Spatial distribution, habitat utilization, and social

Spatial distribution, habitat utilization, and social interactions of humpback whales, Megaptera novaeangliae, off Hawaiyi, determined using acoustic and visual techniques A.S. Frankel, C.W. Clark, LmMm Herman, and C.M. Gabriele

Abstract: Acoustic and visual methods were used to track and observe humpback whales off the island of Hawai'i. Sixty-two singing whales were located acoustically in water depths from 10 to 305 fathoms (mean 126 fathoms; 1 fathom = 1.828 m). This indicates that singers are not confined within the 100-fathom contour, although nearshore waters had a higher density of singers. The separation distance between singers (mean 5.1 km) was found to be significantly greater than that between nonsinging singletons (mean 2.1 km), supporting the hypothesis that song functions to maintain spacing between singers. The mean speed of singers determined from visual data was 1.79 km/h and from acoustic data 1.6 km/h. Some singers actively swam while singing. Other singers continued singing while affiliating with or being joined by other whales. The correlation between breaching and the cessation of singing suggests that the sounds of aerial behavior can convey information to other whales. These observations suggest the need to expand the traditional interpretations of the behavior of singing humpback whales obtained from visual observations alone. RCsumC : Des mCthodes acoustiques et visuelles ont semi a repdrer et a observer des Rorquals a bosse au large de l'ile d'Hawai'i. Soixante-deux rorquals chanteurs ont CtC repCrCs par leurs cris a des profondeurs de 10 a 305 brasses (moyenne 126 brasses; 1 brasse = 1.828 m), ce qui indique que les chanteurs ne sont pas restreints aux 100 premikres brasses, mCme si les chanteurs sont gCnCralement prCsents en plus grand nombre dans les eaux cdtikres. La distance entre deux chanteurs (moyenne 5,l km) s'est avCrCe significativement plus ClevCe que la distance entre deux individus non chanteurs (moyenne 2,l km), ce qui appuie l'hypothkse selon laquelle les cris servent a assurer l'espacement entre les chanteurs. La vitesse moyenne des chanteurs a CtC estimCe a 1,79 km/h d'aprks les donnCes visuelles et a 1,6 km/h d'aprks les donnCes acoustiques. Certains rorquals chanteurs nageaient mCme en Cmettant leurs cris. D'autres ont continud d'dmettre leurs cris tout en s'associant a d'autres rorquals ou a l'approche d'autres rorquals. La corrClation entre les sauts hors de l'eau et 1'arrCt des chants semble indiquer que les bruits reliCs aux comportements aCriens peuvent fournir de l'information aux autres rorquals. Ces observations nous enjoignent de remettre en question les interprktations traditionnelles du comportement des rorquals chanteurs ClaborCes seulement a partir d'observations visuelles. [Traduit par la RCdaction]

Received September 19, 1994. Accepted March 6, 1995.

A.S. Frankel.' Department of Oceanography, University of Hawai'i at Manoa, 1000 Pope Road, Honolulu, HI 96822, T T C AI. V .".I

C.W. Clark. Laboratory of Ornithology, Bioacoustics Research Program, Cornell University, Ithaca, NY 14850, U.S.A. L.M. Herman and C.M. Gabriele. Kewalo Basin Marine Mammal Laboratory and Department of Psychology, University of Hawai'i at Manoa, Honolulu, HI 96814, U.S.A. Present address: Laboratory of Ornithology, Bioacoustics Research Program, Cornell University, Ithaca, NY 14850, U.S.A. Can. J. Zool. 73: 1134- 1146 (1995). Printed in Canada 1 Imprime au Canada

Introduction Humpback whales (Megaptera novaeangliae) are migratory baleen whales that occur in all oceans of the world. They feed in high-latitude waters (Jurasz and Jurasz 1979) and migrate to tropical waters in winter, where mating and calving are thought to occur (Chittleborough 1965; Winn and Winn 1978). Female humpbacks have an average birth interval of 2.4 -2.8 years (Baker et al. 1987; Clapham and Mayo 1987). Given a population sex ratio of approximately 1: 1 (Glockner-Ferrari and Ferrari 1990), the operational sex ratio is therefore approximately 1:3. Reproductively active females therefore constitute a limiting resource. Males appear to compete for reproductive access to females in surface-active pods. The competition between males appears to escalate from

Frankel et al.

low-level agonistic threats and displays to high-level agonism involving physical combat (Ty ack and Whitehead 1983; Baker and Herman 1984). Social sounds produced in these pods may function as acoustic threats between males (Tyack 1983; Silber 1986). Besides producing social sounds, male humpbacks sing long, complex songs. Song consists of a set of themes produced in a consistent sequence (Payne and McVay 1971). Within a season, the songs of all singers typically have the same sequence of themes. However, song structure gradually changes throughout the season, with almost all individuals making the same changes, resulting in a continuously evolving song (Guinee et al. 1983; Payne et al. 1983; Payne and Payne 1985). Song is produced primarily by lone, relatively stationary males on the breeding grounds (Winn and Winn 1978; Tyack 1981). However, some whales sing while in groups (Baker and Herman 1984) and some sing while swimming (Frankel et al. 1989). Song is primarily produced in the wintering grounds, but is occasionally heard in the late fall in high latitudes or along the migratory route (McSweeney et al. 1989; Clapham and Mattila 1990). The common occurrence of song within a presumed reproductive context supports the idea that singing is a component of the humpback mating system. Some investigators have proposed that the broadcast of song may be a sexual advertisement to females (Payne and McVay 1971; Winn and Winn 1978; Tyack 1981). Song may also establish or maintain space between potentially competing males (Winn and Winn 1978; Tyack 1981; Mobley et al. 1988). Baker and Herman (1984) proposed that song may also serve to synchronize ovulation in females. Playback experiments and natural observations have shown that very few whales approach the playback of song (Tyack 1981, 1983; Mobley et al. 1988). Although song may advertise the presence of a male, it does not often appear to attract other whales. Previous attempts to address questions of song function have utilized visual observations or analyses of recordings of individual whales (e.g., Tyack 1981; Mobley et al. 1988). The acoustic location procedure employed in this study records the vocalizations and directly determines the locations of the vocalizing whales. When used in combination with visual tracking, acoustic location allows simultaneous observation of whale surface behavior, movement patterns, vocalizations, and social affiliations. The combined acoustic and visual techniques provide the opportunity to observe whale behavior both underwater and at the surface, potentially offering new insights into the social behavior of humpbacks . Several whale and dolphin species have previously been studied with acoustic location techniques (Watkins and Schevill 1972; Watkins2; Norris and Doh1 1980; Clark 1983; Clark et al. 1986~).For example, acoustic location has revealed that bowhead whales travel under the ice and are often farther offshore than originally believed, at distances where they can be heard but not seen (Clark et al. 1986a; Clark and Ellison 1988). Additionally, through the applica-

*

W. A. Watkins. 1974. Computer measurement of biological sound source locations from four-hydrophone array data. Unpublished technical report for the Office of Naval Research.

tion of passive acoustics arrays, the population size of the Bering-Chukchi-Beaufort stock of bowheads has been shown to be much larger than indicated by visual censuses alone (Zeh et al. 1993). Acoustic tracking of bowheads also revealed that these animals produce calls that may help them navigate through the ice (Ellison et al. 1987; George et al. 1989). Finally, acoustic locations have shown that bowhead singers have two distinct "voices," which indicates that bowheads either sing in closely spaced pairs or that a single bowhead produces two harmonically unrelated sounds simultaneously (Clark et al. 1990). In this paper, we report on the use of passive acoustic location techniques in combination with more traditional visual techniques to study humpback whale behavior on the wintering grounds of Hawai'i.

Methods Data collection Study site The study was conducted from January to early April in 1989 and 1990. Figure 1 shows the study area centered around a hillside vantage point 6 km north of Kawaihae harbor on the northwestern coast of the island of Hawai'i. This vantage point was referred to as the shore station and had an elevation of 65.6 m above sea level. A linear three-element hydrophone array was centered approximately 500 m offshore of the shore station. Our research represents the synthesis of data obtained using several techniques: shore-based passive acoustic location (Clark et al. 1986b), shore-based visual observation, and shore-based theodolite tracking (Tyack 1981). These three techniques were conducted simultaneously. Each technique is described separately. Observations were conducted from approximately 08:OO to 17:OO daily when weather and ocean conditions permitted. Acoustic location A linear three-element hydrophone array was used. The array was 2.1 km long in 1989 and 2.4 krn long in 1990. Each element consisted of a sonobuoy constructed of a PVC shell containing a Johnson Control gel-cell battery pack, a Sippican transmitter, a Tandy VHF antenna, and a connector for the hydrophone cable. The buoy was moored in approximately 30 m of water. The hydrophone cable exited the top of the buoy and was attached to the mooring line. In 1989, the hydrophone element was suspended in the water column with a small float approximately 3 m above the substrate. Water current induced cable fluttering caused us to move the hydrophone element closer to the bottom for the 1990 season. The element was then tie-wrapped to an upright anchor tine approximately 30 cm from the bottom, eliminating the cable flutter. Sounds received at the hydrophones were radio transmitted from the surface buoy to a four-channel radio receiver operated at the shore-station observation site. The signal recording level was adjusted with a custom built multichannel audio monitoring system in 10-dB steps. Data were recorded on a TEAC R-61D cassette tape recorder using Ampex normal-bias cassettes. This data-recording system was similar to that described in Clark et al. (1986~).

1136

The acoustic location procedure recorded and located vocalizing whales even when they were submerged or out of visual range. Acoustic location is based upon the fact that sounds produced by a whale arrived at each hydrophone of an array at a different time. To locate a vocalizing whale, it is necessary to know the geometric positions of the hydrophones, the speed of sound in water, and the arrival times of the sound at the different hydrophones. Theodolite fixes were used to determine ,the location and distance between the three hydrophone array elements. Measurements were conducted 3 times during the season. The hydrophones were fixed in position near the bottom to prevent movement-induced errors. Sound velocity was measured on 2 days (February 9 and March 11, 1989). A 100-300 Hz frequency sweep was played back through a U. S. Navy Research Laboratory J-9 underwater transducer. The distance between buoys divided by the travel time of the

Can. J. Zool. Vol. 73, 1995

signal determined the speed of sound in water at our location. The resulting speed of sound was 1540 mls. Acoustic locations were later calculated in the laboratory using a soundanalysis work station to measure differences in sound arrival times. The work station consisted of low-pass filters and a DEC PDP-11/23 computer equipped with an analog to digital conversion board and running WHALE software (Clark et al. 1 9 8 6 ~ ) .Sounds were played into the work station, which generated spectrograms of the signals. Spectrograms had a frame size of 256 ms and the fast Fourier transform (FFT) size was 256 points. A rectangular window function was used and the spectrogram overlap was 99%, allowing a temporal resolution of 3 ms. The spectrogram pairs were cross-correlated to determine the differences in the arrival times of the signal at the hydrophones. The difference in the arrival of a sound at a pair of hydrophones produced a hyperbolic bearing line pointing toward the vocalizing animal. As

Frankel et al

shown in Fig. 2, measurements from three hydrophones produced three hyperbolic bearing lines and their intersection represents the whale's location. If the three hyperbolic pointing lines did not intersect perfectly, a triangle was formed by the three lines and the centroid of this triangle was used as the whale's position. With this procedure, sounds from up to nine different singers could be located, even when several vocalized simultaneously (Clark et al. 1986b; Frankel et al. 1989). Visual behavioral observation The height of the shore station allowed long-range observation of several pods without influencing the behavior of the whales. A behavioral observer used 7 x 35 binoculars to sight pods within a 5-km radius of the shore station. Each pod was assigned an identification number and its composition was described. The observer noted the time when pods surfaced and dove. Individual behaviors of each pod were described opportunistically with a standard ethogram developed previously (see Baker et al. 1982). The behavioral states of all visible pods were described for each surfacing sequence of that pod with the following definitions.

Stationary

Singerlike

Slow swimming Typical swimming Rapid swimming Low-level agonism High-level agonism

Affiliations and disaffiliations Movement changes Respiration

Aerial behaviors

Behavior characterized by little or no movement between blows and dive times less than 5 min Behavior characterized by little or no movement by a single whale, with dive times greater than 10 min Pods swimming at 1-2 kmlh or less Pods swimming between 2 and 6 kmlh Pods moving at 6 kmlh or more Behavior that includes bubbling, lowintensity head lunges Behavior that includes inflated and strenuous head lunges, striking another whale with any body part, or peduncle slaps Joining or leaving a pod Obvious changes in speed and direction The time of surfacing and submergence; in the absence of a fluke-up dive, dives were defined as any submergence longer than 60 s Aerial behaviors without defined function, including pectoral fin slaps, fluke slaps, breaches, unidentified splash -leap, head slaps

Whale speeds included in these definitions were estimated in the field, although more precise measurements were later made using acoustic and visual tracks. Surfacing and dive times were also recorded to monitor respiration patterns. Behavioral states and individual behaviors were verbally described as 3-digit codes to a computer operator, who entered them onto a Toshiba TlOOO portable computer. The computer allowed accurate and rapid recording of sequential behaviors as well as their times of occurrence.

Visual theodolite tracking Whales and vessels occurring in the study area were tracked with a Topcon DT-20 or a Leitz DT-5 theodolite, which measured horizontal and vertical angles to the target. These angles constituted a "fix," which was labeled with the time and whale or vessel identity. The vertical and horizontal angles of the fix were later converted to x - y coordinates with a computer program that accounted for the earth's curvature. The theodolite can theoretically measure the position of a target with an accuracy of 1 m at 1 km and f90 m at 10 km (Bauer 1986). The real accuracy is also dependent upon the operation of the instrument and environmental conditions. The speed and direction of movement of vessels or whales were calculated from a series of positions over time.

+

Calibration We determined the accuracy of the acoustic location method by comparing acoustically and visually determined locations of singers. Acoustic and visual tracks of two singers were compared at distances of 3-4 km. The mean absolute discrepancy between the two techniques was 0.7" for bearing (37 -49 m) and 3 % (90 - 120 m) for distance. There was some error in both techniques. The bearing accuracy of the theodolite is largely unaffected by range but is subject to operator initialization error. Theodolite distance accuracy decreases with increasing range because of the decreasing declination angle. Acoustic location errors are estimated from the length and width of the triangle formed by intersecting hyperbolic pointing lines. The length of the triangle represents the range error and the width represents the bearing error. In general, bearing error remains reasonably low ( < 1") at distances up to 20 km. The range estimate becomes unreliable at distances greater than approximately four times the overall length of the array (Carter 1987). Analysis The acoustic location analysis produced a dataset of x - y coordinates and time of occurrence, similar to theodolite data. These locations were plotted to visually check the assignment of individual locations to whales. The initial individual assignment was based upon the similarity of bearings, ranges, and difference-in-arrival times. Singers up to 20 km from the shore station were used to examine spatial distribution. Singers at ranges of less than 10 km were represented by the mean of all the acoustically determined locations. Singers located at ranges of 10-20 km were represented by the location with the shortest range. Locations were converted to latitude and longitude and compared to a data base of the bathymetry of Kawaihae Bay. The depth of water at each singer position was determined from the bathymetry data base, and the mean and median depths were calculated. The spacing of singers was compared with the spacing of nonsinging singletons to examine the effect of song on spacing between animals. This was a comparison of visually determined locations with acoustically determined ones, therefore the range was limited to 6 km from the shore station, the limit of visual observation. One separation distance was measured on the first occasion that both were tracked simultaneously. Differences in separation between singers and nonsingers were tested with a t test adjusted for unequal variances.

Can. J. Zool. Vol. 73, 1995

Fig. 2. The three hyperbolic cross-bearings calculated from the delays in time of arrival time of a signal at the three hydrophones. The insert shows the intersections of the hyperbolae. The centroid of the resulting triangle is taken as the location of the whale.

Speed was calculated for all singers within 10 km of the shore station. A speed estimate was determined from acoustic data by dividing the total distance covered by the total time elapsed. This probably produced a slight underestimate, but most singers' tracks were nearly linear. A different technique was used with the visual data. The weighted mean and the median values were calculated from individual legs. These were weighted by the elapsed time between locations so that the shorter, more error prone legs would contribute less to ,the value. Legs shorter than 90 s were eliminated.

Results of acoustic location observations Combined visual and acoustic data were collected during January through April in 1989 and 1990. A total of 76 visual observation sessions were conducted, and 280 h of acoustic recordings were made. Forty-five hours of acoustic data were analyzed for acoustic locations, and a total of 62 singers were located.

Spatial distribution and water depth Figure 3 shows the spatial distribution of all 62 singers located during the study. Singers were found in water depths ranging from 10.4 to 305 fathoms (1 fathom = 1.828 m). The mean water depth was 126 fathoms, the median water depth of singers was 109 fathoms, and the standard deviation was 72.98 fathoms (N = 62). Singers were located from 400 m to 12.9 km from the shoreline. The mean offshore distance was 4.44 km (SD = 3.03 km) and the median distance was 3.76 km. Although whales were found throughout the bay, their distribution did not appear random. As shown in Fig. 3, a 120" sector was drawn over the study area, representing the

theoretical range of operation of the array, or the "arena." The center line of the arena was perpendicular to the axis of the array. The sector was divided with radii of 8.66, 12.25, and 15 km, which subdivided the arena into six equal areas of 39.2 km2. A few whales had mean locations just outside these areas and were included in ,the nearest section. Comparing the two sections closest to shore showed 15 singers in the northern sector and 16 in the southern. Thus, there was no north - south difference in distribution in the nearshore area. The midrange sections had 5 singers north and 13 to the south. Finally, in the farthest offshore area, there were none to the north and 13 to the south. It is relevant to note the higher proportion of shallow water in the southern portion of the arena. Thirty-one of the 62 singers located were offshore of the 100-fathom isobath. The circular sector representing the arena had a total area of 235 km2. The amount of area within the 100-fathom contour was 50 km2, and the remaining offshore area was 185 km2. Thus, the density of whales in the nearshore area was higher than in the offshore area (0.62 vs. 0.17 singers/km2). It is important to remember that these densities represent a collection of locations, so they are only given for comparing onshore with offshore habitat utilization. While most singers are located near shore, there are a substantial number of singers in the offshore, deepwater environment. Singers in very deep water are presumably in a different acoustic environment than shallow-water singers. Important acoustic differences between these areas may include transmission loss, ambient noise, and signal degradation.

Interindividual separation Separation distances were measured for 15 pairs of nonsinging singletons and 12 pairs of singers. All animals were

Frankel et al.

Fig. 3. The bathymetry of the study area, the location of the shore station, and the distribution of all located singers. Fifty percent (31162) of the singers were located outside the 100-fathom (182 m) isobath.

East-West Axis (km) located within a 6-km radius of the shore station. The mean separation was 5076 m (SD = 1888 m) between singers and 2069 m (SD = l a 2 m) between nonsingers. The mean separation for singers was significantly greater than that for nonsingers (t = 4.35, df = 22, two-tailed p < 0.00 1). Figure 4 shows a clear difference in the distribution of separation distances. It can be seen that the distribution of the singleton distances is non-normal; however, the t test is robust to deviations from the normality assumption and the differences in the distribution are distinct, the mean separation for singers being 2.4 times greater than .that for nonsinging singletons.

Moving singers Several singers moved while singing. The speeds of singers located within 10 km of the shore station were calculated from the first and last acoustic locations of each singer. This was done to minimize the effect of error in position measurement. Thirty-one singers were used to produce acoustically derived speeds, with a mean of 1.6 kmlh (SD = 1.03 kmlh) and a median value of 1.2 kmlh. Swimming speeds were also estimated from the visual data. Medians and means weighted by the time interval between successive locations were calculated. Time was used as a weighting factor because short legs tend to have a greater percentage of measurement error. Speeds derived from visual data were slightly higher than those from the acoustic tracks. The median speed was 1.4 kmlh

and the weighted mean was 1.8 kmlh (N = 12 whales). It is possible that some of the observed movements of singers may have been due to passive drift with the current. Shallow current meters ( < 2 0 m deep) placed nearby reported a median current speed of 1.13 kmlh (National Oceanographic Data Center data base). Direct observation of the currents showed that they were primarily parallel to shore. Nevertheless, some singers showed clear signs of active swimming. On several occasions, singing whales were observed to transit the hydrophone array quickly. These whales were probably singing and swimming. Other singers were observed moving with other whales. Four singers were tracked simultaneously on March 24, 1990 (Fig. 5). One of the singers (in pod C) first swam parallel to the shore and then moved directly offshore. Current flow in the area is primarily tidal and parallel to the coastline, both northerly and southerly. Since this track is perpendicular to current flow for at least a portion of the track, it seems likely that the moving singers were actively swimming rather than passively drifting in the current. Furthermore, on the same day, two singers in pods in relatively close proximity were tracked simultaneously moving in opposite directions: pod B was tracked moving north and pod D was moving south. Therefore not all singer's movement can be passive drifting with the current. Singing was occasionally observed in several adult pods.

Can. Zool. Vol. 73, 1995

Fig. 4. The distribution of separation distances between nonsinging singletons and singers. The separation distances of singers are substantially greater than those of nonsingers.

Singers (N=10)

Singletons (N=15)

Separation (km)

Of the 71 pods ,that included a singer, one was a dyad, two were trios, and one was a pod of four animals. The 71 pods include the 62 observed from shore and nine singers observed only from the fluke-photography boat. These observations mirror those of Tyack (198 I), who found 91 lone singers, 3 in dyads, 1 in a trio, and no instances of a singer in a group of 4 whales.

Case studies of social interactions The first observation describes a singer's approach and affiliation with a singleton. At 12:50 on February 26, 1989, we sighted a single whale (pod 4) when it breached while swimming north along the coast from the southern portion of the array (Fig. 6). It was first visually positioned at 13:13. This whale breached again and performed a series of flukeslaps at 13:20. At 13:36, another whale joined pod 4, and they swam northwest until 1355, when only one whale was seen again. This whale was tracked until 14:46. This appeared to be a routine affiliation between two adults. When the acoustic location data were added to the visual data, an entirely different picture emerged. A relatively stationary singer was detected acoustically at 11:40. It stopped singing at 12:44 for 14 min, and resumed at 12:58 while swimming at 3.6 krnlh. Pod 4 breached (13:20) as the singer approached. The acoustic data indicated that it was the singer that joined with pod 4 at 13:36. The singer continued to sing for the short duration of their affiliation, which apparently ended by 13:55. The then solitary singer was tracked visually and acoustically as it swam along the coastline. This example demonstrates how the combination of acoustic and visual data produces a more complete descrip-

tion of a social interaction, in this case a singer joining a lone adult and continuing to sing. This type of interaction has not been reported previously. On March 8, 1989, a single whale affiliated with a singer (Fig. 7). The singer (pod 1) was sighted to the south of the shore station at 10:20, swimming north along the coastline while singing. Meanwhile, another singleton (pod 2) was approaching from the north. The two whales joined at 10:58, and the singer continued to sing as they swam north until 11:14. They then reversed course 180" and swam southeast together. They disaffiliated at 12:02. Pod 1, the singer, continued south until 12:ll and then turned and rapidly swam north, resuming its northerly travel. Pod 1 continued to sing throughout the affiliation. Pod 2 was last seen heading north and away from pod 1, when it was lost from view. Similar to the previous description, this observation presents an example of a singer affiliating with an adult and continuing to sing. While pod 1 continued to sing throughout the course of the affiliation, it did not sing continuously. On two occasions the pod was closely approached by vessels. In both cases the singer stopped singing for a short period of time ( < 15 min), indicating that vessel approach may affect singing.

Interactions between singing and breaching animals On January 21, 1989, two singing whales were heard when only two hydrophones were operating. Therefore we could determine bearings, but not ranges to the whales. One whale was to the north of the array and the other was to the south. Neither whale was seen. Suddenly, a whale breached north of the observation station and both whales stopped singing

Frankel et al.

Fig. 5. The tracks of the four singers (in pods A-D) tracked on 24 March 1990. Pod C is seen moving both parallel and perpendicular to the shore and the typical current regime. Pods B and D are shown moving in opposite directions. These tracks support the conclusion that whales swim actively while singing.

1

Singer B

East-West Axis (km) almost immediately. After about 1- 2 min, the southern singer resumed singing, followed by the northern singer 30 s later. It is not known if the whale that breached was one of the singers. Additionally, on February 26, 1989 (Fig. 6), pod 4 breached before and during the approach of the singer. It is possible that the observed changes in behavior were in response to the acoustic signal of the breaches.

Discussion The combination of acoustic location and visual tracking has revealed several interesting social interactions between singers and other pods that could not have been observed using either method alone. Acoustic location data not only enhance but alter the conclusions that would have been reached from purely visual observation. The results suggest the need to revise the traditional interpretation of exclusively visual observations singing humpback whales.

Habitat utilization Humpback whales are a coastal species while on their wintering grounds (Herman and Antinoja 1977). Their distribution has been described as restricted to shallow water within the 100-fathom contour line (Herman and Antinoja 1977; Chittleborough 1953; Winn et al. 1975). Forsyth et al.

(199 1) found that whales in the Penguin Bank and Maui regions were located at a mean depth of 5 1.4 fathoms. Recent aerial survey data showed that 74% of all pods were seen in waters less than 100 fathoms deep (Mobley et al. 1994). This generalization may hold for many age and sex classes; however, the distribution of singers off the Kohala coast of the Island of Hawai'i appears to differ from the generalized pattern. Singers were found throughout the bay. No dense aggregation of whale locations was observed. Singers were located at least 12.9 km offshore and in water up to 300 fathoms deep. The measurable offshore distance is limited by the functional range of the hydrophone array (Carter 1987). Many whales were heard during the analysis at calculated distances of greater than 20 krn on offshore bearings from the array. While these whales cannot be accurately located, it does suggest that singing whales may be found farther offshore than 12.9 krn. Fifty percent (3 1/62) of singers were located in water deeper than 100 fathoms. This is a significant departure from expected frequencies if 74% of all pods are found within the 100-fathom contour = 18.57, p < 0.05). This indicates that while singers are found close to shore, the proportion of singers found in deep water is higher than for other classes of whales. While equal numbers of singers were found inside and

(Xt,,

Can. J. Zool. Vol. 73. 1995 Fig. 6. An offshore singer is shown approaching and joining with a nonsinging whale that was traveling north along the shoreline. Nonsinging pod 4 moved northwest, while a singer was heard offshore. The singer first moved farther offshore and then stopped singing. The whale then swam approximately 5 km northeast and resumed singing. The singer then swam a loop and approached and affiliated with pod 4. The two whales swam together for 19 min. Pod 4 was then lost from view and presumably disaffiliated. The singer sang throughout and after the affiliation and was tracked both visually and acoustically as it moved northwest.

14:46: End of

13:55: Last Sighting of Pod 3

I

13:36: Singer joins

Singer Resumed Singing

\

13113: First Pod 4 fix

? 13:20:

Pod 4 Breaches

East-West Axis (km) outside the 100-fathom isobath, the densities were not equal. The area of water < 100 fathoms deep that was covered by the array was 50 km2. The offshore area was 185 km2. This converts to a shallow-water density of 0.62 singers/km2 compared with 0.17 singers/km2 for the offshore area. This represents preferential usage of the shallower waters by singers. Such a preference also explains the greater number of singers located in the southern portion of the arena, as the amount of shallow water is greater in the southern portion of the arena. Furthermore, when only the two nearshore subregions of the arena were examined, the numbers of singers were almost equal and the depth distributions in these two areas were similar. It appears that the preference of humpback whales to be near shore applies to singers as well, but not absolutely. Although the density of offshore singers was less, half of the singers were located outside the 100-fathom isobath. Such an onshore -offshore range in distribution opens the possibility that differing oceanographic and acoustic conditions may affect the choice of singing locations. Offshore singing locations provide different acoustic propagation conditions, owing to the deep water and lack of physical obstruc-

tions that absorb sound. However, the factors that determine which area is preferable may be social rather than oceanographic. Pods with a calf are found significantly closer to shore than pods without calves (Smultea 1992). Mothers with calves do mate post partum, but the percentage is small (Chittleborough 1958). Most likely, mature females without calves represent better mating prospects. Mature females, probably estrous, or pre-estrous, can be reliably found in large surface-active or competitive pods (Clapham et al. 1992). These pods are found farther offshore than mothers and calves. Therefore, it might be argued that the region frequented by mature females without calves contains the prime singing areas, and these areas off the Island of Hawai'i are several kilometres offshore. Finally, singers have been detected as far as 300 km offshore in the Caribbean (Levenson and Leapley 1978) and > 100 km offshore in the North Atlantic (Clark et al. 1993). Humpbacks are also known to sing during at least a portion of their migration (Clapham and Mattila 1990). The faroffshore singers in the Caribbean could have been animals continuing on their migration route or moving between islands. Similarly, some of the singers detected offshore of

Frankel et al.

Fig. 7. A singer (pod 1) moved north along the shore and was approached and joined by a nonsinging whale (pod 2). The two whales moved north together, then reversed course. At 12:02, pod 2 disaffiliated and moved north and away from the singer. The singer continued south until 12:11, when it moved north again and continued north until it was lost.

East-West Axis (km) Kohala could be animals moving between the islands of Hawai'i and Maui. The rate of interchange between these islands is unknown, but the rate between the islands of Kauai and Hawai'i, which are farther apart, is low (Cerchio et al. 1991).

Swimming singers Singing humpbacks have been most often described as lone and stationary or slowly moving (Payne and McVay 1971; Winn and Winn 1978; Tyack 1981). While some moving singers have been observed (Tyack 1981; Baker and Herman 1984), it seemed that these were unusual examples. This impression may be due to difficulties in identifying singers from small boats, but these can be overcome with the use of a directional hydrophone (Winn and Winn 1978) or a hydrophone array. The mean speeds of singers reported here are low, with a large proportion of animals moving at speeds low enough that they can be considered stationary or drifting with the current. However, there are a few examples of singers moving at high speeds. Bauer (1986) found that the average speed of singers was 3.86 krn/h, based upon visual tracking alone. This speed is confounded, since Bauer combined "suspected" singers, animals that displayed "singer-like" behavior, such as long down times and limited movement between surfacings, with

singers confirmed by monitoring hydrophones deployed from boats. This inclusion may have biased the speed value downward, since "suspected singers" were by definition nearly stationary. The discrepancy between the speeds determined by Bauer (1986) and ours may be due to one or more factors. A difference in study regions may have contributed to the observed speed differences, Bauer's data were collected off Maui, and the Maui region typically has a greater whale density than Hawai'i, which may affect the behavior of singers. The difference in speeds may be an artifact (G.B. Bauer, personal communication). When acoustic tracks are produced, many individual locations are averaged and locations with apparently erroneous ranges are discarded. The net effect is one of smoothing the path of the singer. This smoothing reduces the length of the path and the calculated speed. Bauer's visual data were error-checked and obviously erroneous fixes were discarded, but there may be less net smoothing in visual tracks than in acoustic tracks. Finally, the most likely scenario is that some of the whales initially described as singers stopped singing and began swimming or increased speed. Nonsinging singletons had a mean speed of 4.62 krn/h. The standard deviation for singers was greater than that for nonsingers (2.89 vs. 1.95). Furthermore, the maximum speed recorded for a singer was 11.07 km/h vs. 8.09 km/h for nonsingers.

Can. J. Zool. Vol. 73, 1995

Observations with the hydrophone array have demonstrated that whales may behave as singers and not sing. For example, one whale observed less than 1 km from the shore station had 18-min downtimes, usually blew 4 times per surfacing, and surfaced in nearly the same location each time. Experienced observers were nearly certain that this whale was singing. However, when the hydrophones were switched on, no song was heard. The whale continued the same behavioral pattern for over an hour after the hydrophones were activated and no singing was heard during that time. The presence of moving singers indicates that the singing display can be spatially dynamic, although the potential significance of moving versus stationary singers has not been resolved.

Aerial behaviors Aerial behaviors of one pod sometimes precede changes in the behavior of other pods. In one observation, a breach was immediately followed by cessation of singing by two animals, suggesting that at least one of the singers detected the sound of the breach and altered its behavior in response. This suggests that the sound of aerial behaviors can intentionally or unintentionally convey information to other whales. Aerial behaviors are known to produce sound in humpbacks and other species (Clark 1983; Whitehead 1985). However, it is unlikely that aerial behaviors function exclusively as acoustic signals, given the energetic cost of aerial behaviors compared with vocally produced signals. Furthermore, many aerial behaviors produce very little sound (Dahlheim et al. 1984; A. S. Frankel, personal observation). Nevertheless, reception of the sound of an aerial behavior such as a breach may indicate the breaching animal's location and perhaps its behavioral state. Social interactions In previous descriptions of affiliations of singers with other pods, singing stopped when the singer joined a pod (Tyack 1981). Observations of singing continuing after affiliation are difficult to interpret, since the gender of the second whale is uncertain. Nevertheless, they are similar to observations of singing escorts accompanying mothers and calves (Baker and Herman 1984), and singing in both of these contexts may represent a component of courtship. Affiliations involving singers tend to be short-lived, leaving the possibility that singers may be affiliating with other males. If the song contains physiological condition or fitness information, then males as well as females should be able to assess the singer. Therefore, these affiliations could be between males of approximately equal fitness. Such affiliations may represent one male attempting to displace another male from the area. However, it is doubtful that such displacements would result in the formation of a dominance hierarchy (Darling 1983) because of the degree of movement of individuals between different islands (Cerchio et al. 1991), and the impermanence of social affiliations of whales in the wintering grounds (Mobley and Herman 1985) probably does not allow sufficient time for such a hierarchy to develop or have much utility. Interindividual separation The separation data show that the distance between singers is approximately 2.4 times that between nonsinging singletons. Such a difference could result if singers were less numerous than singletons. However, there was an average of

two pairs of singers per day compared with only one pair of nonsinging singletons every 2 days. This indicates that the density of singers was actually greater than that of singletons, therefore relative density differences cannot explain the results. While singers are almost certainly all males (Winn et al. 1973; Glockner-Ferrari and Ferrari 1990; Lambertsen et al. 1988), nonsinging singletons could be male or female. It is possible that the separation between male-only singletons would be greater than between all singletons. However, Gabriele (1992) found that single female whales were found alone significantly less often than expected (Xf41 = 13.04, p = 0.01 1). It is therefore likely that the nonsingers observed were primarily males. The more likely explanation is that song production has an effect upon other singers. Tyack (1983) noted an average distance of 6 km between singers off Maui and suggested that song may function to maintain that spacing. The substantially greater separation between singers than between nonsinging singletons suggests that song does function to maintain this spacing. The average spacing of singers found off the Island of Hawai'i was 5.1 km, slightly lower than the value reported by Tyack (1983). It is possible that the spacing may be different on other islands, which have different population densities, physical characteristics, or ambient noise levels. This would certainly seem to be the case in the Caribbean, where singers are reported to be 100-300 m apart, with a mean density of 0.3 whales/km2 (Whitehead 1981) compared with a maximum density of 0.045 whales/km2 off Maui (Whitehead 1981). The visually determined maximum density off Hawai'i was 0.035 whales/km2, similar to that off Maui. Given the much higher population density on Silver Bank than in Hawai'i, some density dependence in spacing seems likely. This is supported by observation of two singers separated by 500 m off Kauai during the seasonal peak in abundance (S. Cerchio, personal communication). Winn and Winn (1978) noted that the separation between singers varied between 1 and 40 km, apparently depending upon local population density. These estimates were based upon the sequential rather than simultaneous detection of singers, which reduces their accuracy. Tyack (1981) rarely observed singers closer to each other than 5 km during his studies off Maui. Both Winn and Winn (1978) and Tyack (1981) suggested that song might function in maintaining that spacing. Winn and Winn (1978) also suggested that song production may allow females to locate displaying males. Tyack (1981) proposed that song facilitates affiliation, at least by advertising location. On the basis of these studies and additional playback experiments, Frankel ( 1987) proposed that males sing songs with messages to both sexes, as do many bird species (Kroodsma and Byers 1991). The intersexual message is one of advertisement: singing conveys location and an indication of reproductive fitness. The intrasexual message is to "stay away," that is, to maintain the spacing between individual singers. The spacing between singers presumably increases the ability of females to localize displaying males. Humpback whales have been shown to be able to locate sound sources (Tyack 1983; Mobley et al. 1988).

-

Implications for the mating system The findings here have refined the description of the behavior and distribution of humpback whale singers. Whales were

Frankel et al.

found to sing while swimming and in more social contexts than previously described. Their distribution ranges farther offshore than that of other classes of whale. The data on spacing and observations of affiliations of singers with both male (Darling 1983) and female (Medrano et al. 1994) whales provoke further speculation on the role played by singers in the mating system of humpback whales. The affiliation of singers with other male whales indicates that intrasexual interactions do occur, although they are unlikely to lead to a dominance hierarchy. Tyack (1981) described whales that stopped singing and pursued other pods which likely contained a female. Medrano et al. (1994) described females approaching and affiliating with singers, suggesting that females are attracted to singers and supporting the female-attraction hypothesis. The social interactions of singers with both males and females suggest that song has both intersexual and intrasexual functions. These data are consistent with the hypothesis of a humpback whale mating system in which males engage in direct physical competition for reproductive access to females or sing, and they may switch strategies. Song functions primarily to attract females and advertise the male's location and, potentially, his quality. Information about the singer, in the form of intraindividual variation in structure, could be incorporated into the song. This would allow males to make assessments of the singer relative to themselves while also providing a potential mechanism for mate choice by females. According to this hypothesis, the intended function of song is primarily intersexual advertisement and attraction. The song also functions to maintain spacing between singers to maximize the females' ability to locate the singers. Males can and do hear song, and can also interact with singers, perhaps in an attempt to displace them. While this may be a likely model for the humpback whale mating system, it is still based on incomplete data. The combination of acoustic and visual observations has provided new information on humpback whale singing behavior, and its continued use in conjunction with other research techniques promises to provide new and important insights into this system.

Acknowledgments The authors profoundly thank the research staff of Thomas R. Freeman, Michael A. Hoffhines, Suzanne Yin, and Barry K. Patterson. Without their tireless efforts and good humor, this study would not have been possible. Bernd Wiirsig, Joe Mobley, Salvatore Cerchio, Mari Smultea, Peter Tyack, and Linda Weilgart provided valuable criticism of the original text. Shannon Atkinson, Brooks Bays, Gordon Bauer, John Bower, Laura Brown, David Helweg, Bill Langbauer, Edward A. Laws, Alexander Malahoff, Suzi Minges, Patrick Moore, Katy Payne, Craig R. Smith, and Dixon Stroup provided assistance and advice. We also thank the Center for Field Studies (Earthwatch) for providing most of the funding and our volunteer help. The Waikoloa Marine Life Foundation and The Whale-Aid Foundation provided support as well. The American Museum of Natural History partially funded the analysis. The North Slope Borough Department of Wildlife Management loaned equipment. Hal Whitehead and an anonymous reviewer helped to improve this paper. This is Contribution SOEST 3901 from the School of Ocean and Earth Science and Technology, University of Hawai'i at Manoa.

References Baker, C.S., and Herman, L.M. 1984. Aggressive behavior between humpback whales (Megaptera novaeangliae) wintering in Hawaiian waters. Can. J. Zool. 62: 1922- 1937. Baker, C.S., Herman, L.M., Bays, B.G., and Stifel, W.S. 1982. The impact of vessel traffic on the behavior of humpback whales in southeast Alaska: 1981 season. Final report to the National Marine Mammal Laboratory, Seattle, Wash. (Contract No. 81ABC-00 114). Baker, C.S., Perry, A., and Herman, L.M. 1987. Reproductive histories of female humpback whales Megaptera novaeangliae in the North Pacific. Mar. Ecol. Prog. Ser. 41: 103- 114. Bauer, G.B. 1986. The behavior of humpback whales in Hawaii and modifications of behavior induced by human interventions. Ph.D. dissertation, University of Hawai'i at Manoa, Honolulu, Hawai'i. Carter, G.C. 1987. Coherence and time delay estimation. Proc. IEEE, 75: 236-255. Cerchio, S., Gabriele, C. and Frankel, A. 1991. Inter-island movements of humpback whales in the Hawaiian Islands: three seasons off Kauai and Hawaii. In Abstracts of the Ninth Biennial Conference on the Biology of Marine Mammals, Chicago, Ill., December 5-9, 1991. p. 13. [Abstr.] Chittleborough, R.G. 1953. Aerial observations of the humpback whale (Megaptera nodosa). Aust. J. Mar. Freshwater Res. 10: 125- 143. Chittleborough, R.G. 1958. The breeding cycle of the female humpback whale, Megaptera novaeangliae (Borowski). Aust . J. Mar. Freshwater Res. 9: 1- 18. Chittleborough, R.G. 1965. Dynamics of two populations of the humpback whale Megaptera novaeangliae (Borowski). Aust. J . Mar. Freshwater Res. 16: 33 - 128. Clapham, P.J., and Mattila, D.K. 1990. Humpback whale songs as indicators of migration routes. Mar. Mammal Sci. 6: 151- 154. Clapham, P.J., and Mayo, C.A. 1987. Reproduction and recruitment of individually identified humpback whales Megaptera novaeangliae, observed in Massachusetts Bay, 1979- 1985. Can. J. Zool. 65: 2853 -2863. Clapham, P.J., Palsboll, P.J., Mattila, D.K., and Vasquez, 0 . 1992. Composition and dynamics of humpback whale competitive groups in the West Indies. Behaviour, 122(3-4): 182- 194. Clark, C.W. 1983. Acoustic communication and behavior of the southern right whale (Eubalaena australis). In Communication and behavior of whales. Edited by R.S. Payne. Westview Press, Boulder, Colo. pp. 163- 198. Clark, C.W., and Ellison, W.T. 1988. Numbers and distribution of bowhead whales, Balaena mysticetus, based on the 1985 acoustic study of Pt. Barrow, Alaska. Int. Whaling Comm. Rep. Comm. SCl39lPS8. pp. 365-370. Clark, C.W., Ellison, W.T., and Beeman, K. 1986a. An acoustic study of bowhead whales, Balaena mysticetus, off Point Barrow, Alaska during the 1984 spring migration. Report to the North Slope Borough, Department of Wildlife Management, from Marine Acoustics, Inc., Cotuit, Mass. Clark, C.W., Ellison, W., and Beeman, K. 19866. Acoustic tracking of migrating bowhead whales. In Proceedings of the IEEE Oceans '86 Conference, Washington, D.C., September 23 -25, 1986. Institute of Electrical and Electronics Engineers, Oceanic Engineering Society, New York. pp. 341 -346. Clark, C.W., Brown, L.M., von der Heydt, K., Baggeroer, A., and Dyer, I. 1990. Songs of the bowhead whale from the western Arctic and Spitsbergen stocks. In Proceedings of the Fifth Conference on the Biology of the Bowhead whale Balaena mysticetus, Anchorage, Alaska, April 1-3 1990. Edited by T.F. Albert. North Slope Borough, Barrow, Alaska. pp. 209 -212. Clark, C.W., Gagnon, C.J., and Mellinger, D.K. 1993. Whales '93: the application of the Navy IUSS for low-frequency marine mammal research. In Abstracts of the Tenth Biennial Confer-

Can. J. Zool. Vol. 73, 1995 ence on the Biology of Marine Mammals, Galveston, Tex. December 11 - 15, 1993. p. 3. [Abstr.] Dahlheim, M.E., Fisher, H.D., and Schempp, J.D. 1984. Sound production by the gray whale and ambient noise levels in Laguna San Ignacio, Baja California Sur, Mexico. In The gray whale. Edited by M.L. Jones, S.L. Swartz, and S. Leatherwood. Academic Press, Orlando, Fla. pp. 5 11- 54 1. Darling, J.D. 1983. Migrations, abundance and behavior of Hawaiian humpback whales, Megaptera novaeangliae (Borowski). Ph.D. dissertation, University of California, Santa Cruz. Ellison, W.T., Clark, C.W., and Bishop, G.C. 1987. Potential use of surface reverberation by bowhead whales, Balaena mysticetus, in under-ice navigation: preliminary considerations (SCl38lPS10). Rep. Int. Whaling Comm. 37: 329 - 332. Forsyth, N., Mobley, J.R., and Bauer, G.B. 1991. Depth preferences in Hawaiian humpbacks. In Proceedings of the Ninth Biennial Conference on the Biology of Marine Mammals. Chicago, Ill. December 5 -9, 1991. p. 24. Frankel, A. S. 1987. Sound playback experiments with humpback whales (Megaptera novaeangliae) in Hawaiian waters. M .S. thesis, University of Hawai'i at Manoa, Honolulu. Frankel, A.S., Clark, C.W., Herman, L.M., Gabriele, C.M., Hoffhines, M.A., Freeman, T.R., and Patterson, B.K. 1989. Acoustic location and tracking of wintering humpback whales off south Kohala, Hawaii. In Abstracts of the Eighth Biennial Conference on the Biology of Marine Mammals, Pacific Grove, Calif., December 7 - 11, 1989. p. 2 1. [Abstr.] Gabriele, C. M. 1992. The behavior and residence characteristics of reproductive classes of humpback whales (Megaptera novaeangliae) in the Hawaiian Islands. M.A. thesis, University of Hawaii. George, J.C., Clark, C., Carroll, G.M., and Ellison, W.T. 1989. observations on the ice-breaking and ice navigation behavior of migrating bowhead whales (Balaena mysticetus) near Point Barrow, Alaska, spring 1985. Arctic, 42: 24-30. Glockner-Ferrari, D .A., and Ferrari, M.J. 1990. Reproduction in the humpback whale Megaptera novaeangliae i n Hawaiian waters 1975- 1988: the life history, reproductive rates and behavior of known individuals identified through surface and underwater photography. In Individual recognition of cetaceans: use of photo-identification and other techniques to estimate population parameters. Edited by P. S. Hammond, S.A. Mizroch, and G.P. Donovan. International Whaling Commission, Cambridge, U.K. pp. 161 -170. Guinee, L.N., Chu, K., and Dorsey, E.M. 1983. Changes over time in the songs of known individual humpback whales (Megaptera novaeangliae). In Communication and behavior of whales. Edited by R.S. Payne. Westview Press, Boulder, Colo. pp. 59-80. Herman, L.M., and Antinoja, R.C. 1977. Humpback whales in Hawaiian waters: population and pod characteristics. Sci. Rep. Whales Res. Inst. Tokyo, 29: 59- 85. Jurasz, C.M., and Jurasz, V. 1979. Feeding modes of the humpback whale, Megaptera novaeangliae, in southeast Alaska. Sci. Rep. Whales Res. Inst. Tokyo, 31: 69-83. Kroodsma, D.E., and Byers, B.E. 1991. The function(s) of bird song. Am. Zool. 31: 318-328. Lambertsen, R.H., Baker, C.S., Duffield, D.A., and ChamberlinLea, J. 1988. Cytogenetic determination of sex among individually identified humpback whales (Megaptera novaeangliae) . Can. J. Zool. 66: 1243-1248. Levenson, C., and Leapley, W.T. 1978. Distribution of humpback whales (Megaptera novaeangliae) in the Caribbean determined by a rapid acoustic method. J. Fish. Res. Board Can. 35: 11501152. McSweeney, D.J., Chu, K.C., Dolphin, W.F., and Guinee, L.N. 1989. North Pacific humpback whale songs: a comparison of

southeast Alaskan feeding ground songs with Hawaiian wintering ground songs. Mar. Mammal Sci. 5: 139- 148. Medrano, L., Salinas, M., Salas, I., Ladron de Guevara, P., Aguayo, A., Jacobsen, J., Baker, C.S. 1994. Sex identification ofhumpback whales, Megaptera novaeangliae, on the wintering grounds of the Mexican Pacific Ocean. Can. J. Zool. 72: 17711774. Mobley , J. R., and Herman, L. M. 1985. Transience of affiliations among humpback whales (Megaptera novaeangliae) on the Hawaiian wintering grounds. Can. J. Zool. 63: 762 - 772. Mobley, J.R., Herman, L.M., and Frankel, A.S. 1988. Responses of wintering humpback whales (Megaptera novaeangliae) to playback of recordings of winter and summer vocalizations and of synthetic sounds. Behav. Ecol. Sociobiol. 23: 2 11-223. Mobley , J.R., Forestell, P. H., and Grotefendt, R. 1994. Results of aerial surveys in Hawaiian waters. Annual report to the Advanced ~esearchProjects Agency, Arlington, ~ a . Norris, K.S., and Dohl, T.P. 1980. The behavior of the Hawaiian spinner porpoise Stenella longirostris. Fish. Bull. 77: 82 1- 877. Payne, K., and Payne, R.S. 1985. Large-scale changes over 17 years in songs of humpback whales in Bermuda. Z. Tierpsychol. 68: 89-114. Payne, K., Tyack, P., and Payne, R.S. 1983. Progressive changes in the song of humpback whales (Megaptera novaeangliae): a detailed analysis of two seasons in Hawaii. In Communication and behavior of whales. Edited by R.S. Payne. Westview Press, Boulder, Colo. pp. 9-57. Payne, R.S., and McVay, S. 1971. Songs of humpback whales. Science (Washington, D.C.), 173: 585 -597. Silber, G. 1986. The relationship of social vocalizations to surface behavior and aggression in the Hawaiian humpback whale (Megaptera novaeangliae). Can. J. Zool. 64: 2075 - 2080. Smultea, M.A. 1992. Habitat utilization patterns of humpback whales, Megaptera novaeangliae, off the Island of Hawaii. Marine Mammal Commission report. National Technical Information Service PB92- 182484. Tyack, P. 1981. Interactions between singing Hawaiian humpback whales and conspecifics nearby. Behav. Ecol. Sociobiol. 8: 105-116. Tyack, P. 1983. Differential response of humpback whales, Megaptera novaeangliae, to playback of song or social sounds. Behav. Ecol. Sociobiol. 13: 49 -55. Tyack, P., and Whitehead, H. 1983. Male competition in large groups of wintering humpback whales. Behaviour, 83: 132 154. Watkins, W.A., and Schevill, W.E. 1972. Sound source location by arrival times on a non-rigid three-dimensional hydrophone array. Deep-Sea Res. 19: 691 -706. Whitehead, H. 1981. The behaviour and ecology of the humpback whale in the northeast Atlantic. Ph.D. dissertation, University of Cambridge, Cambridge. Whitehead, H. 1985. Why whales leap. Sci. Am. 252(3): 84-93. Winn, H.E., and Winn, L.K. 1978. The song of the humpback whale Megaptera novaeangliae in the West Indies. Mar. Biol. (Berl.), 47: 97 - 114. Winn, H.E., Bischoff, W.L., and Taruski, A.G. 1973. Cytological sexing of Cetacea. Mar. Biol. (Berl.), 23: 343 -346. Winn, H.E., Edel, R.K., and Taruski, A.G. 1975. Population estimate of the humpback whale (Megaptera novaeangliae) in the West Indies by visual and acoustic techniques. J. Fish. Res. Board Can. 32: 499 - 506. Zeh, J.E., Clark, C.W., George, J.C., Withrow, D., Carrol, G.M., and Koski, W.R. 1993. Current population size and dynamics. In The bowhead whale. Edited by J. Burns, J.J. Montague, and C.J. Cowles. Spec. Publ. No. 2, Society for Marine Mammalogy , Lawrence, Kans. pp. 409 -489.