USF Libraries
USF Digital Collections

Sound localization abilities of two Florida manatees, trichechus manatus latirostris

MISSING IMAGE

Material Information

Title:
Sound localization abilities of two Florida manatees, trichechus manatus latirostris
Physical Description:
Book
Language:
English
Creator:
Colbert, Debborah E
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Audition
Sirenian
Hearing
West Indian manatee
Localization
Dissertations, Academic -- Psychology -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Florida manatees (Trichechus manatus latirostris) live in the shallow, often turbid inland and coastal waters of the southeastern United States. Since their vision is poor (Bauer et al., 2003), other senses probably guide orientation. Previous studies have found that manatees can hear over 40 kHz (Gerstein et al., 1999) and have the capacity for rapid auditory temporal processing (Mann et al., 2005). However, it is not known if manatees have the ability to localize underwater sounds. Two Florida manatees were trained to identify underwater sound source locations using a four-choice discrimination paradigm. Three broad-band signals ( 0.2 - 20, 6 - 20, and 0.2 2kHz) were tested at four durations (3,000, 1,000, 500, and 200ms) and two tonal signals (4 and 16kHz) were tested with a 3,000ms duration. A total of 1,008 test trials were analyzed per subject.Both manatees learned the task easily, and could localize all of the test signals at a performance rate well above the 25% chance level. Within all of the broad-band conditions, performance accuracy ranged from 93% - 79% for Buffett, and 93% - 51% for Hugh. Broad-band signal duration did not have an effect on performance accuracy with Buffett who ranged from 89% to 87%, but did with Hugh who ranged from 87% - 58%. Broad-band frequency type did not have an effect on performance accuracy with Buffett who averaged 90%, 86%, and 89%, but may have with Hugh who averaged from 76%, 68%, and 65% at the 0.2 20, 6 20, and 0.2 2 kHz conditions. Both animals performed above chance levels with the pure tone signals, but at a much lower accuracy rate with Hugh at 49% and 32% and Buffett at 44% and 33% with the 4 kHz and 16 kHz conditions.Results from this experiment provide information about the manatees ability to localize different types of sounds in a controlled environment. This knowledge is important for understanding how manatees detect and localize noise generated from conspecifics and boat engines and contributes to making competent conservation management decisions about these endangered marine mammals.
Thesis:
Thesis (M.A.)--University of South Florida, 2005.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Debborah E. Colbert.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 90 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001709499
oclc - 68570302
usfldc doi - E14-SFE0001361
usfldc handle - e14.1361
System ID:
SFS0025681:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Sound Localization Abilities of Two Florida Manatees, Trichechus manatus latirostris by Debborah E. Colbert A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Psychology College of Arts and Sciences University of South Florida Major Professor: Sarah Partan, Ph.D Toru Shimizu, Ph.D Steven Stark, Ph.D David Mann, Ph.D Gordon B. Bauer, Ph.D Date of Approval: October 11, 2005 Keywords: audition, Sirenian, hearing, West Indian manatee, localization Copyright 2005, Debborah E. Colbert

PAGE 2

Dedication It is with great pleasure that I dedicate this thesis to the five people who have touched my life in so many wonderful ways my husband Larry, my daughters Katie, Alyssa, and Lauren, and my mother-in-law Mary. To Larry. You have been my partner and best friend for over 23 incredible years. You have stood by me in both times of joy a nd frustration, and have even undertaken the monumental tasks of grocery shopping and c ooking in the last several months! You always make me proud in even the littles t things you do, and I consider myself the luckiest woman in the worl d to be married to you. To my daughters. Katie, you are an am azingly caring, sensitive, and intelligent woman, who has been organized since birth, an d who will most likely keep all of us on the right path for years to come. Alyssa, you are a passionate, independent, and brilliant young lady, who will probably always take the path less travel ed, but oh all the wonderful things you will experience. La uren, you are an incredibly happy, multitalented, extremely smart young lady who be gan singing at two and never stopped, who will no doubt impress the world with your gifts. You all have given me so much joy and made me so proud; I can’t wait to see what you offer the world! To Mary. You have been such an importa nt part of my life and have given so much of yourself. I have learned so much from your car ing gentle demeanor, and I consider myself blessed to have you in my life. I am indeed, a rich woman.

PAGE 3

Acknowledgements This thesis could not have been comple ted without the input and assistance of numerous people. The faculty serving on my committee has been instrumental in the development and completion of this study and I would like to thank my major professor, Dr. Sarah Partan as well as Dr. Toru Shimiz u, and Dr. Steven Stark for their valuable input and unending assistance. I would like to extend a special thank you to committee member Dr. David Mann for working so diligently with me with the computer programming and signal generation issues that arose. You spent many of hours with me at the lab and in your office going over ev erything and your assistance was greatly appreciated. I would also like to particular ly acknowledge Dr. Gordon Bauer. I have worked with Dr. Bauer on various project s for over 8 years now and am proud to consider him my mentor in so many ways Thank you for all of your guidance and patience. This project could not have happened wit hout the assistance of my co-workers at Mote Marine Laboratory, Joseph Gaspard and Ki mberly Dziuk. Kim was instrumental in keeping the whole computer system organi zed throughout testing and has brought a ray of sunshine into our office. I cannot say enough about Joe (no really, what ever I say, he’ll say it wasn’t enough). Joe’s work ethic is incredible and he is a major reason that this project was successful, I am proud to have worked with him. I would also like to acknowledge our wonderful volunt eer trainer Jann Warfield w ho has worked with us for over 6 years, the interns involved, Bethany Augliere and Emily Kane, and a student trainer from New College, Sara Stamper.

PAGE 4

I would like to acknowledge the United States Fish and Wildlife Service for granting the permit. Finally I would like to e xpress my gratitude to those that funded this study, the Avoidance Technology Grant from the Florida Wildlife Res earch Institute, the University of Florida, and the Thurell Family.

PAGE 5

i Table of Contents List of Tables................................................................................................................. ....iv List of Figures................................................................................................................ .....v Abstract....................................................................................................................... .....viii Introduction................................................................................................................... ......1 The Florida Manatee.......................................................................................................1 Manatee Sensory Systems...............................................................................................2 Anatomical and Physiological Studies........................................................................2 Behavioral Studies......................................................................................................4 Sound Localization.........................................................................................................6 In-Air vs. In-Water Acoustic Properties.........................................................................8 Florida Manatees and E nvironmental Noise.................................................................10 Hypotheses....................................................................................................................1 2 Methods........................................................................................................................ .....14 Subjects....................................................................................................................... ..14 Overview....................................................................................................................... 15 Experimental Design.................................................................................................15 Signal Generation......................................................................................................16 Computer and Programming.....................................................................................16

PAGE 6

ii Personnel:..................................................................................................................18 Training Procedures......................................................................................................19 Global Procedures.....................................................................................................19 Phase I: East Wall of the Shelf.................................................................................20 Stationing..............................................................................................................20 Speaker Selection..................................................................................................23 Phase II: Center of the Shelf:....................................................................................27 Testing Procedures........................................................................................................31 Experimental Conditions..........................................................................................31 Personnel Responsibilities........................................................................................33 Data Recording.........................................................................................................34 Environmental Conditions........................................................................................35 Controls.....................................................................................................................36 Signal and Speaker Artifacts.................................................................................36 Trainer Cues:.........................................................................................................36 Motivational Effects..............................................................................................37 Block Criteria........................................................................................................37 Testing Trial Sequence.............................................................................................38 Results:....................................................................................................................... .......40 Performance by Speaker Location................................................................................40 Speaker Calibration.......................................................................................................42 Overall Performance Accuracy.....................................................................................46 Learning by Blocks.......................................................................................................47

PAGE 7

iii Error Distribution..........................................................................................................49 Discussion..................................................................................................................... ....53 Overall Performance.....................................................................................................53 Speaker Artifacts...........................................................................................................55 Learning by Blocks.......................................................................................................56 Error Distribution..........................................................................................................57 Relevance...................................................................................................................... 57 References..................................................................................................................... ....61 Appendices..................................................................................................................... ...69 Appendix A:..................................................................................................................70 The Computer Protocols Used For all Phases of the Experimental Conditions...70 Appendix B:..................................................................................................................73 Data-recording Protocols Used to Docume nt Each Session on a Tank-side Data Sheet......................................................................................................................73 Appendix C:..................................................................................................................75 Error distribution within th e 12 broad-band conditions........................................75

PAGE 8

iv List of Tables Table 1 Conditions tested included three br oad-band signals at four durations and two tonal signals at one duration...........................................................................32 Table 2 Overall Accuracy Performance per Subject by Frequency and Duration Conditions............................................................................................................................47

PAGE 9

v List of Figures Figure 1. Diagram of the 265,000 L manatee e xhibit composed of a Medical Pool, Shelf Area, and Exhibit Area...........................................................................15 Figure 2. Electronic button box used to run the sessions and automatically download each trial into a digital Excel file....................................................18 Figure 3. Training configuration for the east wall of the Shelf Area................................21 Figure 4. Stationing apparatus used in Phase I of the training..........................................22 Figure 5. Speaker holder with attached underwater speaker............................................24 Figure 6. Power spectra (top) and spect rograms (bottom) of the secondary reinforcement signals.......................................................................................25 Figure 7. Training configuration for the east wall of the Shelf Area with two speaker locations..............................................................................................26 Figure 8. Training configuration for the center of the Shelf Area with three speakers represented as the black circles.........................................................28 Figure 9. Training configuration for the cen ter of the Shelf Area with four test speakers, located 105cm from the cen ter of the stationing bar and .75m below the surface.............................................................................................30 Figure 10. The data entry screen used to en ter all of the session’s information into the Access database..........................................................................................35 Figure 11. A comparison of the average performance accuracy on Hugh’s first three test blocks when the test speak ers were in their original locations

PAGE 10

vi and last three test blocks after th e speakers had been moved to their new locations...................................................................................................41 Figure 12. A comparison of the average pe rformance accuracy on Buffett’s first three test blocks when the test speak ers were in their original locations and last three test blocks after th e speakers had been moved to their new locations...................................................................................................42 Figure 13. A power spectra comparison of the 6 20 kHz broadband test signal from each of the four locations when the speakers were in their original positions and when they had been re-located.....................................43 Figure 14. A power spectra comparison of the 0.2 20 kHz broadband test signal from each of the four speakers when they were in their original positions and when they had been re-located..................................................44 Figure 15. Spectrogram comparison of the three broadband test signals played at the 3000 ms duration (sample rate of 97,656 Hz)............................................45 Figure 16. Spectrogram comparison of the tw o pure tone test signals played at the 3000 ms duration..............................................................................................46 Figure 17. Percent correct by duration for each block as testing progressed across each frequency stimulus...................................................................................48 Figure 18. Overall percent correct and dist ribution of errors using only the results from testing with the broad-band signals.........................................................49 Figure 19. Percent correct and dist ribution of errors by frequency..................................51 Figure 20. Percent correct and distribution of errors by duration using only the broadband signals……………………...…………………………..…………………52

PAGE 11

vii Figure A1. The graphical user interface scr een (programmed in Visual C) used to setup the experimental conditions and automatically download the results into an Excel file during the testing sessions........................................71 Figure B1. The tank-side data-recording sheet used to document each session...............74 Figure C1. Percent correct and distributi on of errors by durat ion within the 0.2 20 kHz condition..............................................................................................76 Figure C2. Percent correct and distributi on of errors by durati on within the 6 20 kHz condition...................................................................................................76 Figure C3. Percent correct and distributi on of errors by durati on within the 0.2 2 kHz condition...................................................................................................77

PAGE 12

viii Sound Localization Abilities of Two Florida Manatees, Trichechus manatus latirostris Debborah E. Colbert ABSTRACT Florida manatees ( Trichechus manatus latirostris ) live in the shallow, often turbid inland and coastal waters of the southeastern United States. Sin ce their vision is poor (Bauer et al., 2003), other senses probably guide orientation. Previ ous studies have found that manatees can hear over 40 kHz (Gerstein et al., 1999) and have the capacity for rapid auditory temporal processing (M ann et al., 2005). However, it is not known if manatees have the ability to loca lize underwater sounds. Two Florida manatees were trained to identify underwater sound source locations using a four-choice discrimination paradigm. Three broad-band signals ( 0.2 20, 6 20, and 0.2 – 2kHz) were tested at four durations (3,000, 1,000, 500, and 200ms) and two tonal signals (4 and 16kHz) were test ed with a 3,000ms duration. A total of 1,008 test trials were analyzed per subject. Both manatees learned the task easily, and could localize al l of the test signals at a performance rate well above the 25% chance level. Within all of the broa d-band conditions, performance accuracy ranged from 93% 79% for Buffett, and 93% 51% for Hugh. Broad-band signal duration did not have an effect on performance accuracy wi th Buffett who ranged from 89% to 87%, but did with Hugh who ranged from 87% 58% Broad-band fre quency type did not have an effect on performance accuracy with Buffett who averaged 90%, 86%, and 89%,

PAGE 13

ix but may have with Hugh who averaged from 76%, 68%, and 65% at the 0.2 – 20, 6 – 20, and 0.2 – 2 kHz conditions. Both animals performed above chance levels with the pure tone signals, but at a much lower accuracy rate with Hugh at 49% and 32% and Buffett at 44% and 33% with the 4 kHz and 16 kHz conditions. Results from this experiment provide in formation about the manatees’ ability to localize different types of sounds in a c ontrolled environment. This knowledge is important for understanding how manatees de tect and localize noise generated from conspecifics and boat engines and contri butes to making competent conservation management decisions about these endangered marine mammals.

PAGE 14

1 Introduction The Florida Manatee The Family Trichechidae is composed of the Amazonian manatee ( Trichechus inunguis ), the West African manatee ( Trichechus senegalensis ), and the West Indian manatee ( Trichechus manatus ). All three species have a long, spindle-shaped body that is dark gray and covered with evenly dist ributed hairs. Their agile lips are densely covered with thicker vibrissae. They lack hind limbs but have a large paddle-shaped tail to propel them and flexible front pectoral flippers. The West Indian manatee is divided into two sub-species, the Antillean manatee ( Trichechus manatus manatus ) and the Florida manatee ( Trichechus manatus latirostris ) (Domning and Hayek, 1986). The Florida manatee is the species that this paper focuses upon and will be discussed in greater detail. It is typically found in the coastal waterways surrounding the peninsula of Florida, but can ra nge as far north as Virginia and as far west as Louisiana. It lives in turbid saltw ater habitats in the summer when these waters are warm and grazes primarily on sea grass (Reynolds and Odell, 1991). In the colder months, it needs to migrate to warmer water habitats, such as freshwater springs and power plant discharge sites (Reynolds and Wilcox, 1986) where it feeds primarily on water hyacinth, hydrilla, and other freshwater vegetation. The Florid a manatee has been referred to as a “semi-social” species (Re ynolds, 1979). They are often found grazing or traveling alone, although females with calves can be found together and large numbers of males are found with an estrous female. The Florida manatee is very similar to the

PAGE 15

2 Antillean manatee in appearance and size. It is threatened by naturally occurring events like cold stress and red tide, as well as human -influenced events such as boat strikes, canal lock compression, and habitat degradation (Odell et al., 1979). The Florida manatee is protected by both the Marine Mammal Protection Act (1972) and the Endangered Species Act ( 1973). The January 2005 synoptic survey estimated the endangered Florida manatee population to be 3,142 (Florida Fish and Wildlife Research Institute, 2005). For many years, scientists ha ve studied manatee ecology and population biology through field re search (Hartman, 1979; U.S. Fish and Wildlife Service, 2001). As a result, numerous conservation efforts, all primarily focused on human behavior, have been initiated to help preserve this species including the installation of boater slow speed zones and manatee preservation areas. Although field research provides crucial information about th e manatee’s social structure, habitat usage, and migratory patterns, the sensory processes of this species are just beginning to be understood. Manatee Sensory Systems Historically, few studies ha ve addressed manatee sensory processes, and those that have, have tended to focus on the post-mo rtem physiology of the visual, tactile, and auditory systems. More recently, behavior al studies conducted with captive manatees have provided greater insight into these three sensory systems. Anatomical and Physiological Studies The physiology of the manatee visual syst em has been investigated through the dissection of manatee eyes and visual cortex of the manatee brain. Several studies have shown that the small manatee eye has relativ ely few retinal ganglion cells and lacks an

PAGE 16

3 accommodation mechanism which suggests that the eye has most likely adapted to dim light conditions and is built for sensitivity ra ther than acuity (Walls, 1967; Piggins et al., 1983; West et al., 1991). A later study of ga nglion cell density indicates that the Florida manatee has a limited visual resolution, w ith a minimum angle of resolution of 20 minutes of visual arc (Mass et al., 1997). In addition, two di fferent types of cones have been identified which suggest that manate es possess color vision (Cohen et al., 1982; Ahnelt & Kolb, 2000; Ahnelt & Bauer, unpublished data). The distribution and physiol ogy of the manatee’s vibr issae and body hairs have been investigated through post-mortem dissectio n. Six fields of pe rioral bristles, or vibrissae, have been identified on the face of the Florida manatee. Each follicle is composed of a dense connec tive tissue capsule with a prominent blood sinus complex and substantial innervation, which suggests that the perioral bristles play a tactile sensory role much like that of vibrissae in othe r mammals (Reep et al., 2001). The manatee’s postcranial body hairs were also examined and all were found to contain a blood sinus and were innervated by 20-50 axons. These re sults suggest the possibility that manatees may possess a tactile system that can sense direct ionality in water currents, similar to that of the lateral line of fi sh (Reep et al., 2002). The physiology of the manatee ear has also been studied post-mortem (Klishen et al., 1990). Heffner and Master son (1990) developed a regression of interaural time distances (IATD), the distance sound travels from one ear to the other, divided by the speed of sound, for numerous mammals. These IATD’s have been correlated with the upper frequency limits of the species. This regression found that animals with narrower heads had smaller interaural and intermeatal distances and needed higher the frequency

PAGE 17

4 sensitivity to localize sounds. Studies of the manatee’s inner ear indi cate that their IATD falls between a minimum of 58 sec when measured from intercochlear distances and a maximum of 258 sec when measured from the external intermeatal path, which fall significantly below the regression and sugge st that manatees would not have good directional hearing (Ketten et al., 1992) A more recent i nvestigation of sound conduction through the zygomatic process of the squamosal bone found a lipid-filled channel, similar to that observed in the lower jaw of cetaceans, which may facilitate directional hearing (A mes et al., 2002). However, st udies conducted by Mann et al. (Personal Communication, 2005) which examin ed the manatee’s sound pathways do not support this hypothesis. Behavioral Studies Three behavioral studies of manatee visi on have been invest igated and include brightness discrimination, color discrimina tion, and visual acuity. Brightness discrimination was tested with two West Indian manatees to measure sensitivity. Sensitivity was measured by the relative reflection of targets in a two-choice discrimination procedure, and results suggest that manatees have a Weber fraction of 0.35, similar to that of fur seals although consid erably less than that of humans (Griebel & Schmid, 1997). Color discrimination, also te sted with West Indian manatees using a two-choice discrimination task, found that mana tees were able to discriminate blue and green from a series of comparably bright grays (Griebel and Sc hmid, 1996). A study of visual acuity of two Florida manatees wa s tested using a two-choice discrimination paradigm with various grating widths. Re sults for one of the subjects, who had a minimum angle of resolution measured at 21 mi nutes of visual arc, supported previous

PAGE 18

5 physiological studies conducted by Mass et al. (1997), however, the se cond subject’s was measured at over a degree. These finding s uggested that manatees have poor visual acuity and that vision may be used as an orie ntation cue for large stimuli, but is not of great utility in evaluating fine de tails (Bauer et al., 2003). Three behavioral studies ha ve investigated the tactil e function of the manatee’s facial vibrissae. An observational feedi ng study found that the la rge perioral bristles located on the manatee’s upper lip are used in a prehensile manner during feeding (Marshall et al., 1998 & 2003). A vibrissae sensitivity study conducted with an Antillean manatee, using a two-choice discrimination paradigm with various grating widths, suggested that manatees have good tactile discrimination abil ities with a Weber fraction of 0.14 (Bachteler & Dehnhardt, 1999). A si milar study conducted with two Florida manatees found a Weber fraction of 0.025 for one subject and 0.075 for the other (Bauer et al., 2005). These results s uggest that manatee’s tactile sensitivity probably plays an important sensory role for the species as sensi tivity was found to be comparable to that of the human index finger (Weber fraction of 0.028 ) (Gaydos, 1958) and somewhat better than that of the Antillean manatee (Bachteler & Dehnhardt, 1999) and the harbor seal (Weber fraction of 0.08-0.13) (D ehnhardt & Kaminski, 1995). Three behavioral approaches have b een used in a number of studies which investigate the manatee’s hearing ability: evoke d potential techniques, behavioral testing, and field testing. Evokedpotential techniques demonstrat ed that the largest evoked potentials occurred in the range of 1 – 1.5 kH z but were also found up to 35 kHz (Bullock et al., 1980, 1982; Popov & Supin, 1990). Mo re recent evoked potential techniques indicated that the frequency range of dete ction was from at least 0.2 40 kHz, but

PAGE 19

6 insensitivity of the measurem ent technique suggests some caution on the lower and upper limits (Mann et al., 2005). The temporal reso lution of the manatee’s auditory system was also indirectly measured using Envelope Following Response techniques. Results suggested that manatees could detect cha nges in amplitude modulated rates up 0.6 kHz and had a temporal resolution that was inte rmediate to that of dolphins, who have extremely high levels of temporal resoluti on and can detect changes up to 1.1 kHz, and humans who are sensitive only to 0.2 kHz (Mann et al., 2005). A manatee audiogram was obtained from two Florida manatees used a forcedchoice, two alternative testing paradigm (G erstein et al., 1999). Results showed the hearing thresholds of the subjects to range from 0.5 – 38 kHz for subject 1 and 0.4 – 46 kHz for subject 2. The frequency range of be st hearing for both subjects was reported to be between 6 – 20 kHz. A field study investigated manatee respons es to controlled boater approaches. Results suggested that manatees oriented to wards deeper channel waters and increased their swimming speed when boats were appr oximately within 25 – 50 m of the manatee (Nowacek et al., 2004). The resu lts of both the physiological and behavioral studies of the manatee’s auditory sensory system have demonstrated therefore, that manatees are able to hear sounds within a specific range ; however they do not address one critical component of the manatee auditory system their ability to localize these sounds. Sound Localization Sound localization is the audito ry system’s ability to process the frequency, level, and phase of a sound and associate it with th e spatial location of that sound’s source (Yost, 2000). There are three dimensions from which a sound can be localized, the

PAGE 20

7 vertical plane also called the up-down di mension, distance also called the near-far dimension, and the azimuth plane also called th e horizontal or left-r ight dimension. There are three differential cues that can be used to evaluate and process sound localization within these dimensions, interaural time, inte nsity, and phase differences. Interaural time differences, also known as time of arrival cu es, are a comparison of the sounds time of arrival at each ear. Because the speed of sound is relatively constant, variations in frequency do not have an effect on the pe rception of interaural time differences. Interaural intensity level diffe rences are interpreted when th e sound is one intensity level when it reaches the closest ear but then due to the shadowing effect of the head, is a lower intensity level when it reaches the fart hest ear. The intensity level difference is dependent on the wavelength. Higher freque ncies have shorter wa velengths causing a greater sound shadow. Interaural phase diffe rences are interpreted when the sound that arrives in the first ear is in one period of the frequency but is out of phase and in another period of the frequency when it hits the s econd ear. The phase difference is also dependent on the wavelength. The ability to localize sound is thought to be an evolutionarily conserved trait vital for many species’ ability to find food and conspecifics while avoiding predation. Numerous localization studies that meas ure minimum audible angles have been conducted with terrestrial mammals. These studies measure the smallest detectable difference in an angle of a sound source. Resu lts from these studies have demonstrated that most terrestrial mammals utilize some comb ination of interaural time, intensity level, and phase difference cues, while some have redu ced or lost the ability to use one or more of these differential cues, and others do not seem to be able to use any of them.

PAGE 21

8 Many of the terrestrial mammals studied utilize some combination of interaural cues. These species include humans (Stevens & Newman, 1936; Mill s, 1972), squirrel monkeys (Don & Star, 1972), macaques (Houben & Gourevitch, 1979; Brown et al., 1980), the red fox (Isley & Gysel, 1975) the domestic cat (Casseday & Neff, 1973; Wakeford and Robinson, 1974; Heffner & Heffn er, 1988 B), gerbils (Heffner & Heffner, 1988 A), and Norway rats (Heffner & Heffner 1985). However, all terrestrial animals have not evolved the ability to use th e three types of sound differential cues interchangeably or in combination. The hedgehog (Masterson et al., 1975) and the Northern grasshopper mouse (Heffner & Heffn er, 1988 C) both seem to have reduced or lost the ability to utilize phase difference cu es. In addition, the el ephant, horse, pig, goat and cattle (Heffner & Heffner, 1982, 1984, 1989, a nd 1992 B) appear to have reduced or lost the ability to use interaural intensity level difference cues. At least one species, the pocket gopher ( Geomys bursarius ), is known to be incapable of using any of the interaural time, intensity, and phase differe nce cues (Heffner & Heffner, 1990). Burda (1990) suggested that this may be a result of th is fossorial species’ ad aptation to living in an underground environment where azimuth cues have little meaning. Although similar primitive species have not been studied, they may also lack the use of interaural differential cues. While in-air localizati on may be difficult or impossible for some terrestrial species, the ability to localize sounds underwater may present even more of a challenge to marine mammals. In-Air vs. In-Water Acoustic Properties The speed of sound in air (340m/second) is almost five times slower than in water (1500 m/second) (Urick, 1996). Therefore, mari ne mammal auditory systems need to be

PAGE 22

9 able to process the frequency, intensity level, and phase of sounds that move almost five times faster than the sounds which terrestrial mammals need to process. The cost of evolving such a specialized ability may s eem exorbitant, however acoustic energy in water propagates more efficiently than most other forms of energy such as light, thermal or electromagnetic, which attenuate rapidl y (Au, 1993). Underwater sound localization may provide the best means for marine mammals to find food and conspecifics while avoiding predation in deep water. Shallow water conditions, however, present additional acoustical challenges. Acoustic energy in shallow water does not travel as efficiently as it does in deep water (Medwin and Clay, 1998) Higher frequencies, which have shorter wavelengths, become more directional than lower frequencies a nd the sounds tend to have more reflection off the water’s surface and bottom terrain. Since Florida manatees spend a significant amount of time grazing in shallow water, localization of sound from conspecifics and boats may be particularly challenging. Behavioral testing of underwater so und localization abili ties have been investigated with numerous species of cap tive marine mammals including sea lions (Gentry, 1967; Moore, 1974; Moore & Au, 1975; Holt et al., 2004), bottlenose dolphins (Renaud & Popper, 1975; Moore & Pawloski, 1 993; Moore and Brill, 2001), and harbor seals (Anderson, 1970; Terhune, 1974). Results from these studies suggested that these marine mammals have the ability to local ize underwater sounds by using the same interaural differential sound cues th at terrestrial mammals use. To date, only one behavioral study had been conducted to measure a Florida manatee’s ability to localize sp ecific sounds (Gerstein 1999). In this study, using a fourchoice discrimination paradigm, the subject was required to locali ze a 20ms tonal signal

PAGE 23

10 of 0.5, 1.6, 3, 6, or 12 kHz, which was pulsed for durations of either 200 or 500 ms, paused for 400 ms, and then repeated, from one of four locations. Two speakers were positioned at 45 o and two at 90o angles to the manatee’s head. Results indicated that the manatee was able to localize al l of the signals, but that accur acy decreased with the lower frequencies. Accuracy was better w ith the longer durations and at the 45o angles. Based upon these results, Gerstein has suggested that manatees may not be able to effectively localize the low frequency sounds of boat engi nes to avoid collisions in the wild. Florida Manatees and Environmental Noise The Florida manatee lives in an enviro nment where recreational boats are found in high numbers and conspecifics are often out of visual range. How then do they avoid boat collisions and find conspecifics in thei r vast environment? The manatee olfactory and taste sensory systems have not been i nvestigated, and the pr eviously described sensory system studies have suggested that manatee vision may be poorly adapted for these tasks, and that while their tactile sensi tivity is impressive, it is best utilized to investigate objects nearby. The manatee aud itory system may play a crucial role in accomplishing these challenging tasks. The Florida manatee shares its habitat w ith over 880,000 register ed boats (Florida Department of Highway Safety and Motor Vehi cles). In 2004, watercraft-related injuries accounted for 25% of all manatee mortalitie s, (Florida Fish and Wildlife Research Institute, 2004 B). One question that has b een raised is, if mana tees can localize boat engine noise, why are there so many watercra ft mortalities? Conversely, if manatees cannot localize boat engine noise why are there not more watercraft mortalities? Boat engine noise can be categor ized in two ways: cavitating, which is associated with

PAGE 24

11 propeller rotation and produces higher frequency broad-ba nd noise, and non-cavitating, which is associated with other propulsion machinery that produces lower frequency broad-band noise (Miksis, 2006). Boats trav eling at high speeds usually produce the higher frequency cavitating noise, while boats traveling at idle and slow speeds produce the lower frequency non-cavitating noise (Ro ss, 1976). The dominant recreational boat frequency range is typically 0.01 – 2 kHz but can reach over 20 kHz. The estimated 1/3octave source levels at 1 m for small moto rboats are 120-160 dB re 1 Pa-m (Gerstein, 2002; Richardson et al., 1995). Pe rsonal watercraft, such as je t-skis, utilize jet propulsion rather than outboard propellers and are appr oximately 9 dB quieter than small motorboats (Buckstaff, 2004). The often solitary Florida manatee is ab le to find conspecifics in a wide-ranging habitat. The question of how these semi -social animals find one another remains unanswered, however the ability to localize th e vocalizations of conspecifics would be beneficial. A study that compared the vocal izations of the two sub-species of West Indian manatee (the Florida and Antillean manatees) found no difference between the two, and characterized their voc alizations as short (durati on) tonal harmonic complexes that range from almost pure tones to broad-band noise and have a fundamental frequency that ranges from 2.5 – 5.9 kHz but can extend to 15 kHz (Nowacek, et al., 2003). The question of whether or not manatees are able to localize boat engine noises and conspecific vocalizations has puzzled rese archers for years and is a topic of debate that warrants further investigation. In th is study, two Florida manatees were conditioned and tested to determine their ability to localize sounds of different frequencies and durations from four locations. This res earch extended the loca lization work done by

PAGE 25

12 Gerstein (1999) by testing th ree broad-band frequency ranges, in addition to two tonal signals. Hypotheses Four hypotheses were made. The first pos ited that overall performance accuracy would be higher for the broad-band signals th an for the tonal signals. Since broad-band sounds contain a range of different frequencie s and tonal signals contain a very narrow frequency range, broad-band sounds are easi er to localize (Coren et al., 1994). The second hypothesis stated that perfor mance accuracy would decrease as sound duration decreased. Longer signals often provide subjects enough time to move their heads in different directions to better utilize time, level, or intensity differential cues as a means of localization. Shorter signals redu ce and/or remove the possibility of head movements, thereby limiting th e sound localization cues. The third hypothesis asserted that lear ning would occur as blocks of tests progressed within each conditi on. Learning a task often occurs in a progression as experience with the task increases. The subj ects of this study we re asked to localize a variety of sound conditions that they were not familiar with. As their experience with these sounds increased, the task was expect ed to become easier, thus increasing performance accuracy. Finally, the last hypothesis c ontended that more errors would be made between the 45o and 90 o speakers located on opposite sides of the subject, than between the two 90 o or the two 45o speakers located parallel with and in front of the subject. It was thought that the subjects would be better able to distinguish the sounds originating at the

PAGE 26

13 front two locations from the side two locations than from the front and side locations to each side of them.

PAGE 27

14 Methods Subjects The subjects of this study were tw o captive-born male Florida manatees ( Trichechus manatus latirostris ) that reside at the Mote Ma rine Laboratory and Aquarium in Sarasota, Florida. All procedures used were permitted through the United States Fish and Wildlife Service (Permit # MA837923-6) and approved by the Institutional Animal Care and Use Committee of Mote Marine Labo ratory and Aquarium. At the inception of this study Hugh was 20 years of age, weighed 547 kg, and was 310cm in length, while Buffett was 17 years of age, weighed 773 kg, and was 334 cm in length. They were housed in a 265,000 liter exhibit that was compos ed of three inter-connected sections: a 3.6 x 4.5 x 1.5 m Medical Pool, a 4.3 x 4.9 x 1.5 m Shelf Area, and a 9.1 x 9.1 x 3 m Exhibit Area (Figure 1). Both animals had acquired an extensive training history over the previous seven years and were subjects in an auditory evoked potential study (Mann et al., 2005), making them exce llent candidates for this proj ect. In addition, they had been behaviorally conditioned for husbandry pr ocedures (Colbert et al., 2001), a serum and urine creatinine study (Man ire et al., 2003), a visual acuity study (Bauer et al., 2003), a lung capacity study (Kirkpatric k et al., 2002), and a tactile sensitivity study (Bauer et al., 2005).

PAGE 28

15 Figure 1: Diagram of the 265,000 L manatee exhibit composed of a Medical Pool, Shelf Area, and Exhibit Area. The lines in the Medical Pool represent a distance scale, used in a previous study that was painted on the floor of the exhibit. The oval masses in the Exhibit Areas represent outcroppings in the bottom terrain (built of cement) to conceal the two fl oor-level filtration drains (gratings). The rectangles represent a tree log and stump (built of cement). Overview This section of the methods provides a synopsis of the study to familiarize the reader with the basic experimental design and general logistics, such as signal generation, computer programming, and personnel, requir ed to run the experiment. Specific methodology used for training and then testing th e subjects will be discussed in sections to follow. Experimental Design Testing for this study was conducted in the center of the Shelf Area with the test animal positioned midway between the e xhibit bottom and the surface of the water (approximately .75 m below the surface). The non-test animal was held at station in either 9.1 x 9.1 x 3 m Exhibit Area 4.3 x 4.9 x 1.5 m Shelf Area 4.5 x 3.6 x 1.5 m Medical Pool

PAGE 29

16 the Medical Pool or the Deep Area. The test subject was required to position himself perpendicular to a stationing bar with the crease of his rostrum, approximately 10 cm posterior to his nostrils, pre ssed against it, in response to a specific pulsed-tone being played from a stationing speaker. The manat ee remained stationed until a test signal was played from one of four underwater speakers, whereby he swam to and depressed the speaker from which the sound originated. If correct, a secondary reinforcer tone was emitted from the test speaker and the subject re turned to the stationing device to be fed a primary reinforcement of food. If incorrect, the stationing tone was played from the stationing apparatus speaker and the subject re-positioned correctly with no primary or secondary reinforcement given, and awaited the initi ation of the next trial. All test trials were video-recorded from an overhead camera. Signal Generation Signals were generated digitally by a Tucker-Davis Technologies real-time processor (RP2.1), and attenuated with a pr ogrammable attenuator (PA5) to control intensity level. The signals were amplifie d with a Hafler power amplifier and switched through a power multiplexer (PM2R) that was ca pable of switching the signal to one of the four testing speakers (AquaSynthesis). The status of a switch at each speaker location was monitored by the digital inputs on two RP2’ s. A separate digital to analog channel was used to generate the signal to the stat ioning speaker at the center of the array. Computer and Programming A Dell laptop computer (model Latitude D505) with Windows XP was used to run the signal generation equipment and to automatically download the parameters of each trial into an Excel file.

PAGE 30

17 All signals and conditions of the testing sequence used in the experiments were programmed in RPvds language on the Tucker-D avis signal generator. These programs included the development of two “call to sta tion” signals (one unique to each subject), five “test/training” signals, a nd two “reinforcement” signals (one unique to each subject) used to bridge the subj ect if he was correct. A program was also developed to generate blocks of twelve trials that were counterbalanced between the f our speaker locations. The c ounterbalancing was done in a quasi-random order, meaning that the test signal location was randomized, but had a criterion of no more than two trials in a ro w from the same location. Each test session consisted of three blocks of trials per subject. Each trial was initiated and complete d through an electronic button box which was connected to the RP2 unit, and then into the laptop computer (F igure 2). The button box had four control buttons and four colored LE D lights built into it. The station signal was used to call the specified subj ect to station, to start a trial, or to get the subject to refocus his attention if he was distracted. The actual speaker switching occurred while the station signal was played. The test signa l was unique for each condition and was played only once per trial. The corre ct button was used to play the subject’s unique secondary reinforcement signal when he correctly identif ied the test sound location. This informed him that he was correct and could receive primary reinforcement. The wrong button was used when the subject incorre ctly identified the test signa l location, and was immediately followed by a station signal which informed th e subject that he was incorrect and needed to return to station without receiving primar y reinforcement. Four LEDs were included

PAGE 31

18 on the button box to provide a visual indica tion that their corresponding signals were played and that the trial was downloaded into the Excel file. Figure 2. Electronic button box used to run the sessions and automatically download each trial into a digital excel file. Personnel: Three people were required to run the e xperiment, a “Test Trainer” who worked with the test subject, a “Dat a-recorder” who ran the computer and recorded the data, and a “Station Trainer” who worked with the non-test manatee. Three Manatee Care staff trainers (including the author) were used to run the test animal and/or the computer throughout the experiment. These individuals had extensive experience working with the subjects and were completely familiar with the computer program and the experimental plan. The Data-recorder was seated behind the computer out of both the test subject’s and Test Trainer’s line of sight to avoid inadvertent cuing of the Test Trainer. From this location, the Data-recorder was unable to dete rmine the test subject ’s position in the water. This position was primarily run by one of two experienced staff trainers and occasionally by the author. Station Signal Test Signal Correct/ Brid g e Wrong Station Signal LED Correct LED Wrong LED Test Signal LED

PAGE 32

19 The Test Trainer was positioned on the center of wooden boards that were suspended across the Shelf Area, directly a bove the stationing apparatus. The Test Trainer was “blind” to the test stimulus locations and wore so und-dampening headphones to avoid cueing the subject. This pos ition was run primarily by the author and occasionally by the other most e xperienced staff trainer. The Station Trainer was positioned at either the northeast end of the Medical Pool or Deep Area, out of the test subject’s line of sight. This positi on was run primarily by one of four experienced volunteers/interns and occasionally by one of the three staff trainers. Training Procedures While the previous methods secti on provided a broad overview of the experimental design and logist ics utilized throughout the expe riment, it is important to understand that numerous behaviors needed to be trained and chained together before testing could be initiated. This section provides a thorough description of the global procedures used in both the tr aining and testing portions of this study. In addition, two animal training phases are described in detail. Global Procedures Training and testing sessions were run between 0700 and 1000h five days per week before the Aquarium was open to the public. The manatees’ daily ration of food (72 heads of romaine lettuce and 12 bunches of kale) was fed to the animals from 1200 to 1400 h and was usually consumed by 1700 h, leaving a 14 to 16 hour overnight fast before training was initiated the following morning. All training was completed using standard positive classical and operant condi tioning techniques. The primary reinforcers

PAGE 33

20 used included bite size pieces of apples, beet s and baby peeled carrots. Zupreem monkey biscuits, one of the manatees preferred foods, were used to reward an especially desired behavior during shaping pro cedures. A unique whistle was used as a secondary reinforcer to bridge each animal independently of one another. In addition, verbal and tactile secondary reinforcers were used. Shaping by reinforcement of successive approximations was used to train all beha viors (Pepper & Defran, 1975) and undesirable behaviors were ignored. In addition, timeouts, (Pepper & Defran, 1975; Domjan, 1998) or the removal of the opportunity to receive reinforcement, were used if a string of undesirable behaviors occurred. Phase I: East Wall of the Shelf Stationing Training was initiated with the subject posi tioned perpendicular to the east wall of the Shelf Area and the other animal positioned in the northwest corner of the Medical Pool (Figure 3). The subject was positioned in this way to better allow the Test Trainer to easily reach the subject and physically ma neuver his body and head into the correct position.

PAGE 34

21 Figure 3. Training configuration for the east wall of the Shelf Area. The blue octagon represents the Test Trainer’s location, the green square represents the Data-recorder’s location, and the orange triangle represents the Station Trainer’s location. A stationing apparatus was constructed fr om 2.54 cm diameter polyvinyl chloride pipe (PVC) (Figure 4). This apparatus was designed to fit over the e dge of the exhibit wall for stability, and it extended 30.48 cm belo w the surface of the water. To prevent interference between the sound source and th e manatee’s ears, the stationing apparatus had only a 23 cm stationing bar that the subj ect pushed his rostrum up against instead of hoop or frame which encircles the head co mpletely. An underwater speaker was suspended from the top of the stationing a pparatus and positioned above the manatee’s head, just below the surface of the water. N W E S Medical Pool Shelf Area Exhibit Area

PAGE 35

22 Figure 4. Stationing apparatus used in Phase I of the training. The yellow circle represents the speaker that played the stationing tones. The subject pressed the crease of his rostrum up against the gray stationing bar on the bottom. The test subject’s unique st ation signal was played from the stationing speaker to call the subject to station. Each station si gnal ranged from 10 to 20 kHz and played for a 2000 ms duration, however Buffett’s repe ated at a slower rate of 1.5/s while Hugh’s repeated at a faster rate of 5/s. In re sponse to their station signal, each subject was trained to swim to the stationing apparatus, position himself perpendicular to it, and press the crease of his rostrum against the stationing bar. Both animals had previously been trained to station and follow their own personal targets, and in the ea rly stages of shaping the stationing behavior, the Test Trainer used the subject’s target to guide him to the stationing bar when the station signal was played. Shaping of the correct position was W a t e r Li n e 30.48 cm

PAGE 36

23 facilitated by the Test Trainer’s reaching in to the water to help maneuver the manatee’s head. Over the course of multiple sessions, ea ch learned to station correctly in response to the station sound. The length of time spent in this position was extended to 60 seconds using a fixed-interval schedule of reinforcement (Ramirez, 1999). Speaker Selection Once the manatees had a firm grasp of the stationing behavior, training for the localization of the training s timuli through speaker selection was initiated. The speakers were suspended from speaker holders made of 1.88 cm diameter PVC pipe. The speaker was attached to a long speaker suspension r od which was designed to pivot so that the speaker at the bottom of the rod could be pus hed backwards while the top of the rod tilted forward to touch the speaker holder frame like a pendulum (Figure 5). Initially, one test speaker was placed approximately 20 cm away from the subject’s head at either a 90 o or 270 o position (subject’s head was facing 0o). This speaker’s position was alternated between the two angles so that the subject b ecame familiar with moving to either the left or right side to select the correct location.

PAGE 37

24 Figure 5: Speaker holder with a ttached underwater speaker. The spea ker suspension rod was designed to pivot so that the manatee could push the speaker at the bottom backward while the top tilted forward to touch the speaker holder frame. A 3000 ms broad-band signal, with a frequency range of 0.2 – 20 kHz was programmed in RPvds and used to train th is discrimination task. Two secondary reinforcement signals were programmed to ma tch the unique whistles used to bridge each animal. Buffett’s reinforcement signal ranged from 1.4 to 12 kHz with a peak at 5.3 kHz, while Hugh’s had more of a warble to it and ranged from 1.2 to 11 kHz with a peak at 2.7 kHz (Figure 6). Speaker Suspension Rod Water Line Pivot Point Speaker Frame

PAGE 38

25 Figure 6: Power spectra (top) and spectrograms (botto m) of the secondary reinforcement signals. Buffett’s ranged from 14 to 120 kHz with a peak at 53 kHz, while Hugh’s had more of a warble to it and ranged from 12 to 110 kHz with a peak at 27 kHz. When the subject had stationed correctl y, the training signal was played and the Test Trainer used the subject’s personal target or their hands to help guide the subject to the test speaker. When the subject touche d the test speaker w ith his rostrum, his secondary reinforcer signal was played from the test speaker and he was fed a primary reinforcement of food. Initially, the traini ng signal was played continuously until the subject pressed the test speaker. Once the desired behavior was obtained, the sound duration was then shortened to a series of 1000 ms signals, a nd finally reduced to a single 1000 ms signal. Through successive approxi mations, the subjects were trained to approach and push the test speaker backwards until the top of the speaker suspension rod touched the front of the speak er holder frame when the tr aining signal was played. In addition, the distance of the speaker from th e manatee’s head was gradually increased from 20 cm to approximately 100 cm.

PAGE 39

26 After each subject learned to select the single test sp eaker when it was located 100 cm away and was re-positioned at either the 90o or 270o location, a second test speaker was introduced so that two test speakers were always present, one at the 90o and one at the 270o location. The manatees were then requ ired to localize the training signal source through a two-choice discrimination task. Th e training signal was delivered through one of the two test speakers based on counterb alanced schedules (Gellerman, 1933; Fellows, 1967) that were programmed in RPvds (Figure 7). Figure 7: Training configuration for the east wall of the Shelf Area with two speaker locations. The black circles represent the speakers. The blue octagon represents the Test Trainer’s location, the green square represents the Data-recorder’s location, and the orange triangle represents the St ationing Trainer’s location. When the subjects were able to dependa bly perform the discrimination task at above a 75% accuracy level, the stationing apparatus and speakers were modified to N W E S Medical Pool ShelfArea ExhibitArea

PAGE 40

27 reach a depth of .75m, the mid-way point of th e water depth in the Shelf Area, to equalize the amount of sound reflectivity from the su rface and the bottom. The previously established behaviors were re-shaped to meet the new depth requirements. Phase II: Center of the Shelf: After the subjects were able to reliably di scriminate between th e two test speakers located at the east wall of the Shelf Area with the new depth requirements, the whole setup was moved to the center of the Shelf Area and rotated 90 o to the South. The Test Trainer was positioned on a wooden bridge th at was 15.24 cm above the surface of the water and spanned the width of the Shelf Area. Training was re-established in this new location and when both subjects performed relia bly at a 75% accuracy level or higher, a third test speaker was introduced and positioned at a 0 o angle to the animal’s head (Figure 8).

PAGE 41

28 Figure 8: Training configuration for the center of th e Shelf Area with three sp eakers represented as the black circles. The blue octagon represents the Test Tr ainer’s location, the green square represents the Datarecorder’s location, and the orange triangle represents the Stationing Trainer’s location. Training with the three test speakers commenced, however it quickly became apparent that the subjects had difficulty localizing the training signal, especially when it originated from the 0o location. Both subjects’ perfor mance accuracy declined and they demonstrated various signs of frustration including chuffing, leaving the task multiple times, as well as grabbing, pulling, and break ing the equipment. Interestingly, both manatees (independently of one another) assumed a new body position when at station. N W E S Medical Pool Shelf Area Exhibit Area

PAGE 42

29 Each began to tilt the horizont al axis of their bodies at an angle while still remaining perpendicular to the stationi ng bar. The training signal duration was increased from 1000 ms to 3000 ms to simplify this part of the training and reduce the animals’ level of frustration and their performance improved sl ightly but was still poor with the speaker located at 0 o. This finding was not surprising considering that even humans have difficulty localizing sound when it is directly in front or back of them because there are no interaural differences in time of arri val cues (Yost, 2000). The new stationing postures that both subjects developed may ha ve indicated a strategy they had developed to try to better differentiate time of arrival cues. A fourth test speaker was introduced to compensate for the difficulties observed with the speaker located at 0 o. The four speaker holders were permanently positioned 105cm from the center of the subj ect’s stationing bar, at 45 o, 90 o, 270 o, and 315 o angles (Figure 9).

PAGE 43

30 Figure 9: Training configuration for the center of th e Shelf Area with four te st speakers, located 105cm from the center of the stationing bar and .75m below th e surface. The test speakers are represented as the black circles. The blue octagon represents the Test Tr ainer’s location, the green square represents the Datarecorder’s location, and the orange triangle re presents the Stationing Trainer’s location. The four test speakers used were identif ied and numbered (from 0 to 3) with a permanent marker. Speaker zero was att ached to the speaker holder located at 90 o, speaker one at 270 o, speaker two at 315 o, and speaker three at 45 o. Training for the localization task continued until each subjec t performed the task at a 75% or higher accuracy rate for a minimum of 72 consecutive trials. N W E S Medical Pool Shelf Area Exhibit Area 105cm 105c m 105c m 270 0 90 45 0 315 0

PAGE 44

31 Testing Procedures The preceding methods section summarized the global procedures used in both the training and testing portions of this study and detailed tw o phases of training. All of the methods depicted to this point have led to this final section of the methods: testing procedures. This section provides speci fic information about the experimental conditions, sound calibrations, environmental co nditions, experimental controls, and test block criterion. Additionally, computer pr ogramming, personnel responsibilities, and data and video recording met hodology are expanded upon. Fina lly, a step-by-step testing trial sequence is portrayed. Experimental Conditions Fourteen experimental conditions were te sted (Table 1). These included three broad-band noise bursts of different frequency ranges tested at four duration lengths, and two tonal signals tested at one duration length. All of the test sounds were played at the same spectrum level. Because of this, sounds with broader frequency spectra had louder root mean square (rms) amplitudes. The sound levels were also randomized 1.5 dB to obscure any loudness differences between speakers. Six blocks of twelve trials were run for each of the 14 conditions.

PAGE 45

32 Table 1 Conditions tested included three broadband signals at four durations and two tonal signals at one duration. All signal s were played at the same spectrum level. Frequency (kHz) Duration (ms) Spectrum Level (dB re 1 uPa) 0.2 20 3000, 1000, 500, 200 100 6 20 3000, 1000, 500, 200 100 0.2 2 3000, 1000, 500, 200 100 4 3000 100 16 3000 100 The testing of broad-band noise examined the manatees’ ability to localize sounds that had a variety of differe nt frequencies blended into them. Broad-band sounds are more typically found in all natu ral habitats than tonal signals. The 0.2 20 kHz condition had the widest range of frequencies include d in it and was the broad-band signal used during training, the 6 20 kHz condition contained the highest frequencies, and the 0.2 – 2 kHz condition contained the lowest frequencies. The testing of tonal signals examined the manatees’ ability to localize sounds that were frequency specific. The low frequency 4 kHz tone was similar to the dominant frequency of manatee vocalizations, wh ile the high frequency 16 kHz tone was comparable to the manatees’ peak hearing frequency.

PAGE 46

33 The order of test stimuli presentation wa s as follows: six blocks of each broadband noise, starting with 0.2 – 20 kHz, then 6 – 20 kHz, and finally 0.2 – 2 kHz, were tested at the 3000 ms duration. This orde r was followed throughout each of the sound duration conditions. The tonal signals were only tested at the 3000 ms duration because the subjects performed at a much lower accura cy level and were exhibiting strong signs of frustration. Personnel Responsibilities The Data-recorder, who was positioned out of sight of the Test Trainer and the subject, had six duties. The first was to se t up the correct experime ntal conditions needed for the different portions of the session on th e computer using a gra phical user interface that was programmed in Visual C (see Appendix A for experimental condition set up protocols). The second was to initiate each trial through the button box when instructed to do so by the Test Trainer. The third was to determine which location the subject selected from the Test Trainer and to inform the Test Trainer if this location was correct by leaning out from behind the computer to gi ve a head nod, or if wrong to give a head shake. The fourth was to complete the trial through the button box. The fifth was to record all data on a tank-side session sheet (see Appendix B). The sixth was to run the video equipment. The Test Trainer, who was blind to th e test signal locations and wore sounddampening head phones, was responsible for six duties. The first was to ensure that the subject was positioned correctly before the in itiation of each trial. The second was to ensure that a 25 second minimum inter-trial in terval was met. The third was to let the Data-recorder know when to initiate each trial by verbally stating “tone”. The fourth was

PAGE 47

34 to let the Data-recorder know which location the animal se lected by verbally stating the location number (900 was location 0, 2700 was location 1, 3150 was location 2 and 450 was location 3). The fifth was to determine if the subject was co rrect by looking at the Data-recorder when she came into view for th e appropriate head nod or shake. The sixth was to provide the subject his primary reinforc ement if he was correct and to let the Datarecorder know when the next trial could be initiated by verbally stating “station”. The Station Trainer was responsible for hol ding the non-test animal at station in either the northeast corner of the Medical Pool or Exhibit Area throughout the test animal’s session. The non-test animal was pos itioned out of view of the test animal and could be held in either a dorsa l-up or ventralup layout position. Data Recording Data from each session were recorded in three ways. The first was through the automated computer reports that were recorded in a digital format within the Excel file. The second was through hand r ecorded reports that the Da ta-recorder completed on a tank-side data sheet (see Appendix A). This information was then manually entered into a Microsoft Access database created on a Dell desktop computer (model Dimension 8300) after the completion of each session. A ll data entered into the database were double-checked for accuracy by a second trainer af ter they were entered. This database was designed specifically for this experiment and had a user-friendly data entry screen (Figure 10). This data base was then used to compile and analyze the test data. The third data recording method was th rough the video recording of each test block. A Sony variable zoom, high resolution, outdoor w eather proof, color dome camera (model SCWCD358DVP) was attached to the exhibit’s ceiling di rectly over the subject’s head and

PAGE 48

35 connected to a Sony digital video camera (mode l DCR-TRV50). Pre-printed data sheets that identified the date, s ubject, test frequency, sound durat ion, and speaker locations, were recorded prior to th e initiation of each block. Figure 10. The data entry screen used to enter all of the session’s information into the Access database. Environmental Conditions All sessions were conducted with the manatee’s typical under-water exhibit noise held constant, that is the exhibit’s filtration system a nd pumps ran in their normal capacity. The exhibit noise typically ran be low 500 Hz. The in-air noise level, however, was considerably louder than was typical of previous studies with these animals. Construction for a new 3-story building, loca ted less than 200 feet from the manatee exhibit, began just prior to the initiation of training for this study causing intermittent noise of different intensities.

PAGE 49

36 Controls Numerous controls were put into place to avoid test signal and speaker artifacts, as well as trainer cuing. Contro ls were also established to en sure that the subjects were motivated for the testing and to be able to drop a testing block if specific criteria were met. Signal and Speaker Artifacts All signals had a 100 ms rise -fall time to eliminate tran sients. Signal intensity levels were randomized by +/1.5 dB around the te st level to minimize the possibility of intensity level cues being used to determine speaker location (i.e. associating a particular level from a particular speaker). To control for the manatees using any possible speaker artifacts as cues, the speakers were removed from their original speaker holders and re-connected to the speaker holders diagonally across after th ree blocks had been completed for each condition (i.e. speaker 0 was located at the 900 location for the first three blocks of each condition and then rotated to the 3150 location for the last thre e blocks of each condition and this pattern was repeated with other two speakers). All five test signals were recorded from each of the four speakers in their different positions via a Reson hydrophone to be calibrate d and analyzed to ensure that speaker artifacts were not present. Trainer Cues: Several procedures were followed to avoi d trainer cuing to the subjects. All personnel were positioned out of the test su bject’s line of sight except for the Test Trainer. The Test Trainer was required to wear sound-dampening headphones to avoid the possibility of hearing the test signals and was blind to the test signal’s location. The

PAGE 50

37 Data-recorder was the only individual who knew where each trial’s correct test signal location was, and only obtained that knowledge at the initiation of each trial. To avoid cuing the Test Trainer, the Data-recorde r remained positioned behind the computer screen and was not visible unt il after the subject had made his location selection at the end of each trial. At this point the Test Trainer would look towards the Data-recorder and the Data-recorder would move into view to indicate if the subject’s choice was correct or wrong. Motivational Effects To control for motivational effects, each animal’s session was started with eight “warm-up” trials, two from each location in a randomized order, and ended with four “cool-down” trials, one at from each location in a randomized order. The signal stimulus used for these trials was the same 3000 ms 0.2 – 20 kHz, broad-band noise burst used throughout training. To control for an apparent initial period of confusion that both manatees displayed when changes between frequencies occurred, eight “practice” trials, two from each location in a randomized order, were complete d directly after the eight warm-up trials were completed. The signal stimulus used fo r these trials was the same frequency and duration as the stimulus to be tested in th at session. It was be lieved that although the same spectrum level was used for all of the sound conditions tested, the bandwidth varied with the different stimuli, and the loudness of the test sounds may have been perceived as different to the manatees. Block Criteria Two specific criteria were defined as reasons to drop a test block. The first stipulated that a minimum performance accu racy of 75% was required on the warm-up

PAGE 51

38 and cool-down trials. The second defined a maximum allowance of any combination of three interruptions from the nontest manatee, and/or leaves or attempted leaves from the test subject per block. If a block was dropped, the experiment al condition was repeated in the next session. Testing Trial Sequence In summary, the sequence of steps utilized for each test trial was as follows: 1) The Data-recorder would press the “Sta tion Signal” button to call the subject to station. This sound would orig inate from the stationing speaker. 2) The subject would align himself perpendi cular to and press the crease of his rostrum against the stationing bar. 3) The Test Trainer, after ensuring that a 25 second inter-trial interval had passed and that the subject was corr ectly positioned, would tell the Datarecorder to play the test sound by verbally stating “tone”. 4) The Data-recorder would press the “Tes t Signal” button to play the test sound. This sound would originate from one of the four test speakers in a quasi-random order. 5) Upon hearing this sound, the subject would swim to one of the speaker locations and depress the speaker. 6) The Test Trainer would tell the Data -recorder which location the subject selected by verbally stating the locat ion number. The Test Trainer would then look at the Data-recorder to see if this location was correct. 7) The Data-recorder would determine if the speaker location selected matched the actual location select ed by the computer.

PAGE 52

39 If correct, the data recorder woul d press the “Correct Signal” button so that the secondary reinforcement tone was played to let the subject know he was correct, and would lean into view to give a head nod and let the Test Trainer know that primary reinforcement should be given. The Test Trainer would feed the subject his primary reinforcement and wait for the animal to stop chewing before telling the Datarecorder to call the an imal back to station by verbally stating “station”. The Data-recorder would press the “Station Signal” button and wait for direction to start the next trial. If wrong, the data recorder w ould press the “Wrong” button, immediately followed by the “Station Tone” button to let the subject know he was wrong and that he should re-station, and would lean into view to give a head shake and let th e Test Trainer know that the animal should not receive any primary reinforcement. The Test Trainer would wait fo r the minimum 25 second intertrial interval to pass and for the s ubject to station correctly before starting the next trial.

PAGE 53

40 Results: Training was initiated on January 6, 2005 and completed on July 11, 2005. Testing was initiated on July 12, 2005 a nd completed on August 26, 2005. Both manatees learned the task easily. A total of 1,164 trials were run with Hugh and 1,116 trials were run with Buffett. Nine bloc ks with Buffett and 13 blocks with Hugh were dropped as they met the drop criteria. A total of 60 blocks or 1,008 test trials were kept for each subject. Five data analyses were conducted for each subject. Two analyses, performance by speaker location and speaker calibration, exam ined the possibility of the existence of speaker artifactual cues. Three analys es, overall performance accuracy, progression learning, and error distribution, measured the subject’s capacity to localize the test signals. Performance by Speaker Location Initially, the numbered test speakers were attached to specific speaker holder locations. For each condition tested, 3 blocks of test trials were run with the test speakers in these locations. As a contro l to avoid the use of speaker ar tifact cues, the test speakers were removed from their original speaker holder locations and re-connected to the speaker holders located diagonally across for th e remaining 3 blocks of test trials (i.e. speaker 0 was located at the 900 location for the first three blocks and then rotated to the 3150 location for the last three blocks and th is pattern was repeated with other two speakers). To analyze if the subjects used speaker artifact cues ra ther than the test

PAGE 54

41 signals, the 3 blocks of test trials when the speakers were in their original locations (speaker 0 was located at 900) were averaged together and compared to the those from the remaining 3 blocks (0 was located at 3150) by condition for Hugh (Figure 11) and Buffett (Figure 12). No obvious pattern was observed with either subject. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 90o315o90o315o90o315o90o315o90o315o 0.2 20 kHz6 20 kHz0.2 2 kHz4 kHz16 kHz Frequency and Speaker 0 LocationPercent Correc t 3000ms 1000ms 500ms 200ms Figure 11. A comparison of the average performance accuracy on Hugh’s first three test blocks when the test speakers were in their original locations and last three test blocks after the speakers had been moved to their new locations.

PAGE 55

42 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 90o315o90o315o90o315o90o315o90o315o 02 20 kHz6 20 kHz02 2 kHz 4 kHz16 kHz Frequency and Speaker 0 LocationPercent Correc t 3000ms 1000ms 500ms 200ms Figure 12. A comparison of the average performance accu racy on Buffetts first th ree test blocks when the test speakers were in their original locations and last three test blocks after the speakers had been moved to their new locations. Speaker Calibration All five test signals, recorded via a hydrophone located at the center of the stationing bar, were recorded from each of the four speakers when they were positioned in their original locations (speaker 0 was located at 900) and again when they were repositioned and connected to the speaker holders diagonally across (speaker 0 was located at 3150). The speakers were switched a total of five times throughout testing. This included once after the first three testing blocks of each frequency condition was completed at the first duration condition, a seco nd time after the final three blocks at the first duration and first three blocks at the second duration were complete, a third time when the final three blocks of the second dur ation and first three of the third duration were complete, a fourth time when the final three blocks of the third duration and first three first three blocks of th e fourth duration were complete, and finally a fifth time to complete the last 3 blocks of all frequencies in the fourth duration.

PAGE 56

43 Power spectra were made of all the r ecordings and examined to look for any frequency or intensity cues that might occur at either a specific location (Figure 13) or from a specific speaker (Figure 14). No obvious patterns were observed, and on the contrary, re-location of the speakers produ ced minor signal variations by changing the sound field slightly. Figure 13. A power spectra comparison of the 6 20 kHz broadband test signa l from each of the four locations when the speakers were in their original positions (shown on the left in blue) and when they had been re-located (shown on the right in red). Spkr 0 @ 90oSpkr 1 @ 270oSpkr 2 @ 315oSpkr 3@ 45oSpkr 2 @ 90o Spkr 3 @ 270o Spkr 0 @ 315o Spkr 1@ 45o

PAGE 57

44 Figure 14. A power spectra comparison of the 0.2 20 kHz broadband test signa l from each of the four speakers when they were in their or iginal positions (shown on the left in blue) and when they had been relocated (shown on the right in red). In addition, spectrograms of all the recordings were examined for temporal cues, such as intensity distortions, that might occur within the specific frequencies tested. No obvious patterns or harmonic distortions were observed with either the broad-band (Figure 15) or pure tone signa ls (Figure 16). Interestingl y, the construction hammering can be observed in the pure tone spectrograms. S p kr 0 @ 90oS p kr 1 @ 270oS p kr 2 @ 315oS p kr 3 @ 45o S p kr 0 @ 315oS p kr 1 @ 45oS p kr 2 @ 90oS p kr 3 @

PAGE 58

45 Figure 15. Spectrogram comparison of the three broadband test signals played at the 3000 ms duration (sample rate of 97,656 Hz). 0.2 2 kHz 0.2 20 kHz 6 -20 kHz Time

PAGE 59

46 Figure 16. Spectrogram comparison of the two pure tone test signals played at the 3000 ms duration. The recurrent frequency spikes are from the nearby construction hammering (sample rate of 97,656 Hz). Overall Performance Accuracy Overall performance accuracy was determined and described in Table 2. Percentage correct was calculated for each s ubject based upon 72 trials per condition with a total of 1,008 trials per subj ect. Both subjects performed well above the 25% chance level for all of the broad-band frequency condi tions. Hugh showed a drop in percentage correct as the broad-band signal durations decreased, but this re sult was not observed with Buffett. Both animals also performe d above chance levels with the pure tone signals, but at a much lower accuracy rate than with the broad-band signals. 4 kHz 16 kHz Time

PAGE 60

47 Table 2 Overall Accuracy Performance per Subject by Frequency and Duration Conditions Frequency (kHz) Duration: 0.2 20 6 20 0.2 2 4 16 Hugh 3000 ms 93% 86% 81% 49% 32% 1000 ms 74% 71% 65% 500 ms 71% 63% 57% 200 ms 64% 51% 58% Buffett 3000 ms 88% 82% 92% 44% 33% 1000 ms 93% 79% 92% 500 ms 85% 92% 86% 200 ms 93% 89% 85% Note. The values are based on 72 trials per cond ition with a total of 1,008 trials per subject. Learning by Blocks Learning was assessed for each subject by comparing the percent correct for each of the six test blocks as they progressed with in each of the 14 test conditions (Figure 17). Learning did not appear to occur with Hugh as the individual block accuracy rates had no particular pattern. Buffett demonstrated si milar results for all conditions except the broad-band signals presente d at the 200 ms duration, where some improvement was observed in the 0.2 20 and 6 – 20 kHz conditions.

PAGE 61

48 Figure 17. Percent correct by duration for each block as testing progressed across each frequency stimulus. 200 ms 0.2 20 6 20 0.2 2 3000 ms0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.2 2016 4 0.2 2 6 20 1000 ms 0.2 206 200.2 2 500 ms0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.2 20 6 200.2 2 Frequency (kHz) Frequency (kHz) Hu g h Buffett Percent Correct

PAGE 62

49 Error Distribution The overall error rate, determined from the complete data set collapsed across all conditions, was 29% for Hugh and 19% for Buff ett. The broad-band signal error rate, determined from the same data set excludi ng the tonal signals, was 22% for Hugh and 11% for Buffett. The distribution of errors made by each subject was examined. An overall percent correct and error distribution was determined, however because the subjects performance accuracy was considerably lower with pure tone signals, these data were only derived from the results of the broad-band signal testing (Fi gure 18). Although both subjects had differences in performan ce accuracy, their distributions of location selection were spatially symmetrical. Figure 18. Overall percent correct and distribution of errors using only the results from testing with the broad-band signals. The correct speaker location is notated by double parentheses. Hughs results are always presented below the graph lines in teal and Buffetts are above the lines in maroon. 10050050100 ((45)) 315 270 90 10050050100 45 315 270 ((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 Percent Correct Hugh Buffett

PAGE 63

50 Frequency error distributions were dete rmined by comparing the percent correct and wrong at each speaker location for each of the frequency conditions (Figure 19). The durations were collapsed across the broadband frequencies however the tonal signals were only tested at the 3000 ms duration. Although both subjects had differences in performance accuracy with the broad-band signals their errors were generally consistent, with most equally distributed to the locations adjacent to the correct location. For the pure tone signals, errors were scattered among the locations.

PAGE 64

51 Figure 19. Percent correct and distribution of errors by frequency. The durations were collapsed across the broad-band conditions (top two rows). Tonal conditions are presented in the bottom row. The correct speaker location is notated by double parentheses. Hughs results are always presented below the graph lines in teal and Buffetts ar e above the lines in maroon. 10050050100 ((45)) 315 270 90 10050050100 45 315 270((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 10050050100 ((45)) 315 270 90 10050050100 45 315 270((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 0.2 20 kHz 6 20 kHz 10050050100 ((45)) 315 270 90 10050050100 45 315 270 ((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 16 kHz 4 kHz 0.2 2 kHz Hu g h Buffet t Percent Selected Percent Selected Percent Selected

PAGE 65

52 Duration error distributions were determined by comparing the percent correct and wrong at each speaker location for each of the duration conditions collapsed across the broad-band frequencies (Figure 20). As with the frequency error distributions, the subjects errors were equally distributed to th e locations adjacent to the correct location. Similar results were found when a conditional error distribution compared the percent correct and wrong at each speaker location within the 12 individual broad-band conditions (see Appendix C for all conditional error distributions). Figure 20. Percent correct and distribution of errors by duration using only the results from testing with the broad-band signals. The correct speaker location is notated by double parentheses. Hughs results are always presented below the graph lines in teal and Buffetts are above the lines in maroon. 10050050100 ((45)) 315 270 90 10050050100 45 315 270 ((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 3000 ms 10050050100 ((45)) 315 270 90 10050050100 45 315 270 ((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 1000 ms 10050050100 ((45)) 315 270 90 10050050100 45 315 270 ((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 200 ms 10050050100 ((45)) 315 270 90 10050050100 45 315 270 ((90)) 10050050100 45 315 ((270)) 90 10050050100 45 ((315)) 270 90 500 ms Percent Selected Hu g h Buffet t Percent Selected

PAGE 66

53 Discussion Overall Performance The subjects of this study were able to learn all aspects of the task in the training phases described. The training portion of the study took approximately six months to complete, a bit longer than expected due to so me initial technical problems with the computer programming, the start of the manatee’s spring mating a nd migratory period when they are sexually preoccupied and have a tendency to swim stereo typically, and the presence of a sea turtle temporarily residing in the mana tee’s exhibit. Once these issues were resolved, training was completed rapidly. The testing portion of the study was comp leted in approximately six weeks. Results indicated that the subjects were able to localize all of the si gnals specified within the conditions. These results do not follow the contention that manatees are poor at sound localization as suggested by Ketten (1992). Ketten meas ured the interaural time distance of the manatee and found that the measurements fell significantly below the regression and overlapped that of the pocket gopher, which was found to be incapable of sound localization (Heffner & Heffner, 1992 A). Heffner (1997) has si nce asserted that interaural time distances may not be a good pr edictor of sound localiz ation abilities, and since the results of this study and Gerstein ’s (1999) localizati on study indicate that manatees can localize, other considerations of how this is accomplished should be examined.

PAGE 67

54 The first hypothesis posited that overall performance accuracy would be higher for broad-band signals than for tonal signa ls. This hypothesis was supported with both subjects. Since both types of signals we re only tested at the 3000 ms duration, these results indicated that Buffett ranged betw een 92% and 88% accuracy for broad-band signals, but dropped to 44% and 33% accuracy for tonal signals, likewise, Hugh ranged between 93% and 81% for broad-band signals and 49% and 32% for tonal signals. The second hypothesis stated that perfor mance accuracy would decrease as sound duration decreased. This hypothesis was not supported for Buffett whose overall accuracy was stable (87%, 88%, 88%, and 89% ), but it was for Hugh who showed a drop in overall accuracy (87%, 70%, 64%, and 58%) as signal duration decreased. The discrepancy between the performances of the two subjects is typical of the results found in other sensory studies that have been c onducted with these speci fic animals to date including visual acuity (Bauer et al., 2003), vibrissae tactile sens itivity (Bauer et al., 2005), and auditory evoked potentials (Mann et al., 2005) and sugge sts that individual sensory differences are likely to exist. Similarly, indivi dual variability has been found among dolphins who participated in hearing studies (Ridgway & Carder, 1997; Brill et al., 2001). Both subjects performed above chance leve ls with the 4 and 16 kHz tonal signals, but at a much lower accuracy rate than w ith the broad-band signals. The dominant sounds found in the manatee’s natural habita t, including boat engine noise, conspecific vocalizations, and ambient noise, are typi cally composed of numerous broad-band frequencies. These results suggest that al though manatees can locali ze the tonal signals,

PAGE 68

55 they are better able to localize the broa d-band noises commonly heard in their environment as is typical with many birds and animals (Marler, 1955). Environmental noise outside of the test conditions should be considered. The manatee exhibit background noise was conti nuous and typically below 500 Hz. Masking may have occurred with the lower frequencies, but was unlikely to have occurred at the higher frequencies. Construction noise wa s an unplanned factor that was present throughout the study. The noise level from th e nearby construction was louder on some days than others. Figure 20 illustrated some of the quieter hammering that occurred, however louder construction noise, from larg e machinery such as industrial cranes, cement pump trucks, and gas powered tools, al so transpired. The di fferent frequency and intensity levels did not appear to have an effect on the manatees’ performance. If the background exhibit or construction noise was a f actor that interfered with the subject’s localization ability on some level, the results found in this study might portray an underestimation of their actual abilities. The ambient noise level in the manatee’s natural habitat typically ranges from 1 Hz – 20 kHz a nd from 60 – 90 dB, but can be as loud as 130 dB during heavy rain (Gerstein, 2002). This ever present noise could make localization more difficult by masking sounds su ch as vocalizations from conspecifics, however the results presented here suggest that manatees may be able localize, even with the presence of intermittent noise caused by passing boats. Speaker Artifacts Numerous controls were put in place to avoid the projecti on and recognition of speaker artifact cues, including the incorporation of a 100 ms rise-fall time within signals to eliminate transients, the addition of a +/ 1.5 dB randomization within signal levels to

PAGE 69

56 eliminate intensity level cues, the switchi ng the test signal location during the presentation of the stationi ng tone, and the routine switchi ng of the speaker locations. Analysis of the calibration data showed no obvious temporal or harmonic distortions that might be used as cues. In addition, if the subjects used a frequency or speaker artifact cue, a drop in their perf ormance accuracy would be expected between blocks three and four of each condition (Figure 17). For instan ce, if a particular click was emitted from speaker three and used to identify that speaker’s location in the first three blocks, it would be expected that upon sw itching speaker locatio ns, the subject would continue to select speaker three in its new location and would make incorrect selections. This pattern was not observed a nd the results suggest that the subjects were localizing the actual test signals a nd not artifact cues. Learning by Blocks The third hypothesis asserted that lear ning would occur as blocks of tests progressed within conditions Learning was assessed for each subject by comparing the percent correct for each of the six test blocks as they progressed within each of the test conditions. This hypothesis was not supported as results dem onstrated a random pattern of performance improvement and deteriorati on between blocks. This suggested that learning did not occur as bl ocks progressed with the po ssible exception of Buffett’s performance in the 200 ms condition. Although Bu ffett’s results at the 200 ms duration might provide an indication that learning ha d occurred, the same criterion could be used to suggest that learning was suppressed in the 1000 ms condition. Since each of the blocks consisted of only 12 tria ls, the differences between bl ocks are one to two errors, and this pattern could simply be due to chance.

PAGE 70

57 Although the manatees in this study were asked to localize a variety of sound conditions that they were unfamiliar with, learni ng did not appear to occur as experience with the task increased. It may be that th e broad-band signals used in this study were rather effortless to localize, making it difficult to see improvements after the initial blocks with high accuracy rate s, however learning did not appear to occur with the more difficult tonal signals either. It is possible that adding more blocks within each condition would demonstrate a learning curve. Error Distribution The fourth hypothesis contende d that more erro rs would be made between the 45o and 90 o speakers located on opposite sides of the subject, than between the two 90 o or 45o speakers located parallel with and in front of the subject. This hypothesis was not supported with either subject. Although th e overall error rates w ithin the broad-band conditions were low for both subjects ( 22% for Hugh and 11% for Buffett), their distribution was consistent a nd most errors were equally distributed at the locations adjacent to the correct location. Rather than distinguishing a “l eft vs. right” or “front vs. parallel” strategy, the manatees appeared to use a “nearest neighbor” strategy to localize the broad-band sounds. For the pure tone si gnals, errors were scattered among the locations and no obvious stra tegy could be discerned. Relevance Results from this experiment have pr ovided information about two manatees’ ability to localize specific broad-band and pur e tone signals of diffe rent durations in a controlled environment. This knowledge is important for providing some understanding of how the manatee might dete ct and localize noise from conspecifics and man-made

PAGE 71

58 stimuli such as boats in their natural habita ts. Typical recreationa l boat engine noises are broad-band frequencies that range between 0.01 – 2 kHz, although they can reach as high as 20 kHz (Miksis, 2005; Richardson et al ., 1995). Manatee vocalizations vary from almost pure tones that tend to modulate betw een frequencies to br oad-band noise and have fundamental frequencies that range from 2.5 – 5.9 kHz but can extend to 15 kHz (Nowacek, et al., 2003). The subjec ts in this study were well ab le to localize test signals within these same ranges, suggesting that ma natees can use localization cues as a means to avoid boats and find conspecifics in their environment. The knowledge gained through this study could be advanced by additional behavioral research. Testing with Hugh and Buffett was only conducted within the azimuth plane, and future studies might inves tigate if manatees can also localize sounds within the vertical plane and/or distance di mension. Localization studies conducted with humans suggest that the audito ry system favors binaural cues over spectral shape cues for localization within the azimuth plane, but relies on spectral shape cues for localization within the vertical plane (Middlebrooks & Green, 1991). A minimum audible angle study conducted with bottlenos e dolphins found that the subj ects were able to localize test signal locations within the vertical plan e equally as well as within the azimuth plane (Renaud & Popper, 1975). While the results demonstrated that the subj ects were able to localize, it is unclear if interaural time, intensity level, and/or pha se differences were utilized. Studies with humans have demonstrated that low freque ncy sounds are better lo calized through the use of time of arrival and/or phase difference cu es, while high frequency sounds are localized through intensity level differences (Steve ns & Newman, 1936; Ra yleigh, 1907). Studies

PAGE 72

59 with bottlenose dolphins found that subjects were able to locali ze test signals that ranged 12 from 6 – 20 kHz thru their lower jaw, but used intensity le vel differences to localize signals above 20 kHz (Renaud & Popper, 1975). A study with harbor seals found that the subjects could more readily localize test signals that contained only interaural time differences than those that contained only intens ity level or phase differences (Terhune, 1974). A minimu m audible angle sound localization study conducted with an Indian elephant, one of th e manatee’s closest relatives, found that the elephant only utilized time and intensity leve l difference cues to localize sounds (Heffner & Heffner, 1982). The design of this manat ee study did not specifi cally address what cues were utilized to localize the test sounds presented, howev er several factors should be considered. For instance, th e unique stationing postures de veloped by both subjects when the third test speaker was in troduced and positioned at 0o may have been a way to position themselves at an angle where they could better utilize time of arrival cues. Likewise, because all of the test signals were played at the same spectrum intensity level, those with broader frequency spectra had louder root mean square amplitudes which may have been used to differentiate intensity le vels. Finally, because the manatee’s head is large in comparison to the ear, sound shadowi ng may have caused phase differences that the animals could have utilized as differential cues. The results of this study were based on f our speaker locations, positioned parallel to and in front of the subject, but it is not known if manatees possess the ability to localize sounds in all 360o around them. In addition, the speaker locations were positioned 45o apart. Error distributions suggest ed that the subjects used a “nearest neighbor” strategy to localize the broad-band test sounds. Almost all of the localization

PAGE 73

60 studies conducted with both terrestrial and marine mammals have been minimal audible angle studies, in which the smallest amount of movement of a sound source that can be detected is measured (Mills, 1958). The mi nimum audible angle study design requires that the subject identify a just -detectible change from a part icular reference point. The design used in this study required that the subject locate a sound source relative to his own location. Both designs have resulted in consistent sound localization measurements (Brown, 1994; Brown & May, 1990), however it remains unknown if smaller angles between the speaker locations would result in different performance accuracy rates for manatees. Understanding how the endangered manatee pe rceives its environment is a crucial component to making competent management d ecisions. All of the conservation efforts put in place, including the implementation of boater slow speed zones and manatee preservation areas, have b een based upon field studies th at determine high manatee abundance areas (Reynolds & Wilcox, 1986) and those that focus on boater behaviors (Gorzelany, 2004). Understanding how the manatee’s sensory systems assimilate information and react to environmental stimu li is an important factor that should be considered in conservation management. This research has increased our understa nding of the manatee’s sound localization abilities, but future studies should expand upon these results. Studies which measure minimum audible angles within a 360o area, or investigate localization abilities within the vertical and/or distance planes and measure th e use of interaural time, intensity level, and/or phase differences, or incorporate additional broad-band and pure tone frequencies of various durations and intens ity levels into a similar study would be of great value.

PAGE 74

61 References Ahnelt, P. K. & Bauer, G. B. (2000). Unpublis hed data reported in Ahnelt, P. K. & Kolb, H. (2000). The mammalian photore ceptor mosaic-adaptive design. Progress in Retinal and Eye Research 19, 711-777. Ahnelt, P. K. & Kolb, H. (2000). The ma mmalian photoreceptor mosaic-adaptive design. Progress in Retinal and Eye Research 19, 711-777. Ames, A. L., Van Vleet, E. S. & Reynolds J. E. (2002). Comparison of lipids in selected tissues of the Florida manatee (Order Sirenia) and bottlenose dolphin (Order Cetacea; Suborder Odontoceti). Comparative Biochemistry and Physiology Part B 132, 625-634. Anderson, S. (1970). Directional hearing in th e harbor porpoise, Phocoena phocoena. In G. Pilleri (Ed.), Investigations on Cetacea, Vol II. Berne, Benteli Ag. Au, W. (1993). The Sonar of Dolphins NY: Springer-Verlag. Bachteler, D. & Dehnhardt, G. (1999). Ac tive touch performance in the Antillean manatee: Evidence for a functional differen tiation of the facial tactile hairs. Zoology, 102, 61-69. Bauer, G. B., Colbert, D. E., Gaspard, J.C., & Littlefield, B., Fellner, W. (2003). Underwater visual acuity of Florida manatees ( Trichechus manatus latirostris .). International Journal of Comparative Psychology 16, 130-142. Bauer, G. B., Gaspard III, J. C., Colbert, D.E., Leach, J. B., & Reep, R. (2005, March). Tactile Discrimination of Textures by Florida Manatees, Trichechus manatus latirostris. Paper presented at the 12th annua l International Conference on Comparative Cognition, Melbourne, Florida. Brill, R. L., Moore, W. B., & Dankiewi cz, L. A. (2001). Assessment of dolphin ( Tursiops truncatus ) auditory sensitivity and he aring loss using jawphones. Journal of the Acoustical Society of America 109, 4, 1717-1722. Brown, C. H. (1994). Sound Localization. In R. R. Fay and A. N. Popper (Eds.) Comparative Hearing: Mammals New York, Springer-Verlag, pp. 57-97. Brown, C. H., Beecher, M. D., Moody, D. B., & Stebbins, W. C. (1980). Localization of noise bands by old world monkeys. Journal of the Acoustical Society of America 68, 127-132.

PAGE 75

62 Brown, C. H. & May, B. J. (1990). Sound localiz ation and binaural processes. In M. A. Berkley and W. C. Stebbins (Eds.) Comparative Perception Vol I, New York, John Wiley and Sons, pp. 247-284. Buckstaff, K. C. (2004). Effects of wate rcraft noise on the acoustic behavior of bottlenose dolphins, Tursiops truncatus in Sarasota Bay Florida. Marine Mammal Science 20, 709-725. Bullock, T. H., Domning, D. P., & Best, R. C. (1980). Evoked potentials demonstrate hearing in a manatee ( Trichechus inunguis ). Journal of Mammalogy 61 (1), 130133. Bullock, T. H., O’Shea, T. J., & McClune, M. C. (1982). Auditory evoked potentials in the West Indian manatee (Sirenia: Trichechus manatus ). Journal of Comparative Physiology 148, 547-554. Burda, H., Bruns, V., Muller, M. (1990). Se nsory adaptations in subterranean mammals. In E. Nevo & O. A. Reig (Eds.), Evolution of Subterranean Mammals at the Organismal and Molecular Levels (pp. 269-293). NY: Wiley-Liss. Casseday, J. H. & Neff, W. D. (1973 ). Localization of pure tones. Journal of Acoustical Society of America 54, 365-372. Cohen, J. L., Tucker, G. S., & Odell, D. K. (1982). The photoreceptors of the West Indian manatee. Journal of Morphology, 173, 197-202. Colbert, D.E., Fellner, W., Bauer, G. B., Mani re, C. & Rhinehart, H. (2001). Husbandry and research training of two Florida manatees, Trichechus manatus latirostris Aquatic Mammals 27 (1), 16-23. Coren, S., Ward, L. M., and Enns, J. T. ( 1994). Hearing: Subjective dimensions in sound. In Sensation and Perception 4th Edition, Orlando, Fl, Harcourt Brace & Company, 213-245. Dehnhardt & Kaminski, (1995). Sensitivity of the mystacial vibrissae of harbour seals (Phoca vitulina) for size differences of actively touched objects. The Journal of Experimental Biology 198, 2317-2323. Domjan, M. (1998). The principles of learning and behavior Pacific Grove, CA: Brooks-Cole Publishing Company. Domning, D. P. & Hayek, L. C. (1986). In terspecific and Intraspecific morphological variation in manatees (Sirenia: Trichechus). Marine Mammal Science 2 (2), 87 144.

PAGE 76

63 Don, M. & Starr, A. (1972). Lateralizat ion performance of the squirrel monkey ( Samiri sciureus ) to binaural click signals. Journal of Neurophysiology 35, 493-500. Fellows, B. J. (1967). Change stimulus sequences for discrimination tasks. Psychological Bulletin 67, 87-92. Florida Fish & Wildlife Res earch Institute (2005 A). Annual synoptic surveys http://floridamarine.org/feat ures/view_article.asp?id=15246 Florida Fish & Wildlife Res earch Institute (2005 B). Annual mortality summaries http://www.floridamarine.org/feat ures/category_sub.asp?id=2241. Gaydos, H. F. (1958). Sensitivity in the judge ment of size by finger span, The American Journal of Psychology 71, 557-562. Gellerman, L.W. (1933). Chance orders of al ternating stimuli in visual discrimination experiments. Journal of Genetic Psychology 42, 207-208. Gentry, R.L. (1967). Underwater auditory localization in the California seal lion ( Zalophus californianus ), Journal of Auditory Research 7, 187-193. Gerstein, E. (1999). Psychoacoustic Evaluations of the West Indian manatee (Trichechus manatus latirostris) Unpublished Doctoral Di ssertation, Florida Atlantic University, Boca Raton, FL. Gerstein, E., Gerstein, L., Forsythe, S. & Bl ue, J. (1999). The unde rwater audiogram of the West Indian manatee ( Trichechus manatus ). Journal of the Acoustical Society of America 105, 3575-3583. Gerstein, E. R. (2002). Manate es, bioacoustics and boats. American Scientist 90, 154163. Gorzelany, J. F. (2004). Ev aluation of boater compliance with manatee speed zones along the Gulf coast of Florida. Coastal Management 32, 215-226. Griebel, U., & Schmid, A. (1996) Color vision in the manatee ( Trichechus manatus ). Vision Research, 36, 2747-2757. Griebel, U., & Schmid, A. (1997). Brightness discrimination ability in the West Indian manatee ( Trichechus manatus). Journal of Experimental Biology, 200, 15871592. Hartman, D. S. (1979). Ecology and Behavior of the m anatee (Trichechus Manatus) in Florida. Special publication No. 5 (153 pp.) The American Society of Marine Mammologists.

PAGE 77

64 Heffner, R. S. (1997). Comp arative study of sound localiz ation and its anatomical correlates in mammals. Acta Otolaryngol 532, 46-53. Heffner, R. S. & Heffner, E. H. ( 1982). Hearing in the elephant ( Elephas maximus ): Absolute sensitivity, frequency disc rimination, and sound localization. Journal of Comparative and Physiological Psychology, 96 (6), 926-944. Heffner, R. S. & Heffner, E. H. (1984). Sound localization in large mammals: Localization of complex sounds by horses. Behavioral Neuroscience 98 (3), 541555. Heffner, R. S. & Heffner, E. H. (1985). Sound localization in wild Norway rats ( Rattus Norvegicus ). Hearing Research 19, 151-155. Heffner, R. S. & Heffner, E. H. (1988 A). S ound localization and use of binaural cues by the gerbil ( Meriones unguiculatus ). Behavioral Neuroscience 102 (3), 422-428. Heffner, R. S. & Heffner, E. H. (1988 B). S ound localization acuity in the cat: Effect of azimuth, signal duration and test procedure. Hearing Research 36, 221-232. Heffner, R. S. & Heffner, E. H. (1988 C). Sound localization in a predatory rodent, the Northern Grasshopper mouse ( Onychomys leucogaster ). Journal of Comparative Psychology 102 (1), 66-71. Heffner, R. S. & Heffner, E. H. (1989). Sound localization, use of bi naural cues and the superior olivary complex in pigs. Brain Behavior Evolution 33, 248-258. Heffner, R. S. & Heffner, E. H. (1992 A). Evolution of sound localization in mammals. In, D. Webster, R. Fay and A. Popper (Eds.) The Biology of Hearing SpringerVerlag, pp. 691-715. Heffner, R. S. & Heffner, E. H. (1992 B). Hearing in large mammals: Sound localization acuity in cattle (Bos Taurus ) and goats (Capra hircus ). Journal of Comparative Psychology 106 (2), 107-113. Heffner, R. S. & Heffner, E. H. (1990). Ve stifial hearing in a fossorial mammal, the pocket gopher ( Geomys bursarius ). Hearing Research 46, 239-252. Heffner, R. S. & Masterson, R. B. (1990). Sound localization in mammals: Brain stem mechanisms. In M. A. Berkley, & W. C. Stebbins (Eds.) Comparative Perception New York, John Wiley and Sons, pp. 285-314. Holt, M. M., Schusterman, R. J., Southall, B. L., & Kastak, D. (2004). Localization of aerial broadband noise by pinnipeds. Journal of the Acoustical Society of America 115, 2339-2345.

PAGE 78

65 Houben, D. & Gourevitch, G. (1979). A uditory lateralization in monkeys: An examination of two cues serving directional hearing. Journal of Acoustical Society of America 66, 1057-1063. Isley, T. E. & Gysel, L. W. (1975). Sound source localiza tion by the red fox. Journal of Mammalogy 56, 397-404. Ketten, D. R., Odell, D. K., Domning, D. P. (1992). Structure, function, and adaptation of the manatee ear. In J. A. Thomas, R. A. Kastelein, & A. Y. Supin (Eds.) Marine Mammal Sensory Systems (pp.77-79). NY: Plenum Press. Kirkpatrick, B., Colbert, D. E., Dalpra, D., Newton, E. A. C., Gaspard, J., Littlefield B., & Manire, C. A. (2002). Florida Red Ti des, Manatee Brevetoxicosis, and Lung Models. In K. A. Steidinger, J. H. La ndsberg, C. R. Tomas, and G. A. Vargo (Eds.) Xth International Conference on Harmful Algae. Harmful Algae Florida Fish and Wildlife Conservation Commission and Intergovernmental Oceanographic Commission of UNESCO. Klishen, V. O., Diaz, R. P., Popov, V. V., & S upin, A. Y. (1990). Some characteristics of hearing of the Brazilian manatee, Trichechus inunguis Aquatic Mammals 16 (3), 139-144. Manire, C. A., Walsh, C. J., Rhinehart, H. L., Co lbert, D. E., Noyes, D. R., & Luer, C. A. (2003). Alterations in blood and urine parameters in two Florida manatees, Trichechus manatus latirostris from simulated conditi ons of release following rehabilitation. Zoo Biology 22, 103-120. Mann, D. Personal Communication, August 2005. Mann, D., Colbert, D. E., Gaspard, J. C. III, Casper, B., Cook, M. L. H., Reep, R. L., & Bauer, G. B. (2005). Temporal re solution of the Florida manatee ( Trichechus manatus latirostris ) auditory system. Journal of Comparative Physiology 191, 903-908. Marler, P. (1955). Some charac teristics of animals calls. Nature 176, 6, 6-8. Marshall, C. D., Huth, G. D., Edmonds, V. M. Halin, D. L., & Reep, R. L. (1998). Prehensile use of perioral bristles duri ng feeding and associated behaviors of the Florida manatee (Trichechus manatus latirostris). Marine Mammal Science 14, 274-289. Marshall, C. D., Maeda, H., Iwata, M., Furu ta, M. Asano, A., Rosas, F., & Reep, R. L. (2003). Orofacial morphology and feedi ng behavior of the dugong, Amazonian, West African, and Antillean manat ees (Mammalia: Sirenia): Functional morphology of the muscular-vibrissal complex. Journal of Zoology 259, 245260.

PAGE 79

66 Mass, A. M., Odell, D. K., Ketten, D. R., & Supin, A. Y. (1997). Ganglion layer topography and retinal resoluti on of the Caribbean manatee Trichechus manatus latirostris Doklady Biological Sciences, 355, 392-394. Masterson, B., Thompson, G. C., Bechtold, J. K. & RoBards, M. J. (1975). Neuroanatomical basis of binaural phase-difference analysis for sound localization: A comparative study. Journal of Comparative Physiological Psychology 89, 379-386. Medwin, H., & Clay, C. S. (1998). Fundamentals of Acoustical Oceanography NY: Academic Press. Middlebrooks, J. C. & Green, D. M. (1991) Sound localization by human listeners. Annual Review of Psychology 42, 135-159. Miksis-Olds, J. L. (2006). Manatee Res ponse to Environmental Noise. Unpublished Doctoral Dissertation. University of Rhode Island, Narragansett, RI. Mills, A. W. (1958). On the minimum audible angle. Journal of the Acoustical Society of America 30, 127-246. Mills, A. W. (1972). Auditory localization. In J. V. Tobias (Ed.) Foundations of Modern Auditory Theory: Vol. II, (pp. 3030-348). NY: Academic Press Moore, P. W. B. (1974). Underwater loca lization of click and pul sed pure-tone signals by the California sea lion ( Zalophus californianus ). Journal of the Acoustical Society of America 57 (2), 406-410. Moore, P. W. B. & Au, W. L. (1975). U nderwater localization of pulsed pure tones by the California sea lion ( Zalophus californianus ). Journal of the Acoustical Society of America, 58 (2), 721-727. Moore, P. W. B. & Brill, R. L. (2001). Binaural hearing in dolphins. The Journal of the Acoustical Society of America 109 (5), 2330-2331. Moore, P. W. B. and Pawloski, D. A. (1993) Interaural time discrimination in the bottlenose dolphin. The Journal of the Acoustical Society of America, 94 (3), 1829-1830. Nowacek, D. P., Casper, B. M., Wells, R. W., Nowacek, S. M., & Mann, D. A. (2003). Intraspecific and geographic variat ion of West Indian manatee ( Trichechus manatus spp .) vocalizations Journal of the Acoustical Society of America 114 (1), 66-69.

PAGE 80

67 Nowacek, S. M., Wells, R. S., Owen, E. C. G., Speakman,T. R., Flam, R. O., & Nowacek, D. P. (2004). Florida manatees, Trichechus manatus latirostris respond to approaching vessels. Biological Conservation 119, 517-523. Odell, D. K., & Reynolds, J. E. III. (1979). Observations on manatee mortality in South Florida. Journal of Wildlife Management 28 (3), 572-577. Pepper, R. L., & Defran, R. H. (1975). Dolphin Trainer’s Handbook, Part 1: Basic Training (50p). Naval Undersea Center, San Diego, CA. Piggins, D. J., Muntz, W. R. A., & Best R. C. (1983). Physical and morphological aspects of the eye of the manatee, Trichechus inunguis Marine Behaviour and Physiology 9, 111-130. Popov, V. V. & Supin, A.Y. (1990). Electrophysiological studies on hearing in some cetaceans and a manatee. In J.A. Thomas, & R. A. Kastelein (Eds.), Sensory Abilities of Cetaceans: Laboratory and Field Evidence NY: Plenum Press. Rayleigh, L. (1907). On our perception of sound direction. Philosophical Magazine 13 (6), 214-232. Ramirez, K (1999). Animal Training: Successful animal management through positive reinforcement. Shedd Aquarium, Chicago, Illinois Reep, R. L. Marshall, C. D., Stoll, M. L., Home r, B. L. & Samuelson, D. A. (2001). Microanatomy of facial vibrissae in the Florida manatee: the basis for specialized sensory function and oripulation. Brain Behavior and Evolution 58, 1-14. Reep R. L. Marshall, C. D., & Stoll, M. L. ( 2002). Tactile hairs on the postcranial body in Florida manatees: a mammalian lateral line? Brain Behavior and Evolution 59, 141-154. Renaud, D. L., & Popper, A. N. (1975). Sound localization by the bottlenose porpoise Tursiops truncatus Journal of Experimental Biology 63, 569-585. Reynolds, J. E. III. (1979). The semisocial manatee. Natural History 88 (2), 44-53. Reynolds, J. E. & Odell, D. K. (1991). Manatees and Dugongs (192 pp.). NY: Checkmark Books. Reynolds, J. E. III, & Wilcox, J. R. (1986). Distribution and abundance of the West Indian manatee, Trichechus manatus, around selected Florida power plants following winter cold fronts: 1984-1985. Biological Conservation 38, 103-113. Richardson, W., Greene, C., Malme, C. & Thompson, D. (1995). Marine mammals and noise San Diego, CA: Academic Press.

PAGE 81

68 Ridgway, S. H. & Carder, D. A. (1997). Hearing deficits measured in some Tursiops truncatus and discovery of a deaf/mute dolphin. Journal of the Acoustical Society of America 101, 590-594. Ross, D. (1976). Mechanics of Underwater Noise. NY: Pergamon. Stevens, S. S. & Newman, E. B. (1936). The localization of actual sources of sound. American Journal of Psychology 48, 297-306. Terhune, J. M. (1974). Dir ectional hearing of a harbor seal in air and water Journal of the Acoustical Society of America, 56 (6), 1862-1865. Urick, R. J. (1996). The principles of underwater sound, 3rd Edition (423pp.). CA: Peninsula Publications. U.S. Fish and Wildlife Service (2001). Florida Manatee Recovery Plan, (Trichechus manatus latirostris), Third Revision U.S. Fish and Wildlife Service, Atlanta, Georgia. 144 pp. Wakeford, O. S. and Robinson, D. E. (1974). Lateralization of t onal stimuli by the cat. Journal of the Acoustical Society of America 55, 649-652. Walls, G. L. (1967). The vertebrate eye and it s adaptive radiation NY: Hafner. West, J. A., Sivak, J. G., Murphy, C. J., & Kovacs (1991). A comparative study of the anatomy of the iris and ciliary body in aquatic mammals. Canadian Journal of Zoology, 69, 2594-2607. Yost, W. A. (2000). Sound localiz ation and binaural hearing. In Fundamentals of hearing: An introduction (pp. 179-192). San Diego, CA: Academic Press.

PAGE 82

69 Appendices

PAGE 83

70 Appendix A: The Computer Protocols Used For all P hases of the Experi mental Conditions A graphical user interface, programmed in Visual C was designed to run each phase of the experimental conditions (F igure B1). A drop-down subject menu was designed to distinguish which subject was being tested, and this selection automatically referenced and played that animal’s stat ioning and reinforcement tones throughout the block. A “notes” section allowed any comments to be digitally recorded relative to that block. The “set-up” section defined how many speak er locations were to be tested, how many trials were to be run from each of thos e speakers, and how many of the test sounds could be played from the same location in a row. In addition, broa d-band noise bursts or tonal signals were defined as were the fre quency range to be tested, the sound duration, the dB level (always se t at 3dB), and if the sounds were to be automatically digitally recorded. The “speaker” section provided informa tion about which speaker location each test sound was played from (0 was the 900 location, 1 was the 2700 location, 2 was the 3150 location, and 3 was the 450 location). If needed, a ma nual switching check box was included, which allowed the Data-recorder to se lect the location of the test sound to be played, rather than the randomized location generated by the program. The “status” section defined and digitall y recorded how many trials had been completed within the block, and of those, how many were correct and how many were wrong. The start button initiated the block of twelve trials once the subject and

PAGE 84

71 conditions were defined, and th e stop button was used only if the block had to be ended prior to the completion of the twelve trials. Figure A1: The graphical user interf ace screen (programmed in Visual C) used to setup the experimental conditions and automatically download the results into an Excel file during the testing sessions. The correct experimental conditions needed to be incorporated for each portion of the session, including the warmup, practice, testing, and cooldown trials (a total of 56 trials were run per animal per session). In all portions of a session, four speaker locations, a maximum of two trials in a row per location, and a randomized level of three dB were held constant. In the warm-up trials, two trials were set up per speaker for a tota l of eight trials. The noise button was selected and the fre quency range was defined from 20,000-200 Hz. The sound duration was defined as 3 seconds.

PAGE 85

72 In the practice trials, two trials were set up per speaker for a tota l of eight trials. Either the noise or tone button was se lected depending on which sound condition was being tested in that session. If a broad-ba nd noise was being tested, the frequency range was defined. If a pure tone was being teste d, the single frequency level was entered in the low-pass frequency box and a zero was en tered in the high pass frequency box. The sound duration was defined according to which c ondition was being tested in that session. In the testing trials, all of the settings defined in the practice trials were maintained except for the number of trials pe r speaker, which was changed from two to three, for a total of twelve trials. These settings were maintained for a total of three testing blocks. In the cool-down trials, one trial was set up per speaker for a tota l of four trials. The noise button was selected and the fr equency range was defined from 20,000 to 200 Hz. The sound duration was defined as 3 seconds.

PAGE 86

73 Appendix B: Data-recording Protocols Used to Document Each Session on a Tank-side Data Sheet. All of the session’s general information was documented on the data sheet (Figure B1). This included the date and who the Test Trainer, Data-recorder, and Stationing Trainer were. Specific information was docum ented for all for portions of the session (warm-up, practice, three test blocks and th e cool-down trials) per subject including frequency(ies) and sound duration levels, star t and end time, the location of each trials’ test sound, if the subject was correct or inco rrect, and if incorr ect – the location the subject erroneously selected. Additional information was included for each test block including the video tape numb er and counter start and stop times, the number of times the test subject left or attempted to leave in that block, the number of times the test subject was interrupted by the other animal, the amount of time the other animal was on task, and the test subject’s behavioral rating from a s cale of one to five, where one indicated that the animal did very poorly and was not able to complete the task a nd five indicated that he did an excellent job. A comment sect ion was also provided to add additional information if needed.

PAGE 87

74 Figure B1. The tank-side data-recording sh eet used to document each session.

PAGE 88

75 Appendix C: Error distribution within the 12 broad-band conditions A conditional error distribution was determined by comparing the percent correct to the percent wrong at each speaker loca tion within the 12 individual broad-band conditions. The error distribution is shown fo r all of the duration conditions within the 0.2 20 (Figure C1), 6 15 (Figure C2), a nd 0.2 2 kHz (Figure C3) broad frequency conditions. 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 3,000 ms 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 1,000 ms 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 200 ms 500 ms Percent Selected Percent Selected

PAGE 89

76 Figure C1. Percent correct and distribution of errors by duration within the 0.2 20 kHz condition. The correct speaker location is notated by double parenthesis. Hughs results are always presented below the graph lines in teal and Buffetts are above the lines in maroon. Figure C2. Percent correct and distribution of errors by duration within the 6 20 kHz condition. The correct speaker location is notated by double parenthesis. Hughs results are always presented below the graph lines in teal and Buffetts are above the lines in maroon. 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 903,000 ms 1,000 ms 200 ms 500 ms Percent Selected Percent Selected

PAGE 90

77 Figure C3. Percent correct and distribution of errors by duration within the 0.2 2 kHz condition. The correct speaker location is notated by double parenthesis. Hughs results are always presented below the graph lines in teal and Buffetts are above the lines in maroon. 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 100500050100 ((45)) 315 270 90 100500050100 45 315 270 ((90)) 100500050100 45 315 ((270)) 90 100500050100 45 ((315)) 270 90 Percent Selected Percent Selected 3,000 ms 200 ms 500 ms 1,000 ms


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001709499
003 fts
005 20060614112122.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 060509s2005 flua sbm s000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001361
035
(OCoLC)68570302
SFE0001361
040
FHM
c FHM
049
FHMM
090
BF121 (Online)
1 100
Colbert, Debborah E.
0 245
Sound localization abilities of two Florida manatees, trichechus manatus latirostris
h [electronic resource] /
by Debborah E. Colbert.
260
[Tampa, Fla.] :
b University of South Florida,
2005.
502
Thesis (M.A.)--University of South Florida, 2005.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 90 pages.
520
ABSTRACT: Florida manatees (Trichechus manatus latirostris) live in the shallow, often turbid inland and coastal waters of the southeastern United States. Since their vision is poor (Bauer et al., 2003), other senses probably guide orientation. Previous studies have found that manatees can hear over 40 kHz (Gerstein et al., 1999) and have the capacity for rapid auditory temporal processing (Mann et al., 2005). However, it is not known if manatees have the ability to localize underwater sounds. Two Florida manatees were trained to identify underwater sound source locations using a four-choice discrimination paradigm. Three broad-band signals ( 0.2 20, 6 20, and 0.2 2kHz) were tested at four durations (3,000, 1,000, 500, and 200ms) and two tonal signals (4 and 16kHz) were tested with a 3,000ms duration. A total of 1,008 test trials were analyzed per subject.Both manatees learned the task easily, and could localize all of the test signals at a performance rate well above the 25% chance level. Within all of the broad-band conditions, performance accuracy ranged from 93% 79% for Buffett, and 93% 51% for Hugh. Broad-band signal duration did not have an effect on performance accuracy with Buffett who ranged from 89% to 87%, but did with Hugh who ranged from 87% 58%. Broad-band frequency type did not have an effect on performance accuracy with Buffett who averaged 90%, 86%, and 89%, but may have with Hugh who averaged from 76%, 68%, and 65% at the 0.2 20, 6 20, and 0.2 2 kHz conditions. Both animals performed above chance levels with the pure tone signals, but at a much lower accuracy rate with Hugh at 49% and 32% and Buffett at 44% and 33% with the 4 kHz and 16 kHz conditions.Results from this experiment provide information about the manatees ability to localize different types of sounds in a controlled environment. This knowledge is important for understanding how manatees detect and localize noise generated from conspecifics and boat engines and contributes to making competent conservation management decisions about these endangered marine mammals.
590
Adviser: Sarah Partan, Ph.D.
653
Audition.
Sirenian.
Hearing.
West Indian manatee.
Localization.
690
Dissertations, Academic
z USF
x Psychology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1361