nest 
Crevice 
inspection 
Approach Chase mf 
circling 
mf hover mf 
spawning 
mf 
zig zag 
         
Pulse 
duration 
0.005725 
? 
0.000413 
0.005975 
? 
0.000678 
0.005893 
? 
0.000933 
0.005672 
? 
0.000799 
0.005653 
? 
0.000725 
0.005936 
? 
0.001033 
0.00522 
? 
0.00053 
0.005612 
? 
0.000599 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Pulse interval 0.010715 
? 
0.000637 
0.01089 
? 
0.000611 
0.010488 
? 
0.001149 
0.01008 
? 
0.000831 
0.009646 
? 
0.001091 
0.009920 
? 
0.001248 
0.00968 
? 
0.001177 
0.009673 
? 
0.001187 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Pulse rate 94.643 
? 
7.07 
108.937 
? 
39.167 
98.889 
? 
11.367 
99.891 
? 
6.095 
103.966 
? 
11.808 
100.941 
? 
13.926 
101.469 
? 
9.891 
101.577 
? 
11.889 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Burst 
duration 
0.050706 
? 
0.018929 
0.054071 
? 
0.015526 
0.044801 
? 
0.014275 
0.043427 
? 
0.010944 
0.059994 
? 
0.032841 
0.071781 
? 
0.043438 
0.046699 
? 
0.017035 
0.060793 
? 
0.016351 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Burst interval 0.091799 
? 
0.020639 
0.071837 
? 
0.0134 
0.090004 
? 
0.055715 
0.072042 
? 
0.03241 
0.068969 
? 
0.049904 
0.052074 
? 
0.049117 
0.057766 
? 
0.022342 
0.062494 
? 
0.026878 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Burst rate 8.151 
? 
2.511 
7.813 
? 
1.798 
8.744 
? 
3.037 
9.143 
? 
2.231 
9.127 
? 
2.834 
9.111 
? 
4.677 
8.134 
? 
2.385 
8.037 
? 
2.114 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Burst fund. 
freq. 
112.102 
? 
11.961 
112.98 
? 
13.833 
113.368 
? 
5.361 
112.282 
? 
10.595 
110.384 
? 
9.582 
102.920 
? 
4.899 
114.415 
? 
11.591 
111.758 
? 
9.347 
59 
 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Burst fund. 
freq. dB 
66.187 
? 
5.598 
66.609 
? 
5.78 
68.055 
? 
8.351 
70.033 
? 
6.258 
71.421 
? 
10.829 
74.125 
? 
12.162 
65.887 
? 
14.279 
71.331 
? 
7.877 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Burst harm. 
freq. 
406.000 
? 
36.004 
396.801 
? 
33.458 
392.827 
? 
40.876 
384.431 
? 
40.747 
376.677 
? 
44.58 
350.546 
? 
46.066 
383.972 
? 
73.319 
391.579 
? 
54.797 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
Burst harm. 
freq. dB 
71.368 
? 
5.493 
72.574 
? 
6.129 
72.474 
? 
8.417 
74.803 
? 
7.454 
75.775 
? 
10.619 
79.456 
? 
11.606 
71.231 
? 
14.651 
76.043 
? 
7.43 
 (n=7) (n=7) (n=10) (n=9) (n=7) (n=4) (n=3) (n=7) 
 
Table 1-10 (continued) ? Means and standard deviations of growl signal parameters produced during 8 different behaviors in 
the field.  Multivariate analysis of variance indicated no main effect of behavior on signal parameters for growls (MANOVA: 
F60,204=0.933, P=0.615). 
60 
 
Field Approach mm circle swim mm lateral display 
    
Knock duration 0.00917 ? 0.001667 0.009052 ? 0.00134 0.010826 ? 
0.005253 
 (n=9) (n=7) (n=9) 
Knock fund. freq. 292.735 ? 44.711 286.421 ? 28.358 270.44 ? 28.707 
 (n=9) (n=7) (n=9) 
Knock fund. freq. dB 90.321 ? 11.14 90.485 ? 8.162 91.704 ? 4.595 
 (n=9) (n=7) (n=9) 
Knock harmonic freq. 628.112 ? 54.181 659.912 ? 61.393 665.256 ? 69.809 
 (n=9) (n=7) (n=9) 
Knock harmonic freq. 
dB 
95.487 ? 8.547 98.552 ? 8.356 98.878 ? 4.595 
 (n=9) (n=7) (n=9) 
 
Table 1-11 ? Means and standard deviations of knock signal parameters produced 
during approach, mm circle swim, and mm lateral display behaviors in the field.  
Multivariate analysis of variance indicated no main effect of behavior type on knock 
signal structure (MANOVA: F5,8=2.407, P=0.129). 
  
61 
 
Approach behaviors Laboratory Field F and P values 
    
Pulse duration 0.007481 ? 0.000965 0.005893 ? 0.000933 F = 17.838, P < 0.001 
 (n=18) (n=10)  
Pulse rate 81.296 ? 12.418 98.889 ? 11.367 F = 13.670, P = 0.001 
 (n=18) (n=10)  
Burst duration 0.082731 ? 0.036821 0.044801 ? 0.014275 F = 9.665, P = 0.005 
 (n=18) (n=10)  
Burst interval 0.162006 ? 0.091154 0.090004 ? 0.055715 F = 5.122, P = 0.032 
 (n=18) (n=10)  
Burst rate 5.446 ? 2.087 8.744 ? 3.037 F = 11.581, P = 0.002 
 (n=18) (n=10)  
Burst fund. freq. 81.276 ? 9.578 113.368 ? 5.361 F = 94.690, P < 0.001 
 (n=18) (n=10)  
Knock duration 0.015811 ? 0.005037 0.009169 ? 0.001666 F = 14.309, P = 0.001 
 (n=12) (n=9)  
Knock fund. freq. 165.067 ? 58.638 292.735 ? 44.711 F = 29.595, P < 0.001 
 (n=12) (n=9)  
Knock harm. freq. 587.684 ? 101.435 628.112 ? 54.180 F = 1.169, P = 0.293 
 (n=12) (n=9)  
 
Table 1-12 ? Means and standard deviations of growl and knock signal parameters 
produced during approach behaviors, under laboratory and field conditions.  
Multivariate analysis of variance indicated a significant main effect of location (lab and 
field) on signal structure for both growls (MANOVA: F6,21=15.072, P < 0.001), and knocks 
(MANOVA: F3,17=12.680, P < 0.001).  Results from the Bonferroni corrected univariate 
ANOVA?s are provided in the right column. 
  
62 
 
Lateral Display 
behaviors 
Laboratory Field F and P values 
    
Knock duration 0.014425 ? 0.003546 0.010825 ? 0.005252 F = 3.527, P = 0.076 
 (n=12) (n=9)  
Knock fund. freq. 175.337 ? 52.482 270.439 ? 28.707 F = 23.957, P < 0.001 
 (n=12) (n=9)  
Knock harm. freq. 571.44 ? 71.42 665.25 ? 69.81 F = 9.043, P = 0.007 
 (n=12) (n=9)  
 
Table 1-13 ? Means and standard deviations of knock signal parameters produced 
during lateral display behaviors in both laboratory and field locations.  Multivariate 
analysis of variance indicated a significant main effect of location (lab and field) on 
signal structure (MANOVA: F3,17=9.897, P = 0.001).  Results from the Bonferroni 
corrected univariate ANOVA?s are provided in the right column. 
  
63 
 
Crevice Inspection 
behaviors 
Laboratory Field F and P values 
    
Pulse duration 0.007212 ? 0.000697 0.005975 ? 0.000678 F = 11.354, P = 0.006 
 (n=7) (n=7)  
Pulse rate 82.174 ? 14.584 108.937 ? 39.167 F = 2.870, P = 0.116 
 (n=7) (n=7)  
Burst duration 0.06076 ? 0.026764 0.054071 ? 0.015526 F = 0.327, P = 0.578 
 (n=7) (n=7)  
Burst interval 0.137435 ? 0.042685 0.071837 ? 0.0134 F = 15.050, P = 0.002 
 (n=7) (n=7)  
Burst rate 5.906 ? 1.639 7.813 ? 1.798 F = 4.301, P = 0.060 
 (n=7) (n=7)  
Burst fund. freq. 84.646 ? 9.14 112.98 ? 13.833 F = 20.446, P = 0.001 
 (n=7) (n=7)  
 
Table 1-14 ? Means and standard deviations of growl signal parameters produced 
during crevice inspections, under laboratory and field conditions.  Multivariate analysis 
of variance indicated that there was a significant main effect of location (lab and field) 
on signal structure (MANOVA: F6,7=4.541, P = 0.034).  Results from the Bonferroni 
corrected univariate ANOVA?s are provided in the right column.  
64 
 
 
 
 
 
Chapter 2 
 
The Effect Of Noise On Acoustic Communication In Cyprinella venusta. 
 Noise can be a problem for any acoustically communicating organism due to the 
masking effect it has on acoustic signals.  Most naturally occurring noise is not 
problematic for organisms because they have had time to evolve mechanisms for 
eluding the adverse effects of the noise.  However, rapid expansion of human 
populations, along with the noise that comes with industrialization and motorized 
transportation poses a threat for many acoustically communicating species.  Aquatic 
organisms are no exception to this, and although much work has been devoted to 
marine organisms, relatively little has been directed at freshwater species, especially 
species inhabiting small stream and river systems.  The aim of the current study was to 
determine the effect of elevated noise levels on acoustic signals, associated behaviors, 
and the interfish distance during sound production in the blacktail shiner, Cyprinella 
venusta.  My results were not conclusive, but suggested it is unlikely that C. venusta is 
able to effectively compensate for elevated noise levels by increasing signal intensity, 
redundancy, or by decreasing interfish distance.   
  
65 
 
Introduction 
The ability to communicate acoustically is a mode of communication that has 
been adopted by a wide variety of fishes (Ladich et al. 2006).  Sound can be a useful tool 
for communication, especially in aquatic habitats where visual signals are often 
diminished due to high levels of turbidity, and complex substrates where line of sight is 
not available (Hawkins and Myrberg 1983).  A wide variety of sound producing 
mechanisms and hearing adaptations have evolved in freshwater fishes (Fine et al. 
1977), and sounds are often associated with important life history events such as 
reproduction (Myrberg and Lugli 2006), food localization (Holt and Johnston, 2011), 
territory defense (Lugli 1997), species recognition (Myrberg et al. 1978), and mate 
choice (Myrberg et al. 1986).  Although experimental evidence is sparse, numerous 
studies have correlated acoustic signal production with behaviors critical for the 
reproductive success of numerous fish species, suggesting that acoustic signals play an 
important communicative role in the reproductive success of many fishes (Ladich et al. 
1992; Hawkins and Amorim 2000; Kasumyan 2009). 
 Underwater communication through the use of sound does have its own set of 
unique challenges for fishes, especially those inhabiting shallow, freshwater systems.  
Shallow aquatic systems are subject to low cut-off frequencies, a phenomenon in which 
sounds below a certain frequency, defined mainly by water depth and substrate 
composition (Officier, 1958; Rogers and Cox 1988), attenuate very rapidly away from the 
source.  Studies looking at fish signal propagation in shallow stream habitats have found 
that signals from both C. venusta (D.E. Holt and C.E. Johnston, unpublished data) and 
66 
 
the gobies Padogobius martensii and Gobius nigricans (Lugli and Fine 2003) propagate 
only several decimeters away from their source, even under the best conditions. 
 Even in such challenging conditions, however, there are certain characteristics of 
the underwater environment that fishes are sometimes able to take advantage of in 
order to increase the efficiency of their acoustic communication.  The few studies that 
have investigated the relationship between environmental noise and acoustic 
communication in fishes inhabiting small, freshwater streams have often found a quiet 
window in the natural environmental noise spectrum that matches the dominant 
frequencies utilized for communication by fishes in the habitat (Lugli et al. 2003; 
Speares et al. 2011; D.E. Holt and C.E. Johnston, unpublished data).  This notch in the 
natural ambient noise allows hearing sensitivity within the window to remain relatively 
high and increases the signal-to-noise ratio of acoustic signals inside the window, both 
of which increase the active area and efficiency of acoustic signals.  The use of quiet 
bands in the natural soundscape is not unique to fishes, and has been described in 
numerous other taxa including birds (Ryan and Brenowitz 1985), frogs (Parris et al. 
2009), and primates (Waser and Waser 1977).  It has been suggested that the 
correlation between the quiet window of the environmental natural noise and spectral 
composition of acoustic signals is an evolutionary strategy set up to increase the 
effectiveness of acoustic communication (Wiley and Richards 1982; Ryan and Brenowitz 
1985; Waser and Brown 1986; Slabbekoorn and Smith 2002). 
Elevations in environmental noise caused by anthropogenic sources can 
exacerbate the already limited active area of acoustic signals in shallow waters.  The 
67 
 
relatively recent development and rapid expansion of mass transportation, 
urbanization, and industrialization has contributed to the rise and spread of 
anthropogenic noise at unprecedented rates (Slabbekoorn et al. 2010).  The effect of 
this noise on aquatic organisms is widely unknown, and what research has been 
conducted has focused mainly on marine organisms, especially marine mammals 
(Richardson et al. 1998; Southall et al. 2007).  One area that has received significant 
attention in fishes is the effect of noise on hearing, and numerous studies on fishes have 
reported significant hearing threshold shifts after exposure to elevated noise (Scholik 
and Yan 2001; Scholik and Yan 2002a; Scholik and Yan 2002b; Amoser and Ladich 2003; 
Smith et al. 2004).  Elevated noise levels can also hinder communication by decreasing 
the signal-to-noise ratio of acoustic signals, masking signals altogether, or altering the 
behavior of an organism.  Sun and Narins (2005) found that anthropogenic noise from 
airplane flyby?s and motorcycle playbacks caused a significant decline in call rate among 
a number of pond dwelling frog species.  Researchers have reported that Norwegian 
spring spawning herring (Vab? et al. 2002) and northern cod (Handegard et al. 2003) 
performed avoidance behaviors in response to passing ships that were not likely within 
visual range.  Sara et al. (2007) found that exposure to multiple types of boat noise 
caused behavioral deviations from normal schooling behavior, and increased aggression 
in the bluefin tuna (Thunnus thynnus). 
Animals that are unable to avoid elevated noise levels may be required to 
compensate for them.  This can be accomplished in several ways according to the 
predictions of information theory (Shannon and Weaver 1949).  First, an animal may 
68 
 
increase the intensity of their signals in order to elevate the signal-to-noise ratio of their 
call relative to the background noise.  This phenomenon is termed the Lombard effect 
(Lombard 1911).  Previous research has shown that in several bird species, the Lombard 
effect occurs when there is an increase in environmental noise, possibly as an attempt 
to compensate for the decrease in active space, which in birds, often defines territorial 
boundaries (Warren et al. 2006).  Brumm and Todt (2002) found that the nightingale 
(Luscinia megarhynchos) increases the amplitude of its acoustic signals when exposed to 
noise in the frequency band of its calls.  The Lombard effect has also been shown to 
occur in common marmosets (Callithrix jacchus; Brumm et al. 2004), frogs (Lopez et al. 
1988), whales (Scheifele et al. 2005; Holt et al. 2008), and humans (Lane and Tranel, 
1971).  However, I am unaware of any report, to date, of the Lombard effect in fishes. 
Another method that can be used to compensate for elevated levels of ambient 
noise is altering components of the frequency of calls to avoid frequency bands that are 
more heavily affected by noise.  Feng et al. (2006) showed that the torrent frog 
(Amolops tormotus) has co-evolved an ability to communicate using ultrasonic harmonic 
frequencies in order to avoid the predominantly low frequency ambient noise produced 
by the streams around which it lives.  Frequency shifts can also be plastic, and can occur 
in response to transient noise sources, or noise sources that have not been around long 
enough to have an evolutionary effect.  Fernandez-Juricic et al. (2005) found that house 
finches (Carpodacus mexicanus) occurring in public parks of southern Los Angeles 
County and north Orange County, California, raised the lower frequencies of their calls, 
possibly to help compensate for the masking effects generated by the higher levels of 
69 
 
low frequency ambient noise.  Slabbekoorn and Peet (2003) also found that great tits 
(Parus major) living near road traffic shifted the frequency of their calls upward away 
from the road noise relative to individuals living in quieter areas. 
Animals can also compensate for the effects of elevated ambient noise by 
increasing the temporal or spectral redundancy of acoustic signals.  Lengagne et al. 
(1999) found that king penguins (Aptenodytes patagonicus) increase the number of 
syllables per call at higher levels of ambient noise.  Potash (1972) found that the 
Japanese quail (Coturnix coturnix japonica) employs a similar tactic when exposed to 
higher noise levels.  Brumm et al. (2004) also found that although it did not increase the 
number of syllables per call, the common marmoset (Callithrix jacchus) did increase the 
duration of its calls when exposed to increased levels of white noise.  By increasing the 
call rate or call duration, the probability of receiving a portion of the call containing 
important information embedded within the acoustic signal is improved by increasing 
the redundancy of information in the signal (Shannon and Weaver, 1949).  
Likelihood of successful acoustic communication may also be increased by 
decreasing the distance between sender and receiver when producing a sound.  This 
may not be a possibility if signals are being used for advertisement or other purpose in 
which the location of the potential receiver may not be known.  However, if the location 
of the receiver is known, it should benefit the sender to close the distance under noisy 
conditions in order to maintain a sufficient signal-to-noise ratio and efficiency of 
communication. 
70 
 
I investigated the effects of elevated noise conditions on the vocal behaviors and 
signal parameters in the blacktail shiner (Cyprinella venusta), a common freshwater fish 
found in the southeastern United States.  Cyprinella venusta is a good species for the 
current study because it occupies a wide variety of freshwater habitats that are subject 
to a number of natural and anthropogenic noise sources.  Cyprinella venusta has also 
been documented as a sound producing fish (Delco 1960; D.E. Holt and C.E. Johnston, 
unpublished data), and is easily collected.  My specific objectives were to determine 
whether C. venusta was able to compensate for depressed signal-to-noise ratios under 
noisy conditions by increasing the level of their signals, increasing signal redundancy in 
the temporal domain, or closing the distance between the sender and receiver.  I also 
looked at whether elevated noise levels resulted in an increase in behavior rate, or an 
increase in the percentage of behaviors associated with sound.  An experimental 
approach was taken by manipulating background noise levels in a laboratory setting, 
recording the fish under both conditions, and then measuring sound levels, temporal 
and spectral parameters, interfish distances, behavior rates, and occurrence of sound 
production with behaviors. 
 
Methods 
Fish were captured by sein from tributaries to the Chatahoochee river in Lee 
County, East Alabama, specifically Little Uchee Creek (32.508579? N, -85.184215? W), 
Halawakee Creek (32.697579? N, -85.266951? W, and Wacoochee Creek (32.622799? N, 
-85.132685? W).  Fish were transported back to the lab and stored in 76 l aquariums 
71 
 
with gravel substrates.  Hanging, external filters were used on all tanks and air stones 
were used only when necessary. Fish were kept at approximately 26?C, given a 12L:12D 
hour light cycle, and fed a diet of commercial flake food. Males that had not been tested 
were kept isolated from other males to prevent dominance hierarchies from being 
established prior to experimental trials, and females were stored alone, with other 
females, or with males that were not to be tested.  A 20 gallon experimental tank was 
setup in an acoustic chamber to conduct each experiment. A tile propped up on a brick 
forming a 5 mm crevice served as an artificial nest, and was placed on the wall in the 
center of the tank. An underwater speaker (University Sound UW-30, Oklahoma City, 
OK) was suspended behind a thin mesh barrier on one end of the tank (Figure 2-1). 
A white piece of plexiglass with gridlines at 5 cm intervals was placed on the 
bottom of the tank, and an identical grid was placed on the back of the tank. A 60 W 
blub was hung approximately 1.5 m above the tank, and a camera was placed in front of 
the tank in such a way that both the bottom and rear grids were visible.  The light bulb 
above the tank cast shadows of fish in the tank on the bottom grid, allowing the position 
of the fish along the X and Z axes of the tank to be determined from the video camera. 
The grid on the back of the tank allowed the position of the fish along the Y axis to be 
determined from the video camera.  
 During trials, male and female C. venusta were placed in the experimental tank 
and allowed to acclimate. Acclimation periods varied by trial, and were considered to be 
over when the fish resumed normal behaviors (swimming freely). Typically, two males 
and at least 1 female were placed in the tank for each trial. Trials consisted of two 
72 
 
periods, one quiet and one noisy. Each period usually lasted about 2 hours, but duration 
ranged from 17 min to 2.5 hrs. During noisy periods, band limited white noise (100-1600 
Hz at 73.3 dB re 1?Pa2/Hz at the hydrophone; Figure 2-2) generated in SigGen(Tucker 
Davis Technologies, Gainesville, FL) was amplified using an ART SLA-1, two-channel 
stereo linear amplifier (100 watts per channel), and played through the UW-30 
underwater speaker. During quiet periods, no sound was played through the speaker 
and the amplifier was turned off. Each period was video recorded on a Sony handycam 
digital HD video recorder (HDR-SR11) and sounds were recorded onto Raven 1.3 (Cornell 
University, Ithaca, NY, U.S.A) using a Br?el and Kjaer 8103 hydrophone and a Br?el and 
Kjaer 2635 charge amplifier. Trials were conducted back to back in varying order. When 
the noisy period began, noise was ramped up to full intensity over the course of 5 
minutes to prevent startling the fish. 
 
Analysis 
 Video of each trial was watched while simultaneously reviewing the audio in 
Raven.  Any time a behavior occurred in the video that was accompanied by a sound 
that was suitable for analysis (not contaminated by the fish hitting something, or mostly 
buried under the ambient noise), it was labeled in Raven.  A total of 78 males were 
included in the trials, but only 20 males produced sufficient sounds of acceptable quality 
to be included in statistical analysis.  A total of 1109 calls were produced that were of 
high enough quality to be analyzed.   
73 
 
 All statistical analyses were conducted in PASW 18 statistical software package 
(IBM SPSS Corporation, Chicago Illinois, USA).  Temporal variables were measured in 
Raven from the waveform view and included duration and interval times for knocks, 
bursts, and pulses.  Burst rate and pulse rate were calculated by dividing the number of 
bursts within a call by that call?s total duration (burst rate) or the number of pulses 
within a burst by that burst duration (pulse rate).  Spectral variables including knock and 
burst fundamental frequency and first harmonic frequency were measured from the 
calculated power density spectrum in Raven.  In order to obtain the most accurate 
frequency information from the power density spectrum, sample sizes of fast fourier 
transformations (FFT?s) were maximized for each knock or burst so that the bin 
resolution of the power density spectrum was as small as possible.  Decibel levels of the 
fundamental and first harmonic frequencies from knocks and bursts were also 
measured from the power density spectrum, and were adjusted to represent actual 
levels by taking into consideration the gain imposed by each unit in the recording chain.  
Decibel levels were also corrected to represent spectrum level (1Hz bands; dB re 
1?Pa2/Hz) by subtracting 10 log (bin resolution) of each measured signal (Richardson et 
al. 1998). 
Prior to all statistical analyses of signals recorded under noisy conditions, an 
average value for each signal parameter (pulse duration, burst duration, burst 
fundamental frequency, etc.) was calculated for each behavior, for each individual male.  
Pearson correlation coefficients were calculated to determine if any relationships 
existed between fish standard length and signal parameters (pulse duration, pulse 
74 
 
interval, pulse rate, burst duration, burst interval, burst rate, knock duration, burst and 
knock fundamental and harmonic frequencies, and burst and knock fundamental and 
harmonic levels).  If a correlation was found, body size was used as a covariate in any 
analyses involving that signal parameter that was not a repeated measures or paired 
design.  Correlations were also performed to determine if there was a relationship 
between level and distance from the hydrophone.  If levels were correlated with 
distance from the hydrophone, levels were corrected by multiplying the distance from 
the hydrophone at the moment of sound production by the slope of the linear best fit 
line of the correlation, and adding this value to the original level of the signal.  This 
eliminated the correlation without affecting variability around the best fit line.  Values 
that had been corrected for distance from the hydrophone were used in further 
analysis.  Because growls did not show a correlation between distance from hydrophone 
and level, this correction was only necessary for knocks. 
To determine whether signal parameters differed between contexts under noisy 
conditions, approach and lateral display behaviors were compared between agonistic 
and courtship contexts.  A finding that acoustic signal parameters did not differ between 
agonistic and courtship contexts would justify the lumping of the contexts to bolster 
samples sizes.  Holt and Johnston (unpublished data) have already shown that no 
difference exists between agonistic and courtship contexts under quiet conditions and 
that under quiet conditions, they may be combined.  Under noisy conditions, approach 
and lateral display behaviors were chosen for comparison between contexts because 
they were the only behaviors that were shared between the two contexts with sample 
75 
 
sizes sufficient for statistical analysis.  Because an insufficient number of knocks were 
recorded during approach behaviors performed under courtship contexts, only growls 
were compared between contexts for approach behaviors.  Both knocks and growls 
were compared between contexts for lateral displays.  Multivariate analysis of variance 
with standard length as a covariate (MANCOVA) was used with context (agonistic and 
courtship) as the fixed factors, and signal parameters as the dependent variables.  A 
total of 3 MANCOVA?s were performed to determine whether sounds differed between 
agonistic and courtship contexts, one for growls produced during approach behaviors, 
one for growls produced during lateral display behaviors, and one for knocks produced 
during lateral display behavior. 
To verify that call components (knocks, bursts, and pulses) were distinct under 
noisy conditions, a repeated measures ANOVA was used to determine whether call 
components differed in duration.  Paired T tests were used to determine whether 
knocks and bursts differed in fundamental frequency, first harmonic frequency, or level. 
The only behaviors that were recorded for the same individual with sufficient 
sample sizes for statistical analysis comparing signal parameters between quiet and 
noisy conditions were approaches and lateral displays.  Because sounds from these 
behaviors were shown to not differ between agonistic and courtship contexts, the 
contexts were combined to bolster samples sizes.  Behaviors that did not have sufficient 
sample sizes were excluded from the analysis.  To determine whether C. venusta males 
altered growl acoustic signals under elevated noise conditions, a doubly multivariate 
repeated measures MANOVA design was used for approach behaviors, and paired t 
76 
 
tests with Bonferroni correction were used for lateral display behaviors.  A repeated 
measures MANOVA design was not able to be used for lateral display behaviors because 
of insufficient residual degrees of freedom resulting from a greater number of measures 
than cases.  In the repeated measures MANOVA used for approach behaviors, the quiet 
and noisy conditions were used as the two within subjects factors, with each factor 
containing 10 measures (pulse duration, pulse interval, pulse rate, burst duration, burst 
interval, burst rate, burst fundamental frequency, burst harmonic frequency, burst 
fundamental frequency level, and burst harmonic frequency level).  If an overall effect 
of noise was detected in the MANOVA, Bonferroni corrected post hoc tests were run to 
determine what specific call parameters differed between the noisy and quiet 
conditions. 
Because knocks and growls were not always present together in all behaviors, 
knocks were analyzed separately from growls.  Knocks produced during noisy and quiet 
trials were compared using a repeated measures MANOVA, similarly to the method 
described above for growls.  For the knock analysis, however, there were only 5 
measures (knock duration, knock fundamental frequency, knock harmonic frequency, 
knock fundamental frequency level, and knock harmonic frequency level). 
 Paired t-tests were used to test for differences in inter-fish distance during 
approach and lateral display behaviors between noisy and quiet conditions.  Growls and 
knocks were tested separately and a Bonferroni correction was applied to correct for 
multiple comparisons. 
77 
 
Behavioral analysis was performed by watching an arbitrarily selected portion of 
each video and recording each behavior that occurred, and whether or not the behavior 
was accompanied by sound. The video segment analyzed was not designated by a 
specific time, but by the length necessary to acquire 100-150 behaviors for each male, 
under each treatment (noisy and quiet).  For each male, under each treatment, the total 
number of agonistic and courtship behaviors was divided by the duration of the portion 
of the video in which they occurred to obtain a behavior rate (behaviors/minute) for 
agonistic behaviors and courtship behaviors.  The number of behaviors that were 
associated with a vocalization was also divided by the total number of behaviors 
performed by a particular individual to obtain a percentage of behaviors that were 
associated with sound (% assoc.) under both treatment conditions, for both agonistic 
and courtship contexts. 
 A doubly multivariate repeated measures two-way MANOVA was used to test 
whether a greater number of agonistic and courtship behaviors were performed under 
noisy conditions than quiet conditions, and whether a greater percentage of agonistic 
and courtship behaviors were accompanied by acoustic signals under noisy conditions 
than quiet conditions.  Two within-subject factors (treatment and context), each with 
two levels (noisy and quiet for treatment; agonistic and courtship for context) were 
included in the model.  Additionally, two measures (behaviors/minute and percentage 
of behaviors associated with sound) were included in the model.   
  
78 
 
Results 
For approach behaviors under quiet conditions, fish standard length showed a 
significant positive correlation with pulse interval (Pearson correlation = 0.564, P = 
0.015, N = 18; Figure 2-3b), and a significant negative correlation with burst harmonic 
frequency (Pearson correlation = -0.665, P = 0.002, N = 19; Figure 2-4b).  Under noisy 
conditions, many more parameters showed significant correlations with fish standard 
length including: pulse duration (Pearson correlation = 0.503, P = 0.033, N = 18; Figure 2-
5a), pulse interval (Pearson correlation = 0.498, P = 0.035, N = 18; Figure 2-3a), burst 
duration (Pearson correlation = 0.497, P = 0.036, N = 18; Figure 2-6a), burst rate 
(Pearson correlation = -0.797, P < 0.001, N = 17; Figure 2-7a), knock duration (Pearson 
correlation = 0.659, P = 0.014, N = 13; Figure 2-8a), burst fundamental frequency 
(Pearson correlation = -0.490, P = 0.039, N = 18; Figure 2-9a), burst fundamental 
frequency dB (Pearson correlation = 0.652, P = 0.003, N = 18; Figure 2-10a), and knock 
fundamental frequency dB (Pearson correlation = 0.785, P = 0.001, N = 13; Figure 2-
11a). 
For lateral display behaviors under quiet conditions, fish standard length showed 
a significant positive correlation with pulse interval (Pearson correlation = 0.638, P = 
0.019, N = 13; Figure 2-3d), and a significant negative correlation with burst harmonic 
frequency (Pearson correlation = -0.660, P = 0.007, N = 15; Figure 2-4d).  This was similar 
to approach behaviors under quiet conditions.  Under noisy conditions, however, other 
signal parameters showed significant correlations with fish standard length during 
lateral display behaviors including: pulse duration (Pearson correlation = 0.548, P = 
79 
 
0.042, N = 14; Figure 2-5c), pulse interval (Pearson correlation = 0.709, P = 0.010, N = 12; 
Figure 2-3c), burst fundamental dB (Pearson correlation = 0.595, P = 0.025, N = 14; 
Figure 2-10c), knock fundamental frequency dB (Pearson correlation = 0.570, P = 0.021, 
N = 16; Figure 2-11c), and knock harmonic frequency (Pearson correlation = -0.604, P = 
0.010, N = 17; Figure 2-12c). 
Distance from the hydrophone was not correlated with the level of either the 
fundamental (Pearson correlation: r = -0.028, P = 0.364) or the first harmonic (Pearson 
Correlation: r = -0.042, P = 0.170) frequency of bursts under noisy conditions (Figure 2-
13).  There was, however, a significant negative correlation between the level of the 
fundamental (Pearson Correlation: r = -0.514, P <0.001) and first harmonic (Pearson 
Correlation: r = -0.552, P < 0.001) frequencies of knocks and the distance from the 
hydrophone (Figure 2-14). 
There was not a significant main effect of context (agonistic and courtship) on 
growl signal parameters for approach behaviors (MANCOVA: F10,8 = 2.059, P = 0.159) 
under noisy conditions (Table 2-1).  There was also no significant main effect of context 
on either growl (MANCOVA: F10,2 = 0.590, P = 0.767) or knock (MANCOVA: F5,14 = 0.796, 
P = 0.570) signal parameters for lateral display behaviors under noisy conditions (Table 
2-2). 
An overall difference was found in duration between the three call types for 
approach behaviors under noisy conditions (repeated measures ANOVA: F2,10 = 40.954, 
P <0.001; Table 2-3).  Bonferroni corrected post hoc tests indicated that under noisy 
conditions, burst duration (mean ? SD; 0.057678 ? 0.022026 sec, N = 12) was 
80 
 
significantly longer than both knock duration (0.012270 ? 0.002639 sec, N = 12; 
Bonferroni corrected post hoc: P < 0.001) and pulse duration (0.007708 ? 0.001124 sec, 
N = 12; Bonferroni corrected post hoc: P < 0.001).  Knocks were also significantly longer 
in duration than pulses (Bonferroni corrected post hoc: P < 0.001). 
Paired t-tests indicated that for approach behaviors, under noisy conditions 
(Table 2-3), the fundamental frequency of knocks (189.00 ? 90.19 Hz, N = 12) was 
significantly higher than the fundamental frequency of bursts (83.29 ? 9.16 Hz, N = 12; 
paired t test: t11 = 3.962, P = 0.002).  The first harmonic frequency of knocks (595.46 ? 
99.37 Hz, N = 12) was also significantly higher than the first harmonic frequency of 
bursts (301.13 ? 30.57; paired t test: t11 = 11.947, P < 0.001).  The level of the 
fundamental frequency of knocks (96.98 ? 4.70 dB, N = 12) was significantly greater than 
that of bursts (79.62 ? 4.35 dB, N = 12; paired t test: t11 = 11.766, P < 0.001), and the 
level of the first harmonic frequency of knocks (102.82 ? 5.08 dB, N = 12) was 
significantly greater than that of bursts (77.52 ? 3.08 dB, N = 12; paired t test: t11 = 
13.936, P < 0.001).  These results also corroborated findings under quiet conditions. 
Results from the repeated measures MANOVA showed a significant main effect 
of background noise level on growl acoustic signal parameters associated with approach 
behaviors (repeated measures MANOVA: F10,7 = 9.211, P = 0.004; Table 2-4).  Bonferroni 
corrected post hoc tests indicated that burst duration under noisy conditions (0.054508 
? 0.020176 sec, N = 17) was significantly shorter than under quiet conditions (0.084586 
? 0.037076 sec, N = 17; Bonferroni corrected post hoc: P = 0.003).  Burst rate under 
noisy conditions (7.347 ? 2.447 sec, N = 17) was significantly greater than under quiet 
81 
 
conditions (5.605 ? 2.034 sec, N = 17; Bonferroni corrected post hoc: P = 0.023).  The 
level of both the fundamental frequency (noisy; 78.61 ? 3.92, quiet; 75.84 ? 5.53) and 
the first harmonic frequency (noisy; 77.19 ? 2.90, quiet; 71.03 ? 4.99) of growls was 
significantly greater under noisy conditions than quiet conditions (fundamental 
frequency dB; Bonferroni corrected post hoc: P = 0.037, first harmonic dB; Bonferroni 
corrected post hoc: P < 0.001).   
No significant main effect of noise level on acoustic signal parameters was found 
for knocks associated with approach behaviors (repeated measures MANOVA: F5,4 = 
0.774, P = 0.615), or lateral display behaviors (repeated measures MANOVA: F5,6 = 
4.244, P = 0.054; Table 2-5). 
Paired t tests with Bonferroni correction showed that during lateral display 
behaviors (Table 2-6), burst duration was significantly shorter under noisy conditions 
(0.058557 ? 0.026548 sec, N = 14) than quiet conditions (0.087573 ? 0.040596 sec, N = 
14; paired t test: t13 = -4.471, P = 0.001).  Paired t tests also showed that the level of 
both the fundamental frequency (noisy; 78.79 ? 5.17 dB, N = 14, quiet; 74.71 ? 4.51 dB, 
N = 14) and the first harmonic frequency (noisy; 77.15 ? 3.81 dB, N = 14, quiet; 70.05 ? 
4.55 dB, N = 14) of growls was significantly greater under noisy conditions than quiet 
conditions (fundamental frequency dB; t13 = 3.919, P = 0.002, first harmonic dB; t13 = 
6.984, P < 0.001).  No other parameters were found to be significantly different. 
There was no significant difference in interfish distance during growls for either 
approach behaviors (noisy; 98.48 ? 52.37 mm, N = 18, quiet; 85.19 ? 51.49 mm, N = 19; 
paired t test: t17 = 0.763, P = 0.456) or lateral display behaviors (noisy; 52.88 ? 24.45 
82 
 
mm, N = 14, quiet; 48.86 ? 16.29 mm, N = 17; paired t test: t13 = 0.562, P = 0.584; Figure 
2-15).  There was also no significant difference in interfish distance during knocks for 
approach behaviors (noisy; 70.21 ? 30.38 mm, N = 13, quiet; 57.47 ? 23.82 mm, N = 12; 
paired t test: t8 = 0.992, P = 0.350; Figure 2-15).  Knocks produced during lateral displays, 
however, were made at a significantly closer distance under noisy conditions (39.31 ? 
13.18 mm, N = 17) than quiet conditions (60.08 ? 12.66 mm, N = 12; paired t test: t10 = -
3.984, P = 0.003; Figure 2-15). 
 Behavioral analysis indicated that there was no significant main effect of context 
(agonistic and courtship; doubly multivariate repeated measures MANOVA: F2,14 = 2.323, 
P = 0.134) or treatment (noisy and quiet; doubly multivariate repeated measures 
MANOVA: F2,14 = 0.190, P = 0.829) on behavior rate and percentage of behaviors 
associated with sounds (Figure 2-16). 
 
Discussion 
The general description of the basic signal components (knocks, bursts, and pulses) of C. 
venusta calls under noisy conditions agreed with the findings from sounds produced 
under quiet laboratory conditions and field conditions (D.E. Holt and C.E. Johnston, 
unpublished data).  Knocks, bursts and pulses were all distinct in duration, and knocks 
and bursts showed significant differences in fundamental and harmonic frequencies and 
level under noisy conditions, as was seen under quiet conditions.  I also found that the 
correlations between distance from the hydrophone and level were similar for sounds 
produced under noisy and quiet conditions. 
83 
 
Correlations between fish standard length and signal parameters, although not 
significant in all cases, typically shared similar trends between noisy and quiet 
conditions for both approach and lateral display behaviors.  Pulse duration, burst 
duration, knock duration, burst fundamental frequency, burst fundamental frequency 
level, burst harmonic frequency, knock fundamental frequency level, and knock 
harmonic frequency all showed a statistically significant correlation with fish standard 
length for at least one of the four behavior/noise level combinations (approach/quiet, 
approach/noisy, lateral display/quiet, and lateral display/noisy).  Combinations that 
were not statistically significant did show trends that agreed with the significant 
correlation, suggesting that their statistical significance may have not been detected 
due to the high variability among individuals.  To explain these correlations, I make the 
assumption that the non-significant correlations of the signal parameters listed above 
are probably real due to the fact that they are going in the same direction for both 
behavior types, under both quiet and noisy conditions, and the simplest explanations 
entail a physiological mechanism that has no reason to differ between quiet and noisy 
conditions. 
A possible explanation for the correlation of standard length with pulse duration, 
based on the findings of Connaughton et al. (2000), is that larger males have longer 
sonic muscles, which take a longer period of time to complete a full twitch.  
Connaughton et al. (2000) described the sounds produced by male weakfish (Cynoscion 
regalis), who?s signal structure is very similar to that of C. venusta, and found, as did I for 
C. venusta, a positive correlation between total body length of male C. regalis and pulse 
84 
 
duration.  The same mechanism may also be responsible for the significant correlation 
between knock duration and fish standard length during approach behaviors under 
noisy conditions. 
The significant correlation between fish standard length and burst fundamental 
frequency level, as well as knock fundamental frequency level is not necessarily 
surprising because larger males probably have larger, stronger muscles capable of 
producing more force, thus producing a louder sound.  Complicating the issue is the fact 
that a significant correlation between standard length and level was observed only for 
the fundamental frequency, and not for the harmonic frequency of either bursts or 
knocks.  If body size permits the ability to increase levels, why should it be limited to 
only the fundamental frequency component of the signal, especially when considering 
that the natural quiet window in the environment corresponds best to the harmonic 
frequency of C. venusta?s signals?  I am unsure why, or how male size affects level of the 
fundamental frequency, and not the harmonic frequency within the same acoustic 
signal.  This question, along with the question of why pulse duration increases with body 
size, undoubtedly requires a more thorough knowledge of the anatomical and 
physiological mechanisms responsible for sound production in this species, which are 
currently unknown. 
The positive correlation between fish standard length and pulse interval (the 
length of time between the first peaks in the waveform of two adjacent pulses) was 
significant for both behaviors, under both quiet and noisy conditions, and was the only 
signal parameter that showed a consistent correlation for both behaviors, under both 
85 
 
quiet and noisy conditions.  The possibility that longer muscle fibers in larger fish require 
more time to contract (and presumably relax) may also be an explanation for the longer 
duration of pulse intervals in larger C. venusta. 
Although it seems intuitive that an increased swim bladder size in larger males 
would carry a lower resonant frequency (Harris 1964), thus causing the negative 
correlations seen between standard length and burst fundamental and harmonic 
frequencies  and knock harmonic frequency, swim bladder size does not necessarily 
correlate with signal frequency.  In many fish that produce long duration, tonal calls, 
sonic muscle contraction rate determines the fundamental frequency of the sound 
(Cohen and Winn 1967; Bass and Baker 1991), not the size of the swim bladder.  Sounds 
produced by many fishes (including C. venusta) have a shorter, more pulsed structure 
that is probably produced by a small number of muscle contractions or a single muscle 
twitch (Connaughton et al. 2000).  Connaughton suggests that the longer pulse duration 
of C. regalis signals, caused by a slower muscle twitch, is responsible for the negative 
correlation between fish size and call frequency, which may also be an explanation for 
the correlation between fish size and burst fundamental and harmonic frequencies, and 
knock harmonic frequency seen in C. venusta.   
There are several correlations that I am unable to explain.  First, if muscle fiber 
length is a determining factor for both fundamental and harmonic frequencies in bursts, 
and the harmonic frequency in knocks, I would expect knock fundamental frequency to 
show a similar correlation with body size.  However, no significant correlations were 
found.  Also, the significant negative correlation between fish standard length and burst 
86 
 
rate existed for approach behaviors under noisy conditions only, and was not 
accompanied by a similar trend under quiet conditions.  I am unable to explain this. 
Prior to this experiment, my null hypothesis was that when subjected to elevated 
noise levels, C. venusta would attempt to compensate for decreased signal-to-noise 
ratios using a combination of methods often utilized by other species.  Increasing signal 
redundancy in the temporal domain (by increasing signal duration or signal emission 
rate) is one method for achieving an increased signal-to-noise ratio with a background 
noise that randomly changes in amplitude.  I was not surprised that the most basic 
temporal components of C. venusta signals (pulse duration, pulse interval, and pulse 
rate) remained conserved.  Mann (1997) found that the basic temporal components 
(pulse period specifically) of the damselfish?s acoustic signals are affected least by 
propagation, and probably serve as a reliable source of species identification.  My null 
hypothesis that signal redundancy would increase was not supported by the fact that 
burst duration decreased for both approach and lateral display behaviors under noisy 
conditions.  Shorter burst duration means a decrease in signal redundancy, which goes 
against the predictions of information theory (Shannon and Weaver 1949), and means 
that there is a smaller chance that information encoded in the burst structure will fall 
into a lull in the background noise.  Accompanying the shortened duration of bursts 
under noisy conditions was an increase in burst rate, which could compensate for the 
shorter duration by maintaining redundancy under noisy conditions.  This tradeoff 
between burst duration and rate was not as consistent as the decrease in burst 
87 
 
duration, however, as it was observed in approach behaviors, but not in lateral display 
behaviors.   
The functional significance of burst duration for C. venusta is currently unknown.  
Phillips and Johnston (2008a) found in the whitetail shiner (C. galactura), evidence 
suggesting a potential role of burst duration as an indicator of male quality due to a 
significant positive correlation between burst duration during agonistic behaviors and 
fish standard length.  A significant correlation was not consistently found for C. venusta, 
but burst duration was significantly positively correlated with fish standard length for 
approach behaviors under noisy conditions.  Phillips and Johnston (2008b) also reported 
a significant difference in burst duration between agonistic and courtship contexts in C. 
galactura, suggesting that C. galactura at least partially utilizes burst duration to convey 
information associated with different behavioral contexts.  The fact that burst duration 
showed a significant positive correlation with C. venusta body size for approach 
behaviors under noisy conditions suggests that it may be an indicator of male quality in 
this species.  However, because burst duration was highly variable and did not differ 
between contexts or behaviors, it is unlikely that it is a critical parameter for recognition 
tasks such as species identification, or to convey different behavioral intentions.  It is 
possible that the decrease in burst duration during noisy trials was an energetic 
response to the increase in burst rate.  However, because the pattern was not found for 
both types of behaviors, and sound production is thought to be a relatively low 
energetic investment for fishes (Amorim et al. 2002), I believe this to be an unlikely 
scenario.  Because burst duration decreased for both approach and lateral display 
88 
 
behaviors under noisy conditions relative to quiet conditions, and burst rate increased 
for only approach behaviors, I conclude that C. venusta probably does not effectively 
increase signal redundancy under noisy conditions. 
The results of this study show an elevation in the level of growls under noisy 
conditions, suggesting that C. venusta may utilize the Lombard effect to help overcome 
elevations in ambient noise levels.  Both the fundamental and harmonic frequencies of 
growls showed increased levels under noisy conditions.  There is also, however, some 
evidence suggesting that the difference in level of signals between noisy and quiet 
conditions may be an artifact of experimental design.   The absence of low intensity 
signals in the noisy trials that were present under quiet conditions (Figure 2-17) may 
either be a result of C. venusta not producing lower amplitude sounds during the noisy 
trials, or may be a result of the signals being buried in the background noise in the noisy 
trials.  If the latter is the case, then the average level under noisy conditions would not 
be accurate.  This theory is supported further by the fact that the level below which no 
acoustic signals were detected under noisy conditions was approximately the same level 
as the background noise.  
 The fact that the behavior rate between noisy and quiet trials did not differ, and 
that the percentage of behaviors associated with sound was greater under quiet 
conditions than under noisy conditions, provides further support for the hypothesis that 
the softer signals produced by C. venusta during noisy trials were not detected by the 
hydrophone, and that the apparent elevation in level is an artifact.  It is also likely that 
knocks did not show an elevation in level under noisy conditions because all knocks 
89 
 
were significantly louder than the background noise during noisy trials, and therefore 
did not have any masked signals under noisy conditions.  The softest knocks detected 
during either quiet or noisy trials were more than 10 dB greater than the background 
noise during noisy trials.  Because it is unlikely that any knocks went undetected, and no 
difference was found between the level of knocks produced in noisy and quiet trials, it is 
a possibility that the difference found for growls is due to a number of softer signals that 
went undetected during noisy trials.  
The results of this study are also mixed as to whether C. venusta attempt to close 
the distance between sender and receiver during acoustic communication under 
elevated noise conditions.  Although the inter-fish distance was significantly smaller 
under noisy conditions than quiet conditions for knocks produced during lateral display 
behaviors, distances were not significantly different for growls produced under both 
approach and lateral display behaviors, or knocks produced under approach behaviors.  
One possible explanation for this is that because lateral displays are a behavior 
associated with higher motivation (Phillips and Johnston 2008b) than approach 
behaviors, the importance of successfully transmitting the message may be greater.  
Because knocks are typically more associated with aggression (Phillips and Johnston 
2008b), and more escalated, physical contests may follow if a dispute is not settled at 
the lateral display stage, it may afford males to adjust inter-fish distance for knocks 
produced during lateral display behaviors under noisier conditions.  Growls are more 
typically associated with courtship behaviors, and since there are usually an abundance 
90 
 
of females around a nesting site, a single missed spawning opportunity may not be as 
costly as a physical contest with an equally sized male. 
The results of the current study are not conclusive enough to definitively say 
whether C. venusta is able to compensate for depressed signal-to-noise ratios under 
elevated noise conditions using either temporal redundancy, increased signal levels, or 
by closing interfish distance.  Although the average level increased under noisy 
conditions, other evidence suggesting that some soft signals may have been masked 
under noisy conditions leads us to believe that C. venusta may not possess the ability to 
increase signal sound pressure level to compensate for elevated noise levels.  Future 
studies investigating this question should keep in mind the potential problems 
encountered by the current study, and make attempts to remove the background noise 
from the recordings while preserving the acoustic signals embedded within the noise.  
This could potentially be done by adding the inverse waveform of the artificial noise 
presented during the noisy trials to the final recordings.  This would cancel the artificial 
noise coming from the speaker, but not the sounds being produced by the fishes, and 
may at least allow the researcher to see if any signals were being masked by the noise.   
The possibility that C. venusta may not be capable of compensating for increased 
noise levels is not entirely surprising.  Cyprinella venusta has evolved in small streams 
where the ambient noise has probably changed very little.  When I consider the 
consistent presence of a quiet window in the natural soundscape, and how well it fits 
with the acoustic signals of C. vensuta, it seems that selective pressures for dealing with 
elevated noise levels were probably not strong, preventing the evolution of techniques 
91 
 
for dealing with the types of interference that artificial noise sources provide today.  
Furthermore, because propagation of sounds in stream conditions is so limited to begin 
with due to the shallow depths and high cutoff frequencies, fish are probably forced to 
call near maximum amplitude to begin with just to maintain the small active area 
permitted by their physical environment.  Holt and Johnston (unpublished data) have 
found evidence suggesting that two sources of anthropogenic noise occurring on small, 
non-navigable streams may affect the signal-to-noise ratio of C. venusta signals at 
distances of up to nearly 900 m.  Although C. venusta is very abundant throughout it?s 
large range and is presently in no danger of becoming threatened, other vocal fish 
species that do not have the abundance or range of C. venusta such as the pygmy 
sculpin (Cottus paulus), or threatened species that occur in larger rivers that may be 
exposed to boat traffic such as the pallid sturgeon (Scaphirhynchus albus), could be 
impacted much more by elevated noise levels. 
With worldwide anthropogenic noise levels increasing (Hildebrand 2009; 
Slabbekoorn et al. 2010), so must our understanding of how fishes will respond.  Despite 
a growing concern for the effects of anthropogenic noise on fishes, our understanding of 
the subject is not unequivocal, and far more questions than answers have been brought 
to light (Popper and Hastings 2009).  The current study provides insight into the 
relationship between background noise and vocalizations in a common vocal fish found 
in small freshwater systems.  Although I was unable to definitively answer the question 
of whether C. venusta is able to compensate for decreased signal-to-noise ratios, the 
92 
 
fact that it did not show obvious signs of compensation may be a red flag for other 
species of concern. 
 
References 
Amorim, M.C.P, McCracken, M.L., and M.L. Fine.  2002.  Metabolic costs of sound  
production in the oyster toadfish, Opsanus tau.  Canadian Journal of Zoology.  
80:830-838. 
Amoser, S. and Ladich, F.  2003.  Diversity in noise-induced temporary hearing loss in  
otophysine fishes.  Journal of the Acoustical Society of America.  113:2170-2179. 
Bass, A. and Baker, R.  1991.  Evolution of homologous vocal control traits.  Brain  
Behavior and Evolution.  38:240-254. 
Brumm, H. and Todt, D.  2002.  Noise-dependent song amplitude regulation in a  
territorial songbird.  Animal Behaviour.  63:891-897. 
Brumm, H., Voss, K., K?llmer, I. and Todt, D.  2004.  Acoustic communication in noise:  
regulation of call characteristics in a new world monkey.  The Journal of 
Experimental Biology.  207:443-448. 
Cohen, M.J. and Winn, H.E.  1967.  Electrophysiological observations on hearing and  
sound production in the fish, Porichthys notatus.  Journal of Experimental 
Zoology.  165:355-370. 
Connaughton, M.A., Taylor, M.H. and Fine, M.L.  2000.  Effects of fish size and  
temperature on weakfish disturbance calls: implications for the mechanism of 
sound generation.  The Journal of Experimental Biology.  203:1503-1512. 
93 
 
Delco, E.A.  1960.  Sound discrimination by males of two Cyprinid fishes.  Texas Journal  
of Science.  12:48-54. 
Feng, A.S., Narins, P.M., Xu, C.-H., Lin, W.-Y., Yu, Z.-L., Qiu, Q., Xu, Z.-M. and Shen, J.-X.   
2006.  Ultrasonic communication in frogs.  Nature.  440:333-336. 
Fernandez-Juricic, E.R., Poston, R., DeCollibus, K., Morgan, T., Bastain, B., Martin, C.,  
Jones, K., and Treminio, R.  2005.  Microhabitat selection and singing behavior 
patterns of male house finches (Carpodacus mexicanus) in urban parks in a 
heavily urbanized landscape in the western U.S.  Urban Habitats.  3:49-69. 
Fine, M.L., Winn, H.E., and Olla, B.L.  1977.  Communication in fishes.  In: Sebeok (ed),  
How animals communicate.  Indiana University Press, Bloomington, pp. 472-518. 
Handegard, N.O., Michalsen, K. and Tj?stheim, D.  2003.  Avoidance behavior in cod,  
Gadus morhua, to a bottom trawling vessel.  Aquatic Living Resources.  16:265-
270. 
Harris, G.G.  1964.  Considerations on the physics of sound production by fishes.  In:  
Marine Bio-Acoustics (ed. Tavolga, W.N.).  pp. 233-247.  New York, Pergamon 
Press. 
Hawkins, A.D., and Amorim, M.C.P.  2000.  Spawning sounds of the male haddock,  
Melanogrammus aeglefinus.  Environmental Biology of Fishes.  59:29-41. 
Hildebrand, H.E.  2009.  Anthropogenic and natural sources of ambient noise in the  
ocean.  Marine Ecology Progress Series.  395:5-20. 
Holt, D.E. and Johnston, C.E.  2011.  Can you hear the dinner bell?  Response of cyprinid  
fishes to environmental acoustic cues.  Animal Behaviour.  82:529-534. 
94 
 
Holt, M.M., Noren, D.P., Veirs, V., Emmons, C.K. and Veirs, S.  2008.  Speaking up: Killer  
whales (Orcinus orca) increase their call amplitude in response to vessel noise.  
Journal of the Acoustical Society of America.  125:EL27-EL32. 
Kasumyan, A.O.  2009.  Acoustic signaling in fish.  Journal of Ichthyology.  49:963-1020. 
Ladich, F., Brittenger, W., and Kratochvil, H.  1992.  Significance of agonistic vocalization  
in the croaking gourami (Trichopsis vittatus, Teleostei).  Ethology.  90:307-314. 
Ladich, F., Collin, S.P., Moller, P., and Kapoor, B.G.  2006.  Communication in Fishes.   
Science Publishers. Enfield, NH, USA. 
Lombard, E.  1911.  Le signe de le elevation de la voix.  Ann. Maladies l?Oreille, Larynx,  
Nez, Pharynx.  37:101-119. 
Lane, H. and Tranel, B.  1971.  The Lombard sign and the role of hearing in speech.   
Journal of Speech Hearing Research.  14:677-709. 
Lengagne, T., Aubin, T., Lauga, J. and Jouventin, P.  1999.  How do king penguins  
(Aptenodytes patagonicus) apply the mathematical theory of information to 
communicate in windy conditions.  Proceedings the Royal Society B.  266:1623-
1628. 
Lopez, P.T., Narins, P.M., Lewis, E.R. and Moore, S.W.  1988.  Acoustically induced call  
modification in the white-lipped frog, Leptodactylus albilabris.  Animal 
Behaviour.  36:1295-1308. 
Lugli, M.P.  1997.  Response of male goby, Padogobius martensii, to aggressive sound  
playback following pre-experimental visual stimulation.  Behaviour.  134:1175-
1188. 
95 
 
Lugli, M.P. and Fine, M.L.  2003.  Acoustic communication in two freshwater gobies:  
Ambient noise and short-range propagation in shallow streams.  Journal of the 
Acoustical Society of America.  114:512-521. 
Lugli, M.P., Yan, H.Y. and Fine, M.L.  2003.  Acoustic communication in two freshwater  
gobies: the relationship between ambient noise, hearing thresholds, and power 
spectrum.  Journal of Comparative Physiology A.  189:309-320. 
Mann, D.A. and Lobel, P.S.  1997.  Propagation of damselfish (Pomacentridae) courtship  
sounds.  Journal of the Acoustical Society of America.  101:3783-3791. 
Myrberg, A.A. Jr. and Lugli, M.  2006.  Reproductive behavior and acoustical interactions.   
In: Communication in Fishes.  Ladich, F., Collin, S.P., and Moller, P. (eds).  Science 
Publishers, Enfield, N.H., pp. 149-176. 
Myrberg, A.A. Jr., Mohler, M., and Catala, J.D.  1986.  Sound production by males of a  
coral reef fish (Pomacentrus partitus): its significance to females.  Animal 
Behaviour.  34:913-923. 
Myrberg, A.A. Jr., Spanier, E., and Ha, S.J.  1978.  Temporal patterning in acoustical  
communication.  In: Contrasts in behavior.  Reese, E.S. and Lighter, F.J.  (eds).  
Wiley, New York, pp. 137-179. 
Officier, C.B.  1958.  Introduction to the theory of sound transmission.  McGraw-Hill,  
New York. 
Parris, K.M., Velik-Lord, M. and North, J.M.A.  2009.  Frogs call at a higher pitch in traffic  
noise.  Ecology and Society.  14:25. 
Popper, A.N. and Hastings, M.C.  2009.  The effects of anthropogenic sources of sound  
96 
 
on fishes.  Journal of Fish Biology.  75:455-489. 
Phillips, C.T. and Johnston, C.E.  2008a.  Geographical divergence of acoustic signals in  
Cyprinella galactura, the whitetail shiner (Cyprinidae).  Animal Behaviour.  
75:617-626. 
Phillips, C.T. and Johnston, C.E.  2008b.  Sound production and associated behaviors in  
Cyprinella galactura.  Environmental Biology of Fishes.  82:265-275. 
Potash, L.M.  1972.  Noise-induced changes in calls of the Japanese quail.  Psychonomic  
Science.  26:252-254. 
Richardson, W.J., Greene, R.C. Jr., Malme, C.L., and Thomson, D.H.  1998.  Marine  
Mammals and Noise, Academic Press. 
Rogers, P.H. and Cox, M.  1988.  Underwater sound as a biological stimulus.  In: Sensory  
Biology of Aquatic Animals.  (ed) Atema, J., Fay, R.R., Popper, A.N., and Tavolga, 
W.N.  Springer, New York, pp. 131-149. 
Ryan, M.J. & Brenowitz, E.A.  1985.  The role of body size, phylogeny, and ambient noise  
in the evolution of bird song.  American Naturalist.  126:87?100. 
Sara, G., Dean, J.M., D?Amato, D., Buscaino, G., Oliveri, A., Genovese, S., Ferro, S., Buffa,  
G., Lo Martire, M. and Mazzola, S.  2007.  Effect of boat noise on the behavior of 
bluefin tuna Thunnus thynnus in the Mediterranean Sea.  Marine Ecology 
Progress Series.  331:243-253. 
Scheifele, P.M., Andrew, S., Cooper, R.A., Darre, M., Musiek, F.E. and Max, L.  2005.   
Indication of a Lombard vocal response in the St. Lawrence river beluga.  Journal 
of the Acoustical Society of America.  117:1486-1492. 
97 
 
Scholik, A.R. and Yan, H.Y.  2001.  Effects of underwater noise on auditory sensitivity of a  
cyprinid fish.  Hearing Research.  152:17-24. 
Scholik, A.R. and Yan, H.Y.  2002a.  Effects of boat engine noise on the auditory  
sensitivity of the fathead minnow, Pimephales promelas.  Environmental Biology 
of Fishes.  63:203-209. 
Scholik, A.R. and Yan, H.Y.  2002b.  The effects of noise on the auditory sensitivity of the  
bluegill sunfish, Lepomis macrochirus.  Comparative Biochemistry and Physiology 
A ? Molecular and Integrative Physiology.  133:43-52. 
Shannon, C.E., and Weaver, W.  1949.  The Mathematical Theory of Communication.   
Urbana: Illinois University Press. 
Slabbekoorn, H., Bouton, N., Opzeeland, L.V., Coers, A., Cate, C.T. and Popper, A.N.   
2010.  A noisy spring: the impact of globally rising underwater sound levels on 
fish.  Trends in Ecology and Evolution.  25:419-427. 
Slabbekoorn, H. and Peet, M.  2003.  Birds sing at a higher pitch in urban noise ? great  
 tits hit the high notes to ensure that their mating calls are heard above the 
city?s din.  Nature.  424:267-267. 
Slabbekoorn, H. and Smith, T.B.  2002.  Habitat-dependent song divergence in the little  
 greenbul: an analysis of environmental selection pressures on acoustic signals. 
Evolution, 56, 1849?1858. 
Smith, M.E., Kane, A.S. and Popper, A.N.  2004.  Noise-induced stress response and  
hearing loss in goldfish (Carassius auratus).  The Journal of Experimental Biology.  
207:427-435. 
98 
 
Southall, B.L., Bowles, A.E., Ellison, W.T., Finneran, J.J., Gentry, R.L., Greene Jr., C.R.,  
Kastak, D., Ketten, D.R., Miller, J.H., Nachtigall, P.E., Richardson, W.J., Thomas, 
J.A. and Tyack, P.L.  2007.  Marine mammal noise exposure criteria: initial 
scientific recommendations.  Aquatic Mammals.  33:411?521. 
Speares, P., Holt, D.E. and Johnston, C.E.  2011.  The relationship between ambient noise  
and dominant frequency of vocalizations in two species of darters (Percidae: 
Etheostoma).  Environmental Biology of Fishes.  90:103-110. 
Sun, J.W.C. and Narins, P.M.  2005.  Anthropogenic sounds differentially affect  
amphibian call rate.  Biological Conservation.  121:419-427. 
Vab?, R., Olsen, K. and Huse, I.  2002.  The effect of vessel avoidance of wintering  
Norwegian spring-spawning herring.  Fisheries Research.  58:59-77. 
Warren, P.S., Katti, M., Ermann, M. and Brazel, A.  2006.  Urban bioacoustics: it?s not just  
noise.  Animal Behaviour.  71:491-502. 
Waser, P.M. and Waser, M.S.  1977.  Experimental studies of primate vocalization:  
specializations for long distance propagation.  Zeitschrift fur Tierpsychologie.  
43:239-263. 
Waser, P.M. & Brown, C.H.  1986.  Habitat acoustics and primate communication.  
American Journal of Primatology, 10, 135?154. 
Wiley, R.H. & Richards, D.G.  1982.  Adaptations for acoustic communication in birds:  
sound transmission and signal detection. Acoustic Communication in Birds, Vol. 1 
(eds D.E. Kroodsma & E.H. Miller), pp. 131?181. Academic Press, New York. 
  
99 
 
 
Figure 2-1 ? Experimental tank setup.  Underwater speaker is suspended behind a mesh 
barrier on the right hand side of the tank (according to the viewer?s perspective).  A 
hydrophone is suspended in front of an artificial nest (composed of a tile propped up on 
top of a brick), which is placed against the far wall of the tank.  A light suspended above 
the tank cast shadow?s onto the bottom grid, which together with the grid on the rear 
side of the tank allow the position of the fishes to be determined inside the tank. 
  
100 
 
 
Figure 2-2 ? Power spectrums showing background noise levels during quiet trials 
(broken line) and noisy trials (solid line).  Power spectra were generated from a 1 second 
recording of background noise, in the absence of fish (Hanning window, sampling rate = 
44.1 kHz, FFT samples = 16384, bin resolution = 2.69 Hz).  Hydrophone placement was 
approximately 5 cm in front of the nest for both recordings. 
  
101 
 
 
Figure 2-3 ? Scatterplots showing the relationships between fish standard length (mm) 
and pulse interval (sec) for approach behaviors under noisy (a) and quiet (b) conditions, 
and lateral display behaviors under noisy (c) and quiet (d) conditions.  All correlations 
were statistically significant and positive. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
102 
 
 
Figure 2-4 ? Scatterplots showing the relationships between fish standard length and 
burst harmonic frequency for approach behaviors under noisy (a) and quiet (b) 
conditions, and lateral display behaviors under noisy (c) and quiet (d) conditions.  
Although a negative trend was observed for all conditions, only for approaches under 
quiet conditions (b) was the correlation statistically significant. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
103 
 
 
Figure 2-5 - Scatterplots showing the relationships between fish standard length and 
pulse duration for approach behaviors under noisy (a) and quiet (b) conditions, and 
lateral display behaviors under noisy (c) and quiet (d) conditions.  Although a statistically 
significant correlation was only observed for approach and lateral display behaviors 
under noisy conditions, quiet conditions did show positive trends. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
104 
 
 
Figure 2-6 ? Scatterplots showing the relationships between fish standard length and 
burst duration for approach behaviors under noisy (a) and quiet (b) conditions, and 
lateral display behaviors under noisy (c) and quiet (d) conditions.  The only condition in 
which a statistically significant correlation was observed was during approach behaviors 
under noisy conditions (a). 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
105 
 
 
Figure 2-7 ? Scatterplots showing the relationships between fish standard length and 
burst rate for approach behaviors under noisy (a) and quiet (b) conditions, and lateral 
display behaviors under noisy (c) and quiet (d) conditions. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
106 
 
 
Figure 2-8 ? Scatterplots showing the relationships between fish standard length and 
knock duration for approach behaviors under noisy (a) and quiet (b) conditions, and 
lateral display behaviors under noisy (c) and quiet (d) conditions. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
107 
 
 
Figure 2-9 ? Scatterplots showing the relationships between fish standard length and 
burst fundamental frequency for approach behaviors under noisy (a) and quiet (b) 
conditions, and lateral display behaviors under noisy (c) and quiet (d) conditions. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
108 
 
 
Figure 2-10 ? Scatterplots showing the relationships between fish standard length and 
burst fundamental frequency level for approach behaviors under noisy (a) and quiet (b) 
conditions, and lateral display behaviors under noisy (c) and quiet (d) conditions. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
109 
 
 
Figure 2-11 ? Scatterplots showing the relationships between fish standard length and 
knock fundamental frequency level for approach behaviors under noisy (a) and quiet (b) 
conditions, and lateral display behaviors under noisy (c) and quiet (d) conditions. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
110 
 
 
Figure 2-12 ? Scatterplots showing the relationships between fish standard length and 
knock harmonic frequency for approach behaviors under noisy (a) and quiet (b) 
conditions, and lateral display behaviors under noisy (c) and quiet (d) conditions. 
  
Standard Length (mm) Standard Length (mm) 
Standard Length (mm) Standard Length (mm) 
111 
 
 
Figure 2-13 ? Scatterplot of burst fundamental frequency level (top) and first harmonic 
level (bottom) with distance of sound producing fish from the hydrophone under noisy 
conditions.  Neither showed a significant correlation (fundamental frequency; Pearson 
Correlation: r = -0.028, P = 0.364; first harmonic; Pearson Correlation: r = -0.042, P = 
0.170). 
  
112 
 
 
Figure 2-14 ? Scatterplot with best fit line (equation included in graph) of fundamental 
frequency and first harmonic levels of knocks and distance of fish from hydrophone 
under noisy conditions.  There was a significant negative correlation between the level 
of the fundamental frequency of knocks and distance from the hydrophone (top-right; 
Pearson Correlation: r = -0.514, P <0.001), and the level of the first harmonic frequency 
of knocks and distance from the hydrophone (top-left; Pearson Correlation: r = -0.552, P 
< 0.001).  Values were corrected by multiplying the distance from the hydrophone at 
each point by the slope of the best fit line, and adding the product to the original dB 
value.  Corrected plots for both the first harmonic frequency level (bottom-left) and 
fundamental frequency level (bottom-right) are provided. 
  
113 
 
 
Figure 2-15 ? Error bar plots showing interfish distances for growls and knocks, 
produced during approach and lateral display behaviors, under noisy and quiet 
conditions.  The circles (noisy) and triangles (quiet) represent means, and the whiskers 
represent the 95% confidence interval around the mean.  Knocks produced during 
lateral display behaviors represented the only situation in which interfish distance was 
significantly different between noisy and quiet conditions (paired t test: t10 = -3.984, P = 
0.003). 
  
114 
 
 
Figure 2-16 ? Error bar plot showing the percentage of courtship and agonistic behaviors 
that were associated with sound production, under noisy and quiet contidions (top), and 
behavior rates (number of behaviors/minute) for both agonistic and courtship contexts, 
under noisy and quiet conditions (bottom).  The ?n? and ?q? represent the noisy and 
quiet trials, respectively.  The circles represent means, and the whiskers represent the 
95% confidence interval around the mean.  Doubly multivariate repeated measures 
MANOVA indicated that there was no significant main effect of context (agonistic and 
courtship; F2,14 = 2.323, P = 0.134) or treatment (noisy and quiet; F2,14 = 0.190, P = 0.829) 
on behavior rate and percentage of behaviors associated with sounds. 
  
115 
 
 
Figure 2-17 ? Errorplot showing level of growl fundamental and harmonic frequencies, 
for approach and lateral display behaviors, under both noisy and quiet conditions.  
Empty triangles represent the average level under noisy conditions, and filled circles 
represent the average level under quiet conditions.  Whiskers above and below each 
mean represent the 95% confidence interval around that particular mean.  The dark 
grey bar running across the graph between 70 and 75 dB represents the level of the 
background noise during noisy trials.  Notice that no confidence intervals for the noisy 
trials fall below the background noise level during noisy trials. 
  
116 
 
Noisy Conditions Agonistic Courtship 
Approach behaviors   
   
Pulse duration 0.00698 ? 0.001425 0.007704 ? 0.000971 
 (n=7) (n=13) 
Pulse interval 0.012843 ? 0.002255 0.013015 ? 0.000908 
 (n=7) (n=13) 
Pulse rate 89.828 ? 17.452 81.923 ? 7.485 
 (n=7) (n=13) 
Burst duration 0.041803 ? 0.015677 0.058638 ? 0.017182 
 (n=7) (n=13) 
Burst interval 0.174018 ? 0.080841 0.121138 ? 0.055354 
 (n=7) (n=13) 
Burst rate 7.84 ? 3.469 7.401 ? 2.384 
 (n=7) (n=13) 
Burst fund. freq. 88.728 ? 16.34 81.493 ? 4.975 
 (n=7) (n=13) 
Burst fund. freq. dB 76.792 ? 3.746 79.773 ? 3.293 
 (n=7) (n=13) 
Burst harmonic freq. 326.295 ? 42.44 294.995 ? 26.039 
 (n=7) (n=13) 
Burst harmonic freq. 
dB 
76.123 ? 1.552 77.861 ? 2.891 
 (n=7) (n=13) 
 
Table 2-1 ? Means and standard deviations of growl parameters produced during 
approach behaviors under noisy conditions, for agonistic and courtship contexts.  
Multivariate analysis of co-variance with fish standard length as the covariate indicated 
no main effect of context on signal structure (MANCOVA: F10,8=2.059, P=0.159). 
  
117 
 
Noisy Conditions 
Lateral Display behaviors 
Agonistic Courtship 
   
Pulse duration 0.006557 ? 0.001159 0.006894 ? 0.000908 
 (n=7) (n=7) 
Pulse interval 0.0121 ? 0.002017 0.013412 ? 0.000853 
 (n=7) (n=7) 
Pulse rate 123.087 ? 63.547 81.094 ? 5.905 
 (n=7) (n=7) 
Burst duration 0.038887 ? 0.010311 0.065696 ? 0.019027 
 (n=7) (n=7) 
Burst interval 0.141973 ? 0.096844 0.095912 ? 0.039052 
 (n=7) (n=7) 
Burst rate 7.86 ? 2.813 6.338 ? 1.151 
 (n=7) (n=7) 
Burst fund. freq. 91.887 ? 16.822 78.265 ? 4.151 
 (n=7) (n=7) 
Burst fund. freq. dB 76.211 ? 4.263 80.473 ? 3.453 
 (n=7) (n=7) 
Burst harmonic freq. 314.756 ? 51.923 285.753 ? 19.137 
 (n=7) (n=7) 
Burst harmonic freq. dB 75.943 ? 2.857 79.289 ? 3.141 
 (n=7) (n=7) 
Knock duration 0.013752 ? 0.003668 0.013192 ? 0.004249 
 (n=12) (n=9) 
Knock fund. freq. 195.192 ? 77.194 209.932 ? 102.426 
 (n=12) (n=9) 
Knock fund. freq. dB 97.504 ? 3.212 98.908 ? 5.822 
 (n=12) (n=9) 
Knock harmonic freq. 596.811 ? 86.703 564.844 ? 89.416 
 (n=12) (n=9) 
Knock harmonic freq. dB 102.463 ? 2.722 105.048 ? 6.784 
 (n=12) (n=9) 
 
Table 2-2 ? Means and standard deviations of growl and knock signal parameters 
produced during lateral display behaviors under noisy conditions, for agonistic and 
courtship contexts.  Multivariate analysis of co-variance with fish standard length as the 
covariate indicated no main effect of context on signal structure for either growls 
(MANCOVA: F10,2=0.590, P=0.767), or knocks (MANCOVA: F5,14=0.796, P=0.570). 
  
118 
 
Approach 
Behavior 
Knock Burst Pulse  
     
Duration (s) 0.012270 ? 0.057678 ? 0.007708 ? ANOVA: 
F2,10=40.954 
 0.002639 0.022026 0.001124 P < 0.001 
 (n=12) (n=12) (n=12)  
Fund. Freq. (Hz) 188.999 ? 83.288 ? - t11=3.962 
 90.191 9.162 - P = 0.002 
 (n=12) (n=12) -  
Fund. Freq. (dB) 96.978 ? 79.623 ? - t11= 11.766 
 4.702 4.349 - P < 0.001 
 (n=12) (n=12) -  
First Harmonic 
(Hz) 
595.456 ? 301.134 ? - t11= 11.947 
 99.368 30.566 - P < 0.001 
 (n=12) (n=12) -  
First Harmonic 
(dB) 
102.820 ? 77.516 ? - t11= 13.936 
 5.076 3.080 - P < 0.001 
 (n=12) (n=12) -  
 
Table 2-3 ? Means and standard deviations of frequency and dB properties of the three 
call types for approach behaviors produced under noisy conditions.  The far right 
column provides statistics (repeated measures ANOVA for duration, and paired-samples 
t tests for fundamental frequency, first harmonic, and levels of each). 
  
119 
 
Approach Behaviors Noisy Quiet  
    
Pulse duration 0.007419 ?  
0.001206 
0.007557 ? 
 0.000938 
F = 0.261,  
P = 0.616 
 (n=17) (n=17)  
Pulse interval 0.012911 ? 
 0.001422 
0.013253 ? 
 0.002029 
F = 0.834, 
P = 0.375 
 (n=17) (n=17)  
Pulse rate 84.832 ?  
12.509 
81.464 ?  
12.779 
F = 2.591,  
P = 0.127 
 (n=17) (n=17)  
Burst duration 0.054508 ?  
0.020177 
0.084587 ?  
0.037076 
F = 11.807,  
P = 0.003 
 (n=17) (n=17)  
Burst interval 0.133218 ?  
0.064244 
0.146928 ?  
0.066934 
F = 0.360,  
P = 0.557 
 (n=17) (n=17)  
Burst rate 7.348 ?  
2.448 
5.606 ?  
2.034 
F = 6.277,  
P = 0.023 
 (n=17) (n=17)  
Burst fund. freq. 84.676 ?  
11.044 
81.782 ?  
9.622 
F = 2.158,  
P = 0.161 
 (n=17) (n=17)  
Burst fund. freq. dB 78.606 ?  
3.925 
75.843 ?  
5.526 
F = 5.181,  
P = 0.037 
 (n=17) (n=17)  
Burst harmonic freq. 307.026 ?  
34.105 
305.522 ?  
45.185 
F = 0.056,  
P = 0.816 
 (n=17) (n=17)  
Burst harmonic freq. 
dB 
77.186 ?  
2.898 
71.035 ?  
4.987 
F = 29.617,  
P < 0.001 
 (n=17) (n=17)  
 
Table 2-4 ? Means and standard deviations of growl signal parameters produced during 
approach behaviors under noisy and quiet conditions.  Multivariate analysis of variance 
indicated a significant main effect of background noise on signal structure for growls 
(MANOVA: F10,7=9.211, P=0.004).  The column on the far right provides P values from 
Bonferroni corrected post hoc tests for the corresponding signal parameters in the left 
column.  Significant results are in bold. 
  
120 
 
Approach behaviors Noisy Quiet 
   
Knock duration 0.012445 ? 0.002726 0.013815 ? 0.002918 
 (n=9) (n=9) 
Knock fund. freq. 151.515 ? 58.925 169.856 ? 59.844 
 (n=9) (n=9) 
Knock fund. freq. dB 97.305 ? 4.190 97.812 ? 5.933 
 (n=9) (n=9) 
Knock harmonic freq. 628.851 ? 79.729 589.781 ? 91.924 
 (n=9) (n=9) 
Knock harmonic freq. 
dB 
103.119 ? 3.462 100.569 ? 6.560 
 (n=9) (n=9) 
   
Lateral Display 
Behaviors 
  
Knock duration 0.014161 ? 0.004978 0.014266 ? 0.003674 
 (n=11) (n=11) 
Knock fund. freq. 170.913 ? 73.232 174.966 ? 55.027 
 (n=11) (n=11) 
Knock fund. freq. dB 98.933 ? 3.641 96.524 ? 4.590 
 (n=11) (n=11) 
Knock harmonic freq. 594.726 ? 75.211 578.366 ? 70.556 
 (n=11) (n=11) 
Knock harmonic freq. 
dB 
103.815 ? 2.681 98.814 ? 5.566 
 (n=11) (n=11) 
 
Table 2-5 ? Means and standard deviations of knock signal parameters produced during 
approach and lateral display behaviors under noisy and quiet conditions.  Multivariate 
analysis of variance indicated no significant main effect of background noise on signal 
structure for knocks produced during approach behaviors (MANOVA: F5,4=0.774, 
P=0.615), or lateral display behaviors (MANOVA: F5,6=4.244, P=0.054). 
  
121 
 
Lateral Display 
Behaviors 
Noisy Quiet  
    
Pulse duration 0.006832 ?  
0.000947 
0.006989 ?  
0.00094 
t12 = -0.611,  
P = 0.552 
    
Pulse interval 0.012679 ?  
0.001544 
0.014019 ?  
0.004057 
t10 = -1.191,  
P = 0.261 
    
Pulse rate 100.063 ?  
52.561 
103.067 ? 
68.768 
t12 = -0.443,  
P = 0.665 
    
Burst duration 0.058557 ?  
0.026548 
0.087573 ?  
0.040597 
t13 = -4.471,  
P = 0.001 
    
Burst interval 0.128898 ?  
0.076923 
0.122995 ?  
0.049235 
t12 = 0.216,  
P = 0.833 
    
Burst rate 6.971 ?  
1.93 
6.791 ?  
3.264 
t12 = 0.158,  
P = 0.877 
    
Burst fund. freq. 85.289 ?  
12.747 
81.787 ?  
9.498 
t13 = 1.290,  
P = 0.220 
    
Burst fund. freq. dB 78.786 ?  
5.17 
74.706 ?  
4.506 
t13 = 3.919,  
P = 0.002 
    
Burst harmonic freq. 306.961 ?  
38.232 
308.126 ?  
42.235 
t13 = -0.168,  
P = 0.869 
    
Burst harmonic freq. 
dB 
77.146 ?  
3.807 
70.051 ?  
4.547 
t13 = 6.984,  
P < 0.001 
 
Table 2-6 ? Means and standard deviations of growl signal parameters produced during 
lateral display behaviors under noisy and quiet conditions.  Because of a lack of sample 
size, repeated measures MANOVA was unable to be performed, and paired t tests with 
Bonferroni correction were used instead.  The column on the far right provides T and P 
values for the corresponding signal parameters in the left column.  Significant results are 
in bold  
122 
 
 
 
 
 
Chapter 3 
Hearing Sensitivity, Environmental Soundscape, And Propagation Of Acoustic Signals 
In Cyprinella venusta. 
In order for an acoustic signal to be an effective source of communication, it 
must be not only be emitted properly by a sender, but must also be successfully 
detected by the intended receiver.  One potential barrier to acoustic communication is 
background noise.  The habitat of Cyprinella venusta varies greatly across it?s large 
range, and in some areas containing riffles, shoals and waterfalls, can become quite 
noisy.  The presence of C. venusta in larger, navigable bodies of water and streams with 
road and train crossings also presents potential anthropogenic noise sources with which 
C. venusta did not evolve.  In the present study, I investigated the relationship between 
the vocalizations, hearing sensitivity, and natural soundscape of C. venusta.  I also 
investigated, for the first time, the potential effects of anthropogenic noise from bridge 
crossings on the ability to communicate acoustically in a small, freshwater fish.  The 
results of this study revealed a particularly close association of a naturally occurring 
quiet window in the natural soundscape of C. venusta and a harmonic frequency within 
the courtship vocalization of C. venusta.  I found that C. venusta?s acoustic signals 
propagate very short distances following predictions based on the calculated cutoff 
frequency of the streams they inhabit, and may be easily disrupted by the noise 
generated from interstate and train bridge crossings.  
123 
 
Introduction 
 Sound can be found anywhere a medium in which it may propagate exists.  
Anything that moves produces sound, and anything with a device able to convert 
pressure fluctuations or particle displacements into action potentials may be able to 
detect the sounds.  Because sound has the capacity to carry information, can be used 
intermittently, and does not require a line of sight, it is used extensively by animals as a 
mode of information transfer.  However, numerous factors concerning the properties of 
the projected sound, hearing abilities of the receiver, constraints imposed by the 
physical environment, and ambient noise levels of the environment must fit together 
properly in order for acoustic communication to be effective.   
 The correlation of sound frequency and intensity with hearing sensitivity is a 
critical first step in the successful transfer of information through acoustics.  If an 
organism?s hearing sensitivity at a specific frequency is not sufficiently below the 
amplitude of the acoustic signal at that frequency, the sound will not be detected.  In 
many cases, the peak hearing sensitivity of an organism is correlated with the frequency 
of the sounds they produce for intraspecific communication (Bullock 1977; Capranica 
1978; Stabentheiner, 1988; Vasconcelos et al. 2007; Fay and Edds-Walton, 1997; Ladich 
and Yan, 1998).  This correlation will typically maximize the distance at which the sound 
can be detected by the receiver and minimize the required amplitude for successful 
detection.  However, this correlation is not always the case.  Mismatches in peak 
hearing sensitivity and sound production have been documented in birds (Sachs et al. 
1978; Dooling et al. 1978), humans (Geldard 1972), and fishes (Fine 1981; Ladich 1999).  
124 
 
The reason for mismatches is not always known, but is likely the result of conflicting 
selective pressures.  Fine (1981) suggested that in the oyster toadfish (Opsanus tau), the 
mismatch may promote the response of females to relatively louder calls of nearby 
males, which is often an indicator of male size. 
 The physical environment can influence the usefulness of acoustic signals.  In 
free field conditions (the deep ocean or open air), sound attenuates according to 
spherical spreading (a decrease in 6 dB for each doubling of distance).  Dense vegetation 
can have a scattering effect on sound, resulting in reverberations that degrade 
information encoded in amplitude and frequency modulations (Richards and Wiley 
1980).  The location of a sound producing individual with reference to the ground or 
other reflective boundary, as well as the type of reflective boundary can also affect 
emitted acoustic signals (Forrest, 1994).  The medium in which a sound is propagating 
can also affect transmission.  Absorption, the loss of energy from a propagating sound in 
the form of heat and the relaxation of molecules in the medium (Bass, 1991), is 
frequency dependent, and is about 100 times greater in air than in water (Michelsen, 
1978).  However, the close presence of two boundaries in shallow water, the water 
surface and the substrate, present a unique challenge for organisms attempting to 
communicate underwater using acoustics.  Sounds below a specific cutoff frequency, 
defined mainly by the water depth and substrate composition (Officier, 1958; Rogers 
and Cox, 1988), experience severe attenuation in power by a factor of up to 100,000 
relative to frequencies above the cutoff frequency (Forrest, 1994).   
125 
 
Ambient noise from both biotic and abiotic sources can also have a large effect 
on acoustic communication by decreasing the signal-to-noise ratio of signals, making 
temporal information more difficult to extract (Wysocki and Ladich, 2005), or by 
masking signals altogether.  Abiotic noise can come from sources such as wind, rainfall, 
or turbulence while biotic noise comes from the acoustic signals of other organisms.  
Studies on terrestrial species have shown that these environmental noise sources can 
act as strong selective pressures in the evolution of signal structure (Waser and Waser, 
1977; Wiley and Richards, 1982; Narins et al. 2004).  Jouventin et al. (1999), for 
example, found that king penguin, Aptenodytes patagonicus, chicks find their parents in 
a crowded, noisy colony through a highly specialized call that takes advantage of a brief 
(0.23 s) inflection point containing all the information needed for recognition, along 
with high degrees of redundancy in time and frequency domains.  Narins et al. (2004) 
also found that an arboreal frog (Amolops tormotus) and a leaf warbler (Abroscopus 
albogularis) possess ultrasonic harmonics in their calls, and suggest that their function 
may be to avoid masking in the wideband river noise of their habitat. 
Unlike natural biotic and abiotic noise sources, the relatively recent development 
and rapid expansion of human activities such as urbanization, shipping, motorized 
recreational activities, drilling, and seismic explorations (Myrberg, 1990; Popper, 2003) 
may not be providing the time necessary for the evolution of acoustic signals in some 
animals.  While efforts have been made to determine the effect of anthropogenic noise 
on marine mammals (Richardson et al. 1995), investigations on the effect on fishes has 
been sparse.  Elevated noise levels have also been shown to reduce egg survival, reduce 
126 
 
reproduction and reduce growth rates in fishes (Banner and Hyatt, 1973; Lagard?re, 
1982).  Several studies have also shown that anthropogenic noises can affect hearing 
abilities or fish behaviors (Fernandes et al. 2000; Vab? et al. 2002; Handegard et al. 
2003), which can have detrimental effects on fitness.  Amoser et al. (2004) found, for 
example, that noise from powerboats racing on an alpine lake was loud enough to be 
detected by otophysine fishes (fishes possessing a hearing specialization; see Popper 
and Fay, 2010) at up to 400 m away.  Vasconcelos et al. (2007) found that the noise from 
ferry boats in the Tagus River estuary (Portugal) caused significant hearing threshold 
shifts in the Lusitanian toadfish (Halobatrachus didactylus), and that females ability to 
detect male signals would be significantly diminished under ship noise.   
Despite the efforts that have been made elsewhere, I am currently unaware of 
any study that has looked at potential anthropogenic noise sources in small streams, 
and how these noise sources may impact the ability of small, vocal cyprinids to 
communicate acoustically.  The current study was aimed at describing the interaction 
between the natural soundscape and acoustic repertoire and hearing of the blacktail 
shiner (Cyprinella venusta) in east Alabama, and examining the effect of two 
anthropogenic noise sources that occur in low order streams that are not typically 
exposed to motorized boat traffic.  The results provide a better understanding of how 
small, freshwater fishes cope with the seemingly noisy natural soundscape of their 
environment, and how two common sources of anthropogenic noise may affect 
communication. 
 
127 
 
Methods 
ABR 
Hearing thresholds were determined using the auditory brainstem response 
(ABR; aka auditory evoked potential , AEP).  Methods and equipment were identical to 
those used in Holt and Johnston (2011) with a few exceptions that will be mentioned 
here.  Test subjects (seven C. venusta and five C. auratus) were wrapped in bandage 
gauze and pinned (through the gauze) to a flat bed of clay in lieu of restraining the fish 
with clay alone.  This was done in response to concerns that the heavy clay would affect 
the sound field at the fish?s body.  Also, frequencies used in this experiment included 
100, 200, 300, 400, 600, 800, 1000, 2000, and 3000 Hz. 
 
Natural and Anthropogenic Ambient Noise Measurements and Propagation 
A large shoal on Little Uchee Creek (Lee Co., AL, U.S.A., 32.508579? N, -
85.184215? W), a tributary of the Chattahoochee River, was chosen to describe the 
natural soundscape of C. venusta.  This location was chosen because it offered a wide 
variety of suitable spawning habitats for C. venusta such as pools, runs, and chutes.  
During the reproductive summer months, the water is also periodically shallow and 
clear enough to allow a researcher to locate the exact locations of nests by watching the 
fish behave from the bank.  A total of 9 active nest sites were identified by observation 
from the creek bank using polarized sunglasses.  Nest sites were typically found at the 
confluence of a chute or waterfall and a pool.  However, spawning aggregations were 
also observed directly within rapidly flowing chutes, and pools with negligible flow.   
128 
 
A hydrophone (Hi-Tech HTI-96-MIN, sensitivity -164.4 re 1V/?Pa, frequency 
response: 0.002-30 kHz) and digital recorder (Marantz PMD 661, sampling rate 44.1 kHz) 
were used to record 1 minute of ambient noise in each of the 10 sites.  In sites with 
substantial flow, an effort was made to place the hydrophone in a still area adjacent to 
the flow, or in the eddy of a rock to reduce hydrodynamic noises.  Sounds were 
imported into Raven 1.4 (Cornell University, Ithaca, NY), where three, 1 sec segments 
were randomly selected from the recording made at each site.  Two pressure density 
spectra of each 1 sec segment were then calculated using the power spectrum function 
of Raven (Hanning window, FFT length: 2048 samples, bin resolution: 21.5 Hz and 
Hanning window, FFT length: 512 samples, bin resolution: 86.1 Hz).  Decibel levels were 
corrected to represent actual levels by taking into consideration the gain applied to the 
signal by the Marantz recorder and the sensitivity of the hydrophone.  Decibel levels 
were also corrected to represent dB re 1?Pa2/Hz by subtracting 10log(sampling rate/FFT 
sampling size) dB.  The pressure density spectra levels were exported into an excel file, 
where the three, 1 sec segments from each nest site were averaged to produce a single 
power density spectra for each of the 10 nest sites, which were then averaged to 
generate a single power spectrum for the natural ambient noise.  Kendall?s concordance 
test was used to determine whether the spectrum shape (using the power density 
spectrum curve with bin resolution of 21.5 Hz) of natural ambient noise between 21.5 
and 1999.5 Hz was significantly different across active nesting sites (Lugli and Fine 
2003). 
129 
 
Source levels and propagation of train and road noise were measured at the 
crossing of Interstate-85, and a CSX Railroad crossing at Uphapee Creek (Macon County, 
AL, interstate: 32.47441?N, -85.68656?W, train: 32.48429?N, -85.70605?W).  Both 
crossings were beam bridges with at least one piling exposed to the water in the creek.  
For all recordings at both the train and interstate crossing, the hydrophone was placed 
approximately 10 cm off the substrate.  This depth was chosen because the substrate 
was mostly sand and gravel, and most potential nests (logs, rocks, or other items that 
could present suitable crevices) in this type of habitat are usually close to the substrate.  
Also, when spawning aggregations were observed, behaviors that were associated with 
sound were usually performed near the nest.  
At the interstate crossing, a hydrophone was fixed 1.3 m upstream of the piling 
that entered the water.  A second hydrophone was moved 2, 4, 6, 9, 12.2, and 14 m 
upstream of the piling and both hydrophones recorded simultaneously on two channels 
of the Marantz digital recorder.  Because of hydrophone cable length constraints, 
measurements at 22, 48.8, and 82.3 m upstream of the piling were performed with a 
single hydrophone and levels were compared to the average semi-trailer truck levels at 
1.3 m.  Recording was performed for several minutes at each distance, and the time at 
which semi-trailer trucks passed was noted.  A total of three semi-trailer truck passes 
were selected from the recordings at each distance.  To determine the noise level of the 
semi-trailer trucks at the source, two power density spectra were generated (bin 
resolution of 21.5 and 86.1 Hz) for each measurement made at 1.3 m away from the 
bridge (n=17).  Data were exported to excel and the same corrections to level that were 
130 
 
used in the natural ambient noise at nesting sites was applied.  The 17 spectra were 
then averaged to generate two final power density spectra (one at a bin resolution of 
21.5 Hz and one at 86.1 Hz) representing semi-trailer truck noise 1.3 m downstream of 
the bridge (Figure 3-7). 
To calculate attenuation of noise at the interstate crossing, three semi-trailer 
truck passes at each distance were selected from the audio files.  Pressure density 
spectra were calculated for each of the three passes using Raven (Hanning window, FFT 
length: 32768 samples, bin resolution: 1.35 Hz).  The pressure density spectra were 
exported to Excel where they were adjusted so that they represented dB re 1?Pa2/Hz.  
The average power in 1/3 octave bands was then calculated for bands with center 
frequencies between 63.1 and 5011.9 Hz for each of the semi-trailer truck passes.   
Attenuation of the noise in each of the 1/3 octave bands was calculated by subtracting 
the level between the two hydrophones for each 1/3 octave and averaging this 
attenuation across the three semi-trailer truck passes at each distance. 
At the train crossing, a source level was taken at 3 m downstream of the piling 
that entered the water.  Subsequent recordings were taken at 8, 12.5, and 18 m 
downstream of the piling.  All recordings lasted between a few seconds before the 
locomotive started upon the bridge, to a few seconds after the last car had left the 
bridge.  Similar methods were used to determine levels at the source, and attenuation 
of the train noise as interstate noise, except that instead of selecting three semi-trailer 
truck passes at each distance, five, 1 sec segments were selected from the train noise at 
each distance.  One addendum was that the portion of the sound that included the 
131 
 
locomotive crossing was excluded due to its transient nature relative to the duration of 
the rest of the train crossing, and that it?s level was significantly different than the rest 
of the train. 
Bin resolutions of 21.5 and 86.1 were included in the analysis to allow for signal-
to-noise ratios (SNR?s) comparison between C. venusta acoustic signals and the different 
noise sources.  When comparing SNR?s, it was necessary to analyze all sounds to be 
included in the analysis at the same frequency resolution.  The bin resolutions of 21.5 
and 86.1 Hz result from the typical duration of growls and knocks.  Decibel levels of 
knocks and bursts were measured in the lab, and further details describing the acoustic 
signals of C. venusta will be presented in a forthcoming paper.   
 
Propagation of C. venusta acoustic signals 
Propagation of C. venusta signals was measured both directy from the fish, and 
by playing the sounds back through an underwater speaker.  To measure propagation 
directly, a nest guarding male was found, and a hydrophone was placed as close to the 
nest as possible without chasing the sound producing male away.  A second hydrophone 
was placed 20 ? 30 cm away from the first hydrophone, and both recorded 
simultaneously to the Marantz digital recorder.  Video of the sound producing male at 
the nest site was recorded using an underwater camera (Aqua-Vu ZT 60, Nature Vision 
Inc., Brainerd, MN, U.S.A) linked to a mini digital video recorder (Super Circuits, 
MDVR25).  Raven 1.4 was used to analyze the sounds.  Sounds included in the analysis 
of propagation were limited to those that were made when the male was within 1-2 cm 
132 
 
directly above the hydrophone closest to the nest.  The sound level at this hydrophone 
was considered the source level of the sounds.  Using Raven, the individual burst that 
occurred when the male was positioned directly above the source hydrophone was 
isolated on both channels.  The pressure density spectra level of the burst was 
calculated at both hydrophones using the power spectrum function of Raven (Hanning 
window, n-point FFT where n is the number of points in the signal, bin resolution: 
44100/n), and was corrected to represent actual dB re 1?Pa2/Hz as described 
previously.  The average power was then calculated in 13 one-third octave bands 
(center frequencies from 63 to 1000 Hz) at both hydrophones.  Attenuation in each one-
third octave band, at each distance was calculated by subtracting the level at the distant 
hydrophone from the source hydrophone. 
Propagation was also measured by playing the sounds back through an 
underwater speaker (University Sound UW-30, Oklahoma City) and recording at 
different distances.  A sound clip consisting of a growl preceded and followed by a single 
knock was generated in Raven (the knocks were from agonistic behaviors and the growl 
was from a courtship behavior).  The sound was 2.2 sec in duration, and the dominant 
frequency of the knocks was 17 dB louder than the dominant frequency of the growl 
(the typical difference in level between growls and knocks).  The sound was played from 
a Blackberry curve 8520, and amplified by a Pioneer GM-X372 240W amplifier.  One 
hydrophone was placed 5.08 cm directly in front of the speaker, and a second 
hydrophone was moved away from the first in increments of 5.08 cm up to 96.52 cm.  
At some nest sites, measurements were not able to be taken along the entire 96.52 cm, 
133 
 
and analysis was limited to distances that were included at all sites.   The generated C. 
venusta sound was recorded by both hydrophones at each distance and attenuation was 
calculated using the same methods described above for propagation of fish-produced 
sounds. 
 
Results 
The audiogram of C. venusta was very similar to that of C. auritus (Figure 3-1).  
Evoked potential traces at 100 and 200 Hz appeared different than traces at higher 
frequencies (Figure 3-2).  Lower frequencies (100 and 200 Hz) were associated with 
longer latencies, and showed multiple spikes within a longer dip and rise.  Frequencies 
above 200 Hz generally showed a cleaner evoked potential that lacked the series of 
smaller spikes following the initial dip and rise. 
 The pressure and particle acceleration audiograms for C. venusta and C. auratus 
both showed a U shaped sensitivity curve typical of hearing specialists (Figure 3-1a,b).  
Cyprinella venusta sensitivity decreased relatively linearly from 100 Hz (95.7 dB; 2.5e-3 m 
s2) to 300 Hz (87.1 dB; 9.2e-4 m s2), increased slightly to 89.3 dB (1.2e-3 m s2) at 400 Hz, 
then dropped to the peak sensitivity of 83.6 dB (6.5e-4 m s2) at 600 Hz.  Above 600 Hz, 
thresholds increased relatively linearly to 116.4 dB (3.0e-2 m s2) at 2000 Hz, which was 
the highest attainable threshold for C. venusta.  Carassius auratus differed from C. 
venusta in slightly lower thresholds at 100 and 400 Hz. 
134 
 
 At frequencies above 100 Hz, background noise was typically between 45 -50 dB, 
more than 30 dB lower than the lowest threshold.  Spikes in the background noise were 
found at 60, 120, 180, 240, 300, 420, 540, 660, and 780 Hz. 
Noise levels of semi-trailer trucks crossing Uphapee Creek at Interstate-85 were 
relatively high at the source (1.3m upstream of the bridge piling), with the greatest 
power occurring at low frequencies below 180 Hz.  The average power between 0-56, 
the 63 Hz band, and the 80 Hz band at 1.3 m from the bridge was 103.8, 108.8, and 
105.7 dB re 1?Pa, respectively.  Power declined linearly from 106.2 dB at the 100 Hz 
band, to 69.6 dB at the 500 Hz band at a rate of approximately 0.1 dB/Hz.  Power 
increased to 71.2 dB at the 800 Hz band, and then gradually decreased until levels were 
comparable to natural background levels at Uphapee Creek (approximately 47 dB) at 2.4 
kHz and above.  Attenuation of semi-trailer truck noise showed some variation with 
frequency, but generally followed attenuation rates between spherical and cylindrical 
theoretical rates (spherical spreading based on 20log(distance from source); cylindrical 
spreading based on 10log(distance from source);Figure 3-3). 
At the train crossing, noise levels were considerably higher than the semi-trailer 
truck noise.  Train noise showed more variation than semi-trailer truck noise at 
frequencies below 540 Hz.  The average power between 0-56, the 63 Hz band, and the 
80 Hz band at 3 m from the bridge was 115.3, 111.9, and 109.4 dB respectively.  There 
was a distinct gap in the train noise spectrum at the 400 and 500 Hz one-third octave 
bands where the levels were 97.7 and 101.9 dB respectively.  Surrounding this window 
were spikes in the spectrum of 109.0 dB at the 250 Hz band, and 104.1 dB at the 630 Hz 
135 
 
one-third bands.  Between the 630 and 1250 Hz bands, power levels dropped relatively 
linearly at a rate of approximately 0.054 dB/Hz.  Above the 1250 Hz band, levels 
continued to drop, but very slowly, and stayed about 7.4 ? 3.5 dB above the ambient 
noise levels up to and beyond 4.0 kHz.  Attenuation of train noise generally followed 
attenuation rates between spherical and cylindrical theoretical rates, similar to semi-
trailer truck noise (Figure 3-4) 
The peak level of natural ambient noise averaged across 9 active nesting sites 
was 73.6 ? 5.6 dB at 30 Hz.  Above 30 Hz, power levels dropped rapidly at a rate of 
approximately 0.13 dB/Hz into a quiet window between 180 ? 330 Hz, where the 
average power was 54.3 ? 1.3 dB (average power in 1.35 Hz bands).  Above this quiet 
window, power levels increased at approximately 0.026 dB/Hz to a second peak at 690 
Hz, where the average power was 66.9 ? 15.5 dB.  Power levels gradually declined above 
690 Hz to levels comparable to the quiet window at 2.2 kHz and above.  Significant 
concordance in spectrum shape of naturally occurring ambient noise between 21.5 and 
1999.5 Hz was found among the active nesting sites at Moffits Mill (W=0.428, chi-
square=232.64, P<0.001, df=8). 
Attenuation rates of natural C. venusta calls were very similar to those played 
through the speaker (Figure 3-5).  Assuming losses from cylindrical spreading as 3 
dB/distance doubled, the distance from 0 to 40.6 cm (with the first measurement at 
5.08 cm) should result in a transmission loss of 12 dB, which equates to 16 dB/m.  
Although some variation with frequency was apparent (Figure 3-6), attenuation at all 
frequencies was much more rapid than what cylindrical spreading would predict, and 
136 
 
follows what would be expected for signals below the cutoff frequency.  A linear best fit 
line plotted through the average attenuation of all frequencies for knocks and growls 
showed an attenuation of 55.9 dB/m.  At the 1000 Hz one-third octave band, the 
frequency at which attenuation was the slowest, the slope of the best fit linear line was 
38.9 dB/m.  At the 315 Hz one-third octave band, where attenuation was the most 
rapid, the slope was 62.8 dB/m.   
The sounds produced by C. venusta are broadband, with the greatest power 
existing at lower frequencies.  With growls, the dominant frequency typically occurs 
around 70 Hz (which is the pulse repetition rate), but can also be found around 300 Hz 
(which is typically the dominant frequency of an individual pulse).  Knocks are also 
broadband in nature with the majority of their power in the low frequency range.  The 
dominant frequency of knocks is also variable and often occurs between 70 and 300 Hz 
or 500 and 700 Hz.  In the laboratory, under quiet conditions, bursts are produced with 
levels of 70.42 ? 7.05 dB at 86 Hz and 67.41 ? 8.08 dB at 301 Hz (bin resolutions of 21.5 
Hz; Figure 3-7).  Knocks are produced with levels of 96.89 ? 5.08 dB at 172.2 Hz and 
97.16 ? 8.24 dB at 602.7 Hz (corrected for hydrophone distance, bin resolutions of 86.1 
Hz; Figure 3-8).   
Applying attenuation rates of C. venusta signals observed in the field to the 
average level of sounds produced by C. venusta in the lab, and the average power 
spectra of naturally occurring noise their environment, I was able to approximate the 
potential active area of the signals under different noise conditions. Fishes are generally 
unable to detect pure tones with signal-to-noise ratio below 15 - 20 dB (10 - 15 dB if the 
137 
 
signal and noise are coming from the different directions; Buerkle, 1969; Chapman, 
1973; Chapman and Sand, 1974; Fay, 1974; Hawkins and Sand, 1977; Fay 1988; Mann 
and Lobel, 1997).  At 172.2 and 602.7 Hz, the natural noise is 49.91 ? 4.91 and 51.96 ? 
14.12 dB, respectively.  Based on the average knock levels and attenuation rates,C. 
venusta knocks should propagate approximately 50.73 cm at 172.2 Hz and 65.19 cm at 
602.7 Hz before reaching a signal-to-noise ratio of 15.0 dB (SNR calculated as Signal level 
? Noise level, which is equilivant to 20log(signal voltage/noise voltage)).  At 86 and 301 
Hz, the natural noise is 53.75 ? 3.82 and 43.19 ? 10.02 dB, respectively.  Based on the 
average burst levels and attenuation rates, growls should propagate approximately 2.99 
cm at 86 Hz and 16.49 cm at 301 Hz before reaching a SNR of 15 dB.   
Losses based on spherical spreading show that in the presence of semi-trailer 
truck noise at the interstate crossing (96.64 dB at 86 Hz and 67.82 dB at 301 Hz; 1.3 m 
away from the bridge piling), the growls should emerge from the noise with a SNR 
above 15 dB at distances beyond 117.8 m from the bridge at 86 Hz and 5.9 m from the 
bridge at 301 Hz .  At 172.2 and 602.7 Hz, the semi-trailer truck noise is 87.98 and 50.94 
dB, respectively.  The 172 Hz portion of knocks should reach a SNR above 15 dB at 2.0 m 
from the bridge.  At 602.7 Hz, knocks already have a SNR of 31.76 dB at the closest 
distance to the bridge that semi-trailer truck noise was measured (1.3 m). 
Three meters away from the bridge piling at the train crossing, the noise level 
was 105.64 and 101.67 dB at 86 and 301 Hz, respectively.  Based on spherical spreading 
rates of the train noise, the 86 Hz portion of growls should reach a SNR above 15 dB at 
distances beyond 324.3 m from the bridge.  The 301 Hz portion of growls should reach a 
138 
 
SNR above 15 dB at distances beyond 290.4 m.  Noise levels were 102.14 and 90.76 dB 
at 172.2 and 602.7 Hz at the train crossing, respectively.   The 172.2 Hz portion of knocks 
should reach a SNR above 15 dB at distances beyond 10.3 m.  The 602.7 Hz portion of 
knocks should reach a SNR above 15 dB at distances beyond 2.7 m. 
The values presented above only indicate at what distance the signals should be 
detectable over the semi-trailer truck and train noise at the point source of the 
biological sound (aka, at the sound producing fish).  It may be more meaningful to 
report the distance at which anthropogenic noise levels should attenuate to levels 
comparable to natural noise, which would be the point at which the signals regain the 
active area observed under natural conditions.  Semi-trailer truck noise should 
attenuate to natural noise levels in 142.7 m at 86 Hz, 52.8 m at 172 Hz, 17.0 m at 301 
Hz, and 1.8 m at 602.7 Hz.  Train noise should attenuate to natural levels in 393.1 m at 
86 Hz, 269.5 m at 172 Hz, 839.5 m at 301 Hz, and 178.6 m at 602 Hz (Table 3-1). 
 
Discussion 
It was surprising that attenuation of the four highest one-third octave bands of 
semi-trailer truck noise followed attenuation rates intermediate to spherical and 
cylindrical spreading.  Because all of the bands were below the calculated cut-off 
frequency of the stream, my null hypothesis was that the frequencies would have 
shown attenuation values much greater than spherical or cylindrical spreading, and 
more similar to the attenuation rates of C. venusta signals.  One possibility for the 
persistence of the sounds is that they are traveling through the air and transmitting into 
139 
 
the stream at farther distances.  The critical angle of transmission from the air into the 
stream is 76.6? (based on Snell?s law; cos?c=c1/c2).  Only sounds with incident angles less 
than the critical angle are capable of being transmitted into the stream.  The bridge is 
8.5 m above the surface of the water, meaning that the incident angle of airborne 
sounds at the air/water boundary relative to the vertical would become greater than 
76.6? at 35.7 m away from the bridge.  Because an elevation in attenuation beyond this 
distance is not observed, the sounds must also be propagating through the substrate 
and re-emerging into the water column.  Because the composition of the bottom below 
the immediate substrate (sand) is unknown, I cannot address propagation in this respect 
beyond what was observed, which was attenuation between spherical and cylindrical 
spreading.  Because the train noise was not measured beyond a distance of 18 m, I 
cannot address the transmission from air into the water at longer distances.  However, 
because both the train and interstate bridges had pillars entering the water, were 
similar heights, and attenuation of noise was similar between the two bridges, I assume 
that the method of propagation is similar. 
The spawning strategy of Cyprinella venusta involves the deposition of demersal 
eggs into crevices formed by bedrock, woody debris, or other structure along the 
substrate.  This strategy is an effective method of protecting the eggs from many 
potential predators, but does not allow fanning by the parental fish, a technique used by 
many fish species that removes silt and circulates oxygenated water around the 
developing eggs.  It is not surprising, therefore, that nest sites associated with flow are 
preferred by C. venusta (Baker et al. 1994).  All of the active nesting sites observed at 
140 
 
Moffits Mill were associated with substantial flow, and occurred within or just below a 
chute, or in a run just below a waterfall or riffle, an observation that corroborates 
findings of Baker et al. (1994).  Under natural conditions in shallow stream habitats, 
noise levels are usually dependent on hydrological factors such as water speed, volume, 
sediment transport, and cavitation (Wysocki et al. 2007).  Quiet areas typically occur 
where flow is minimal and the water surface is unbroken (Lugli and Fine 2003).  The 
presence of considerable flow and pieces of substrate breaking the surface of the water 
at C. venusta nesting sites most likely contributes to elevated levels of natural ambient 
noise at spawning sites. 
The low frequency spike in natural ambient noise around 20 Hz is caused mainly 
by turbulence from water flowing around submerged objects such as rocks (Strasberg 
1979; Lugli and Fine 2007).  The broadband elevation in the natural ambient noise 
around 690 Hz is probably due to bubble noise caused by the surface being broken by 
turbulence or objects, forcing pockets of air underwater (Lugli and Fine 2003).  Lugli and 
Fine (2003) suggested the quiet window found in the shallow streams in which they 
recorded was formed by the attenuation of the low frequency turbulence noise before 
the emergence of bubble noise at higher frequencies.  
Despite the presence of elevated natural noise at nesting sites, C. venusta are 
able to maintain a SNR above 15 dB at both the dominant, and sub-dominant 
frequencies of both call types, albeit at different distances.  Knocks showed levels that 
were drastically above the ambient noise at all frequencies relevant to C. venusta with 
SNR?s of 43.36 dB and 51.44 dB at 172 Hz and 602.7 Hz, respectively.  The presence of a 
141 
 
quiet window in the natural noise power spectrum around approximately 300 Hz was an 
important finding of the current study.  Because of large SNR of knocks at all 
frequencies, this quiet window is irrelevant with reference to knocks.  The subdominant 
frequency of growl type calls, however, occurs directly within this quiet window with a 
SNR of 24.22 dB.  This quiet window is especially relevant due to the low SNR of growls 
relative to the natural ambient noise at frequencies outside the window (Figure 3-7).  
The dominant frequency of growls, for example, is 3 dB greater than the subdominant 
frequency, but does not overlap as well with the quiet window, and is thus only 
detectable up to 2.99 cm, which is 13.5 cm less than the subdominant frequency.   
Several other studies have found quiet windows in the natural background noise 
of stream environments (Wysocki et al. 2007; Speares et al. 2011; Lugli and Fine 2003; 
Amoser 2006; Crawford et al. 1997).  Lugli et al. (2003) found a match between the peak 
hearing, quiet window, and dominant frequencies utilized during sound production in 
two freshwater gobies (Padogobius martensii and Gobius nigricans).  Speares et al. 
(2011) found that the dominant frequency of two darter species (Etheostoma flabellare 
and E. crossopterum) falls within a relatively quiet area of the natural ambient noise 
spectrum.  Crawford et al. 1997 also found that the sounds produced by the mormyrid 
Pollimyrus isidori fit into a very broad quiet window (200-3000Hz) in the Niger River 
(Mali, Africa).  Because, for the most part, hydrology dictates natural environmental 
noise, and fish are often bound to certain habitat types due to specific life history 
strategies (such as for C. venusta), the utilization of a naturally occurring quiet window 
in the ambient noise is probably an adaptation for intraspecific communication.  The 
142 
 
almost perfect match between growl subdominant frequency and the naturally 
occurring quiet window in C. venusta habitat may be considered as potential evidence 
that the environmental soundscape acted as a selective pressure in the evolution of 
signal structure for C. venusta. 
Detection distances of the different call types do not appear to be arbitrary and 
seem to fit appropriately with the behaviors they accompany.  The ability of C. venusta 
to theoretically detect knocks at over half of a meter away indicates that males are 
capable of relatively long distance acoustic communication under normal ambient noise 
conditions.  This correlates well with the types of behaviors that are typically associated 
with knocks.  Knocks are usually produced by males during aggressive encounters with 
other males around the nesting site.  Knocks were also observed during chases between 
nest guarding males and sunfish that would attempt to eat eggs from the nest.  If knocks 
serve as an honest signal of male fitness to other males, a larger active area of the 
sound should benefit the sound producer.  Because there are often many other 
reproductive males at a given nesting site, a territorial male should benefit from sending 
honest signals about their intentions and motivational level to both the intruding male 
and peripheral males simultaneously. 
Growls are produced more often during courtship behaviors with females.  
During courtship, males will often approach and circle a potential female quickly, 
making frequent trips back and forth between the nest and the female while all the time 
producing growls.  Sound production continues up to the moment of spawning, ceases 
during the 1 ? 2 sec of actual spawning, then immediately resumes.  During the course 
143 
 
of this courting behavior, the distance between male and female varies, but is often 
close (within 15 cm).  During the period immediately prior to and during spawning, the 
male is very close (within 1-2 cm) to the female.  Due to the small active area of growls, 
it is unlikely that they are used as advertisement to distant females.  They are most 
likely strictly intended for the individual female being courted, possibly for spawning 
synchronization, or as a signal of male fitness. 
Semi-trailer truck and train noise was shown to have the ability to diminish the 
normal active area of both growls and knocks, depending on the frequency of concern 
and the distance from the noise source (Table 3-1).  Growls showed the potential to be 
more severely affected than knocks, due to their lower amplitude and greater overlap 
with anthropogenic noise frequency spectrum.  The location of the subdominant 
frequency fits ideally within the natural ambient noise spectrum, as might be expected 
in a natural system.  However, both sources of anthropogenic noise significantly 
increase the noise levels within the quiet window. 
It is difficult to assess just how much the anthropogenic noise sources may affect 
communication and overall reproductive success of C. venusta.  It is not known whether 
there is a difference in the information content of the dominant and subdominant 
frequencies, which are affected differently by the anthropogenic noises.  Several studies 
have shown that dominant frequency is correlated with male size, which can be 
considered an indicator of male fitness.  In the case of C. venusta, the first harmonic 
frequency is negatively correlated with fish standard length while the fundamental 
frequency is not.  This, along with the fact that the quiet window corresponds more 
144 
 
closely to the first harmonic frequency, suggests that the fundamental frequency may 
not carry as much importance as the first harmonic frequency.  Other signal parameters 
such as call duration have also been found to be correlated with male condition 
(Amorim et al. 2010).  If call duration carries such information in C. venusta, a smaller 
active area of growls induced by anthropogenic noise events could cause growls to be 
heard for a shorter period of time by a female, which could diminish perceived male 
fitness and have adverse effects on reproductive success.  Unfortunately, I was unable 
to measure call duration as the fish were often moving in and out of hydrophone range 
while vocalizing, frequently causing the beginnings and ends of calls to be undefinable. 
It is also possible that the periodic nature of both semi-trailer truck and train 
noise allow sufficient communication between noise events.  However, the ability to 
effectively communicate even between noise events may require that hearing is not 
significantly affected by exposure to the noise.  Numerous studies have shown that 
exposure to elevated noise can cause temporary threshold shifts (TTS?s) in fishes (Smith 
et al. 2004; Popper and Clarke, 1976; Scholik and Yan, 2001; Amoser and Ladich, 2003).  
However, most of these studies have looked at relatively long term exposure to 
constant noise at very high levels, with the shortest exposure times being 10 minutes at 
122 dB re 1 ?Pa2/Hz (after which C. auratus exhibited a 5 dB TTS; Smith et al. 2004).  
Although the recovery time is not specified for that particular exposure, recovery from a 
greater TTS followed an exponential curve with initial rapid recovery rates slowing as 
the threshold neared original levels.  It is not known how hearing thresholds would be 
affected by continued exposure to relatively low level transient noise sources such as 
145 
 
the anthropogenic sources investigated in this paper.  Lugli et al. (2003), however, found 
that the hearing thresholds of P. martensii and G. nigricans matched (albeit about 30 dB 
higher) the upper standard deviation spectrum level curve of the stream locations with 
the highest ambient noise levels.  This suggests that periodic elevations in ambient noise 
could dictate hearing thresholds more than average levels.  In the current study, 
background noise levels at 301 Hz are elevated to 67.82 dB at 1.3 m by semi-trailer truck 
crossings, and 101.67 dB at 3 m by train crossings.  Although these are not exceptionally 
high levels, and values will decrease as distance is increased from the bridge, numerous 
studies have found that threshold shifts in response to a white noise are greatest at 
frequencies where hearing sensitivity is best (Smith et al. 2004; Popper and Clarke, 
1976; Amoser and Ladich, 2003), which most likely occurs within the quiet window 
(around 300 Hz) for C. venusta.  Further research focused on the effect of these 
particular noise sources and the avoidance rate of fishes around the bridges is needed 
for a more clear understanding of how these noise sources affect hearing. 
The ABR?s in this study provide similar audiograms between C. venusta and C. 
auratus.  Descriptions of hearing sensitivity in fishes using both behavioral and 
electrophysiological methods such as the ABR have shown very high variation between 
researchers, even for identical species.  Carassius auratus is probably the most 
extensively tested species of fish, and it has been suggested by Higgs et al. (2003) that 
authors describing the hearing sensitivity of a new species include an audiogram of C. 
auratus using the same system to facilitate interspecific comparisons.  Because of the 
high degree of variation in audiogram sensitivity seen across numerous studies, the 
146 
 
notion that ABR provides a conservative estimate of sensitivity, the fact that the 
greatest sensitivity on the audiogram produced for C. venusta in the current study is 13 
dB above the average peak dB of growls, and the fact that the audiograms for C. venusta 
and C. auratus were so similar, I feel that a more realistic picture of C. venusta hearing 
abilities may be obtained by looking at historical C. auratus audiograms obtained 
through behavioral methods.  Popper (1971) found an audiogram with a shape similar to 
the current study with a peak sensitivity of 52 dB re 1?Pa at 500 Hz and 54 dB at 300 Hz.  
Jacobs and Tavolga (1967) also found a similarly shaped audiogram with a peak 
sensitivity of 55 dB at 500 Hz.  I believe that because of physiological similarities, similar 
audiogram shapes, and the fact that hearing sensitivities should be below the level of 
the acoustic signals, it is not unreasonable to assume that hearing sensitivities of C. 
venusta should more closely resemble the behavioral audiograms found historically for 
C. auratus. 
The findings of the current study offer a new perspective at which to look at 
reproductive behaviors in C. venusta.  The noisy environment in which they often live 
has been shown to have a convenient window in the noise spectrum, which has been 
exploited by C.venusta for the purpose of communication with females during 
reproductive behaviors, albeit at very short distances.  Two sources of anthropogenic 
noise were determined to have potentially detrimental effects on the communicative 
ability of C. venusta during critical periods in it?s life history.  This work is the first to 
explore the effect of bridge crossing anthropogenic noise sources on the acoustic 
communication of fishes in a small, unnavigable stream.  Future work should attempt to 
147 
 
determine whether C. venusta are avoiding areas impacted by such noise and if not, 
what affect the noise is having on hearing ability.  Investigations should also look into 
the effect of different types of stream crossings on noise levels, and whether more 
sedentary species such as darters are affected. 
 
References 
Amorim, M.C.P, Sim?es, J.M., Mendon?a, N., Bandarra, N.M., Almada, V.C., and  
Fonseca, P.J.  2010.  Lusitanian toadfish song reflects male quality.  The Journal 
of Experimental Biology.  213:2997-3004. 
Amoser, S.  2006.  Year-round changes of ambient noise in freshwater habitats and  
temperature effects on hearing in fishes.  Journal of the Acoustical Society of 
America.  120:3056-3056. 
Amoser, S. and Ladich, F.  2003.  Diversity in noise-induced temporary hearing loss in  
otophysine fishes.  Journal of the Acoustical Society of America.  113:2170-2179. 
Amoser, S. Wysocki, L.E., and Ladich, F.  2004.  Noise emission during the first  
powerboat race in an alpine lake and potential impact on fish communities.  The 
Journal of the Acoustical Society of America.  116:3789-3797. 
Baker, J.A., Killgore, K.J. and Foster, S.A.  1994.  Population variation in spawning current  
speed selection in the blacktail shiner, Cyprinella venusta (Pisces: Cyprinidae).  
Environmental Biology of Fishes.  39:357-364. 
Banner, A., and Hyatt, M.  1973.  Effects of noise on eggs and larvae of two estuarine  
fishes.  Transactions of the American Fisheries Society.  1:134-136. 
148 
 
Bass, H.E.  1991.  Atmospheric acoustics.  In: Encyclopedia of Applied Physics.  Vol. 2.   
pp. 145-179.  VCH Publishers. 
Buerkle, U. 1969.  Auditory masking and the critical band in Atlantic Cod (Gadus  
morhua).  J. Fish. Res. Brd. Can.  26: 1113-1119. 
Bullock, T.H.  1977.  Recognition of complex acoustic signals.  Life sciences research  
report 5.  Dahlem Konferenzen, Berlin, 1977. 
Capranica, R.R.  1978.  Symposium on auditory processing and animal sound  
communication.  Fed. Proc.  37:2315-2359. 
Chapman, C. J. 1973.  Field studies of hearing in teleost fish.  Helg. Meer. 24: 371-390 
Chapman, C. J., and Sand, O.  1974.  Field studies of hearing ink two species of flatfish  
Pleuronectes platessa (L) and Limanda limanda (L) (Family Pleuronectidae).  
Comp. Biochem. Phys. A.  47:371-385. 
Crawford, J. D., Jacob, P., and B?nech, V.  1997.  Sound production and reproductive  
ecology of strongly acoustic fish in Africa: Pollimyrus isidori, Mormyridae.  
Behaviour.  134:677?725.  
Dooling, R.I., Zoloth, S.R., and Baylis, J.R.  1978.  Auditory sensitivity, equal loudness,  
temporal resolving power, and vocalizations in the house finch (Carpodacus 
mexicanus).  Journal of Comparative Physiology and Psychology.  92:867-876. 
Fay, R.R.  1974.  Masking of tones by noise for the goldfish (Carassius auratus).  J. Comp.  
Physiol. Psych.  87:708-716. 
Fay, R.R.  1988.  Hearing in Vertebrates: A Psychophysics Databook (Hill-Fay Associates,  
Winnetka, IL). 
149 
 
Fay, R.R., and Edds-Walton, P.L.  1997.  Diversity in frequency properties of saccular  
afferents of the toadfish, Opsanus tau.  Hearing Research.  113:235-246. 
Fernandes, P.G., Brierley, A.S., Simmonds, E.J., Millard, N.W., McPhail, S.D., Stevenson,  
P., and Squires, M.  2000.  Fish do not avoid survey vessels.  Nature.  604:35-36. 
Fine, M.L.  1981.  Mismatch between sound production and hearing in the oyster  
toadfish.  In: Hearing and Sound Communication in Fishes.  (Ed) Tavolga, W.N., 
Popper, A.N. and Fay, R.R.  Springer, New York.  pp 257-263. 
Forrest, T.G.  1994.  From sender to receiver: Propagation and environmental effects on  
acoustic signals.  Journal of Integrative and Comparative Biology.  34:644-654. 
Geldard, F.A.  1972.  The human senses.  New York: Wiley. 
Handegard, N.O., Michalsen, K., and Tjostheim, D.  2003.  Avoidance behavior in cod  
(Gadus morhua) to a bottom-trawling vessel.  Aquat. Living Resour.  16:265-270. 
Hawkins, S.D., and Sand, O.  1977.  Directional hearing in the median vertical plane by  
the cod.  J. Comp. Physiol. A.  122:1-8. 
Hawkins, A.D. and Myrberg, A.A. Jr.  1983.  Hearing and sound communication under  
water.  In: Lewis B (ed) Bioacoustics: a comparative approach.  Academic Press, 
London, pp 347-405. 
Higgs, D.M., Rollo, A.K., Souza, M.J., and Popper, A.N.  2003.  Development of form and  
function in peripheral auditory structures of the zebrafish (Danio rerio).  Journal 
of the Acoustical Society of America.  113:1145-1154. 
Holt, D.E., and Johnston, C.E.  2011.  Hearing sensitivity in two black bass species using  
150 
 
the auditory brainstem response approach.  Environmental Biology of Fishes.  
91:121-126. 
Jacobs, D.W., and Tavolga, W.N.  1967.  Acoustic intensity limens in the goldfish.  Animal  
Behaviour.  15:324-335. 
Jouventin, P., Aubin, T., and Lengagne, T.  1999.  Finding a parent in a king penguin  
colony: the acoustic system of individual recognition.  Animal Behaviour.  
57:1175-1183. 
Ladich, F.  1999.  Did auditory sensitivity and vocalization evolve independently in  
Otophysan fishes?  Brain Behavior and Evolution.  53:288-304. 
Ladich, F., and Yan, H.Y.  1998.  Correlation between auditory sensitivity and vocalization  
in anabantoid fishes.  Journal of Comparative Physiology A.  182:737-746. 
Lagard?re, J. P.  1982.  Effects of noise on growth and reproduction of Crangon crangon  
in rearing tanks. Mar. Biol.  71:177-185. 
Lugli, M. and Fine, M. L.  2003.  Acoustic communication in two freshwater gobies:  
Ambient noise and short-range propagation in shallow streams.  Journal of the 
Acoustical Society of America.  114:512-521. 
Lugli, M. and Fine, M. L.  2007.  Stream ambient noise, spectrum and propagation of  
sounds in the goby Padogobius martensii: sound pressure and particle velocity.  
Journal of the Acoustical Society of America.  122:2881-2892. 
Lugli, M., Yan, H.Y., and Fine, M.L.  2003.  Acoustic communication in two freshwater  
gobies: the relationship between ambient noise, hearing thresholds and sound 
spectrum.  Journal of Comparative Physiology A.  189:309-320. 
151 
 
Mann, D.A. and Lobel, P.S.  1997.  Propagation of damselfish (Pomacentridae) courtship  
sounds.  J. Acoust. Soc. Am.  101:3783-3791. 
Michelsen, A.  1978.  Sound reception in different environments.  In: Ali, M.A. (ed).   
Sensory Ecology.  pp. 345-372.  Plenum, New York. 
Myrberg, A.A.  1990.  The effects of man-made noise on the behavior of marine  
mammals.  Environ. Int.  16:575-586. 
Narins, P.M., Feng, A.S., Lin, W., Schnitzler, H-U, Denzinger, A., Suthers, R.A., Xu, C.   
2004.  Old world frog and bird vocalizations contain prominent ultrasonic 
harmonics.  The Journal of the Acoustical Society of America.  115:910-913. 
Officier, C.B.  1958.  Introduction to the theory of sound transmission (McGraw-Hill,  
New York). 
Popper, A.N.  1971.  The effects of size on the auditory capacities of the goldfish.  The  
Journal of Auditory Research.  11:239-247. 
Popper, A.N.  2003.  Effects of anthropogenic sounds on fishes.  Fish.  Res.  28:24-31. 
Popper, A.N., and Clarke, N.L.  1976.  The auditory system of the goldfish (Carassius  
auratus): effect of intense acoustic stimulation.  Comparative Biochemistry and 
Physiology Part A: Physiology.  53:11-18. 
Popper, A.N., and Fay, R.R.  2010.  Rethinking sound detection by fishes.  Hearing  
Research.  273:25-36. 
Richards, D.G., and Wiley, R.H.  1980.  Reverberations and amplitude fluctuations in the  
propagation of sound in a forest: implications for animal communication.  The 
American Naturalist.  115:381-399. 
152 
 
Richardson, W.J., Green, C.R., Malme, C.J., and Thomson, D.H.  1995.  Marine Mammals  
and Noise.  Academic, London. 
Rogers, P.H., and Cox, M.  1988.  Underwater sound as a biological stimulus.  In: Atema,  
J., Fay, R.R., Popper, A.N., and Tavolga, W.N. (eds).  Sensory Biology of Aquatic 
Animals.  pp. 131-149.  Springer-Verlag, New York. 
Sachs, M.B., Sinnott, J.M., and Heinz, R.D.  1978.  Behavioral and physiological studies of  
hearing in birds.  Fed. Proc.  37:2329-2335. 
Scholik, A.R., and Yan, H.Y.  2001.  Effects of underwater noise on auditory sensitivity of  
a cyprinid fish.  Hearing Research.  152:17-24. 
Slabbekoorn, H., and Peet, M.  2003.  Birds sing at higher pitch in urban noise.  Nature.   
424:267. 
Smith, M.E., Kane, A.S., and Popper, A.N.  2004.  Noise-induced stress response and  
hearing loss in goldfish (Carassius auratus).  The Journal of Experimental Biology.  
207:427-435. 
Speares, P., Holt, D., and Johnston, C.  2011.  The relationship between ambient noise  
and dominant frequency of vocalizations in two species of darters (Percidae: 
Etheostoma).  Environmental Biology of Fishes.  90:103-110. 
Stabentheiner, A.  1988.  Correlations between hearing and sound production in  
piranhas.  Journal of Comparative Physiology A.  162:67-76. 
Strasberg, M.  1979.  Nonacoustic noise interface in measurements of infrasonic  
ambient noise.  Journal of the Acoustical Society of America.  66:1487-1493. 
Vasconcelos, R.O., Amorim, C.P., and Ladich F.  2007.  Effects of ship noise on the  
153 
 
detectability of communication signals in the Lusitanian toadfish.  The Journal of 
Experimental Biology.  210:2104-2112. 
Vab?, R., Olsen, K., and Huse, I.  2002.  The effect of vessel avoidance of wintering  
Norwegian spring spawning herring.  Fish.  Res.  58:59-77. 
Waser, P.M., and Waser, M.S.  1977.  Experimental studies of primate vocalization:  
specializations for long distance propagation.  Z. Tierpsychol.  43:239-263. 
Wiley, R.H., and Richards, D.G.  1982.  Adaptations for acoustic communication in birds:  
sound transmission and signal detection.  In: Kroodsma, D.E., Miller, E.H., and 
Quellet, H. (eds).  Acoustic Communication in Birds, vol 1.  Academic Press, New 
York, pp 131-278. 
Wysocki, L. E., Amoser, S., and Ladich, F.  2007.  Diversity in ambient noise in European  
freshwater habitats: noise levels, spectral profiles, and impact on fishes.  Journal 
of the Acoustical Society of America.  121:2559-2566. 
Wysocki, L. E. and Ladich, F.  2005.  Effects of noise exposure on click detection and the  
temporal resolution ability of the goldfish auditory system.  Hearing Research.  
201:27-36. 
  
154 
 
 
 
Figure 3-1 ?  Audiograms of Carassius auratus and Cyprinella venusta presented in 
(a)pressure and (b) particle acceleration.  Baseline noise recorded at the position of the 
fish?s head within the ABR chamber is shown as the power spectrum in (a). 
  
155 
 
 
Figure 3-2  ? Evoked potential traces at threshold, 5 dB below threshold, and 20 dB 
above threshold to (a) 100 Hz tone burst and (b) 400 Hz from C. venusta.  Plates (c) and 
(d) are traces from the same individual being presented with the same stimulus, but 
after the individual had been sacrificed. 
  
156 
 
 
Figure 3-3 ? Attenuation of semi-trailer truck noise at the crossing of Interstate-85 over 
Uphapee Creek.  The four one-third octave bands above 60 Hz with the highest energy 
are shown.  Attenuation of the 315 and 630 Hz bands was highly irregular due to 
interference from ambient noise.  Theoretical attenuation values for spherical and 
cylindrical spreading are shown for reference. 
  
157 
 
 
Figure 3-4 ? Attenuation of train noise at the CSX crossing of Uphapee Creek.  
Attenuation of all one-third octave frequency bands from 63 to 1000 Hz were included 
in the average.  Error bars represent a standard deviation.  Calculated theoretical 
attenuation values from spherical and cylindrical spreading are presented for reference. 
  
158 
 
 
Figure 3-5 ? Attenuation of actual and speaker playbacks of C. venusta signals.  The solid 
line with standard deviation bars represents the average attenuation of the 13 
frequency bands between 60 and 1000 Hz for knocks and bursts played back by the 
speaker at five sites.  The frequency bands containing the dominant frequencies of 
knocks (80 and 630 Hz) are shown, as well as for growls (80 and 315 Hz).  The calculated 
theoretical attenuation rate based on cylindrical spreading (3 dB/distance doubled) is 
also shown for reference.  Average attenuation across the 13 frequency bands of signals 
actually produced by real fish are shown as empty squares occurring at hydrophone 
separations of 15.24, 30.48, and 45.72 cm.  Measurements of signals from actual fish 
were taken at one distance per site at three different sites. 
  
159 
 
 
Figure 3-6 ? Attenuation of one-third octave bands from speaker playbacks of C. venusta 
signals. 
  
160 
 
 
Figure 3-7 ? Solid line shows average power spectra of bursts under quiet conditions in 
the laboratory.  Other lines represent the average power spectra of semi-trailer truck 
noise and train crossings, and the average level of natural ambient noise at active 
nesting sites.  A bin resolution of 21.5 Hz was used for all power spectra and SNR?s for 
bursts were calculated based on these levels. 
  
161 
 
 
Figure 3-8 - Solid line shows average power spectra of knocks under quiet conditions in 
the laboratory.  Other lines represent the average power spectra of semi-trailer truck 
noise and train crossings, and the average level of natural ambient noise at active 
nesting sites.  A bin resolution of 86.1 Hz was used for all power spectra and SNR?s for 
knocks were calculated based on these levels. 
  
162 
 
 Growls Knocks 
 Dominant Hz Subdominant 
Hz 
Subdominant 
Hz 
Dominant Hz 
Frequency (Hz) 86 301 172.2 602.7 
 
A.A. in Natural 
Noise (m) 
 
0.0299 
 
0.1649 
 
0.5073 
 
0.6519 
 
Distance from 
semi-trailer 
truck noise (m) 
 
117.8 
(142.7) 
 
5.9 
(17.0) 
 
2.0 
(52.8) 
 
--- 
(1.8) 
 
Distance from 
train noise (m) 
 
324.3 
(393.1) 
 
290.4 
(839.5) 
 
10.3 
(269.5) 
 
2.7 
(178.6) 
 
Table 3-1 ? The active area (A.A.) of the dominant and subdominant frequencies of both 
call types from C. venusta in natural noise conditions as measured in 9 active nesting 
sites at Moffits Mill.  Active area is based on a threshold of detection at a signal-to-noise 
ratio of 15 dB.  For the rows containing a set of numbers, the top numbers show the 
closest distance to the noise sources at which the signal frequency component would 
theoretically be detectable (i.e. above a SNR of 15 dB).  Numbers in parentheses show 
the distances at which active areas of the dominant and subdominant frequencies 
should re-establish themselves at levels seen under natural noise conditions due to 
attenuation of truck and train noise.  Attenuation rates of semi-trailer truck and train 
noise are based on spherical spreading, which is a conservative estimate of attenuation 
based on the results of the current study.  Distance from semi-trailer truck noise at 
602.7 Hz is blank because knocks were already above a SNR of 15 dB at the source of the 
noise.