© 2009 paul august andersonufdcimages.uflib.ufl.edu/uf/e0/02/47/76/00001/anderson_p.pdf ·...

1

THE FUNCTIONS OF SOUND PRODUCTION IN THE LINED SEAHORSE, HIPPOCAMPUS ERECTUS, AND EFFECTS OF LOUD AMBIENT NOISE ON ITS

BEHAVIOR AND PHYSIOLOGY IN CAPTIVE ENVIRONMENTS

By

PAUL AUGUST ANDERSON

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT

OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2009

2

© 2009 Paul August Anderson

3

To my parents, Josephine J. McLaughlin and Paul F. Anderson, whose unconditional love and support have paved the way for me to be everything that I want to be

4

ACKNOWLEDGMENTS

First and foremost, I owe a debt of gratitude to my Ph.D. advisor, W.J. Lindberg

(University of Florida/UF), who accepted the daunting role of advising a student with complex,

multidisciplinary research interests. Among the many contributions made to the project, Dr.

Lindberg provided wisdom and teachings in the scientific process, helped guide complex

negotiations among many stakeholders involved, offered funding via the D.M. Smith Fellowship,

and assisted in the acquisition of additional funds along the way. Dr. Lindberg provided

enlightened guidance all along the route of this circuitous adventure to its destination!

Similarly, I offer many thanks to each of my Ph.D. committee members, I. Berzins (The

Florida Aquarium) , H. Masonjones (University of Tampa), D. Murie (UF), D. Parkyn (UF), C.

St. Mary (UF), each of whom played important roles in shaping the project into its final form. In

particular, I thank I. Berzins for offering a fruitful and collaborative relationship with The

Florida Aquarium Center for Conservation, and for generating ideas about the effects of noise on

stress in fishes (that led to co-authorship in Chapter 4). I. Berzins also provided funding to

conduct The Seahorse Sound Survey (Chapter 2) and has created a position for me at The Center

for Conservation that has to date enabled me to fulfill mutual career goals; it is an opportunity

that I truly cherish and enjoy. The concept of the Dissertation was developed in consult with H.

Masonjones, who first suggested effects of noise on seahorse health and reproduction, having

previously detected effects in her study system (H. zosterae) and suggesting further exploration

of the phenomenon. H. Masonjones also captained the vessel (and furnished a crew of

competent shipmates, S., G., and K. Masonjones) upon which ambient noise measurements were

taken from the field. D. Murie provided thorough critique of study design and also suggested the

study of acoustic roles in prey capture behavior, that led to Chapter 5 of this Dissertation. D.

Parkyn offered wisdom and teachings in the complex world of acoustic neuroethology, and was

5

instrumental in securing stipend funding along the way. Finally, C. St. Mary also offered

valuable critique of study design, assisted with fund acquisition, and advised statistical methods

in Chapter 4. In the company of these mentors, I also wish to acknowledge R. Francis-Floyd,

who opened the door for me at The University of Florida and helped me to get started on my

path.

Some Chapters in this Dissertation are the result of fruitful collaborations with co-authors,

who provided knowledge and resources instrumental to the work. I am indebted to D. Mann

(University of South Florida), who provided the analytical system necessary to conduct the study

in Chapter 3, along with patient teachings in the methods of AEP audiometry and sound analysis

in general. I also acknowledge his students, B. Casper, M. Hill-Cook, R. Hill, and J. Locascio,

who provided guidance along the way. L. J. Guillette and H. Hamlin (UF) taught enzyme

immunoassay methods and offered their laboratory to conduct cortisol assays included in

Chapter 4. They also provided valuable advice in considering the stress phenomenon, advising

the examination of variance among response measures, that led to a fruitful examination of

results. I also acknowledge members of their lab, in particular T. Edwards and B. Moore, for

guidance. E. Adams and F. Fogarty (UF) served as undergraduate research assistants along the

way. Among the many (acknowledged below), these individuals invested tremendous effort and

original thought over an extended period of time that contributed greatly to Chapters 6 and 4

(respectively). I am very thankful for their sustained hard work.

The Dissertation was further molded by fruitful conversations with, and wisdom

contributed by, professionals in diverse fields. A. Slater (The Florida Aquarium) molded my

skills in syngnathid husbandry. I owe many thanks to J. Pattee (Pioneer Hill Software) and R.

Shrivastav (UF), who taught concepts of acoustics and sound analysis. A. Noxon (Acoustic

6

Sciences Corporation) shared his knowledge in acoustic engineering that shaped soundproofing

methods that were used in the methods of Chapters 4-6. The muting surgery that was employed

in Chapters 5 and 6 was first developed and advised by D. Colson (Rhode Island ENT

Physicians, Inc.), who first published a study using the muting method. K. Harr (UF) advised

hematological methods used in Chapter 4. A. Dove (Georgia Aquarium) and E. Greiner (UF)

assisted with parasite identification in Chapter 4. L. Farina (UF) guided histological

interpretation. Statistical analyses employed in the Chapters were selected and shaped as a result

of consultation with several people in addition to those previously mentioned, including M.

Allen, M. Brennan, J. Colee, D. Dutterer, J. Hill, and R. Littell (UF).

Several people provided technical guidance and offered access to technical resources that

made elements of methodology possible. I am especially grateful to D. Petty (UF) and her

laboratory (including J. Holloway and T. Crosby), who provided invaluable veterinary services

all along the way to maintain the health of the research collection. They also participated in, and

helped shape the methodology for, the physiological methods used in Chapter 4 and the muting

surgery used in Chapters 5 and 6. D. Samuelson and P. Lewis (UF) taught histological

techniques and provided access to their laboratory for histological processing and evaluation. H.

Rutherford (The Pier Aquarium) loaned hydrophones that were used extensively for sound

recording.

I appreciate the generous contributions of people in the fishing and marine ornamental

industries that provided animals to study as well as resources to care for them. I owe many

thanks to M. Helmholtz (Above The Reef), R. Stevens (Mary Jane Shrimp Co.), and the crew of

The Twin Rivers Marina, who believed in the project and donated many animals to build the

research collection. I am grateful to A. and M. Maxwell and the crew of Sea Critters, Inc., who

7

provided a live food supply for the collection over a period of several years. R. Lewis (Aquatic

Indicators) provided Mysidopsis bahia for the experiment described in Chapter 5.

I enjoyed the opportunity to engage undergraduate students of the University of Florida in

research experience. In return, this team provided a powerful workforce, ensuring proper care of

the research collection as well as the collection of quality data that are at the base of the results

reported here. These people are E. Adams, F. Bastos, J. Bound, F. Fogarty, S. Koka, J. Liu, B.

Macke, A. Maness, J. Moriarty, K. Nuessly, S. Osborn, Z. Punjani, J. Rosenbaum, B. Slossberg,

D. Snipelsky, and A. Weppelman.

Nine public aquaria from throughout the United States participated in The Seahorse Sound

Survey (Chapter 2); for their contributions I am grateful. Thanks to R. Doege (Dallas World

Aquarium), R. Curttright (Kingdom of the Seas Aquarium), K. Dobson (Maritime Aquarium), J.

Reynolds and S. Reiner (Moody Gardens), S. Spina (New England Aquarium), J. Moffatt

(Pittsburgh Aquarium), J. Skoy (Ripley’s Aquarium of the Smokies), J. Rawlings (Riverbanks

Zoo), and K. Alford (Tennessee Aquarium). I also acknowledge N. Dunham (Florida Fish and

Wildlife Conservation Commission), who graciously offered data and maps on sites in Tampa

Bay where seahorses have been collected, and R. Watkins (UF), who provided Figure 2-2.

Many funding sources contributed to the success of this Dissertation. As previously

mentioned, the D.M. Smith Fellowship funded laboratory construction and research operations,

and The Florida Aquarium Center for Conservation funded The Seahorse Sound Survey. The

Mulvihill Scholarship (granted by Aquaculture Research/Environmental Associates, Inc. and the

United States Aquaculture Association), Project A.W.A.R.E., and the Twin Rivers Marina also

funded operations. The American Association of Zoo Veterinarians’ Mazuri Fund fully

sponsored Chapter 4. My scholarship and stipend was provided by the University of Florida’s

8

Alumni Fellowship, the Morris Animal Foundation, The Spurlino Foundation, and The Florida

Aquarium Center for Conservation.

Animal collection was authorized by the Florida Fish and Wildlife Conservation

Commission Special Activities License #05SR-944. Husbandry and experimental protocols

were authorized by the University of Florida IACUC Protocol #D-432, the University of South

Florida IACUC Protocol #2118, and The Florida Aquarium Animal Care and Use Committee.

9

TABLE OF CONTENTS page

ACKNOWLEDGMENTS ...............................................................................................................4

LIST OF TABLES .........................................................................................................................13

LIST OF FIGURES .......................................................................................................................14

ABSTRACT ...................................................................................................................................16

CHAPTER

1 INTRODUCTION ..................................................................................................................18

The Aquaculturist’s Challenge ...............................................................................................18 The Acoustic Sense: Is It Important to Consider? .................................................................18 Effects on Hearing ..................................................................................................................20 The Stress Response ...............................................................................................................22 The Seahorse: A Model .........................................................................................................24 Questions ................................................................................................................................27 Objectives and Outline ...........................................................................................................27 Contribution to Science and Industry .....................................................................................28

2 THE SEAHORSE SOUND SURVEY ...................................................................................30

Introduction .............................................................................................................................30 Materials and Methods ...........................................................................................................31

Aquarium Data Collection ...............................................................................................31 Data Collection in the Wild .............................................................................................32 Sound Analysis ................................................................................................................32

Calibration ................................................................................................................32 Sound processing ......................................................................................................33

Statistical Analysis ..........................................................................................................34 Results .....................................................................................................................................35

Questionnaire Data ..........................................................................................................35 Tank dimensions, materials, and habitat ..................................................................35 Aeration and drainage ..............................................................................................35 Life support equipment ............................................................................................36 Other noise sources ..................................................................................................36 Animal collection .....................................................................................................36 Health and reproduction records ..............................................................................37

Sound Analysis ................................................................................................................37 Mid-water level recordings ......................................................................................37 Bottom level recordings ...........................................................................................38 Wild ambient noise recordings .................................................................................38

10

Ambient Noise Comparisons ...........................................................................................38 Effects of Design Variables on Total Power ...................................................................39

Discussion ...............................................................................................................................40

3 AUDITORY EVOKED POTENTIAL OF THE LINED SEAHORSE, HIPPOCAMPUS ERECTUS, AND ITS RELATIONSHIP TO CONSPECIFIC SOUND PRODUCTION .....54


Animal Accession, Holding, and Husbandry Procedures ...............................................56 AEP ..................................................................................................................................58

Experimental setup ...................................................................................................58 Sound generation, calibration, and AEP acquisition ................................................59 Data analysis ............................................................................................................60

Click Recordings .............................................................................................................60 Experimental setup ...................................................................................................60 Data analysis ............................................................................................................61

Results .....................................................................................................................................63 Ambient Noise .................................................................................................................63 AEP ..................................................................................................................................63 Click Recordings .............................................................................................................64

Pressure ....................................................................................................................64 Particle acceleration .................................................................................................65 Comparisons between pressure waveforms and particle acceleration

waveforms .............................................................................................................65 Comparisons among individuals and between sexes ...............................................66

Discussion ...............................................................................................................................66

4 SOUND, STRESS, AND SEAHORSES: THE CONSEQUENCES OF A NOISY ENVIRONMENT TO ANIMAL HEALTH ...........................................................................82


Animal Accession and Husbandry Procedures ................................................................83 Laboratory and Experimental Tank Design ....................................................................84 Sound Recording and Analysis .......................................................................................86 Animal Assignment and Preparation ...............................................................................87 Ethological Methodology ................................................................................................87

Data collection ..........................................................................................................87 Measures and statistical analyses .............................................................................88

Physiological Methodology .............................................................................................89 Necropsy ...................................................................................................................89 Blood processing ......................................................................................................90 Bacterial culture .......................................................................................................91 Measures & statistical analyses ................................................................................92

Results .....................................................................................................................................93

11

Sound Analysis ................................................................................................................93 Ethological Results ..........................................................................................................93

Ethogram ..................................................................................................................93 State analyses ...........................................................................................................93 Event analyses ..........................................................................................................94

Physiological Results .......................................................................................................96 Morphological indices ..............................................................................................96 Hematological count ................................................................................................96 Leukocyte differential ..............................................................................................96 Blood glucose concentration ....................................................................................97 Plasma cortisol concentration ...................................................................................97 Incidence of disease .................................................................................................97

Discussion ...............................................................................................................................98 The Stress Concept ..........................................................................................................98 Primary Stress Indices .....................................................................................................99 Secondary Stress Indices ...............................................................................................100 Tertiary Stress Indices ...................................................................................................102

Physiology ..............................................................................................................102 Behavior .................................................................................................................104

Conclusions ...........................................................................................................................107

5 ACOUSTIC ROLES AND EFFECTS IN PREY CAPTURE BEHAVIOR OF LINED SEAHORSES (HIPPOCAMPUS ERECTUS) IN AQUARIA .............................................118

Introduction ...........................................................................................................................118 Materials and Methods .........................................................................................................120

Experimental Tank Design ............................................................................................120 Sound Recording and Analysis .....................................................................................120 Muting ...........................................................................................................................121

Surgical procedure ..................................................................................................121 Click recording and analysis ..................................................................................121

Prey Capture Experiment ..............................................................................................122 Results ...................................................................................................................................123

Sound Analysis ..............................................................................................................123 Ambient noise analysis ...........................................................................................123 Click analysis .........................................................................................................124

Prey Capture ..................................................................................................................124 Discussion .............................................................................................................................125

6 SOUND, SEX, AND SEAHORSES: ACOUSTIC ROLES AND EFFECTS IN COURTSHIP BEHAVIOR OF LINED SEAHORSES (HIPPOCAMPUS ERECTUS) IN AQUARIA .......................................................................................................................129

Introduction ...........................................................................................................................129 Manifestations of Stress in Courtship Behavior ............................................................129 The Functions of Clicking in Courtship Behavior of the Lined Seahorse,

Hippocampus erectus .................................................................................................131

12

The Effects of Tank Noise on Acoustic Communication ..............................................131 Study Objectives ............................................................................................................132

Materials and Methods .........................................................................................................132 Animal Accession, Laboratory Design, and Husbandry Procedures ............................132 Experimental Tank Design ............................................................................................133 Sound Recording and Analysis .....................................................................................133 Muting ...........................................................................................................................134 Courtship Experiment ....................................................................................................134

Experimental design & observational methods ......................................................134 Ethological analysis ................................................................................................135

Results ...................................................................................................................................137 Sound Analysis ..............................................................................................................137 Ethological Analysis ......................................................................................................138

Tests of means ........................................................................................................138 Tests of mean deviations ........................................................................................139 Occurrence of clicking ...........................................................................................141

Discussion .............................................................................................................................142 The Effect of Tank Noise on Courtship Behavior .........................................................142 The Functions of Clicking in Courtship Behavior ........................................................143 The Effects of Tank Noise on Acoustic Communication .............................................146

7 DISCUSSION .......................................................................................................................157

The Hearing Ability of Hippocampus erectus in the Context of Wild and Captive Acoustic Environments .....................................................................................................157

Pressure vs. Displacement ....................................................................................................158 The Function of Clicking ......................................................................................................160 Variability in Signaling .........................................................................................................163 Masking ................................................................................................................................164 The Lateral Line ...................................................................................................................165 The Stress Response .............................................................................................................166

Physiology .....................................................................................................................166 Behavior ........................................................................................................................167

Solutions ...............................................................................................................................168

REFERENCES ............................................................................................................................172

BIOGRAPHICAL SKETCH .......................................................................................................190

13

LIST OF TABLES

Table page 2-1 Spectra Plus processing settings ........................................................................................46

2-2 Public aquarium participants of the Seahorse Sound Survey ............................................46

2-3 Frequency distribution of tank design specifications and animal inhabitants ...................47

2-4 Descriptive statistics of spectrum level SPLs, peak amplitude, total power, and peak frequency ............................................................................................................................48

2-5 Spectrum-level SPLs at peak or dominant frequencies of representative fish sounds ......49

3-1 Spectra Plus recording settings. .........................................................................................72

4-1 Categorization of physiological measures for statistical analysis ....................................109

4-2 Ethogram of the lined seahorse, Hippocampus erectus: Maintenance behavior ............110

4-3 Summary of physiological results ....................................................................................111

5-1 Partial ethogram of the lined seahorse, Hippocampus erectus: Prey capture behavior ..127

6-1 Partial ethogram of courtship behaviors of the lined seahorse, Hippocampus erectus ...148

6-2 Summary of significant results of Type III tests of means of fixed effects in SAS PROC GLIMMIX ............................................................................................................149

6-3 Summary of significant results of Type III tests of mean deviations of tank treatment effects or its interactions in SAS PROC GLIMMIX .......................................................150

14

LIST OF FIGURES

Figure page 2-1 The Seahorse Sound Survey kit .........................................................................................50

2-2 Acoustic sampling sites in Tampa Bay, FL. ......................................................................51

2-3 Power spectra from representative tanks ...........................................................................52

2-4 Total power vs. selected design specifications ..................................................................53

2-5 Mid-level total power vs. wall material type among tanks with gravel bottoms ...............53

3-1 Soundproofed laboratory tank ............................................................................................73

3-2 AEP experimental setup .....................................................................................................74

3-3 Auditory evoked potentials to a 400 Hz tone pip ..............................................................75

3-4 Click recording chamber ....................................................................................................76

3-5 Click characteristics measured in the time domain waveform ..........................................76

3-6 Click characteristics measured in the frequency domain waveform .................................77

3-7 Representative power spectra of sound pressure of long-term holding tanks and the sound-proofed laboratory tank ...........................................................................................78

3-8 Power spectrum of particle acceleration of the sound-proofed laboratory tank. ...............79

3-9 Audiograms of the lined seahorse, H. erectus, for sound pressure and particle acceleration ........................................................................................................................79

3-10 A resonant click, depicted in the time domain ...................................................................80

3-11 Sound pressure audiograms of representative hearing generalist fishes, measured by the AEP technique ..............................................................................................................80

3-12 Comparison of the broadband sound pressure audiogram against peaks of recorded clicks ..................................................................................................................................81

3-13 Comparison of the broadband particle acceleration audiogram against peaks of recorded clicks ...................................................................................................................81

4-1 Experimental tank design schematic: Loud tank ............................................................112

4-2 Experimental tank design schematic: Quiet tank ............................................................113

15

4-3 Power spectra of ambient noise in representative tanks ..................................................114

4-4 Proportion of time spent stationary ..................................................................................115

4-5 Number of adjustments made per hour while stationary .................................................115

4-6 Occurrence of piping ........................................................................................................116

4-7 Occurrence of clicking .....................................................................................................116

4-8 Occurrence of gaping .......................................................................................................117



6-2 Comparisons of means (+ SE) of male approach ............................................................152

6-3 Comparisons of means (+ SE) of female point ................................................................152

6-4 Comparisons of means (+ SE) of clicking between males and females ..........................153

6-5 Comparisons of standard deviations of brightening between males in quiet and loud tanks .................................................................................................................................153

6-6 Comparisons of standard deviations of display between males in quiet and loud tanks .154

6-7 Comparisons of standard deviations of display among females ......................................154

6-8 Comparisons of standard deviations of pointing between males in loud and quiet tanks .................................................................................................................................155

6-9 Comparisons of standard deviations of pouch pumping between males in loud and quiet tanks ........................................................................................................................155

6-10 Comparisons of standard deviations of pointing among females ....................................156

6-11 Comparisons of standard deviations of clicking among males ........................................156

7-1 Comparisons of ambient noise in tank environments with broadband hearing in Hippocampus erectus .......................................................................................................170

7-2 Comparisons of ambient noise in the wild with broadband hearing in Hippocampus erectus ..............................................................................................................................171

16

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

THE FUNCTIONS OF SOUND PRODUCTION IN THE LINED SEAHORSE,

HIPPOCAMPUS ERECTUS, AND EFFECTS OF LOUD AMBIENT NOISE ON ITS BEHAVIOR AND PHYSIOLOGY IN CAPTIVE ENVIRONMENTS

By

Paul August Anderson

August 2009 Chair: William J. Lindberg Major: Fisheries and Aquatic Sciences

Loud noise in aquaria represents a cacophonous environment for captive fishes. I tested

the effects of loud noise on acoustic communication, feeding behavior, courtship behavior, and

the stress response of the lined seahorse, Hippocampus erectus. Total Root Mean Square (RMS)

power of ambient noise to which seahorses are exposed in captivity varies widely but averages

126.1 + 0.8 deciBels with reference to one micropascal (dB re: 1 µPa) at the middle of the water

column and 133.7 + 1.1 dB at the tank bottom, whereas ambient noise in the wild averages 119.6

+ 3.5 dB. Hearing sensitivity of H. erectus, measured from auditory evoked potentials,

demonstrated maximum spectrum-level sensitivities of 105.0 + 1.5 dB and 3.5 X 10-3 + 7.6 X

10-4 m/s2 at 200 Hz; which is characteristic of hearing generalists. H. erectus produces acoustic

clicks with mean peak spectrum-level amplitudes of 94.3 + 0.9 dB at 232 + 16 Hz and 1.5 X 10-3

+ 1.9 X 10-4 m/s2 at 265 + 22 Hz. Frequency matching of clicks to best hearing sensitivity, and

estimates of audition of broadband signals suggest that seahorses may hear conspecific clicks,

especially in terms of particle motion.

Behavioral investigations revealed that clicking did not improve prey capture proficiency.

However, animals clicked more often as time progressed in a courtship sequence, and mates

17

performed more courtship behaviors with control animals than with muted animals, lending

additional evidence to the role of clicking as an acoustic signal during courtship. Despite loud

noise and the role of clicking in communication, masking of the acoustic signal was not

demonstrated.

Seahorses exposed to loud noise in aquaria for one month demonstrated physiological,

chronic stress responses: reduced weight and body condition, and increased heterophil to

lymphocyte ratio. Behavioral alterations were characterized by greater mean and variance of

activity among animals housed in loud tanks in the first week, followed by habituation. By week

four, animals in loud tanks demonstrated variable performance of clicking and piping, putative

distress behaviors. Despite the physiological stress response, animals in loud tanks did not

reduce feeding response or courtship behavior, suggesting allostasis.

18

CHAPTER 1 INTRODUCTION

The Aquaculturist’s Challenge

In general, the aim of finfish aquaculture is to maximize production of healthy finfish for

economic gain. Biologically, this translates into maximizing growth, growth rate, rate of

reproduction, and fecundity of target organisms. The culturist must give consideration to the

many factors needed to ensure survival and reproduction of his fish. This includes considering

the animal’s nutritional needs that vary over the animal’s life history (Watanabe, 1988),

designing and maintaining a filtration system and husbandry schedule that keeps water quality

within acceptable parameters (Bromage et al., 1988), giving consideration to stocking density

and tank size (Pillay, 1990), and providing an environment with necessary environmental cues

sufficient to induce spawning (Bardach and Magnuson, 1980), among other considerations.

The strategies that an aquaculturist may use to improve one parameter might adversely

affect another. For example, while a high rate of water flow might facilitate filtration, the

increased current may adversely affect delicate fry (e.g., Opstad et al., 1998). A feed that is high

in protein may accelerate growth but also may increase biological load on the filtration system

and adversely affect water quality (e.g., Tidwell et al., 1996). Somehow, the aquaculturist must

balance numerous factors to produce an optimal environment for his organism. This can be

particularly challenging when some strategies may adversely affect an organism in ways that the

aquaculturist is unaware.

The Acoustic Sense: Is It Important to Consider?

I suggest that the acoustic sense of fish is a sensory modality that is often overlooked in

aquaculture. The earliest evolved and most general role of the fish ear is to gain information

about the environment through its acoustic signature (Popper and Fay, 1999). Some fishes have

19

taken further advantage of the acoustic sense by evolving sound production mechanisms for

intraspecific communication, as in the plainfin midshipman fish (Porichthys notatus), that

utilizes acoustic communication in courtship (Bass and Clark, 2003). An intensive culture

system may significantly interfere with both functions. Fish in intensive culture systems are

exposed to ambient noise from many sources, perhaps most notably the pump motors that run to

maintain water quality. Sounds from water pumps, air bubbles, air pumps, chiller motors, and

the like can combine to create a loud, cacophonous sound in a tank across a broad frequency

range (Bart et al., 2001; Davidson et al., 2007). What effect this has on fish health, growth,

courtship, and reproduction in the context of aquaculture has been explored only cursorily in the

literature.

Banner and Hyatt (1973) exposed eggs and larvae of Cyprinidon variegatus and Fundulus

similis to noise (at 118 peak decibels with reference to one micropascal, or dBpeak re: 1 μPa)

from a submersible water pump and airstones. Tested against controls in quiet tanks (at 103

dBpeak re: 1 μPa), they discovered greater mortality of eggs and fry of C. variegatus in noisy

tanks, and slower growth rates of fry in noisy tanks in both species. Lagardère (1982) exposed

brown shrimp (Crangon crangon) to noise (at 128 dBpeak re: 1 μPa) by aquarium air pumps

placed adjacent to culture tanks. Tested against controls in quiet tanks (at 88 dBpeak re: 1 μPa),

animals in noisy tanks demonstrated slower growth, less food consumption, reduced

reproduction (less egg-bearing females), and higher mortality due to higher rates of cannibalism

and higher incidence of disease. In a later study, Regnault and Lagardère (1983) also

documented increased oxygen consumption and ammonia excretion (suggesting increased

metabolism) among brown shrimp in loud tanks. In contrast, Wysocki et al. (2007) found no

20

effect of chronic exposure to loud tank noise (at up to 150 dBrms re: 1 µPa) on growth or

mortality of rainbow trout (Oncorhynchus mykiss).

Outside the context of aquaculture, it is clear that sound can affect fish at all levels of

biological organization, ranging from biochemical perturbations (e.g., Santulli et al., 1999), to

physiological changes (e.g., Sverdrup et al., 1994), to behavioral modifications (e.g., Pearson et

al., 1992). Most of this work has examined the effects of either laboratory-produced sounds, that

are unlikely to be heard in a natural or culture environment, or anthropogenic sounds

encountered in nature.

Effects on Hearing

Some of the earliest work has studied the effects of noise on the auditory system of fish.

Popper and Clarke (1976) used classic behavioral training techniques (avoidance conditioning,

threshold tracking, and classical conditioning of respiratory suppression) to show that goldfish

(Carassius auratus) exposed to intense tonal stimulation increased hearing thresholds; threshold

shifts varied with different frequencies of exposure and testing. Also using classical

conditioning techniques, Fay (1974) measured threshold increases of tones masked by broadband

noise.

More recently, workers have applied the auditory evoked potential (AEP) technique to

measure fish hearing. The AEP technique is a non-invasive far-field recording of synchronous

neural activity in the eighth nerve and brainstem auditory nuclei elicited by acoustic stimuli

(Jacobson, 1985). Originally invented and developed for use in clinical evaluations of human

hearing (Jacobson, 1985), the AEP technique has since been applied to many animals, including

mammals (e.g., Wolski et al., 2003) and fish (e.g., Yan, 2001). Scholik and Yan (2001a&b,

2002a&b) took advantage of the AEP technique to study the effects of noise on the auditory

sensitivity of the fathead minnow (Pimephales promelas), a hearing specialist (a fish that has

21

evolved a morphological specialization to increase hearing sensitivity, such as a bony or gaseous

vesicular connection between the gas bladder and the inner ear), and the bluegill sunfish

(Lepomis macrochirus), a hearing generalist (without such a morphological specialization).

Using this technique, they exposed both fish to white noise (a broadband noise in which all

frequencies in the noise spectrum are of the same sound pressure level) between 0.3 and 4.0kHz

at 142 dB re: 1μPa for the minnow and 0.3 to 2.0 kHz at 142 dB re: 1μPa for the sunfish for

various durations. In addition, they exposed minnows to two hours of boat engine noise, at a

sound pressure level of 142 dB re: 1 μPa. In noise-exposed minnows, auditory thresholds were

significantly elevated at several frequencies tested within the minnow’s most sensitive hearing

range. The auditory effect of noise exposure was dependent on duration of exposure. In

contrast, the sunfish demonstrated a slight, but insignificant, threshold increase. They concluded

that ambient noise in natural environments could have negative impacts on fish, but that effects

may differ between fish based on hearing sensitivity.

Fishes may still fail to detect biologically relevant sounds even if a temporary hearing

threshold shift (i.e., hearing loss) does not occur. Signals may simply be masked by the

presence of loud ambient noise. The masking effect in fishes has been recognized since 1961

(Tavolga, 1967). Among hearing specialist fishes, masking can be quite pronounced at very low

levels of noise. In goldfish, Fay (1974) documented an increase in signal detection threshold of

approximately 20 dB when a spectrum level masking noise of only 51 dB re: 1 µPa was present.

Masking also occurs in hearing generalist fishes, though the relatively poor hearing sensitivity of

generalists requires that masking noise be substantially louder to produce an effect. A few

studies demonstrated masking among representative hearing generalists (Lagodon rhomboides,

Caldwell and Caldwell, 1967, in Tavolga, 1974; Tilapia macrocephala, Haemulon sciurus,

22

Tavolga, 1974; Gadus morhua, Hawkins and Chapman, 1975; Sebastes schlegeli, Motomatsu et

al., 1998; Lepomis gibbosus, Wysocki and Ladich, 2005; Perca fluviatilis, Amoser and Ladich,

2005) at critical ratios (the difference between hearing threshold and spectrum level of masking

noise) ranging from 10 to 60 dB (median around 20 dB), with spectrum level masking noise

ranging from 70 to 90 dB re: 1 µPa and total power of masking noise ranging from 100 to 110

dB re: 1 µPa (where measured). Furthermore, broadband noises cause a more pronounced

masking effect than do narrowband noises (Hawkins and Chapman, 1975), thus the broadband

nature of ambient noise is of additional concern.

Masking may pose substantial problems for animals that rely on acoustic communication.

Many fish produce sounds during aggression, defense, territorial advertisement, courtship, and

mating (for an overview, see Zelick et al., 1999). Masking of these signals in aquaria may result

in, for example, excessive and unnecessary aggressive encounters (and subsequent social stress)

if acoustic territorial signals are not perceived (e.g., Myrberg and Riggio, 1985), inhibition of

species recognition (e.g., Spanier, 1979), or inhibition of phonotaxis, courtship, and/or mating

behavior (e.g., Stout, 1975; McKibben and Bass, 1998).

The Stress Response

Alternatively, ambient noise should be considered as a possible stressor in a culture

system. Stress is an important consideration in the successful husbandry of finfish. Severe stress

can result in mortality, but even sublethal stress can compromise various physiological and

behavioral functions, leading to suppressed disease resistance, growth rate, and fecundity, all

contributing to suboptimal production (Iwama et al., 1997).

The stress response is an evolutionarily adapted cascade of physiological and behavioral

changes that occur to enable the animal to react adaptively to a stressful stimulus, or to cope

long-term in a stressful environment (the latter scenario has been coined “allostasis,” essentially

23

an alternative to homeostasis in a suboptimal environment, by Sterling and Eyer, 1988).

Ultimately, if the persistence of a chronic stressor overcomes the ability of an animal to cope,

pathological conditions and mortality can ensue (per the General Adaptation Syndrome, or GAS,

Selye, 1950). The adaptive components of the stress response are mediated endocrinologically

via the secretion of catecholamines and cortisol, often considered as indices of a primary stress

response (Sumpter, 1997). Together, these hormones cause a suite of changes in the physiology

of the animal, considered as secondary stress indices. These include (but are not limited to)

changes in the matrix of ionic, osmotic, and acid-base components of the blood (McDonald and

Milligan, 1997), immune system depression (Balm, 1997), downregulation of reproductive

hormones, and variable changes in growth hormone plasma concentrations (Pankhurst and Van

Der Kraak, 1997). Tertiary, or whole-organism indices, include measures such as behavioral

changes (Schreck et al., 1997), and depression in growth, weight gain, fecundity, and survival of

offspring (Pankhurst and Van Der Kraak, 1997).

While acoustic stress has been suggested as a possible factor affecting aquaculture (e.g.,

Lagardère, 1982; Regnault and Lagardère, 1983; Wysocki et al., 2007), other literature has

supported anthropogenic noise as a cause of biochemical, physiological, and behavioral changes,

all markers of stress response (e.g., Blaxter and Batty, 1987; Knudsen et al., 1992; Skalski et al.,

1992; Sverdrup et al., 1994; Knudsen et al., 1997; Santulli et al., 1999).

Santulli et al. (1999) and Sverdrup et al. (1994) examined the effects of intense acoustic

stimulation emitted by air guns used in seismic surveys on biochemical and physiological

responses in European Sea Bass (Dicentrarchus labrax L.) and Atlantic Salmon (Salmo salar),

respectively. Santulli et al.’s (1999) post-shock serum analyses indicated increases in cortisol,

variations in glucose and lactate, and decreased adenylate concentrations. Skeletal muscle and

24

liver ATP concentration fell, ADP rose, while AMP did not significantly change. Sverdrup et al.

(1994) demonstrated an immediate post-shock decrease in cortisol followed by a consistent rise

to 40 nmol/L over pre-shock levels. Adrenaline levels spiked at 75 nmol/L over pre-shock

levels. Structurally, the vascular endothelium of the ventral aorta and the coeliaco mesenteric

artery (CMA) revealed signs of injury within the first 30 minutes after the experimental shock.

Functionally, the cholinergic and adrenergic vasoconstrictor responses in the CMA were

markedly reduced during the first day after the shock. The loss of structural integrity and the

reduced functional responses indicated a temporary impairment of the vascular endothelium in

response to the seismic shock.

Behavioral responses to sound stimulation have also been examined. Blaxter and Batty

(1987) demonstrated startle responses in herring (Clupea harengus harengus) when exposed to

transient sound stimuli. Knudsen et al. (1992, 1997) demonstrated spontaneous avoidance

responses of Atlantic and Pacific salmon (Salmo salar and Oncorhynchus tshawytscha,

respectively) in response to 10 Hz stimulation, hypothesizing acute awareness in the infrasound

range to the evolutionary importance of detecting swimming predators. Skalski et al. (1992)

demonstrated alarm and startle responses of rockfish (Sebastes spp.) to acoustic stimuli emitted

by geophysical survey devices, and tied these results to a decreased catch-per-unit effort.

The Seahorse: A Model

Seahorses (Hippocampus spp.) have received much attention in recent years in several

fields of study. Classical biologists have taken advantage of the genus’ unique, monogamous

mating system to test predictions involving sexual selection and sex roles (Vincent et al., 1992;

Vincent, 1994a&b; Masonjones and Lewis, 2000), while exploitation of seahorses in the

Traditional Chinese Medicine and aquarium markets (Lourie et al., 1999) has fueled interest in

25

seahorse aquaculture as a means to provide alternative sources to wild-caught animals (Lockyear

et al., 1997; Woods, 2000; Job et al., 2002; Woods, 2003).

Several detailed studies on seahorse courtship and mating behavior suggest that seahorses

use visual cues in mate choice and timing of mating. Seahorses pair size-assortively (Vincent

and Sadler, 1995; Jones et al., 2003), and may engage in ritualistic courtship behaviors to

synchronize reproductive states between paired individuals (Vincent, 1995).

However, the role of sound communication in seahorse pairing or courtship has remained

largely unstudied. Dufossé (1874, cited in Fish, 1953) first reported “a monotonous noise

analogous to that of a tambour, especially during the breeding season” in courting Hippocampus

brevirostris. Fish (1953) also provided two other accounts of seahorses (species unknown)

making clicking noises when placed in proximity to one another in separate jars, suggesting an

intraspecific signaling system. She first characterized the clicking noise made by H. hudsonius

(= erectus) when placed in a new aquarium environment. Her recordings indicated broadband

signals ranging from 0 to 4.8 kHz, with maximum energy between 300 to 600 Hz and 400 to

800 Hz. She hypothesized that sound production may be used in new surroundings for

orientation, and perhaps to find conspecifics. Fish (1954, cited in Marshall, 1966) later observed

clicking in courtship and mating. More recently, Vincent (1994b) observed H. fuscus “snapping”

in the context of male-male competition during courtship with a female, while Woods (2000)

observed clicking by H. abdominalis while rising in the water column, prior to (or perhaps

during) egg transfer. But clicking is not included among courtship behaviors described between

mating pairs of H. fuscus (Vincent, 1994b), H. whitei (Vincent and Sadler, 1995), or H. zosterae

(Masonjones and Lewis, 1996).

26

Colson et al. (1998) conducted a detailed study of sound production during feeding in H.

zosterae and H. erectus. They characterized a similar clicking sound, stridulatory in origin,

produced by an articulation between the ridge on the dorsal posterior region of the supraoccipital

and the medial groove on the anterior margin of the coronet. Peak frequencies of clicks

produced by H. erectus ranged from 1.96 to 2.37 kHz. Furthermore, they found that peak

frequencies of clicks were inversely correlated with body weight, at least in H. zosterae. The

higher range of peak frequencies (2.65 to 3.43 kHz) for H. zosterae, a much smaller species at 3

vs. 15 cm maximum height (Kuiter, 2000), further supports this inverse relationship. While this

clicking-sound was associated with feeding, the authors also observed this sound and related

behavior (a rapid, upward jerk of the head) when seahorses were placed into a new aquarium,

and among competing males during courtship.

From the perspectives of conservation and aquaculture, wild seahorse populations are

declining (Project Seahorse, 2009), prompting collection regulations (CITES, 2001).

Stakeholders are thus exploring aquaculture as an alternative to wild collection (e.g., Wilson and

Vincent, 1998). However, seahorses have proven particularly challenging to culture. Matsunaga

and Rahman (1998) provided evidence suggesting that seahorses do not have gut-associated

lymphoid tissue (GALT), an important component of adaptive immunity. Seahorses are also

particularly vulnerable to disease in aquaculture conditions (Vincent, 1998, Berzins and

Greenwell, 2001). Stressors such as crowding, handling (e.g., Barnett and Pankhurst, 1998), and

ambient noise (e.g., Lagardére, 1982) might further compromise the already immunologically

disadvantaged seahorse, potentially rendering it vulnerable to pathogens commonly encountered

in aquaculture conditions (e.g., Leonard et al., 2000).

27

The seahorse is thus an opportune model to measure the effects of ambient noise on the

health, welfare, and behavior of fish in an aquaculture setting. The vulnerability of wild seahorse

populations urges research and development of aquaculture techniques to improve production, in

the hopes of alleviating fishing pressure. Their documented and stereotypical courtship behavior

facilitates behavioral measurements of courtship and reproduction and the effects that noise

stress may have on them. Their unusual immune system may predispose them to stress-induced

disease, prompting the evaluation of stress response to chronic noise exposure. Seahorses are

known to produce sounds, but the function of sound production is still unclear in this group of

animals. This presents an opportunity to test sound production in the context of feeding (prey

capture) and courtship, among other behaviors, and, as it relates to ambient noise, potential

masking effects of ambient noise on sound production and communication.

Questions

Does ambient noise impact fish health, behavior, and physiology in aquaculture settings?

If so, are effects mediated via a stress response that may be manifested in measures of health,

behavior, or physiology? Or, does ambient noise mask acoustic signals that may be important

components of behavior in fishes?

Objectives and Outline

The objectives of this dissertation are severalfold. In Chapter 2, I seek to characterize the

range of ambient noise to which captive seahorses are exposed in aquaria, to ascertain the effect

of tank design elements on ambient noise, and to explore potential correlations between ambient

noise and seahorse morbidity or mortality. In Chapter 3, I seek to 1) characterize the hearing

range of the lined seahorse, H. erectus, 2) characterize the acoustic properties of the seahorse

click, and 3) assess audition of conspecific sound production. In Chapter 4, I explore the effects

of chronic loud noise exposure on physiological and behavioral measures of stress in H. erectus.

28

In Chapters 5 and 6, I assess the functions of sound production (clicking) on prey capture and

courtship behavior, respectively, the effect of chronic loud noise exposure on these behaviors,

and explore alternative hypotheses explaining the mechanism of any observed effect: 1) that

ambient noise may contribute to reductions in these behaviors via stress response, or 2) that

ambient noise may mask acoustic signaling, disrupting the acoustic component of these

behaviors.

Contribution to Science and Industry

This dissertation addresses questions that are important and interesting to the fields of

ethology and aquaculture. Behaviorally, this proposal addresses an important umwelt question.

To understand the biological systems of an animal, it is important to study how an animal

perceives its environment. Failing to do so could mean the failure to detect information that an

animal is processing to make adaptive decisions. This shortcoming can hamper the

understanding of signaling and communication systems, sexual selection, foraging, prey capture,

and orientation (Wersinger and Martin, 2009).

Failing to understand an animal’s umwelt may have important consequences in

application/industry. The field of aquaculture relies on animals to reproduce naturally (or

sometimes induced with hormones). While efforts are made to provide an environment

conducive to spawning, some operators may be overlooking the acoustic environment. Breeding

tanks may be exposed to sound and vibration from heavy circulation and filtration equipment.

The underwater cacophony that results may either mask acoustic signals that are important

components of behaviors such as courtship and prey capture, or may be stressing fish, resulting

in reduced health, growth, and reproduction.

Results of this dissertation may prompt aquaculturists to consider the acoustic environment

of their systems, and to reduce ambient noise where possible via system design modifications.

29

The results presented here may be useful not only to the stakeholders located across the globe

who have recently begun culturing seahorses to provide an alternative source for product in the

traditional Chinese medicine and aquarium markets, but also to culturists of all types of aquatic

animals found worldwide.

30

CHAPTER 2 THE SEAHORSE SOUND SURVEY

Introduction

Given the effects of chronic noise exposure on fish hearing, behavior, and physiology, it is

important to understand whether or not ambient noise in aquarium/aquaculture conditions is loud

enough to induce such deleterious effects. However, ambient noise levels commonly

encountered in aquarium and aquaculture settings remain largely uncharacterized.

Bart et al. (2001) surveyed ambient noise in several different aquaculture settings; indoor

circular fiberglass and concrete tanks, indoor concrete and wooden-frame raceways, and outdoor

ponds. Of all enclosure types, ponds without running aerators were the quietest at a total power

of 94 dB re: 1 µPa, but pond aerators increased total power to 135 dB. In a frequency range of

25-1000 Hz, concrete tanks were typically more quiet (at 110 dB SPL re: 1 µPa total power) than

fiberglass tanks (at 130 dB). Some of these SPLs clearly fall into a range known to induce

hearing loss and stress, though physical inner ear damage occurs at higher SPLs (e.g., 180 dB

SPL re: 1 µPa, Hastings et al., 1996; McCauley et al., 2003).

Here, I characterize the range of ambient noise to which seahorses (Hippocampus spp.,

Family Syngnathidae) are exposed in public aquaria. This survey offers data on an assortment of

smaller tanks more likely to be in use among aquarists and culturists of ornamental fishes. It

includes summarized data on the variety of tank sizes, materials used for tanks and stands, as

well as drainage, aeration, and associated noise-producing life support equipment. Animal

collections, health, and reproduction records are summarized. Ambient noise profiles are

characterized, and relationships between tank design specifications and ambient noise levels are

explored and presented. Finally, I present ambient noise data collected from geographic areas in

31

Tampa Bay, FL (USA) within which wild populations of H. erectus occur, to provide a

comparison between noise encountered in the wild and noise encountered in captivity.

Materials and Methods

Aquarium Data Collection

The Seahorse Sound Survey was a kit developed for participation by public aquaria

throughout the United States. The kit contained a questionnaire, an HTI-96min pressure-

sensitive hydrophone (High Tech Instruments, Inc., sensitivity: -165 dB re: 1 V/µPa,

bandwidth: 2-30,000 Hz), a Creative NOMAD Jukebox 3 digital audio recording device, a

digital camera, an instruction manual, headphones and measuring tape (Figure 2-1). The

questionnaire requested the following data:

• Contact Information

• Tank Specifications: Volume, dimensions, tank wall and stand materials, substrate type

• Aeration: Air blower vs. pump, type, number of air endpoints, type of air endpoint

• Drainage: Number, location

• Heaters/Chillers: Types, distance from tank, method of plumbing/attachment

• Pumps: Types, distance from tank, method of plumbing/attachment

• Other Noise Sources: Types, distance from tank, method of plumbing/attachment

• Animal Collection: Species, number of males, females, juveniles, unknown sex, and breeding pairs

• Breeding Records: Number of broods per pair per year, number of fry

• Health & Disease Records: Type and number of sick and/or dead animals per year

In addition to completing the questionnaire, participating aquarists were asked to take two

1 min recordings with the hydrophone in two positions: 1) in the middle of the water column,

and 2) touching the tank bottom. During recording, the hydrophone was connected to the 9-volt

32

battery amplifier, and to the NOMAD Jukebox 3. Aquarists were asked to hold the NOMAD

Jukebox 3 and excess cord still during recording, and not to allow recording equipment to come

in contact with any other solid during recording. Recordings were typically taken with a 0 dB

gain, but gain was adjusted (+ up to 12 dB) if ambient noise was too quiet or too loud at 0 dB

gain, and corrected in calibration for subsequent analysis. Recordings were digitized as .wav

files with a sampling rate of 44.1 kbps.

Finally, photos were taken of tanks and systems to corroborate questionnaire data.

Data Collection in the Wild

On July 15, 2008, acoustic recordings were taken from ten sites within and around Tampa

Bay, FL (USA) at GPS locations where trawls had previously collected 3 or more seahorses,

between the years of 2006 and 2007 (N. Dunham, Florida Fish and Wildlife Conservation

Commission, pers. comm., Figure 2-2). Acoustic recording equipment consisted of an HTI-96

min hydrophone (specs as above) with a 17 m cable and the Creative NOMAD Jukebox 3 digital

audio recording device. At each recording site, Beaufort number (Wenz, 1962) and depth were

recorded for characterization of environmental condition and site, respectively. The hydrophone

was lowered overboard until resting on the seafloor, and 1-min acoustic recordings were made

with the Jukebox set at 0 dB Gain. The boat engine was shut off during recordings, and the cable

was held still to prevent acoustic artifacts due to jostling of the cable/hydrophone.

Sound Analysis

Calibration

All recordings were analyzed with SpectraPlus (Pioneer Hill Software) signal analysis

software. For calibration, I generated a 500 Hz sine wave at 1 Vpeak with a Leader LG1301 2

MHz function generator connected to a Leader LS1020 20 MHz oscilloscope for signal

visualization. The function generator was also connected to the NOMAD Jukebox 3, and

33

recordings of the signal were made at all gain levels, and digitized as .wav files with a sampling

rate of 44.1 kbps. Calibration setting files in SpectraPlus were created using these known-

voltage signals, according to program instructions, for all gain levels of the NOMAD Jukebox 3.

Sound processing

Sound files were post-processed by removing putative artificial electrical peaks in

frequency spectra (at 60 Hz or its harmonics) using the Fast Fourier Transform (FFT) notch filter

function in CoolEdit (Syntrillium Software Corporation). These notch-filtered files were then

post-processed in SpectraPlus using the analysis settings summarized in Table 2-1. I chose a

decimation ratio1 of 22 along with an FFT-size of 2048 to achieve a spectral line resolution of

approximately 1 Hz (0.979 Hz) and an upper frequency detection limit of 1,002 Hz. These

settings enable accurate spectrum level SPL measurements and output total power SPLs that are

summed over a frequency range of 2-1,002 Hz. Hearing generalist fishes, without connections

between the gas bladder and inner ear, generally have insensitive hearing above 1,000 Hz (Fay,

1988b). Hearing tests of H. erectus using the auditory evoked potential (AEP) technique are

consistent with this observation (Chapter 3). In addition, there are no observable connecting

structures between the swimbladder and the inner ear in this species (pers. obs.). Both lines of

evidence thus suggest that seahorses, or at least H. erectus, are/is (a) hearing generalist(s). In

their survey of aquaculture systems, Bart et al. (2001) also analyzed recordings in the low

frequency region (25-1000 Hz), that provides an equivalent frame of reference for comparison.

It should be noted that a low pass filter reduces the total power, but the resulting total power SPL

is summed only over the fish’s range of environmentally relevant hearing ability. Unless

1 Decimation reduces the sampling rate of the signal by averaging (in this case) every 22 samples as one sample. In combination with an FFT-size of 2048, this procedure improves the frequency resolution of the power spectrum to a 1 Hz bandwidth.

34

otherwise indicated, all amplitudes reported in the results are peak amplitudes in deciBels with

reference to 1 micropascal (dB SPL re: 1 µPa).

Statistical Analysis

Questionnaire data were tabulated and descriptive statistics were computed. Power spectra

of all ambient noise recordings were generated; from these, I obtained spectrum level SPLs at 10,

20, 30, 40, 50, 70, 80, 90, 100, 200, 400, 500, 700, 800, and 980 Hz, and recorded peak

frequencies, peak amplitudes, and total power SPLs. These data were tabulated and descriptive

statistics computed. I conducted a paired t test between total power SPLs at the middle of the

water column and at the tank bottom, and unpaired t tests between total power SPLs of wild

ambient noise recordings and tank recordings at both positions. To examine possible trends

between tank/filtration design variables and ambient noise, or animal breeding/health records and

ambient noise, I constructed scatterplots setting total power SPLs on the y-axis. I fit regression

lines to plots with continuous x-variables. For ordinal or categorical x-variables, plots were

visualized for trends. I also conducted unpaired t tests for mid-level and bottom-level total

power SPLs between tanks containing adult males and female seahorses that produced at least

one brood and tanks that did not produce broods. Only stand material type, tank wall material

type, and substrate type revealed possible visual trends among all constructed plots. These three

variables were tested in a three-way MANOVA using SAS software according to the following

model:

y1 + y2 = w + s + b + w*s + w*b + s*b + w*s*b + ε (2-1)

where: y1 = mid-level total power (dB SPL re: 1 µPa), y2 = bottom-level total power, w = wall

material type, s = stand material type, b = bottom habitat type, and ε = experimental error. All

three independent variables are class variables. Data were not balanced, so Type III sums of

squares were used. For MANOVA results, the Wilks’ Lambda statistic was used to assess

35

significance. Within component ANOVAs, Tukey’s test comparisons were made among

treatment means of significant design variables. Significant interaction effects were further

explored using PROC SORT to test simple effects of one factor at fixed levels of the other,

followed with Tukey’s test comparisons among sorted treatment means.

Results

Questionnaire Data

Nine public aquaria returned data from 42 tanks (Table 2-2). Results of questionnaire data

described below may be cross-referenced with Table 2-3.

Tank dimensions, materials, and habitat

Tank sizes ranged from 19 to 10,978 L, with a median of 291 L. Most tanks were

rectangular (n = 31), others were cylindrical, hexagonal, other polygonal, crescent, bubble or

bow-shaped. Tank walls were constructed of fiberglass (n = 14), glass (n = 13), acrylic (n = 11),

or concrete (n = 3). Stands on which tanks were supported were constructed of plastic (n = 15),

wood (n = 13), metal (n = 9), or concrete (n = 4). The bottom habitat of tanks were primarily

gravel (n = 26), but some tanks had bare bottoms (n = 9), while others had a plenum underneath

gravel beds (n = 6).

Aeration and drainage

Twenty-two tanks were aerated. Seventeen tanks were aerated by a remote blower, three

by air pump (2 tanks had unreported sources). Of aerated tanks, 13 had one air endpoint, 8 had

two air endpoints, and one had four endpoints. Twelve of the aerated tanks had open end airlines

in the tank, and ten had airstones. Most tanks had one drain (n = 29), six had two drains, one

tank had three drains, and five tanks had no drainage. Of the tanks with drainage, most (n = 32)

were surface skimmers, and just four were subsurface drains.

36

Life support equipment

Twenty-nine tanks were serviced by heaters or heat exchange units that were either

plumbed inline (n = 12) at 0.6 + 0.3 m (mean + SE) from the tank, installed in the sump (n = 11),

or in the tank itself (n = 6). Only four tanks were serviced by chillers, either plumbed inline or

located in the sump, at a median of 3 m from the tank (range: 0.5 to 7.6 m). Eighteen tanks were

serviced by one water pump, nineteen were serviced by two water pumps, and two by three water

pumps (or had multiple water pumps servicing one system with multiple tanks). Wattage of

primary pumps ranged from 20 to 1492 W, with a median of 265 W. Thirty-five primary pumps

were plumbed inline using PVC, one plumbed with vinyl tubing, one submersed in the sump, and

two submersed in the tank. Primary pumps were located up to 7.6 m away from tanks, with a

median distance of 0.9 m. Wattage of secondary pumps ranged from 25 to 746 W, with a

median of 322 W. Fourteen were plumbed inline using PVC, six were plumbed inline with vinyl

tubing, and one was submersed in the tank. Secondary pumps had the same range and median

distance as primary pumps. Tertiary pumps for two tanks were 218.5 W pumps plumbed in-line

with PVC and located 0.3 m from the tank.

Other noise sources

Other noise sources not associated with tanks included pumps, chillers, television and

audio, and drainage overflows. These noise sources were located between 0.3 to 3 m from tanks,

with a median distance of 1.5 m.

Animal collection

Hippocampus species kept included H. abdominalis, H. erectus, H. fuscus, H. kuda, H.

procerus, H. reidi, H. subelongatus, and H. zosterae. Only five tanks kept more than one

Hippocampus species. Tanks held anywhere from zero to 80 individuals, with a mean of 8.5.

Tanks containing adults were stocked to a mean density of 19 + 3 adults/m2. Average sex ratio

37

(M:F) of animals of known sex was 0.97. Ten tanks housed between 2 and 54 juveniles, with a

median of 7. Tanks containing juveniles were stocked to a mean density of 38 + 14 juveniles/

m2. Some tanks also held a variety of other aquatic animals, including other syngnathids (n =

10), non-syngnathid fishes (n = 8), and invertebrates (n = 12). Twenty-two tanks housed only

Hippocampus spp.

Health and reproduction records

Annual mortality ranged from 0 to 100 %, with an average of 22.7 + 4.5 %. Sixteen tanks

held seahorses that mated and produced broods; of these, breeding frequency ranged from 0.1 to

6 broods per pair per year, with a mean of 1.6 + 0.3. Total power SPLs of tanks containing adult

males and females that produced at least one brood were not significantly different than total

power SPLs of tanks that did not produce broods at either the middle of the water column (123.2

+ 1.2 vs. 125.4 + 1.0 dB respectively, p = 0.143) or the tank bottom (130.9 + 1.3 vs. 132.8 + 2.6

dB respectively, p = 0.231), but sample sizes were small (15 and 6 tanks, respectively).

Sound Analysis

Mid-water level recordings

At the middle of the water column, spectrum level SPLs ranged from 54.6 to 123.9 dB

across the frequency range (Figure 2-3a, Table 2-4). Spectrum levels were highest from 10-30

Hz, with a mean spectrum level SPL of 109.0 dB. Between 40 to 100 Hz, spectrum level SPLs

averaged 93.6 dB. As frequencies increase from 200 to 980 Hz, spectrum level SPL gradually

declined from 84.0 dB to 71.8 dB. Total power of tanks ranged from 116.3 to 142.9 dB, with a

mean of 126.1 + 0.8 dB. Peak frequencies were very low, predominantly located between 7.8

and 25.5 Hz (interquartile range), with a median of 9.8 Hz. Peak amplitudes ranged from 107.4

to 140.2 dB, with a mean of 120.1 + 0.9 dB.

38

Bottom level recordings

At the bottom of the tanks, spectrum level SPLs ranged from 57.2 to 129.8 dB across the

frequency range (Figure 2-3b, Table 2-4). Spectrum level SPLs were highest between 10 and 30

Hz with a mean SPL of 113.3 dB. Between 40 and 100 Hz, spectrum level SPLs averaged 104.4

dB. Spectrum level SPLs dropped precipitously from 102.0 dB at 100 Hz to 81.6 dB at 500 Hz,

and leveled off to 77.3 dB by 980 Hz. Total power of tanks ranged from 122.3 to 146.1 dB, with

a mean of 133.7 + 1.1 dB. Peak frequencies were higher than in the middle of the water column,

predominantly located between 9.8 and 56.8 Hz (IQ range), with a median of 54.8 Hz. Peak

amplitudes ranged from 113.4 to 144.5 dB, with a mean of 127.8 + 1.3 dB.

Wild ambient noise recordings

Ten sites had a median Beaufort number of 2 (range 1-3) and a mean depth of 3.9 + 0.5

m (range 1.2 to 6.7 m). Spectrum level SPLs ranged from 64.9 to 121.9 dB across the frequency

range (Figure 2-3c, Table 2-4). Spectrum level SPLs were highest at 10 Hz with a mean SPL of

105.3 + 4.2 dB. Spectrum level SPLs declined most precipitously as frequencies increased to

100 Hz, at which point spectrum level SPLs averaged 86.1 dB + 3.3 dB, then further declined

gradually to 76.1 + 1.5 dB by 980 Hz. Total power ranged from 103.3 to 132.6 dB, with a mean

of 119.6 + 3.5 dB. Peak frequencies were very low, predominantly located between 7.1 and 13.5

Hz (IQ range), with a median of 9.3 Hz. Peak amplitudes ranged from 87.2 to 123.4 dB, with a

mean of 109.0 + 4.4 dB.

Ambient Noise Comparisons

Total power SPLs of bottom-level recordings were significantly louder than mid-level

recordings among aquaria (p < 0.0001). Total power SPLs of ambient noise in the wild were not

significantly different from total power SPLs of ambient noise in aquaria at the middle of the

39

water column (p = 0.095), but were significantly quieter than total power SPLs of ambient noise

in aquaria at tank bottom (p = 0.002).

Effects of Design Variables on Total Power

Wall material, bottom habitat type, and their interaction had significant effects on total

power (F6,40 = 3.69, p = 0.0052; F4,40 = 5.36, p = 0.0015; and F2,20 = 12.73, p = 0.0003,

respectively; Figure 2-4). Wall material, bottom habitat type, and their interaction had

significant effects on mid-level power (F3,21 = 3.79, p = 0.0256; F2,21 = 11.03, p = 0.0005; and

F1,21 = 23.04, p < 0.0001, respectively), whereas only bottom habitat type had significant effects

on bottom-level power (F2,21 = 5.43, p = 0.0126). For mid-level recordings, glass tanks were

significantly louder (at a mean of 128.4 dB) than acrylic (123.6 dB) and concrete (122.5 dB)

tanks, but not fiberglass (126.4 dB) tanks. For mid-level and bottom recordings, bare-bottom

tanks (at a mean of 131.0 dB at mid-level and 139.7 dB at bottom) were significantly louder than

gravel-bottom (125.0 dB and 132.2 dB respectively) and plenum (122.4 dB and 130.2 dB

respectively) tanks.

Among the simple effects contributing to the interaction effect of wall material*bottom

habitat type at mid-level recordings, one bare bottom tank at 139.9 dB was significantly louder

than gravel bottom tanks at 120.9 + 1.2 dB when tank walls were composed of acrylic (F1,21 =

47.61, p < 0.01). There was significant variation in mid-level power among wall material types

of tanks with gravel bottoms (F3,21 = 8.00, p < 0.01, Figure 2-5). There was significant variation

in mid-level power among wall material types of tanks with bare bottoms (F1,21 = 21.46, p <

0.01). Here, one acrylic tank at 142.9 dB was louder than one glass tank at 130.7 dB and six

fiberglass tanks at 129.1 + 0.7 dB (there were no bare bottom tanks with concrete walls).

40

Discussion

These results demonstrate a wide range of ambient noise to which seahorses are exposed in

public aquaria. The loudest tanks demonstrate total power SPLs that are 4.4 and 3.9 times louder

than the quietest tanks at the middle of the water column and the tank bottom, respectively (as a

6 dB increase is a doubling of pressure). These results are generally consistent with and in the

range of Bart et al.’s (2001) tank data, despite the fact that the tanks measured in the Seahorse

Sound Survey were smaller overall.

Noise at the bottom of the tank was significantly louder than in the middle of the tank.

This pattern was also documented in Wysocki et al.’s (2007) thorough characterization of the

ambient noise field in a rectangular tank, but is in contrast to Bart et al.’s (2001) results that

showed no significant differences among tank recording locations. However, Bart et al.’s (2001)

hydrophone never touched any surfaces, whereas in this study, bottom recordings were taken

with the hydrophone touching the bottom habitat of the tank. Ambient noise was studied at both

mid-level and bottom surface locations to (1) demonstrate the variation in sound pressure at

different locations in the tank, and (2) to generate hypotheses about the differences in ambient

noise to which fish are exposed, based on their positioning in the water column. Fishes that are

pelagic or spend most of their time in the water column are most likely to be exposed to the

lower levels of sound measured in this region of the tank. However, there are many bottom-

dwelling fishes (e.g., catfish, eels, hawkfish, jawfish, loaches, grouper, seahorses, etc.) that spend

much of their lives in contact with the substrate. Loud ambient noise from the tank bottom might

transfer efficiently through spines, rays, or bones in contact with the bottom habitat to the

skeleton, skull, and inner ear. Sound transfers more efficiently between materials of similar

impedance; as such, solid-to-solid transfer is more efficient than liquid-to-solid transfer. It

41

would be worthwhile to study this hypothesis further and to characterize, compare and contrast

perceived sound pressure levels among water-column vs. bottom dwelling fishes.

This distinction is especially important in light of ambient noise levels in the middle of the

water column of aquaria not being significantly different from levels in the wild, but tank

bottoms are significantly louder. Thus, benthic dwelling animals such as the seahorse are

chronically exposed to louder ambient noise in a tank environment than they would normally be

exposed to in the wild, but the ambient noise to which mid-water column dwelling fishes are

exposed in tank environments are similar to ambient noise levels in the wild.

These results suggest that some ambient noise levels can be loud enough to mask hearing

in fishes. Masking noise levels discussed in Chapter 1 are within the range of ambient noise

encountered in the Seahorse Sound Survey. At the middle of the water column and in the

seahorse’s best hearing range (100-400 Hz, Chapter 3), spectrum level ambient noise ranged

from 60 to 107 dB, with a mean of 85 dB. At tank bottoms, spectrum level ambient noise in the

same frequency window ranged from 69 to 122 dB, with a mean of 94 dB. Mean total power

SPLs were also well above 100-110 dB. Furthermore, broadband noises cause a more

pronounced masking effect than do narrowband noises (Hawkins and Chapman, 1975), thus the

broadband nature of ambient noise is of additional concern. Masking, then, is likely to be

common for both hearing specialists and generalists in many aquarium environments.

Though sound production in diverse representative fishes among actinopterygians is

widely reported, very few publications report SPLs. Table 2-5 lists spectrum level SPLs at peak

or dominant frequencies of calls from several marine and freshwater fishes. Calls range from 90

to 126 dB SPL re: 1 µPa. Masking studies indicate critical ratios averaging around 20 dB; thus,

calls must be at least 20 dB louder than spectrum level ambient noise to be detected. It is clear,

42

then, that quieter calls (from H. erectus, P. martensii, and the drumming sounds of several

otophysans) are at risk of being masked by typical ambient noise in aquaria. Other species (e.g.,

M. undulates, O. tau, C. nebulosus, T. vittata, B. modesta) may be able to overcome the critical

ratio in an aquarium with mean levels of ambient noise. However, signals can be designed in

other ways to increase the likelihood of detection. For instance, a broadband sound (e.g.,

knocking or stridulation) is more likely to be detected than a sound at the same spectrum level

SPL limited to a narrow frequency band (tone, Egner and Mann, 2005). Conversely, maximizing

peak energy within the most sensitive hearing range of fishes is another strategy evident in

seahorses (Chapter 3) and gobies (Lugli et al., 2003), for example. Sudden onset (such as the

seahorse stridulation, Chapter 3), repetition (such as the pulse train of gobies, Barimo and Fine,

1998; Lugli et al., 2003), and increased duration (such as the plainfin midshipman boatwhistle,

Bass and McKibben, 2003) also increase likelihood of detection (Dusenbery, 1992; Bass and

Clark, 2003). Direct hearing tests of masking of conspecific signals by aquarium ambient noise

(using, for example, the auditory evoked potential method) are warranted.

Prolonged exposure to loud ambient noise can lead to hearing loss that persists even after

noise is removed. Temporary threshold shifts (TTS) have been documented for hearing

specialists (e.g., Popper and Clarke, 1976; Scholik and Yan, 2001a&b, 2002b; Amoser and

Ladich, 2003; Smith et al., 2004) when exposed to noise with spectrum level SPLs as low as 134

dB re: 1 µPa that persist for up to two weeks or longer, but a hearing generalist did not exhibit

TTS when exposed to noise with a spectrum level of 142 dB re: 1 µPa (Scholik and Yan, 2002a).

In either case, the ambient noise levels of most aquaria are lower than those required to induce

TTS even among hearing specialists; thus TTS is not likely to be of concern for fishes in aquaria.

43

The potential for fish ear damage from exposure to ambient aquarium noise is also

unlikely. Damage was documented after exposure to 180 dB re: 1 µPa (spectrum level and total

power, respectively, per Hastings et al., 1996 and McCauley et al., 2003), but not at 140 dB

(spectrum level, Hastings et al., 1996). These values are far above ambient noise levels

encountered in the Seahorse Sound Survey.

It is clear, however, that the ambient noise encountered in some aquaria can induce a

chronic stress response in fishes. As detailed in Chapter 4, I demonstrate a chronic stress

response (evidenced by behavioral differences, reduced growth and body condition and altered

leukocyte differential profiles) among H. erectus housed in tanks with total power SPLs of 123

dB re: 1 µPa at the middle of the water column and 137 dB at the bottom. These are

approximately 3 dB below and above (respectively) mean ambient noise levels encountered

among public aquaria. Chronic stress is likely endured by seahorses in many captive aquaria at

or above mean ambient noise levels. Given the high mortality rate suffered by seahorses in

public aquaria (22.7%), it is critical to minimize stimuli that induce chronic stress, in order to

improve health, welfare, and immunocompetence.

Questionnaire data revealed a variety of system designs. The variability inherent in a

survey of this nature may obscure trends associated with any one particular variable, hence, the

majority of design variables did not reveal patterns in relation to ambient noise. It is notable,

however, that wall material type and bottom habitat type had significant effects on ambient

noise, despite the variability in tank/system design among surveyed tanks. My results are

generally consistent with Bart et al. (2001), who found lowest SPLs among concrete tanks and

highest SPLs in fiberglass tanks (acrylic and glass tanks were not measured in their study).

44

Reverberations (described by Yost, 1994, in Akamatsu et al., 2002, as the persistence of

sound in an enclosed space as a result of multiple reflections after sound generation has stopped)

and resonance (the tendency of a system to oscillate at maximum amplitudes at certain

frequencies) are acoustic phenomena that have long been recognized in small tanks (Parvulescu,

1964, 1967). Together, these phenomena tend to increase sound duration, frequency distribution,

and amplitude of signals (Akamatsu et al., 2002). For captive fishes in aquaria, this poses two

potential problems: ambient noise may be amplified, and conspecific signals may be severely

distorted, as vividly demonstrated by Yager’s (2002) recordings of the click of a pipid frog in

pond vs. small tank environments. In effect, signal distortion poses the same problems to social

behavior in fishes as does masking in small tanks, as discussed earlier. A troublesome

consequence of this phenomenon is the questionable results it may yield for behavioral acoustic

experiments in laboratory aquaria (Akamatsu et al., 2002).

Based on these results and corroborating literature, aquarists and aquaculturists are advised

to choose tanks with concrete or acrylic walls, to provide substrate, and to choose tanks with

minimum resonant frequencies above the hearing range of resident fishes. Akamatsu et al.

(2002) provide equations for calculating minimum resonant frequencies of rectangular and

cylindrical tanks. Based on their equations, rectangular tanks up to 4,000 L have minimum

resonant frequencies above 1,000 Hz; tanks must be a volume of approximately 38,000 L before

minimum resonant frequencies fall into the range of 500-600 Hz. Thus, resonance is not likely

to be a problem in small tanks for hearing generalist fishes, that demonstrate lowest hearing

thresholds below 500 Hz (Popper et al., 2003). In contrast, for hearing specialists, some of

which have sensitive hearing up to 2000 Hz and less sensitive hearing up to 4,000 Hz (Popper et

al., 2003), rectangular tanks as small as 570 L may have minimum resonant frequencies that fall

45

into their range of best hearing sensitivity (2,000 Hz), with very small tanks (38 to 570 L)

exhibiting minimum resonant frequencies in the range of 2,000 to 4,000 Hz. The point here is

that tank size can be chosen to avoid resonant noise in a fish’s range of hearing.

Davidson et al. (2007) make sensible recommendations to reduce sound impinging upon

fiberglass tanks. Additionally, I suggest the following, based on trials in my laboratory: Choose

a smaller, less powerful, and thus quieter pump to perform filtration and circulation; place the

pump on the floor or a surface that is not directly in connection with the tank (such as on the

shelf of the tank stand); move noisemakers (such as airstones or powerheads) to the sump of the

tank; install subsurface drains (thereby minimizing noise associated with water movement at the

surface); and install soundproofing loops in flexible plumbing lines (per A. Noxon, Acoustic

Sciences Corporation, pers. comm.). Soundproofing loops decouple sound and vibration

traveling through the water and pipe walls from the pump end to the tank end. Loops must

“float,” they may be supported by a bungee cord tied to a wall, for example, not in connection

with the tank, and the ends of the loops cannot touch. Finally, furnish an aquarium with

substrate, rockwork, and other habitat to absorb reverberant soundwaves in tanks. These

common sense design modifications may reduce ambient noise in tank environments and

ameliorate some consequences of loud noise exposure to tank inhabitants.

46

Table 2-1. Spectra Plus processing settings. Scaling……………………………………........ Sampling Rate (Hz)………..………………......

Logarithmic, peak amplitude 44100

Sampling Format……………..……………….. 16-bit, Stereo Standard Frequency Weighting…. …………… Flat (none) Decimation Ratio……………………………... 22 Frequency Limit………………………………. 1002 Hz,

Low-pass filter enabled FFT Size………………………......................... 2048 Spectral Line Resolution…………………….... 0.979 Hz Smoothing Window…………………………... Hann Averaging Settings……………………………. Infinite, Linear,

Disable Peak Hold FFT Overlap…………………………………... 0 Time Resolution………………………………. 1.022 seconds Input Signal Overload…………........................ Enable Overload Detection

Exclude Overloaded Data from Processor

Table 2-2. Public aquarium participants of the Seahorse Sound Survey Dallas World Aquarium………………………. Kingdom of the Seas Aquarium………………. Maritime Aquarium…………………………… The Aquarium at Moody Gardens……………. New England Aquarium………………………. Pittsburgh Zoo and PPG Aquarium…………... Ripley’s Aquarium of the Smokies…………… Riverbanks Zoo……………………………….. Tennessee Aquarium…………………………..

Dallas, TX Omaha, NE Norwalk, CT Galveston, TX Boston, MA Pittsburgh, PA Gatlinburg, TN Columbia, SC Chattanooga, TN

47

Table 2-3. Frequency distribution of tank design specifications and animal inhabitants. n represents number of tanks. Tank Dimensions, Materials, and Habitat

n

Aeration and Drainage n

Life Support Equipment n

Other Noise Sources & Animal Collection n

Tank Shape Rectangle Other polygon Bow-shaped Cylinder Bubble Crescent Hexagon Tank Wall Material Fiberglass Glass Acrylic Concrete Stand Material Plastic Wood Metal Concrete Bottom Habitat Gravel Bare Plenum

31 3 2 2 1 1 1

14 13 11 3

15 13 9 4

26 9 6

Aeration Remote Blower Air Pump Unreported # Air Endpoints One Two Four Air Endpoint Type Open End Airstones # Drains None One Two Three Drain Type Surface Skimmer Subsurface

1732

1381

1210

52961

324

Heater Installation Plumbed In-Line Sump Tank Chiller Installation Plumbed In-Line Sump # Pumps One Two Three 1st Pump Installation PVC Pipe Tank Submersion Sump Submersion Vinyl Tubing 2nd Pump Installation PVC Pipe Vinyl Tubing Tank Submersion 3rd Pump Installation PVC Pipe

12116

22

18192

35211

1461

2

Other Noise Sources Pumps TV/Audio Chillers Drainage Overflows Fans Air Piston Animal Collection H. erectus H. reidi H. abdominalis H. fuscus H. subelongatus H. zosterae H. kuda H. procerus Other sygnathids Non-sygnathid fishes Invertebrates

2442211

1210852211

108

12

48

Table 2-4. Descriptive statistics of spectrum level SPLs, peak amplitude, total power (all in dB SPL re: 1 µPa), and peak frequency (in Hz).

Mid-Level Recordings Bottom-Level Recordings Wild Recordings Hz Min Max Mean + SE Min Max Mean + SE Min Max Mean + SE

10 88.67 123.75 113.23 + 1.31 92.27 129.81 115.22 + 1.49 82.07 121.91 105.29 + 4.1820 94.21 123.88 107.92 + 1.28 100.80 129.18 111.93 + 1.26 74.67 118.78 98.07 + 5.3730 91.40 121.02 105.93 + 1.28 101.65 129.11 112.66 + 1.26 73.21 115.80 94.77 + 5.2440 82.76 107.28 97.67 + 0.89 93.83 119.29 106.24 + 1.10 68.03 110.97 92.89 + 4.8850 82.76 116.25 97.95 + 1.15 95.82 125.90 108.00 + 1.44 69.12 108.33 90.35 + 4.1970 75.74 107.83 93.24 + 1.18 90.61 120.84 105.09 + 1.41 66.79 103.61 87.67 + 3.7980 76.15 110.69 91.33 + 1.34 89.81 120.55 103.05 + 1.37 65.75 101.46 86.78 + 3.6190 73.54 106.15 90.50 + 1.32 89.27 118.25 101.96 + 1.30 66.24 100.91 87.13 + 3.25

100 76.10 105.51 90.70 + 1.25 85.40 121.75 101.94 + 1.36 64.90 99.09 86.05 + 3.34200 66.44 106.98 83.98 + 1.42 78.62 118.27 94.61 + 1.39 67.42 96.57 82.59 + 2.90400 60.34 102.35 80.14 + 1.72 69.45 117.64 86.76 + 1.79 70.54 89.75 80.33 + 2.33500 60.83 101.68 76.57 + 1.65 64.07 111.29 81.62 + 1.82 70.85 88.18 79.68 + 1.98700 54.55 91.70 73.66 + 1.60 57.65 105.14 80.27 + 1.67 69.60 84.08 76.97 + 1.63800 56.97 97.42 72.58 + 1.65 57.66 104.22 78.37 + 1.85 70.46 83.39 77.05 + 1.33980 54.63 98.46 71.77 + 1.94 57.15 96.64 77.30 + 1.70 68.57 83.65 76.12 + 1.54

Peak Hz 4.89 178.14 22.96 + 5.13 4.89 178.14 45.58 + 6.49 4.89 17.62 10.18 + 1.30Peak Amp 107.36 140.24 120.06 + 0.89 113.36 144.52 127.82 + 1.30 87.16 123.39 109.04 + 4.35Tot. Power 116.34 142.86 126.09 + 0.84 122.33 146.05 133.68 + 1.06 103.34 132.62 119.57 + 3.45

49

Table 2-5. Spectrum-level SPLs at peak or dominant frequencies of representative fish sounds. Some values reported as mean + SE.

Species Call Peak Hz SPL (dB re: 1 µPa)

Dist (cm)

Reference

Marine Species

Hippocampus erectus Stridulation 232 + 16 94 + 1 7.5 P. Anderson and D. Mann, forthcoming

Micropogonius undulatus Pulse train 460 114 100 Barimo and Fine, 1998 Cynoscion nebulosus Pulse Train 350 120 5 Locascio and Mann,

2008 Opsanus tau Grunt 134-170 123 + 2.3 100 Barimo and Fine, 1998 Opsanus tau Boatwhistle 230-270 126 + 1.2 100 “

Freshwater Species Corydoras paleatus Stridulation 2000 85 5 Ladich, 1999

Padogobius martensii Pulse Train Tone

67-104 120-200

90 98

5-10 5-10

Lugli et al., 2003 “

Serrasalmus nattereri Drum 200 95 5 Ladich, 1999 Pimelodus pictus Stridulation

Drum 1500-3000

100 102 105

5 5

“ “

Agamyxis pectinifrons Stridulation Drum

800-1500 100

105 90

5 5

“ “

Pimelodus blochii Stridulation Drum

2000 200-400

105 97

5 5

“ “

Platydoras costatus Stridulation Drum

400-800 100

110 105

5 5

“ “

Gobius nigricans Tone 73-103 113 5-10 Lugli et al., 2003 Botia modesta Knock 100-300 120 5 Ladich, 1999

Trichopsis vittata Pulse train 1500 124 3 Wysocki and Ladich, 2001

50

Figure 2-1. The Seahorse Sound Survey kit.

51

Figure 2-2. Acoustic sampling sites in Tampa Bay, FL.

52

Figure 2-3. Power spectra from representative tanks at the minimum, maximum (both in gray) and median (in black) of the range.

a = Recordings from the middle of the water column, b = recordings from tank bottom, c = recordings from Tampa Bay.

a b c

53

105110115120125130135140145

Concre

teAcry

lic

Fibergl

ass

Glass

Wood

Concre

teMeta

l

Plastic

Plenum

Gravel

Bare

Wall Material Stand Material Bottom Habitat

SPL

(dB

re: 1

uPa

)

AA A,B

B

C C

DE E

F

Figure 2-4. Total power vs. selected design specifications. Mid-level SPLs in white, bottom-

level SPLs in gray. Error bars are + SE. Letters denote significant differences among treatments within a subgroup (Tukey’s test, p < 0.05).

115

120

125

130

Glass Fiberglass Concrete Acrylic

Wall Material Type

SPL

(dB

re: 1

uPa

)

A

B,C

A,B

C

Figure 2-5. Mid-level total power vs. wall material type among tanks with gravel bottoms. Error

bars are + SE. Letters denote significant differences among wall material types (Tukey’s test, p < 0.05).

54

CHAPTER 3 AUDITORY EVOKED POTENTIAL OF THE LINED SEAHORSE, HIPPOCAMPUS ERECTUS, AND ITS RELATIONSHIP TO CONSPECIFIC SOUND PRODUCTION

Introduction

The auditory evoked potential (AEP) technique to measure hearing ability is widely

practiced among human clinicians (e.g., Davis, 1976; Picton et al., 1981; Schroeder and Kramer,

1989) and has been expanded to test hearing ability of representatives from many vertebrate taxa

(Corwin, 1982), including fish (Kenyon et al., 1998). Among fishes, this method has advantages

over behavioral methods such as cardiac suppression (e.g., Chapman and Sand, 1974),

ventilatory suppression (e.g., Fay, 1995), stereotyped defense responses (e.g., Kenyon, 1996),

classical conditioning (Fay, 1988a), instrumental avoidance conditioning (e.g., Tavolga and

Wodinsky, 1963), and operant conditioning (e.g., Yan and Popper, 1993). Behavioral methods

pose various drawbacks, including inconsistency of response, excessively long training periods,

and stressful stimuli (Kenyon et al., 1998). In contrast, AEP is a non-invasive far-field recording

of synchronous neural activity in the eighth nerve and brainstem auditory nuclei elicited by

acoustic stimuli (Jacobson, 1985).

The objectives in this study are two-fold. First, I aim to characterize the hearing ability of

the lined seahorse (Hippocampus erectus) in a comprehensive manner by measuring both the

pressure and particle acceleration of the acoustic stimuli in hearing tests. These components of

sound contribute in different ways in the near-field and far-field of sound sources.

A vibrating sound source produces two physical changes in the surrounding environment:

particle motion and pressure wave propagation of the surrounding medium (Dusenberry, 1992).

The near-field of a sound source is dominated by local hydrodynamic flow, established by the

displacement of water molecules (particle motion) adjacent to the sound source (Bass and Clark,

55

2003). In the far-field, propagating waves begin and can generate a pressure wave in addition to

particle motion.

The inner ears of fishes include three otolithic end organs (the saccule, utricle, and lagena)

containing a calcium-carbonate otolith, encased within a sac lined with a sensory epithelium.

The sensory unit of the epithelium is the neuromast, or hair cell. It is a mechanosensor that

detects particle motion. When the particle motion component of a propagating sound wave

approaches a fish, the same degree of particle motion is generated in the watery tissues of the

fish as in the surrounding medium. The otololith, however, moves at a different amplitude and

phase due to its greater density. This sets up a relative displacement between the otolith and the

neuromasts that is proportional to acoustic particle motion (Popper and Fay, 1999).

The gas bladder, thought to have originally evolved as a mechanism for buoyancy control,

incidentally affects the hearing sensitivity of fishes. When a volume of gas is exposed to

oscillating pressure changes, it will display larger volume pulsations than a comparable volume

of water. So, it acts as an amplifier, converting pressure fluctuations to motions detectable by

the ears. This enables some fishes to detect the pressure component of sound as well, especially

hearing specialist fishes, or fishes which have evolved a bony or gaseous vesicular connection

between the gas bladder or inner ear (Popper et al., 2003).

It is generally well-accepted that hearing specialist fishes respond to both particle

acceleration and pressure, but are more sensitive to pressure particularly in the far-field and at

frequencies above 70 Hz (Fay et al., 1982). Hearing generalist fishes, that have no specialized

connections between the swimbladder and inner ear, have yielded equivocal data concerning the

relative importance of pressure sensitivity to sound detection and processing (Cahn et al., 1968;

Sand and Enger, 1973; Chapman and Johnstone, 1974; Fay and Popper, 1975; Jerkø et al., 1989;

56

Lovell et al., 2005). The consensus that might be drawn from this literature is that both acoustic

modalities may be detected and processed by hearing generalist fishes, though the relative

contributions of each may vary with respect to distance, frequency, and sound pressure level;

both modalities are thus reported here.

I also aim to characterize conspecific sound production and to draw inference about

audition of these sounds. Seahorses produce clicks that were first described by Fish (1953) and

later expanded upon by Colson et al. (1998). Clicks are short duration, broadband sounds

(Zelick et al., 1999) that, among seahorses, are characterized by a peak frequency that is

inversely related to body size (Colson et al., 1998). The seahorse click is a stridulatory sound

produced from the bony articulation between the supraoccipital and the coronet (Fish, 1953;

Colson et al., 1998) and is demonstrated in many behavioral contexts, including feeding (Colson

et al., 1998), aggression and competition for mates (Vincent, 1994b), distress (Fish, 1953), and

stress (Chapter 4). Audition of conspecific clicks must be demonstrated as a precursor to testing

hypotheses of conspecific acoustic communication.


Animal Accession, Holding, and Husbandry Procedures

Lined seahorses (H. erectus) were collected as bycatch from shrimp trawl nets and donated

by local fishermen. Upon accession, animals were quarantined for one month prior to transfer to

a sound-dampened holding system. Clear round acetate tags (approx. 1 cm diameter) were

marked with alphanumeric codes, hung on monofilament line collars, and tied around the necks

of seahorses (tagging methods modified from Vincent and Sadler, 1995). Animals were fed

frozen mysids (Piscine Energetics) in the mornings and live Artemia sp. enriched with Roti-Rich

in the afternoons. Tanks were siphoned clean of debris twice daily and system water changes of

10% were performed weekly. Water quality parameters remained within the following ranges

57

during holding: Temperature, 25-27°C; salinity, 28.5 to 31.5 ppt; ammonia-nitrogen, 0 ppm;

nitrite-nitrogen, 0 ppm; nitrate-nitrogen, 2.8-22.7 ppm.

Eleven animals were transferred to a sound-proofed tank with an established biofilter 2 to

11 days prior to testing. Soundproofing was accomplished by resting the frame of the tank on a

sturdy lab bench with sections of bearing felt, installing a subsurface drain that transferred water

to a sump resting on the floor where filtration occurred, and using a quiet, 15W water pump with

a flexible return pipe that returned water to the tank below the water surface. A loop was

suspended in the flexible return line; this attenuated vibration and sound traveling through the

return water and pipe walls (A. Noxon, Acoustic Sciences Corp., personal communication,

Figure 3-1). The ambient noise profiles of both the holding system and the sound-proofed tank

were measured with an HTI-96-min hydrophone (High Tech Instruments, Inc., sensitivity = -

164.1 dB re: 1V/µPa, bandwidth = 2-30,000 Hz), for sound pressure level (SPL) measurements.

The ambient noise profile of the sound-proofed tank was also measured with an Acoustech

geophone probe (Acoustech Corporation, sensitivity = 212 V/m/s, bandwidth = 100-1000 Hz) for

measurements of particle motion. Both instruments, when in use, were connected to the line-in

port of a laptop computer running CoolEdit (Syntrillium Software). Hydrophone recordings

were collected from the middle and bottom of tanks. Geophone recordings were collected from

the bottom of tanks in the center, along three orthogonal axes, because particle motion is a vector

quantity (as opposed to pressure, which is a scalar quantity). Resulting sound files were

calibrated according to manufacturer instructions and post-processed with SpectraPlus (Pioneer

Hill software). Analysis settings used in SpectraPlus are summarized in Table 3-1. The

Acoustech geophone probe measures particle velocity. To convert to acceleration, Fast-Fourier

58

Transforms (FFT’s) processed by SpectraPlus were exported into a spreadsheet program and

particle velocity values converted to particle acceleration using the following formula:

Acceleration (m/s2) = v2πf (3-1)

where v = velocity (m/s), π = pi (~3.14), and f = frequency (Hz) (Casper and Mann, 2006). The

magnitude of particle acceleration was calculated by vector averaging according to the following

equation:

222 zyxonAccelerati ++= (3-2)

where x, y, and z refer to acceleration (m/s2) in each of three orthogonal axes.

AEP

Experimental setup

The testing chamber consisted of a steel tube (1.22 m high, 20.32 cm diameter, 0.9525 cm

thickness), closed at the bottom with a square steel plate (60.96 X 60.96 cm), and oriented

vertically. Four anti-vibration floor mounts (Tech Products Corp., 51700 Series) were placed

under each corner of the base of the tank. The tube was filled with saltwater of approximately

26°C up to a height of 1.12 m. A laboratory stand was supported on an adjacent vibration-

isolated table and scaffolding descended into the tube for animal suspension. A University

Sound UW30 speaker was placed at the bottom of the tube in the center. This setup was

enclosed inside an audiology booth.

For testing, individual fish were secured in a harness constructed from Nitex mesh,

fastened with clamps to scaffolding 2.5 cm below the water surface. The harness restricted

movement while allowing normal respiration. Subdermal stainless steel needle electrodes

(Rochester Electro-Medical) were used to record the AEP signal. An electrode was inserted

about 1 mm into the head, over the medulla region. Reference and ground electrodes were

59

placed directly in the water in close proximity to the fish. Evoked potentials recorded by the

electrode were fed through a Tucker Davis Technologies (TDT) HS4 fiber obtic headstage and

into a TDT RP 2.1 processor, routed into the computer and averaged by TDT BioSig software.

All eleven seahorses were tested for AEP’s to tone stimuli (described below); also, a dead

goldfish (Carassius auratus) was run to generate control AEP signals.

Sound stimuli and AEP waveform recordings were produced with a TDT AEP workstation

running SigGen and BioSig software. Sounds were generated by an RP 2.1 Enhanced Real-

Time processor, fed through a PA5 programmable attenuator to control sound level, and

amplified by a Hafler Trans.Ana P1000 110 W professional power amplifier before being sent to

the UW30 speaker, where sound was emitted (Figure 3-2).

Sound generation, calibration, and AEP acquisition

Calibration. Acoustic stimuli were calibrated with the HTI-96-min hydrophone for

pressure measurements of tones and the Acoustech geophone in three orthogonal axes for

particle motion, all connected to the RP 2.1 processor. During calibration, the hydrophone or

geophone was positioned in the experimental setup in place of the fish at the level of the

animal’s head, and the stimulus presentation protocol as described for AEP acquisition (below)

was executed, except without phase alternation. Signals were captured and averaged by BioSig.

Resulting time domain averaged signals were exported as ASCII formatted files, imported in

SpectraPlus, and 4096 point FFTs run (SpectraPlus settings in Table 3-1) to generate power

spectra, from which peak amplitude measurements were taken. Particle acceleration was

calculated per Equations 3-1 and 3-2. Calibration runs were conducted daily.

AEP acquisition. Stimuli consisted of 60 ms pulsed tones gated with a Hann window.

The phase of the tone was alternated between presentations to minimize electrical artifacts from

the recordings. During each trial, 9 different frequencies were presented: 100, 200, 300, 400,

60

600, 800, 1000, 1500, and 2000 Hz. Amplitudes at each frequency were presented within a

range of approximately 74 to 148 dB re: 1 μPa and 8.60 X 10-6 to 0.67 m/s2, beginning at

amplitudes below threshold and increasing in 6 dB steps until a threshold was visually detected

in the digital signal output (see “Data Analysis”). Post-hoc trials were run at amplitudes that

were 3 dB below visual threshold to increase the accuracy of threshold determination. Up to

2,000 signal presentations (or until detection was visually confirmed) were averaged to measure

the evoked response at each level of each frequency.

Data analysis

Evoked potential traces were transformed with the Hann window function and converted to

power spectra with a 2048-point FFT in BioSig. Evoked potentials are visualized as peaks that

occur at twice the frequency of the presented stimulus (Figure 3-3). This is a well-established

phenomenon in evoked potentials of fish to pure tones in the frequency domain (Egner and

Mann, 2005). Visualized peaks were considered true evoked potentials if they were at least 3 dB

above the average of all peaks occurring within a window of 50 Hz above the presented stimulus

frequency. AEP thresholds were defined as the lowest amplitude at which a true evoked

potential, according to these criteria, was visualized. AEP waveforms of live seahorses were

checked against AEP waveforms of dead goldfish to ensure that the identified peak was not a

stimulus or electrical artifact.

Click Recordings

Experimental setup

A rectangular polystyrene fish shipment box with inner dimensions of 37.7 cm L X 37.7

cm W X 19 cm H and wall thickness of 2.3 cm was marked with a 5 cm X 5 cm square grid on

the floor of the box, placed on the floor, and filled to 17 cm high with saltwater. The Acoustech

probe was placed against one of the walls of the box at its center, with the motion sensitive axis

61

oriented perpendicular to the wall (facing toward the center of the box, Figure 3-4). I utilized the

Acoustech probe’s incorporated hydrophone (sensitivity = -176 dB re: 1 V/µPa, bandwidth = 10

– 1,000 Hz) for pressure sensitive measurements and the geophone for particle motion sensitive

measurements. The hydrophone was fed to the left-channel and the geophone to the right-

channel of the line-in port of a laptop computer running CoolEdit (Syntrillium Software) for

recording at a sampling rate of 44.1 kbps. The recording system of the laptop was previously

calibrated with a 1.0 Vpeak sine wave.

One to three seahorses were placed inside the box at any one time. An aliquot of live

Artemia sp. was added to the box to stimulate a feeding response. Recording began when the

first seahorse made its first feeding strike. During recording, as each animal struck, the animal

ID, distance to the Acoustech geophone, and the timestamp of the strike was recorded.

Recording ceased after 30 min or earlier if animals stopped feeding for an excessive period of

time.

Data analysis

Signal processing and analysis. Using CoolEdit, documented clicks from resulting .wav

files were isolated within the center of 5 s windows, copied and pasted into new, individual

sound files for each click. The first two seconds of each click file were used to obtain a noise

profile using a 4096 pt FFT. This noise profile was used to reduce noise in the entire click file

by 40 dB. Noise-reduced files were then calibrated for voltage in SpectraPlus according to

manufacturer instructions and post-processed. Click analysis settings used in SpectraPlus are

summarized in Table 3-1.

For both the pressure-sensitive waveform (left-channel) and the particle-motion sensitive

waveform (right-channel), I took the following cursor measurements in the time domain:

62

• Duration: Duration of entire click waveform (in milliseconds)

• Rise: Time from onset of waveform to largest non-resonant peak (in milliseconds)

• Fall: Time from largest non-resonant peak to end of waveform (in milliseconds)

• Peak Amplitude: Amplitude at largest non-resonant peak (in dB SPL (re: 1 µPa) or m/s)2

• Peak-to-Peak Amplitude: Amplitude differential from largest non-resonant positive peak to largest non-resonant negative peak (in dB SPL (re: 1 µPa) or m/s, Figure 3-5)2

For both waveforms, I took the following cursor measurements in the frequency domain,

examining a window from 2-998 Hz:

• Frequency at 25% Peak Amplitude: On both sides of the broadband signal (in Hz)

• Peak Frequency: Frequency of 1st, 2nd, and 3rd highest amplitude peaks (in Hz)

• Peak Amplitude: Amplitude of 1st, 2nd, and 3rd highest amplitude peaks (in dB SPL (re: 1 µPa) or m/s2)2

• Total RMS Power: From 2 to 998 Hz (in dB SPL (re: 1 µPa) or m/s2, Figure 3-6)2

Statistical analysis. Clicks were categorized from visual inspection as either resonant or

non-resonant, characterized by the presence or absence (respectively) of a series of high

amplitude, high frequency waveforms occurring during or immediately after the rise of the click

in the time domain. For each measure, descriptive statistics were computed for total, resonant,

and non-resonant clicks from each animal and for summed clicks, for both pressure and

acceleration. F-tests were run between non-resonant and resonant clicks for all measures to

examine heterogeneity of variance. t-tests (assuming equal variances only where F-tests were

non-significant) were run between non-resonant and resonant clicks for all measures to examine

differences in click characteristics in the time and frequency domains. Paired t-tests were run

between pressure-sensitive waveforms and particle-motion sensitive waveforms to examine 2 The Acoustech geophone probe measures particle velocity. In the time domain, particle motion measurements are represented in terms of particle velocity. In the frequency domain, particle motion measurements are represented in terms of particle acceleration, and calculated using Equation 3-1.

63

differences in the time domain and in peak frequencies. Finally, a repeated measures ANOVA

model was employed using SAS (The SAS Institute) to examine possible differences in peak

frequencies, peak-to-peak amplitudes and click durations, between the sexes and among

individuals, for both pressure-sensitive waveforms and particle-motion sensitive waveforms,

according to the following model:

y = sex + animal(sex) + ε (3-3)

where sex is a fixed factor--male or female, animal(sex) is a random factor referring to the

individual animal (nested within sex), and ε is the error term. Data were not balanced, so Type

III Sums of Squares were used.

Results

Ambient Noise

The long-term holding tanks in which animals were housed prior to transfer to the AEP

laboratory demonstrated an average total RMS power (within the 2 to 998 Hz frequency range)

of 117.4 + 0.9 dB SPL (re: 1 µPa) at the middle of the water column and 128.8 + 1.4 dB at the

bottom (Figure 3-7a). The soundproofed AEP laboratory tank demonstrated an average total

RMS power of 115.8 + 0.5 dB at the middle of the water column and 120.5 + 0.2 dB at the

bottom (Figure 3-7b), and a vector-averaged total RMS power of 4.58 X 10-3 m/s2 (Figure 3-8).

AEP

The AEP audiograms averaged from 11 H. erectus are plotted in Figure 3-9. For sound

pressure, this species’ most sensitive hearing range is below 400 Hz, with a minimum threshold

of 105.0 + 1.5 dB SPL re: 1 µPa at 200 Hz (mean + SE). After 600 Hz, hearing thresholds

increase to levels above most environmentally relevant noise (Urick, 1975). For particle

acceleration, this species’ most sensitive hearing range is below 300 Hz, with a minimum

threshold of 3.46 X 10-3 + 7.64 X 10-4 m/s2 at 200 Hz.

64

Click Recordings

I characterized up to ten clicks from ten seahorses, for a total of 85 clicks.

Pressure

Characterization. Seahorses produced clicks at an average distance of 7.5 + 0.4 cm from

the center of the Acoustech geophone. Clicks had an average duration of 109.8 + 7.6 ms, with a

rise time of 48.5 + 5.2 ms and a fall time of 59.9 + 3.7 ms. 71.7% of clicks presented peaks in

the positive phase, and 28.2% in the negative phase. Clicks averaged peak amplitudes of 136.8 +

1.0 dB SPL (re: 1 µPa) and peak-to-peak amplitudes of 141.6 + 1.0 dB in the time domain. In

the frequency domain, the power distribution was broad, with amplitudes at 25% of peak

beginning at 89 + 5 Hz and ending at 741 + 21 Hz. Peak frequency averaged 232 + 16 Hz, with

an average peak amplitude of 94.3 + 0.9 dB SPL (re: 1 µPa). The most prominent peaks (I

selected three peaks with the highest amplitudes from each click) were found within an

interquartile range of 166 to 343 Hz, with a mean of 271 + 10 Hz. Amplitudes of these peaks

were within an interquartile range of 86.2 to 97.3 dB dB SPL (re: 1 µPa), with a mean amplitude

of 92.5 + 0.5 dB.

Resonant vs. non-resonant clicks. Pressure waveforms were particularly susceptible to

resonance artifacts, a common phenomenon in small tanks (Parvulescu, 1964, 1967; Yager,

1992; Akamatsu et al., 2002). Clicks were judged as resonant if a series of high amplitude, high

frequency waveforms occurred during or immediately after the rise of the click in the time

domain (Figure 3-10). Clicks whose resonance masked the peak (and the delineation between

rise and fall time) were not included in analysis of measures in the time domain, but in most

cases, resonance occurred after the peak and during the fall time of the click. There were no

significant differences in the duration, rise time, or fall time between resonant and non-resonant

peaks. However, resonant peaks demonstrated louder peak amplitudes (140.1 + 1.6 dB vs. 134.7

65

+ 1.2 dB, p = 0.009) and peak-to-peak amplitudes (144.8 + 1.4 dB vs. 139.6 + 1.2 dB, p = 0.008)

in the time domain waveform. This trend is corroborated in the frequency domain; resonant

clicks had higher peak amplitudes (96.7 + 1.3 dB) than non-resonant clicks (92.4 + 1.2 dB, p =

0.015).

Particle acceleration

Clicks measured from the particle acceleration recordings had an average duration of 154.3

+ 10.7 ms, with a rise time of 55.2 + 5.4 ms and a fall time of 100.8 + 6.9 ms. 58.8% of clicks

presented peaks in the positive phase, and 41.2% in the negative phase. Clicks averaged peak

velocities of 1.06 X 10-4 + 1.65 X 10-5 m/s and peak-to-peak velocities of 1.91 X 10-4 + 2.96 X

10-5 m/s in the time domain. In the frequency domain, the power distribution was also broad,

with amplitudes at 25% of peak beginning at 81 + 3 Hz and ending at 681 + 17 Hz. Peak

frequency averaged 265 + 22 Hz, with a peak acceleration of 1.52 X 10-3 + 1.87 X 10-4 m/s2.

The most prominent peaks (I selected three peaks with the highest amplitude from each click)

were found within an interquartile range of 148 to 372 Hz, with a mean of 290 + 13 Hz.

Amplitudes of these peaks were within an interquartile range of 4.90 X 10-4 to 1.33 X 10-3 m/s 2,

with a mean acceleration of 1.20 X 10-3 + 8.99 X 10-5 m/s 2.

Comparisons between pressure waveforms and particle acceleration waveforms

In the time domain, particle motion waveforms exhibited longer overall duration (p = 1.65

X 10-8), as well as longer rise (p = 2.50 X 10-4) and fall (p = 2.37 X 10-8) times. In the frequency

domain, pressure waveforms tended to have more energy distributed at higher frequencies;

frequencies at 25% of peak amplitude to the right of the largest peak were higher in pressure

motion waveforms (p = 1.98 X 10-4). However, there were no significant differences in the

frequencies of either the highest amplitude peak (p = 0.180) or the frequencies of the three

highest amplitude peaks combined (p = 0.208) between the two waveform types.

66

Comparisons among individuals and between sexes

The sex of the animal had no effect on click duration (pressure, F1,8.18 = 0.39, p = 0.547;

particle motion, F1,8.30 = 1.95, p = 0.199), peak-to-peak amplitude (pressure, F1, 8.97 = 1.35, p =

0.276; particle motion, F1, 75 = 0.11, p = 0.745), or peak frequency (pressure, F1, 10.35 = 0.17, p =

0.686; particle motion, F1, 10.344 = 3.70, p = 0.082) of waveforms, although sample sizes were

small (n♀ = 7, n♂ = 3). Clicks differed significantly among individuals in duration (pressure,

F8,75 = 15.76, p < 0.0001; particle motion, F8,75 = 9.52, p < 0.0001) and peak-to-peak amplitude

of both waveforms (pressure, F8,69 = 6.36, p < 0.0001; particle motion, F8,75 = 3.58, p = 0.0015),

but not in peak frequency (pressure, F8,75 = 1.29, p = 0.261; particle motion, F8,75 = 1.29, p =

0.260).

Discussion

My results demonstrate that H. erectus is a hearing generalist. Its hearing sensitivity falls

within the range of sensitivities documented for other hearing generalist fishes (e.g., Kenyon et

al., 1998; Yan, 2001; Scholik and Yan, 2002a; Lugli et al., 2003; Egner and Mann, 2005; Lovell

et al., 2005; Casper and Mann, 2006, Figure 3-11). While I had the ability to test this animal’s

hearing sensitivity at up to 2,000 Hz (for sound pressure), thresholds at frequencies above 600

Hz begin to rise into a range of high SPL’s that animals are not likely to encounter in the natural

environment (Urick, 1975). My conclusion that this seahorse is a hearing generalist is

corroborated by the lack of specialized connections between the gas bladder and the inner ear (P.

Anderson, pers. obs.), that is generally a requirement for hearing specialist fishes (Popper et al.,

2003).

There is remarkable similarity between the shapes of the audiograms for sound pressure

and particle acceleration. Despite the fundamental differences between the pressure and particle

motion component of sound, and the fundamental differences in the way each modality is

67

processed by fishes, both hearing specialists (e.g., Ictalurus punctatus, Fay and Popper, 1975)

and generalists (e.g., Ginglystoma cirratum, Casper and Mann, 2006) demonstrate similarities

between the shape of audiograms for each acoustic modality. Some tests have shown

dissimilarities (e.g., Hawkins and Johnstone, 1978; Kelly and Nelson, 1975) that may be due to

artifactual acoustic discontinuities between sound pressure and particle motion in constrained

testing environments.

My results of seahorse click characteristics differ considerably from previous

characterizations, though sample sizes in previous studies are small. Fish (1953) documented

maximum energy of clicks from one female H. erectus to occur in 300-600 Hz and 400-800 Hz

frequency bands. Colson et al. (1998) documented peak frequencies ranging from 1.96 to 2.37

kHz in four clicks produced by one individual. The response bandwidth of Colson et al’s

hydrophone was not stated; whether the frequency range of 0-1000 Hz was investigated is

unknown. Even if peak frequencies above 1,000 Hz are real, they are unlikely to be of

intraspecific communicative value, given the audiogram presented here. The clicks I recorded

may have contained energy above 1,000 Hz, particularly among resonant clicks (that, in this

case, may be artifactual and not a true component of the click), but given the sensitivity of H.

erectus and other hearing generalist fishes, I considered the bandwidth of 0-1,000 Hz to be most

worthy of investigation.

The resonance artifacts I documented in sound pressure among clicks are typical of

problems of sound distortion in small tanks, that have been known for some time (Parvulescu,

1964, 1967). Yager (1992) also documented resonance artifacts when recording clicks from the

African Pipid frog (Xenopus borealis) in small tanks; showing a broad band of high-amplitude

energy between 5-40 kHz that is not present when clicks are recorded in a pond. However, my

68

tests showed no significant differences among the frequency components of the clicks in the 0-

1000 Hz frequency band, that is within the range of hearing of H. erectus and other hearing

generalists. It is likely that resonance artifacts seen in the time domain would have appeared as

peaks at higher frequencies that I did not consider, as predicted by Akamatsu et al.’s (2002)

equations for calculating minimum resonant frequencies, but because these peaks are outside of

the hearing range of most fishes, these distortions are not likely to be audible to them.

Resonance did not alter temporal components of clicks, but resonant clicks were louder than

non-resonant clicks. I hypothesize that resonance did not result in louder clicks, but rather that

louder clicks were more likely to propagate standing waves in the recording chamber and

generate resonance. Hearing generalist fishes may thus still be able to detect and correctly

process conspecific sounds in small tanks, because resonance components that distort the

original signal occur outside of the hearing range of these fishes. However, acoustic

communication modalities may be hampered among hearing specialists in small tanks, where

resonance components are likely to occur within their range of hearing.

The seahorse click is an example of a broadband sound that increases likelihood of

reception by listeners. Broadband signals are detected at lower amplitudes than tonal signals

(Yost, 2000). In an environment such as a marine coastal zone where substrate, habitat, surface

waves, and ambient noise serve as obstacles to sound propagation (via reflection, refraction,

attenuation, masking, absorption, scattering, etc.), broadband signals promote the reliability of

sound propagation where the spectral content of the signal is often distorted (Gerald, 1971).

Sudden onset of the click, a temporal characteristic, also boosts likelihood of detection (Hall,

1992).

69

A visually striking argument for the evolution of sound production for the purpose of

conspecific communication in H. erectus is made when comparing peak frequencies of the sound

with the audiogram. Because the audiogram measures hearing at one frequency while clicks are

broadband, I adjusted the audiogram thresholds to estimate the animal’s ability to detect

broadband sounds using a calculation proposed by Yost (2000) and employed by Egner and

Mann (2005) for hearing in damselfishes. Specifically, the audiogram was adjusted by an

estimated critical bandwidth that is assumed to be 10% of the center frequency. Because sound

pressure is expressed on a logarithmic scale (dB), this adjustment is:

( )FrequencyThreshold %10log10∗− (3-4)

Because acceleration is a linear quantity, this adjustment is:

FrequencyThreshold%10

(3-5)

Additionally, because clicks were only characterized for particle acceleration with the

geophone in one axis, I employed audiogram threshold measurements from only one axis with

the greatest acceleration values. These comparisons are represented in Figures 3-12 and 3-13.

For both pressure and particle acceleration, peak frequencies cluster around the lowest threshold

at 200 Hz. While only 34% of the peaks are apparently audible in sound pressure, 93% are

audible in particle acceleration. Furthermore, AEP methodology underestimates hearing

thresholds in comparison to behavioral methods by up to 19 dB (Hill, 2005), so it is likely that

most clicks are audible. These observations, coupled with the conclusion that H. erectus is a

hearing generalist, lead to the conclusion that H. erectus may hear and process conspecific

clicks. Correlation between auditory sensitivity and vocalization has been documented for other

sound-producing fishes as well (e.g., Stabentheiner, 1988; Ladich and Yan, 1998; Ladich, 1999;

Wysocki and Ladich, 2001; Lugli et al., 2003).

70

Vocalization in fishes may communicate a wealth of information, including identity

(Myrberg and Riggio, 1985), sex (Ladich, 1997), size (Myrberg et al., 1993), species (Spanier,

1979), location (Ladich, 1997), and behavioral state (Crawford, 1997). Among many of the

generalists that create broadband sounds, modulation of the temporal component of sound (as

opposed to the frequency component) is considered to be the most important communicative

feature (Popper et al., 2003), but this is generally in reference to sounds that can vary in duration,

interpulse interval of pulse trains, or number of vocalizations (e.g., Spanier, 1979; McKibben and

Bass, 1998). My results show no detectable difference in the time or frequency domains

between males and females, though sample sizes were small. Individual differences were

detected in click duration and amplitude, that provides for the opportunity to discriminate among

individuals in these measures. A seahorse may evaluate clicks produced by nearby seahorses to

differentiate between mates and non-mates in populations where extra-pair encounters occur

(Vincent and Sadler, 1995). Likewise, an animal may click to advertise its presence or location

to its mate. Clicking may also assist unpaired animals in the advertisement and location of

potential mates in a sparsely distributed population. The click is also associated with feeding

strikes (Colson et al., 1998); advertising the presence of a food source to a monogamous mate

may be a strategy for increasing a mate’s reproductive fitness. Laboratory studies have

demonstrated clicking as part of an aggressive interaction between males competing for a mate

(Vincent, 1994b). Ladich (1997) suggests, in Trichopsis vittata, that fish emitting louder

agonistic signals may win competitive encounters; this scenario may also be plausible in H.

erectus. Though correlations of signal parameters with body size were not examined in this

study, Colson et al. (1998) documented a negative correlation between animal size and peak

frequency. As seahorses exhibit size-assortative mating (Foster and Vincent, 2004), the click

71

may also assist potential mates in the selection of a mate of suitable size. While my study

suggests that the seahorse click may have evolved to signal to conspecifics, further studies on the

function of clicking in these various contexts are warranted to test the hypotheses proposed.

72

Table 3-1. Spectra Plus recording settings. Ambient Noise Analysis AEP Stimulus Analysis Click Analysis Sampling Rate (Hz) 44100 48828 44100 Sampling Format 16-bit, Stereo 16-bit, Stereo 16-bit, Stereo Standard Hz Weighting Flat (none) Flat (none) Flat (none) Decimation Ratio 11 12 11 Frequency Limit 2004.545 Hz,

Low-pass filter enabled 2034.500 Hz, Low-pass filter enabled

2004.545 Hz, Low-pass filter enabled

FFT Size 4096 4096 4096 Spectral Line Resolution 0.979 Hz 0.993 Hz 0.979 Hz Smoothing Window Hann Hann Hann Averaging Settings Infinite, Linear,

Disable Peak Hold 1, Linear, Enable Peak Hold 1, Linear, Enable Peak Hold

FFT Overlap 0% 99% 99% Time Resolution 1021.68 ms 10.07 ms 10.22 ms Input Signal Overload Enable Overload Detection

Exclude Overloaded Data from Processor

Enable Overload Detection Exclude Overloaded Data from Processor

Enable Overload Detection Exclude Overloaded Data from Processor

73

Figure 3-1. Soundproofed laboratory tank.

74

Figure 3-2. AEP experimental setup (modified from Egner and Mann, 2005). P1000 = amplifier, PA5 = attenuator, RP2.1 = processor, RFE = reference electrode, RE = recording electrode, GE = ground electrode, HS4 = headstage.

75

Figure 3-3. Auditory evoked potentials (AEPs) to a 400 Hz tone pip depicted in the time domain (left) and in the frequency domain (right). a = Control AEP waveform (dead goldfish). b,c,d = AEP waveforms of H. erectus at progressively lower amplitudes. d represents amplitude at threshold. Asterisks denote AEPs that occur at twice the frequency of the presented stimulus (in this case, 800 Hz).

76

Figure 3-4. Click recording chamber (top view). The Acoustech geophone probe was placed against the wall of the chamber, with the recording axis facing toward the center.

Figure 3-5. Click characteristics measured in the time domain waveform. Horizontal arrows depict time measurements, vertical arrows depict amplitude measurements. P-P = Peak to peak.

77

Figure 3-6. Click characteristics measured in the frequency domain waveform. 1,2, and 3 refer to the 1st, 2nd, and 3rd highest amplitude peaks.

78

Figure 3-7. Representative power spectra of sound pressure of (a) long-term holding tanks and (b) the sound-proofed laboratory tank. Gray = Ambient noise recorded from the middle of the water column. Black = Ambient noise recorded from the tank bottom.

b a

79

0.00001

0.00010

0.00100

2 100 198 296 393 491 589 687 785 883 981

Frequency (Hz)

Acc

eler

atio

n (m

/s^2

)

Figure 3-8. Power spectrum of particle acceleration of the sound-proofed laboratory tank.

100

105

110

115

120

125

130

135

140

100 200 300 400 600 800 1000 1500 2000

Frequency (Hz)

SPL

(dB

re: 1

uP

a)

0.001

0.01

0.1

1

Acc

eler

atio

n (m

/s^2

)

Figure 3-9. Audiograms of the lined seahorse, H. erectus, for sound pressure and particle acceleration. Black trace = sound pressure, Gray trace = particle acceleration.

80

Figure 3-10. A resonant click, depicted in the time domain. Compare with Figure 3-5.

60

70

80

90

100

110

120

130

140

150

100 200 300 400 500 600 700 800 900 1000

Frequency (Hz)

Soun

d Pr

essu

re L

evel

(dB

re: 1

uPa

)

Hippocampus erectus (this study)

Lepomis macrochirus (Scholik and Yan, 2002a)

Opsanus tau (Yan, 2001)

Padogobius martensii (Lugli et al., 2003)

Astronotus ocellatus (Kenyon et al., 1998)

Carassius auratus (Kenyon et al., 1998)

Figure 3-11. Sound pressure audiograms of representative hearing generalist fishes, measured by the AEP technique. The audiogram of a representative hearing specialist, C. auratus, is also shown for comparison.

81

70

75

80

85

90

95

100

105

110

115

120

0 500 1000 1500 2000 2500

Frequency (Hz)

SPL

(dB

re: 1

uPa

)

Figure 3-12. Comparison of the broadband sound pressure audiogram against peaks of recorded clicks, represented as gray points.

0.0001

0.001

0 200 400 600 800 1000 1200

Frequency (Hz)

Acc

eler

atio

n (m

/s^2

)

Figure 3-13. Comparison of the broadband particle acceleration audiogram against peaks of recorded clicks, represented as gray points.

82

CHAPTER 4 SOUND, STRESS, AND SEAHORSES: THE CONSEQUENCES OF A NOISY

ENVIRONMENT TO ANIMAL HEALTH

Introduction

Here, I examine the hypothesis that chronic exposure to loud ambient noise acts as a

chronic stressor to aquarium fish, resulting in responses among primary (e.g., plasma cortisol),

secondary (e.g., blood glucose, hematological measures), and/or tertiary (e.g., growth, behavior,

mortality) stress indices.

Stress is an important consideration in the successful husbandry of fish. Severe stress

results in mortality, but even sublethal stress compromises various physiological and behavioral

functions, leading to suppressed immune function and disease resistance, growth rate, and

reproduction, all contributing to suboptimal production (Iwama et al., 1997).

Anthropogenic noise to which wild and captive fishes are exposed is variable, but can have

negative impacts at all levels of biological organization. Santulli et al. (1999) examined the

effects of intense acoustic stimulation emitted by air guns used in seismic surveys on

biochemical and physiological responses in European sea bass. Their post-shock serum analyses

indicated increases in cortisol, variations in glucose and lactate, decreased skeletal muscle and

liver ATP concentrations, and increased ADP concentrations. Even exposure to vessel noise at

lower volumes (153 dBLeq re: 1 μPa) elicited increased plasma cortisol concentrations in both

hearing generalists and hearing specialists (Wysocki et al., 2006), but white noise exposure (at

160-170 dBrms re: 1 μPa) did not lead to sustained elevated cortisol concentrations in goldfish

(Smith et al., 2004).

Behaviorally, Blaxter and Batty (1987) demonstrated startle responses in herring when

exposed to transient sound stimuli. Popper and Carlson (1998) review several studies

demonstrating avoidance behavior of several groups of fishes to varying sound stimuli as a

83

method to discourage fishes from entering intakes at power plants or dams. Pearson et al. (1992)

demonstrated alarm and startle responses of rockfish to acoustic stimuli emitted by geophysical

survey devices (at 180peak dB re: 1 μPa), and Skalski et al. (1992) subsequently demonstrated a

decreased catch-per-unit effort.

White noise exposure (at 158rms dB re: 1 μPa) induced temporary hearing threshold shifts

of hearing specialist fishes (Amoser and Ladich, 2003), and very intense exposures (at levels

ranging from 180 to 204 dB re: 1 μPa) have led to neuromast damage and loss in the inner ear of

fishes (e.g., Enger, 1981; Cox et al., 1986a, b, 1987; Hastings et al., 1996; McCauley et al.,

2003).

The diversity of literature exploring effects of anthropogenic noise on fishes renders a

prediction of effects in an aquaculture setting difficult to estimate. Anthropogenic noise to

which wild fishes are exposed varies greatly in sound pressure level (SPL), frequency

composition, and duration of exposure. Literature pertaining specifically to effects of

aquaculture noise is, on the other hand, quite sparse, and has been reviewed in Chapter 1.

I examined effects of chronic exposure to loud ambient aquarium noise on primary,

secondary, and tertiary stress indices of a popular marine ornamental aquarium fish, the lined

seahorse (Hippocampus erectus). Masonjones and Babson (unpublished) demonstrated

increased incidence of gas bladder disease, behavioral differences, longer gestation lengths, and

fewer, smaller, and slower growing offspring in dwarf seahorses (H. zosterae) exposed to low

frequency boat motor noise, suggesting that seahorses are also prone to effects of ambient noise.


Animal Accession and Husbandry Procedures

Animals were collected as bycatch from shrimp trawl nets and donated by local fishermen.

Upon accession, animals were quarantined for one month prior to transfer to the research system,

84

receiving two treatments of chloroquine diphosphate at 10 mg/L spaced 21 days apart (for

Amyloodinium spp.) and 10 minute freshwater dips (for external parasites).

Animals were tagged as described in Chapter 3. Seahorses were held in same-sex groups

for a minimum period of 20-21 days (the gestation period of H. erectus, Lourie et al., 1999) prior

to experimentation. This waiting period standardizes reproductive status by allowing males to

give birth and females to reach the end of their postmating refractory period (Masonjones and

Lewis, 1996), eliminating the potentially confounding factor of variation in reproductive state on

measures of stress.

Animals were fed frozen mysids (Piscine Energetics™) in the mornings and live Artemia

sp. enriched with Roti-Rich™ in the afternoons. Tanks were siphoned clean of debris twice daily

and system water changes of 10% were performed on a weekly basis. Animals received 11

hours of fluorescent light daily, with low-intensity incandescent dawn and dusk lights

illuminated ½ hour before and ½ hour after fluorescent lighting transition. Animals were

maintained in Atlantic Ocean seawater that was diluted, chlorinated, neutralized, and pH

adjusted prior to use, at 25-27ºC, 27-30 ppt salinity, and at a pH of 8.0 to 8.4. Ammonia-

nitrogen and nitrite-nitrogen levels remained at 0 ppm throughout holding and experimental

periods. Nitrate-nitrogen levels ranged from 2.8 to 5.6 ppm.

Laboratory and Experimental Tank Design

Up to fourteen 76 L glass holding aquaria held up to five animals each in same-sex groups,

when not in experimentation. All exterior surfaces (except the top and front) were painted with

an opaque light blue paint to standardize the visual environment. Each tank contained five

plastic Vallisneria plants. Holding tanks were connected to a central filtration system consisting

of coarse mechanical prefilter pads, wet/dry biofiltration, carbon, and ultraviolet sterilization.

85

I employed several soundproofing design modifications to minimize ambient noise in the

system and associated holding tanks. The water level in each tank was adjusted to remain

approximately 2 cm above the strainer/bulkhead fitting drain assembly in the top corner of the

tank, as well as above the flexible water feed tube, to minimize ambient noise that might be

generated by water surface disturbance. Air stones were placed only in the sump, not in

individual tanks. The pump was soundproofed by placement within a plastic box lined with

acoustic insulation material, and an ellipsoid loop of flexible PVC pipe plumbed inline to

decouple vibration traveling through the rigid PVC pipe walls from the pump side to the tank

side. The loop was supported by plastic strapping that fastened to the ceiling. Acoustic

dampening felt strips were fitted along the corners and at the center of each tank bottom frame.

Experimental tanks consisted of a row of sixteen 76 L glass aquaria, set-up alternately as

quiet or loud tanks. Experimental tanks were painted, plumbed to the main system, and outfitted

as above, except that bottom exterior surfaces were painted with a sand-color paint and felt strips

were not applied.

Each loud tank sat on a commercially available wrought iron stand. A commercially

available magnetic drive inline aquarium pump (25 W, 1931 max LPH) was chosen as a noise

stimulus circulation pump. The base of the circulation pump was fastened to the center of a 1.9

cm thick plywood shelf that rested on the bottom frame of the stand. Each pump was outfitted

with a PVC ball valve at the output, and connected with hose barbs at both ends to 1.6 cm ID

(internal diameter) flexible tubing that led to the tank. Rigid PVC drain and return pipes were

fastened via hose barbs to the flexible tubing at the tank end. Return pipes were rotated at a 45o

angle, depositing above the water level (Figure 4-1).

86

Each quiet tank sat on a rectangular polystyrene pad (1.3 cm thick) resting on a wooden

frame stand modeled after a commercially available stand. The stand sat on a second

polystyrene rectangular frame pad (1.9 cm thick). A smaller, commercially available magnetic

driven inline aquarium pump (15W, 1022 max LPH) was chosen as a circulation pump. This

pump sat on the cement floor inside the polystyrene base frame. The pump input and output was

connected via hose barbs and flexible tubing to PVC drain and return pipes as in the loud tanks,

except return pipes were submerged 15 cm inside the tank, depositing below the water level.

Loops of flexible tubing were incorporated into the circulation pump input and output lines, as

well as the system feed line. Loops were supported by PVC collars connected to small bungee

cords that fastened to the ceiling (for the system feed loops) or the wall (for the pump loops,

Figure 4-2). These loops attenuated sound vibrations traveling inline (A. Noxon, Acoustic

Sciences Corporation, pers. comm.).

Tank flow rates were measured at the beginning of each trial and two weeks into each trial.

Flow rates of noisy tanks were then adjusted to match the average flow rate of the quiet tanks at

the time of each measurement.

For behavioral observations, an opaque white curtain was hung 1 m away from tank fronts

to conceal observers from animals. Square holes (6.5 X 6.5 cm) were cut in the curtain in front

of each tank to allow for behavior recording with a videocamera.

Sound Recording and Analysis

The ambient noise profile of each test tank was measured at the beginning and end of each

of two trials. Recordings were taken with an HTI-96-min hydrophone (High Tech Instruments,

Inc., sensitivity = -164.1 dB re: 1V/μPa) connected to the line-in port of a laptop computer

running SpectraPlus (Pioneer Hill Software), calibrated according to manufacturer’s instructions.

87

One minute recordings were collected from the middle and bottom of each tank. The

recording/analysis settings used in SpectraPlus are summarized in Table 2-1.

Hearing generalist fishes, without connections between the gas bladder and inner ear,

generally have insensitive hearing above 1,000 Hz (Fay, 1988b). Hearing tests of H. erectus

using the Auditory Evoked Potential (AEP) technique (e.g., Egner and Mann, 2005) are

consistent with this observation (Chapter 3). In their survey of aquaculture systems, Bart et al.

(2001) also analyzed recordings in the low frequency region (25-1000 Hz), which provides a

frame of reference for the sound levels to which the fish in this study were exposed. Thus, a

decimation ratio of 22 was chosen in analysis settings, which low-pass filtered the signal at a

cutoff of 1002 Hz. It should be noted that this reduces the total rms power measure, but the

sound pressure level that is output is averaged only over the fish’s range of environmentally

relevant hearing ability. Resulting sound files were post-processed by removing putative

artificial electrical peaks in frequency spectra (at 60 Hz or its harmonics) using the FFT filter

function in CoolEdit (Syntrillium Software Corporation).

Animal Assignment and Preparation

Tagged seahorses were randomly chosen from the holding population for each of two trials

and then randomly assigned to tanks, with the restriction that each sex was evenly distributed

between treatments for each trial. On Day 0, identified seahorses were anesthetized with 100

ppm of 2:1 buffered MS-222, weighed, measured for standard length (head length + trunk length

+ tail length, per Lourie, 2003), and placed in their assigned tanks.

Ethological Methodology

Data collection

For each of the two trials, 1 h observations were conducted for three animals daily for 5

days, followed by one additional animal observation on the 6th day, each week for 4 weeks. This

88

schedule enabled all 16 test animals to be observed once per week. The weekly order of

observations was kept constant for each of the 4 weeks so that each animal was observed at 7-

day intervals. Orders of the 3 animals observed each day were randomized daily. Observations

occurred in the afternoons, at least 1 h after morning husbandry (siphoning and feeding) but

before afternoon husbandry. Observations were videotaped with a Sony Handycam Video Hi8.

An ethogram was developed and programmed into the JWatcher program (Blumstein et al.,

2000) to subsequently score behaviors quantitatively from videotapes.

Measures and statistical analyses

The following behavioral measures were chosen for analysis:

• The proportion of time each animal spent at the top of the tank while in sight (of the observer)

• The proportion of time each animal spent stationary while in sight

• The proportion of time the animal spent holding a holdfast over the total time the animal was stationary while in sight

• The number of adjustments made at rest while the animal was in sight (standardized to number of adjustments per hour while stationary for each observation session)

• Total number of clicks

• Total number of pipes

• Total number of gapes

A repeated measures analysis of variance (ANOVA) was performed on each measure using

the SAS GLM procedure (the SAS Institute, Cary, NC) according to the following model:

y = noise + week + seahorse(noise) + noise*week + error (4-1)

Of the main effects, noise contained two levels (loud, quiet), week contained four levels

(weeks 1-4), and seahorse (nested within noise) was a random factor that contained 32 levels (for

each animal tested). Data were not balanced, so Type III sums of squares were used. Measures

89

containing significant interaction effects were subsequently sorted by time and ANOVAs were

run on the effect of noise at each week for each measure.

I conducted a post-hoc analysis to examine heterogeneity of variance in behavioral data.

To do this, I obtained the absolute value of the difference of each individual measurement from

the mean (per treatment per week) and tested these individual deviations in the above specified

repeated measures ANOVA. I then log transformed original data for measures that showed

significant heterogeneity of variance due to noise or the noise*week interaction and tested log-

transformed data in the ANOVA model (Equation 4-1). Reported results are from log-

transformed data for which I found significant heterogeneity of variance, and raw data for which

variance was found to be homogenous.

Physiological Methodology

Necropsy

After 30 days, animals were euthanized for diagnostic necropsy and physiological

measurements. On the day of necropsy, no animal husbandry was completed and activity in the

room was limited to entry and exit to retrieve animals for necropsy. Human movement in front

of tanks still containing animals did not occur.

For retrieval, a researcher entered the room with a bucket containing a solution of 1,000

ppm of 2:1 buffered MS-222 in seawater. Animals in each tank were quickly removed by hand

and placed in the bucket. Animals were then transported to an adjacent satellite necropsy

laboratory with a total of three workers present. After 1 min exposure to MS-222, the animal

was immediately decapitated with a heparinized scalpel, followed by amputation of the entire

tail. One researcher obtained two glucose readings from blood expressed by the cut surface of

the head using an Ascensia Contour Glucometer (Bayer HealthCare). One researcher prepared a

blood smear from blood expressed by the cut surface of the tail. All three researchers then

90

collected remaining blood from the cut surfaces of the head, trunk, and tail using labeled

heparinized microhematocrit tubes. Microhematocrit tubes were sealed with clay and

refrigerated until later processing. All blood was collected within 4 minutes of retrieving the

animal from the experimental tank to avoid acute stress response due to handling (cortisol

concentrations rise typically rise five to ten minutes after the onset of an acute stressor in fishes,

Sumpter, 1997).

The animal was then reassembled, weighed, and measured for standard length (per Lourie,

2003). External health observations were noted, and wet mounts of skin tissue samples prepared

and observed under compound microscope if grossly visible abnormalities were present. Heads

and tails were subsequently discarded. The cutting board and animal trunk were then wiped with

alcohol pads and flame-sterilized. Using sterilized instruments and sterile technique, the trunk

wall was removed, exposing the coelomic cavity. Organs were displaced as necessary to expose

the posterior kidney. One blood agar plate (TSA II 5% Sheep’s Blood, Becton, Dickinson, and

Company) and two Lowenstein-Jensen slants (Becton, Dickinson, and Company) were

inoculated with anterior kidney culture samples to check for bacterial infection. Seahorse trunks

were subsequently eviscerated, organs separated, and the condition of individual organs were

noted. Individual organs were weighed, and wet mount slides of organ samples were prepared

and observed under compound microscope if grossly visual abnormalities were present.

Blood processing

Microhematocrit tubes were spun at 14,000 rpm for 10 minutes. Using dial calipers, the

height of the red cell column was measured and compared to the total height of the column of

whole blood to obtain packed cell volumes (PCV). Tubes were then broken at the packed

cell/plasma interface and plasma was transferred to labeled cryovials that were subsequently

stored at -80oC for cortisol analysis.

91

Blood smears were stained with Dip Quick (Jorgensen Laboratories). Smears were

evaluated for leukocyte count, differential, and heterophil:lymphocyte (H:L) ratio under

microscopy at 1,000X. Each cell type (erythrocytes, lymphocytes, monocytes, heterophils, and

other granulocytes) was counted in one optical field. This process was repeated for each smear

until at least 2,000 cells were enumerated.

Cortisol was measured using a commercially available Cortisol Enzyme Immunoassay

(EIA) Kit (Cayman Chemical, Inc.). All seahorse plasma samples were extracted twice with

diethyl ether prior to assay. The cortisol EIA kit for the lined seahorse was validated by the

parallelism of the dilution curve of a pooled plasma sample of six lined seahorses (not used in

the experiment) with the standard curve of the kit. Based on validation results, a dilution factor

of 1:30 was chosen for assay of experimental animals. Assays were completed according to

manufacturer’s instructions and plates were read with a microplate reader connected to a PC with

Microplate Manager software. Samples that yielded out-of-range or questionable results were

rerun in a second assay, at either a 1:300 or 1:30 dilution factor (respectively). One in-range

sample from the first assay was also run in the second assay to evaluate interassay variation. The

coefficients of variation intra-assay were (mean + SE) 10.6 + 2.1% for assay 1 and 5.4 + 1.6%

for assay 2. The coefficient of variation interassay was 25.6%.

Bacterial culture

Blood agar plates were incubated at 25oC and checked for growth once daily for 4 d.

Positive growth was subcultured and evaluated for morphology and motility, stained with a

Protocol Gram Stain (Fisher Diagnostics), and then submitted to All Florida Veterinary

Laboratory (Archer, FL) for identification. One of the two Lowenstein-Jensen slants per animal

was incubated at 25oC and checked for growth at days 1 & 2, then once a week for 6 weeks. The

other was incubated at 37oC and checked for growth daily for 6 days. Positive growth was

92

evaluated for colony color, cell morphology, and stained with an acid-fast Ziehl-Neelsen Stain

(Becton, Dickinson, and Company).

Measures & statistical analyses

Measures were categorized, and in some cases transformed, as listed in Table 4-1. Among

the morphological indices, the change in Fulton condition factor (ΔK) was calculated as the

difference of the Fulton condition factor at the end of the trial minus the Fulton condition factor

at the beginning of the trial. Fulton condition factor was calculated according to the following

equation, after Anderson and Neumann (1996):

000,1003 ×⎟⎠⎞

⎜⎝⎛=

LWK (4-2)

Where W = weight (g) and L = length (mm).

Data for categories Categories 1-5 were tested in a 3-way factorial MANOVA in SAS

(The SAS Institute, Cary, NC) according to the following model:

y1 y2…yk = noise + sex + trial + noise*sex + noise*trial + sex*trial + noise*sex*trial + error

(4-3)

Of the main effects, noise contained two levels (loud, quiet), sex contained two levels (male,

female), and trial contained two levels (1, 2). Data were not balanced, so Type III sums of

squares were used. Interpretation of MANOVA results are based on the Wilks’ Lambda statistic.

ANOVAs of component measures within each MANOVA were also assessed and are reported.

Results focus on the main effect of noise and any significant interaction effects involving noise.

Other main effects and interactions were included in the model only to partition out variability;

any significant effects not pertaining to noise are not critical to the question and are thus not

reported in detail.

93

As category 6 only included one dependent variable (incidence of disease), this measure

was run as a 3-way factorial ANOVA according to the above model. Chi-square tests were also

employed to test for dependence of pathogen incidence on tank treatment.

Results

Sound Analysis

Ambient noise in loud tanks averaged 123.3 + 1.0 dB SPL (re: 1 μPa) in the middle of the

water column and 137.3 + 0.7 dB at the tank bottom. Ambient noise in quiet tanks averaged

110.6 + 0.6 dB in the middle of the water column and 119.8 + 0.4 dB at the tank bottom. In both

positions, loud tanks were significantly louder than quiet tanks (T-test, p < 0.001, Figure 4-3).

Ethological Results

Ethogram

I classified all observed behaviors as maintenance behavior, as they occurred in the

absence of any external stimulus (except for noise and surrounding habitat). No other animals,

food, or observers were present/visible during observation sessions; thus, feeding and

intra/interspecific interaction behaviors were neither expected nor observed. Maintenance

behaviors were categorized with regard to the animal’s position in relation to the tank, stationary

postures, locomotion, adjustments of the tail while attached to a holdfast, and head movements

(Table 4-2). Position is an artifact of confinement and not translatable to seahorse behavior in

the wild, but in an aquarium setting, I hypothesized that seahorses may position themselves in

quieter areas in the tank; thus, I include it here and have subsequently quantified it (those results

follow).

State analyses

There were no overall differences in the percentage of time animals spent in the behavioral

states that were measured. Animals in both loud and quiet tanks spent 72 + 5% (mean + SE) of

94

observed time in the top half of the tank (F1,30 = 0.01, p = 0.914). There were no differences

among weeks (F3,88 = 1.06, p = 0.372) and no treatment*week interaction effects (F3,88 = 1.38, p

= 0.254). There was no significant heterogeneity of variance observed due to noise (F1,30 = 0.03,

p = 0.855) or interaction (F3,88 = 2.14, p = 0.101) in this measure.

Animals spent 93 + 2% of observed time stationary in loud tanks and 95 + 1% of observed

time stationary in quiet tanks (F1,30 = 0.48, p = 0.492, Figure 4-4). Animals spent more time

stationary in both treatments as weeks progressed (F3,88 = 3.96, p = 0.011). There were no

interaction effects (F3,88 = 1.54, p = 0.211). There was no significant heterogeneity of variance

observed due to noise in this measure (F1,30 = 0.89, p = 0.354), but there was a significant

interaction effect (F3,88 = 4.64, p = 0.005). This is attributable to significantly greater variance in

this measure among animals in loud tanks in week 1 (SDloud = 29.6% vs. SDquiet = 14.3%, F 1,105 =

9.07, p < 0.01).

While stationary, animals spent 74 + 4% of observed time holding in loud tanks and 75 +

3% of time holding in quiet tanks (F1,31 = 0.03, p = 0.857). There were no differences among

weeks (F3,86 = 1.18, p = 0.322) and no interaction effects (F3,86 = 1.80, p = 0.153). There was no

significant heterogeneity of variance observed due to noise (F1,31 = 0.22, p = 0.645) or interaction

(F3,86 = 0.47, p = 0.701) in this measure.

Event analyses

There were no overall differences in the number of adjustments made per hour of rest

between treatments (37 + 6 adjustments in loud tanks and 31 + 3 adjustments in quiet tanks, F1,31

= 0.05, p = 0.825) or among weeks (F3,86 = 1.62, p = 0.191, Figure 4-5). However, there was a

significant interaction effect (F3,86 = 4.14, p = 0.009). At week 1, animals in loud tanks made

significantly more adjustments (69 + 18) than animals in quiet tanks (27 + 7, F1,120 = 7.40, p <

0.01). At week 2, animals in quiet tanks made significantly more adjustments (47 + 7) than

95

animals in loud tanks (26 + 5, F1,120 = 4.70, p < 0.05). There were no differences between

treatments observed at weeks 3 and 4 (p > 0.25 for both weeks). There was significant

heterogeneity of variance observed due to the interaction between noise and week (F3,86 = 4.43, p

= 0.006), driven by greater variance observed in this measure among animals in loud tanks in

week 1 (SDloud = 71.5 vs. SDquiet = 26.8, F 1,120 = 3.92, p < 0.01).

No differences were observed between treatments (F1,30 = 0.03, p = 0.864) or among weeks

(F3,86 = 0.26, p = 0.856) in the number of times animals piped during observations (Figure 4-6).

There was no significant interaction effect (F3,86 = 1.93, p = 0.130). However, animals in loud

tanks demonstrated significantly greater heterogeneity of variance in this measure overall (SDloud

= 142.1 vs. SDquiet = 9.6, F 1,31 = 5.14, p = 0.031), and especially in week 4 (SDloud = 373.6 vs.

SDquiet = 4.1, F 1,103 = 16.50, p < 0.01).

No differences were observed between treatments (F1,30 = 0.45, p = 0.508) or among weeks

(F3,86 = 0.93, p = 0.428) in the number of times animals clicked during observations (Figure 4-7).

There was no significant interaction effect (F3,86 = 2.15, p = 0.100). However, animals in loud

tanks demonstrated significantly greater heterogeneity of variance in this measure overall (SDloud

= 6.7 vs. SDquiet = 3.0, F 1,31 = 5.52, p = 0.025), and especially in week 4 (SDloud = 12.3 vs. SDquiet

= 1.4, F 1,110 = 23.13, p < 0.01).

Animals gaped significantly more often in quiet tanks (12.5 + 1.2) than in loud tanks (7.8 +

0.8, F1,31 = 7.41, p = 0.011, Figure 4-8). No significant differences were observed among weeks

(F3,86 = 1.34, p = 0.267) and no significant interaction effects were observed (F3,86 = 0.29, p =

0.835). Animals in quiet tanks also demonstrated greater heterogeneity of variance overall in

this measure (SDloud = 6.4 vs. SDquiet = 9.3, F 1,31 = 7.41, p = 0.011).

96

Physiological Results

Morphological indices

Morphological indices were significantly different between animals in loud tanks and

animals in quiet tanks (p = 0.024). Animals in loud tanks declined in weight (∆Wt = -2.22 +

0.30 g) and condition factor (∆K = -0.0449 + 0.009) significantly more than did animals in quiet

tanks (∆Wt = -1.44 + 0.28 g, p = 0.039, ∆K = -0.0019 + 0.0126, p = 0.007). There was no

significant difference in HSI (p = 0.812) or GSI (p = 0.326) between tank treatments. Pertinent

interaction effects were not significant.

Hematological count

There were no significant differences between treatments for hematological count

measures (p = 0.660). There were no significant effects of noise or any interactions pertaining to

noise for any individual dependent variable.

Leukocyte differential

There were no significant differences between treatments for leukocyte differentials as a

group (p = 0.254). While lymphocytes constituted a smaller proportion of the lymphocyte

population in loud tanks than in quiet tanks (59.0 + 5.3% vs. 71.0 + 3.4%), differences between

treatments were marginally significant (p = 0.051). Heterophils constituted a significantly

greater proportion of the lymphocyte population in loud tanks than in quiet tanks (35.8 + 4.9%

vs. 22.5 + 3.6%, p = 0.028). Consequently, the heterophil:lymphocyte (H:L) ratio was

significantly greater among animals in loud tanks (0.88 + 0.25) than animals in quiet tanks (0.36

+ 0.7, p = 0.029). Pertinent interaction effects were insignificant.

97

Blood glucose concentration

Blood glucose concentrations were not significantly different between treatments (31.1 +

1.8 mg/dL in loud tanks vs. 34.0 + 2.5 mg/dL in quiet tanks, p = 0.310). There were no pertinent

interaction effects.

Plasma cortisol concentration

Plasma cortisol concentrations were not significantly different between treatments (median

values of 7.23 ng/mL in loud tanks vs. 5.75 ng/mL in quiet tanks, p = 0.113). There were no

pertinent interaction effects.

Incidence of disease

In general, animals in both treatments exhibited variable levels of parasitism, bacterial

infection, and organ pathology. Six of 16 animals in each treatment presented with dermal cysts

that revealed infection by Glugea heraldi as described by Blasiola (1979) and Vincent and

Clifton-Hadley (1989). The digenean Dictysarca virens (described in Manter, 1947) was found

in the swimbladders of 7 of 16 animals in loud tanks and 9 of 16 in quiet tanks; prevalence did

not depend on treatment (X2 test, p > 0.1). Nematodes and encysted metazoan parasites were

found variably on the surfaces of organs, and in the coelom around the anus. Presence of

metazoan parasites only depended on treatment for kidneys (metazoans found on 10 of 16

animals in loud tanks and 7 of 16 animals in quiet tanks, X2 test, p < 0.05). Other gross

abnormalities were commonly observed in the liver, that was enlarged, pale, vascularized, and

exhibited bile accumulation in the majority of animals from both treatments.

Bacteria were cultured on blood agar from three animals. Aeromonas hydrophila (Gram

negative motile bacilli) were cultured from the posterior kidneys of two males in quiet tanks.

Vibrio vulnificus (Gram negative motile bacilli) was cultured from the posterior kidney of one

female in a loud tank. These results were not sufficient for statistical testing between treatments.

98

Lowenstein-Jensen media incubated at 37oC yielded no growth from any animals.

Lowenstein-Jensen media incubated at 25 oC yielded pale yellow colored growth of acid-fast

positive bacilli by 2 weeks or thereafter (indicative of Mycobacterium sp. infection) from 7

animals in loud tanks and 6 animals in quiet tanks; these proportions were not dependent on

treatment (X2 test, p > 0.05).

The number of organs affected by pathogens were not significantly different between

treatments or sexes, but trial 1 animals had fewer organs infected than trial 2 animals (2.9 + 0.4

vs. 4.1 + 0.3 organs respectively, p = 0.007).

Physiological results are summarized in Table 4-3.

Discussion

The Stress Concept

Within the paradigm of Selye’s (1950) General Adaptation Syndrome (GAS), organisms

initially exhibit an alarm response to a stressor (whether acute or chronic). In the presence of a

persistent (i.e., chronic) stressor, the organism adjusts or compensates for the disturbance to

achieve allostasis, essentially making physiological compromises to achieve stability in a

suboptimal environment (Sterling and Eyer, 1988). If the organism is unable to cope with the

chronic stressor, it enters the third stage of the syndrome, exhaustion, that can lead to the

development of a pathological condition or death. The aquaculture literature has abundant

evidence of acute stressors in aquaria eliciting primary (e.g., plasma cortisol), secondary (e.g.,

hematology and clinical chemistry), and tertiary (e.g., avoidance behavior) stress responses in

fishes. Chasing (e.g., Papoutsoglou et al., 1999), net confinement (e.g., Kubokawa et al., 1999),

exposure to air (e.g., Barcellos et al., 1999; Belanger et al., 2001), handling and transport (e.g.,

Iversen et al., 1998; Belanger et al., 2001,), and acute exposure to toxicants (e.g., Jian-yu et al.,

99

2005) have all been shown to elicit stress reponses at these levels. The stress responses to

chronic stressors encountered in aquaculture settings, however, is less clear.

In a chronic stress situation, an organism could be at any point along a continuum of stress

response in either the second or third stage of Selye’s GAS. The results of indices of stress

measured in fishes exposed to chronic stressors are highly variable, and reflect points along this

continuum. This picture is further confounded by variability in the nature, magnitude, and

duration of the stressor, the species of fish tested, and even the individual’s genetic background.

My results demonstrate the complexity of evaluating chronic stress from a mosaic of primary,

secondary, and tertiary response indices.

Primary Stress Indices

Cortisol, the primary stress response, is the indicator most often measured in studies of

chronic stress. The common pattern that emerges in chronic stress studies of aquacultured fish,

especially when cortisol is measured on a time series, is a peak in concentration that occurs

within minutes to hours after the onset of the stressor, followed by a decline that occurs over

days to weeks to resting levels that are nevertheless higher than unstressed controls. Such is the

case in reported studies of chronic confinement stress, altered photoperiods, daily chasing, and

subordinate fish in a dominance hierarchy (e.g., Pottinger et al., 2002; Biswas et al., 2004;

Barcellos et al., 1999; Sloman et al., 2001; respectively).

In other studies, the cortisol peak in response to the onset of a chronic stressor resolves to

control levels despite the continued presence of the stressor. Crowding stress elicited this

phenomenon in carp (Cyprinus carpio, Ruane et al., 2002; Ruane and Komen, 2003), rainbow

trout (Oncorhynchus mykiss), and brown trout (Salmo trutta, Pickering and Pottinger, 1987),

despite differences from controls demonstrated in the measurement of other physiological stress

indices. This phenomenon of resolution of corticoid titers to pre-stress levels in the continued

100

presence of a stressor has been termed ideal compensation (Precht, 1958, in Schreck, 1981).

Despite ideal compensation, Pickering and Pottinger (1987) were concerned that chronically

stressed fish might still be limited in their performance capacities, noting increased mortality

rates in other studies, reduced growth rates in other studies as well as theirs, and reduced

immunocompetence as indicated by reduced lymphocyte counts in blood of their study fish.

I did not detect any significant elevations in plasma cortisol concentrations in seahorses

after one month of exposure to noise. Inherent individual variability (cortisol levels ranged

widely in both treatments) may have reduced statistical power to detect an effect, but I also

believe that plasma cortisol concentrations, if they had risen at the onset of exposure, may have

resolved over time via coping to achieve allostasis. It would have been informative to document

cortisol dynamics throughout the course of the experiment; this warrants future work.

Secondary Stress Indices

Of note among the secondary indices that were measured in this study is the response in

leukocyte differential, that fell just short of demonstrating lymphocytopenia, but did demonstrate

significant heterophilia among animals in loud tanks. Cortisol induces karyorrhexis of

lymphocytes (Dougherty, 1960) and increases blood heterophils but inhibits the migration of

these cells to injured sites or inflammatory lesions and slows down wound healing (Wendelaar

Bonga, 1997). A suite of studies have demonstrated lymphocytopenia and/or heterophilia in

response to acute and chronic stress in fish (e.g., McLeay, 1973; McLeay and Brown, 1974;

Ellsaesser and Clem, 1986; Barcellos et al., 2004; Svobodová et al., 2006). The leukocyte

differential results reported here are thus indicative of a chronic stress response.

This pattern of change in the leukocyte differential has led scientists to invoke the

heterophil to lymphocyte ratio (H:L ratio) as a measure of stress in animals. Gross and Siegel

(1983) established this measure for use in birds, demonstrating that administered exogenous

101

corticosterone (the primary stress hormone in birds) led to correlative increases in plasma

corticosterone and H:L ratio. Since then, the H:L ratio has been used in studies of acute and

chronic stress in a variety of animals such as birds (e.g., Al-Murrani et al., 2002; Huff et al.,

2005; Campo et al., 2005, 2007), reptiles (e.g., Case et al., 2005), and mammals (e.g., Hansen

and Damgaard, 1993; Fisher et al., 1997), including people (e.g., Duffy et al., 2006). Among

these studies, increased H:L ratios were associated with measures of primary, secondary, and

tertiary stress indices, including increased plasma cortisol/corticosterone concentrations (Gross

and Siegel, 1983; Hansen and Damgaard, 1993; Fisher et al., 1997), increased susceptibility to

infectious disease (Al-Murrani et al., 2002; Huff et al., 2005), increased occurrence or duration

of distress behaviors (Hansen and Damgaard, 1993; Case et al., 2005; Campo et al., 2005, 2007),

reduced weight, and mortality (Huff et al., 2005; Duffy et al., 2006). This study demonstrates

an association between increased H:L ratios and weight loss, reduced body condition, and

increased variability in the occurrence of distress behaviors among animals in the loud tank

group. To my knowledge, this paper is the first to employ the H:L ratio as a measure of chronic

stress in fishes.

Hyperglycemia is a consistent and well-documented response to acute stressors (Mazeaud

and Mazeaud, 1981), but blood glucose responds inconsistently in studies of chronic stress. Fish

can become hypoglycemic or hyperglycemic in response to various chronic stressors (McLeay,

1973; McLeay and Brown, 1974; Pottinger et al., 2002; Sala-Rabanal et al., 2003). In other

studies, blood glucose variations in fish were not detected in response to altered photoperiod

regimes, despite cortisol profiles indicative of ideal compensation (Biswas et al., 2006) or even

chronically elevated cortisol (Biswas et al., 2004). The non-remarkable results presented here are

102

consistent with the referenced studies that also measured blood glucose concentrations after

chronic exposure to a stimulus affecting sensory systems.

Acute stress typically results in hemoconcentration and therefore increased hematocrit,

mainly due to the action of catecholamines secreted by chromaffin cells (a primary stress

response). Catecholamines cause erythrocytes to swell and also increase erythrocyte numbers by

causing splenic contraction with release of erythrocytes into the circulation (Wendelaar Bonga,

1997). The hematocrit response in chronic stress situations, on the other hand, is quite variable,

showing fluctuations in either direction (e.g., Pickering and Pottinger, 1987; Montero et al.,

1999; Barcellos et al., 2004) or no response at all (McLeay, 1973; McLeay and Brown, 1974;

Sala-Rabanal et al., 2003; Biswas et al., 2004). Chronic ambient noise exposure did not elicit a

hematocrit response in this study.

Tertiary Stress Indices

Physiology

Despite variations in the leukocyte differential between treatments, this study did not

demonstrate differences in incidence of disease between treatments. Animals in loud and quiet

tanks had similar prevalence of infection by acid-fast positive bacilli, indicative of

Mycobacterium sp., identical prevalence of infection by Glugea heraldi, and similar prevalence

of infection by the swimbladder parasite Dictysarca virens. Virtually all animals, regardless of

treatment, presented with encysted internal metazoan parasites, whether they were located in the

coelom or in/on organs. Results demonstrating greater incidence of encysted metazoans on

kidneys in loud tanks may be coincidental; often cysts were found on the capsules of organs and

may have originated from elsewhere within the coelom. The prevalence of pathogens in both

treatments was quite surprising and speaks to the general prevalence of disease in wild

specimens of this species. This variable but prevalent baseline pathogen load may have masked

103

differences due to immunosuppression, especially in presence/absence data, that does not

characterize or differentiate the magnitude of pathogen load in animals between treatments.

Furthermore, the time course over which animals were exposed to their treatments (one month)

may not have been long enough to allow disease processes to develop differentially between

treatments. Other studies (reviewed in Schreck, 1996; Balm, 1997; Wendelaar Bonga, 1997)

have shown differential prevalence of infection between acutely or chronically stressed and non-

stressed fish when inoculated with a specific pathogen, but this methodology was outside of the

scope of the present study and could have obfuscated the causative factor of other observed

stress response indicators.

The hepatosomatic index (HSI) in this study was not affected by ambient noise. HSI’s in

fishes are commonly altered in response to chronic stressors that directly affect liver function,

such as fasting (e.g., Kakizawa et al., 1995), malnutrition (e.g., Montero et al., 2001;

Papoutsoglou et al., 2005), and toxicant exposure (e.g., Datta and Kaviraj, 2003; Porter and Janz,

2003; Gagnon et al., 2006; Khan, 2006). However, the HSI can also decrease in response to

other stressors that do not intuitively affect liver function, such as crowding stress (Montero et

al., 1999) and cortisol implantation (Vijayan and Leatherland, 1989). But Papoutsoglou et al.

(2000, 2005) did not detect decreased HSI’s in fishes exposed to stressful (as evidenced by

chronically elevated plasma cortisol concentrations) tank background colors. Stress-inducing

sensory stimuli may not have a significant effect on HSI. It is worth noting, however, that many

of the seahorses, regardless of treatment, presented with pale, enlarged, fatty livers, a common

phenomenon among captive fishes (e.g., Grant et al., 1998). This may have masked any chronic

stress response due to ambient noise in the measure of HSI.

104

It is well-established that repeated acute and chronic stress can downregulate reproduction

in fishes, decreasing reproductive indices, including gonad size and GSI (Billard et al., 1981;

Pankhurst and Van Der Kraak, 1997). Though a significant decrease in GSI was not

demonstrated among seahorses housed in loud tanks, a trend in this direction was nonetheless

apparent among female seahorses.

Seahorses demonstrated decreases in two of the arguably most important measures of

success in (food) fish culture, in weight gain and change in condition factor. Cortisol is known

to depress growth rate both directly and through reduced food intake (e.g., Gregory and Wood,

1999). These mechanisms of weight/condition factor reduction could not be teased apart due to

the methodology employed in this study. However, body condition results coupled with results

from the leukocyte differential suggest that animals in loud tanks may have responded with

increased plasma cortisol profiles earlier in the experimental trial, with subsequent resolution

under the aforementioned ideal compensation scenario. It should be noted that another sensory

stressor, tank color, elicited both elevated cortisol profiles and reduced growth in summer

flounder (Paralichthys dentatus) (Cotter et al., 2005). In the context of the variability observed

in other measures, the ability to subtract initial values from final values in these two measures

eliminated a substantial degree of variability due to individual differences, perhaps offering more

power in the statistical tests employed for these measures. Van Weerd and Komen (1998) also

point out the masking effect of individual differences in assessing chronic stress response in

fishes.

Behavior

A trend that became apparent after reviewing quantitative behavioral data was increased

variability in the loud tank group for several measures. Noting this, I examined heterogeneity of

variance more rigorously, partitioning the contribution of factors to this phenomenon. Orlando

105

and Guillette’s (2001) review of population responses to environmental stressors documents

greater variance of responses in comparison to control populations. In fact, this phenomenon has

led other workers to document sub-populations of species with different stress-coping styles

(e.g., stress proactive or stress reactive). These populations demonstrate bimodal distributions in

physiological and behavioral stress response measures that co-vary (Koolhaas et al., 1999; Øverli

et al., 2007). This complex response pattern within a population can lead to Type II errors if

only measures of central tendency, and not dispersion, are compared. In this section, I discuss

measures of central tendency and heterogeneity of variance in behavioral results.

Animals consistently display avoidance responses to acute stressors (e.g., Beitinger, 1990),

including anthropogenic noise (e.g., Blaxter et al., 1981; Blaxter and Batty, 1987; Dunning et al.,

1992; Knudsen et al., 1992; Knudsen et al., 1997), provided that an escape to a location where

the acute stressor is no longer present is available. While seahorses were confined in tanks,

limiting avoidance capability, sound gradients within the tank provided the opportunity for

animals to choose quieter locations. These sound gradients were demonstrated by hydrophone

recordings, in which mid-water column recordings were 9.2 dB quieter than bottom recordings in

quiet tanks and 14.0 dB in loud tanks. Behavioral results showed no differences in the

percentage of time spent in the top half of the tank vs. the bottom half of the tank between

treatments. However, animals in both treatments spent the majority of time in the top half of the

tank, where ambient noise is quieter. This result should be interpreted with caution, as other

factors (e.g., different holdfast or habitat types) could explain this preference.

Masonjones and Babson (unpublished) conducted an experiment exposing the dwarf

seahorse (Hippocampus zosterae) to boat motor noise in tanks. They found, among other

measures, that H. zosterae spent less time attached to a holdfast in noise-exposed tanks than in

106

quiet tanks. I expected to see a similar result in this study. I found similar responses, but only in

week one, when animals in loud tanks made significantly more adjustments of the tail than did

animals in quiet tanks. This measure was also more variable among loud tank animals. I also

partitioned out a greater variability in the percentage of time spent while stationary among the

loud tank animals in the first week. Collectively, increased occurrence or variability in these

behaviors may have indicated an irritation response by some animals in loud tanks when attached

to holdfasts emanating potentially irritating vibrations. The fact that these measures were not

significantly greater (in mean or variance) among loud tank animals in weeks 2-4 indicate that

habituation may have occurred, whereby the stimulus was no longer perceived as irritating. A

habituation response is consistent with the results of Dunning et al. (1992), who reported that

alewives (Alosa pseudoharengus) habituated when exposed to continuous tones, but not to pulses

of sound. It is difficult, however, to explain why this trend reversed in the measure of

adjustments in week 2, when animals in quiet tanks made significantly more adjustments.

Some animals in loud tanks displayed extremely high frequencies of clicking and piping,

leading to heterogeneity of variance in these measures between treatments overall, and especially

in week 4. Piping occurs among sick captive fishes and is thought to ameliorate hypoxia by

making use of a thin layer of oxygen-rich water at the surface (Francis-Floyd, 1988). Piping may

not serve the same function in seahorses. Seahorses extend the snout beyond the air-water

interface and into the air during piping. Occasionally seahorses expelled bubbles from the snout

after piping, indicative of air intake. Piping may nonetheless constitute a pathologic behavior. I

have witnessed this behavior to occur frequently in moribund syngnathids.

Clicking in seahorses occurs in several contexts. It is known to occur during feeding

(Colson et al., 1998), and aggression/competition (Vincent, 1994b), though its frequency of

107

occurrence in response to a stressor has never been tested prior to this study. Fish (1953) noted

clicking in H. erectus in response to transfers to new seawater containers, and in response to

rapid transfer between two water bodies that differed in temperature by 12oC; these transfers may

have represented acute stressors. In the context of a stressful environment, clicking might serve

as a distress vocalization. Distress vocalizations are elicited in mammals and birds when 1) in

the clutches of a predator (Högstedt, 1983), and 2) separated from their mothers or peers (for

social animals); the latter phenomenon has been associated with a physiological stress response

and serves to advertise location to lost conspecifics (Norcross and Newman, 1999; Stahl et al.,

2002; Feltenstein et al., 2003). Most seahorse species (including H. erectus) are social animals,

remaining pair-bonded to mates throughout at least one breeding season and perhaps longer

(Foster and Vincent, 2004); a distress vocalization could serve the latter function and be

enhanced under stress. Seahorses in this experiment were housed in isolation, providing

separation stimulus, and the observed trend of increased numbers of clicks occurring among

some animals in loud tanks is concordant with physiological indices of increased stress status.

Further experimentation may yield interesting insights on the function of clicking in this and/or

other social contexts.

Animals in quiet tanks gaped more frequently and variably than did animals in loud tanks.

Limited observations on gaping in fishes in scientific literature suggests that gaping is associated

with low activity levels (Rasa, 1971), that is somewhat consistent with these results

demonstrating less frequent adjustments in animals in quiet tanks, but only in week 1.

Conclusions

The suite of physiological and behavioral stress measurements examined in this study

suggest that ambient noise from pumps in aquaculture settings at the levels tested constitutes a

chronic, subtle stressor to seahorses. Seahorses responded both behaviorally and physiologically

108

with altered secondary and tertiary stress indices indicative of a chronic stress response that has

resolved to ideal compensation in an environment necessitating allostasis. As a result, though

the primary stress index (e.g., plasma cortisol) was not detected after one month of exposure, the

chronic stressor still exerted deleterious long-term effects, including altered leukocyte profiles

and reduction in weight and condition factor. Behavioral measures suggest initial disturbance

followed by habituation, until the duration of exposure eventually elicits pathological and

concomitant distress behaviors among some susceptible animals (e.g., piping and clicking,

respectively). In light of these results, seahorse aquarists and aquaculturists are advised to

consider the acoustic environment of their aquaria, and to incorporate soundproofing

modifications during design and set-up of facilities to avoid the debilitating consequences that

chronic loud noise exposure can have on fish health, growth, and welfare.

109

Table 4-1. Categorization of physiological measures for statistical analysis. 1) Morphological Indices a) Change in weight (Δwt) = End wt – Beginning wt (g) b) Change in condition factor (ΔK) = Kend – Kstart (per Anderson and Neumann, 1996) c) Hepatosomatic Index (HSI) = liver wt / (body wt – gonad wt) d) Gonadosomatic Index (GSI) = gonad wt / (body wt – gonad wt) 2) Hematological Count a) Leukocytes per 2,000 cells b) Lymphocytes per 2,000 cells c) Monocytes per 2,000 cells d) Heterophils per 2,000 cells e) Packed Cell Volume (%), averaged over microhematocrit tubes collected for each animal 3) Leukocyte Differential a) Lymphocytes per total leukocytes observed (%) b) Monocytes per total leukocytes observed (%) c) Heterophils per total leukocytes observed (%) d) Heterophil to Lymphocyte (H:L) ratio (square-root transformed) 4) Blood Glucose a) First reading b) Second reading 5) Plasma Cortisol: Non-parametric results were ranked and categorized as: a) First reading b) Second reading 6) Incidence of Disease: Number of tissues/organs in which pathogens were found

110

Table 4-2. Ethogram of the lined seahorse, Hippocampus erectus: Maintenance behavior. Position Position of animal in relation to the tank.

Top Any component of the animal’s body is located in the top half of the vertical

height of the tank. Bottom The entire animal is located in the bottom half of the vertical height of the tank. Stationary Postures while stationary.

Perch Animal is attached to holdfast with less than ½ of tail. The rest of the tail, trunk,

and head are extended and not in contact with the holdfast, and angled above the horizontal plane.

Hold Animal is attached to holdfast with ½ of tail or greater. Trunk and head are either touching or are less than 1cm away from holdfast.

Extend Animal is attached to holdfast with less than ½ of tail. Body is outstretched away from the holdfast, angled at or below the horizontal plane.

Sit Animal is stationary on the bottom of the tank, with tail curled on the bottom. Head and trunk are in an upright position.

Locomotion Postures while locomoting.

Swim Animal travels horizontally, obliquely, or vertically through the water in an obliquely upright or upright position, undulating pectoral and dorsal fins. The tail may either be curled forward, extended below, or extended trailing behind the animal.

Skim Animal travels across the bottom of the tank with trunk in an obliquely upright or upright position, undulating pectoral and dorsal fins. The tail drags on the bottom, outstretched behind the animal.

Adjustments Movements of the tail with respect to the animal’s holdfast.

Shift Animal undulates tail while attached to the holdfast. Head and trunk positions remain stationary in relation to the holdfast. There is no net displacement of the tail with respect to its initial position on the holdfast.

Rotate Animal rotates tail and body around holdfast. Trunk remains upright during this event.

Slide Animal ascends or descends a vertical holdfast by inching upward or downward through tail movements on the holdfast. Head or trunk movement is minimal. Animal may either be perched or extended, thus only the lower half of the tail is in contact with the holdfast.

Head Movements

Movements of the head or its parts.

Click Head tilts upward instantaneously, hyoid protrudes, mouth opens. Associated with a click sound.

Pipe Animal’s snout breaks water surface while opercular movements continue. Gape Animal opens mouth slowly, slightly tilts head upward, and protrudes hyoid.

There is no click sound.

111

Table 4-3. Summary of physiological results. Significant p-values are in bold. F = MANOVA (for categories) or ANOVA (for individual measures and “organs affected”), X2 = Chi-Square Test, subscripts denote degrees of freedom. NV = Not valid. Non-parametric plasma cortisol reported as 1st Quartile<Median<3rd Quartile.

Physiological Measure Loud Tanks Quiet Tanks Test Mean + SE Mean + SE Statistic p

Morphology F4,21 = 3.51 0.024∆ Wt (g) -2.22 + 0.30 -1.44 + 0.28 F1,24 = 4.79 0.039∆K -0.0449 + 0.0092 -0.0019 + 0.0126 F1,24 = 8.70 0.007HSI 0.028 + 0.003 0.026 + 0.002 F1,24 = 0.06 0.812GSI

0.003 + 0.001 ♂0.021 + 0.006 ♀

0.002 + 0.001 ♂0.033 + 0.008 ♀

F1,24 = 1.00 0.326

Haematological Count F5,17 = 0.66 0.660Leukocytes 73.9 + 8.3 71.7 + 9.7 F1,21 = 0.07 0.790Lymphocytes 43.3 + 6.3 51.1 + 7.2 F1,21 = 1.54 0.229Monocytes 2.2 + 0.5 2.4 + 0.6 F1,21 = 0.00 0.951Heterophils 27.1 + 6.4 16.8 + 3.7 F1,21 = 1.38 0.253PCV (%)

21.5 + 1.3 20.6 + 1.3 F1,21 = 0.20 0.663

Differential F4,19 = 1.88 0.254Lymphocytes (%) 59.0 + 5.3 71.0 + 3.4 F1,22 = 4.26 0.051Monocytes (%) 3.0 + 0.7 4.3 + 1.2 F1,22 = 1.16 0.294Heterophils (%) 35.8 + 4.9 22.5 + 3.6 F1,22 = 5.53 0.028H:L Ratio 0.88 + 0.25 0.36 + 0.7 F1,22 = 5.43 0.029

Clinical Chemistry Blood Glucose (mg/dL) 31.1 + 1.8 34.0 + 2.5 F2,22 = 1.24 0.310Plasma Cortisol (pg/mL)

4.76<7.23<23.09 2.53<5.75<21.69 F2,21 = 2.42 0.113

Pathogen Incidence Glugea heraldi n = 6 n = 6 χ2

1 = 0.000 >0.9

Dictysarca virens n = 7 n = 9 χ21

= 0.500 >0.1Mycobacterium sp.

n = 7 n = 6 χ21

= 0.130 >0.5

Organs Affected 3.75 + 0.35 3.31 + 0.40 F1,24 = 1.02 0.322Liver n = 10 n = 7 χ2

1 = 1.129 >0.1

Kidney n = 13 n = 7 χ21

= 4.800 <0.05GI Tract n = 3 n = 8 χ2

1 = 3.580 >0.05

Gonad n = 2 n = 5 NV NVCoelom n = 5 n = 3 NV NV

112

Figure 4-1. Experimental tank design schematic: Loud tank (see text for additional details).

113

Figure 4-2. Experimental tank design schematic: Quiet tank (see text for additional details).

114

Figure 4-3. Power spectra of ambient noise in representative tanks. a = recordings from the middle of the water column, b = recordings from tank bottom. Black trace = loud tank, gray trace = quiet tank.

a b

115

50

60

70

80

90

100

1 2 3 4

Week

% T

ime

Stat

iona

ry(w

hile

in s

ight

)

††

Figure 4-4. Proportion of time spent stationary. Bars represent mean values, error bars represent standard deviations. Gray bars indicate loud tanks, white bars indicate quiet tanks. Of type III tests of variance sorted by week, †† p < 0.01.

0

20

40

60

80

100

120

140

1 2 3 4

Week

N A

djus

tmen

ts p

er H

our

whi

le S

tatio

nary

††**

*

Figure 4-5. Number of adjustments made per hour while stationary. Bars represent mean values, error bars represent standard deviations. Gray bars indicate loud tanks, white bars indicate quiet tanks. Of type III tests of means sorted by week, * p < 0.05, ** p < 0.01. Of type III tests of variance sorted by week, †† p < 0.01.

116

0

100

200

300

400

500

1 2 3 4

Week

Pipe

s (n

) ††

Figure 4-6. Occurrence of piping. Bars represent mean values, error bars represent standard deviations. Gray bars indicate loud tanks, white bars indicate quiet tanks. Piping was significantly more variable among animals in loud tanks overall (type III test of variance, p = 0.031). Of type III tests of variance sorted by week, †† p < 0.01.

0

2

4

6

8

10

12

14

16

18

20

1 2 3 4

Week

Clic

ks (n

) ††

Figure 4-7. Occurrence of clicking. Bars represent mean values, error bars represent standard deviations. Gray bars indicate loud tanks, white bars indicate quiet tanks. Clicking was significantly more variable among animals in loud tanks overall (type III test of variance, p = 0.025). Of type III tests of variance sorted by week, †† p < 0.01.

117

0

5

10

15

20

25

30

1 2 3 4

Week

Gap

es (n

)

Figure 4-8. Occurrence of gaping. Bars represent mean values, error bars represent standard deviations. Gray bars indicate loud tanks, white bars indicate quiet tanks. Overall, animals yawned significantly more often, and more variably, in quiet tanks than in loud tanks (type III tests of means and variance, p = 0.011).

118

CHAPTER 5 ACOUSTIC ROLES AND EFFECTS IN PREY CAPTURE BEHAVIOR OF LINED

SEAHORSES (HIPPOCAMPUS ERECTUS) IN AQUARIA

Introduction

It is well known that one of the myriad effects of stress on animals is a decrease in energy

balance, as animals reallocate metabolic energy away from investment activities (i.e., growth and

reproduction) and toward homeostatic mechanisms that require further input in a stress situation

(i.e., respiration, locomotion, hydromineral regulation, tissue repair, etc., Wendelaar-Bonga,

1997). This altered maintenance has been termed allostasis (Sterling and Eyer, 1988). There are

multiple insults to energy balance during stress, including reductions in appetite, food intake,

food assimilation, metabolic rate (Wendelaar-Bonga, 1997), and potentially via alterations of

growth promoting hormone concentrations, though the latter effect is quite variable among fishes

(Pankhurst and Van Der Kraak, 1997). Though reduced growth is a common indicator of

chronic stress (Morgan and Iwama, 1997), the contribution of each of the above-mentioned

mechanisms to this effect is often unknown.

In the aquaculture industry, fish growth and condition are important contributors to product

marketability. As such, it behooves aquaculturists to design and maintain systems and

husbandry practices that minimize exposure of fish to acute and/or chronic stressors that would

negatively impact these qualities. In the design of filtration systems to optimize water quality for

fish health, aquaculturists may be inadvertently exposing fish to loud ambient noise, a chronic

stressor resulting in reduced growth and body condition (Chapter 4).

Is this effect mediated via reduced food intake? If so, would a sound-induced reduction of

intake stem from the downregulation of appetite and feeding response, or in the case of

soniferous fishes that produce sound associated with feeding behavior, can loud ambient noise

interfere with acoustic stunning of prey?

119

The last question leads to closer examination of the prey stunning hypothesis. Also termed

the acoustic predation hypothesis, it suggests that some aquatic animals (odontocetes and

snapping shrimp, in particular) stun or kill prey with high amplitude sound. Norris and Møhl

(1983) first proposed the hypothesis based on morphological considerations and observations in

the wild. Other observations in the wild (Marten et al., 2001; Simon et al., 2005) and acoustic

simulation experiments (Zagaeski, 1987; Mackay and Pegg, 1988; Martin et al., 2001)

corroborate the hypothesis, but it is contentiously debated. In some cases, sounds associated

with prey capture behavior cannot be decoupled from physical disturbance caused by the

behavior, rendering the cause of prey-stunning effects unknown (Simon et al., 2005). In the case

of snapping shrimp, prey damage has been linked to water jets produced by snapping claws and

not to acoustic snaps caused by cavitating bubbles (Herberholz and Schmitz, 1999; Versluis et

al., 2000). Furthermore, at least one acoustic simulation experiment has yielded no effects on

prey behavior (Benoit-Bird et al., 2006).

To the author’s knowledge, the prey stunning hypothesis has not yet been proposed for any

fish, but sound production is known to occur concurrently with feeding among seahorses. The

seahorse click is a broadband sound caused by the stridulation of the posterior process of the

supraoccipital against the coronet and is a component of the seahorse feeding strike (Colson,

1998). Clicks can be quite loud, reaching peak sound pressure levels of 136.8 + 1.0 dB SPL (re:

1 µPa) and peak spectral accelerations of 1.52 X 10-3 + 1.87 X 10-4 m/s2 (Chapter 3). The

seahorse thus merits consideration as a model for testing the prey stunning hypothesis.

This study tests the prey stunning hypothesis in the lined seahorse, Hippocampus erectus,

while also testing the effects of ambient noise on prey capture rate and success.

120


Animals were accessioned, tagged, quarantined, and maintained as described in Chapter 4,

except that animals received 12 hours of fluorescent light daily, with low-intensity incandescent

dawn and dusk lights illuminated ½ hour before and ½ hour after fluorescent lighting transition.

Throughout the experiment, temperature was maintained at 24ºC, salinity at 27 ppt, and pH at

8.3. Ammonia-nitrogen and nitrite-nitrogen levels remained at 0 ppm throughout holding and

experimental periods. Nitrate-nitrogen levels ranged from 2.8 to 5.6 ppm.

Experimental Tank Design

Two experimental tanks were designed as loud tanks and two as quiet tanks, as described

in Chapter 4. Tank flow rates were measured on each day and averaged 439 L/h; flow rates of

loud tanks were adjusted daily to match flow rates of quiet tanks. For behavioral observations,

an opaque white curtain was hung 1 m away from tank fronts to conceal observers from animals.

Small rectangular holes (10.5 X 7.0 cm) were cut in the curtain in front of each tank to allow for

behavior recording with a videocamera.


The ambient noise profile of each test tank was measured daily. One minute recordings

were taken from the middle of the water column and also at the bottom of the tank, at a sampling

rate of 44.1 kbps, with an HTI-96-min hydrophone (High Tech Instruments, Inc., sensitivity = -

164.1 dB re: 1V/μPa) connected to a NOMAD Jukebox3 digital audio recorder. The NOMAD

Jukebox3 was calibrated with a 1.0Vpeak sine wave.

Digital sound files (.wav) were post processed by removing putative artificial electrical

peaks in frequency spectra (at 60 Hz or its harmonics) using the FFT filter function in CoolEdit

(Syntrillium Software Corporation). These noise reduced files were then analyzed with

SpectraPlus (Pioneer Hill Software), calibrated according to manufacturers instructions.

121

SpectraPlus analysis settings are listed in Table 2-1. From the frequency domain spectrum, peak

frequency (in Hz), peak amplitude, and total RMS power (both in dBpeak re: 1 µPa) were

documented. Signals were low-pass filtered at 1,002 Hz; total RMS power levels are thus

summed within this frequency range, which is pertinent for hearing generalist fishes (including

H. erectus, Chapter 3) that have insensitive hearing above 1,000 Hz (Fay, 1988b). These

measures were tested between loud and quiet tanks for both recording positions with t-tests.

Muting

Surgical procedure

Nine animals were surgically muted after Colson et al. (1998). Briefly, animals were

anesthetized in a solution of 175 ppm of tricaine methanesulfonate (MS-222) in tank water. The

surgical site was cleaned with a diluted betadine solution prior to surgery, and sterile technique

was observed. With the upper head of the animal held above the water line, a longitudinal

incision was made immediately anterior to the coronet, exposing the posterior process of the

supraoccipital. The process was clipped with rongeurs. Skin was apposed and sealed with

cyanoacrylate. Eight additional animals were subjected to a control surgical procedure, where

the right post-temporal process was removed using the same procedure (this process is not

involved in sound production).

All surgical wounds were treated once daily with topical application of 1% silver

sulfadiazine cream for a minimum of 2 weeks or until fully healed.

Click recording and analysis

After healing, muting effect was checked by recording seahorse clicks during feeding. To

do this, a polysterene fish shipping box was set up as a recording chamber within a quiet room.

The box was filled with system water and the HTI-96min hydrophone was positioned in the

center of the box at the middle of the water column and connected to the NOMAD Jukebox3 for

122

digital audio recording. Up to three seahorses were placed in the box at any one time and an

aliquot of live Artemia sp. was added to the box. Recording began when any one animal made a

first strike and ceased when animals stopped feeding. The fish ID and timestamp of each strike

were documented during recording.

Documented clicks were isolated in the center of windows of 5 s duration, copied and

pasted into new, individual sound files for each click. Resulting sound files were then low-pass

filtered at 1,002 Hz using CoolEdit software. Files were then calibrated for voltage in

SpectraPlus according to manufacturer instructions and post-processed. Peak amplitude (in

dBpeak re: 1 µPa) was measured in the time domain waveform. Five to 18 clicks were measured

from each animal (with a mean of 9.9 + 0.7 clicks per animal) and peak amplitudes were

averaged. Average peak amplitudes were tested between muted and control animals with a t-test.

Prey Capture Experiment

Each animal was tested in both a loud and a quiet tank with randomized order assignment,

spaced at least 2 days apart and tested at the same time of day on each day. Animals were not

fed on test days. Each animal was placed into a test tank and allowed to acclimate for 1 h prior

to test. For each test trial, 100 live mysids (consisting primarily of Mysidopsis bahia) were

introduced to test tanks. A SONY Handycam 8mm videocamera was used to film feeding

behavior behind a blind for a 10 min period immediately following mysid introduction.

Video footage was subsequently scored using JWatcher (Blumstein et al., 2007), according

to the partial ethogram in Table 5-1. Two measures were calculated for analysis:

% Successful Strikes (%SS) = 100∗+ MSS

SS (5-1)

123

%SS was calculated as a measure of feeding proficiency. The outcomes of Unknown

strikes (US) were unknown due to observational error and were not likely dependent upon

treatment; these strikes were thus excluded from the above calculation.

Total Successful Strikes (TSS) = ( )TimeIS

SSUSSS%

%+ (5-2)

TSS were calculated as a measure of feeding motivation. The construct US(%SS) was

calculated to estimate the number of unknown strikes that were successful in order to include

these in the total count of successful strikes. %TimeIS normalizes for variation in the amount of

time an animal was in sight during filming.

%SS was arcsine transformed according to Zar (1974). %SS and TSS were tested in a

repeated measures ANOVA using Minitab (v. 15) according to the following model:

y = A + T + S(A) + A*T + ε (5-3)

where y is the response variable, A is the animal treatment at two levels (muted or

control), T is the tank treatment at two levels (loud or quiet), S(A) is the animal subject, nested

within animal treatment, and ε is the error term. A and T are fixed factors, S(A) is a random

factor.

Results

Sound Analysis

Ambient noise analysis

Loud tanks demonstrated a peak amplitude of 120.3 + 1.0 dBpeak SPL (re: 1 µPa, mean

+ SE) at a peak frequency of 55.9 + 20.0 Hz, with a total RMS power of 125.7 + 0.9 dBpeak at the

middle of the water column. At tank bottom, loud tanks demonstrated a peak amplitude of 137.7

+ 2.2 dBpeak SPL (re: 1 µPa, mean + SE) at a peak frequency of 137.8 + 15.6 Hz, with a total

RMS power of 143.1 + 1.3 dBpeak. Quiet tanks demonstrated a peak amplitude of 114.3 + 0.7

124

dBpeak SPL (re: 1 µPa, mean + SE) at a peak frequency of 7.3 + 0.6 Hz, with a total RMS power

of 118.7 + 0.5 dBpeak at the middle of the water column. At tank bottom, quiet tanks

demonstrated a peak amplitude of 118.2 + 1.1 dBpeak SPL (re: 1 µPa, mean + SE) at a peak

frequency of 5.5 + 0.5 Hz, with a total RMS power of 124.5 + 0.6 dBpeak. Loud tanks had

significantly higher peak amplitudes than quiet tanks at the middle of the water column (p <

0.0001) and at the tank bottom (p < 0.0001). Loud tanks had significantly higher total RMS

power than quiet tanks at the middle of the water column (p < 0.0001) and at the tank bottom (p

< 0.0001, Figure 5-1).

Click analysis

Clicks of control animals averaged 137.8 + 2.8 dBpeak SPL (re: 1 µPa). Clicks of muted

animals averaged 125.6 + 2.4 dB. Control animals made significantly louder clicks than muted

animals (p = 0.004). As the deciBel (dB) scale is a logarithmic scale, a 6 dB reduction is

equivalent to a 50% reduction in sound pressure. Clicks of muted animals were thus, on average,

approximately 25% as loud as clicks of control animals.

Prey Capture

Control animals made 35 + 8 (Mean + SE) successful strikes with a 94 + 3% success rate

in loud tanks and 26 + 5 successful strikes with a 98 + 1% success rate in quiet tanks. Muted

animals made 35 + 4 successful strikes with a 96 + 1% success rate in loud tanks and 33 + 5

successful strikes with a 95 + 2% success rate in quiet tanks.

There were no significant differences between muted and control animals (F1,15 = 1.07, p =

0.316), loud and quiet tanks (F1,15 = 0.67, p = 0.424), nor among individuals (F15,15 = 1.39, p =

0.267) in the percentage of successful strikes. There was no significant interaction effect of

animal treatment and tank treatment (F1,15 = 0.00, p = 0.987).

125

There were no significant differences between muted and control animals (F1,15 = 0.33, p =

0.574) nor between loud and quiet tanks (F1,15 = 1.62, p = 0.223) in the number of successful

strikes made, though there were significant differences in the number of successful strikes made

among individual animals (F15,15 = 2.48, p = 0.044). There was no significant interaction effect

of animal treatment and tank treatment (F1,15 = 0.71, p = 0.411).

Discussion

Results of this study suggest that loud ambient noise in aquaculture settings does not

suppress feeding response. Effects of noise on growth and body condition are instead likely to

be mediated by physiological changes in energy balance required to maintain allostasis in a

stressful environment (Sterling and Eyer, 1988; Wendelaar-Bonga, 1997), or perhaps via changes

in expression of growth hormone (Pankhurst and Van Der Kraak, 1997).

This study has also submitted the prey-stunning hypothesis to a novel test with a novel

model. The muting surgery allowed comparison of prey capture dynamics with and without its

acoustic component (or with a markedly reduced acoustic component). The reduction of sound

pressure from the prey capture mechanism did not significantly reduce prey capture proficiency,

discounting the prey-stunning hypothesis at least in this species and for this particular prey item.

Because of this, the question of whether or not ambient noise interferes with the acoustic

stunning mechanism is not applicable for assessment in this model.

The prey stunning hypothesis should be further explored in seahorses with different prey

items. I chose live mysids in the context of this experiment because of their rapid escape

response and their widespread use among public aquarists maintaining syngnathids (pers. obs.).

This test should be repeated with live amphipods, that comprise the majority of the diet of wild

H. erectus (Teixeira and Musick, 2001).

126

For the seahorse, this evidence discounting an acoustic function of clicking on prey capture

suggests, by the process of elimination, other proposed functions. While clicking prominently

occurs during feeding, it also occurs during distress (Fish, 1953; Chapter 4), and in aggressive

encounters among males competing for females (Vincent, 1994b). Clicking has not been

examined in the courtship behavior of seahorses (Vincent and Sadler, 1995; Masonjones and

Lewis, 1996); but in the context of seahorses’ sparse populations (Foster and Vincent, 2004),

pair bonding (Vincent and Sadler, 1995), and small home ranges (Foster and Vincent, 2004);

clicking, even if occurring while feeding, may signal an animal’s presence and/or location to a

mate, or could advertise the presence of a food source to a mate, ultimately providing for more

fit offspring via increased nourishment of both parents. On balance, the function(s) of the

seahorse click is/are still unknown and merit(s) further study. This study discounted its function

in prey capture proficiency.

127

Table 5-1. Partial ethogram of the lined seahorse, Hippocampus erectus. Prey capture behavior Behavior Code Definition Successful strike SS Animal rapidly lifts head, depresses hyoid, opens mouth, followed by

visually confirmed suction of mysid into snout Unknown strike US Animal rapidly lifts head, depresses hyoid, opens mouth, but mysid

ingestion unknown. Miss M Animal rapidly lifts head, depresses hyoid, opens mouth, followed by

visually confirmed escape of mysid (suction of mysid into snout does not occur)

128

Figure 5-1. Power spectra of ambient noise in representative tanks. a = recordings from the middle of the water column, b = recordings from tank bottom. Black trace = loud tank, gray trace = quiet tank.

a b

129

CHAPTER 6 SOUND, SEX, AND SEAHORSES: ACOUSTIC ROLES AND EFFECTS IN COURTSHIP

BEHAVIOR OF LINED SEAHORSES (HIPPOCAMPUS ERECTUS) IN AQUARIA

Introduction

The acoustic sense of fishes is a sensory modality that can be overlooked in aquaculture.

Fishes in aquaria are exposed to ambient noise from many sources, such as water pumps, air

bubbles, air pumps, and chiller motors, that can create a loud, cacophonous sound in a tank (Bart

et al., 2001; Chapter 2). Constant exposure to loud noise may represent a chronic stressor, or

may mask acoustic communication signals among soniferous fishes. If either (or both)

mechanism(s) are at play, results may be deleterious for reproduction in captive fishes, which is

a critical component of a successful aquaculture operation.

Manifestations of Stress in Courtship Behavior

Effects of stress on measures of reproductive success at several levels of biological

organization are documented among fishes. Cortisol, the primary stress response hormone,

downregulates the production of gonadotropins, androgens, estrogens, and vitellogenin in fishes

(Carragher et al., 1989; Pankhurst and Van Der Kraak, 1997; Pickering et al., 1987).

Downstream effects of cortisol and/or chronic stress exposure include reduced oocyte diameter

and gonad size, and lower survival and quality of progeny (Campbell et al., 1994; McCormick,

1998; Pankhurst and Van Der Kraak, 1997). But studies examining impaired courtship behavior

as a tertiary stress response in fishes are rare. Morgan et al. (1999) tested the effects of

capture/confinement stress on courtship behavior in Atlantic cod (Gadus morhua). Stressed cod

continued to display courtship behavior, but they initiated fewer courtships and were more likely

to move directly to the final activity in the courtship sequence without performing prior courtship

behaviors that commonly occur in non-stressed fish.

130

Elsewhere in the animal kingdom, the effect of stress on courtship behavior is well-

founded. Male courtship behavior is reduced or altered among male rodents exposed to chronic

stress (D’Aquila et al., 1994; Retana-Marquez et al., 1996) and inescapable stress (Holmer et al.,

2003). Psychosocial stress in sheep (Ovis aries) inhibits the ability of ewes to attract rams and

the motivation of ewes to seek rams and initiate mating (Pierce et al., 2008). In a terrestrial

example of sound stress, low-level military jet over-flights reduced courtship duration in

harlequin ducks (Histrionicus histrionicus) for up to 1.5 h after exposure (Goudie and Jones,

2004). Among terrestrial invertebrates, thermal stress decreased courtship and mating behavior

in fruit flies (Drosophila spp.; Patton and Krebs, 2001; Fasolo and Krebs, 2004).

The hormones involved in the stress response and their direct effect on courtship behavior

have also been investigated. Administration of corticosterone inhibits clasping in roughskin

newts (Taricha granulosa), a very stereotypical courtship behavior that is easily triggered by

applying pressure to the cloaca (Moore et al., 1994; Rose and Moore, 1999; Rose, 2000).

Corticotropin-releasing factor (CRF), an upstream hormone that signals the release of

corticotropins (e.g., cortisol, corticosterone) in response to stress, suppresses solicitation (a

courtship behavior) in female white-crowned sparrows (Zonotrichia leucophrys; Maney and

Wingfield, 1998). Cortisol also inhibits calling in green treefrogs (Hyla cinerea; Burmeister et

al., 2001).

The effect of stressors or stress hormones do not always result in a simple decrease in

courtship behavior, but may elicit more complex alterations of the courtship and mating

behavioral sequence. In some cases, stress (or exogenous corticotropin induction) increases

early courtship behaviors in animals, such as mounting in rats (Retana-Marquez et al., 1996),

anogenital nosing of ewes by rams (Pierce et al., 2008), oral, tactile contact, and dart-shooting in

131

land snails (Arianta arbustorum; Locher and Baur, 2002), despite reductions in later courtship

behaviors or actual mating.

The Functions of Clicking in Courtship Behavior of the Lined Seahorse, Hippocampus erectus

The lined seahorse (Hippocampus erectus) makes a click sound, caused by a stridulation of

the posterior process of the supraoccipital against the coronet (Colson et al., 1998). The click

has been observed in many behavioral contexts, including feeding (Colson et al., 1998), distress

(Fish, 1953; Chapter 4), and in aggressive encounters among males competing for females

(Vincent, 1994b). Clicking is not commonly observed in the context of courtship behavior

(Vincent and Sadler, 1995; Masonjones and Lewis, 1996). But in the context of seahorses’

sparse populations (Foster and Vincent, 2004), pair bonding (Vincent and Sadler, 1995), and

small home ranges (Foster and Vincent, 2004), clicking, even if occurring while feeding, may

signal an animal’s presence and/or location to a mate, or could advertise the presence of a food

source to a mate, ultimately providing for more fit offspring via increased nourishment of both

parents. Whether intentional or incidental, clicking may communicate useful information to a

potential or current mate that may increase reproductive success.

The Effects of Tank Noise on Acoustic Communication

Examining the effects of loud noise exposure on courtship behavior is confounded in

species that employ acoustic communication in courtship behavior. In this scenario, sound may

not only act as a stressor, but it may also mask acoustic communication. Sound production is

exhibited by a variety of fishes (e.g., Fish et al., 1952), and occurs in a variety of behavioral

contexts, including courtship and reproduction. Fishes use sounds produced by potential mates

to discriminate among species (e.g., Delco, 1960; Gerald, 1971; Myrberg and Spires, 1972;

Spanier, 1979), and to choose high quality mates (e.g., Myrberg et al., 1986). In the context of

132

courtship, sound production also elicits mate phonotaxis (Tavolga, 1958; Delco, 1960; Ibara et

al., 1983; Myrberg et al., 1986) and interception of mating by competing males (Stout, 1975;

Kenyon, 1994). A loud environment in which acoustic signaling is masked may disrupt

appropriate courtship and reproductive behaviors, potentially resulting in reduced rates of

reproduction.

Masking increases the hearing threshold of fishes in an ensonified environment, requiring a

signal to be louder in order to be heard. Masking noise levels presented in Chapter 1 are within

the range of ambient noise encountered in aquaria and aquaculture environments (Chapter 2, Bart

et al., 2001). Furthermore, broadband noises cause a more pronounced masking effect than do

narrowband noises (Hawkins and Chapman, 1975), thus the broadband nature of ambient noise is

of additional concern.

Study Objectives

The objectives of this study are thus three-fold, (1) to examine the effect of chronic loud

noise exposure on courtship behavior in the lined seahorse, (2) to examine the effect of sound

production by mates on courtship behavior, and (3) to test the hypothesis that loud ambient noise

masks putative acoustic communication signals between mates, resulting in altered courtship

behaviors.


Animal Accession, Laboratory Design, and Husbandry Procedures

Animals were accessioned, tagged, quarantined, and maintained as described in Chapter 4,

except that animals received 12 hours of fluorescent light daily, with low-intensity incandescent

dawn and dusk lights illuminated ½ hour before and after fluorescent lighting transition. Water

quality was tested weekly; median results are as follows: Temperature, 26.1 ºC; Salinity, 29.8

ppt; pH, 8.2; NH3-N, 0.0 ppm; NO2-N, 0.0 ppm; NO3-N, 2.8 ppm.

133

Experimental Tank Design

Four 170 L aquaria were designed as experimental tanks. Tanks were painted on three

sides with light blue paint and on the bottom with sand color paint to standardize the visual

environment. Four plastic Vallisneria sp. plants were added to the holding tanks to provide

holdfasts/habitat.

Two experimental tanks were designed as loud tanks and two as quiet tanks, as described

in Chapter 4. Tank flow rates were measured at the introduction of each pair (monthly for each

tank) and averaged 436.3 + 4.2 liters per hour. Flow rates of loud tanks were adjusted to match

flow rates of quiet tanks at each measurement. For behavioral observations, an opaque white

curtain was hung 1m away from tank fronts to conceal observers from animals. Small square

holes (11 X 11 cm) were cut in the curtain in front of each tank to allow for behavior recording

with a videocamera.


The ambient noise profile of each test tank was measured at the introduction of each pair

(monthly for each tank). One minute recordings were taken from the middle of the water column

and also at tank bottom, at a sampling rate of 44.1 kbps, with an HTI-96-min hydrophone (High

Tech Instruments, Inc., sensitivity = -164.1 dB re: 1V/μPa) connected to a NOMAD Jukebox3

digital audio recorder. The NOMAD Jukebox3 was calibrated with a 1.0Vpeak sine wave.

Digital sound files (.wav) were post processed by removing putative artificial electrical

peaks in frequency spectra (at 60 Hz or its harmonics) using the FFT filter function in CoolEdit

(Syntrillium Software Corporation). These noise reduced files were then analyzed with

SpectraPlus (Pioneer Hill Software), calibrated according to manufacturers instructions.

SpectraPlus analysis settings are listed in Table 2-1. From the frequency domain spectrum, peak

frequency (in Hz), peak amplitude, and total RMS power (the latter two measurements in dBpeak

134

re: 1 µPa) were documented. Signals were low-pass filtered at 1,002 Hz; total RMS power

levels are thus summed within this frequency range, which is pertinent for hearing generalist

fishes (including H. erectus, Chapter 3) that have insensitive hearing above 1,000 Hz (Fay,

1988b). These measures were tested between loud and quiet tanks for both recording positions

with t-tests.

Muting

Eight males and eight females were surgically muted after Colson et al. (1998), as

described in Chapter 5. Another 8 males and 8 females were subjected to the control surgical

procedure.

Courtship Experiment

Experimental design & observational methods

Animals were paired as size-matched male-female muted pairs and male-female control

pairs. The courtship behavior of both types of pairs were observed in four trials in quiet tanks

and in four trials in loud tanks. Each animal was used in only one trial.

Animals were placed in assigned tanks on the morning of the first day of observation

before lights turned on. Pairs were observed and videotaped with a SONY 8mm Handycam for

the first hour of light on each day for five days, beginning with the day of introduction. This

time period was chosen because seahorses court most actively at dawn (Vincent, 1994b; Vincent

and Sadler, 1995; Masonjones and Lewis, 1996). Observations ceased prior to day 5 if animals

successfully mated (as evidenced by a full male brood pouch) or if a female dropped eggs (as

evidenced by the presence of eggs at the bottom of the tank). Tanks were siphoned clean and

feed introduced before lights on with the aid of red LED headlamps, to minimize disturbances at

first light.

135

To capture behaviors of both animals in sufficient detail, the following rules were applied

during videotaping (refer to Table 6-1 for terms). Both animals were judged as together and

filmed within the frame when they were < 15 cm of each other and when at least one animal

exhibited interactive behavior (usually brightening). When an animal moved >15 cm away from

the other animal, then the videographer zoomed in on only one animal. The choice of animal

(male or female) to film when animals were apart alternated between bouts of together. These

videotaping rules resulted in complete coverage of time spent together during the observation

period, and partial and variable coverage of time spent apart for each sex during the observation

period.

Ethological analysis

An ethogram, modified from Masonjones and Lewis (1996), Vincent (1994b), and Vincent

and Sadler (1995), was constructed from observed behaviors and programmed into the JWatcher

program (Blumstein et al., 2000) to subsequently score behaviors quantitatively from videotapes.

Of the series of behaviors described in these works and observed by me, I sampled representative

behaviors (Table 6-1) for statistical analysis from both sexes, from across the stages of the

courtship sequence, and that occurred with sufficient frequency to permit statistical tests among

treatments. While clicking is not described as a component of courtship behavior in the

published ethograms referenced, I chose to analyze clicking behavior when not associated with

feeding to further examine the hypothesis that animals may click to present an acoustic signal to

a mate.

Behavioral states were measured as total time (in minutes) over the observation period.

Behavioral events were measured as counts. All behavioral states and most behavioral events

tended to occur when animals were together; so to exclude variability in these measures due to

variation in observational coverage of single animals, occurrence of these behaviors when

136

animals were apart were excluded from analysis. Pouch pumping occurred primarily while

males were apart. To normalize for variation in videotape coverage of single animals while apart

across trials, this behavior was only counted when animals were apart, converted to a frequency

measure (events per minute), and multiplied by the total time animals were apart during the

observation period. Clicking also occurred primarily while animals were apart, but in order to

address questions of the role of clicking in courtship behavior, clicking was counted separately

for animals while together and for animals while apart; the same conversion as for pouch

pumping applied to clicking that occurred while animals were apart.

To compare courtship behavior among treatments and interactions, behavioral states were

log transformed; behavioral events were square-root transformed (per Zar, 1974). Each behavior

was tested in a repeated measures model with three factors using PROC GLIMMIX in SAS (The

SAS Institute, 2006). Fixed factors included animal treatment (muted or control), tank treatment

(quiet or loud), both class variables, and day (1-5), a quantitative variable. Animals were

modeled as random subjects for both G and R-side effects, and the first order auto-regressive

structure was chosen to model co-variance structure of R-side effects. Because data

transformations approximate normal distributions, all data distributions were modeled as

Gaussian distributions. The Kenward-Roger method was chosen to calculate denominator

degrees of freedom. If results demonstrated significant interaction effects with time, two

methods were followed for further analysis and partitioning of effects: 1) Least-square mean

differences were computed between significant treatments for each day and p-values compared

against a Bonferroni-adjusted alpha value of α = 0.05/5 = 0.01 for 5 comparisons (days 1-5), and

2) Time was modeled to examine linear (D), quadratic (D*D), cubic (D*D*D), or quartic

(D*D*D*D) patterns.

137

I also examined heterogeneity of variance in behavorial data. To do this, I obtained the

absolute value of the difference of each individual measurement from the mean of four animals

per treatment combination per day, square-root transformed these individual deviations, and

tested them in the repeated measures model described above, following the same procedures for

model parameterization and methods for further examination of interaction effects.

Finally, to test for differences in occurrence of clicking while animals were together vs.

apart, counts were averaged over days and paired t-tests run between clicks that occurred while

together and clicks that occurred while apart for both males and females, regardless of animal or

tank treatment.

Results

Sound Analysis

Loud tanks demonstrated a peak amplitude of 115.8 + 1.4 dBpeak SPL (re: 1 µPa, mean +

SE) at a peak frequency of 61.5 + 7.1 Hz, with a total RMS power of 123.6 + 1.0 dBpeak at the

middle of the water column. At tank bottom, loud tanks demonstrated a peak amplitude of 132.8

+ 0.9 dBpeak at a peak frequency of 98.5 + 28.6 Hz, with a total RMS power of 137.6 + 0.9

dBpeak. Quiet tanks demonstrated a peak amplitude of 108.8 + 0.8 dBpeak at a peak frequency of

7.1 + 1.2 Hz, with a total RMS power of 114.7 + 0.6 dBpeak at the middle of the water column.

At tank bottom, quiet tanks demonstrated a peak amplitude of 115.9 + 1.8 dBpeak at a peak

frequency of 14.6 + 6.3 Hz, with a total RMS power of 122.9 + 1.3 dBpeak. Loud tanks had

significantly higher peak amplitudes than quiet tanks at the middle of the water column (p <

0.0001) and at the tank bottom (p < 0.0001). Loud tanks had significantly higher total RMS

power than quiet tanks at the middle of the water column (p < 0.0001) and at the tank bottom (p

< 0.0001, Figure 6-1).

138

Ethological Analysis

Tests of means

Thirteen behaviors described in Table 6-1 were chosen for analysis. To maintain focus on

the questions posed in this study, only results of behaviors demonstrating significant effects of

animal treatment, tank treatment, or their interactions are presented in detail here. Results of

clicking are described in full detail as its role in courtship behavior was a central question of this

study. Most behaviors demonstrated either no significant effects of any fixed factor (male

display, male point, female brighten, female display, female bow, female click) or demonstrated a

significant effect of day only (time spent together, male pouch pump, male click, female

approach). However, as elaborated below, the behavioral event of male approach demonstrated

a significant effect of day (F1,25.16 = 4.39, p = 0.046) as well as an interaction effect between

animal treatment and day (F1,25.16 = 7.74, p = 0.010). Female point also demonstrated a

significant interaction effect between animal treatment and day (F1,42.2 = 5.64, p = 0.022).

Significant results of Type III tests of means of fixed effects for selected behaviors are

summarized in Table 6-2.

In days 1-3 of the courtship sequence, males approach females with statistically similar

frequencies (with mean number of approaches ranging from 1.6 + 0.7 to 4.6 + 2.5, Figure 6-2).

By days 4 and 5, control males continue to approach females with elevated occurrence (with

mean number of approaches at 4.3 + 1.1 and 6.3 + 1.2, respectively) while approach behavior

declines significantly (p < 0.003) among muted males (with mean number of approaches at 2.1 +

0.5 and 0.8 + 0.4, respectively).

In days 1-3 of the courtship sequence, pointing occurs rarely among females, regardless of

treatment (with mean number of points ranging from 0.3 + 0.3 to 0.8 + 0.6, Figure 6-3). In days

4 and 5, females of control pairs increase pointing behavior dramatically (with mean number of

139

points at 4.7 + 4.7 and 6.7 + 3.7, respectively) while females of muted pairs cease pointing

entirely (p < 0.005).

As there were no significant effects of animal treatment, tank treatment, or their interaction

on clicking, data from these groups were pooled for presentation (Figure 6-4). Male clicking

demonstrated a significant effect of day (F1,36.27 = 11.11, p = 0.002), starting from 0.71 + 0.28

clicks produced in the first day and escalating to 5.16 + 1.48 clicks by day 5. In general, female

clicks were produced in varying frequency over the 5 day period, ranging from 1.11 + 0.41 to

6.96 + 3.76 clicks. There was no significant effect of day among females (F1,26.08 = 1.61, p =

0.216).

Tests of mean deviations

While tests of means showed few effects of animal treatment and no effects of tank

treatment, variance in many courtship behaviors differed significantly among tank treatments.

There were also several significant interaction effects. As heterogeneity of variance was

examined as a stress response measure due to tank treatment, only significant main effects of

tank treatment and significant interaction effects involving tank treatment are described below.

Significant type III results of mean deviations of these effects and interactions are summarized in

Table 6-3.

Behavioral states: There were significant differences in time trends of mean deviations of

time spent together by pairs in quiet tanks and loud tanks over time (F1,25.46 = 5.07, p = 0.033),

but least square means comparisons did not reveal significant differences of variance at the

Bonferroni-corrected alpha value. Rather, the pattern of change across time differed between

treatments; variation in time spent together followed no significant time trend among animals in

loud tanks (there were no significant linear, quadratic, cubic, or quartic effects of day), but

followed a quartic trend among animals in quiet tanks (F1,11.91 = 12.41, p = 0.004).

140

There was significant heterogeneity of variance in male brightening and male display

between animals in loud tanks and in quiet tanks in the latter days of observation, demonstrated

by a significant interaction effect of day and tank treatment (male brightening, F1,19.53 = 10.61, p

= 0.004, Figure 6-5; male display, F1,15.75 = 10.78, p = 0.005, Figure 6-6). Least square means

comparisons revealed that males brightened and displayed in loud tanks for more variable

durations than in quiet tanks (p < 0.007).

There was no significant heterogeneity of variance due to treatments or their interactions in

female brightening, but female display behavior showed significant heterogeneity of variance in

the interaction of animal treatment and tank treatment (F1,31.47 = 9.38, p = 0.005); day and tank

treatment (F1,22.81 = 6.90, p = 0.015); and a 3-way interaction in day, animal, and tank treatment

(F1,22.81 = 37.22, p < 0.001, Figure 6-7). With regard to tank treatment, females in quiet tanks

displayed for significantly more variable durations than females in loud tanks on days 3, 4, and 5

(p < 0.002). With regard to the interaction of animal treatment and tank treatment, animal

treatment type was only heterogeneous in variance in quiet tanks (F1,12.24 = 9.90, p = 0.008) and

tank treatment type was only heterogeneous in variance among muted pairs (F1,11.51 = 19.31, p =

0.0001); these are due to a large standard deviation in this behavior among muted pairs in quiet

tanks (at SD = 1.89), compared to a smaller range of standard deviations in other treatment

groups (0.28 to 0.61).

Behavioral events: There was significant heterogeneity of variance in pointing and pouch

pumping between males in loud tanks and in quiet tanks. Males in quiet tanks pointed more

variably in quiet tanks than in loud tanks overall (F1,37.91 = 12.33, p = 0.001, Figure 6-8). For

both behaviors, there was a significant interaction effect of day and tank treatment (pointing,

F1,42.59 = 8.72, p = 0.005; pumping, F1,46.25 = 6.05, p = 0.018, Figure 6-9). For pointing, least

141

square means comparisons revealed significant heterogeneity of variance in days 1 and 2 (p <

0.003). For pumping, least square means comparisons significant heterogeneity of variance that

differed in magnitude between quiet and loud tanks in a variable pattern over days 2-5 (p <

0.001).

Female pointing displayed significantly greater variance in quiet tanks (SD=5.56) than

loud tanks (SD=4.56) overall (F1,43.31 = 7.77, p = 0.008, Figure 6-10).

Female clicking displayed no heterogeneity of variance among any treatments or

combinations, but male clicking displayed heterogeneity of variance over time (F1,31.07 = 41.17,

p < 0.001, Figure 6-11). There was also an interaction of tank treatment and day (F1,31.07 = 4.80,

p = 0.036); however, least square means comparisons revealed no significant differences of

variance between tank treatments for any day when animal treatments are pooled. Rather,

variance in male clicking followed different time trends between animals in loud and quiet tanks;

there was no significant time trend among animals in loud tanks, but variance progressed

according to a quartic model among animals in quiet tanks (F1,14.08 = 6.60, p = 0.022). There was

also a significant 3-way interaction (F1,31.07 = 5.43, p = 0.027). When the quartic time trend

among animals in loud tanks was parsed further for analysis in the 3-way interaction, only the

behavior of muted males demonstrated this time trend in loud tanks (F1,8.51 = 64.04, p < 0.001).

Further parsing of the three-way interaction revealed least square means differences of variance

at days 3,4, and 5 between control males in loud and quiet tanks vs. muted males in quiet tanks

(p < 0.011), and at days 4 and 5 between muted males in loud tanks vs. muted males in quiet

tanks (p < 0.002).

Occurrence of clicking

Males clicked an average of 0.7 (+ 0.15) times while together and 2.6 (+ 0.6) times while

apart. These differences were significant (p = 0.007). Females clicked an average of 0.6 (+ 0.2)

142

times while together and 3.8 (+ 1.2) times while apart. These differences were also significant

(p = 0.007).

Discussion

Comprehensively, these results reveal that loud ambient tank noise does not impair

courtship behavior in H. erectus via stress response. Clicking is an acoustic communication

signal employed at least by males (and possibly by both sexes) in courtship behavior, and the

signal elicits courtship responses by mates. However, loud ambient tank noise also apparently

does not mask acoustic communication signals.

The Effect of Tank Noise on Courtship Behavior

Loud tanks had no apparent effects on courtship behavior means by either sex. This is not

entirely unexpected. Orlando and Guillette’s (2001) review of population responses to

environmental stressors underscores the importance of comparing variance in response between

stressed and control populations. In Chapter 4, significant behavioral results among stressed fish

are dominated by changes in the variance of the behavior as opposed to the mean of the behavior.

The hypothesis examined here is the potential for chronic loud noise exposure to generate a

stress response that is manifested in altered courtship behavior. It was sensible, then, to examine

heterogeneity of variance in courtship behaviors between animals in loud and quiet tanks.

The predictions, then, are that animals in loud tanks should exhibit greater variance in

courtship behaviors than animals in quiet tanks. Heterogeneity of variance between animals in

loud and quiet tanks was demonstrated, primarily as an interaction with day in 8 of 13 behaviors

measured (but two behaviors, male and female point, demonstrated a significant main effect of

tank treatment). However, in only two of these measures were variances greater in animals of

loud tanks (male brightening and male display); other measures either showed greater variance

among animals in quiet tanks or variable variance between treatments across days.

143

Results suggest that heterogeneity of variance may arise from the distribution of data.

Among these data, group variances are likely to be proportional to means. This was suggested

by histograms created from pooled data for each measure when these data were first explored;

for almost all measures, pooled data demonstrated histograms with a high degree of kurtosis at

the low end of the range and a large positive skew. This is intuitive when thinking about animal

behavior; animals that tend not to participate actively in courtship behavior produce behavioral

data with many zeros and perhaps some low-values, all within a narrow distribution. Animals

that participate actively in courtship behavior, on the other hand, may vary widely with respect to

frequency of behaviors (and/or time spent in behavioral states).

The lack of apparent patterns in heterogeneity of variance across behavioral measures for

tank treatment preclude sensible rationalization of these results. Based on the lack of patterning,

it is not prudent to conclude from these data that loud tanks affect courtship behavior by means

of a stress response.

The Functions of Clicking in Courtship Behavior

One of the observations that was readily apparent to observers without the aid of statistics

was the more vigorous courtship behavior exhibited by males than by females, as is consistent

with other seahorse species (e.g., Masonjones and Lewis, 1996). In general, courtship behavior

by males was robust regardless of treatment combination, while courtship behavior by females

was comparatively rare. This general observation is reflected in the number of male courtship

behaviors (four) that increased over the five day period (as evidenced by a significant effect of

day) vs. the number of female courtship behaviors (one) that increased over time. These time

trends are to be expected of the progressive courtship sequence as described by Masonjones and

Lewis (1996).

144

For most of the courtship behaviors examined, the effect of day was to be expected but

otherwise not pertinent to this study in the absence of any interactions. However, the effect of

day in male clicking behavior is noteworthy. The increased occurrence of clicking over time co-

occurs with increased occurrence of other documented male courtship behaviors over time and

therefore suggests that clicking may be involved in the courtship sequence. Clicking followed an

upward trend over time with male seahorses, but not with female seahorses. This may suggest

that if clicking is performed to communicate acoustic signals to a mate, that male to female

signaling may be more prominent in the courtship sequence than female to male signaling.

In this study, clicks were chosen for analysis when not associated with feeding. This

stipulation was chosen to examine clicks that may be produced intentionally by a mate for the

purposes of communication, as opposed to clicks that may be produced unintentionally during

feeding behavior (Colson et al., 1998).

The relative paucity of click behavior when animals are together are corroborated in gobies

(Tavolga, 1958); in Tavolga’s study system, males called frequently when females were outside

of the nest but calling ceased once females entered; other courtship and mating activities ensued.

This may also be the case with seahorses; acoustic communication may signal readiness to court

when animals are apart and not within visual range, but visual signaling prevails when animals

are close.

Because male clicking increases over time during the courtship behavior but while animals

are apart presents some plausible hypotheses regarding the function of the click. In this scenario,

it may serve to advertise presence and location to a mate, as well as readiness to mate. Courting

males elicit phonotaxis of females by acoustic signaling in damselfish (Myrberg et al., 1986),

gobies (Tavolga, 1958), midshipmanfish (Ibara et al., 1983), sunfish (Gerald, 1971), and toadfish

145

(Bass, 1990). The acoustic nature of the click may also signal mate quality, as suggested by

Colson et al. (1998) in seahorses and as demonstrated by Myrberg et al. (1986) for damselfishes.

These results, coupled with the results of Chapter 3 demonstrating a frequency match

between the peak amplitude of the click with the most sensitive hearing range of H. erectus,

strongly suggest that the click is used in acoustic communication.

So, the trend of increasing click behavior among males over the course of the courtship

sequence suggests that males may be signaling acoustically during courtship. But are females

responding? This study was designed to address this question as well, and one result in

particular suggests that they are. Given the rare performance of courtship behavior by females in

general (as discussed earlier), the significant interaction effect of muting over time in the female

courtship behavior of pointing is quite important. Pointing is a behavior that occurs later in the

courtship sequence, usually on the day of mating (Masonjones and Lewis, 1996). As such, an

increase in female pointing behavior during the latter days of observation is expected; this was

indeed the case among control animals (as evidenced by Figure 6-3). However, among muted

pairs, pointing ceased entirely in females during the latter days. Note that the latter days also

correspond with increased clicking behavior among males. The removal of the male’s acoustic

signal in the courtship repertoire by muting may thus prohibit reciprocal courtship behaviors in

females.

Because female clicks exhibited no time trends, the argument for female signaling and its

importance in the courtship repertoire is less strong, though the hypothesis for acoustic signaling

while apart fits clicking patterns in females as well. It is noteworthy that males did not increase

approach behavior over time in muted pairs (Figure 6-2), whereas this behavior progressed over

time in control pairs. The courtship sequence is a complicated reciprocal signaling system that is

146

thought to synchronize reproductive states in preparation for copulation (Vincent and Sadler,

1995). A failure of muted males to escalate number of approaches as time progresses might

suggest a downregulation of response due to lack of acoustic signals received by muted females,

or it may reflect a lack of visual signals received by females over time, perhaps because females

are not receiving acoustic signals by males. Further courtship studies in which muted and non-

muted partners are paired may help to further elucidate mechanisms behind these results.

The Effects of Tank Noise on Acoustic Communication

Though these results strongly suggest acoustic communication as a functional component

of courtship behavior in Hippocampus erectus, there appears to be no masking effect of tank

noise on signal reception. If masking occurred, animal treatment by tank treatment interactions

would be expected. Specifically, the masking hypothesis leads to the prediction that courtship

behaviors would be reduced among muted animals (in loud or quiet tanks) AND control animals

in loud tanks, but not among control animals in quiet tanks. Only one significant animal

treatment by tank treatment interaction was detected in the tests of mean deviations of female

display behavior, but this effect was triggered by an unusually large standard deviation among

displays by muted pairs in quiet tanks, whereas other treatment combinations demonstrated

comparably low standard deviations. This pattern does not match the prediction.

The lack of a masking effect is surprising in light of the masking phenomenon among

hearing generalists and in light of a comparison of sound pressure levels of loud tanks to sound

pressure levels of clicking. At around 200-300 Hz, which brackets the best hearing sensitivity of

H. erectus (Chapter 3), loud tanks in this study demonstrated mean spectrum level amplitudes of

90-120 dBpeak SPL (re: 1 µPa) at tank bottom. Results from Chapter 3 demonstrate that seahorse

clicks reach peak spectrum level amplitudes of 94.3 + 0.9 dBpeak SPL (re: 1 µPa). In Chapter 1, I

present a range of critical ratios that must be met in order to preclude masking among hearing

147

generalist fishes. As discussed, signals need to be from 10-60 dB (median around 20 dB) louder

than background noise to be heard. In this experimental case, the spectral amplitude of clicking

does not clear the minimum critical ratio of this range.

However, as discussed in Chapter 3, the click has several additional acoustical properties,

including a sudden onset, and its broadband nature, that boost likelihood of detection. It is also

possible that seahorses are signaling while located higher in the water column, away from the

louder noise profile at the tank bottom. Finally, as discussed in Chapter 2, seahorses may be

responding to the particle acceleration component of the click as opposed to the sound pressure

component. Unfortunately, lack of appropriate instrumentation precluded the opportunity to

measure the particle acceleration of ambient noise in test tanks; but the signal-to-noise ratio may be

quite different in this acoustic modality. In general, scientific investigation of fish hearing

response to the particle acceleration modality of sound underwater is very sparse, presumably due

to the lack of appropriate equipment for measuring this modality. This frustration has been

expressed by others as well (Fay and Simmons, 1999; Casper et al., 2003). Future examination of

this modality, for the questions proposed in this Dissertation, and elsewhere in the realm of fish

bioacoustics, are likely to yield new insights to the sense of hearing in fishes.

148

Table 6-1. Partial ethogram of courtship behaviors of the lined seahorse, Hippocampus erectus. Behavior Description States Together

Animals are within 15 cm of one another and at least one animal is engaged in interactive behavior (e.g., brightening).

Brighten

The body of an animal blanches in coloration, with the exception of the head and characteristic dark dorsal and ventral midlines

Display

An animal tucks its head downward and in, slightly off to one side, while brightened.

Events Approach One animal approaches to within 15 cm of another animal Point An animal raises its head toward the water surface at an approximately 135º

angle with respect to the body axis and then lowers it to its starting position

Pouch Pump

Male opens brood pouch and jackknifes its tail forward rapidly, followed by a return of the tail to its starting position.

Click

An animal rapidly raises head, opens mouth, and depresses hyoid, with a rapid return to starting position. This behavior is often accompanied with a click sound. In this study, only clicking not associated with a food item was quantified.

Bow An animal angles its body downward, orienting its lateral body surface parallel to the tank floor, often with the head pointed in the direction of a courting mate.

149

Table 6-2. Summary of significant results of Type III tests of means of fixed effects in SAS PROC GLIMMIX. All F values have 1 degree of freedom in the numerator. den df = denominator degrees of freedom.

Day (D) Animal Treatment * Day (A*D) Interaction den df F p Trend den df F p Trend States Together 24.59 8.75 0.007 Increase over time Male Brighten 14.49 10.76 0.005 Increase over time Events Male Approach 25.16 4.39 0.046 Increase over time 25.16 7.74 0.010 Control > Muted @ Days 4,5 Male Pouch Pump 46.03 7.08 0.011 Increase over time Male Click 42.39 11.11 0.002 Increase over time Female Approach 22.64 14.17 0.001 Increase over time Female Point 42.2 5.64 0.022 Control > Muted @ Days 4,5

150

Table 6-3. Summary of significant results of Type III tests of mean deviations of tank treatment effects or its interactions in SAS PROC GLIMMIX. All F values have 1 degree of freedom in the numerator. A = animal treatment, T = tank treatment, D = day, den df = denominator degrees of freedom, NS = not significant

Effect den df F p Trend States Together T*D 25.46 5.07 0.033 Loud tanks: NS time trend, quiet tanks: Quartic time trend Male Brighten T*D 19.53 10.61 0.004 > variance among males in loud tanks @ days 4, 5 Male Display T*D 15.75 10.78 0.005 > variance among males in loud tanks @ days 4, 5 Female Display A*T

T*D A*T*D

31.4722.8122.81

9.38 6.90

37.22

0.0050.015

<0.001

> variance among muted females in quiet tanks only > variance among females in quiet tanks @ days 3, 4, 5 Large variance among muted females in quiet tanks relative to other groups

Events Male Point T

T*D 37.9142.59

12.33 8.72

0.0010.005

> variance among males in quiet tanks > variance among males in quiet tanks @ days 1, 2

Male Pouch Pump T*D 46.25 15.06 <0.001 Variable trends over days 2-5 Male Click T*D

A*T*D 31.0731.07

4.80 5.43

0.0360.027

Loud tanks: NS time trend, quiet tanks: Quartic time trend Quartic time trend demonstrated only for muted males in loud tanks > variance among muted males in quiet tanks vs. control males in all tanks @ days 3, 4, 5 > variance among muted males in quiet tanks vs. muted males in loud tanks @ days 4, 5

Female Point T 43.31 7.77 0.008 > variance among females in quiet tanks

151

Figure 6-1. Power spectra of ambient noise in representative tanks. a = recordings from the middle of the water column, b = recordings from tank bottom. Gray trace = quiet tank, black trace = loud tank.

a b

152

0

1

2

3

4

5

6

7

8

1 2 3 4 5

Day

Cou

nt*

*

Figure 6-2. Comparisons of means (+ SE) of male approach. Data pooled for tank treatment. Gray = control pairs, black = muted pairs. * Significant least-squares means differences, p < 0.003.

0

2

4

6

8

10

12

1 2 3 4 5

Day

Cou

nt

* *

Figure 6-3. Comparisons of means (+ SE) of female point. Data pooled for tank treatment. Gray = control females, black = muted females. * Significant least-squares means differences, p < 0.0005.

153

0

2

4

6

8

10

12

1 2 3 4 5

Day

Cou

nt

Figure 6-4. Comparisons of means (+ SE) of clicking between males and females. Data pooled for tank and animal treatment. Gray = females, black = males.

0

2

4

6

8

10

12

14

16

18

1 2 3 4 5

Day

SD T

ime

(min

)

* *

Figure 6-5. Comparisons of standard deviations of brightening between males in quiet and loud tanks. Data pooled for animal treatment. Gray = quiet tanks, black = loud tanks. * Significant least-squares means differences, p < 0.005.

154

0

2

4

6

8

10

12

14

16

18

1 2 3 4 5

Day

SD T

ime

(min

)* *

Figure 6-6. Comparisons of standard deviations of display between males in quiet and loud tanks. Data pooled for animal treatment. Gray = quiet tanks, black = loud tanks. * Significant least-squares means differences, p < 0.007.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1 2 3 4 5

Day

SD T

ime

(min

) *

*

*

Figure 6-7. Comparisons of standard deviations of display among females. Open circles = control females, closed circles = muted females, gray lines = quiet tanks, black lines = loud tanks. * Significant least-squares mean differences between tank treatments (p < 0.002).

155

0

1

2

3

4

5

6

7

8

9

10

1 2 3 4 5

Day

SD C

ount

**

Figure 6-8. Comparisons of standard deviations of pointing between males in loud and quiet tanks. Data pooled for animal treatment. Gray = quiet tanks, black = loud tanks. * Significant least-squares means differences, p < 0.003.

0

5

10

15

20

25

30

35

40

1 2 3 4 5

Day

SD C

ount

**

*

*

Figure 6-9. Comparisons of standard deviations of pouch pumping between males in loud and quiet tanks. Data pooled for animal treatment. Gray = quiet tanks, black = loud tanks. * Significant least-squares means differences, p < 0.001.

156

0

5

10

15

20

25

1 2 3 4 5

Day

SD C

ount

Figure 6-10. Comparisons of standard deviations of pointing among females. Open circles = control females, closed circles = muted females, gray lines = quiet tanks, black lines = loud tanks.

0

2

4

6

8

10

12

14

16

1 2 3 4 5

Day

SD C

ount

*

*

Figure 6-11. Comparisons of standard deviations of clicking among males. Open circles = control males, closed circles = muted males, gray lines = quiet tanks, black lines = loud tanks. * Significant least-squares means differences between muted males in loud tanks vs. muted males in quiet tanks, p < 0.002.

157

CHAPTER 7 DISCUSSION

The Hearing Ability of Hippocampus erectus in the Context of Wild and Captive Acoustic Environments

The lined seahorse is best described as a hearing generalist, as discussed in Chapter 3. As

such, it has relatively insensitive hearing abilities (especially in comparison to hearing

specialists). Sounds must therefore be relatively loud in order to be audible to the seahorse.

Figure 7-1 overlays the estimated broadbound sound pressure audiogram with the range of

ambient noise encountered at the bottom of seahorse aquaria, where seahorses spend most of

their lives in contact with the substrate or a holdfast resting on the substrate. Within the

seahorse’s best hearing range (below 400 Hz), it is evident that ambient noise profiles among

many aquaria are audible to the seahorse. Compare this to Figure 7-2, which overlays the

estimated broadband sound pressure audiogram with the range of noise encountered in the wild.

Only the loudest sites where wild seahorses are collected are loud enough to be audible, and

within a narrower frequency range of 100 to 300 Hz.

An environment with audible ambient noise is not necessarily deleterious. To the contrary,

Popper and Fay (1999) suggest that hearing in fishes evolved to evaluate the “auditory scene.” It

may be beneficial for fishes to take advantage of the acoustic sense to monitor the scene for

sounds that might signal approaching predators, prey, or mates, for example. But some fishes,

both hearing generalists and hearing specialists, live in natural environments with inaudible

ambient noise (Amoser and Ladich, 2005).

The audible acoustic environment of aquaria to seahorses presents possible complications

of masking of acoustic signals, or, alternatively, a stress response to chronic noise exposure.

158

Pressure vs. Displacement

Sound energy is carried in two physical modalities, as pressure waves and as particle

motion. These components of sound contribute in different ways in the near-field and far-field

of sound sources. The near-field of a sound source is dominated by local hydrodynamic flow,

established by the displacement of water molecules (particle motion) adjacent to a sphere

vibrating in place (as in the case of a monopole sound source) or a sphere vibrating along an axis

(as in the case of a dipole sound source, Bass and Clark, 2003). For a monopole sound source,

the near-field predominates to a distance of:

λ/2π (7-1)

from the sound source, where λ = wavelength (m) and π (pi) ≈ 3.14. So this distance is

frequency dependent, but can be large underwater owing to the speed of sound in water, as:

λ = c/f (7-2)

where λ = wavelength (m), c = speed of sound (~340 m/s in air, ~1,500 m/s in water,

Dusenberry, 1992), and f = frequency (Hz). For example, a 100-Hz soundwave has a near-field

that propagates 2.4 m away from the sound source underwater and only 0.5 m in air. Beyond

this distance, the far-field becomes dominant (by λ/π) and is characterized by propagating waves

that can generate a pressure wave in addition to particle motion.

The inner ears of fishes are particle motion detectors (Popper et al., 2003). They do not

respond directly to sound pressure. The swimbladder acts as a monopole resonator that converts

sound pressure to particle motion for detection by the inner ear. In effect, the swimbladder

renders some fishes sensitive to sound pressure; it also acts as an amplifier of sound (Popper et

al., 2003). Some fishes have taken evolutionary advantage of this physical property of the

swimbladder and have evolved bony or gaseous vesicular connections between the swimbladder

and the inner ear to boost sound detection. It is generally well-accepted that hearing specialist

159

fishes respond to both particle motion and pressure, but are more sensitive to pressure

particularly in the far-field and at frequencies above 70 Hz (Fay et al., 1982). Hearing generalist

fishes, that have no specialized connections between the swimbladder and inner ear, have yielded

equivocal data concerning the relative importance of pressure sensitivity to sound detection and

processing (Cahn et al., 1968; Sand and Enger, 1973; Chapman and Johnstone, 1974; Fay and

Popper, 1975; Jerkø et al., 1989; Lovell et al., 2005). The same problem is true for sharks, that

have no swimbladders at all (van den Berg and Schuijf, 1983; Myrberg, 2001). The consensus

that might be drawn from this literature is that both acoustic modalities may be detected and

processed by hearing generalist fishes, though the relative contributions of each may vary with

respect to distance, frequency, and amplitude.

Nonetheless, the vast majority of the primary literature concerning hearing in fishes

describes, measures, and tests sound in terms of sound pressure, even for hearing generalist

fishes, that ought to be more sensitive to particle acceleration. This is presumably due to the lack

of specialized equipment available to measure particle acceleration underwater; this frustration

has been expressed by others (Fay and Simmons, 1999; Casper et al., 2003). I was fortunate to

have access to the Acoustech Geophone for measurement of particle acceleration in Chapter 3.

However, this unit was not available for use in any of the other studies presented in this

Dissertation. The Acoustech Geophone was an expensive instrument retailing for approximately

$3,000 in 2006, as opposed to the pressure-sensitive hydrophone used in this study that retailed

for approximately $300 (D. Mann, pers. comm.; pers. obs.). The Acoustech Geophone unit has

since malfunctioned and the company has gone out of business; another example of the difficulty

in obtaining appropriate equipment for these measurements.

160

The caveat to characterizing sound in terms of sound pressure only as it relates to hearing

in generalists is that authors run the risk of describing a modality that animals are either not

detecting and processing, or that pressure contributes relatively little information as opposed to

particle acceleration, which may be the main modality of acoustic reception in generalists.

Furthermore, the morphology of the sound waves of particle acceleration and particle motion do

not necessarily correlate, especially in tank environments. Due to the long wavelengths of

underwater sounds at low frequencies, a progressive sound wave is extremely difficult or

impossible to produce in small tanks (Popper et al., 2003). Thus, the relationships between

sound pressure and particle acceleration cannot be calculated with certainty. Sound pressure

thresholds measured in the lab among hearing generalist fishes may therefore be inadequate and

misleading to describe or predict the animal’s sensitivity to sound sources in its natural

environment.

The Function of Clicking

Previous literature has reported the click to occur in several behavioral contexts, including

feeding (e.g., Colson et al., 1998) and competition for mates (Vincent et al., 1994b). Its role in

courtship was previously unclear; Woods (2000) suggested a role in his observations of courtship

behavior in Hippocampus abdominalis, but clicking was not included among courtship behaviors

described in H. fuscus (Vincent, 1994b), H. whitei (Vincent and Sadler, 1995), or H. zosterae

(Masonjones and Lewis, 1996).

This dissertation elucidated the role of clicking in several behavioral contexts. It occurred

more variably among animals in loud tanks in Chapter 4. Coupled with physiological evidence

of a chronic stress response, this suggests that the click may be a distress behavior; these results

are corroborated by early observations made by Fish (1953), who noted clicking in H. erectus in

response to transfers to new containers of water with different water quality parameters (perhaps

161

representing an acute stressor). Whether or not the distress click is of communicative value to

conspecifics is unclear from the results of the Dissertation, but it is possible (see Chapter 4 for a

more complete discussion of this hypothesis).

Because clicking occurs so often and predictably in the context of feeding, it was prudent

to evaluate any acoustic role of clicking in prey capture behavior. However, it appeared not to

improve the prey capture success of H. erectus, at least when foraging on Mysidopsis bahia. The

acoustic role of the click in prey capture behavior should be further examined with other prey

items (e.g., amphipods) that make up a larger component of the wild seahorse’s diet (Texeira and

Musick, 2001).

This Dissertation provides compelling evidence for the role of the click in intraspecific

communication, however, especially in the context of courtship behavior. The first line of

evidence stems from Chapter 3. As previously stated, hearing in fishes is thought to have first

evolved to evaluate the auditory scene (Popper and Fay, 1999). This theory is supported by

several lines of evidence. First, most hearing generalist fishes exhibit optimum hearing

sensitivity below 500 Hz (Popper et al., 2003). This is concurrent with the observation that most

shallow water ambient noise occurs in the range of 50 to 500 Hz (Bass and Clark, 2003). The

inner ear evolved to a successful design early on in the evolutionary time table; it is found among

the most primitive of jawless vertebrates and is ubiquitous in the vertebrate subphylum (Popper

and Fay, 1999). By contrast, sound production is found in only a subset of fishes (Popper et al.,

2003) and it has independently evolved multiple times (Bass and Clark, 2003), suggesting its

origins of evolution after the evolution of the inner ear. The evolution of sound production for

intraspecific communication is thus constrained by the limitations of the auditory system.

162

To boost the likelihood of detection by a limited auditory sense, signals may have evolved

several features. In the case of the seahorse click, sudden onset and its broadband nature boost

likelihood of detection, as previously discussed. An especially compelling line of evidence is the

frequency match between the peak frequency of the click (at 271 + 10 Hz for sound pressure and

265 + 22 Hz for particle acceleration) and the animal’s optimum hearing sensitivity between 200

and 300 Hz (Figures 3-12 and Figure 3-13). This correspondence between auditory sensitivity

and vocalization has been demonstrated for other sound-producing fishes as well (e.g.,

Stabentheiner, 1988; Ladich and Yan, 1998; Ladich, 1999; Wysocki and Ladich, 2001; Lugli et

al., 2003). The evolutionary shaping of the signal in both the temporal and frequency domains to

accommodate the hearing ability of the animal lends strong evidence for the role of the click in

intraspecific communication.

The second line of evidence stems from Chapter 6. Male seahorses increased clicking

behavior as the courtship sequence progressed over time. This correlates with other documented

courtship behaviors that also increase over time (Masonjones and Lewis, 1996; Chapter 6). This

suggests that clicking may be a component behavior of the courtship sequence. The lack of

discussion or examination of click behavior in courtship in other literature may be because

clicking most often occurs while animals are apart. This is not necessarily counter-intuitive;

clicking may serve as an alternate signaling system when distance or obstructions prevent visual

signaling. This pattern also occurs in gobies (Tavolga, 1958); males call frequently when

females are outside of the nest but cease calling once females enter; other courtship and mating

activities take precedence.

It is also evident that animals are responding to acoustic signals in the context of courtship;

male approach and female pointing did not escalate over the course of the courtship sequence

163

among muted pairs. This suggests a downregulation of behavioral response in the reciprocal

signaling system of the courtship repertoire (Vincent and Sadler, 1995) as a result of the removal

of acoustic signals from the repertoire.

Clicking may signal any one of a number of messages to a mate, including species

discrimination (e.g., Delco, 1960; Gerald, 1971; Myrberg and Spires, 1972; Spanier, 1979),

potential mate quality (e.g., Myrberg et al., 1986; Colson et al., 1998), location and/or

reproductive readiness (Tavolga, 1958; Delco, 1960; Ibara et al., 1983; Myrberg et al., 1986),

and because clicking occurs so reliably with feeding behavior (Colson et al., 1998), presence of a

food source. Further studies in which one sex is muted while the other is not are suggested to

further elucidate the functions of the click in courtship behavior; this is especially sensible in

light of observed sex differences in the trend of clicking over time during the courtship sequence.

Variability in Signaling

In this Dissertation, only feeding clicks were recorded and quantified, as they could be

reliably produced upon the presentation of prey. It may be worthwhile to record and quantify

clicking signals that occur in the absence of a food source; these clicks may be intended by the

signaler as communication signals and may therefore be shaped differently in order to optimize

reception by conspecifics (e.g., they may be louder, have peaks at different frequencies, etc.).

In several instances, I noted that the lined seahorses made a very low frequency vibration

that originated from within the body cavity when handled out of water. Occasionally, this

vibration was of sufficient intensity to cause the head of the animal to vibrate visibly. This was a

“signal” that was felt, as opposed to heard, and may occur in context of the animal’s natural

history (I hypothesize that this vibration might occur during quivering behavior of courtship as

described in Masonjones and Lewis, 1996; and Vincent, 1994b; and may be detected by the

lateral line of mates). The functional significance of this signal deserves further exploration.

164

Masking

In order for biologically important signals to be heard, they must not only be audible (i.e.,

above the hearing threshold of the receiving fish at a given frequency), but they must also

overcome a critical signal to noise ratio. As discussed in Chapter 2, signals must be anywhere

between 10 to 60 dB (median around 20 dB) louder than the background environmental noise in

order to be audible to hearing generalist fishes. In the case of the seahorse, mean spectrum-level

amplitude of ambient noise is 82.6 + 2.9 dB SPL re: 1 µPa at 200 Hz in the wild. The putative

communication signal, the click, demonstrates a mean spectrum-level peak amplitude of 94.3 +

0.9 dB at a nearby mean peak frequency of 271 + 10 Hz. This is a signal-to-noise ratio of

roughly 12 dB. Other elements of the click, such as its sudden onset and its broadband nature,

also boost its likelihood of detection (Hall, 1992; Yost, 2000). It must also be remembered that

AEP methodology underestimates hearing thresholds in comparison to behavioral methods by up

to 19 dB (Hill, 2005). Together, the evidence suggests that clicks are likely to be audible to

conspecifics, even in the background of ambient noise encountered in the wild. Testing the AEP

response to click stimuli in the presence of environmentally relevant background noise would

provide more conclusive evidence.

By contrast, in aquarium environments, mean spectrum-level amplitude of ambient noise is

94.6 + 1.4 dB at 200 Hz at the tank bottom, which is virtually the same amplitude of the mean

spectrum-level peak amplitude of the click. This would suggest that the click is much less likely

to overcome the critical signal-to-noise ratio required in order to be heard by conspecifics in this

artificial environment, rendering masking a likely phenomenon occurring among seahorses in

aquaria. But results of Chapter 6 do not support masking as a hypothesis. Furthermore, results

in Chapter 3 suggest that, in the absence of the consideration of ambient noise, seahorse clicks

are more likely to be audible in terms of particle acceleration than in terms of sound pressure

165

(Figures 3-12 and 3-13). In light of these lines of evidence, it is suggested that animals may be

responding primarily to the particle acceleration modality of sound, and not the pressure

modality. In the particle acceleration modality, where clicks are more likely to be audible,

masking is less likely to occur. Further examination of the ambient noise fields in both wild and

captive environments in terms of particle acceleration is needed. Also, most masking studies

document sound in terms of sound pressure; masking studies of particle acceleration are also

needed, as there is no a priori reason to expect that critical ratios would match those for sound

pressure, especially in light of the lack of correlation between the two parameters in some

environments.

The Lateral Line

The lateral line is a sensory system that was not examined in this Dissertation but may

detect ambient noise and/or signals. Like the inner ear, the lateral line is also an acceleration

detector. But the lateral line is only stimulated by local, near-field, hydrodynamic flow. The

lateral line is optimally sensitive at up to 70 Hz for free neuromasts or 180 Hz for canal

neuromasts (Kalmijn, 1988). In the wild, ambient noise is unlikely to originate within the near-

field of the fish and is thus unlikely to be detected by the lateral line. Signals generated within

close proximity of a fish (e.g., by a mate in close courtship exchanges, by a competitor in close

aggressive exhanges, or by an approaching predator) may be detected by the lateral line if they

have energy below 100 or 200 Hz. In the wild, then, lateral lines may be exempt from some of

the signal-to-noise ratio problems that the inner ear must resolve and may therefore be a reliable

detector of low-frequency acoustic signals.

In most small to medium sized aquaria, ambient noise is likely to be generated within the

near-field of the fish, and therefore low-frequency components of ambient noise are likely to be

detected by the lateral line. I invite the reader to examine all the power spectra among the

166

Figures presented in this Dissertation. They depict sound energy in a range of 0-1000 Hz. The

reader will note that in most cases, the region of loudest sound energy is below 100 Hz, precisely

where lateral lines exhibit optimum sensitivity.

The consequence of these observations is that masking of signals could occur with the

lateral line sense (if signal amplitude is important in this frequency region); and/or the stress

response may actually be mediated via the lateral line in addition to, or instead of, the inner ear.

In general, the lateral line has received much less attention than the inner ear over time. As

late as 1984, Sand acknowledged the lack of experimental evidence to support the notion that the

lateral line detects low-frequency sound (this is now accepted in the scientific community).

Thorough investigation of a fish’s experience of the lateral line sense would be fruitful here. In

particular, I suggest comparison of lateral line detection of ambient noise in confined vs. open

environments, masking of signals by noise in the lateral line system, and the potential for chronic

excitation of the lateral line system to induce chronic stress in fishes. The relative roles of the

lateral line and inner ear in detecting communication signals by conspecifics is also worthy of

investigation; though in the lined seahorse, the peak frequency of the click (at 271 + 10 Hz in the

pressure modality and 265 + 22 Hz in the particle acceleration modality) is outside the range of

sensitivity of the lateral line.

The Stress Response

Physiology

The suite of physiological measures examined in response to chronic noise exposure in

Chapter 4 suggests that seahorses present a chronic stress response.

In particular, higher heterophil to lymphocyte ratios were observed; this is a secondary

stress response representing an alteration of the immune system. Seahorses are already known to

have a reduced immune system, characterized by the lack of gut-associated lymphoid tissue

167

(Matsunaga and Rahman, 1998). In aquarium and aquaculture settings, syngnathids are known

to be particularly susceptible to stress-induced disease (e.g., Berzins and Greenwell, 2001, Frasca

et al., 2005, Vincent, 1998). This result in lieu of these predispositions warns aquarists of the

potential for loud ambient noise to render seahorses vulnerable to opportunistic disease.

Seahorses also demonstrated tertiary responses in two important measures in aquaculture,

in weight and body condition. As the goal of aquaculture (particularly food-fish aquaculture) is

to rapidly produce fish muscle mass as the product of sale, these results suggest that sound stress

can hinder this effort.

Behavior

Behavioral stress responses were more clearly elucidated when heterogeneity of variance

was examined between treatments. This method is guided by abundant observations and reviews

by others who have documented increased variance in responses to stress in populations of

animals (e.g., Orlando and Guillette, 2001), perhaps due to alternative coping mechanisms

employed by different subpopulations of animals (e.g., Koolhaas et al., 1999; Øverli et al., 2007).

Behavioral alterations exhibited in Chapter 4 were subtle. In the first week of exposure,

animals made more adjustments on holdfasts among loud tanks, and the number of adjustments

made was more variable among animals housed in loud tanks. The lack of differences in this

measure in subsequent weeks suggests behavioral habitation to the presence of the noise

stimulus. It wasn’t until week 4 that clicking and piping, putative distress behaviors, began to

appear more variably among animals in loud tanks.

Despite a documented physiological stress response, chronic loud noise exposure did not

appreciably affect feeding behavior in Chapter 5 or courtship behavior in Chapter 6. The results

of Chapter 5 are interesting in that they are counterintuitive to the physiological results of

Chapter 4. Animals in loud tanks lost weight and body condition over time, but the results of

168

Chapter 5 suggest that this may not be due to a reduced feeding response; instead, weight loss

may be the result of an increased metabolic demand on the animal to maintain homeostasis in a

suboptimal environment (or allostasis, Sterling and Eyer, 1988).

The sum of these subtle and variable behavioral results in spite of clear physiological

results suggests that animal behavior may not always be a reliable indicator of chronic stress in

fishes.

Solutions

This dissertation documents a chronic stress response among fishes housed in loud tanks in

aquaculture. The sound pressure levels that produced a stress response are well within the range

of ambient noise encountered among public aquaria. In monitoring chronic stress in fishes,

behavior may not always be a reliable indicator. For aquaculturists and aquarists, the non-lethal

method of measuring body lengths and weights is suggested to monitor the physiological status

of fishes in culture over time as a potential indicator of chronic stress.

In the design of fish-holding systems, it’s important first to think about the natural history

and the umwelt of the animal(s) intended to be kept. Does the animal spend the majority of its

time swimming in the water column, or is it a benthic-dwelling animal? Is the animal a hearing

generalist or a hearing specialist?

Chapter 2 demonstrates that ambient noise profiles differ at different locations within an

aquarium environment. Specifically, the ambient noise encountered at the bottom of a tank is

significantly louder than the ambient noise encountered in the middle of the water column. In

the case of H. erectus, ambient noise at the middle of the water column is not significantly louder

than ambient noise in the wild, but ambient noise at the tank bottom is. Thus, benthic dwelling

animals (such as H. erectus) may be more susceptible to chronic loud noise exposure in this

region of the aquarium environment, and soundproofing measures should be focused on reducing

169

noise at tank bottoms. In contrast, tank noise may not be loud enough to be of concern for some

mid-water column dwelling fishes.

Hearing specialists have greater hearing sensitivity, and hear at higher frequencies, than do

hearing generalists. In light of the evidence demonstrated here that even a hearing generalist is

susceptible to noise-induced stress, the acoustic environments of hearing specialists should

especially be scrutinized. The dichotomy of hearing generalists vs. hearing specialists is

important in the consideration of resonance in aquaria. Resonance is the tendency of a system to

oscillate at maximum amplitudes at certain frequencies, and tends to increase sound duration,

frequency distribution, and amplitude of signals (Akamatsu et al., 2002). Fortunately for hearing

generalist fishes, resonant frequencies tend to fall above the hearing range in most tanks,

according to the minimum resonant frequency calculations of Akamatsu et al. (2002). In

contrast, for hearing specialists, some of which have sensitive hearing up to 2,000 Hz and less

sensitive hearing up to 4,000 Hz (Popper et al., 2003), rectangular tanks as small as 570 L may

have minimum resonant frequencies that fall into their range of best hearing sensitivity (2,000

Hz), with very small tanks (38 to 570 L) exhibiting minimum resonant frequencies in the range

of 2,000 to 4,000 Hz. Tank size is thus a variable to consider, especially when housing hearing

specialist fishes.

Once these variables are considered, soundproofing design modifications can be employed

according to recommendations suggested by Davidson et al. (2007) and by the author in Chapter

2 of the Dissertation. These sensible recommendations promise to reduce ambient noise in

aquarium and culture environments, thereby reducing chronic stress exposure among captive

fishes.

170

Figure 7-1. Comparisons of ambient noise in tank environments with broadband hearing in

Hippocampus erectus. Power spectra of ambient noise recorded from the bottom of representative tanks of the seahorse sound survey at the minimum and maximum of the range (both in solid gray) and at the median of the range (in solid black). Audiogram of H. erectus (adjusted to estimate audition of broadband noise) is represented by the dashed black line.

171

Figure 7-2. Comparisons of ambient noise in the wild with broadband hearing in Hippocampus

erectus. Power spectra of ambient noise recorded from the wild at the minimum and maximum of the range (both in solid gray) and at the median of the range (in solid black). Audiogram of H. erectus (adjusted to estimate audition of broadband noise) is represented by the dashed black line.

172

REFERENCES

Akamatsu, T., Okumura, T., Novarini, N., Yan, H.Y., 2002. Empirical refinements applicable to the recording of fish sounds in small tanks. J. Acoust. Soc. Am. 112(6), 3073-3082.

Al-murrani, W.K., Al-Rawi, I.K., Raof, N.M., 2002. Genetic resistance to Salmonella typhimurium in two lines of chickens selected as resistant and sensitive on the basis of heterophil/lymphocyte ratio. Brit. Poult. Sci. 43, 501-507.

Amoser, S., Ladich, F., 2003. Diversity in noise-induced temporary hearing loss in otophysine fishes. J. Acoust. Soc. Am. 113(4), 2170-2179.

Amoser, S., Ladich, F., 2005. Are hearing sensitivities of freshwater fish adapted to the ambient noise in their habitats? J. Exp. Biol. 208, 3533-3542.

Anderson, R.O., Neumann, R.M., 1996. Length, weight, and associated structural indices. In: Murphy, B.R., Willis, D.W. (Eds.), Fisheries Techniques. American Fisheries Society, Bethesda, pp. 447-482.

Balm, P.H.M., 1997. Immune-endocrine interactions. In: Iwama, G.K., Pickering, A.D., Sumpter, J.P., Schreck, C.B. (Eds.), Fish Stress and Health in Aquaculture. Cambridge University Press, New York, pp. 195-221.

Banner, A. Hyatt, M., 1973. Effects of noise on eggs and larvae of two estuarine fishes. Trans. Am. Fish. Soc. 1973(1), 134-136.

Barcellos, L.J.G., Kreutz, L.C., de Souza, C., Rodrigues, L.B., Fioreze, I., Quevedo, R.M., Cericato, L., Soso, A.B., Fagundes, M., Conrad, J., Lacerda, L. dA., Terra, S. 2004. Hematological changes in jundiá (Rhamdia quelen Quoy and Gaimard Pimelodidae) after acute and chronic stress caused by usual aquacultural management, with emphasis on immunosuppressive effects. Aquaculture 237, 229-236.

Bardach, J.E., Magnuson, J.J., 1980. Introduction and perspectives. In: Bardach, J.E., Magnuson, J.J., May, R.C., Reinhart, J.M. (Eds.), Fish Behavior and Its Use in the Capture and Culture of Fishes. International Center for Living Aquatic Resources Management, Manila, pp. 1-31.

Barimo, J.F., Fine, M.L., 1998. Relationship of swim-bladder shape to the directionality pattern of underwater sound in the oyster toadfish. Can. J. Zool. 76, 134-143.

Barnett, C.W., Pankhurst, N.W., 1998. The effects of common laboratory and husbandry practices on the stress response of the greenback flounder Rhombosolea tapirina (Günther, 1862). Aquacult. 162, 313-329.

Bart, A.N., Clark, J., Young, J., Zohar, Y., 2001. Underwater ambient noise measurements in aquaculture systems: A survey. Aquacult. Eng. 25, 99-110.

Bass, A.H., 1990. Sounds from the intertidal zone: Vocalizing fish. BioScience 40(4), 249-258.

173

Bass, A.H., Clark, C.W., 2003. The physical acoustics of underwater sound communication. In: Simmons, A.M., Popper, A.N., Fay, R.R. (Eds.), Acoustic Communication. Springer-Verlag, New York, pp. 15-64.

Bass, A.H., McKibben, J.R., 2003. Neural mechanisms and behaviors for acoustic communication in teleost fish. Prog. Neurobiol. 69, 1-26.

Beitinger, T.L. 1990., Behavioral reactions for the assessment of stress in fishes. J. Great Lakes Res. 16(4), 495-528.

Benoit-Bird, K.J., Au, W.W.L., Kastelein, R., 2006. Testing the odontocete acoustic prey debilitation hypothesis: No stunning results. J. Acoust. Soc. Am. 120 (2), 1118-1123.

Berzins, I.K., Greenwell, M., 2001. An overview of common syngnathid health problems. Proceedings of the 2nd International Conference on Marine Ornamentals, 26 November-1 December 2001, Lake Buena Vista, FL, 117.

Billard, R., Bry, C., Gillet, C., 1981. Stress, environment and reproduction in teleost fish. In: Pickering, A.D. (Ed.), Stress and Fish. Academic Press, New York, pp. 185-208.

Biswas, A.K., Maita, M., Yoshizaki, G., Takeuchi, T., 2004. Physiological responses in Nile tilapia exposed to different photoperiod regimes. J. Fish Biol. 65, 811-821.

Biswas, A.K., Seoka, M., Takii, K., Maita, M., Kumai, H., 2006. Stress response of red sea bream Pagrus major to acute handling and chronic photoperiod manipulation. Aquaculture 252, 566-572.

Blasiola, G.C., 1979. Glugea heraldi n. sp. (Microsporidea, Glugeidae) from the seahorse Hippocampus erectus Perry. J. Fish Dis. 2, 493-500.

Blaxter, J.H.S., Batty, R.S., 1987. Comparisons of herring behaviour in the light and dark: Changes in activity and responses to sound. J. Mar. Biol. Assoc. UK 67, 849-860.

Blaxter, J.H.S., Gray, J.A.B., Denton, E.J., 1981. Sound and startle responses in herring shoals. J. Mar. Biol. Assoc. UK 61, 851-869.

Blumstein, D.T., Evans, C.S., Daniel, J.C., 2000. JWatcher 0.9. http://www.jwatcher.ucla.edu/

Bromage, N., Shepherd, J., Roberts, J., 1988. Farming systems and husbandry practice. In: Shepherd, C.J., Bromage, N.R. (Eds.), Intensive Fish Farming. BSP Professional Books, Oxford, pp. 50-102.

Burmeister, S., Somes, C., Wilczynski, W., 2001. Behavioral and hormonal effects of exogenous vasotocin and corticosterone in the green treefrog. Gen. Comp. Endocrinol. 122, 189-197.

http://www.jwatcher.ucla.edu/�

174

Cahn, P.H., Siler, W., Wodinsky, J., 1968. Acoustico-lateralis system of fishes: Tests of pressure and particle-velocity sensitivity in grunts, Haemulon sciurus and Haemulon parrai. J. Acoust. Soc. Am. 46(6), 1572-1578.

Caldwell, D.K., Caldwell, M.C., 1967. Underwater sounds associated with aggressive behavior in defense of territory by the pinfish, Lagodon rhomboides. Bull. S. Calif. Acad. Sci. 66, 69-75.

Campbell, P.M., Pottinger, T.G., Sumpter, J.P., 1994. Preliminary evidence that chronic confinement stress reduces the quality of gametes produced by brown and rainbow trout. Aquacult. 120, 151-169.

Campo, J.L., Gil, M.G., Dàvila, S.G., Muñoz, I., 2005. Influence of perches and footpad dermatitis on tonic immobility and heterophil to lymphocyte ratio of chickens. Poult. Sci. 84, 1004-1009.

Campo, J.L., Gil, M.G., Dàvila, S.G., Muñoz, I., 2007. Effect of lighting stress on fluctuating asymmetry, heterophil-to-lymphocyte ratio, and tonic immobility duration in eleven breeds of chickens. Poult. Sci. 86, 37-45.

Carragher, J.F., Sumpter, J.P., Pottinger, T.G., Pickering, A.D., 1989. The deleterious effects of cortisol implantation on reproductive function in two species of trout, Salmo trutta L. and Salmo gairdneri Richardson. Gen. Comp. Endocrinol. 76, 310-321.

Case, B.C., Lewbart, G.A., Doerr, P.D., 2005. The physiological and behavioural impacts of and preference for an enriched environment in the eastern box turtle (Terrapene carolina carolina). Appl. Anim. Behav. Sci. 92, 353-365.

Casper, B.M., Mann, D.A., 2006. Evoked potential audiograms of the nurse shark (Ginglystoma cirratum) and the yellow stingray (Urobatis jamaicensis). Environ. Biol. Fish. 76, 101-108.

Casper, B.M., Lobel, P.S., Yan, H.Y., 2003. The hearing sensitivity of the little skate, Raja erinacea: A comparison of two methods. Environ. Biol. Fish. 68, 371-379.

Chapman, C.J., Johnstone, A.D.F., 1974. Some auditory discrimination experiments on marine fish. J. Exp. Biol. 61, 521-528.

Chapman, C.J., 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.

CITES, 2001. Convention on International Trade in Endangered Species of Wild Fauna and Flora. http://www.cites.org, accessed May 2009.

Colson, D.J., Patek, S.N., Brainerd, E.L., Lewis, S.M., 1998. Sound production during feeding in Hippocampus seahorses (Syngnathidae). Env. Biol. Fish. 51, 221-229.

Corwin, J.T., Bullock, T.H., Schweitzer, J., 1982. The auditory brain stem response in five vertebrate classes. Electroencephalogr. Clin. Neurophysiol. 54, 629-641.

http://www.cites.org/�

175

Cotter, P.A., Harris, J.O., McLean, E., Craig, S., Schwarz, M., Rasmussen, M.R., 2005. The effect of tank color upon growth performance and stress response of summer flounder Paralichthys dentatus. Poster, Aquaculture America 2005, 17-21 Jan 2005, New Orleans, LA.

Cox, M., Rogers, P.H., Popper, A.N., Saidel, W.M., Far, R.R., 1986a. Frequency regionalization in the fish ear. J. Acoust. Soc. Am. Suppl. 1 79, S80.

Cox, M., Rogers, P.H., Popper, A.N., Saidel, W.M., 1986b. Anatomical effects of intense tone stimulation in the ear of bony fishes. J. Acoust. Soc. Am. Suppl. 1 80, S75.

Cox, M., Rogers, P.H., Popper, A.N., Saidel, W.M., Fay, R.R., Coombs, S., 1987. Anatomical effects of intense tone stimulation in the goldfish ear: Dependence on sound pressure level and frequency. J. Acoust. Soc. Am. Suppl. 1 89, S7.

Crawford, J.D., 1997. Hearing and acoustic communication in mormyrid electric fishes. Mar. Fresh. Behav. Physiol. 29, 65-86.

D’Aquila, P.S., Brain, P., Willner, P., 1994. Effects of chronic mild stress on performance in behavioural tests relevant to anxiety and depression. Physiol. Behav. 56(5), 861-867.

Datta, M., Kaviraj, A., 2003. Ascorbic acid supplementation of diet for reduction of deltamethrin induced stress in freshwater catfish Clarias gariepinus. Chemosphere 53, 883-888.

Davis, H., 1976. Principles of electric response audiometry. Ann. Otol. Rhinol. Laryngol. Suppl. 29 85, 1-95.

Davidson, J., Frankel, A.S., Ellison, W.T., Summerfelt, S., Popper, A.N., Mazik, P., Bebak, J., 2007. Minimizing noise in fiberglass aquaculture tanks: Noise reduction potential of various retrofits. Aquacult. Eng. 37, 125-131.

Delco, E.A. Jr., 1960. Sound discrimination by males of two cyprinid fishes. Tex. J. Sci. 12, 48-54.

Dougherty, T.F., 1960. Lymphocytokaryorrhectic effects of adrenocortical steroids. In: Rebuck, J.W. (Ed.), The Lymphocyte and Lymphocytic Tissue. Haper (Hoeber), New York, pp. 112-124.

Duffy, B.K., Gurm, H.S., Rajagopal, V., Gupta, R., Ellis, S.G., Bhatt, D.L., 2006. Usefulness of an elevated neutrophil to lymphocyte ratio in predicting long-term mortality after percutaneous coronary intervention. Am. J. Cardiol. 97(7), 993-996.

Dufossé, A., 1874. Recherches sur les bruits et les sous expressifs que font entendre les poisons d’Europe et sur les organs producteurs de ces phenomenes acoustiques ainsi que sur les appareils de l’audition de plusieurs de ces animaux. Annal. Sci. Nat. 19, 53pp., 20, 134pp.

Dunning, D.J., Ross, Q.E., Geoghegan, P., Reichle, J.J., Menezes, J.K., Watson, J.K., 1992. Alewives avoid high-frequency sound. N. Am. J. Fish. Mana. 12, 407-416.

176

Dusenbery, D.B., 1992. Sensory ecology: How organisms acquire and respond to information. W.H. Freeman and Co., New York, 558pp.

Ellsaesser, C.F., Clem, L.W., 1986. Haematological and immunological changes in channel catfish stressed by handling and transport. J. Fish Biol. 28, 511-521.

Egner, S., Mann, D., 2005. Auditory sensitivity of sergeant major damselfish (Abudefduf saxatilis) from post-settlement juvenile to adult. Mar. Ecol. Prog. Ser. 285, 213-222.

Enger, P.S., 1981. Frequency discrimination in teleosts—central or peripheral? In: Tavolga, W.N., Popper, A.N., Fay, R.R. (Eds.), Hearing and Sound Communication in Fishes. Springer-Verlag, New York, pp. 243-255.

Fasolo, A.G., Krebs, R.A., 2004. A comparison of behavioral change in Drosophila during exposure to thermal stress. Biol. J. Linn. Soc. 83, 197-205.

Fay, R.R., 1974. Masking of tones for the goldfish (Carassius auratus). J. Comp. Physiol. Psychol. 87(4), 708-716.

Fay, R.R., 1988a. Hearing in vertebrates: A Psychophysics Databook. Hill-Fay, Winetka, IL.

Fay, R.R., 1988b. Peripheral adaptations for spatial hearing in fish. In: Atema, J., Fay, R.R., Popper, A.N., Tavolga, W.N. (Eds.), Sensory Biology of Aquatic Animals. Springer-Verlag, New York, pp. 711-731.

Fay, R.R., 1995. Psychoacoustical studies of the sense of hearing in goldfish using conditioned respiratory suppression. In: Klump, G.M., Dooling, R.J., Fay, R.R., Stebbins, W.C. (Eds.), Methods in Comparative Psychoacoustics. Birkhäuser, Basel, pp. 249-261.

Fay, R.R., Popper, A.N., 1975. Modes of stimulation of the teleost ear. J. Exp. Biol. 62, 379-387.

Fay, R.R., Hillery, C.M., Bolan, K., 1982. Representation of sound pressure and particle motion information in the midbrain of the goldfish. Comp. Biochem. Physiol. A 71, 181-191.

Fay, R.R., Simmons, A.M., 1999. The sense of hearing in fishes and amphibians. In: Fay, R.R., Popper, A.N. (Eds.), Comparative Hearing: Fish and Amphibians. Springer-Verlag, New York, pp. 269-318.

Feltenstein, M.W., Lambdin, L.C., Webb, H.E., Warnick, J.E., Khan, S.I., Khan, I.A., Acevedo, E.O., Sufka, K.J., 2003. Corticosterone response in the chick separation-stress paradigm. Physiol. Behav. 78, 489-493.

Fish, M.P., Kelsey, A.S.Jr., Mowbray, W.H., 1952. Studies on the production of underwater sound by North Atlantic coastal fishes. J. Mar. Res. 11(2), 180-193.

Fish, M.P., 1953. The production of underwater sound by the Northern seahorse, Hippocampus hudsonius. Copeia 1953, 98-99.

177

Fish, M.P., 1954. The character and significance of sound production among fishes of the Western North Atlantic. Bull. Bing. Oceanogr. Coll. 14.

Fisher, A.D., Crowe, M.A., Ó Nuallàin, E.M., Monaghan, M.L., Prendiville, D.J., O’Kiely, P., Enright, W.J., 1997. Effects of suppressing cortisol following castration of bull calves on adrenocorticotropic hormone, in vitro interferon γ production, leukocytes, acute-phase proteins, growth, and feed intake. J. Anim. Sci. 75, 1899-1908.

Foster, S.J., Vincent, A.C.J., 2004. Life history and ecology of seahorses: Implications for conservation and management. J. Fish Biol. 65, 1-61.

Francis-Floyd, R., 1988. Behavioral Diagnosis. Vet. Clin. N. Am.-Small 18(2), 305-316.

Frasca, S., Nyaoke, A., Hinckley, L., de Hoog, S., Wickes, B., Sutton, D., Weber, E.S., Keller, C., 2005. An extreme example of common seahorse diseases. Proceedings of the 30th Eastern Fish Health Workshop, Shepherdstown, WV, p. 42.

Gagnon, A., Jumarie, C., Hontela, A., 2006. Effects of Cu on plasma cortisol and cortisol secretion by adrenocortical cells of rainbow trout (Oncorhynchus mykiss). Aquat. Toxicol. 78, 59-65.

Gerald, J.W., 1971. Sound production during courtship in six species of sunfish (Centrarchidae). Evolution 25, 75-87.

Goudie, R.I., Jones, I.L., 2004. Dose-response relationships of harlequin duck behaviour to noise from low-level military jet over-flights in central Labrador. Environ. Conserv. 31(4), 289-298.

Grant, S.M., Brown, J.A., Boyce, D.L., 1998. Enlarged fatty livers of small juvenile cod: a comparison of laboratory-cultured and wild juveniles. J. Fish Biol. 52, 1105-1114.

Gregory, T.R., Wood, C.M., 1999. The effects of chronic plasma cortisol elevation on the feeding behaviour, growth, competitive ability, and swimming performance of juvenile rainbow trout. Physiol. Biochem. Zool. 72(3), 286-295.

Gross, W.B., Siegel, H.S., 1983. Evaluation of the heterophil/lymphocyte ratio as a measure of stress in chickens. Avian Dis. 27(4), 972-979.

Hall, J.W., 1992. Handbook of Auditory Evoked Responses. Allyn and Bacon, Boston, 886 pp.

Hansen, S.W., Damgaard, B.M., 1993. Behavioural and adrenocortical coping strategies and the effect on eosinophil leucocyte level and heterophil/lymphocyte-ratio in beech marten (Martes foina). Appl. Anim. Behav. Sci. 35, 369-388.

Hastings, M.C., Popper, A.N., Finneran, J.J., Lanford, P.J., 1996. Effects of low-frequency underwater sound on hair cells of the inner ear and lateral line of the teleost fish Astronotus ocellatus. J. Acoust. Soc. Am. 99 (3), 1759-1766.

178

Hawkins, A.D., Chapman, C.J., 1975. Masked auditory thresholds in the cod, Gadus morhua L. J. Comp. Physiol. 103, 209-226.

Hawkins, A.D., Johnstone, A.D.F., 1978. The hearing of the Atlantic salmon, Salmo salar. J. Fish. Biol. 13, 655-673.

Herberholz, J., Schmitz, B., 1999. Flow visualization and high speed video analysis of water jets in the snapping shrimp (Alpheus heterochaelis). J. Comp. Physiol. A 185, 41-49.

Hill, R.J., 2005. Standardizing The Auditory Evoked Potential Technique: Ground-truthing against behavioral conditioning in the goldfish, Carassius auratus. M.S. Thesis, University of South Florida, St. Petersburg, 174pp.

Högstedt, G., 1983. Adaptation unto death: Function of fear screams. Am. Nat. 121(4), 562-570.

Holmer, H.K., Rodman, J.E., Helmreich, D.L., Parfitt, D.B., 2003. Differential effects of chronic escapable versus inescapable stress on male Syrian hamster (Mesocricetus auratus) reproductive behavior. Horm. Behav. 43, 381-387.

Huff, G.R., Huff, W.E., Balog, J.M., Rath, N.C., Anthony, N.B., Nestor, K.E., 2005. Stress response differences and disease susceptibility reflected by heterophil to lymphocyte ratio in turkeys selected for increased body weight. Poult. Sci. 84, 709-717.

Ibara, R.M., Penny, L.T., Ebeling, A.W., van Dykhuizen, G., Cailliet, G., 1983. The mating call of the plainfin midshipman fish, Porichthys notatus. In: Noakes, D.L.G., Lindquist, D.G., Helfman, G.S., Ward, J.A. (Eds.), Predators and Prey in Fishes. Dr. W. Junk Publ., The Hague, pp. 205-212.

Iversen, M., Finstad, B., Nilssen, K.J., 1998. Recovery from loading and transport stress in Atlantic salmon (Salmo salar L.) smolts. Aquaculture 168, 387-394.

Iwama, G.K., Pickering, A.D., Sumpter, J.P., Schreck, C.B. (Eds.), 1997. Fish stress and health in aquaculture. Society for Experimental Biology Seminar Series 62, Cambridge University Press, Cambridge, 278pp.

Jacobson, J.T., 1985. The Auditory Brainstem Response. College-Hill Press, San Diego, 444pp.

Jerkø, H., Turunen-Rise, I., Enger, P.S., Sand, O., 1989. Hearing in the eel (Anguilla anguilla). J. Comp. Physiol. A 165, 455-459.

Jian-yu, X., Xiang-wen, M., Ying, L., Shao-rong, C., 2005. Behavioral response of tilapia (Oreochromis niloticus) to acute ammonia stress monitored by computer vision. J. Zheijiang Univ. Sci. 6B(8), 812-816.

Job, S.D., Do, H.H., Meeuwig, J.J., Hall, H.J., 2002. Culturing the oceanic seahorse, Hippocampus kuda. Aquacult. 214, 333-341.

179

Jones, A.G., Moore, G.I., Kvarnemo, C., Walker, D., Avise, J.C., 2003. Sympatric speciation as a consequence of male pregnancy in seahorses. PNAS 100, 6598-6603.

Kakizawa, S., Kaneko, T., Hasegawa, S., Hirano, T., 1995. Effects of feeding, fasting, background adaptation, acute stress, and exhaustive exercise on the plasma somatolactin concentrations in rainbow trout. Gen. Comp. Endocr. 98, 137-146.

Kalmijn, A.J., 1988. Hydrodynamic and acoustic field detection. In: Atema, J., Fay, R., Popper, A., Tavolga, W. (Eds.), Sensory Biology of Aquatic Animals. Springer-Verlag, New York, pp. 83-130.

Kelly, J.C., Nelson, D.R., 1975. Hearing thresholds of the horn shark, Heterodotus francisci. J. Acoust. Soc. Am. 58(4), 905-909.

Kenyon, T.N., 1994. The significance of sound interception to males of the bicolor damselfish, Pomacentrus partitus, during courtship. Env. Biol. Fish. 40, 391-405.

Kenyon, T.N., 1996. Ontogenetic changes in the auditory sensitivity of the bicolor damselfish, Pomacentrus partitus (Poey). J. Comp. Physiol. A 179, 553-561.

Kenyon, T.N., Ladich, F., Yan, H.Y., 1998. A comparative study of hearing ability in fishes: the auditory brainstem response approach. J. Comp. Physiol. A 182, 307-318.

Khan, R.A., 2006. Assessment of stress-related bioindicators in winter flounder (Pleuronectes americanus) exposed to discharges from a pulp and paper mill in Newfoundland: A 5-year field study. Arch. Environ. Con. Tox. 51, 103-110.

Knudsen, F.R., Enger, P.S., Sand, O., 1992. Awareness reactions and avoidance responses to sound in juvenile Atlantic salmon, Salmo salar L. J. Fish. Biol. 40, 523-534.

Knudsen, F.R., Schreck, C.B., Knapp, S.M., Enger, P.S., Sand, O., 1997. Infrasound produces flight and avoidance responses in Pacific juvenile salmonids. J. Fish Biol. 51, 824-829.

Koolhaas, J.M., Kort, S.M., De Boer, S.F., Van Der Vegt, B.J., Van Reenen, C.G., Hopster, H., De Jong, I.C., Ruis, M.A.W., Blokhuis, H.J. 1999. Coping styles in animals: Current status in behavior and stress-physiology. Neurosci. Biobehav. Rev. 23, 925-935.

Kubokawa, K., Watanabe, T., Yoshioka, M., Iwata, M., 1999. Effects of acute stress on plasma cortisol, sex steroid hormone and glucose levels in male and female salmon during the breeding season. Aquaculture 172, 335-349.

Kuiter, R.H., 2000. Seahorses, pipefishes, and their relatives. TMC Publishing, Chorleywood, 240pp.

Ladich, F., 1997. Agonistic behaviour and significance of sounds in vocalizing fish. Mar. Fresh. Behav. Physiol. 29, 87-108.

180

Ladich, F., 1999. Did auditory sensitivity and vocalization evolve independently in otophysan fishes? Brain Behav. Evol. 53, 288-304.

Ladich, F., Yan, H.Y., 1998. Correlation between auditory sensitivity and vocalization in anabantoid fishes. J. Comp. Physiol. A 182, 737-746.

Lagardère, J.P., 1982. Effects of noise on growth and reproduction of Crangon crangon in rearing tanks. Mar. Biol. 71, 177-185.

Leonard, N., Blancheton, J.P., Guiraud, J.P., 2000. Populations of heterotrophic bacteria in an experimental recirculating aquaculture system. Aquacult. Eng. 22(1-2), 109-120.

Locascio, J.V., Mann, D.A., 2008. Diel periodicity of fish sound production in Charlotte Harbor, FL. Trans. Am. Fish. Soc. 137, 606-615.

Locher, R., Baur, B., 2002. Nutritional stress changes sex-specific reproductive allocation in the simultaneously hermaphroditic land snail Arianta arbustorum. Funct. Ecol. 16, 623-632.

Lockyear, J., Kaiser, H., Hecht, T., 1997. Studies on the captive breeding of the Knysna seahorse, Hippocampus capensis. Aq. Sci. Conserv. 1, 129-136.

Lourie, S.A., 2003. Measuring seahorses. Technical Report No. 4. Project Seahorse, British Columbia, 15pp.

Lourie, S.A., Vincent, A.C.J., Hall, H.J., 1999. Seahorses: An identification guide to the world’s species and their conservation. Project Seahorse, London.

Lovell, J.M., Findlay, M.M., Moate, R.M., Nedwell, J.R., Pegg, M.A., 2005. The inner ear morphology and hearing abilities of the Paddlefish (Polyodon spathula) and the Lake Sturgeon (Acipenser fulvescens). Comp. Biochem. Physiol. A 142, 286-296.

Lugli, M., Yan, H.Y., Fine, M.L., 2003. Acoustic communication in two freshwater gobies: The relationship between ambient noise, hearing thresholds, and sound spectrum. J. Comp. Physiol. A 189, 309-320.

Mackay, R.S., Pegg, J., 1988. Debilitation of prey by intense sounds. Mar. Mamm. Sci. 4 (4), 356-359.

Maney, D.L., Wingfield, J.C., 1998. Neuroendocrine suppression of female courtship in a wild passerine: Corticotropin-releasing factor and endogenous opioids. J. Neuroendocrinol. 10, 593-599.

Manter, H.W., 1947. The digenetic trematodes of marine fishes of Tortugas, Florida. Am. Midl. Nat. 38(2), 257-416.

Marshall, N.P., 1966. The life of fishes. World Publishing Company, Cleveland.

181

Marten, K., Herzing, D., Poole, M., Allman, K.N., 2001. The acoustic predation hypothesis: Linking underwater observations and recordings during odontocete predation and observing the effects of loud impulsive sounds on fish. Aquat. Mamm. 27 (1), 56-66.

Masonjones, H.D. Lewis, S.M., 1996. Courtship behavior in the dwarf seahorse, Hippocampus zosterae. Copeia 1996, 634-640.

Masonjones, H.D., Lewis, S.M., 2000. Differences in potential reproductive rates of male and female seahorses related to courtship roles. Anim. Behav. 59, 11-20.

Matsunaga, T., Rahman, A., 1998. What brought the adaptive immune system to vertebrates? The jaw hypothesis and the seahorse. Immunol. Rev. 166, 177-186.

Mazeaud, M.M., Mazeaud, F., 1981. Adrenergic responses to stress in fish. In: Pickering, A.D. (Ed.), Stress and Fish. Academic Press, New York, pp. 49-75.

McCauley, R.D., Fewtrell, J., Popper, A.N., 2003. High intensity anthropogenic sound damages fish ears. J. Acoust. Soc. Am. 113(1), 638-642.

McCormick, M.I., 1998. Behaviorally induced maternal stress in a fish influences progeny quality by a hormonal mechanism. Ecology 79(6), 1873-1883.

McDonald, G., Milligan, L., 1997. Ionic, osmotic, and acid-base regulation in stress. In: Iwama, G.K., Pickering, A.D., Sumpter, J.P., Schreck, C.B. (Eds.), Fish, Stress, and Health in Aquaculture. Cambridge University Press, Great Britain, pp. 119-144.

McKibben, J.R., Bass, A.H., 1998. Behavioral assessment of acoustic parameters relevant to signal recognition and preference in a vocal fish. J. Acoust. Soc. Am. 104(6), 3520-3533.

McLeay, D.J., 1973. Effects of a 12-hr and 25-day exposure to kraft pulp mill effluent on the blood and tissues of juvenile coho salmon (Oncorhynchus kisutch). J. Fish Res. Board Can. 30, 395-400.

McLeay, D.J., Brown, D.A., 1974. Growth stimulation and biochemical changes in juvenile coho salmon (Oncorhynchus kisutch) exposed to bleached kraft pulpmill effluent for 200 days. J. Fish Res. Board Can. 31, 1043-1049.

Moore, F.L., Lowry, C.A., Rose, J.D., 1994. Steroid-neuropeptide interactions that control reproductive behaviors in an amphibian. Psychoneuroendocrinol. 19(5-7), 581-592.

Morgan, J.D., Iwama, G.K., 1997. Measurements of stressed states in the field. In: Iwama, G.K., Pickering, A.D., Sumpter, J.D., Schreck, C.B., (Eds.), Stress and Health in Aquaculture. Society for Experimental Biology Series 62. Cambridge University Press, Cambridge, pp. 247-268.

Morgan, M.J., Wilson, C.E., Crim, L.W., 1999. The effect of stress on reproduction in Atlantic cod. J. Fish Biol. 54, 477-488.

182

Montero, D., Izquierdo, M.S., Tort, L., Robaina, L., Vergara, J.M., 1999. High stocking density produces crowding stress altering some physiological and biochemical parameters in gilthead seabream, Sparus aurata, juveniles. Fish Physiol. Biochem. 20, 53-60.

Montero, D., Robaina, L.E., Socorro, J., Vergara, J.M., Tort, L., Izquierdo, M.S., 2001. Alteration of liver and muscle fatty acid composition in gilthead seabream (Sparus aurata) juveniles held at high stocking density and fed an essential fatty acid deficient diet. Fish Physiol. Biochem. 24, 63-72.

Motomatsu, K., Hiraishi, T., Yamamoto, K., Nashimoto, K., 1998. Auditory masking by ambient noise in black rockfish Sebastes schlegeli. Nippon Suisan Gakk. 64(5), 792-795.

Myrberg, A.A. Jr., 2001. The acoustical biology of elasmobranchs. Env. Biol. Fish. 60, 31-45.

Myrberg, A.A., Riggio, R.J., 1985. Acoustically mediated individual recognition by a coral reef fish (Pomacentrus partitus). Anim. Behav. 33, 411-416.

Myrberg, A.A. Jr., Spires, J.Y., 1972. Sound discrimination by the bicolor damselfish, Eupomacentrus partitus. J. Exp. Biol. 57, 727-735.

Myrberg, A.A. Jr., Ha, S.J., Shamblott, M.J., 1993. The sounds of bicolor damselfish (Pomacentrus partitus): Predictors of body size and a spectral basis for individual recognition and assessment. J. Acoust. Soc. Am. 94(6), 3067-3070.

Myrberg, A.A. Jr., Mohler, M., Catala, J.D., 1986. Sound production by males of a coral reef fish (Pomacentrus partitus): Its significance to females. Anim. Behav. 34, 913-923.

Norcross, J.L., Newman, J.D., 1999. Effects of separation and novelty on distress vocalizations and cortisol in the common marmoset (Callithrix jacchus). Am. J. Primatol. 47, 209-222.

Opstad, I., Bergh, Ø., Skiftesvik, A.B., 1998. Large-scale rearing of Atlantic halibut, Hippoglossus hippoglossus L., yolk sac larvae: Effects of flow rate on growth, survival, and accumulation of bacteria. Aquacult. Res. 29(12), 893-898.

Orlando, E.F., Guillette, L.J.Jr., 2001. A re-examination of variation associated with environmentally stressed organisms. Hum. Reprod. Update 7(3), 265-272.

Øverli, Ø., Sørensen, C., Pulman, K.G.T., Pottinger, T.G., Korzan, W., Summers, C.H., Nilsson, G.E., 2007. Evolutionary background for stress-coping styles: Relationships between physiological, behavioral, and cognitive traits in non-mammalian vertebrates. Neurosci. Biobehav. Rev. 31, 396-412.

Norris, K.S., Møhl, B., 1983. Can odontocetes debilitate prey with sound? Am. Nat. 122 (1), 85-104.

183

Pankhurst, N.W., Van Der Kraak, G., 1997. Effects of stress on reproduction and growth of fish. In: Iwama, G.K., Pickering, A.D., Sumpter, J.P., Schreck, C.B. (Eds.), Fish Stress and Health in Aquaculture. Society for Experimental Biology Seminar Series 62, Cambridge University Press, New York, pp. 73-93.

Papoutsoglou, S.E., Karakatsouli, N., Chiras, G., 2005. Dietary l-tryptophan and tank colour effects on growth performance of rainbow trout (Oncorhynchus mykiss) juveniles reared in a recirculating water system. Aquacult. Eng. 32, 277-284.

Papoutsoglou, S.E., Miliou, H., Chadio, S., Karakatsouli, N., Zarkada, A., 1999. Studies on stress responses and recovery from removal in gilthead sea bream Sparus aurata (L.) using recirculated seawater system. Aquacult. Eng. 21, 19-32.

Papoutsoglou, S.E., Mylonakis, G., Miliou, H., Karakatsouli, N.P., Chadio, S., 2000. Effects of background color on growth performances and physiological responses of scaled carp (Cyprinus carpio L.) reared in a closed circulated system. Aquacult. Eng. 22, 309-318.

Parvulescu, A., 1964. Problems of propagation and processing. In: Tavolga, W.N. (Ed.), Marine Bio-Acoustics. Pergamon Press, New York, pp. 87-100.

Parvulescu, A., 1967. The acoustics of small tanks. In: Tavolga, W.N. (Ed.), Marine Bio-Acoustics (Vol. 2). Pergamon Press, New York, pp. 7-13.

Patton, Z.J., Krebs, R.A., 2001. The effect of thermal stress on the mating behavior of three Drosophila species. Phsyiol. Biochem. Zool. 74(6), 783-788.

Pearson, W.H., Skalski, J.R., Malme, C.I., 1992. Effects of sounds from a geophysical survey device on behavior of captive rockfish (Sebastes spp.). Can. J. Fish. Aquat. Sci. 49, 1343-1356.

Pickering, A.D., Pottinger, T.G., 1987. Crowding causes prolonged leucopenia in salmonid fish, despite interrenal acclimation. J. Fish Biol. 30, 701-712.

Pickering, A.D., Pottinger, T.G., Carragher, J., Sumpter, J.P., 1987. The effects of acute and chronic stress on the levels of reproductive hormones in the plasma of mature male brown trout, Salmo trutta L. Gen. Comp. Endocrinol. 68, 249-259.

Picton, T.W., Stapells, D.R., Campbell, K.B., 1981. Auditory evoked potentials from the human cochlea and brainstem. J. Otolaryngol. Suppl. 9 10, 1-41.

Pierce, B.N., Hemsworth, P.H., Rivalland, E.T.A., Waenmaker, E.R., Morrissey, A.D., Papargiris, M.M., Clarke, I.J., Karsch, F.J., Turner, A.I., Tilbrook, A.J., 2008. Psychosocial stress suppresses attractivity, proceptivity and pulsatile LH secretion in the ewe. Horm. Behav. 54, 424-434.

Pillay, T.V.R., 1990. Aquaculture: Principles and practices. Fishing News Books, Oxford, 575pp.

184

Popper, A.N., Carlson, T.J., 1998. Application of sound and other stimuli to control fish behavior. Trans. Am. Fish. Soc. 127(5), 673-707.

Popper, A.N., Clarke, N.L., 1976. The auditory system of the goldfish (Carassius auratus): Effects of intense acoustic stimulation. Comp. Biochem. Physiol. 53A, 11-18.

Popper, A.N., Fay, R.R., 1999. The auditory periphery in fishes. In: Fay, R.R., Popper, A.N. (Eds.), Comparative Hearing: Fish and Amphibians. Springer, New York, pp. 43-100.

Popper, A.N., Fay, R.R., Platt, C., Sand, O., 2003. Sound detection mechanisms and capabilities of teleost fishes. In: Collin, S.P., Marshall, N.J. (Eds.), Sensory Processing in Aquatic Environments. Springer-Verlag, New York, pp. 3-38.

Porter, C.M., Janz, D.M., 2003. Treated municipal sewage discharge affects multiple levels of biological organization in fish. Ecotox. Environ. Safe 54, 199-206.

Pottinger, T.G., Carrick, T.R., Yeomans, W.E., 2002. The three-spined stickleback as an environmental sentinel: Effects of stressors on whole-body physiological indices. J. Fish Biol. 61, 207-229.

Precht, H., 1958. Concepts of the temperature adaptation of unchanging reaction systems of cold-blooded animals. In: Prosser, C.L. (Ed.), Physiological Adaptations. Lord Baltimore Press, Inc. Washington, pp. 50-78.

Project Seahorse, 2009. http://seahorse.fisheries.ubc.ca, accessed May 2009.

Rasa, O.A.E., 1971. The causal factors and function of ‘yawning’ in Microspathodon chrysurus (Pisces: Pomacentridae). Behaviour 39(1), 39-57.

Regnault, M., Lagardère, J-P., 1983. Effects of ambient noise on the metabolic level of Crangon crangon (Decapoda, Natantia). Mar. Ecol. Prog. Ser. 1, 71-78.

Retana-Marquez, S., Dominguez Salazar, E., Velazquez-Moctezuma, J., 1996. Effect of acute and chronic stress on masculine sexual behavior in the rat. Psychoneuroendocrino. 21(1), 39-50.

Rose, J.D., 2000. Corticosteroid actions from neuronal membrane to behavior: Neurophysiological mechanisms underlying rapid behavioral effects of corticosterone. Biochem. Cell Biol. 78, 307-315.

Rose, J.D., Moore, F.L., 1999. A neurobehavioral model for rapid actions of corticosterone on sensorimotor integration. Steroids 64, 92-99.

Ruane, N.M., Carballo, E.C., Komen, J., 2002. Increased stocking density influences the acute physiological stress response of common carp Cyprinus carpio (L.). Aquacult. Res. 33, 777-784.

Ruane, N.M., Komen, H., 2003. Measuring cortisol in the water as an indicator of stress caused by increased loading density in common carp (Cyprinus carpio). Aquaculture 218, 685-693.

http://seahorse.fisheries.ubc.ca/�

185

Sala-Rabanal, M., Sánchez, J., Ibarz, A., Fernández-Borrás, J., Blasco, J., Gallardo, M.A., 2003. Effects of low temperatures and fasting on hematology and plasma composition of gilthead sea bream (Sparus aurata). Fish Physiol. Biochem. 29, 105-115.

Sand, O., 1984. Lateral-line systems. In: Bolis, L., Keynes, R.D., Maddrell, S.H.P. (Eds.), Comparative Phyiology of Sensory Systems. Cambridge University Press, Cambridge, pp. 3-32.

Sand, O., Enger, P.S., 1973. Evidence for an auditory function of the swimbladder in the cod. J. Exp. Biol. 59, 405-414.

Santulli, A., Modica, A., Messina, C., Ceffa, L., Curatolo, A., Rivas, G., Fabi, G., D’Amelio, V., 1999. Biochemical responses of European sea bass (Dicentrarchus labrax L.) to the stress induced by offshore experimental seismic prospecting. Mar. Poll. Bull. 38(12), 1105-1114.

Scholik, A.R., Yan, H.Y., 2001a. Effects of underwater noise on auditory sensitivity of a cyprinid fish. Hear. Res. 152, 17-24.

Scholik, A.R., Yan, H.Y., 2001b. The effects of underwater noise on auditory sensitivity of fish. Proc. Inst. Acoust. UK, 23(4), 27-36.

Scholik, A.R., Yan, H.Y., 2002a. The effects of noise on the auditory sensitivity of the bluegill sunfish, Lepomis macrochirus. Comp. Biochem. Physiol. A 133, 43-52.

Scholik, A.R., Yan, H.Y., 2002b. Effects of boat engine noise on the auditory sensitivity of the fathead minnow, Pimephales promelas. Env. Biol. Fish. 63, 203-209.

Schreck, C.B., 1981. Stress and compensation in teleostean fishes: Response to social and physical factors. In: Pickering, A.D. (Ed.), Stress and Fish. Academic Press, New York, 295-321.

Schreck, C.B., 1996. Immunomodulation: Endogenous factors. In: Iwama, G., Nakanishi, T. (Eds.), The Fish Immune System: Organism, Pathogen, and Environment. Academic Press, New York, pp. 311-337.

Schreck, C.B., Olla, B.L., Davis, M.W., 1997. Behavioral responses to stress. In: Iwama, G.K., Pickering, A.D., Sumpter, J.P., Schreck, C.B. (Eds.), Fish Stress and Health in Aquaculture. Cambridge University Press, Great Britain, pp. 145-170.

Schroeder, L.L., Kramer, S.J., 1989. The Very Basics of ABR. The Interstate Printers and Publishers, Inc., Danville, IL, 41pp.

Selye, H., 1950. Stress and the general adaptation syndrome. Brit. Med. J. June 17, 1950, 1383-1392.

Shepherd, S., 2002. Seahorses and education in aquaria. In: Bull, C.D., Mitchell, J.S. (Eds.), Seahorse Husbandry in Public Aquaria: 2002 Manual. Project Seahorse and the John G. Shedd Aquarium, Chicago, pp. 11-12.

186

Simon, M., Wahlberg, M., Ugarte, F., Miller, L.A., 2005. Acoustic characteristics of underwater tail slaps used by Norwegian and Icelandic killer whales (Orcinus orca) to debilitate herring (Clupea harengus). J. Exp. Biol. 208, 2459-2466.

Skalski, J.R., Pearson, W.H., Malme, C.I., 1992. Effects of sounds from a geophysical survey device on catch-per-unit effort in a hook-and-line fishery for rockfish (Sebastes spp.). Can. J. Fish. Aquat. Sci. 49, 1357-1365.

Sloman, K.A., Metcalfe, N.B., Taylor, A.C., Gilmour, K.M., 2001. Plasma cortisol concentrations before and after social stress in rainbow trout and brown trout. Physiol. Biochem. Zool. 74(3), 383-389.

Smith, M.E., Kane, A.S., Popper, A.N., 2004. Noise-induced stress response and hearing loss in goldfish (Carassius auratus). J. Exp. Biol. 207, 427-435.

Spanier, E., 1979. Aspects of species recognition by sound in four species of damselfishes, Genus Eupomacentrus (Pisces: Pomacentridae). Z. Tierpsychol. 51, 301-316.

Stabentheiner, A., 1988. Correlation between hearing and sound production in piranhas. J. Comp. Physiol. A 162, 67-76.

Stahl, C.E., Redei, E., Wang, Y., Borlongan, C.V., 2002. Behavioral, hormonal and histological stress markers of anxiety-separation in postnatal rats are reduced by prepro-thyrotropin-releasing hormone 178-199. Neurosci. Lett. 321, 85-89.

Sterling, P., Eyer, J., 1988. Allostasis: A New Paradigm to Explain Arousal Pathology. In: Fisher, S., Reason, J. (Eds.), Handbook of Life Stress, Cognition and Health. John Wiley & Sons Ltd., New York, pp. 629-649.

Stout, J.F., 1975. Sound communication during the reproductive behavior of Notropis analostanus (Pisces: Cyprinidae). Am. Mid. Nat. 94(2), 296-325.

Sumpter, J.P., 1997. The endocrinology of stress. In: Iwama, G.K., Pickering, A.D., Sumpter, J.P., Schreck, C.B. (Eds.), Fish Stress and Health in Aquaculture. Cambridge University Press, Great Britain, pp. 95-118.

Sverdrup, A., Kjellsby, E., Kruger, P.G., Floysand, R., Knudsen, F.R., Enger, P.S., Serck-Hanssen, G., Helle, K.B., 1994. Effects of experimental seismic shock on vasoactivity of arteries, integrity of the vascular endothelium and on primary stress hormones of the Atlantic salmon. J. Fish Biol. 45, 973-995.

Svobodová, Z., Vykusová, B., Modrá, H., Jarkovský, J., Smutná, M., 2006. Haematological and biochemical profile of harvest-size carp during harvest and post-harvest storage. Aquacult. Res. 37, 959-965.

Tavolga, W.N., 1958. The significance of underwater sounds produced by males of the gobiid fish, Bathygobius soporator. Physiol. Zool. 31(4), 259-271.

187

Tavolga, W.N., 1967. Masked auditory thresholds in teleost fishes. In: Tavolga, W.N. (Ed.), Marine Bio-Acoustics (Vol. 2). Pergamon Press, New York, pp. 233-245.

Tavolga, W.N., 1974. Signal/noise ratio and the critical band in fishes. J. Acoust. Soc. Am. 55(6), 1323-1333.

Tavolga, W.N., Wodinsky, J., 1963. Auditory capacities in fishes. Pure tone thresholds in nine species of marine teleosts. Bull. Am. Mus. Nat. Hist. 126, 177-240.

Teixeira, R.L., Musick, J.A., 2001. Reproduction and food habits of the lined seahorse, Hippocampus erectus (Teleostei: Syngnathidae) of Chesapeake Bay, Virginia. Rev. Braz. Biol. 61, 79-90.

The SAS Institute, 2006. The GLIMMIX Procedure, June 2006. The SAS Institute, Cary, NC, 255pp.

Tidwell, J.H., Webster, C.D., Coyle, S.D., 1996. Effects of dietary protein level on second year growth and water quality for largemouth bass (Micropterus salmoides) raised in ponds. Aquacult. 145 (1-4), 213-223.

Urick, R.J., 1975. Principles of Underwater Sound. McGraw-Hill Book Company, New York, 384pp.

van den Berg, A.V., Schuijf, A., 1983. Discrimination of sounds based on the phase difference between particle motion and acoustic pressure in the shark Chiloscyllium griseum. Proc. R. Soc. Lond. B 218, 127-134.

Van Weerd, J.H., Komen, J., 1998. The effects of chronic stress on growth in fish: A critical appraisal. Comp. Biochem. Physiol. A 120, 107-112.

Versluis, M., Schmitz, B., von der Heydt, A., Lohse, D., 2000. How snapping shrimp snap: Through cavitating bubbles. Science 289, 2114-2117.

Vijayan, M.M., Leatherland, J.F., 1989. Cortisol-induced changes in plasma glucose, protein, and thyroid hormone levels, and liver glycogen content of coho salmon (Oncorhynchus kisutch Walbaum). Can. J. Zool. 67(11), 2746-2750.

Vincent, A.C.J., 1994a. Operational sex ratios in seahorses. Behav. 128, 153-167.

Vincent, A.C.J., 1994b. Seahorses exhibit conventional sex roles in mating competition, despite male pregnancy. Behav. 128, 135-151.

Vincent, A.C.J., 1995. A role for daily greetings in maintaining seahorse pair bonds. Anim. Behav. 49, 258-260.

Vincent, A.C.J., 1998. Seahorse culturing. Austasia Aquacult. 11(5), 72-73.

188

Vincent, A.C.J., Clifton-Hadley, R.S., 1989. Parasitic infection of the seahorse (Hippocampus erectus)—A case report. J. Wildlife Dis. 25(3), 404-406.

Vincent, A.C.J., Sadler, L.M., 1995. Faithful pair bonds in wild seahorses, Hippocampus whitei. Anim. Behav. 50, 1557-1569.

Vincent, A.C.J., Ahnesjö, I., Berglund, A., Rosenqvist, G., 1992. Pipefishes and seahorses: Are they all sex role reversed? Trends Ecol. Evol. 7(7), 237-241.

Watanabe, T., 1988. Nutrition and growth. In: Shepherd, C.J., Bromage, N.R. (Eds.), Intensive Fish Farming. BSP Professional Books, Oxford, pp. 154-197.

Wendelaar Bonga, S.E., 1997. The stress response in fish. Physiol. Rev. 77(3), 591-626.

Wenz, G.M., 1962. Acoustic ambient noise in the ocean: Spectra and sources. J. Acoust. Soc. Am. 34(12), 1936-1956.

Wersinger, S.R., Martin, L.B., 2009. Optimization of laboratory conditions for the study of social behavior. ILAR 50(1), 64-80.

Wilson, M.J., Vincent, A.C.J., 1998. Preliminary success in closing the life cycle of exploited seahorse species, Hippocampus spp., in captivity. Aq. Sci. Cons. 2, 179-196.

Wolski, L.F., Anderson, R.C., Bowles, A.E., Yochem, P.K., 2003. Measuring hearing in the harbor seal (Phoca vitulina): Comparison of behavioral and auditory brainstem response techniques. J. Acoust. Soc. Am. 113(1), 629-637.

Woods, C.M.C., 2000. Preliminary observations on breeding and rearing the seahorse Hippocampus abdominalis (Teleostei: Syngnathidae) in captivity. New Zeal. J. Mar. Fresh. Res. 34, 475-485.

Woods, C.M.C., 2003. Factors affecting successful culture of the seahorse, Hippocampus abdominalis Leeson, 1827. In: Cato, J.C., Brown, C.L. (Eds.), Marine Ornamental Species: Collection, Culture, and Conservation. Iowa State Press, Ames, pp. 277-288.

Wysocki, L.E., Ladich, F., 2001. The ontogenetic development of auditory sensitivity, vocalization, and acoustic communication in the labyrinth fish Trichopsis vittata. J. Comp. Physiol. A 187, 177-187.

Wysocki, L.E., Ladich, F., 2005. Hearing in fishes under noise conditions. J. Assoc. Res. Otolaryngol. 6, 28-36.

Wysocki, L.E., Davidson, J.W.III., Smith, M.E., Frankel, A.S., Ellison, W.T., Mazik, P.M., Popper, A.N., Bebak, J., 2007. Effects of aquaculture production noise on hearing, growth, and disease resistance of rainbow trout Oncorhynchus mykiss. Aquacult. 272, 687-697.

Wysocki, L.E., Dittami, J.P., Ladich, F., 2006. Ship noise and cortisol secretion in European freshwater fishes. Biol. Conserv. 128, 501-508.

189

Yager, D.D., 1992. Underwater acoustic communication in the African pipid frog Xenopus borealis. Bioacoust. 4, 1-24.

Yan, H.Y., 2001. A non-invasive electrophysiological study on the enhancement of hearing ability in fishes. Proc. Inst. Acoust. UK 23(2), 15-25.

Yan, H.Y., Fine, M.L., Horn, N.S., Colón, W.E., 2000. Variability in the role of the gasbladder in fish audition. J. Comp. Physiol. A 186, 435-445.

Yan, H.Y., Popper, A.N., 1993. Acoustic intensity discrimination by the cichlid fish Astronotus ocellatus (Cuvier). J. Comp. Physiol. A 173, 347-351.

Yost, W.A., 1994. Fundamentals of Hearing: An Introduction. 3rd Ed. Academic Press, San Diego, 326pp.

Yost, W.A., 2000. Fundamentals of Hearing: An Introduction. 4th Ed. Academic Press, San Diego, 349pp.

Zagaeski, M., 1987. Some observations on the prey stunning hypothesis. Mar. Mamm. Sci. 3 (3), 275-279.

Zar, J.H., 1974. Biostatistical Analysis. Prentice-Hall, Inc., Englewood Cliffs, 620pp.

Zelick, R., Mann, D.A., Popper, A.N., 1999. Acoustic communication in fishes and frogs. In: Fay, R.R., Popper, A.N. (Eds.), Comparative Hearing: Fish and Amphibians. Springer-Verlag, New York, pp. 363-411.

190

BIOGRAPHICAL SKETCH

Paul August Anderson was born and raised in Massachusetts. At an early age, he

developed a love for the marine world and became practiced in the hobby of marine aquarium

keeping. His fascination led him to pursue an undergraduate education in marine science at

Eckerd College in St. Petersburg, FL. While there, Paul had the opportunity to explore many

facets of marine science, conducting experiments and activities in nitrogen cycling in marine

aquaria, marine plant physiology, marine mammal photo identification, and his undergraduate

thesis in coral reef ecology where he examined and characterized a commensal relationship

between territorial damselfishes and mysid schools on the reefs of Belize, Central America.

After graduating college, Paul plunged back into his first love of aquaria and worked as an

aquarist and educator at Mote Marine Aquarium in Sarasota, FL, and subsequently as a lead

aquarist and education coordinator for The Pier Aquarium in St. Petersburg, FL. The additional

skills he learned in these positions in animal care and project management prepared him to

pursue graduate studies at the University of Florida, that gave him the opportunity to explore an

interesting question in the aquarium hobby and in aquaculture from a fish’s perspective: what is

the umwelt of a fish in an aquarium world?

Most recently, Paul has accepted the position of Conservation and Research Coordinator

for The Florida Aquarium’s Center for Conservation in Tampa, FL. It is a thoroughly rewarding

career opportunity for Paul; one in which his wide-ranging teaching and research background is

employed to advance the diverse and exciting marine research and conservation initiatives of the

Center.

© 2009 paul august andersonufdcimages.uflib.ufl.edu/uf/e0/02/47/76/00001/anderson_p.pdf ·...

Documents