USF Libraries
USF Digital Collections

Standardizing the auditory evoked potential technique

MISSING IMAGE

Material Information

Title:
Standardizing the auditory evoked potential technique ground-truthing against behavioral conditioning in the goldfish carassius auratus
Physical Description:
Book
Language:
English
Creator:
Hill, Randy J
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Auditory brainstem response
Audiogram
Threshold
Masking
Modified staircase method
Dissertations, Academic -- Marine Science -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Auditory evoked potentials (AEPs) have become commonly used to measure hearing thresholds in fish. However, it is uncertain how well AEP thresholds match behavioral hearing thresholds and what effect variability in electrode placement and tank composition has on AEPs. In the first experiment, the effect of testing tank composition and electrode placement on AEPs was determined by recording AEPs in the same individual fish in a steel and PVC cylindrical testing tank, and simultaneously recording AEPs from four locations and two different depths on each of 12 goldfish, Carassius auratus. Results from these studies show that tank composition has an effect AEP strength and hearing thresholds, with steel producing lower thresholds for all frequencies. Electrode placement and depth showed no significant effect on hearing thresholds.In the second experiment, the hearing sensitivity of 12 goldfish was measured using both classical conditioning and AEPs in the same setup.For behavioral conditioning, the fish were trained to reduce their respiration rate in response to a 5s sound paired with a brief shock. Once the behavioral audiogram was completed, the AEP measurements were made without moving the fish. The same sound stimuli were presented and the resultant evoked potentials were recorded for 1,000-6,000 averages. AEP input-output functions were then compared to the behavioral audiogram to compare techniques for estimating behavioral thresholds from AEP data. Results show a large range in variability between behavioral and evoked potential thresholds between fish, with the linear regression evoked potential analysis method producing closer thresholds to behavioral methods. In the third study, the effects of masking were examined on the behavioral and evoked potential audiograms. Behavioral thresholds were first determined with a constant masking noise for two frequencies, followed by threshold measurements with no masking noise.
Thesis:
Thesis (M.S.)--University of South Florida, 2005.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Randy J. Hill.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 174 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001670349
oclc - 62287776
usfldc doi - E14-SFE0001211
usfldc handle - e14.1211
System ID:
SFS0025532:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001670349
003 fts
005 20051216093308.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 051116s2005 flu sbm s000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001211
035
(OCoLC)62287776
SFE0001211
040
FHM
c FHM
049
FHMM
090
GC11.2 (Online)
1 100
Hill, Randy J.
0 245
Standardizing the auditory evoked potential technique
h [electronic resource] :
b ground-truthing against behavioral conditioning in the goldfish carassius auratus /
by Randy J. Hill.
260
[Tampa, Fla.] :
University of South Florida,
2005.
502
Thesis (M.S.)--University of South Florida, 2005.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 174 pages.
520
ABSTRACT: Auditory evoked potentials (AEPs) have become commonly used to measure hearing thresholds in fish. However, it is uncertain how well AEP thresholds match behavioral hearing thresholds and what effect variability in electrode placement and tank composition has on AEPs. In the first experiment, the effect of testing tank composition and electrode placement on AEPs was determined by recording AEPs in the same individual fish in a steel and PVC cylindrical testing tank, and simultaneously recording AEPs from four locations and two different depths on each of 12 goldfish, Carassius auratus. Results from these studies show that tank composition has an effect AEP strength and hearing thresholds, with steel producing lower thresholds for all frequencies. Electrode placement and depth showed no significant effect on hearing thresholds.In the second experiment, the hearing sensitivity of 12 goldfish was measured using both classical conditioning and AEPs in the same setup.For behavioral conditioning, the fish were trained to reduce their respiration rate in response to a 5s sound paired with a brief shock. Once the behavioral audiogram was completed, the AEP measurements were made without moving the fish. The same sound stimuli were presented and the resultant evoked potentials were recorded for 1,000-6,000 averages. AEP input-output functions were then compared to the behavioral audiogram to compare techniques for estimating behavioral thresholds from AEP data. Results show a large range in variability between behavioral and evoked potential thresholds between fish, with the linear regression evoked potential analysis method producing closer thresholds to behavioral methods. In the third study, the effects of masking were examined on the behavioral and evoked potential audiograms. Behavioral thresholds were first determined with a constant masking noise for two frequencies, followed by threshold measurements with no masking noise.
590
Adviser: David A. Mann.
653
Auditory brainstem response.
Audiogram.
Threshold.
Masking.
Modified staircase method.
690
Dissertations, Academic
z USF
x Marine Science
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1211



PAGE 1

Standardizing the Auditory Evoked Potential Technique: Ground-Truthing Against Behavioral Conditioning in the Goldfish, Carassius auratus by Randy J. Hill A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science College of Marine Science University of South Florida Major Professor: David A. Mann, Ph.D. Zhongmin J. Lu, Ph.D. Joseph J. Torres, Ph.D. Date of Approval: May 24, 2005 Keywords: Auditory Brainstem Response, Audiogram, Threshold, Masking, Modified Staircase Method Copyright 2005, Randy J. Hill

PAGE 2

ACKNOWLEDGEMENTS This project would not have been completed if it was not for the numerous individuals who have contributed. First, I would like to thank my advisor, Dr. David Mann. Without his support, guidance, and patience, I would have never been able to succeed in completing this thesis. Secondly, I would like to thank Dr. Jose Torres and Dr. John Lu for serving on my committee as well as providing constructive criticism and sharing ideas on how to make this project better. I would also like to thank my fellow students in the marine sensory ecology lab, especially Bryan Nichols for his digital photo expertise, Brandon Casper and Mandy Hill-Cook for sharing their knowledge about AEP’s, and Jim Locascio for our stimulating statistical discussions as well as making the lab an enjoyable environment to work. Lastly, although most importantly, I would like to thank my wife, Heather, for her love and understanding throughout this project. She proofread and edited everything I wrote, and put up with all my ups and downs that accompany graduate school. The support from my parents has been tremendous as well.

PAGE 3

i TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES v ABSTRACT viii INTRODUCTION 1 Investigations in Fish Hearing 1 Detection of Sound 1 Purpose 6 CHAPTER1: EFFECTS OF TANK COMPOSITION AND ELECTRODE PLACEMENT ON AUDIDTORY EVOKED POTENTIAL MEASUREMENTS IN THE GOLDFISH, CARASSIUS AURATUS 10 ABSTRACT 10 INTRODUCTION 12 MATERIALS & METHODS 14 Experimental Setup 14 Sound Generation and AEP Acquisition 15 Experiment 1: Tank Composition 16 Experiment 2: Recording Electrode Location 16 Experiment 3: Recording Electrode Depth 17 RESULTS 17 Tank Comparison 17 Electrode Location 18 Electrode Depth 19 DISCUSSION 19 Tank Comparison 19 Electrode Location 21 Electrode Depth 23 Conclusions 24

PAGE 4

ii CHAPTER 2: GROUND-TRUTHING EVOKED POTENTIAL THRESHOLDS AGAINST BEHAVIORAL CONDITIONING IN THE GOLDFISH, CARASSIUS AURATUS WITH AND WITHOUT MASKING NOISE 81 ABSTRACT 81 INTRODUCTION 83 MATERIALS & METHODS 86 Experimental Setup 86 Fish Setup 87 Behavioral Conditioning 88 Evoked Potential Acquisition 89 Masking Study 90 RESULTS 91 Behavior vs. Evoked Potential Thresholds 91 Masking vs. No Masking Thresholds 93 DISCUSSION 95 Behavior vs. Evoked Potentials 95 Masking vs. No Masking 97 Conclusion 101 Future Direction 102 LITERATURE CITED 156

PAGE 5

iii LIST OF TABLES Table 1.1 Standard length and weight of fish in tank comparison, electrode placement, and electrode depth studies 26 Table 1.2 Frequencies and sound levels at maximum sound pressure for tank comparison/electrode placement studies 28 Table 1.3 r2 values obtained from linear regressions on EP data in PVC 30 Table 1.4 r2 values obtained from linear regressions on EP data in steel 32 Table 1.5 Average r2 values obtained from linear regressions on EP data from four brain regions 34 Table 1.6 Average r2 values obtained from linear regressions on EP data from two depths 36 Table 1.7 Sound speed and wavelengths for each frequency tested in steel, PVC, and unconfined water 38 Table 2.1 Standard length, weight, and freque ncies tested of fish in behavioral vs. EP threshold/masking comparisons 105 Table 2.2 r2 values obtained through EP linear regression analysis from behavioral vs. EP comparison 107 Table 2.3 Results of Wilcoxon paired-rank test comparing behavioral and evoked potential thresholds 109 Table 2.4 Results of two-tailed paired-sample T-test comparing behavioral thresholds with and without masking 111 Table 2.5 Results of two-tailed paired-sample T-test comparing evoked potential thresholds with and without masking 113 Table 2.6 Results of two-tailed paired-sample T-test comparing behavioral and evoked potential thresholds without masking 115 Table 2.7 Results of two-tailed paired-sample T-test comparing behavioral and evoked potential thresholds with masking 117

PAGE 6

iv Table 2.8 Amount of time needed to determine evoked potential thresholds at various signal presentation averages 119

PAGE 7

v LIST OF FIGURES Figure 1 Audiograms of six fish species 3 Figure 2 Relationship between sensory epithelium and otolith 5 Figure 3 Diagram of goldfish Weberian Ossicles 8 Figure 1.1 Diagram of the AEP recording setup 40 Figure 1.2 Sound stimulus used throughout evoked potential recordings 42 Figure 1.3 AEP waveforms in the time and frequency domains 44 Figure 1.4 Three-point linear regression for EP threshold determination 46 Figure 1.5 Schematic of internal goldfish brain 48 Figure 1.6 Internal and external images of the goldfish brain labeled with four tested brain regions 50 Figure 1.7 Average evoked potential comparisons between tanks 52 Figure 1.8 Average audiogram for PVC and steel 56 Figure 1.9 Average threshold differences between steel and PVC 58 Figure 1.10 Evoked potential comparison between brain regions for one fish at 150 Hz and 2400 Hz 60 Figure 1.11 Average evoked potential diffe rences for different brain regions relative to medulla 62 Figure 1.12 Average audiogram comparing four brain regions 66 Figure 1.13 Average threshold differences between brain regions relative to the medulla 68 Figure 1.14 Evoked potential comparison of electrode depths for one fish at 150 Hz and 2400 Hz 70

PAGE 8

vi Figure 1.15 Average evoked potential differences between electrode depths 72 Figure 1.16 Average audiogram comparing electrode depths 76 Figure 1.17 Average threshold differences between electrode depths 78 Figure 1.18 Comparison of steel and PVC thresholds with threshold results obtained in a plastic test tank 80 Figure 2.1 Diagram of behavioral conditioning and EP setup 121 Figure 2.2 Diagram of electric shock setup for behavioral conditioning 123 Figure 2.3 Respiratory rate for control pe riod and test period showing untrained and trained fish 125 Figure 2.4 Histograms showing behavioral threshold criterion 127 Figure 2.5 Modified staircase method for determining behavioral thresholds 129 Figure 2.6 AEP waveform in the time and frequency domain 131 Figure 2.7 Three-point linear regression for EP threshold determination 133 Figure 2.8 Average audiogram comparing behavioral thresholds, visual EP thresholds, and linear regression EP thresholds 135 Figure 2.9 Average differences between threshold techniques 137 Figure 2.10 Evoked potential input-output f unctions with behavioral thresholds for two fish at 600 Hz 139 Figure 2.11 Evoked potentials generated from one fish at 150 Hz with 100, 1000, and 5000 signal averages 141 Figure 2.12 Number of signal averages re quired to bring EP at behavioral threshold to noise floor 143 Figure 2.13 Average noise floor level at each frequency 145 Figure 2.14 Average behavioral and EP thresholds for two frequencies with and without masking 147 Figure 2.15 EP input-output functions with be havioral thresholds for one fish at 150 Hz with and without masking 149

PAGE 9

vii Figure 2.16 EP input-output functions with be havioral thresholds for one fish at 600 Hz with and without masking 151 Figure 2.17 Comparison of behavioral a nd EP audiograms with audiograms from other studies 153 Figure 2.18 EP audiograms with and without masking compared with results from another study 155

PAGE 10

viii Standardizing the Auditory Evoked Po tential Technique: Ground-Truthing Against Behavioral Conditioning in the Goldfish, Carassius auratus Randy J. Hill ABSTRACT Auditory evoked potentials (AEPs) have become commonly used to measure hearing thresholds in fish. However, it is uncertain how well AEP thresholds match behavioral hearing thresholds and what effect variability in electrode placement and tank composition has on AEPs. In the first experiment, the effect of testing tank composition and electrode placement on AEPs was determined by recording AEPs in the same individual fish in a steel and PVC cylindri cal testing tank, and simultaneously recording AEPs from four locations and two di fferent depths on each of 12 goldfish, Carassius auratus Results from these studies show that tank composition has an effect AEP strength and hearing thresholds, with steel producing lower thresholds for all frequencies. Electrode placement and depth showed no si gnificant effect on hearing thresholds. In the second experiment, the hearing sensitivity of 12 goldfish was measured using both classical conditioning and AEPs in the same setup. For behavioral conditioning, the fish were trained to reduce their respiration rate in response to a 5s sound paired with a brief shock. Once the be havioral audiogram was completed, the AEP measurements were made without moving the fish. The same sound stimuli were presented and the resultant evoked potentials were recorded for 1,000-6,000 averages. AEP input-output functions were then compared to the behavioral audiogram to compare techniques for estimating behavioral thresholds from AEP data. Results show a large range in variability between behavioral and evoked potential thresholds between fish, with the linear regression evoked potential analysis method producing closer thresholds to behavioral methods. In the third study, the effects of masking were examined on the behavioral and evoked potential audiograms. Behavioral thresholds were first determined with a

PAGE 11

ix constant masking noise for two frequencies, followed by threshold measurements with no masking noise. After behavioral conditioni ng, evoked potentials were conducted without moving the fish, first with masking and then without masking. Results show that masking has a larger effect on the behavioral audiogram than on evoked potentials, and at 600 Hz, the masking evoked potential threshold is significantly lower than the behavioral masking threshold.

PAGE 12

1 INTRODUCTION Investigations in Fish Hearing The auditory system is one of the most vital sensory systems for aquatic animals because it provides a wealth of information regarding prey, predation, competition, and locating potential mates through the use of acoustic signals in the environment (Myrberg, 1978). Determining fish hearing thresholds, therefore, is vital to assess how natural and anthropogenic noise affect fish hearing. Hearing in fishes was first established in the early part of the 20th century in cyprinids (Parker, 1903; Bigelow, 1904). Heari ng in different families of fish, as well as investigations into the range of frequencies, was performed a short time after by von Frisch and colleagues (von Frisch, 1936). Through the use of various techniques, fish audiograms have been obtained on approximately 70 of the over 25,000 extant fish species (Chapman and Sand, 1974; Hawkins and Johnstone, 1978; Myrberg and Spires, 1980; Fay, 1988, Popper and Carlson, 1998)(Figure 1). Detection of Sound Sound detection in fishes involves the inner ear, and depending on the species, structures peripheral to the ear that enhance sound detection (Popper and Lu, 2000). Three otolithic organs, the utricle, saccule, and lagena are used as acoustic receptors. Each of these organs contains an otolith, which is a dense, calcareous structure. The wall of the chamber in each otolithic organ contains a sensory epithelium called the macula. The macula is separated from the otolith by a thin, otolithic membrane that is connected to each structure, which holds them in place relative to each other. The remaining portion is filled with fluid (Rogers et al., 1988)(figure 2). The macula contains hair cells, which are transducers of acoustic information (Popper and Carlson, 1998; Popper and Lu, 2000). Each macula may contain tens or

PAGE 13

2 Figure 1 Audiogram demonstrating the wide range of hearing abilities between species. Carassius auratus (goldfish) is a hearing specialist (Jacobs and Tavolga, 1967), Salmo Salar (Atlantic salmon; Hawkins and Johnstone, 1978), Gadus morhua (Atlantic cod; Chapman and Hawkins, 1973), Stegastes leucostictus (beau-gregory; Myrberg and Spires, 1980), and Euthynnus affinis (kawakawa, Iversen, 1969) do not possess hearing specializations and are considered hearing generalists. Alosa sapidissima (American Shad) is considered a hearing specialist due to it’s broad hearing range (to over 180 kHz) but has poor sensitivity at lower frequencies compared to hearing specialists such as Carassius (Mann et al., 1998) (Redrawn from Popper and Lu, 2000).

PAGE 14

3 40 60 80 100 120 140 160 10100100010000100000 Frequency (Hz)Threshold (dB re 1 Pa) Carassius Gadus Salmo Stegastes Euthynnus Alosa

PAGE 15

4 Figure 2 Schematic illustration of the relationship between the sensory epithelium and the overlying otolith. The ciliary bundles from the sensory hair cells inundate the lumen of the otolithic chamber and contact, or come close to contacting, the otolith. A thin otolithic membrane separates, and connects, the otolith and the sensory epithelium (redrawn from Popper and Lu, 2000).

PAGE 16

5

PAGE 17

6 even hundreds of thousands of sensory cells, and as the fish grows, the number of hair cells increases (Corwin, 1977; Popper and Hoxer, 1984; Lombarte and Popper, 1994). Each hair cell has an apical ciliary bundle containing a single kinocillium and more than 40 sterocillia. The otolith is at least three times more dense than the fish’s body, which causes the otolith to move at a different amplitude and phase from the sensory epithelium. This bends the ciliary bundles located on the sensory hair cells, which generates a receptor potential in the hair cell and excites the neurons of the eighth nerve (Roberts et al., 1988). Fishes have been categorized into two groups dependent upon the connection or proximity of the swimbladder to the ear (Popper and Platt, 1993, Yan et al., 2000). Fishes with the best hearing sensitivity are called hearing specialists, and they possess specializations that acoustically couple the swimbladder (or other gas bubble) to the inner ear, and can detect sound both directly (sound hits the otolith organ directly) and indirectly (sound resonates off the swimbladder and hits the inner ear)(Fay, 1988). The acoustic coupling may involve special bones, such as the Weberian ossicle in otophysan fishes (Figure 3), or rostral projections of the swimbladder that end at the inner ear, such as in soldierfish (Popper and Lu, 2000). Hearing generalists primarily hear via direct stimulation and lack a swimbladder-ear connection. Hearing generalists typically have poorer hearing sensitivity than hearing specialists, and most fish species fall into this category (Yan et al., 2000). Purpose The auditory evoked potential (AEP) method is an electrophysiological technique used to measure hearing thresholds in fishes and other vertebrates (Kenyon et al., 1998). The intent of measuring nerve impulses (evoked potentials) generated by the eighth nerve in response to a sound stimulus is to obtain behavioral conditioning thresholds in lieu of extensive training of the individual. However, inconsistencies in evoked potential measurements between studies may result in different threshold levels for the same species, and may not represent the behavioral audiogram. Behavioral and evokedpotential hearing thresholds have been obtained for the same species (e.g. Yan and Popper, 1991; Kenyon et al., 1998) but never on the same individuals in the same setup.

PAGE 18

7 Figure 3 Diagram illustrating the connection of the swimbladder to the inner ear in otophysan fishes via Weberian Ossicles (modified from Chranilov, 1927).

PAGE 19

8

PAGE 20

9 To determine if behavioral thresholds can be predicted from evoked potential measurements, behavioral and evoked potential thresholds must be conducted on the same individual in the same laboratory set-up. Aside from inconsistencies between setups and procedures, a majority of evoked potential measurements are conducted in quiet environments, which does not reflect the noisy habitats where fish occur. The purpose of this study was to 1) measure evoked potential strength and audiograms in goldfish in two different test tanks and with different recording electrode positions and de pths and 2) to compare behavioral and evoked potential thresholds in the same fish with and without the presence of a masking noise. Different evoked potential analysis methods were compared to determine which technique results in thresholds closer to those measured with behavioral conditioning. Based on these results, the number of signal averages required to bring the evoked potential at behavioral threshold to the noise floor were determined.

PAGE 21

10 Chapter 1: Effects of tank composition a nd electrode placement on auditory evoked potential measurements in the goldfish, Carassius auratus Randy J. Hill ABSTRACT Auditory evoked potentials (AEPs) have recently been used to determine fish hearing thresholds in lieu of behavioral thresholds. This method allows for rapid threshold determination and can be used on uncooperative or inattentive subjects that can not be trained behaviorally. However, inconsistencies in laboratory setups between studies can potentially cause discrepancies in hearing thresholds, such as different testing tanks and variations in the position of the recording electrode. In the first study, evoked potential strengths and hearing thresholds were measured for 12 goldfish, Carassius auratus in a PVC tank and a steel tank at five different frequencies. The recording electrode was not removed from the fish dur ing transfer between tanks, and the speaker and water depths were the same for each tank. Average evoked potential levels were greatest in steel except at 600 Hz. The greatest difference was at 2400 Hz, the highest frequency tested. Most of the fish produced lower hearing thresholds in steel for all frequencies, with a mean difference ranging between 0.7 and 18 dB. In the second study, simultaneous recordings from four electrodes placed above the telencephalon, optic tectum, cerebellum, and medulla were measured at five frequencies. At 150 Hz, average evoked potential strengths were strongest in the medulla and became weaker towards the rostral parts of the brain. For 300 and 600 Hz, evoked potential strengths were slightly stronger over the optic tectum than the medulla, and for 1200 Hz and 2400 Hz, the strongest signals were recorded over the optic tectum and telecephalon respectively. Despite these differences, thresholds from each region were very similar with the largest mean difference between locations being slightly over 1 dB. In the third study, two electrodes of varying lengths were inserted simultaneously over the medulla region to determine the effects of depth on evoked potential strength.

PAGE 22

11 Average evoked potential strengths were stronger for the deeper electrode at all frequencies, but thresholds were not statistically different, with the largest mean difference between electrode thresholds of 0.6 dB. In summary, tank composition can significantly affect hearing threshold measurements, whereas electrode placement has little influence.

PAGE 23

12 INTRODUCTION Auditory evoked potentials (AEPs) have become a common method to measure fish hearing thresholds in lieu of behavioral training. The AEP is an electrophysiological technique for measuring hearing thresholds in fishes and other vertebrates (Kenyon et al., 1998). Electrodes placed cutaneously or inserted subdermally in proximity to the organism’s brainstem directly measure neural impulses created in the eighth nerve and brain in response to sounds (Corwin et al., 1982). Evoked potential signals are extracted from background noise by averaging the evoked potentials from repeated signal presentations. This allows for rapid hearing measurements without the need for training. However, the effect of tank composition and recording electrode placement on AEP and threshold calculations is not known. Ideally, hearing measurements should be conducted in an open body of water with a depth and width exceeding the wavelength of the targeted sound to minimize distortions due to reverberations. However, environmental parameters (e.g. water temperature, ambient noise) can not be controlled in field situations, and turbidity can hinder behavioral observations (Akamatsu et al., 2002). The advantages to using a small tank are the increased ability to control environmental factors and yield precise behavioral observations. The composition of the testing chamber can play a large role in whether the sound stimulus accurately represents the natural environment, and therefore produces accurate hearing thresholds that can be extrapolated to natural situations. Various testing tank compositions have been used, such as glass (Akatmatsu et al, 2002), concrete (Akatmatsu et al., 2002), plastic (Yan, 2001), and PVC (Egner and Mann, 2005). Various linings, such as horse hair, fibers, and sand (Tavolga, personal communications), have also been used in an attempt to minimize reverberations. A material and tank size that would permit sound to travel 1500 m/s, which would simulate the natural environment, would be ideal. If the material is not very stiff relative to water, the tank walls may move in the presence of the sound field, which will dissipate acoustic energy and hence slow the wave propagation speed, resulting in a shorter wavelength than might occur in a

PAGE 24

13 natural environment. This could potentially affect hearing measurements in fishes, like goldfish, that rely on gasbladders to convert acoustic pressure to particle motion. Recording electrode location and depth within the brain is another variable that may result in discrepancies in hearing threshold measurements between studies. The sense of hearing in fishes is mediated by the inner ear, which is located in the cranial cavity approximately at the level of the medulla (Popper and Carlson, 1998). The inner ear consists of three otolithic organs: the saccula, utricle, and lagena. In the sleeper goby the utricle has been found to be largely vestibular in that it is less sensitive than the saccule or lagena (Lu and Xu, 2002). The saccule and lagena lie closer to the medulla region, with the utricle positioned more rostrally in the brain. The eighth nerve innervates the otolithic organs and transmits auditory information to the brain for processing (Fay and Popper, 1974). In most vertebrates, the auditory fibers terminate within various parts of the brain, including the cerebellum, reticular formation, and the octaval column in the medulla. The ocataval column is the main auditory region involved in transmitting the input to more frontal parts of the brain, and in fishes the nerves from all of the octaval endorgans primarily end in the octaval column (Butler and Hodos, 1996). Due to the location of the inner ear and termination of the auditory nerves in the octaval column, the typical recording electrode location when measuring evoked potentials is over the medulla region of the brain (e.g. Kenyon, 1998; Ladich and Yan, 1998; Yan, 2001; Egner a nd Mann, 2005; Kojima et al., 2005), although there is no set standard. Auditory nerves extend from the octaval column to more frontal parts of the brain, and therefore, other brain locations should produce evoked potentials in response to a sound stimulus. Lu and Xu (2002) were the first to reco rd evoked potentials from multiple areas of the brain, and to compare ABR responses with a recording electrode placed on the skull versus in the fluid of the brain cavity. They found the amplitude of ABR responses depend on electrode location, and the amplitude of the responses were significantly smaller on the skull relative to the brain cavity measurements. This was only conducted for one frequency and level, and thus audiograms were not calculated. Evoked potential levels and hearing thresholds may also vary with electrode depth as well. Previous studies have positioned the recording electrode on the exterior of the fish above the

PAGE 25

14 midline (e.g. Kenyon et al., 1998; Higgs et al., 2003), subdermally in the fish (e.g. Mann et al., 2001; Egner and Mann, 2005), and in the brain cavity (Lu and Xu, 2002). Three sets of experiments were performed. In the first study, the EP’s of goldfish were measured in a PVC tank and a steel tank to measure the effects of tank composition on evoked potential levels and thresholds calculated from the EP’s. In these experiments, the same individual was measured in each tank, without changing the electrode configuration. In the second study, the effect of electrode placement along the brain were measured from goldfish in a steel tank. In the third study, the effect of electrode depth above the medulla was measured with two electrodes. In the second and third studies, recordings from each electrode were made simultaneously, which eliminates the effects of variations in the acoustic field on measurements of variability at each electrode position. MATERIALS AND METHODS Experimental setup Three separate experiments were conducted in this study to determine the effect of different variables on EP measurements: 1) tank comparison, 2) location of EP recording electrode and 3) depth of reco rding electrode. For each experiment, 12 different goldfish were utilized (table 1.1) Goldfish were obtained from a local aquarium fish supplier. Animals were maintained in a 29-gallon filtered aquarium at 25 + 1 C and were fed one pinch of commercially prepared goldfish flakes daily (Tetramin). Once the fish were tested, they were either euthanized or placed in a separate tank and used in following experiments. All procedures were approved by the University of South Florida Institutional Animal Care and Use Committee. Evoked potential measurements were determined by securing an individual fish in a harness made from vinyl mesh fastened with clamps and suspe nded from laboratory stands. Harnesses were adjustable to fit each animal, allowing for uninhibited respiration and minimal movement.

PAGE 26

15 Sound generation and AEP acquisition Sound stimuli were produced by an AEP workstation (Tucker-Davis Technologies TDT) using TDT SigGen and BioSig software through an RP2.1 Enhanced Real-Time Processor and a PA5 Programmable Attenuator. The signal was then amplified by a Hafler Trans.Ana P1000 110 Watt Professional Power Amplifier sent to a UW-30 (University Sound) speaker (figure 1.1). Sound stimuli consisted of 50 ms tones shaped with a Hanning window. Sounds were presented 13 times per second (figure 1.2). The phase of each presentation of the tone stimulus was alternated to reduce electrical artifacts in the recorded signals. The sound stimuli were calibrated using a Reson hydrophone (sensitivity –212 dB V/1 Pa) connected to a Reson VP1000 Voltage Preamplifier with a high-pass filter of 5 Hz and 32 dB gain. The hydrophone was positioned in the experimental setup in place of the fish, and lowered to the same water depth (15 cm below the surface) for calibration. Subdermal stainless steel needle electrodes (Rochester Electro-Medical) were used for recording the evoked potential signal. Evoked potentials recorded were fed through an RA16 Medusa Base Station and aver aged by BioSig software (figure 1.1). The reference electrode was placed within the fish’s dorsal musculature and the ground electrode was placed directly into the water in proximity to the fish (figure 1.1). Hearing thresholds were determined for five frequencies (150 Hz, 300 Hz, 600 Hz, 1200 Hz, 2400 Hz) and each fish was presented with the maximum sound pressure level for each frequency, with the sound level decreasing in 6dB steps until 90 dB attenuation of the loudest sound level was achieved (table 1.2). A discrete Fourier transform (DFT; using MATLAB) of all AEP waveforms was calculated and analyzed for the presence of significant peaks, which were defined as peaks at twice the frequency of the sound stimulus 3 dB above the background noise within a 20 Hz window of the dominant frequency (figure 1.3). A three-point linear regression was performed using the lowest detected evoked potential value and the evoked potentials measured at the two previous louder sound pressure levels (SPL). The SPL where the regression crossed zero volts was defined as the hearing threshold (figure 1.4).

PAGE 27

16 Experiment 1: Tank Comparison The testing chambers consisted of a PVC tube (1.2 m high, 30 cm in diameter, 0.4 cm. thickness), closed at the bottom, and oriented vertically, and a steel tube (1.22m high, 20.32 cm. in diameter, 0.9525 cm. thickness), closed at the bottom with a square steel plate (60.96 cm x 60.96 cm), and oriented vertically. The PVC tank was placed within a sand-filled cylindrical PVC container for s upport and four anti-vibration floor mounts (Tech Products Corp 51700 series) were placed under the base of the steel tank. The PVC and steel tanks were both filled with fresh water at approximately 26 C up to a height of 111.76 cm. Both tanks were in proximity to a table vhich held the laboratory stands. The testing chambers were located in an audiology booth. The recording electrode was inserted 1 mm into the head, over the medulla region. Each fish was lowered to a depth of 15 cm in each tank. Up to 200 signal presentations were averaged from each tank to obtain the evoked potential at each frequency and sound level. Once all five frequencies and sound levels were tested in one tank, the fish was immediately transferred to the other tank without removing the electrodes, and the schedule was repeated. Si x fish were tested in PVC first, and the other six fish were first tested in steel. A Wilcoxon paired-sample rank test (alpha=0.05) was used to test whether there were significant differences between hearing thresholds measured in each tank. Experiment 2: Recording Electrode Location Four recording electrodes were placed subdermally using surface landmarks to record simultaneous evoked potentials from each brain region to a sound stimulus. The four locations for recording electrodes were above the medulla in the hindbrain region of the fish, over the cerebellum located in the hindbrain, over the optic tectum in the midbrain area, and over the telecephalon in the forebrain (figures 1.5 and 1.6). Each electrode was inserted subcutaneously one millimeter. Electrical tape on the electrode shaft was used to keep electrode depths accurate between electrodes and fish. Each fish was tested in the steel tank and lowered to a depth of 15 cm. Signal presentations were averaged up to 2000 times to obtain evoked potentials at each frequency and sound level.

PAGE 28

17 Evoked potential measurements generated from the four brain regions were analyzed with MATLAB. A Friedman repeated measures ANOVA on ranks (alpha = 0.05) was used to determine if there was a significant difference in thresholds between brain areas. If a difference existed, a Tukey multiple comparison test was used to determine which brain areas were different. Experiment 3: Recording Electrode Depth Two recording electrodes were glued together with 0.7 mm spacing between them using epoxy. One electrode extended 2mm farther down than the other. The electrode pair was inserted subcutaneously over the medulla region directly behind the cerebellum, with the longer electrode extending to a depth of 2.5 mm, and the shorter to a depth of 0.5 mm. Electrical tape on the electrode shaft was used to keep electrode depths accurate between electrodes and fish. Each fish was test ed in the steel tank and lowered to a depth of 15 cm. Signal presentations were averaged up to 2000 times to obtain evoked potentials at each frequency and sound level. Evoked potential measurements generated from the two electrode depths were analyzed with MATLAB. A Wilcoxon paired-sample rank test (alpha = 0.05) was used to determine if hearing thresholds determined with the linear regression technique varied significantly between electrodes. RESULTS AEP waveforms indicate that as the sound level of the stimulus decreased, the amplitude of the AEP waveforms decreased. The dominant frequency in the power spectra was approximately twice the stimulus frequency (figure 1.3). Tank Comparison A comparison of average evoked potential measurements from each testing tank (figure 1.7) demonstrated that fish generally produce a stronger evoked potential in the steel tank than in the PVC tank at 150 Hz, 300 Hz, and 2400 Hz. At 600 Hz fish in the PVC tank produced stronger signals at higher SPLs. At 1200 Hz, the evoked potentials measured in the two tanks were almost identical

PAGE 29

18 The coefficients of determination from the linear regressions used to calculate thresholds were generally high: 92% of r2 values generated from fish tested in PVC were greater than 0.9, 88% of r2 values from steel were greater than 0.9 (table 1.3 and 1.4). Mean hearing thresholds for all fish in the two tanks suggest that fish have lower hearing thresholds when tested in the steel tank for all frequencies tested except at 1200 Hz, where fish tested in PVC had a lower mean threshold by less than 1 dB, which is not statistically significant (figure 1.8). There wa s a large amount of variation between fish in the magnitude of the difference between thresholds measured in steel versus PVC especially at 300 and 2400 Hz (figure 1.9). However, most fish had lower hearing thresholds in the steel tank than in the PVC tank. Three of the frequencies tested (150, 600, and 2400) showed a statistically significant difference in thresholds between te sting tanks with an alpha level of 0.05. The other two frequencies (300 and 1200) were not significantly different (Wilcoxon pairedsample rank test) (figure 1.9). Electrode Location The EP input-output functions show variation in EP levels recorded from the different brain regions that is frequenc y dependent (figure 1.10). Low frequencies showed the medulla producing the strongest evoked potentials, with signals weakening with distance away from the medulla (figure 1.11). At the two highest frequencies tested, however, an opposite effect occurred, where the medulla produced weaker evoked potential strengths than the other three regions (figure 1.11). At 1200 Hz, the optic tectum produced the strongest evoked potentials and at 2400 Hz, the telencephalon showed the strongest evoked potential levels (figure 1.11). Variations in EP levels also appeared to be sound pressure level dependent. Average hearing thresholds calculated for each of the electrode positions showed only a small difference between brain regions (figure 1.12). High r2 values were obtained from each linear regression performed (table 1.5) (88% from all channels were greater than 0.9). The response of each brain area relative to the medulla showed the greatest difference between the telecephalon and medulla region at 2400 Hz (approximately 1 dB) (figure 1.13). The medulla produced lower average thresholds at three of five frequencies

PAGE 30

19 (300, 1200, and 2400 Hz). Statistical comparison of hearing thresholds showed no significant difference between brain regi ons (Friedman Repeated Measures ANOVA on Ranks). Electrode Depth The EP magnitude was generally greater for the deeper electrode in comparison to the shallow electrode (figure 1.14 and 1.15). Each frequency produced relatively similar magnitude differences, although they were level dependent (figure 1.15). High r2 values were obtainted from each linear regression performed (table 1.6) (87% of total r2 values for depth test were greater than 0.9). The evoked potential audiograms calculated from the EPs at each electrode depth were similar (figure 1.16). The largest average difference between electrodes was at 150 Hz, with a difference of only 0.64 dB (figure 1.17). Comparison of hearing thresholds measured at the two recording depths showed no statistically significant difference (Wilcoxon paired-sample rank test). DISCUSSION Tank Comparison Previous studies using auditory evoked potentials have used various tank compositions, sizes, and shapes. However, until now no study has tested the same individual fish in different tanks to determine their effects on hearing threshold measurements. The data from this study suggest that different tanks can produce statistically different hearing threshold results at speci fic frequencies, with steel producing lower hearing thresholds than PVC (see figure 1.9). The large amount of variability in hearing threshold differences in each of the tanks between different fish suggests that even though a majority of fish showed lower hearing thresholds in steel, the magnitude of differences in hearing thresholds can greatly vary between tanks and fish. Goldfish audiograms collected from previous studies (Enger, 1966; Fay, 1969; Jacobs and Tavolga, 1967; Offutt, 1968; Poppe r, 1971; Sawa, 1976; Weiss, 1966) show a tremendous amount of variability, with differences of over 60 dB in thresholds at some frequencies. Experiments by Popper et al., ( 1973) suggested that these differences were

PAGE 31

20 not due to the different conditional techniques employed, but were likely caused by different acoustic conditions under which the experiments were performed, as well as the degree of masking by background noise. Goldfish, which are hearing specialists, use their swimbladder to pick up the pressure component of a passing sound wave and couple it to the otolithic organs to enhance hearing abilities (Yan, 2001). The pressure component is converted to particle velocity via the vibrating swimbladder. Thus the wavelength of sound could affect how the swimbladder responds to a sound. To estimate the speed of sound in a tank one can determine the tank’s effective bulk modulus, which is a measure of how much the tank will compress under a given amount of external pressure. The bulk modulus of the material is the ratio of the change in pressure to the fractional volume compression (Junger and Feit, 1993). If the bulk modulus is low, much of the sound wave energy will be absorbed by the tank, lowering sound speed. To determine the bulk modulus of the steel and PVC tanks, the following equation was used: e = /1 + (2 /E)(ro 2 + ri 2/ro 2 ri 2) + where is the bulk modulus of the fluid, E is the elastic modulus, is Poisson’s ratio of the pipe material, and ro and ri are the outer and inner radius of the testing tank (Junger and Feit, 1993). Table 1.7 shows the estimated speed of sound and the resultant wavelength for each frequency tested in the two tanks and unconfined water. Steel tanks will produce wavelengths closer to open water than PVC tanks. Even though hearing thresholds should not be compared between studies to test the effects of tank composition, bandwidth, shape of the audiogram, and frequency of best sensitivity can be compared between laboratories and testing tanks (Higgs et al., 2002). The shape of the audiograms produced in the steel and plastic tanks are very similar (Kenyon et al. 1998) (figure 1.18), with the best hearing sensitivity in the plastic and steel being 600 Hz. PVC did not have the same audiogram shape, and best sensitivity was achieved at 1200 Hz. Although steel produced lower hearing thresholds and sound characteristics closer to that of open water, it is impossible to say whether hearing thresholds obtained in the steel more closely represent actual hearing thresholds in natural environments. Obtaining hearing thresholds in open water and directly comparing the results can only determine this.

PAGE 32

21 Electrode Location Fish hearing studies using the evoked potential method generally place the recording electrode externally over the medulla region. Data from this study suggests that different electrode positions and depths do have an effect on the evoked potential level, but do not have an appreciable effect on the calculated hearing thresholds. The Weberian ossicles, used to enhance hearing abilities in otariophysans, connect the anterior end of the swimbladder to the perilymph-filled transverse canal that leads directly into the sacculus of the inner ear (von Frisch, 1936). This more intimate relationship with the sacculus over the lagena and utriculus is the likely reason that the sacculus has been most frequently identified as the major acoustic organ (Popper and Coombs, 1980). The location of the inner ear in fishes suggests that the location of strongest evoked potentials would be the medulla region, where the eighth nerve enters the brain. This trend appears to be the case at the lower frequencies, but an opposite effect occurs at the higher frequencies, where the electrode over the medulla produces the weakest signals. Lu and Xu (2002) mapped evoked potential strengths in 21 different brain locations in the sleeper goby, Dormitator latifrons and found that ABR amplitudes were significantly different between recording sites, and the ABR responses with intact crania tended to decrease when the recording site varied from the anterior to the posterior regions of the brain. This suggests that the medulla region produced the weakest signals at this frequency. When the cranium was exposed, the amplitudes remained relatively stable until they reached the posterior end of the brain, where they declined. This was conducted at 500Hz, and the results in the sleeper goby are similar to the results from this study on goldfish at 1200 and 2400 Hz. However, the closest frequency tested to the Lu study was 600 Hz, and at this frequency, the results did not coincide with the 500 Hz results from their study. This difference could potentially be due to species differences, especially because the sleeper goby does not have a swimbladder-inner ear connection. Results from the low frequencies in this study correspond with Corwin et al. (1982) showing the difference in amplitudes between the telecephalon and medulla in the leopard frog, Rana pipiens Corwin et al (1982) showed the decrease of evoked potential amplitudes measured from the telencephalon region relative to the medulla.

PAGE 33

22 Although the frequency response in various parts of the fish brain have never been analyzed, the ability of fish to discriminate between different frequencies, and the mechanisms for frequency coding in the inner ear, have been extensively studied (e.g. Stetter, 1929; Dijkgraaf and Berheijen, 1950; Enger, 1963; Furukawa and Ishii, 1967; Jacobs and Tavolga, 1968; Fay, 1970a; Sand, 1974; Fay, 1978a; Fay, 1981; Fay and Passow, 1982; Moeng and Popper, 1984; Platt and Popper, 1984). Possible mechanisms for frequency analysis by fishes include temporal analysis, which may involve converting a sound stimulus into a spike rate or sequence of spike intervals related to the temporal nature of the sound stimulus (Fay, 1981; Fay and Passow, 1982), the length of ciliary bundles in hair cells, with different lengths responding to different frequencies (Saunders and D ear, 1983), and frequency-to-spatial mapping, where different regions of the saccular macula may respond to different frequencies. This method is analogous to the place method in mammals, where frequency analysis is a matter of the central processing of information coded by the cochlea as spatial patterns of activity across large fibers arrays (reviewed in Lewis et al., 1985). Although a mechanism analogous to the tonotopically organized structures in other vertebrates have not been found in fish, there is some evidence that the response characteristics of different regions of the saccule and lagena may vary in several species (Enger, 1981; Sand and Michelse n, 1978; Cox et al., 1987). Sand (1974) demonstrated that microphonics recorded from the perch sacculus show frequency response functions that vary with location of the electrode along the macula. The type of fibers found in different regions of the otolithic organs, in particular the sacculus, have been associated with frequency discrimination abilities of fish. Furukawa and Ishii (1967) associated high fre quency units with large diameter fibers of the anterior sacculus and lower frequency sensitive units with small diameter fibers, found in both the anterior and posterior sacculus. The presence of four types of neural units in the sculpin was demonstrated (Enger, 1963), some of which displayed a following response to the acoustic stimulus. He concluded frequency discrimination takes place in part by a following response (place theory) and partly by a separation into low and high frequency sensitive units (volley theory).

PAGE 34

23 For fishes that use swim bladders to aid in detecting the pressure components of sound, it has also been suggested that different modes of input to the ear (the lowfrequency inertial route and high-frequency sw imbladder route) may result in different patterns of otolith movement, allowing frequency discrimination (Fay, 1981). The frequency variation in evoked potentia l strengths generated from different regions of the brain found in this study indi cate that frequency discrimination may not occur solely in the inner ear, but may take place in auditory fibers generated from the eighth nerve in the brainstem that transmit information throughout the brain. The results indicate the possibility of tonotopical organization in the brain. The evoked potential results obtained from various regions of the brain indicate the difference between brain regions varies with sound pressure (see figure 1.12). The largest magnitude differences between brain locations at 300 Hz occurs at the quieter sound pressure levels, whereas at 600 Hz and 1200 Hz the largest difference occurs at the louder sound pressure levels. The differences between locations at 150 Hz and 2400 Hz are somewhat uniform throughout the tested sound pressure levels. The source of this variability is not known, but could be studied with single unit recordings in different brain areas. Electrode Depth The results obtained with different depth electrodes over the medulla are typical of what would be expected. The deeper el ectrode was closer to the eighth nerve and brain which generates the evoked potential, therefore producing a stronger evoked potential signal for all frequencies tested (figur e 1.16). A majority of fish hearing studies that utilized the AEP method placed the recording electrode on the exterior of the fish above the medulla (e.g. Kenyon et al., 1998; Ladich and Yan, 1998; Yan, 2001; Mann et al, 2001; Lu and Tomchik, 2002;Yan, 2002; Hi ggs et al., 2003; Wysocki and Lacich, 2003). Lu and Xu (2002) found that the amplitude of the ABR responses recorded with the recording electrode placed on the skull was significantly smaller than that recorded with the electrode placed in the fluid of the brain cavity. This difference in amplitudes could potentially have an effect on hearing thresholds obtained from these studies.

PAGE 35

24 Corwin et al. (1982) found that the AEPs can be recorded without drilling a hole in the skull, either from subcutaneous needles or the outer surface of the skin if good electrical contact can be made. They conclude d that in aquatic animals, AEPs could be recorded from the water some millimeters from the animal. This may be possible, but according to the study conducted by Lu and Xu (2002), it would produce a weaker signal. It must be noted that electrodes placed externally on the skull were not examined in this study. These electrodes could have possibly produced hearing thresholds that were significantly different from the obtained results. CONCLUSIONS The results of the tank study show that tank composition can have a significant effect on measured fish hearing thresholds. To minimize the negative effects of small tank acoustics on threshold measurement, a tank composition that can mimic the natural environment is ideal. Although evoked potential strengths vary with frequency and sound pressure level, hearing thresholds generated from the four different regions of the brain and two different electrode depths were almost id entical, with no significant difference between locations (see figure 1.13). Thus, the placement and depth of electrode can be arbitrary, and can be eliminated as a potential cause of variation in threshold differences between studies. The frequency-specific and sound pressure level effects on evoked potential strength in different regions of the brain is une xpected and should be investigated further.

PAGE 36

25 Table 1.1 Standard length and weight of the 12 fish used in the tank comparison, electrode location, and electrode depth studies.

PAGE 37

26 Fish Standard Length (cm) Weight (g) Tank Comparison Electrode Placement Electrode Depth Tank Comparison Electrode Placement Electrode Depth 1 5.2 4.2 5.4 3.49 2.73 4.67 2 5.2 5.8 5.8 3.61 5.43 5.92 3 5.2 5.5 6.6 4.34 4.65 7.58 4 5.3 5.8 5.7 4.37 4.57 5.28 5 5.4 6.2 6.1 4.40 6.55 6.93 6 5.4 5.7 5.1 4.65 4.66 4.29 7 5.5 6.4 5.3 4.85 8.01 4.48 8 5.5 5.6 5.6 5.38 5.36 4.71 9 5.6 5.4 4.8 5.38 5.39 3.95 10 5.9 5.6 4.9 5.79 5.62 3.22 11 6.1 5.6 4.6 5.90 5.08 2.69 12 6.0 6.3 4.3 7.02 6.80 2.37 Mean 5.5 5.7 5.4 4.93 5.40 4.67 Standard Deviation 0.32 0.57 0.66 1.01 1.32 1.58

PAGE 38

27 Table 1.2 Frequencies and sound levels at maximum sound pressure for the tank comparison/electrode placement test.

PAGE 39

28 Frequency (Hz) Maximum sound level for PVC (dB) Maximum sound level for steel (dB) 150 180 166 300 174 162 600 146 175 1200 135 169 2400 139 164

PAGE 40

29 Table 1.3 r2 values obtained from linear regression on evoked potential data for all fish and all frequencies in PVC.

PAGE 41

30 Fish 150 Hz 300 Hz 600 Hz 1200 Hz 2400 Hz 1 0.9993 0.9793 0.9600 0.9894 0.9974 2 0.9977 0.8250 0.9804 0.9715 0.9807 3 0.9464 0.9635 0.9992 0.9584 0.9607 4 0.9468 0.9602 0.9941 0.9695 0.9846 5 0.9998 0.9699 0.8288 0.9564 0.9627 6 0.9950 0.9994 0.9304 0.9667 0.9056 7 0.9998 0.9919 0.8960 0.9780 0.9514 8 0.9369 0.9995 0.9991 0.9498 0.9977 9 0.9825 0.9286 0.9909 0.9708 0.8840 10 0.9725 0.9108 0.7977 0.9115 0.9901 11 0.9991 0.9422 0.9644 0.9644 0.9822 12 0.9968 0.9950 0.9822 0.9687 0.9673 Mean 0.9811 0.9554 0.9436 0.9629 0.9637

PAGE 42

31 Table 1.4 r2 values obtained from linear regression on evoked potential data for all fish and all frequencies in steel.

PAGE 43

32 Fish 150 Hz 300 Hz 600 Hz 1200 Hz 2400 Hz 1 0.8003 0.8565 0.9823 0.9211 0.8672 2 0.8601 0.9767 0.9983 0.9800 1.0000 3 0.9994 0.9990 0.7885 0.9193 0.9535 4 0.8817 0.9562 0.9683 0.9152 0.9786 5 0.8201 0.9331 0.9663 0.9563 0.9899 6 0.9051 0.9862 0.9210 0.9275 0.8827 7 0.9847 0.9532 0.9932 0.9037 0.9364 8 0.8332 0.9795 0.7569 0.9584 0.9799 9 0.9209 0.8688 1.0000 0.9774 0.9811 10 0.9694 0.8577 0.9563 0.9463 0.9899 11 0.9935 0.9103 0.9995 0.9250 0.9923 12 0.9957 0.9481 0.9567 0.9834 0.9000 Mean 0.9137 0.9384 0.9406 0.9428 0.9543

PAGE 44

33 Table 1.5 Average r2 obtained through linear regression threshold determination for four brain regions at all frequencies (n=12).

PAGE 45

34 150 Hz 300 Hz 600 Hz 1200 Hz 2400 Hz Telecephalon 0.9503 0.9304 0.9790 0.9417 0.9453 Optic Tectum 0.9495 0.9387 0.9750 0.9399 0.9434 Cerebellum 0.9525 0.9296 0.9700 0.9365 0.9478 Medulla 0.9483 0.9310 0.9626 0.9305 0.9443

PAGE 46

35 Table 1.6 Average r2 values obtained through linear regression threshold determination for electrode depths at all frequencies (n=12).

PAGE 47

36 150 Hz 300 Hz 600 Hz 1200 Hz 2400 Hz Shallow 0.9234 0.9118 0.9795 0.9471 0.9567 Deep 0.9196 0.9350 0.9852 0.9493 0.9435

PAGE 48

37 Table 1.7 Estimated sound speed and resultant wavelengths for each frequency tested in the PVC tank, steel tank, and unconfined water. Freshwater sound speed obtained from Clay & Medwin (1977).

PAGE 49

38 Wavelength Material Sound Speed (m/s) 150 300 600 1200 2400 PVC 247.2877 m/s 1.65 m 0.82 m 0.41 m 0.21 m 0.10 m Steel 1329.588 m/s 8.86 m 4.43 m 2.22 m 1.11 m 0.55 m Fresh H20 1438 m/s 9.59 m 4.79 m 2.40 m 1.20 m 0.60 m

PAGE 50

39 Figure 1.1 Diagram of the EP-recording setup ( RP2.1 Enhanced Real-Time Processor, PA5 programmable attenuator, P1000 110 Watt Professional Power Amplifier, RA16 Medusa Amplifier, RA16 Medusa Base Station, REC recording electrode, REF reference electrode, GRO ground electrode).

PAGE 51

40

PAGE 52

41 Figure 1.2 Recordings of sound stimuli used in the evoked potential recordings (A) 300 Hz sound stimulus for one second, showing approximately 13 pulsed tones (B) 150 Hz tone (C) 300 Hz tone (D) 600 Hz tone (E) 1200 Hz tone (F) 2400 Hz tone .

PAGE 53

42 A B C D E F

PAGE 54

43 Figure 1.3 Example of an ABR response from fish #1 when played a tone at 600 Hz A) Response in time domain. B) Response in frequency domain (DFT). Vertical bar indicates 1 V.

PAGE 55

44 A B

PAGE 56

45 Figure 1.4 Linear regression on the evoked potential generated at the visual SPL threshold and two previous measurements from fish #4 at 2400 Hz. The point where the linear regression crosses 0 defined as the hearing threshold. Arrow indicates evoked potential at visual threshold. r2 = 1.0

PAGE 57

46 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 8090100110120130140150160170180SPL (dB re 1 uPa)Evoked Potential ( V)

PAGE 58

47 Figure 1.5 Diagram of the goldfish brain indicating the four recording locations for this study. OB = olfactory bulb, T = telecephal on, Cb = cerebellum, H = hindbrain, D = delecephalon (from Butler & Hodos, 1996).

PAGE 59

48

PAGE 60

49 F igure 1.6 (A) Internal image of the goldfish brain: A telecephalon; B optic tectum; C Cerebellum; D vagel lobes (directly above the medulla) (B) External image of the goldfish showing brain locations.

PAGE 61

50 A B

PAGE 62

51 Figure 1.7 Evoked potential comparisons (mean SD) between tanks for the five frequencies tested (n=12).

PAGE 63

52 150 Hz-170 -160 -150 -140 -130 -120 -110 -100 -90 8090100110120130140150160170SPL (dB re 1 Pa)Evoked Potential (dB V) Steel PVC 300Hz -160 -150 -140 -130 -120 -110 -100 -90 8090100110120130140150160170SPL (dB re 1 Pa)Evoked Potentail (dB V) Steel PVC

PAGE 64

53 600 Hz-160 -150 -140 -130 -120 -110 -100 8090100110120130140150SPL (dB re 1 Pa)Evoked Potential (dB V) steel PVC 1200 Hz-165 -155 -145 -135 -125 -115 -105 -95 708090100110120130140SPL (dB re 1 Pa)Evoked Potential (dB V) steel PVC

PAGE 65

54 2400 Hz-190 -180 -170 -160 -150 -140 -130 -120 708090100110120130140150SPL (dB re 1 Pa)Evoked Potential (dB V) steel PVC

PAGE 66

55 Figure 1.8 Average audiogram for all fish at each tested frequency for PVC and steel (n = 12).

PAGE 67

56 75 80 85 90 95 100 105 110 115 120 05001000150020002500Frequency (Hz)SPL (dB re 1 Pa) Steel PVC

PAGE 68

57 Figure 1.9 Differences between steel and PVC (mean SD) for each fish at each frequency. Positive values indicate fish measured in steel have lower hearing thresholds than when measured in PVC (PVC – steel). Asterisks indicate values significantly different with alpha =0.05 (n=12).

PAGE 69

58 -15 -10 -5 0 5 10 15 20 25 30 05001000150020002500300 0 Frequency (Hz)SPL Difference (dB re 1 Pa)* *

PAGE 70

59 Figure 1.10 Evoked potential comparisons for one fish for all four brain regions at 150 Hz and 2400 Hz.

PAGE 71

60 150 Hz-165 -155 -145 -135 -125 -115 -105 -95 7090110130150170190SPL (dB re 1 Pa)Evoked Potential (dB V) Telecephalon Optic Tectum Cerebellum Medulla 2400 Hz-185 -175 -165 -155 -145 -135 -125 -115 7090110130150170SPL (dB re 1 Pa)Evoked Potential (dB V) Telecephalon Optic Tectum Cerebellum Medulla

PAGE 72

61 Figure 1.11 Evoked potential strengths from the telecephalon, optic tectum, and cerebellum relative to the medulla region (mean SD) for five frequencies tested. Positive values denote weaker evoked potentials relative to the medulla. (n=12).

PAGE 73

62 150 Hz-2 -1 0 1 2 3 4 7090110130150170190SPL (dB re 1 Pa)EP difference (uV) Medulla Telecephalon Medulla Optic Tectum Medulla Cerebellum 300Hz -3 -2 -1 0 1 2 3 4 5 6 7 6080100120140160180SPL (dB re 1 Pa)EP difference (uV) medulla telecephalon medulla optic tectum medulla Cerebellum

PAGE 74

63 600Hz -3 -2 -1 0 1 2 3 4 5 7090110130150170SPL (dB re 1 Pa)EP difference (uV) Medulla Telecephalon Medulla Optic Tectum Medulla Cerebellum 1200Hz -4 -3 -2 -1 0 1 2 3 4 7090110130150170190SPL (dB re 1 Pa)EP difference (uV) Medulla Telecephalon Medulla Optic Tectum Medulla Cerebellum

PAGE 75

64 2400Hz-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 6080100120140160180SPL (dB re 1 Pa)EP difference (uV) Medulla Telecephalon Medulla Optic Tectum Medulla Cerebellum

PAGE 76

65 Figure 1.12 Average audiogram from four brain locations for all fish tested (n=12).

PAGE 77

66 75 80 85 90 95 100 05001000150020002500Frequency (Hz)SPL (dB re 1 Pa) Telecephalon Optic Tectum Cerebellum Medulla

PAGE 78

67 Figure 1.13 Threshold differences in brain regions (mean SD) relative to the medulla. Positive values denote a higher threshold than that obtained from the medulla (n = 12).

PAGE 79

68 -6 -4 -2 0 2 4 6 050010001500200025003000Frequency (Hz)Threshold difference (dB re 1 Pa) Telecephalon Medulla Optic Tectum Medulla Cerebellum Medulla

PAGE 80

69 Figure 1.14 Evoked potential comparisons for one fish at two electrode depths at 150 Hz and 2400 Hz.

PAGE 81

70 150 Hz-160 -150 -140 -130 -120 -110 -100 6080100120140160180SPL (dB re 1 Pa)Evoked Potential (dB V) Shallow Deep 2400 Hz-200 -190 -180 -170 -160 -150 -140 -130 -120 -110 -100 6080100120140160180SPL (dB re 1 Pa)Evoked Potential (dB V) Shallow Deep

PAGE 82

71 Figure 1.15 Within-individual evoked potential strength differences from the shallow electrode relative to the deep electrode (mean SD). Positive values denote weaker evoked potentials relative to the deep electrode (n=12).

PAGE 83

72 150 Hz-5 -4 -3 -2 -1 0 1 2 3 4 708090100110120130140150160170SPL (dB re 1 Pa)EP difference (dB V) 300 Hz-6 -4 -2 0 2 4 6 8 10 12 6080100120140160SPL (dB re 1 Pa)EP Difference (dB V)

PAGE 84

73 600 Hz-5 -4 -3 -2 -1 0 1 2 3 4 5 8090100110120130140150160170180SPL (dB re 1 Pa)EP Difference (dB V) 1200 Hz-3 -2 -1 0 1 2 3 4 5 6 7090110130150170SPL (dB re 1 Pa)EP Difference (dB V)

PAGE 85

74 2400 Hz-8 -6 -4 -2 0 2 4 6 708090100110120130140150160170SPL (dB re 1 Pa)EP Difference (dB V)

PAGE 86

75 Figure 1.16 Average audiograms at each electrode depth for all fish tested (n=12).

PAGE 87

76 75 77 79 81 83 85 87 89 91 93 95 05001000150020002500Frequency (Hz)SPL (dB re 1 Pa) Shallow Deep

PAGE 88

77 Figure 1.17 Threshold differences (mean SD) in electrode depth (shallow-deep) Positive values denote a lower threshold obtained with the deeper electrode (n = 12).

PAGE 89

78 -4 -3 -2 -1 0 1 2 3 4 05001000150020002500Frequency (Hz)Threshold Difference (dB re 1 Pa) Shallow-Deep

PAGE 90

79 Figure 1.18 Audiogram showing results from this experiment (n = 12) and goldfish hearing thresholds obtained in a plastic tub (Kenyon, 1998). From the Kenyon experiment (n = 4)

PAGE 91

80 60 70 80 90 100 110 120 05001000150020002500Frequency (Hz)SPL (dB re 1 Pa) Kenyon et al. Steel PVC

PAGE 92

81 Chapter 2: Ground-truthing evoked potential thresholds against behavioral conditioning in the goldfish, Carassius auratus with and without masking noise Randy J. Hill ABSTRACT Auditory evoked potentials (AEPs) are commonly used to measure hearing thresholds in fishes and other animals. However, it is uncertain how well AEP thresholds match behavioral hearing thresholds, and if obtaining evoked potential measurements in a noisy background can affect threshold levels. The hearing sensitivity of 11 goldfish, Carassius auratus were measured using both classical conditioning and AEPs in the same setup at five frequencies. For behavi oral conditioning, fish were trained to reduce their respiration rate in response to a 5s sound presentation paired with a brief shock. A modified staircase method was utilized to determine hearing thresholds, and the last 12 reversals were averaged to calculate thresholds. AEP measurements were made immediately following behavioral conditioning without moving the fish. The two most common AEP analysis techniques (visual inspection and linear regression to 0 V) were used to estimate AEP thresholds. Comparison of these thresholds with behavioral thresholds indicates that the regression techni que produces thresholds closer to behavioral thresholds. Behavioral methods produced lower hearing thresholds, on average, than AEP methods, especially at lower frequencies. However, there is a large amount of variability between methods at all frequencies, with behavioral versus visual inspection producing variations in standard deviations between 9 and 21 dB and behavioral and linear regressions producing variations between 9 and 27 dB. The same procedure was also conducted using six goldfish with and without the presence of a broadband background noise. Fish were tested at two frequencies (150 Hz and 600 Hz) with classical conditioning with the masking noise present and then in quiet. Immediately after behavioral conditioning, AEP measurements were conducted at the same two frequencies with and without th e masking noise without moving the fish.

PAGE 93

82 Behavioral thresholds increased proportionally to the increase in background noise. AEP measurements showed a masking effect, however, the magnitude of the masking effect was much less than measured with behavioral techniques. In fact, at 600 Hz, the evoked potentials produced significantly lower hearing thresholds than the behavioral technique.

PAGE 94

83 INTRODUCTION Fish audiograms have been measured for over 50 species, using either behavioral or electrophysiological methods. Behavioral methods include instrumental avoidance conditioning, where a fish is trained to cross a barrier when a sound is detected to avoid an electric shock (e.g. Horner et al. 1961; Behrend and Bitterman, 1962; Tavolga and Wodinsky, 1963; Popper, 1971), operant conditioning which utilizes positive reinforcement upon pecking a paddle in response to a sound stimulus (Yan and Popper, 1991,1992,1993), and classical conditioning, which involves pairing a sound stimulus with a mild electrical shock and measuring avoidance responses (Myrberg and Spires, 1980), cardiac suppression (Fay, 1969; Chapman and Sand, 1974), or ventilatory suppression (Banner, 1967). Another widely used method of evaluating fish auditory thresholds is to measure the extracellular microphonics from the auditory organs while playing an acoustic stimulus (Saidel and Popper, 1987) or using single-unit recordings to measure single nerve fiber discharge patterns (Enger and Anderson, 1967). Behavioral conditioning and the use of microphonics have been widely used, but have some serious drawbacks: it takes an extensive amount of time to condition the fish to the sound and physiological stress may result from the electric shock (Kojima et al., 2005). Auditory evoked potentials (AEPs) have been extensively used in studies of mammalian audition, and only recently have been utilized to measure fish hearing (Kenyon et al., 1998; Yan et al., 2000; Yan and Curtsinger, 2000; Mann et al., 2001; Scholik and Yan, 2001,2002; Lu and Tomchik, 2002; Wysocki and Ladich, 2002,2003; Akamatsu et al., 2002; Egner and Mann, 2005; Kojima et al., 2005). AEPs are noninvasive far-field recordings of nerve impulses generated in the eighth nerve and brain in response to an acoustic stimulus (Jewett, 1970; Jewett and Williston, 1971; Jacobson, 1985; Kenyon et al., 1998). AEPs have several advantages over behavioral and singleunit methods: thresholds can be determined from animals that cannot be trained, and there is the potential for repeated use of the same animal (Kenyon et al., 1998; Yan and

PAGE 95

84 Popper, 1991). AEPs usually yield higher thresholds than behavioral techniques (Katz, 1994) but no study has ever tested the same animal in the same setup using both methods. Behavioral methods to measure fish hearing thresholds are generally considered to be the most reliable method as they indicate the animal’s perception of the sound (Fay, 1988). The often unstated purpose of AEPs is to estimate the behavioral audiogram from the evoked potential data. However, there are several variables, such as the number of signal averages and methods of evoked potential threshold calculation, in AEP data collection and analysis that may produce evoked potential audiograms that significantly vary from the behavioral hearing thresholds. AEP measurements are made by presenting an acoustic stimulus repeatedly and averaging the resultant evoked potentials. Averaging the evoked potentials reduces the level of uncorrelated noise from all sources (including neural and electronic) in the evoked potential measurement as a function of the square root of the number of averages. This is a powerful method to detect weak neural signals in background noise. However, this could also be one of the biggest sources of inconsistency in evoked potential studies. The number of evoked potential sweeps that are averaged is usually arbitrary, and is based on balancing the amount of time needed to conduct the measurement with obtaining a reasonable threshold measurement. The major criteria for stopping the averaging process are 1) when a response is detected, 2) when there appears not to be a response, 3) when the subject appears too noisy to continue, and 4) when the maximum number of sweeps allocated for a run have been acquired (Don and Elberling, 1996). When measuring the evoke d potential, it is impossible to determine whether averaging more EP sweeps would dr aw an evoked potential from the background noise. Thresholds from evoked potential techniques have been derived from either the amplitude of the evoked potential or latency measurements in evoked potential peaks. The easiest way to estimate the threshold is visually determining the lowest sound level for which an evoked potential response can still be detected and differentiated from background noise (Higgs et al., 2003). This method is potentially variable and depends on the experience of the observer, amount of residual noise, and usually requires repeated measures (Don and Elberling, 1996). Modifications of the visual determination technique include estimates of the variance of the averaged signal compared to the

PAGE 96

85 variance of the background noise (Don et al., 1984), comparing two different trials to determine if the evoked potential is repeated (Kenyon et al., 1998), and transforming the AEP waveform to the frequency domain with a FFT (Fast Fourier Transform) and visually analyzing for the presence of significant peaks at twice the frequency of the stimulus, known as frequency doubling (Egner & Mann, 2005). However, this method is also arbitrary in that a specific signal to noise ratio must be specified in order to determine whether a peak is present. A second technique to calculate AEP thresholds is to plot the input-output function (evoked potential strength vs. sound pressure level) and extrapolate to where the evoked potential would reach 0 volts, assuming a linear relationship between sound level and evoked potential amplitude (Ridgeway et al., 1981; Eberling and Don, 1987; Popov and Supin, 1990). Behavioral audiograms and AEP generated evoked potentials have been conducted on the same species, such as goldfish, Carassius auratus (ex. Yan and Popper, 1991; Kenyon et al., 1998), oscar, Astrontus ocellatus (Kenyon, 1998), American Shad, Alosa sapidosomma (Mann et al., 1997,1998,2001), harbor seal, Phoca vitulina (Wolski et al., 2003), killer whale, Orcinus orca (Szymanski et al., 1999), and little skate, Raja erinacea (Casper et al., 2003). However, none of the measurements with fishes were conducted on the same individuals in the same laboratory setups. Different individuals may have varying hearing abilities, and therefore comparing the two methods from different fish may include variability from other sources in addition to the technique being used. This study had two main objectives. First, conduct behavioral and evoked potential measurements on the same individual to determine if behavioral audiograms can be predicted from evoked potential measurements, and which method of evoked potential analysis is more accurate in predicting behavioral thresholds. Second, compare behavioral and evoked potential thresholds in the same individual with and without the presence of a background masking noise to determine the effects of masking on evoked potentials and the behavioral audiogram.

PAGE 97

86 MATERIALS AND METHODS Experimental setup Two experiments were conducted in this study: 1) comparison of behavioral and EP thresholds and 2) comparison of thresholds with and without masking noise. For threshold comparison, 11 different goldfish were used, and for the masking comparison, six goldfish were used (table 2.1). Goldfish were obtained from a local aquarium fish supplier. Animals were maintained in a 29-gallon filtered aquarium at 25 + 1 C and were fed one pinch of commercially prepared goldfish flakes daily (Tetramin). Once the fish were tested, they were either euthanized or placed in a separate tank and used in subsequent experiments. All procedures were approved by the University of South Florida Institutional Animal Care and Use Committee. The test tank was a cylindrical steel tube (1.22m high, 20.32 cm in diameter, 0.9525 cm thickness) closed at the bottom with a square steel plate (60.96 cm by 60.96 cm) and oriented vertically. Four antivibration floor mounts (Tech Products Corp, 51700 series) were placed under each corner of the base of the tank. The tank was filled with fresh water at approximately 26 C up to a height of 111.76 cm. The tank was located in proximity to a table that held the suspended fish from laboratory stands. The testing tank was located in an audiology booth. Sound stimuli used to obtain hearing thresholds for evoked potentials and behavioral conditioning were produced by an AEP workstation (Tucker-DavisTechnologies – TDT). The sound was generate d by TDT SigGen software and presented by TDT BioSig software through an RP2.1 Enhanced Real-Time processor and a PA5 programmable attenuator. The signal was then amplified by a Hafler Trans.Ana P1000 110 Watt Power Amplifier before reaching the fish through a speaker located at the bottom of the testing tank (University Sound UW30) (figure 2.1). Sound stimuli consisted of 50 ms tones shaped with a (5 ms) cosine squared window. Sounds were presented 13 times/second. For AEP measurements the tone stimulus was presented in alternating phase to reduce electrical artifacts. For behavioral measurements the same tone signal was used (50 ms in duration repeated 13 times per second), however the tone pip train was 5 s in duration.

PAGE 98

87 The sound stimulus was calibrated using a Reson hydrophone (sensitivity –212 dB V/1 Pa) connected to a Reson VP1000 Voltage Preamplifier with a high-pass filter of 5 Hz and 32 dB gain. The hydrophone was positioned in the experimental setup in place of the fish, and lowered to the same water depth (15 cm below the surface). Calibration was conducted each time a fish was tested. The ambient noise in the tank was also calibrated without the presence of sound stimulus using an HTI 96-min hydrophone (sensitivity –164 dBV/ Pa, 20 Hz to 32 kHz) connected to a RP2.1 processor which sent the signal to custom software for recording. The hydrophone was positioned in the experimental setup in place of the fish, and lowered to the same water depth (15 cm below the surface). Fish Setup Each individual fish was secured in a harness made from vinyl mesh fastened with clamps and suspended from laboratory stands. Harnesses were adjustable to fit each animal, allowing for uninhibited respiration and minimal movement. This same setup was used for both procedures. A stainless steel needle electrode (Roche ster Electro-Medical) was glued to the binder clip holding the fish so that the tip was placed in between the operculum and the gills to measure respiration. The signal from this electrode was amplified with a DB4HS4 bioamp headstage (TDT) with a lowpass filter at 40Hz and 70,000x amplification. Two alligator clips were located on either side of the fish and connected to an AC supply with a solid state relay to deliver the shock. A recording electrode to measure evoked potentials was inserted into the fish’s head directly over the medulla region to a depth of 1mm. Inserting the recording electrode prior to behavioral conditioning allowed the recording of evoked potentials directly following behavioral conditioning wit hout moving, and thus disturbing, the fish. The reference electrode was inserted into the dorsal musculature and the ground electrode was located in the water in close proximity to the fish. Electrodes were insulated with enamel except for the tip to reduce electrical artifacts. Once the setup was complete, the fish was submerged 15 cm below the water surface.

PAGE 99

88 Behavioral Conditioning Behavioral thresholds in 11 fish were determined by means of a modified staircase method measuring reduction in respiration upon detecting a sound stimulus. Recording of the respiration rate and delivery of the shock were conducted with a custom MATLAB program. This program measured ten-seconds of a control respiration, followed by recording five seconds of control respiration and five seconds of respiration during the sound stimulus. Immediately following the five-second sound stimulus, a 50 millisecond AC electric shock was delivered via alligator clips using a variable autotransformer (Staco Energy Product Co.) (f igure 2.2) The lowest voltage that initiated respiratory suppression was used (4-8V). Suppression of respiration was determined by comparing the RMS amplitude of the respiratory signal during the sound stimulus with that obtained during the five-second pre-stim ulus control period (figure 2.3) If the amplitude during the stimulus decreased by at least 0.9 it was counted as a detection. Trials were presented randomly with 60-120 seconds between trials. A threshold criterion was established by testing three different fish with and without the presence of a sound stimuli and electric shock at one frequency and sound level (600 Hz, 135 dB). A total of 717 control respiratory rates were calculated, where two five-second control periods were compared. Of these 717 trials, 519 were used to measure control and test respiration, with a sound and electric shock. Ratios between control periods (figure 2.4 A) and between the control and test period (figure 2.4 B) were compared to determine a threshold criterion in which a majority of the control periods were above, and the test ratios were below the criterion. A ratio of 0.9 was selected which resulted in correct rejection of 93% of the control trials, and correct detection of 90% of the sound trials. Thus, this should produce approximately 7% false positives and 10% false negatives. Before behavioral thresholds were determined, each fish was trained to suppress its respiration upon hearing a sound stimulus. The training stimulus consisted of a pulsed tone played at the same frequency and sound pressure level for each trial (150 Hz, 122 dB). After the five-second sound stimulus, a 50 millisecond AC electric shock was administered. The fish was said to be trained when it suppressed its respiration at or below the criterion for a detection for five c onsecutive trials. Once the fish was trained, it

PAGE 100

89 remained in the test setup at the exact location, and threshold measurements were conducted. Behavioral thresholds were determined using a modified staircase method (e.g. Jacobs and Tavolga, 1968; Mann et al., 1998). Each frequency tested began at 40 dB attenuation from the loudest sound pressure level that was generated. If a detection occurred on the first trial, the second trial decreased in sound level by 6 dB. If the detection did not occur on the first trial, the second trial increased in sound level by 6 dB. After the first eight reversals, amplitude changes were made in 3 dB steps. Trials were continued until 20 reversals, alternations between detection and no detection, occurred. The threshold was determined as the average of the last twelve amplitudes where reversals occurred (figure 2.5). Evoked Potential Acquisition Immediately following behavioral conditi oning, evoked potentials were measured without disturbing the fish. The same sound stimulus and presentation of sound was used for both procedures. Evoked potential measurements recorded by the electrode were fed through an RA16 Medusa Amplifier (TDT) to the RA16 Medusa Base Station, routed into the computer and averaged by BioSig software. The number of signal presentations averaged to measure the evoked response at each level of each frequency varied between fish from between 1,000 and 6,000 averages. At louder sound pressure levels the test conditions were advanced once an obvious EP was present. Sound level at each frequency was decreased in 6 dB steps until 120 dB attenuation was achieved. Two different methods were utilized to calculate evoked potential thresholds. Hearing thresholds were first determined by calculating power spectra with a discrete Fourier transform (DFT) for all AEP waveforms and analyzed for the presence of significant peaks (peaks at twice the frequency of the stimulus that were at least 3 dB above background levels) within a 20 Hz window of the dominant frequency (figure 2.6 A & B). Analysis of significant peaks was done using the visual inspection, which is the traditional means of determining evoked potential thresholds (e.g. Kenyon et al., 1998; Wysocki and Ladich, 2003; Yan, 2002). The last sound pressure level where a significant peak could be identified was considered the threshold.

PAGE 101

90 Hearing thresholds were also determined by performing a three-point linear regression on the evoked potential value obtained at the visual threshold and the evoked potentials generated at the two previous sound pressure levels. The sound pressure level where the regression crossed 0 V was calculated (figure 2.7). Thresholds for the masking experiment were determined with the same methodology. Differences between behavioral thresholds and evoked potential thresholds obtained visually and by linear regression were compared using a Wilcoxon pairedsample test with an alpha level of 0.05. To determine the approximate number of signal presentation averages that would be needed for the evoked potential at the behavioral threshold to be at the noise floor, a three-point regression was first calculated on the evoked potential (in dBV) at the visual threshold and two previous evoked potential measurements. The difference between the noise floor (calculated from the last 20 ms of the EP waveform from the last six sound pressure levels) and the behavioral evoked potential value were used to calculate the number of averages needed. Sound pressure levels for the ambient background noise and for the masking background noise were determined by performing a 4028-point Fast Fourier Transform (FFT) on five separate recordings, and taking an average FFT of the recordings. Masking study To determine the behavioral thresholds with and without the presence of masking noise, six fish were tested using the same experimental set-up and procedures for the behavioral conditioning threshold determinations mentioned above. During masking, a continuous broadband background noise was presented to the fish out of the same loudspeaker with an SM5 signal mixer. The sound pressure level of the masking noise at the location of the fish was meausured with an HTI 96-min hydrophone (sensitivity –164 dBV/ Pa, 20 Hz to 32 kHz) connected to a RP2.1 processor. Once fish were trained as described above the behavioral hearing thresholds with masking noise were obtained for two frequencies (150 Hz and 600 Hz). After the masking behavioral thresholds were determined, the masking noise was disconnected and the procedure repeated. Once behavioral thresholds were determined, evoked potential

PAGE 102

91 measurements were conducted as described above with the same two frequencies without moving the fish, first with the masking noise and then without the noise. Differences between behavioral cond itioning with and without masking noise, evoked potential thresholds with and without masking noise, behavioral and evoked potential thresholds with masking noise, and behavioral and evoked potential thresholds without masking noise were compared using two-tailed paired-sample t tests. Due to repetition of statistical tests on the same subjects, a Bonferroni correction was used to determine significant values. The alpha level was set at 0.05, so taking the Bonferroni correction into account (0.05/4), values were considered significant when p<0.0125. RESULTS Behavioral vs. EP thresholds Behavioral conditioning training t ook approximately 10-15 trials. RMS amplitude of respiration generally decreased well beyond the established criterion ratio of 0.9. AEP waveforms in the time (SPL/ms) and frequency domain (DFT) show that as the sound level of the stimulus decreased, the amplitude of the AEP waveforms decreased (figure 2.6). The dominant frequency in the power spectra was approximately twice the stimulus frequency. Comparison of audiograms generated from behavioral conditioning, visual evoked potential analysis, and linear regression evoked potential analysis (figure 2.8) suggests behavioral conditioning produces lower hearing thresholds than evoked potential measurements at all frequencies tested with the exception of 1200 Hz, in which evoked potential measurements via linear regression produce lower thresholds. The standard error bars indicate that at the lower frequencies (150 Hz and 300 Hz) the deviation of the mean does not overly with the mean deviations of evoked potential measurements, with the distribution of means for behavioral always being lower in sound pressure level. However, at the other three frequencies (600 Hz, 1200 Hz, and 2400 Hz) there is overlap of standard errors, suggesting a wide range in variability in estimated means of the population with the three different threshold measurements. This corresponds with individual fish data, in which all ten fish tested at 150 Hz produced lower behavioral thresholds, 9 of 10 fish pr oduced lower behavioral thresholds at 300 Hz,

PAGE 103

92 six of 11 had lower behavioral thresholds at 600 Hz, 4 of 10 had lower behavioral thresholds at 1200 Hz, and 3 out of 4 fish ha d lower behavioral thresholds at 2400 Hz. The average difference between methods for each individual fish was plotted to account for differences in hearing thresholds between fish (figure 2.9). High r2 values (93% of total r2 values were greater than 0.9)(table 2.2) obtained via linear regression threshold method for each frequency suggest the visual threshold typically lies on a linear portion of the evoked potential curve. The audiogram and threshold differences with standard error and standard deviations show that the largest difference between behavioral and evoked potential thresholds measured with both techniques occurs at the lowest frequency, and gets smaller as the frequencies move higher, with an exception at 2400 Hz. They also demonstrate that measuring evoked potentials using the linear regression technique produces threshold values that are closer to behavioral thresholds at all frequencies tested except for 1200 Hz. However, large variations in threshold differences between these two methods exist as well between individual fish. Behavioral thresholds were also plotted on the evoked potential input-output function to determine if behavioral thresholds can be predicted from the curve (figure 2.10). When this was conducted on several different fish in the same frequency, the behavioral threshold varied tremendously from being located in the middle of the linear portion of the curve to being located in the noise floor. This variation in behavioral threshold location on the evoked potential curve indicates the invalidity of predicting the behavioral threshold based on the evoked potential curve. When statistically comparing thresholds obtained behaviorally with visual evoked potentials using the Wilcoxon paired-sample test (table 2.3), three of the four frequencies analyzed were significantly different from each other with an alpha level of 0.05. Only 1200 Hz produced thresholds that were not si gnificantly different. Differences between thresholds were not tested at 2400 Hz due to the small sample size at this frequency. Differences between thresholds obtained behaviorally and linear regression evoked potentials were also compared using the same statistical method (table 2.3) and show that at the two lowest frequencies tested, 150 Hz and 300 Hz, there is a statistically significant difference between obtained thresholds with an alpha level of 0.05. However,

PAGE 104

93 at the next two higher frequencies, 600 Hz and 1200 Hz, there was no difference in thresholds. Visual observation of the differences in threshold methods along with statistical results show that overall evoked potentials measured with the linear regressions are closer to the behavioral thresholds than visual detection of evoked potential thresholds. Conducting different numbers of signal averages on the same individual demonstrates how the noise floor is lowered with an increase in averages (figure 2.11). The approximate number of averages that would be needed to bring the evoked potential at behavioral threshold to the noise floor (figure 2.12) suggests that the mean number of signal presentation averages is extremely large at the lower frequencies, and decreases with an increase in frequency. The amount of variability between fish is also the largest at 150 Hz, and decreases as frequency increases. The average noise floor level at each frequency (figure 2.13) coincides with the number of averages needed, showing the noise floor to be louder at lower frequencies and decreasing towards higher frequencies. This suggests that more averages would be needed at lower frequencies to extract the evoked potential signal from background noise since the noise floor has a higher sound pressure level. Masking vs. No Masking An audiogram of mean hearing thresholds for six fish tested at two frequencies (150 Hz and 600 Hz) using behavioral and linear regression evoked potential techniques with and without the presence of a background masking noise was constructed (figure 2.14). The audiogram also includes the background noise during non-masking and masking threshold recordings. Average behavioral thresholds without masking were much lower than behavioral thresholds with masking, with a difference of approximately 24 dB for 150 Hz and 27 dB for 600 Hz. This difference is equivalent to the difference in the noise floor level between the ambient noise level without any sound and background levels with masking noise. Behavioral and evoked potential thresholds without the masking noise followed the same trend as seen when comparing evoked potentials and behavioral thresholds, with the largest difference at 150 Hz and the difference getting smaller with an increase

PAGE 105

94 in frequency to 600 Hz. This same trend was seen when comparing evoked potential average threshold measurements with and without the masking noise, showing thresholds with no masking producing lower thresholds at both frequencies and the difference decreasing from the lower to the higher frequency. However, the difference between these two measurements are much smaller than the difference between behavioral thresholds with and without masking noise, with a difference of approximately 8 dB at 150 Hz and 4 dB at 600 Hz. When comparing masking behavior and masking evoked potential thresholds, masking behavior produces lower threshol ds at 150 Hz, but only by approximately 4 dB. However, at 600 Hz, evoked potential measurements are much lower than behavioral thresholds, with an average difference of approximately 19 dB. This suggests that measuring hearing thresholds at 600 Hz are much lower when measuring evoked potentials compared with behavioral thresholds when masking noise is presented at the level shown in the audiogram. The standard error bars on the audiogram indicate that the lower evoked potential thresholds at 600 Hz with masking occurred in all six fish tested, and this can be seen when behavioral thresholds are plotted on the evoked potential input-output function for each frequency with and without the presence of background masking noise (figures 2.15 and 2.16). At 600 Hz, the behavioral thresholds fall towards the lower portion of linear part of the curve and into the noise floor without the masking noise. When masking noise is presented, the behavioral threshold lies in the middle of the linear portion, illustrating how evoked potential thresholds are achieved at lower sound pressure levels. To determine significant differences, a two-tailed paired-sample t test was used to compare behavioral thresholds with and without masking (table 2.4), evoked potential thresholds with and without masking (table 2.5), behavioral and evoked potential thresholds without masking (table 2.6), and behavioral and evoked potential thresholds with masking for 150 Hz and 600 Hz (table 2.7). Behavioral thresholds were significantly different with and without masking noise for both frequencies at an alpha level of 0.05. However, evoked potential thresholds were not significantly different with and without masking at both frequencies. When comparing behavioral thresholds with evoked potential thresholds without masking, there was a significant

PAGE 106

95 difference at 150 Hz, but not at 600 Hz. DISCUSSION Behavioral vs. Evoked Potential Thresholds It has been well documented that behavioral thresholds produce lower hearing sensitivities than evoked potential measurements at low frequencies, but higher sensitivities than evoked potential measurements at high frequencies (Figure 2.17) (Kenyon et al., 1998; Casper et al., 2003; Wols ki et al., 2003). The shape and bandwidth characteristics of the evoked potential and behavioral audiograms are similar, however. Behavioral and evoked potential analyses from this study show consistently higher thresholds than previous work, with the exception of the 2400 Hz frequency, and the amplitude difference between lower and highe r frequencies for both EP and behavioral methods is reduced compared to previous studies. Additionally, the shape of the behavioral audiogram is not consistent with other studies in the higher frequency range. Differences in shape and threshold levels may result from different fish hearing abilities, various testing setups, level of ambient noise, and number of fish tested. Comparison of evoked potential and behavioral thresholds in this study suggests that measuring evoked potential hearing thres holds cannot be used to obtain behavioral hearing thresholds, regardless of the evoked potential analysis chosen. Ideally, the average difference between behavioral and e voked potential thresholds could be used to determine how far off the evoked potential thresholds are from behavioral thresholds. However, the large standard deviations associated with the differences show a wide range in variation of the differences in th reshold techniques between fish (figure 2.9). The differences are not associated with electrode placement (chapter 1), which may have varied slightly between individuals, but could be accounted for by motivation between individuals responding to a sound stimulus and electric shock. Threshold criterion for measuring evoked potentials and behavioral thresholds are arbitrary, and may also be the cause of variability. Regardless of which method of EP is employed, visual or linear regression, the observer define s the visual threshold or on which part of the curve the regression begins. Even though the visual threshold appears to lie on a linear portion of the EP curve, extrapolating to 0 V is still an assumption and may

PAGE 107

96 produce inaccurate results. When establishing behavioral thresholds, the percent correct level used for the criterion is chosen arbitrarily. In this study, a 50% correct detection for determining thresholds was used, in which the fish had to suppress it’s respiration 50% of the time to be considered a detection. Another detection percentage criterion, such as 75% or 99%, may have produced more accurate behavioral thresholds, minimizing variability. Even though exact behavioral hearing thresholds can not be determined from evoked potential measurements with the established criterion, a correction can be used to obtain a range in which the behavioral threshold is found. For example, since 68% of the population falls within one standard deviation of the mean threshold, it can be said that 68% of evoked potential thresholds measured fall within 9.2 dB of the behavioral threshold for 150 Hz (figure 2.9) This correction, however, can only be applied when measuring goldfish hearing thresholds, as other fish species will produce varying standard deviations from the mean. As mentioned, this study performed from 1,000 to 6,000 signal presentation averages on evoked potential measurements. The number of signal averages varies between investigations, which can alter hearing thresholds obtained if the signal is hidden within background noise. Kojima et al (2005) used up to 300 waveform averages, which was the amount required to distinguish the ABR from background noise, whereas Wolski et al. (2003) used 1,000 averages, and Kenyon et al. (1998) and Yan (2002) used 2,000 waveform averages to produce evoked potential thresholds. The number of averages used in this study were higher than most previous studies and still displayed a large amount of variability in threshold data between methods. Since the level of uncorrelated noise can be reduced as a function of the square root of the number of averages (Picton and Hink, 1974)(figure 2.11), the number of averages needed to bring the evoked potential at behavioral threshold to the noise floor can be calculated. The number of averages needed to obtain the evoked potential at behavioral threshold was extremely large at lower frequencies, and decreases significantly with an increase in frequency (figure 2.12). This directly correlates to the background noise level measured at each frequency, which showed a decrease with an increase in frequency. In essence, therefore, thresholds obtained by measuring evoked potentials

PAGE 108

97 might be more of a function of background noise than hearing abilities of animals, and the current number of averages typically used to determine thresholds may not be enough to extract the signal from the background noise, leading to higher thresholds. The higher background SPL at lower frequencies can also account for larger difference between behavioral and EP thresholds (figure 2.13). The increase in background noise can hide the EP signal, producing inaccurately high hearing thresholds, and as the noise decreases with an increase in frequency, the EP is easier to extract, narrowing the gap between threshold techniques. The number of averages to obtain behavioral thresholds shows that with enough signal averaging, the time-locked firing of a single neuron should be detectable with EP techniques. However, one of the advantages to measuring evoked potential thresholds over behavioral conditioning is the ability to rapidly measure hearing thresholds without time-consuming training (Kenyon et al., 1998). By determining the amount of time needed for 1,000 signal averages at one sound pressure level, the time requirement for running one animal starting at the loudest sound pressure level and attenuating in 6dB steps until 90 dB attenuation for five frequencies (same as in this study) was calculated for several signal presentation averages (table 2.8). The length of time required for higher signal presentation averages, such as what would be needed at 150 Hz, is not practical and the fish would not be able to survive for this length of time. Also, since behavioral conditioning was s hown to take approximately 1-2 hours per frequency and is a generally considered a more accurate method to determine fish hearing, this would be a more ideal method to determine thresholds as opposed to measuring evoked potentials with this many signal averages. It should also be noted that the number of averages calculated were to show how many were needed to bring the evoked potential at behavioral threshold to the noise floor. To bring the evoked potentials 3 dB above the noise floor, which was used to determine thresholds in this study, you would have to double the number of averages needed. Masking vs. No Masking Masking occurs when the detection of one signal is impaired by another (Wysocki and Ladich, 2004). In aquatic environments, sound is one of the most

PAGE 109

98 important signal carriers, for it is transported five times faster than in air, is not attenutated as fast as light or chemical substances, and is propagated over large distances due to existing sound channels (Hawkins and Myrberg, 1983). Sounds from different sources provide marine animals with information relevant for survival, such as finding mates and prey, avoiding predators, a nd learning about their environments. In most studies of auditory sensitivity, thresholds have been determined in quiet environments very uncharacteristic of the noisier environment of fish (Popper and Clarke, 1979). There have been several studies that have looked at the masking effects on auditory sensitivity in mammals (e.g. Tavolga, 1967, 1974; Buerkle, 1969; Chapman and Johnstone, 1974; Fay, 1974; Fay et al., 1978) However, the physiological basis for masking may be very different (Poppe r and Clarke, 1979) and not many studies have been completed looking at masking effects on auditory sensitivity in fish. Amoser and Ladich (2003) tested the effects of intense white noise on two otophysine fish species, the goldfish, Carassius auratus and the catfish, Pimelodus pictus and discovered that both species displayed a significant loss of sensitivity (up to 26 dB in goldfish and 32 dB in catfish) immediately afte r noise exposure, with the greatest hearing loss in the range of their most sensitive frequencies. Goldfish were shown to recover within three days, whereas catfish took 14 days to fully recover, showing how hearing specialists are affected differently by noise exposure. However, there have not been many studies testing hearing thresholds dur ing a background masking noise, which would be the case for fish detecting signals in their natural environments. Hearing thresholds obtained with and without the presence of a background noise have been obtained in goldfish measuring evoked potentials (Wysocki and Ladich, 2004). Their results show that continuous white noise at 110 dB and 130 dB significantly influenced auditory thresholds in the goldfish, except at 4 kHz. The auditory thresholds of goldfish increased linearly with the background noise level within the best hearing range, with a 20 dB increase in white noise (110 dB vs. 130 dB) increasing masking thresholds by about 20 dB, except at 4 kHz (figure 2.18). The evoked potential data obtained from this study with and without masking noise were plotted against this data, and show an opposite effect, where the difference between masking and no masking thresholds decreases with an increase in frequency. Our results compared with those obtained by Wysocki and

PAGE 110

99 Ladich (2004) show the amount of variability and the inaccuracies of predicting the effects of masking on hearing thresholds using evoked potential methods. Evoked potential measurements were shown to be unreliable in predicting behavioral thresholds. Therefore, to truly assess the effects of masking on auditory sensitivity and determine how accurate evoked potential measurements are in determining masking thresholds, behavioral and evoked potential measurements need to be recorded in the same fish with and without the presence of masking. Our results indicate that behavioral thresholds with and without the presence of masking do increase linearly and reflect the shift from ambient noise to masking noise (figure 2.14). Evoked potential measurements with and without masking show a much smaller magnitude difference and the difference being largest at the lower frequency. These discrepancies between evoked potential and behavioral thresholds under masking and no masking conditions again reflect the inaccuracy in measuring hearing thresholds and determining masking effects using evoked potentials. Behavioral thresholds were shown to be statically different using a pairedsample t -test with an alpha level of 0.05 for both frequencies (table 2.3). However, evoked potential thresholds were not significantly different at either frequency level. This demonstrates another major flaw of using evoked potential measurements to assess masking thresholds. If masking vs. no masking is tested using evoked potential methods and the results indicate there is no statistical difference, one may conclude that masking has no effect on determining hearing thresholds, when in all actually behavioral methods would have shown a mu ch greater magnitude difference. Using different statistical tests, however, may yield varying results. Aside from comparing masking and no masking thresholds behaviorally and with evoked potentials against each other, thresholds were compared between behavioral and evoked potential methods with masking, and between behavioral and evoked potential methods without masking (figure 2.14). When no masking noise was present, the thresholds obtained at 150 Hz and 600 Hz show the trend obtained in other studies as well as this study where behavioral methods produce lower thresholds at both frequencies, and the difference decreases with an increase in frequency. Thresholds obtained via behavioral conditioning and evoked potential techniques also showed

PAGE 111

100 behavioral thresholds lower at 150 Hz than evoked potentials. However, the difference between the two decreased significantly from 20 dB with no masking to 4 dB with masking. At 600 Hz, an opposite effect occurred, where evoked potential thresholds were lower than behavioral thresholds by a large average difference of approximately 19 dB. One hypothesis to why the difference is small at 150 Hz and evoked potentials are much stronger at 600 Hz than behavioral thresholds is due to the time-locking effect of measuring evoked potentials. When obtaining hearing thresholds via evoked potentials, the observer is averaging the signal noise away from the signal, and knows exactly where to look for the signal, since it would be at twice the stimulus frequency. By averaging long enough, the neuron signal firing at the frequency that is being analyzed will show as a detection, since all others are averaged away. The fish, on the other hand, does not have the ability to average away signals of different frequencies, and therefore can not detect the sound if it is being masked by another noise. This result has severe consequences for measuring evoked potentials in environments where noise may be present. If the noise is loud enough and at the correct frequency, it could prevent the fish from detecting the testing sound stimulus, yet evoked potentials will average away other frequency responses producing a signal at the desired frequency level. This will greatly overestimate auditory sensitivity and lead to false assumptions on hearing abilities and the effects of noise on hearing sensitivity. However, some fish have been shown to be less limited by naturally occurring noise levels than others (Hawkins and Myrberg, 1983). For example, field studies on the cod (Chapman and Hawkins, 1973; Hawkins and Chapman, 1975) have shown that altering the ambient environmental noise greatly affects their hearing, whereas sunfish are much less affected by ambient noise. Otophysines, such as goldfish, have been shown to be much more affected by ambient noise (Wysocki and Ladich, 2004, current study). Although goldfish live in freshwater environments, these habitats, e.g. rivers (Lugli and Fine, 2003), contain considerable background noise and give rise to the speculation that ambient sounds in the environment have influenced the evolution of sound detection or source segregation in fishes (Schellart and Popper, 1992). These results also have implications for human hearing. Evoked potential

PAGE 112

101 thresholds not being significantly different with and without masking implies that hearing tests do not need to be conducted in absolute quiet conditions, and may be administered outside an audiology room. Th is may only apply to specific frequencies however, and may depend on the level of ambient noise. One method to quantify auditory masking and to understand the effects of environmental noise on signal detection and acoustic communication is the thresholdto-noise (T/N) ratio (Wysocki and Ladich, 2004). This is defined as the difference (in dB) between the masked hearing threshold and the spectrum level of the masking noise (Chapman and Hawkins, 1973). Fay and Coombs (1983) obtained T/N ratios for goldfish ranging from 14 to 25 dB using a masking noise with a flat spectrum in the frequency range of the signals used. Resu lts from Wysocki and Ladich (2004) varied, with T/N ratios differing by 6 to 11 dB. T/N ratios from this study were 22 dB for behavioral masking and 26 dB for evoked potential masking at 150 Hz. At 600 Hz, T/N ratios were 38 dB for behavioral and 19 dB for evoked potentials. The evoked potential T/N ratios are different from the evoked potential ratios obtained by Wysocki and Ladich, 2004, and could be the result of different methodological differences concerning the acoustic stimuli as well as differences between fish. The differences between fish may also account for the T/N ratio discrepancy at 600 Hz with that found in Fay and Coombs (1983). CONCLUSION Results from this study conclude that there is a large amount of variability between evoked potential and behavioral thresholds within individual fish, and therefore evoked potential measurements can not be used to predict behavioral thresholds with the number of averages used and the current methods of evoked potential analysis. The data, however, can be used to roughly predict a range of how closely the evoked potential is to behavioral threshold. This range could only be applied to goldfish in this case, since other species may produce different variability’s in thresholds between methods. More averages could bring the evoked potential at behavioral threshold to and above the noise floor, but running this many signal averages would be impractical and behavior al conditioning would be faster and more

PAGE 113

102 accurate at obtaining hearing thresholds. The audiograms obtained via behavior al conditioning and measuring evoked potentials with and without the presence of a background noise demonstrate the ineffectiveness of using evoked potentials to analyze effects of masking on auditory sensitivity at certain frequencies. The difference between evoked potential thresholds with and without masking are much smaller in magnitude than behavioral threshold differences, and are not significantly different This could mislead results and indicate masking has no effect on auditory sensitivity, when it all actuality it does. Comparison of behavioral and evoked potential thresholds with masking show that evoked potential thresholds produces a significantly lower threshold than behavioral thresholds at 600 Hz. This could possibly be due to time-locking of the evoked potential signal, allowing us to average away all the other noise. This could result in evoked potential thresholds overestimating the actual hearing threshold when background noise is present. FUTURE DIRECTION This study provided important data on ground-truthing evoked potential measurements against behavioral conditioning with and without the presence of a masking noise. Goldfish were an ideal fish to use because of their extensive hearing abilities. More behavioral and evoked potential audiograms should be produced for other fish species, including hearing generalists and hearing specialists, to determine if the same amount of variability exists. Even though the number of signal averages to produce evoked potentials at behavioral thresholds to the noise floor was extremely high, especially at lower frequencies, the number of signal averages should more extensively be examined in different fish species to determine if that many averages is always needed to obtain the behavioral thresholds. The method of analysis commonly used to determine evoked potential hearing thresholds is by visual analysis. This study showed that evoked potential thresholds obtained through a linear regression to 0 volts were closer for a majority of frequencies than visual thresholds. However, both methods produced thresholds that were significantly different from each other, and both methods are arbitrary since they rely

PAGE 114

103 on the observer to determine visual threshold. More analysis on optimal evoked potential analysis would be beneficial to aid in predicting behavioral thresholds. This was the first time masking and no masking was conducted on the same individual with behavioral and evoked potential methods. Our study showed that evoked potential thresholds can be significantly lower than behavioral thresholds at certain frequencies. However, only two frequencies were tested. More studies comparing the two methods on the same individual with and without masking noise need to be conducted on several different frequencies to determine if this occurs at other levels. Other species should also be tested, including hearing generalists and specialists to see if this affect occurs in other fish. Measuring evoked potentials is a very rapid way to determine fish hearing thresholds, and is extremely useful in comparing the hearing abilities between fish. The data from this study, however, show that there are several inconsistencies that can result in different thresholds between studies. In order for evoked potentials to be accurate in determining hearing thresholds and comparing results between studies, all of these inconsistencies need to be addressed further.

PAGE 115

104 Table 2.1 Standard length, weight, and frequencies tested of the 11 fish used in comparing EP and behavioral thresholds and 6 fish used in masking study.

PAGE 116

105 EP versus Behavioral Conditioning Masking versus No Masking Fish Standard Length (cm) Weight (g) Freq tested Standard Length (cm) Weight (g) 1 5.0 2.82 600, 1200, 2400 4.7 2.42 2 4.9 3.43 150, 300, 600, 1200, 2400 4.7 2.31 3 4.8 3.25 150, 300, 600, 1200 5.0 3.08 4 4.5 2.21 150, 300, 600, 1200 4.8 3.26 5 5.2 4.45 150, 300, 600, 1200, 2400 4.3 1.86 6 5.2 3.20 150, 300, 600, 1200 4.7 3.13 7 5.0 3.01 150, 300, 600, 1200 --------8 4.7 3.10 150, 300, 600 --------9 4.7 1.96 150, 300, 600, 1200 --------10 5.0 2.58 150, 300, 600, 1200, 2400 --------11 2.2 4.5 150, 300, 600, 1200 --------Mean 4.7 3.14 ---------4.7 2.68 Standard Deviation 0.85 0.80 ---------0.23 0.56

PAGE 117

106 Table 2.2 r2 values obtained from linear regression on evoked potential data for all fish and all frequencies comparing evoked potential thresholds with behavioral thresholds.

PAGE 118

107 Fish 150 Hz 300 Hz 600 Hz 1200 Hz 2400 Hz 1 0.9473 0.9985 0.9098 0.9979 -------2 0.9997 0.9585 0.9992 0.9889 -------3 ----------------1 0.9647 0.9351 4 0.9731 0.9414 0.9996 0.9921 0.9368 5 1 0.9999 0.9998 0.8993 0.8780 6 0.9275 0.9942 1 0.9987 -------7 0.9998 1 0.9997 1 -------8 0.9995 0.9463 0.9934 0.9439 -------9 0.9964 0.9914 0.9639 1 -------10 0.9997 0.9969 0.9669 ---------------11 0.8685 0.9474 0.9566 0.9938 0.9998 Mean 0.9712 0.9775 0.9808 0.9779 0.9374

PAGE 119

108 Table 2.3 Results of Wilcoxon paired-rank test between behavioral evoked potential thresholds. Asterisks denote values signifi cantly different (P<0.05). NT = Not Tested

PAGE 120

109 Frequency n Behavior (Mean + SD) Visual EP (Mean + SD) p-Value Regression EP (Mean + SD) p-Value 150 10 84.70 + 8.51 103.80 + 8.63 p < 0.005 99.95 + 9.14 p < 0.005 300 10 82.64 + 9.95 99.90 + 11.41 p < 0.005 96.17 + 11.81 p < 0.005 600 11 90.34 + 16.46 98.91 + 13.48 p = 0.05 93.54 + 14.33 p = 0.05 1200 10 96.31 + 25.06 97.10 + 10.37 p > 0.50 93.72 + 9.61 p > 0.50 2400 4 101.48 + 16.02 115.75 + 24.78 NT 111.73 + 25.40 NT

PAGE 121

110 Table 2.4 Results of the two-tailed paired-sample t-tests comparing behavioral thresholds with and without masking. Asterisks denote values significantly different (P< 0.0125).

PAGE 122

111 Frequency Masking Behave (Mean + SD) No Masking Behave (Mean + SD) P-Value 150 102.02 + 1.87 78.27 + 2.35 3.96E-06 600 120.52 + 3.15 93.77 + 8.53 0.002

PAGE 123

112 Table 2.5 Results of the two-tailed paired-sample t-tests comparing evoked potential thresholds with and without masking. Aste risks denote values significantly different (P< 0.0125).

PAGE 124

113 Frequency Masking EP (Mean + SD) No Masking EP (Mean + SD) P-Value 150 106.66 + 4.22 98.66 + 5.66 0.07 600 101.49 + 5.18 97.44 + 10.85 0.21

PAGE 125

114 Table 2.6 Results of the two-tailed paired-sample t-tests comparing behavioral and evoked potential thresholds without masking. Asterisks denote values significantly different (P< 0.0125).

PAGE 126

115 Frequency No Masking Behave (Mean + SD) No Masking EP (Mean + SD) P-Value 150 78.27 + 2.35 98.66 + 5.66 0.0003 600 93.77 + 8.53 97.44 + 10.85 0.175

PAGE 127

116 Table 2.7 Results of the two-tailed paired-sample t-tests for behavioral and evoked potential thresholds with masking. Asterisks denote values significantly different (P< 0.0125).

PAGE 128

117 Frequency Masking Behave (Mean + SD) Masking EP (Mean + SD) P-Value 150 102.02 + 1.87 106.66 + 4.22 0.05 600 120.52 + 3.15 101.49 + 5.18 0.001

PAGE 129

118 Table 2.8 Amount of time needed to determine fish hearing thresholds using various signal presentation averages. Each time is calculated from 16 sound pressure levels at 5 frequencies.

PAGE 130

119 Number of Averages Time (16 sound pressure levels for five frequencies) 1000 118.67 minutes (1.98 hours) 5000 593.33 minutes (9.89 hours) 10000 1186.67 minutes (19.78 hours) 100000 11866.67 minutes (197.78 hours = 8.24 days) 500000 59333.33 minutes (988.89 hours = 41.20 days)

PAGE 131

120 Figure 2.1 Diagram of the behavioral conditioning and AEP setup ( RP2.1 Enhanced Real-Time Processor, PA5 programmable attenuator, P1000 110 Watt Professional Power Amplifier, HS4 Bioamp Headstage, DB4 Bioamp Headstage Controller, REC recording electrode, REF reference electr ode, GRO ground electrode, RES Respiration recording electrode). AEP setup would not include the HS4 and DB4, and instead use an RA16 Medusa Amplifier and RA16 Medusa Base Station connected directly into the computer.

PAGE 132

121

PAGE 133

122 Figure 2.2 Diagram of the electric shock set-up used to condition fish ( RP2.1 Enhanced Real-Time Processor, PA5 programmable attenuator ).A A

PAGE 134

123

PAGE 135

124 Figure 2.3 Respiration during a ten-second control period (top) and five-second control period followed by a five-second test period, which included a sound stimulus followed by an electric shock (bottom). A) fish that has not yet trained to the sound B) Typical response of a trained fish suppressing respiration upon initiation of the sound stimulus.

PAGE 136

125 A B

PAGE 137

126 Figure 2.4 Histograms showing ratios A) with no sound presented (n = 717) and B) with suprathreshold sound presented (150 & 600 Hz, 122 dB) (n = 519) (0.9 was used as detection criterion).

PAGE 138

127 A 0 20 40 60 80 100 120 1400 5 0 5 6 0 6 2 0 6 8 0 7 4 0 8 0 8 6 0 9 2 0 9 8 1 0 4 1 1 1 1 6 1 2 2 1 2 8 1 3 4 1 4 1 4 6 1 5 2Control ratioNumber of trials thresold ratio = 0.9 B 0 5 10 15 20 25 300 02 0 1 0 18 0 26 0 34 0 42 0 5 0 58 0 66 0 74 0 82 0 9 0 98 1 06 1 14 1 22 1 3 1 38 1 46 1 54Test vs. control ratioNumber of trials threshold ratio = 0.9

PAGE 139

128 Figure 2.5 Modified staircase method for determining behavioral thresholds. Squares represent the last 12 reversals, in which the average of these reversals determined threshold.

PAGE 140

129 40 60 80 100 120 140 160 05101520253035 Number of trialsSound Pressure (dB re 1 uPa)

PAGE 141

130 Figure 2.6 Example of an AEP response when played pulsed tone at 600 Hz in A) Response in time domain B) Response in frequency domain (DFT).

PAGE 142

131 A B

PAGE 143

132 Figure 2.7 Linear regression on the evoked potential generated at the visual SPL threshold and two previous measurements at 2400 Hz. Where linear regression crosses 0 is the determined hearing threshold. Arrow indicates evoked potential at visual threshold. r2 = 1

PAGE 144

133 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 8090100110120130140150160170180SPL (dB re 1 uPa)Evoked Potential (uV)

PAGE 145

134 Figure 2.8 Audiograms (mean SE) comparing behavioral thresholds, visual evoked potential thresholds, and linear regression thresholds (n = 10 for 150 Hz, 300 Hz, and 1200 Hz; n = 11 for 600 Hz; n = 4 for 2400 Hz).

PAGE 146

135 70 80 90 100 110 120 130 050010001500200025003000Frequency (Hz)SPL (dB re 1 uPa) Behavior EP(visual) EP (regression)

PAGE 147

136 Figure 2.9 Individual difference between behavioral thresholds (mean SE) using A) visual evoked potential methods (mean SD) and B) linear regression evoked potential methods (mean SD).

PAGE 148

137 A -20 -10 0 10 20 30 40 050010001500200025003000 Frequency (Hz)SPL Difference (dB re 1 uPa) EP(visual)-Behave B -30 -20 -10 0 10 20 30 40 050010001500200025003000 Frequency (Hz)SPL Difference (dB re 1 uPa) Regression-Behave

PAGE 149

138 Figure 2.10 Evoked potential input-output functions with behavioral threshold for two fish at 600 Hz with A) behavioral threshol d in middle of linear portion of curve and B) behavioral threshold in the noise floor. Circle represents behavioral threshold.

PAGE 150

139 A -150 -140 -130 -120 -110 -100 -90 -80 507090110130150170190 Sound Pressure (dB re 1 uPa)Evoked Potential (dB V) B -150 -140 -130 -120 -110 -100 -90 -80 507090110130150170190Sound Pressure (dB re 1 uPa)Evoked Potential (dB V)

PAGE 151

140 Figure 2.11 Evoked potentials generated from one fish at 150 Hz with 100, 1000, and 5000 signal averages.

PAGE 152

141 -160 -150 -140 -130 -120 -110 -100 -90 406080100120140160180SPL (dB re 1 Pa)Evoked Potential (dB V) 100 signal averages 1000 signal averages 5000 signal averages

PAGE 153

142 Figure 2.12 Number of signal presentations (mean SD) needed to bring the evoked potential noise floor down to the estimated EP level at the behavioral threshold. Insert shows average number of signal presentation averages on a logarithmic scale.

PAGE 154

143 -1000000 -500000 0 500000 1000000 1500000 2000000 050010001500200025003000 Frequency (Hz)Number of Averages 1 10 100 1000 10000 100000 1000000 010002000 Frequency (Hz)Number of Averages

PAGE 155

144 Figure 2.13 Noise floor level (mean SD) at each frequency.

PAGE 156

145 -170 -165 -160 -155 -150 -145 -140 -135 -130 050010001500200025003000 Frequency (Hz)Noise Floor Level (dB re 1 uPa)

PAGE 157

146 Figure 2.14 Hearing thresholds for behavioral conditioning and evoked potential measurements (mean SD) with and without the presence of a background masking noise. Background noise floor with and wit hout the presence of masking noise also shown. Open markers represent evoked potential measurements, closed markers represent behavioral thresholds, squares represent thresholds with no masking, and triangles represent thresholds with masking.

PAGE 158

147 40 50 60 70 80 90 100 110 120 130 100200300400500600 Frequency (Hz)Threshold (dB re 1 uPa) No Masking Behavior Masking Behavior No Masking EP Masking EP Masking noise floor No masking noise floor

PAGE 159

148 Figure 2.15 Evoked potential input-output functions with behavioral thresholds for one fish at 150 Hz showing A) no masking and B) masking.

PAGE 160

149 A -150 -145 -140 -135 -130 -125 -120 -115 -110 -105 -100 406080100120140160180Sound Pressure (dB re 1 uPa)Evoked Potential (dB V) B -190 -180 -170 -160 -150 -140 -130 -120 -110 -100 -90 406080100120140160180Sound Pressure (dB re 1 uPa)Evoked Potential (dB V)

PAGE 161

150 Figure 2.16 Evoked potential input-output functions with behavioral thresholds for one fish at 600 Hz showing A) no masking and B) masking.

PAGE 162

151 A -160 -150 -140 -130 -120 -110 -100 -90 406080100120140160180Sound Pressure (dB re 1 uPa)Evoked Potential (dB V) B -180 -170 -160 -150 -140 -130 -120 -110 -100 -90 -80 406080100120140160180200Sound Pressure (dB re 1 uPa)Evoked Potential (dB V)

PAGE 163

152 Figure 2.17 Comparison of behavioral and evoked potential audiograms with other studies. Redrawn from Kenyon et al (1998). Open symbols represent EP data.

PAGE 164

153 40 50 60 70 80 90 100 110 120 130 100100010000 Frequency (Hz)Sound Pressure (dB re 1 uPa) ABR (Kenyon et al., 1998) Enger 1966 Fay 1969 Jacobs and Tavolga 1967 Popper 1971 Behavioral audiogram (current study) ABR audiogram (current study)

PAGE 165

154 Figure 2.18 Evoked potential audiograms with and without masking noise compared with results obtained from Wysocki and Ladich (2004).

PAGE 166

155 30 40 50 60 70 80 90 100 110 120 130 100100010000 Frequency (Hz)Sound Pressure (dB re 1 uPa) normal hearing thresholds (Wysocki & Ladich, 2004) 110 dB hearing thresholds (Wysocki & Ladich, 2004) 130 dB hearing thresholds (Wysocki and Ladich, 2004) Normal hearing thresholds (this study) 121 dB hearing thresholds (this study)Normal Lab Conditions (Wysocki & Ladich, 2004) Normal Lab Conditions (This study) 110 dB (Wysocki and Ladich, 2004) 130 dB (Wysocki and Ladich, 2004) 121 dB (This study)

PAGE 167

156 Literature Cited: Akamatsu T, Okumura T, Novarini N, Yan H (2002) Empirical refinements applicable to the recording of fish sounds in small tanks. J Acoust Soc Am 112(6): 3073-3082 Amoser S, Ladich F (2003) Diversity in noise-induced temporary hearing loss in otophysine fishes. J Acoust Soc Am 113: 2170-2179 Banner A (1967) Evidence of sensitivity to acoustic displacements in the lemon shark, Negaprion brevirostris (Poey). In: Cahn PE (ed) Lateral line detectors. Indiana University Press, Bloomington, Indiana, pp 265-273 Behrend ER, Bitterman ME (1962) Avoidance conditioning in the goldfish: exploratory studies of the CS-US interval. Am J Psychol 75: 18-34 Bigelow HB (1904) The sense of hearing in the goldfish Carassius auratus L. Am Nat 38: 275-284 Buerkle BU (1969) Auditory masking and the critical band in Atlantic cod ( Gadus morhua ). J Fish Res Bd Can 26: 1113-1119 Butler A, Hodos W (1996) Comparative Vertebrate Neuroanatomy. Wiley-Liss, Inc. New York, New York Casper BM, Lobel PS, Yan HY (2003) The hearing sensitivity of the little skate, Raja erinacea : A comparison of two methods. Env Biol Fishes 68: 371-379. Chapman CJ, Hawkins AD (1973) A field study of hearing in the cod, Gadus morhua L. J Comp Physiol A 85: 147-167 Chapman CJ, Johnstone ADF (1974) Some auditory discrimination experiments on marine fish. J Exp Biol 61: 521-528 Chapman CJ, Sand O (1974) Field studies of hearing in two species of flatfish Pleuronectes platessa (L.) and Limanda limanda (L.)(family Pleuronectidae). Comp Biochem Physiol A 47: 371-385 Corwin JT (1977) Morphology of the macula neglecta in sharks of the genus Carcharhinus. J Morphol 152(3): 341-362

PAGE 168

157 Corwin JT, Bullock TH, Schweitzer J (1982) The auditory brainstem response in five vertebrate classes. Electroencephalogr Clin Neurophysiol 54: 629-641 Cox M, Rodgers PH, Popper AN, Saidel WN ( 1987) Anatomical effects of intense tone stimulation in the goldfish ear: dependence on sound pressure level and frequency. J Acoust Soc Am 81, Suppl 1:7 Dijkgraff S, Berheijen F (1950) Neue Vers uche ber das Tonunterscheidungsv ermogen der Elritze. Z Vergl Physiol 32: 248-256 Don M, Eberling C (1996) Use of quantitative measures of auditory brainstem response peak amplitude and residual background noise in the decision to stop averaging. J Acoust Soc Am 99: 491-499 Don M, Eberling C, Waring M (1984) Objective detection of averaged auditory brainstem responses. Scand Audiol 13: 219-228 Eberling C, Don M (1987) Threshold characteristics of the human auditory brainstem response. J Acoust Soc Am 81: 115-121 Egner SA, Mann DA (2005) Auditory sensitivity of Sergeant Majors ( Abudefduf saxatilis ) from post-settlement juvenile to adult. Mar Ecol Prog Ser 285: 213-222 Enger PS (1963) Single unit activity in the peripheral auditory system of a teleost fish. Actaphysiol scand 59: suppl. 1-48 Enger PS (1966) Acoustic threshold in goldfish and its relation to the sound source distance. Comp Biochem Physiol 18: 859-868 Enger PS (1981) Frequency Discrimination in Teleosts-Central or Peripheral? In: Tavolga WN, Popper AN, Fay RR (eds) Hearing and sound communication in fish. Springer-Verlag, New York, New York, pp 243-253 Enger PS, Anderson R (1967) An electrophysio logical field study of hearing in fish. Comp Biochem Physiol 22: 517-525 Fay RR (1969) Behavioral audiogram for the goldfish. J Aud Res 9: 112-121 Fay RR (1970a) Auditory frequency discrimination in the goldfish. J Comp Physiol Psychol 73: 175-180 Fay RR (1978a) Coding information in single auditory nerve fibers of the goldfish. J Acoust Soc Am 63: 136-146

PAGE 169

158 Fay RR (1981) Coding of acoustic information in the eighth nerve. In: Tavolga WN, Popper AN, Fay RR (eds) Hearing and sound communication in fishes. Springer, New York, pp 189-222 Fay RR (1988) Hearing in vertebrates: A psychophysics databook. Hill-Fay Associates, Winnetka, IL Fay RR, Ahroon WA, Orawski AA (1978) Auditory masking patterns in the goldfish ( Carassius auratus ): psychophysical tuning curves. J Exp Biol 74: 83-100 Fay RR, Coombs S (1983) Neural mechanisms in sound detection and temporal summation. Hearing Res 10: 69-92 Fay RR, Passow B (1982) Temporal discrimina tion in the goldfish. J Acoust Soc Am 72: 753-760 Fay RR, Popper AN (1974) Acoustic stimulation of the ear of the goldfish ( Carassius auratus ). J Exp Biol 61: 243-260 Furukawa T, Ishii Y (1967) Neurophysiol ogical studies on hearing in goldfish. J Neurophysiol 30: 1377-1403 Hawkins AD, Chapman CJ (1975) Masked auditory thresholds in the cod Gadus morhua L. J Comp Physiol 103A: 209-226 Hawkins AD, Johnstone ADF (1978) The hearing of the Atlantic salmon, Salmo salar J Fish Biol 13: 655-673 Hawkins AD, Myrberg AA (1983) Hearing and sound communication underwater. In: Lewis B (ed) Bioacoustics, a comparative approach. Academic Press, London, pp 347-405 Higgs DM, Brittan-Powell EF, Soares D, Souza MJ, Carr CE, Dooling RJ, Popper AN (2002) Amphibious auditory responses of the American alligator ( Alligator mississipiensis) J Comp Physiol A 188: 217-223 Higgs DM, Rollo AK, Souza MJ, Popper AN (2003) Development of form and function in peripheral auditory structures of the zebrafish ( Danio rerio ). J Acoust Soc Am 113: 1145-1154 Horner JL, Longo N, Bitterman ME (1961) A shuttlebox for fish and a control circuit of general applicability. Am J Psychol 74: 144-120 Iversen, RTB (1969) Auditory thresholds of the scombrid fish Euthynnus affinis with comments on the use of sound in tuna fishing. FAO (Food and Agriculture Organizations of the United Nations) Fisheries Report 62: 849-859

PAGE 170

159 Jacobs DW, Tavolga WN (1967) Acoustic intensity limens in the goldfish. Anim Behav 15: 324-335 Jacobs DW, Tavolga WN (1968) Acoustic fr equency discrimination in the goldfish. Anim Behav 16: 67-71 Jacobson JT (1985) An overview of the auditory brainstem response. In: Jacobson JT (ed) The auditory brainstem response. College-Hill Press, San Diego, pp 3-12 Jewett DL (1970) Volume conducted potentials in response to auditory stimuli as detected by averaging in the cat. Electroencephalogr Clin Neurophysiol 28: 609618 Jewett DL, Williston JS (1971) Auditory evoked far fields averaged from the scalp of humans. Brain 94: 681-696 Junger MC, Feit D (1993) Sound, structures, and their interaction. Acoustical Society of America, Melville, New York Katz J (1994) Handbook of Clinical Audiology. Williams and Williams, Baltimore, MD Kenyon TN, Ladich F, Yan H (1998) A comparative study of hearing ability in fishes: the auditory brainstem response approach. J Comp Physiol A 182: 307-318 Kojima T, Ito H, Komada T, Taniuchi T, Akamatsu T (2005) Measurements of auditory sensitivity in common carp Cyprinus carpio by the auditory brainstem response technique and cardiac conditioning method. Fish Sci 71: 95-100 Ladich F, Yan HY (1998) Correlation between auditory sensitivity and vocalization in anabantoid fishes. J Comp Physiol A 182: 737-746 Lewis ER, Leverenz EL, Bialek WS (1985) Th e Vertebrate Inner Ear. CRC Press, Boca Raton, FL Lombarte A, Popper AN (1994) Quantitative analyses of postembryonic hair cell addition in the otolithic endorgans of the inner ear of the European hake, Merluccius merluccius (Gadiformes, Teleostei). J Comp Neurol 345: 419-428 Lu Z, Tomchik SM (2002) Effects of a redtide toxin on fish hearing. J Comp Physiol A 188: 807-813 Lu Z, Xu Z (2002) Effects of saccular otolith removal on hearing sensitivity of the sleeper goby ( Dormitator latifrons ). J Comp Physiol A 188: 595-602

PAGE 171

160 Lugli M, Fine ML (2003) Acoustic communication in two freshwater gobies: ambient noise and short-range propagation in shallow streams. J Acoust Soc Am 114: 512521 Mann DA, Higgs DM, Tavolga WN, Souza MJ, Popper AN (2001) Ultrasound detection by clupeiform fishes. J Acoust Soc Am 109(6): 3048-3054 Mann DA, Lu Z, Hastings MC, Popper AN ( 1998) Detection of ultrasonic tones and simulated dolphin echolocation clicks by a teleost fish, the American shad ( Alosa sapidissima ). J Acoust Soc Am 104(1): 562-568 Mann DA, Lu Z, Popper AN (1997) A clupeid fish can detect ultrasound. Nature 389: p 341 Myrberg AA (1978) Ocean noise and the behavior of marine animals: relationships and implications. In: Fletcher JL, Busnel RB (eds) Effects of noise on wildlife. Academic Press, New York, pp. 169-208 Myrberg AA, Spires JY (1980) Hearing in dams elfishes: an analysis of signal detection among closely related species. J Comp Physiol 140: 135-144 Moeng RS, Popper AN (1984) Auditory response of saccular neurons of the catfish, Ictalurus punctatus J Comp Physiol A 155: 615-624 Offutt GC (1968) Auditory response in the goldfish. J Aud Res 8: 391-400 Parker GH (1903) The sense of hearing in fishes. Am Nat 37: 185-204 Picton TW, Hink RF (1974) Evoked potentials: How? What? & Why? Am J EEG Technol 14: 9-44 Platt C, Popper AN (1984) Variation in lengths of ciliary bundles on hair cells along the macula of the sacculus in two species of teleost fishes. Scanning Electron Microsc 1984: 1915-1925 Popper AN (1971) The effects of size on auditory capacities of goldfish. J Aud Res 11: 239-247 Popper AN, Carlson TJ (1998) Application of sound and other stimuli to control fish behavior. Trans Am Fish Soc 127: 673-707 Popper AN, Clarke NL (1979) Non-simultaneous auditory masking in the goldfish, Carassius auratus J Exp Biol 83: 145-158 Popper AN, Chan AT, Clarke NL (1973) An evaluation of methods for behavioral investigations of teleost audition. Behav Res Methods Instrum 5: 470-472

PAGE 172

161 Popper AN, Coombs S (1980) Auditory mechanisms in teleost fish. Am Sci 68:429-440 Popper AN, Fay RR (1993) Sound detection and processing by fish: critical review and major research questions. Brain Behave Evol 41: 14-38 Popper AN, Hoxter B (1984) Growth of a fish ear: 1. Quantitative analysis of sensory hair cell and ganglion cell proliferation. Hear Res 15: 133-142 Popper AN, Lu Z (2000) Structure-function relati onships in fish otolith organs. Fish Res 46: 16-25 Popper AN, Platt C (1993) Inner ear and lateral line. In: Evan DH (ed) The physiology of fishes. CRC Press, Boca Raton, FL, pp 99-136 Popov VV, Supin AY (1990) Auditory brainstem responses in characterization of dolphin learning. J Comp Physiol A 166: 385-393 Ridgeway SH, Bullock TH, Carder DA, Seeley RL, Woods D, Galambos R (1981) Auditory brainstem response in dolphins. Proc Natl Acad Sci 78: 1943-1947 Roberts WM, Howard J, Hudspeth AJ (1988) Hair cells: transduction, tuning and transmission in the inner ear. Annual Rev Cell Biol 4: 63-92 Rogers PH, Popper AN, Hastings MC, Saidel WM (1988) Processing of acoustic signals in the auditory system of bony fish. J Acoust Soc Am 83: 338-349 Saidel WM, Popper AN (1987) Sound perception in two anabantid fishes. Comp Biochem Physiol A 88: 37-44 Sand O (1974) Directional sensitivity of microphonic potentials from the perch ear. J Exp Biol 60: 901-908 Sand O, Michelsen A (1978) Vibration measurement of the perch otolith. J Comp Physiol 123: 85-89 Saunders JC, Dear SP (1983) Comparative morphology of stereocilia. In: Fay RR, Gourevitch (eds) Hearing and other senses: Presentations in honor of E.G. Wever. Amorpha Press, Groton, CT, pp 175-198 Sawa M (1976) Auditory responses from single neurons of the medulla oblongata in the goldfish. Bull Jpn Soc Sci Fish 42: 141-152 Schellart NAM, Popper AN (1992) Functional aspects of the evolution of the auditory system of actinopterygian fish. In: Webster DB, Fay RR, Popper AN (eds) The evolutionary biology of hearing. Springer-Verlag, New York, pp 295-322

PAGE 173

162 Scholik AR, Yan HY (2001) Effects of underwater noise on auditory sensitivity of a cyprinid fish. Hear Res 152: 17-24 Scholik AR, Yan HY (2002) Effects of boat engine noise on the auditory sensitivity of the fathead minnow, Pimephales promelas Stetter H (1929) Untersuchungen ber den Gehrsinn der Fische, besonders von Phoxinus laevis L. und Amiurus nebulosus Raf. Z Vergl Physiol 9: 339-447 Szymanski MD, Bain DE, Keihl K, Pennington S, Wong S, Henry KR (1999) Killer whale ( Orcinus orca ) hearing: Auditory brainstem response and behavioral audiograms. J Acoust Soc Am 106(2): 1134-1140 Tavolga WN (1967) Masked auditory thresholds in teleost fishes. In: Tavolga WN (ed) Marine bio-acoustics, volume 2. Pergamon Press, Oxford, UK, pp 233-245 Tavolga WN (1974) Signal/noise ratio and the critical band in fishes. J Acoust Soc Amer 55: 1323-1333 Tavolga WN, Wodinski J (1963) Auditory capacities in fishes. Bull Amer Mus Nat Hist 126: 177-240 von Frisch K (1936) ber den Gehrsi nn der Fische. Biol Rev 11: 210-246 Weiss BA (1966) Auditory sensitivity in the goldfish. J Aud Res 6: 321-335 Wolski LF, Anderson RC, Bowles AE, Yochem PK (2003) Measuring hearing in the harbor seal ( Phoca vitulina ): Comparison of behavioral and auditory brainstem response techniques. J Acoust Soc Am 113: 629-637 Wysocki LE, Ladich F (2002) Can fishes resolve temporal characteristics of sound? New insights using auditory brainstem responses. Hear Res 169: 36-46 Wysocki LE, Ladich F (2003) The representation of conspecific sounds in the auditory brainstem of teleost fishes. J Exp Biol 206: 2229-2240 Wysocki LE, Ladich F (2004) Hearing in fishes under noise conditions. J Assoc Res Otolaryn (in press). Yan (2001) A non-invasive electrophysiologi cal study on the enhancement of hearing abilities in fishes. Proc I.O.A. 23(4): 15-26 Yan HY (2002) The use of acoustically evoked potentials for the study of hearing in fishes. Bioacoustics 12: 325-328

PAGE 174

163 Yan HY, Curtsinger WS (2000) The otic gasbladder as an ancillary auditory structure in mormyrid fish. J Comp Physiol A 186: 595-602 Yan HY, Fine ML, Horn NS, Coln WE (2000) Variability in the role of the gas bladder in fish audition. J Comp Physiol A 186: 435-445 Yan HY, Popper AN (1991) An automated positive reward method for measuring acoustic sensitivity in fish. Behav Res Methods Instrum 23: 351-356 Yan HY, Popper AN (1992) Auditory sensitivity of the cichlid fish Astronotus ocellatus (Cuvier). J Comp Physiol A 171: 105-109 Yan HY, Popper AN (1993) Acoustic intensity discrimination by the cichlid fish Astronotus ocellatus (Cuvier). J Comp Physiol A 173: 347-351