DOLPHIN SOUND AND COMMUNICATION ABSTRACTS

ODONTOCETE SOUND AND COMMUNICATION ABSTRACTS - INDEX
(From: WMMSC, Monaco 1998, Abstract Volume)

Others

COMPARISON BETWEEN RISSO'S DOLPHIN VOCAL BEHAVIOUR

THE VOCAL REPERTOIRE OF THE BELUGA WHALE

Mysticeti (Baleen whales)

EVIDENCE FOR THE SOCIAL FUNCTION OF THE HUMPBACK WHALE SONG

Sonar

CLICK INTERVALS OF BAIJI AND FINLESS PORPOISE

REAL-TIME MEASUREMENT OF THE ECHOLOCATION SIGNALS OF WILD STENELLA F.

DISCRIMINATION OF ACOUSTICALLY SIMULATED AND REAL UNDERWATER TARGETS BY DOLPHINS

SONAR RELEVANT STRUCTURES IN THE FETAL NARWHAL. (MONODON MONOCEROS)

(INDEX)

JOURNEY INTO THE TOOTHED WHALE'S HEAD: STRUCTURE AND FUNCTION OF THE SONAR APPARATUS by Dr. Ted W. Cranford

Where: NASA Ames Research Center Moffett Field, CA Numerical Aerodynamic Simulation Facility (N-258) Room 127 Date: Tuesday, December 16, 1997 Time: 12:00 Noon Point-of-Contact: Ron Reisman rreisman@mail.arc.nasa.gov 650-604-1459

NOTE: If you are planning to attend, ESPECIALLY IF YOU ARE NOT A US CITIZEN AND DO NOT HAVE PERMANENT RESIDENT STATUS, please contact Mr. Reisman and make arrangements for entry to NASA Ames.

Discovery of the sonar signal generator(s) within the dolphin nasal apparatus has eluded cetologists and bioacousticians for more than three decades. Recent interest in developing a bionic sonar system has reinvigorated the effort to characterize the apparatus and its operation in all toothed whales.

The anatomy in the heads of several toothed whales were collected using modern medical imaging techniques (CT and MR). The head of a neonate sperm whale was too large for these devices and had to be CT scanned in the non-destructive testing facility at the Naval Air Warfare Center in China Lake, CA. All these images show an anatomic complex that is thought to be the site of sonar signal generation.

Functional verification of this location was conducted using high-speed video endoscopy within the nasal airway of a dolphin performing an echolocation task. A visual demonstration of the synchronous activity at the internal "monkey lips"(dorsal bursae) with signal repetition rate will be presented.

The probable site of sonar signal generation in a sperm whale will be shown using a dynamic computer reconstruction of the anatomy inside the whale's head.

The U.S. Navy has been recording the sonar signals of dolphins and conducting psychoacoustic experiments of their sonar capabilities for the past three decades. As a complement to the Navy's "skin-out" biosonar research, Dr. Cranford works on what he calls, "skin-in" echolocation, using various imaging and computer techniques to investigate the source of odontocete sonar signals. He produced quantitative descriptions of the biosonar anatomy in a wide array of odontocete taxa (porpoises to sperm whales) and provided the essential geometric components for modeling the echolocation system. Subsequently, he identified and verified the biosonar source(s) using high-speed video endoscopy and computer modeling techniques.

Dr. Cranford's s research is concerned with: (1) testing his "unified hypothesis," which states that all odontocetes produce biosonar signals by the same mechanism, (2) using computers to simulate sound generation, propagation and reception in toothed whales, (3) recording and analysis of the loud impulse sounds produced by odontocetes and their potential role in acoustic prey debilitation, and (4) studying the functional implications and phylogenetic pathways of (directional) cranial asymmetry in odontocetes.

******************************************************

Ted Cranford received a Ph.D. in Biology from the University of California at UC Santa Cruz in 1992, where he produced the first quantitative description of the biosonar anatomy in the heads of toothed whales (odontocetes).

He was awarded a three year National Research Council fellowship to work at the Naval Research and Development laboratory in Hawaii and continues now as a consultant to the Navy's SPAWAR laboratory in San Diego, California.

His work has been recognized by an international group of his peers with two Outstanding Paper Awards and a special commendation for Innovative Research.

(INDEX)

CLICK INTERVALS OF BAIJI AND FINLESS PORPOISE IN SHISHOU SEMI NATURAL RESERVE AND IN A TANK

( 1 ) Akamatsu, T. and (2) Wang, D.

(I) National Research Institute of Fisheries Engineering, Hasaki, Kashima, Ibaraki 3l4-04, Japan (2) Institute of Hydrobiology, The Chinese Academy of Sciences, Wuhan, 430072, China

It is known that click intervals of captive dolphins are larger than two-way transiting time at target discrimination task. Also, free-ranging dolphins are supposed to change their navigation range of echolocation in the wild. Click intervals of free-ranging baiji and finless porpoises were observed in Shishou Semi-natural Reserve adjacent to the Yangtze River in China and in a tank. A baiji in a l3 meter circular tank frequently produced 26 ms to 28 ms click intervals and 90 % of 329,940 intervals were less than 90ms. Two finless porpoises in a rectangular tank (8 m x 5 m x 2 m) frequently produced 8 ms to 10 ms click intervals and 90 % of 36,647 intervals were Iess than 18ms. On the other hand, a baiji and finless porpoises in the Semi-natural Reserve (21 km in length and I-2 km in width) produccd a wide variety of click intervals, up to 286 ms (90 % of 14,294 intervals from baiji) and 276 ms (90% of 2,506 intervals from finless porpoise), respectively. Reverberated pulse signals were mostly recorded within 2 ms after direct path signals in both the tank and the Reserve, so that click intervals smaller than 2 ms were not analyzed. Also, most sound pressure levels of reverberated signals were smaller than the threshold Ievel (127 dB re 1uPa) of the data acquisition system. Two-way transiting times of 26 ms and 286 ms correspond with distances of 20 m and 215 m, respectively. Navigation or target detection range of echolocation are thought to be less than these distances. Baiji and finless porpoises in a small enclosure only used short range adapted echolocation signals. In the Reserve, their echolocation signals were adapted for a wide variety of distances.

(INDEX)

RHYTHMIC SIGNATURE IN SPERM WHALE CLICK TRAINS: A FUNCTION OF INDIVIDUAL IDENTIFICATION AND COMMUNICATION

( 1 ) Andre, M. and (?) Kamminga, C.

(l) Dept. of Biology. University of Las Palmas de Cran Canaria. Spain (1) Information Theory Group, TU Delft, The Netherlands

Sperm whale (Physeter macrocephalus) usual click trains have been generally restricted to echolocation functions, while codas are thought to serve as the main communicative support within the sperm whale acoustic repertoire. To test if a specific structure was present in the series of usual clicks. cross-correlation techniques were used to separate the clicks of four individuals out of an 85 seconds duration sequence of usual clicks. Of each isolated sequence, a click repetition pattern is constructed, showing a more or less random succession of click intervals with an average click repetition rate ranging from 1.44-1.93 Hz. If this sequence is, however, considered as a random event process and a Fourier transform is applied to this series of -functions, we note a strong modulating frequency in the spectrum of each sperm whale. This deterministic modulation in the order of 0.111-0.167 Hz suggests a cohesive longitudinal structure in their vocalizations along the whole recorded click sequence. The data strongly support the hypothesis that a continuous rhythmic signature is present in the series of usual click trains, thus allowing the individual identification of the whales. The polyrhythmic structure of the repertoire of a social group of sperm whales must carry all the information necessary to maintain the observed cohesion of the group when foraging, thus giving to the series of usual clicks an important communication function.

(INDEX)

BEHAVIOURAL SIGNIFICANCE AND STRUCTURE OF WHISTLES FROM WILD KILLER WHALES (ORCINUS ORCA) IN THE WATERS AROUND VANCOUVER ISLAND, BRITISH COLUMBIA

(1) Thomsen,F., (2) Ford, J.K.B.

(!) Zoologie. Universität Hamburg, 20146 Hamburg, Germany

(2) Vancouver Aquarium. Vancouver. BC, Canada V6B 3X8

Killer whales (Orcinus orca) off Vancouver Island. British Columbia, produce two classes uf sounds which function in underwater communication: whistles and burst pulsed calls. As yet research on killer whale communication has focused mainly on sterotyped, pod specific burst pulsed calls. Recent studies on whistles indicated that they are variable and graded. furthermore whistles ar most frequent during social interactions within the pod. This Ied to the conclusion that whistles are emotive signals indicating the emotional state uf the signaller. Despite this first results there are many questions remaining on the function of whistles. It is unknown if whistles appear only in interactions within the pod or also when different pods meet. Further the analysis of whistle structure is still in its infancy. In this study the behavioural significance and structure of whistles are analysed in more detail. Simultaneous underwater recordings and surface behavioural observations were made on resident killer whales in the Johnstone Strait, British Columbia in 1996. Earlier sound and behavioural recordings from l978-94 were included in the analysis. Whistles were investigated with the RTS and SIGNAL computer programs for sound analysis. Whistle emission is highest not only during social interactions within members of thc same pod but also during interactions of members from different pods. Most whistles appear to be variable. However some stereotyped whistle types could be identified which are stable over more than one decade. This whistle types are emitted in regular series, similar to graded vocal emissions in other mammalian species. Whistle types are pod specific. We conclude from this results that pod specific, graded whistles in wild killer whales serve a similar function as graded close range emissions in other mammalian species. Whistles probably indicate internal emotional states (level of arousal, friendliness, aggressiveness) of the signaller. Whistling thus coordinates close range interactions in killer whales.

(INDEX)

THE VOCAL REPERTOIRE OF THE BELUGA WHALE IN BRISTOL BAY, ALASKA

( I ) Angiel, N.M., (2) DeMaster, D.P. and (3) VanBlaricom, G.R.

The vocalizations of the Alaskan beluga whale (Delphinapterus leucas) stocks have never been catalogued and little is known about the acoustic behavior of these groups. This study characterizes the acoustic repertoire of the Bristol Bay beluga whale population, based on underwater vocalizations recorded in July 1995. The aim of this research was to find a repeatable and consistent classification process for beluga whale vocalizations. With such a process, it would be possible to compare the vocal repertoires of different beluga whale populations without fear of observer bias. A study comparing the vocalizations of different Alaskan beluga stocks might yield evidence of geographic variation and dialects and could help to elucidate stock structure. The repertoire was classified by two independent bioacousticians in order to examine the subjectivity and bias of the individual observer. Cluster analysis was proposed as a consistent and repeatable method to classify vocalizations, using time and frequency measurements from spectrograms. The results contribute to our understanding of beluga acoustic behavior and demonstrate the inconsistency of current classification methods. The two independent observers found significantly different numbers of vocalizations and. the cluster analysis failed to provide an unbiased and consistent method of repertoire classification.

(INDEX)

REAL-TIME MEASUREMENT OF THE ECHOLOCATION SIGNALS OF WILD DOLPHINS USING A 4-HYDROPHONE ARRAY

(1) Au, Whitlow W. L., (2) Herzing, Denise L., ( I ) Aubauer, Roland

(1) Hawaii Institute of Marine Biology. P. O. Box 1106, Kuilua, HI 96734 USA

(2) Wild Dolphin Project. P.O. Box 8436, Jupiter, FL 33468 USA

An array of four hydrophones arranged in a symmetrical star configuration was used to measure the echolocation signals of Atlantic spotted dolphins (Stenella Frontalis) in the Bahamas. The spacing between the center hydrophone and the other hydrophones was 45.7 cm. A four-channel simultaneous analog-to-digital (A/D) data acquisition system operating at 500. kHz and controlled by a personal computer was used to detect, digitized and store the echolocation signals. A video camera was attached to the array and a video tape recorder time synchronized to the computer. The video camera was used to ascertain the aspect of the dolphins to the array. The array was either suspended from the boat or held by a skin diver and the acoustic and video signals were cabled back to a boat. The echolocation signals had bi-modal frequency spectra with a low-frequency peak between 40-50 kHz and a high-frequency peak between 130-140 kHz. The low-frequency peak dominated when the signal source level (signal amplitude 1 m from the dolphin) was low and the high-frequency peak dominated when the source level was high. The specific frequency of the low and high frequency peaks also depended on the source level. Simultaneous A/D sampling was necessary to determine arrival time differences between hydrophones so that the distance of the echolocating dolphins along with the source levels could be accurately calculated. Peak-to-peak source levels as high as 210 dB were measured. The characteristics of the signals are similar to those of Tursiops truncatus and Pseudorca crassidens measured in open waters under controlled conditions. The technique of using a multi-element array system along with video recording is an excellent method to accurately measure echolocation signals of f'ree-swimming dolphins.

(INDEX)

DISCRIMINATION OF ACOUSTICALLY SIMULATED AND REAL UNDERWATER TARGETS BY DOLPHINS

Aubauer, R. , Au, W. W. L. , Pawloski, J. L. , Pawloski, D. A. , Nachtigall, P. E.

The complex interdependence of the physical dimensions and the reflection characteristics of underwater targets makes it diffcult to extract relevant echo parameters that dolphins use for discrimination or classification. For that reason a target simulator was developed that acoustically simulates the echoes of underwater targets. Dolphin echolocation clicks are received with a hydrophone and fed into the target simulator, where they are transformed in real time with the fast convolution method into'phantom' echoes. The backscattering process of an underwater target is described by the target impulse response, which is independent of the incident signal. The phantom echo is played back to the animal after a preset time delay relative to the signal recording. The target simulator is implemented on a digital signal processing unit, which is programmable over a connected personal computer. The signals are sampled with 1MHz and 12Bit resolution. The acoustic simulation of several 7.6cm spheres has shown very good agreement with the original echoes measured in a tank (cross-correlation coefficients larger than 0.98). In a behavioral experiment an echolocating dolphin was trained to discriminate between several 7.6cm metal spheres and their phantom echo replicas. By changing the impulse response of a simulated target the intemal structure of the phantom echoes can be manipulated, so that echo parameters that force the dolphin's decision can be distinctly isolated from each other. This new method gives precise control over the stimuli parameters in a dolphin behavioral experiment and thus allows the development of an understanding of the process of echolocation discrimination.

(INDEX)

ACOUSTIC CENSUS OF SPERM WHALES IN THE EASTERN TEMPERATE NORTH PACIFIC

Barlow. Jay and Taylor, Barbara L.

Southwest Fisheries Science Center. Nationa! Marine Fisheries Service, P.O. Box 27l. La Jolla. Cf USA

A line-transect survey was conductcd from 9 March to 9 June 1997 to estimate the abundance of sperm whales in the eastem temperate North Pacific between the west coast of the United States and Hawaii using a 52 m research vessel. Acoustic methods were used concurrently with visual methods during daylight hours and were also used at night. A hydrophone array was towed at l00m depth as the vessel surveyed approximately 25,000 km of predetermined transect lines at eight knots. Acoustic signals received by a pair of hydrophones were digitized and sent up to the ship via a coaxial tow cable. Bearing angles to the locations of sperm whales were calculatcd in real-time based on differences in signal arrival times between the two hydrophones and whale locations were determined by the convergence of bearing lines as the ship continued along its course. The ship was maneuvered to resolve the left/right ambiguity in whale locations and, during daylight hours, was directed to this estimated location to obtain visual estimates of group size when the whales surfaced. A total of 102 distinct groups of sperm whales were acoustically detected with sufficient amplitude to estimate bearing lines. Whale locations were estimated for 51 of these detections. Groups sizes ranged from one to approximately 100. "Slow clicks" (with a period of 4-8 sec) were detected at ranges up to 20 nmi; whereas the more typical sperm whale clicks (with a period of 0.5-l.0 sec) were typically heard at less than 5 nmi. Only two sperm whale detections were made first by the visual team and. in both cases, whales were heard shortly after being seen. Acoustic techniques substantially increased the number of sperm whales detected on this line-transect survey and allowed the detection of submerged whales that would have otherwise been missed by visual observers; however, acoustics could not replace visual methods for estimating group size.

(INDEX)

COMPARISON BETWEEN RISSO'S DOLPHIN VOCAL BEHAVIOUR IN SCOTTISH WATERS AND IN THE MEDITERRANEAN SEA (1) Benoldi, C., (2) Gill'A., (3) Evans, P.G.H., (4) Manghi, M., (5) Pavan, G. and (6) Priano M.

(I) Department of Zoology, University of Milano, Italy (2) The Western Isles Risso's Dolphin Project, P. O. Box 9902. Stornoway, Isle of Gewis. HS2 OHQ, Scotland. (3) Department of Zoology, University of Oxford U.K. (./,5,6) Centro Interdisciplinare di Bioacustica e Ricerche Ambientali, University of Pavia, Italy.

During summer 1996 recordings of Risso's dolphin (Grampus griseus) vocalisations were gathered for The Risso's Dolphin Project in Scottish waters of the Isle of Lewis, Outer Hebrides, to understand more about this elusive species. Comparison with recordings of the same species, collected in previous years in the Meditenanean sea, belonging to the Sound Library of the Centro Interdisciplinare di Bioacustica e Ricerche Ambientali, University of Pavia, Italy, was made to find differences and similarities between these two different areas. Risso's dolphins were sighted and recorded for 12 days in the coastal waters of the Isle of Lewis, collecting about seven hours of useful recordings. Four sightings in the Mediterranean resulted in about two hours recordings collected. ln Scotland two stereo hydrophones plunged from a small survey boat were used, while a dipole-array of hydrophones was towed during Mediterranean cruises. Digital recordings were analysed with the real-time Digital Signal Processing Workstation developed by Dr. G. Pavan, except for the analysis of whistles recorded in Hebridean waters that was done using Canary v. l.2 (Comell University's Bioacoustics Laboratory). At first all the vocalisations collected in Scotland, analysed by frequency, duration and repetition rate, have been grouped in categories in a catalogue with their relative spectrograms. In addition analysis of Mediterranean recordings were compared with the established categories. A general trend of similarities has been found but some categories are not represented in Meditenanean recordings. Risso's dolphin is a well spread out species, living in tropical and warm temperate waters world-wide, but at present little effort has heen done to study its vocal behaviour. Our future aim is to enrich the present catalogue with further recordings from North Sulawesi, Indonesia.

(INDEX)

MULTIPLE SPERM WHALES TRACKED BY COMBINING A TOWED DIPOLE HYDROPHONE AND FREE-DRIFTING SPAR-BUOY ARRAYS (1) Borsani, J.F., (2) Flayes, S.A., (3) Molinari, A. and (4) Costa, D.

(1,3) Tethys Research Institute, Viale G.B. Gadio 2, 20l2l Milano. (l.d) University of California, Santa Cruz, CA 95064, USA

The feasibility of tracking individual large cetaceans acoustically while monitoring distribution and acoustic behavior of other cetaceans within the same area or group was tested in the Ligurian Sea during April 1997. A towed dipole hydrophone array (4 hydrophones switchable to pairs, Casio DA-7 DAT recorder, COLMAR amplifier- system response 500 Hz to 17 kHz) and two semi-submersible free-drifting spar buoys (each equipped with a Garmin 45 GPS data logger, Sony D-8 DAT recorder and HTI SSQ-41B hydrophone- system response 10 Hz to 22 kHz and a VHF radio for time synchronization) were deployed in the presence of sperm whales (Physeter macrocephalus). While the buoys drifted free the dipole array was towed under sail on a track parallel to the drifting front of the buoys; vessel speed and heading were kept constant and its position logged automatically. One sperm whale was chosen as the reference bearing and tracked throughout the recording session (100 min. of 4 channel digital recording were obtained). Six other whales could be discriminated and located in space by computing arrival time differences of clicks at the sensors and calculating bearings of the sources during post-processing. Two-dimensional locations were obtained for one-minute samples throughout the recording session, three-dimensional locations were obtained when the depths of several hydrophones were known. Advantages of this combination are: 1) the freedom of the towing vessel to pursue other objectives after buoy deployment (an individual female sperm whale was tracked to obtain a fluke-id); 2) the active variation of the array aperture to provide virtually infinite spacing combinations required for locating different cetacean species; 3) the use of the spar buoys to eliminate the right/left ambiguity of towed arrays while maintaining real-time tracking abilities; 4) the enhanced recording quality of buoys unaffected by towing noise and 5) the extension of detection ranges by the cross-correlation of sounds from different sensors to extract masked signals. The result is a system that permits the simultaneous tracking of focal animals while conducting acoustic surveys of cetaceans over large geographic areas.

(INDEX)

SONAR RELEVANT STRUCTURES IN THE FETAL NARWHAL. (MONODON MONOCEROS)

(1) Comtesse-Weidner, P. and (2) Oeschläger, H.A.

Adult toothed whales possess forehead structures such as the blowhole, nasal air sacs, a peculiar facial musculature, acoustic fat bodies as the melon and dorsal bursae and monkey lips (Cranford et al.1996) that are thought to function in the generation and emission of biosonar signals. The signals are reflected and modified in the environment, received by the lower jaw (acoustic fat body) and transmitted via the tympanic bone and the auditory ossicles to the inner ear (Norris 1968, 1988). The ear bones are united in a tympanoperiotic complex and uncoupled from the skull (acoustic isolation; cf. Oelschläger 1990). We investigated sections of a microslide series (coronal plane; thickness: 40 µm) of an early narwhal fetus of 137 mm total length (Hubrecht Laboratory, Utrecht, The Netherlands) in order to determine the morphological status of those structures which presumably are part of the sonar system. The most important observations are as follows: 1. No facial asymmetry; blowhole transverse, in front of cranial vault; no nasal conchae; three buds of nasal air sacs on each side; nasal plugs present; future melon consisting of mesenchyme interweaved by numerous bands of muscle fiber bundles. 2. Rostrum very short, telescoping of skull in early stage;jugal bone rod-like; dental with large alveolar canal, containing mesenchyme (future mandibular fat body) and Meckel's cartilage; posterior part of dental as "pan bone". 3. Rudimentary external auditory meatus; tympanic membrane funnel-shaped; middle ear cavity without accessory sacs; plane of incudomallear joint horizontal; tensor tympani and stapedius muscle present. Cochlea large, with two turns; vestibular apparatus comparatively very small; cochlear aris pointing ventrad. Tympanoperiotic complex large; tympanic bone more or less shell-like; uncoupling of periotic incomplete. In principle, the morphology of this narwhal fetus is intermediate between the mammalian bauplan and the neonate condition. However, many of the future sonar-relevant structures seem to be in an advanced stage of development without being mature histologically. The morphology of the temporal region is impressive, in particular the striking size difference between the cochlea and the vestibular system, which recalls the situation in the adult.

(INDEX)

EVIDENCE FOR THE SOCIAL FUNCTION OF THE HUMPBACK WHALE SONG

Darling; J.D.

West Coast Whale Research. l200-925 W Georgia St. Vancouver, BC V6R2L3. CAN

A prevailing hypothesis on thc mating system of the humpback whale (Megaptera novaeangliae) is that the song functions to attract females to the male singer. Here I present evidence which suggests that the social function of the song is primarily a display between males. This idea initially arose from observations made in the early 1980s on the characteristics of the song, on the gender and behavior of singers and on a small sample of 4 known males that joined singers. To evaluate these alternate explanations, between 2 February and l0 April 1997 off Maui Hawaii, 83 singers were located, audio-recorded, photo- identified and monitored for interactions with other whales. In cases when a lone adult joined a singer, both animals were biopsied when possible in order to determine sex genetically. The most common interaction of lone adults and singers, observed in 28 instances, included: 1) a single adult moved into the vicinity of a singer; 2) the animals was eventually joined; 3) the singing stopped; 4) the pair interacted for a period ranging from 30 seconds to 10 minutes; 5) the pair split apart with one or both animals moving quickly from the area. Interactions ranged from a single pass-by within a whales length, to side-by-side swimming, to thrashing and tail slapping by either animal. During this interaction, 16 joiners were sexed: 12 genetically and 4 behaviorally (joiner began singing after the interaction). All joiners sampled during this 'typical' interaction were found to be males. Few exceptions to this pattern were noted. One explanation for the regular meeting of male humpbacks is that it is a necessary behaviour pattem for the establishment and maintenance of a dominance order, typical of many land mammals.

(INDEX)

ASSESSING COMPLEX PATTERNS OF ACOUSTIC SIMILARITY: USING A NEURAL NETWORK TO INVESTIGATE KILLER WHALE (ORCINUS ORCA) DIALECTS.

(1) Deecke, V., (2) Ford, J.K.B. and (3) Spong, P.

(1) University of British Columbia, Vancouver BC V6T IZ4, Canada (2) Vancouver Aquarium, Vancouver, BC V6B 3X8 (3) Orcalab, Alert Bay BC VON 1AO, Canada.

A quantitative measure of similarity of acoustic signals is crucial to any study attempting to describe and compare the vocal signals of different species, social groups, or individual animals. Previous approaches to this problem have ranged from univariate and maltivariate statistical analyses to ratings of similarity by human observers. In this study, we developed and tested a non-linear indicator of acoustic similarity based on neural network analysis. Since the performance of a neural network depends on the amount of consistent variation in the training set, this technique can be used to assess and integrate the presence of such variation from samples of acoustical signals and can be used to derive an index of acoustic similarity. We tested the performance of this index on the calls of 8 different maternal subgroups of killer whales belonging to A-clan of the northern resident community of British Columbia, Canada. A total of 1420 calls were digitized from recordings in which only one of the groups was present and assigned to one of 14 different call types produced by these groups. Similarity of the 6 most common call types shared by these groups was assessed in pairwise comparisons using a backpropagation neural network. For comparison, acoustic similarity . of the N4 call from the 8 groups was also judged by human observers in pairwise discrimination tasks, in which the subjects were asked to classify test sets of 32 calls from two different groups using training sets of 16 known calls for reference. As in the neural network analysis, the discrimination error for each pairwise comparison was used as an index of similarity. Both the neural network index and the results of the discrimination tasks divided the 8 groups into 3 major acoustic clusters, which correlated well with the association patterns of thc groups. Neural network analysis therefore represents a useful and biologically meaningful tool to quantify patterns of acoustic similarity. Furthermore, the overall degree of similarity varied between different call types, implying different rates of change for different calls.

(INDEX)

WHISTLE USAGE IN BOTTLENOSE DOLPHINS (TURSlOPS TRUNCATUS): SIGNATURE WHISTLES AS CONTACT CALLS

Janik, V. M. and Slater. P. J. B.School oJBiological and N/edica! Sciences. University oJSt Andre,vs,FiJe KYl6 9TS, UK

The signature whistle hypothesis assumes that bottlenose dolphins use signature whistles for individual recognition and to maintain group cohesion. This assumption is based on the fact that isolated individuals produce stereotyped whistles. To date, no study has tried to compare call usage in group and isolation contexts to investigate functional aspects of dolphin whistles. In this study whistle type usage in a group of four captive bottlenose dolphins was compared in two contexts. Individuals were recorded while they were separate from the group and while they all swam in the same pool. Separations occurred spontaneously when one animal swam into another pool. No partitionings were used. Calling animals were identified by an amplitude comparison of the same sound in the two different pools. To avoid observer bias whistles were classified into types independently by five different observers. Each individual produced its own stereotyped signature whistle when it was separated from the group. The remaining dolphins also produccd primarily their signature whistles if one animal was in a separation like that. However, if all animals swam in the same pool whistling rates were the same as in separation but only non-signature whistles were used. Whistle copying was rare and did not initate reunions or specific vocal responses. No aggressive interactions were observed prior to separations. 'These results strongly support the hypothesis that signature whistles are contact calls that are used in individual recognition and the maintenance of group cohesion.

(INDEX)

DISCRIMINATION OF SIGNATURE WHISTLES BY A BOTTLENOSE DOLPHIN

Harley, H. E.

Division of Social Scienes. New College of US.A. Sarasota, FL 34243, USA & The Living Seas, Walt Disney World's EPCOT Center. Orlando. FL, 32830-1000, USA

Although researchers have long believed that dolphins may use individual-speeific signature whistles to identify each other, well documented laboratory evidence suggesting that dolphins can discriminate among these whistles has been scanty. To address this issue, the ability of a captive Atlantic bottlenose dolphin (Tursiops truncatus) to discriminate among the signature whistles of six wild Atlantic bottlenose dolphins is being assessed. The whistle stimuli are digitized versions of recordings (made by Tyack, Wells and their associates) of vocalizations produced by dolphins in Sarasota Bay, FL. The captive dolphin is performing a conditional matching task in which each whistle is paired with a specific object. The dolphin has recently learned to discriminate one whistle (Whistle A) from the other five. ln the last ten 18-trial sessions, the dolphin has accurately matched Whistle A to its object 100% (30/30) of the time and has erroneously matched other whistles to Whistle A's object only 4% (6/150) of the time. Hence, the data suggest that a bottlenose dolphin can discriminate among signature whistles. (And it is probable that more whistle object associations will be mastered within the next six months.)

*************

(same abstract in other version)

Poster Summary: Signature Whistle Discrimination by the Bottlenose Dolphin Heidi Harley New College of the University of South Florida & The Living Seas, Disney's EPCOT Center

Although researchers have long believed that dolphins may use individual-specific signature whistles to identify each other, well-documented laboratory evidence suggesting that dolphins can discriminate among these whistles has been scanty. To address this issue, the ability of a captive Atlantic bottlenose dolphin (Tursiops truncatus) to discriminate among the signature whistles of six wild Atlantic bottlenose dolphins is being assessed. The whistle stimuli are digitized versions of recordings (made by Tyack, Wells, and their associates) of vocalizations produced by dolphins in Sarasota Bay, FL. The captive dolphin is performing a conditional matching task in which each whistle is paired with a specific object. The dolphin has recently learned to discriminate four whistles, from the others. In the last five sessions, the dolphin has accurately matched these four whistles to their objects 89% (49/55) of the time. (Chance is 25%.) Hence, the data suggest that a bottlenose dolphin can discriminate among signature whistles. (Preliminary data also suggest that dolphins can identify new exemplars of familiar whistles.)

The Ouestion: Can dolphins discriminate among signature whistles'

Method: Stimuli: The stimuli were digitized whistles provided by Peter Tyack and his associates. The whistles were collected with Randy Wells from wild dolphins Iiving in Sarasota Bay, Florida. Conditional Matching-to- Sample: Each whistle was uniquely paired with an object presented in an array. (For example, Dolphin 148's whistle was paired with Flower at the far left of the array.) At the start of each trial, the dolphin was positioned in front of a hydrophone at the center of the six-object array, and she listened to a whistle. Then she chose an object. If she was correct, she received 2 fish. Session Organization: Each whistle was presented 3 times during each session. Presentation order was randomized. The trainer could not hear the whistle presented, and thus was ignorant of its identity. Results At this time, the dolphin has learned to associate four whistle-object pairs. Performance accuracy on these pairs in the last five sessions was 89%. (Chance accuracy for four choices is 25%.)

The Answer: Dolphins can discriminate among signature whistles.

More Questions: Do Dolphins categorize different exemplars of the same dolphin's whistle together?

At this time, the dolphin has been exposed to a new exemplar of Dolphin 12's whistle eight times. (New exemplars are presented once in each session. The dolphin is never reinforced for a response to a new exemplar.) The dolphin has responded by touching the object normally associated with 12's whistle five (63%) times. (Chance with six choices iS 17%.)

Do dolphins classify whistles by referent or whistler ?

At this time, the dolphin has been exposed to male-dyad partners' imitations of each other's whistles. (These new exemplars are presented once in each session. The dolphin is never reinforced for a response to a new exemplar.) Dolphin 73's imitation of 74 has been presented five times, and the dolphin has responded by touching the object normally associated with 74 one (20%) time and the object associated with 73 three (60%) times. On the other hand, Dolphin 74's imitation of 73 has been presented Six times, and the dolphin has responded by touching the object normally associated with 73 four (67%) times and the object associated with 74 one (17%) time.

(INDEX)

WHISTLE CONTOUR TYPES ASSOCIATED WITH SPECIFIC AGONISTIC CONTEXTS IN BOTTLENOSE DOLPHINS

(1) Veit, F.

(I) Institut für Verhaltensbiologie, Freie Universität Berlin, Hoderslebener Str. 9,

12163 Berlin, Germany

The 'signature whistle' hypothesis is prevailing among researchers of dolphin communication, but in the last years criticism arose to this hypothesis. It proved to be incomplete probably due to recording conditions, where animals were in isolation or under restrain and it was therefore demanded to investigate dolphin communication under more natural conditions. Following this idea, I recorded the vocalisations and the concurrent behaviours of a group of eight bottlenose dolphins (Tursiops truncatus) over a period of four month in diverse social contexts. The animals, two males, three females and three juveniles, were housed under semi-free conditions in the 'Dolphin Reef Eilat', Israel. The occurrence of four clearly distinctive whistle contour types (wh-A, wh-4, wh-C, wh-D), different from the presumed signature whistles of the observed individuals, was compared for three agonistic contexts. Context 1: Male-a chasing Male-b (n=24); Context 2: Male-a chasing Female (n=16); Context 3: Male-b chasing Male-a (n=11). Vocalisations could not be generally addressed to a certain individual, nevertheless the sender was identified in many cases by several means, including relative position to hydrophone. Of all recorded whistles, 261 belonged to the types analysed here (100 wh-A, 26 wh-B,120 wh-C,15 wh-D). In Context 3 (Male-b chasing Male-a) only 2 wh-D occurred and none of the other whistles. Therefore only the two other contexts, where Male-a was the pursuer, are compared in the following. All 4 whistle types occurred in these contexts. In Context 1, the relative frequencies of occurrence of wh-A and wh-D were significantly higher, whereas in Context 2 the relative frequency of wh-C was significantly higher (p<0,05 for all). An interestihg feature of Context 1 was that the pursued Male-b was sometimes closely accompanied by females (8 out of 24 chases). All occurrences of wh-B and wh-C in this context were during these 8 chases. Whenever the sender of one of the whistles could be assigned it was Male-a. The frequency contours of wh-B and wh-C showed interesting similarities with the signature whistle of this male. The results show that the vocal communication system of bottlenose dolphins is indeed very sophisticated and this holds even for applications of their various whistles.

(INDEX)

CUES FROM RESPONSES OF BOTTLENOSE DOLPHINS TO WHISTLE PLAYBACK

(1)·;Todt, D., (2) Veit, F., (3) Hultsch, H. and (4) Zilber, R.

(1,1,3) Institut f. Verhaltensbiologie, FU Berlin, Haderslebener Str. 9, 12163 Berlin, Germany.

(4) Dolphin Reef Southern Beach, P.O.B. 104, Eilat 88100, Israel

The majority of publications on acoustical signals of bottlenose dolphins (Tursiops truncatus) deals with the so-called signature whistle which is considered to encode a particular individual's identity and to serve thereby as a kind of contact call. Currently, however, a growing number of investigators extended the research focus to the repertoires of whistle types and analysed, for example, relationships between signal parameters and features of social contexts. We have supplemented these studies by an approach which included a series of playback experiments designed to elucidate the potential meaning of particular whistle parameters by an analysis of the dolphins' responses. Our study was conducted at the Dolphin Reef, Eilat, Red Sea (size of site: > l0.000m² ; size of group: 5 adults (1/4), 3 calves (2/1); for details see: Todt & Hultsch, l996, European Research on Cetaceans 9: 287-291). Playbacks were done through an underwater loudspeaker placed at the Northern part of the net around the site. Specifically synthesized whistles served as auditory stimuli (control: signature whistle of a dolphin who had left the group some weeks before the experiments). As a prerequisite for the start of a given test, we ascertained that all dolphins were assembled (e.g. for fishing) in a distance of about 50 m apart from the loudspeaker. This allowed us to assess especially whether and how quickly which dolphin approached the sound source. Our data showed that the number of adults who approached the loudspeaker and also their response latency depended on the frequency contour of the played whistle type. The signature whistle and down-sweeps had a stronger effect than up-sweeps. However, subjects finally habituated towards the playback and this effect concerned all types of whistles. No dishabituation occurred when new whistles were used. Our results suggest that reinforcement of responses to auditory signals is a prerequisite for their persistence.

(INDEX)

First Page Previous Page Go Top