THE VOCAL AND BEHAVIORAL REACTIONS OF THE WHITE WHALE , OR BELUGA , ( DELPHINAPTERUS LEUCAS ) TO UNDERWATER PLAYBACK OF NATURAL AND SYNTHETIC BELUGA SOUNDS

Underwater playback of natural and synthetic sounds was shown to be an effective tool for investigation of communication in Delphinapterus leucas, the beluga. Natural sounds were used to determine the significance of the sounds to the animals, and synthetic sounds to determine some of the parameters of the sounds that had an effect on their significance. The general, overall response of captive animals was an increase of interest in the sound source during the playback, with decreasing interest in the three minutes following playback. The general response of free-swimming belugas was a decrease of vocal emissions. A strong, stereotyped, vocal response was elicited from one captive animal by the Harmonic Long, Loud Whistle; and synthetic sounds, based on this natural sound, showed that both duration and frequency affected the significance of this sound. It is suggested that the Jaw Clap or Bang should be regarded as a general 'alerting' or 'attention' call, permiting it to serve as either an alarm signal or a threat. The Squeals of the free-swimming herd may have been associated with the calves, being produced either by the calves themselves or by accompanying adults. Both syntax (the combination of individual sounds into sound-series) and context were important in the conveyance of information by the playbacks. 'Scouting behavior' occurred during the playback of some sounds. A functional classification of animal sounds is proposed. Suggestions are advanced for further work with the vocal behavior of the beluga.

9. Significant increases and decreases in frequency of emission of five sound types during and after playback of FldPBS 2,3,4,6,9,and    Playback of pure tones has been used to investigate parameters of an acoustic system not directly concerned with communication. These parameters include frequency limits of hearing, effectiveness of auditory masking, direction-finding capabilities, frequency discrimination capabilities, and intensity limits of frequency detection (audiograms).
The reseaEch described in this report was designed to determine whether or not the playback technique could be used as an experimental tool for the investigation of the acoustical system of the beluga, or white whale (Delphinapterus leucas, Pallas), by modification of sound emissions and/or behavioral patterns. The beluga was chosen as the experimental subject because of its availability in captivity and in 2 the free state, and because of its known ability to produce a wide range of sounds Lawrence, 1949, 1950;Fish and Mowbray, 1962).
Natural sounds were played back to study the significance and meaning of these sounds, and synthetic sounds were used to determine some of the parameters of the sounds which had a bearing on their significance.
Playback experiments were first used with cetaceans in 1952 to determine the upper limits of hearing of the bottlenose porpoise, Tursiops truncatus. Kellogg and Kohler (1952) and Kellogg (1953) determined that the upper limit of hearing in this species reached at least to 80 kHz by using playbacks of pure tones. This limit was further extended to 120 kHz by Schevill and Lawrence (1953) and to 150 kHz by Johnson (1967), both of whom also used playback of pure tones as their experimental technique.
Pure-tone playbacks were used by Dudok van Heel (1959) to determine auditory direction finding in Phocoena phocoena and by Johnson (1968) to measure masked frequency thresholds in T. truncatus. Other uses of this technique have included determinations of audiograms for several species, for example Inia geoffrensis (Jacobs and Hall, 1972) and Orcinus orca (Hall and Johnson, 1972), and of auditory frequency discrimination limens in!· truncatus (Jacobs, 1972).
Playback of the sounds of a conspecific animal has been used since 1961 as a means of demonstrating acoustical exchanges between dolphins. In that year, Lilly and Miller (1961) showed that acoustic stimuli from one T. truncatus immediately elicited whistles and click trains from another, isolated animal of the same species. Lang and Smith (1965) also showed that an isolated !· truncatus would respond to the sounds of a second individual until one particular sound was 3 played back. Response then ceased for some time, suggesting that different sounds had different significances. Dreher (1966) also found varied vocal and behavioral reactions to playback of six different whistle contours of T. truncatus to that same species.  used conspecific playbacks of eight individuals of T. truncatus to show that one dolphin could discriminate between a random assortment of individuals on the basis of a wide assortment of their whistle emissions, and concluded that certain whistles were specific to the individual producing them (signature whistles).
A third type of playback experiment has been the playback of sounds of one cetacean species to another. Fish and Vania (1971) used playbacks of the sounds of the killer whale (Orcinus orca) to keep belugas from entering the Kvichak River in Alaska during the salmon spawning run, and Cununings and Thompson (1971) used these sounds to affect the behavior of the gray whale (Eschrichtius robustus) during their southward migration past California. In both cases, the animals responded with a flight reaction.  found that T. truncatus was able to discrimina~e between individuals of Del~hinus delphis solely on the basis of the Q. delphis whistles played back. Davies (1962) carried this type of playback even further by playing back killer whale sounds to animals of a different class, the Zambezi River shark (Carcharhinus zambezensis). He found that the largest of five sharks was disturbed by the sounds, swimming around the tank at greatly increased speed.
Thus sound playbacks are seen to be a powerful tool in the investigation of many aspects of cetacean sound. However, the playback of conspecific sounds to animals in captivity and in the field for the purposes of correlating behavior and vocal emissions, and determining the significance of sounds, has hardly been utilized. As mentioned above, Lilly and Miller (1961), Dreher (1966),  have started work in thi s direction with captive animals, but no results of investigations of this nature with cetaceans in the field have been published. In fact, the only published field work of this type done with any of the marine mammals is that carried out by Watkins and Schevill (1968)  through an LTV University MM-2PPS underwater loudspeaker or a Chesapeake J-9 omnidirectional sound transducer. The University loudspeaker was used only for the first series of playbacks at the Aquarium and as a backup sound source during the rest of the study in case of failure of the J-9 transducer. I n the field, recordings were made using a Uher 4400-S Report Stereo tape recorder, and the playback amplifier used was a Realistic PA-25. Frequency response of the system was 50-10,000 Hz, with the possible exception of the University loudspeaker. All field experiments were carried out from a 16-foot Boston Whaler, and all recordings were made at 7t ips tape speed.

Methods
The captive population was recorded over a period of three days in July of 1968. From these recordings, twelve sounds and a control (backgvound tank noise) were chosen as the primary playback sounds.
These sounds are designated throughout this report as PBS 1-13 (Play-Back Sounds 1-13), and are described in Appendix I. Each sound was placed on a tape loop in combination with a ten-second piece of blank tape and re-recorded for three minutes, resulting in alternation of the sound and ten-seconds of silence. Six series of playback experiments were performed: (1) Playback of sounds from the captive population to the captive animals.--The first series of experiments involved the playback of the PBS series of sounds to the four captive whales, all in the same tank.
(2) Playback of sounds from the captive population to the Saguena.y herd.--In July and August of 1970, eight of the captive sounds, plus a 4,8 kHz pure tone, were played back to the Saguenay herd. This series is designated as FldPBS 1 4; 6-10 (Field PlayBack Sounds 1-4; 6-10) and is marked with an asterisk ("k) in Appendix I.
(3) Playback of sounds from the Saguenay animals to the Saguenay herd.--Seven of the sounds recorded from the Saguenay herd in July of 1970 were re-recorded in the manner described above. These seven sounds, Appendices. Descriptions of other sounds that were recorded, but were not affected by playback, and accounts of concurrent behavioral observations will be present~d in later papers.
The nonparametric sign test (Siegel, 1956)   there was no significant change in head orientation toward the hydrophone, but that head orientation toward the sound source increased for every playback sound in Pb, and also decreased for all 13 sounds in PtPb to near-PrPb levels. The whales approached the hydrophone significantly less in Pb, but approached the sound source significantly more in Pb, with a decrease in approaching once more in PtPb.
As can be seen from   The changes of interest shown in Table 3a are sununarized in Table 3b . For a key to the symbols used, see Table 1.  For a key to the symbols used, see Table 1.  (Table 4b).
Alex was also observed to display the least overall interest in the hydrophone at all times.
The strength of the changes of interest from PrPb to Pb toward the sound source by each whale for each playback sound are compared in      The changes of interest shown in Table 4a are summarized in Table 4b. For a key to the symbols used, see Table 1.     The changes of interest shown in Table Sa are summarized in  Table Sb. For a key to the symbols used, see Table 1.       The changes of interest shown in Table 7a are summarized in Table 7b. For a key to the symbols used, see Table 1. Ring was very similar to the Ping, but had a longer reverberation time, and the Click Train was a series of rapid-repetition-rate clicks similar to those used for echolocation in some cetaceans. All changes of significance occurring between PrPb and Pb were decreases in numbers of sounds, while all significant changes between Pb and PtPb were increases.
Since, in the field situation, the sound source was relatively close to the hydrophone, it is possible that the playback sounds themselves masked some of the sounds emitted by the herd during playback. However, all significant changes between PrPb and PtPb, with the exception of one sound, were also decreases and may indicate that the overall decrease of sound production observed between PrPb and Pb was the true situation.
There was no strong specific response elicited from the Saguenay herd by any one of the New York Aquarium sounds such as was observed from the captive animals in response to playback of the Harmonic LLW.
Behavioral reactions.--There was no stereotyped overt behavioral reaction to any of the sounds from the captive animals that were played back to the free-swimming herd. When the herd was traveling up or down the river, the whales continued their transit. When they were milling about in a bay or quiet area, they neither approached nor withdrew from the boat. However, a gray animal, the gray color indicating a young whale, was sometimes observed close to the boat during or shortly after PtPb. It could not be determined whether or not this was always the same individual. This happened most frequently after playback of the Jaw Claps, being observed in five of the ten instances that this sound was used. With the other playback sounds, the most often that this event was recorded was one out of ten or two out of twelve repetitions.
Diving times.--Diving times were taken for single animals or for a single group of animals within the herd during each playback experiment. The mean dive times for each of the three periods, averaged over all repetitions of each sound, are presented in Table 10. Use of the modified t-test for unequal sample sizes and variances showed no significant changes in diving times in response to any of the playback sounds.
The Jaw Claps were associated with the greatest change in the mean dive, showing a trend toward increase in length of dive in Pb, followed by a decrease in PtPb to PrPb levels. Neither of these trends were statistically significant.
To obtain a mean overall diving time for the beluga, the diving times recorded during all PrPb periods (761 dives) were averaged. This mean dive for an undisturbed, free-swimming beluga was found to be 25.7 seconds. However, all changes of significance involved decreases in sound production, whether in Pb or in PtPb. Once again, the playback sounds may have masked sounds being made by the whales during Pb, but the overall decreases shown in PtPb (to below the levels of PrPb or Pb) suggest that the overall reaction in most cases was a decrease in sound production.
Behavioral reactions.--Observations were recorded concerning any whale or whales coming toward or retreating from the boat, any deviation in the path of the animals in the cases when they were traveling up-or down-river, and any pause in such a transit. Table 12 is a summary of these observations, summed over the two years of field playbacks. The playback situations were divided into two categories: milling (circling or seemingly random swimming in a bay or estuary) or transiting (traveling up or down the river). Reactions to the playbacks of certain sounds were noted much more commonly when the animals were milling than when making their daily passage up and down the Saguenay River. A positive response (+) when milling indicates that at least some of the herd was headed directly toward the boat during Pb and/or early PtPb, and that  For a key to the symbols used, see Table 1.  ., Each figure represents the mean duration, in seconds, ef all dives during that period for all repetitions of that SagPBS. 31 jured by Blanchon, the dominant male, and to enable the start of a training program with Alex. The sounds from the Saguenay herd were played back to the three animals remaining in the large beluga tank, and also to Alex alone in the separate pool.
It was immediately apparent when working with the three belugas, that Alex had been either the most vocal of the four whales when they were all together, or that his presence had caused more vocalizations from the other three animals. Overall numbers of sounds were much lower in the group tank after Alex' removal. In fact, Alex was found to be more vocal alone than were the other three belugas together.
Vocal reaction.--There were no significant changes (see Appendix VIII) in the vocal emissions of Blanchon, Frances, and Ethel in response to any of the Saguenay playback sounds.
Changes of interest directed toward hydrophone and sound source.--A summary of the significant changes of interest directed toward the hydrophone and sound source by all three whales considered together is presented in Table 14. The complete data for all seven sounds are presented in Appendix IX. As shown in columns one and three of Table 14, there were no significant changes of interest shown toward the hydrophone in reaction to any of the Saguenay playback sounds. Orientation toward the sound source increased significantly in Pb for five of the Saguenay sounds, the exceptions being the Moans and the Screams-and-Wails, These increases were all followed by decreases in PtPb. No significant changes in approaching or touching the sound source were elicited.   For a key to the symbols used, see Table 1.
The changes of interest shown in Table 15a are sununarized in Table 15b. For a key to the symbols used, see Table 1. 34 the hydrophone or sound source. The whales oriented toward the sound source more often in Pb than in PrPb in response to all seven playback sounds (Table 15b), and approached the sound source more often in Pb in response to all sounds except the Squeals. There were no significant changes associated with either orienting toward or approaching the hydrophone. Again, as was the case when the sounds from the captive animals were played back to the four captive animals, the overall pattern was an increase of interest in the sound source in Pb, regardless of which sound was being played back, followed by a decrease of interest in PtPb .

35
Considering each beluga separately, no whale showed any signi- The changes of interest shown in Table 17a are summarized in Table 17b. For a key to the symbols used, see Table 1. Since the frequencies of greatest energy content in the natural

40
Harmonic LLW were about 4.8 and 2.4 kHz, these were two of the frequencies used in the synthesized sounds. Both of these frequencies produced highly significant increases of Harmonic LLW emissions during Pb when presented at a 30-second duration, followed by an equally significant decrease in PtPb.
The third frequency used for synthetic playbacks was 3.3 kHz, an atypical frequency in the sense that the Harmonic LLW was never emitted Pb when only the 6 completed playbacks were considered. When projected to a repetition of 12 playbacks, these data would have been expected to show a highly significant increase in Pb, and a highly significant decrease in PtPb.
In        Table 2la are summarized in Table 2lb. For a key to the symbols used, see Table 1 The overall tendency, when total number of circuits during all repetitions of each sound were compared (Table 22), was an increase of activity in Pb and a decrease a n PtPb. The Harmonic LLW was never heard from the free-swimming herd in the Saguenay River, nor did it elicit any reaction from the Saguenay animals. It also was not recorded from any of the other animals held captive at the New York Aquaritun, and did not elicit a vocal response from any of them.

50
The biological significance or 'meaning' of the Harmonic LLW is not known, but it seemed to occur in situations that might have been described in htunans as productive of 'impatience' or 'expectation', such as at the expected time of feeding or before times of training sessions or public exhibitions. The very fact that it occurred so seldom, but was so uniformly elicited by playback of the same sound, suggested that it occurred only in a very specific context. This was in sharp contrast to such commonly occurring sounds as the shorter less strident 'whistle' which was heard at any time of day and under almost any circtunstance.
Also, since the effect of the Harmonic LLW playback carried over into PtPb, it can be assumed that the effect on Alex was not merely that of a stimulus-response reflex action, but was instead an increase in his overall level of arousal that continued after cessation of the initiating stimulus.
The Contact Sound-Series was an extended combination of various sound types (Barks, Squawks, Jaw Claps, Whistles, Squawls, Buzzes, Whinnys, and Chirps) emitted at times of physical contact or close proximity between two or more whales, or at times of major disturbance.
Apparently the Contact Sound-Series was indicative of a high state of arousal. Further, the specific sounds responsible for conveying this context may have been the Squawk and the Jaw Clap, two of the more prominent sounds occurring at the time of maximtun disturbance during the emission of the Contact Sound-Series (Morgan, 1970 have been expected to show such a decrease. The Squawk was not a common sound out of the context of the Contact Sound-Series. It was apparently a sound associated with a high state of arousal (Morgan, 1970) brought on by either fright or inter-individual contact or proximity, and was thus of uniform low occurrence as a separate sound in all playback experiments. The situation with regard to the Whistle and the Chirp was exactly the reverse. These sounds were associated with any disturbance, however slight, inside or outside the tank, and were thus taken as being indicative of a very low state of arousal when occurring alone (Morgan, 1970). They were of uniform high occurrence throughout most of the playback experiments and would not have been expected to decrease in occurrence during playback. They were also the 'finishing sounds' of nearly all Contact Sound-Series, seeming to occur for some time after the cessation of the Contact Sound-Series, much as a small bird will occasionally 'peep' while calming down after being frightened.
Changes of interest directed toward hydrophone and sound source.--Whenever any sound was played back to the captive whales the general, overall response was an increase of interest in the sound source during Pb with consequent decrease of interest in other objects in the tank.
This was followed by a decrease of interest in the sound source in PtPb (Tables 2 and 3). Tables 4-7 demonstrate that this general reaction was true for all four belugas. This reaction showed some especially interesting relations existing between three separate sounds: the Buzz, the Whinny, and the Jaw Clap (refer to Table 2 for the following discussion). The Whinny was never observed to occur alone; it was always emitted in combination with the Buzz, and may have been incidental to the production of the Buzz. The Buzz was often heard without the ac-companying Whinny. The PBS designated the Whinny was actually such a Buzz-and-Whinny combination, with the Whinny being louder relative to the Buzz than was generally the case. The Whinny was found to elicit significant increases, during Pb, of both orientation and approach 53 toward the sound source, and also to elicit a decrease in total sound production in PtPb (Table 1). During PtPb there was a significant decrease of both orientation and approach toward the sound source. With the Buzzes alone, the only significant reaction observed was an increase in orientation toward the sound source in Pb. There was no significant change in vocal emissions in response to the Buzzes. The overall reaction of the animals to the Buzz was thus less than to the Whinny.
When the Buzz-and-Whinny combination (with the Buzz being dominant, as was usually the case) was played back, the reaction seemed to be a combination of those seen in response to the Whinny and the Buzz separately (Table 2). Interest during Pb increased significantly with respect to both orientation and approach toward the sound source. This was followed by a decrease in orientation toward the sound source during PtPb , but not by a decrease in approaching the sound source such as was observed with the Whinny. Thus, interest remained higher in PtPb than in PrPb.
Therefore the Buzz-and-Whinny combination was as effective or more effective with regard to increasing interest in the sound source than were either of the component sounds alone, and produced a longer-lasting effect, carrying over into PtPb.
The Jaw Clap, when played back to the four captives, produced no significant changes of interest toward the sound source. When a combination of the Jaw Clap, the Buzz, and the Whinny was played back, there was also no significant changes of interest observed toward the sound source (Table 2). Thus, the Jaw Clap inhibited the effect of the Buzz-and-Whinny. In fact, the Jaw Clap-Buzz-and-Whinny combination was one of only 2 sounds of the 13 in the PBS series 1 tlhat showed a trend toward producing a decrease of interest in the sound source during Pb (Table 3a). The other sound showing this tendency was the Contact Sound-Series which, like the Jaw Clap-Buzz-and-Whinny combination, was a combination of various sound types, was of relatively longer duration, and included the Jaw Clap. As discussed above (p. 51), the Contact Sound-Series may have affected the whales by redirection of interest from sound source to tankmates. Perhaps a similar response to the Jaw Clap-Buzz-and-Whinny combination was responsible for the decrease of interest shown toward the equipment during playback of that sound also, There was, however, no voeal response associated with playback of the Jaw Clap-Buzz-and-Whinny combination such as occurred in the case of the Contact Sound-Series playbacks ( Table 1).
One of the major differences between the two types of sound series just mentioned was the occurrence of the Squawk as a dominant sound in the Contact Sound-Series, while Squawks aid not occur in the Jaw Clap-Buzz-and-Whinny combination. When two types of Squawks were played back to the four belugas, interest toward the sound source increased significantly during Pb in reaction to both sounds (Table 2) The inhibitory effect of the Jaw Clap on the Buzz-and-Whinny is further demonstrated in Table 8, which shows the strengths of the changes from PrPb to Pb associated with each sound for each animal. The Buzz, the Whinny, and the Buzz-and-Whinny produced large changes, while the Jaw Clap-Buzz-and-Whinny combination showed the least response toward the equipment for all four whales.
These intersound relationships may be summarized in the following manner: (1) Combinations of sounds apparently had a different significance for the belugas than did the component sounds by themselves.
This was seen with the Buzz-and-Whinny combination, which had a longer lasting effect than either the Whinny or the Buzz, and with the Jaw  Behavioral reactions.--There were no significant behavioral reactions observed to the playback of any of the 'captive' sounds to the free-swimming animals, nor were there any significant changes of duration of diving times (Table 10). Either these sounds carried no significance for these animals or, conversely, whatever significance was normally associated with each sound was not conveyed ' b o the animals in the conditions under which the playbacks were carried out. Morgan (1970) suggested four possible reasons to explain the absence of behavioral reaction. These are: (1) The background noise levels in the recordings from the aquarium tank may have masked the sounds or affected their 'reliability' to the animals.
(2) The sounds themselves may have been modified by the standing waves or multiple echoes in the concrete aquarium tank, thus affecting their 'reliability' to the unrestrained animals.
(3) The sounds produced by the captive belugas may have been modified by the animals themselves during their period of captivity.
(4) Animals from different localities or herds may have differ-ent dialects.
All four of these possibilities concern the suggestion that the sounds, as presented to the animals, carried no significance due to modification by the tank conditions or the animals, or by being unfamiliar sounds. The fact that a young animal was observed close to the boat during or shortly after playback is puzzling, but may be regarded as supportive evidence for 1~3 above. This reaction, if indeed it was a reaction, was observed most often after playback of the Jaw Clap, an abrupt, loud 'crack' or 'bang' usually associated in the beluga with rapid closing of the jaws. This sound has been associated with alarm or threat in several cetacean species (Wood, 1953;Caldwell, Haugen, and Caldwell, 1962;Fish and Mowbray, 1962), including the beluga.
There were no frequency variations or amplitude modulations associated with this sound, and it probably conveyed its meaning by being a very loud sound with a sharp onset. When heard in the Saguenay River, the Jaw Clap resembled a rifle shot and dominated any other sounds being emitted at the same time. Since this sound did not rely upon subtle frequency variations or amplitude modulations to convey its significance, it, of all the aquarium sounds used, would have been the least modified by standing waves or background noise.
With reference to the nearness of the younger whales after playback of the Jaw Claps, perhaps tihey were less able to determine the 'reliability' or 'unreliability' of sounds than were the adult whales.
Thus, it would have been expected that if any reaction was to have been elicited by the playback of sounds recorded in aquaria, it would have been from the younger animals and in reaction to the least modified of the sounds. If true, this would mean that learning plays a part in the 58 59 developing effectiveness of the sonic system of the beluga.
There are two more possibilities that could have accounted for the absence of any reaction to the aquarium soundsJ both having to do with the idea that althoggh the sound types may have been familiarJ and not modified before or during playbackJ the normal significance of the sounds was not conveyed in the conditions under which they were played back: (5) The significance of the sounds may have been context-specific. That isJ they may have had no meaning to the animals if not received in the same context in which they were produced.
( 6) The sounds may have had to be presented within a particular syntaxJ at the same time dependent upon contextJ in order to be of significance.
If either of these possibilities was the caseJ a weaker or no reaction would have been expected to the playback of single sounds recorded in captivity. As was suggested above (p.56)J syntax may well have been of importance in the effectiveness of information transfer.
Evidence that context was also important will be presented in the following section of this paper.

II. Playback of Sounds Recorded From the Saguenay Herd
A. Playback to the Saguenay River herd.
Vocal reaction.--None of the seven sounds recorded from the Saguenay River herd that were used as playbacks elicited a consistent vocal reaction from these same animals. All of the few significant changes during the playbacks of both 1970 and 1971 (Tables lla and llbJ respectively) were decreases in numbers of sounds emitted in Pb or PtPb as compared with PrPb. As was discussed above, the decreases that occurred between PrPb and Pb could have been due to the close proximity 60 of hydrophone and sound source. However, as was the case with the sounds from the captive animals, decreases were also observed between PrPb and PtPb, and these are regarded as a reaction to the playq ack sounds. Since the results can not be attributed to one particular sound, but were all decreases in sound emissions, it seems that the belugas reacted to extraneous sounds in their environment by decreasing the number of sounds that they emitted themselves. This interpretation was supported by the observation that the vocal activity of the free-swimming belugas decreased markedly in the presence of the noise from passing boats. This decrease was noted well in advance of the time the boats approached the beluga herd.
Behavioral reactions.--The behavioral reactions to the playbacks depended on the context in which the sounds were played back (Table 12).
Three sounds drew the whales toward the research boat a significant number of times when the animals were milling in a bay or estuary, but had no effect when the whales were traveling up or down the river. On the other hand, one sound was found effective in drawing the whales toward the boat during transit, but had no effect while they were milling.
No sound was found effective in both the milling and the transiting contexts, and no sound elicited a significant change in durations of dives (Table 13).
Two of the sounds that attracted the whales while milling were in reality combinations of sounds (Table 12). The Blats-and-Ping was a combination of three Blats with one Ping at the end, and the Screams-and-Wails was an extended series of several sounds (see Appendix II for complete descriptions). Again, combinations of sounds were more effective in eliciting a reaction from the belugas, and thus presumably were more effective in conveying information to the animals.
The third sound found effective during milling was the Jaw Clap (Table 12), which in this case attracted both young and adult belugas.

61
As was discussed above (p. 58), the Jaw Clap , Crack, or Bang has been described as an alarm or fright call for several cetacean species. In the light of the present experiments, it is felt that a better description of the significance of this sound might be as an 'attention' or 'alerting' cc a ll, produced in reaction to an alarm or antagonistic context. A sound that had evolved for this purpose would be expected to have been startling, loud, and definitive. The Jaw Clap, with its abrupt onset , wide frequency spectrum, and high intensity fulfilled these requirements. The reaction to such an 'alerting' sound might have been expected to be an approach toward the animal producing the sound in order to gather more information concerning the cause of the disturbance , or for mutual protection. The Jaw Clap was the only single sound used as a playback that produced a significant response in the milling context (Table 12) . In captivity, when combined with other sounds, it inhibited reaction to the other sounds ( Only one of the seven Saguenay sounds used as playbacks, the Squeals, elicited an approach from at least a part of the herd while it was moving up or down the river. When the whales were making such a transit the calves were found concentrated toward the rear of the herd, which was usually very strung out. These calves were nearly always accompanied by an adult beluga, presumably the mother. This was the portion of the herd from which the most Squeals were recorded, and it is suggested that these sounds were associated with the calves. They coa ld have been produced by the young themselves, by the females accompanying the calves, or by both. Although the Squeals were heard while the whales were either milling or transiting, the playback was effective in attracting them only while they were transiting. Possibly it was more likely that a young beluga would have become separated from its attendant female while the herd was moving than while it was milling in quiet water. If the Squeal functions for maintenance of contact between calf and mother or between calf and entire herd, or as a general distress call of the you~g, it would have been most effective as a playback in the situation where loss of contact or distress was most likely. The response was usually elicited from more than one adult beluga, often accompanied by calves. There are numerous reports concerning several species of cetaceans that more than one female is involved in the care of one calf (Caldwell and Caldwell, 1966).
Usually, when a positive reaction was elicited by one of these playback sounds only a few of the animals were involved. However , on one occasion during the 1970 season, the response was shown by the en-63 tire herd. This was the strongest response observed during both years of field work and was elicited by two playback sounds, the Screams-and-Wails and the Blats-and-Ping. Since this is regarded as further evidence that syntax was important in the transfer of information and thus in eliciting responses, it will be described in some detail. The entire series, from start to finish, lasted approximately one hour.
On The Blats-and-Ping was immediately followed by the playback of the Squeals. During the first two minutes of this three-minute playback no whales were observed on the surface. Three groups then surfaced about thirty yards from the boat and remained at that distance until the playback ended. After the Squeals ended the animals moved on past the boat and started to swim away. The Screams-and-Wails was then played back again, and the herd turned once more to approach the boat. At this time an adult white animal passed beneath the sound source at an esti-mated depth of six or seven feet. Once again, when this playback ended the belugas started to move off. When the Screams-and-Wails started again, the herd turned and approached to within twenty-five yards, where they stopped and began milling about. At this time two large adult belugas separated from the herd and swam directly and steadily toward the boat. One of these two was not seen again, but the other came to a position directly under the sound source, stopped, turned onto its side, then onto its back, and inclined its head upward toward the sound source.
This animal then swam from view, and although the Screams-and-Wails playback continued for more than five minutes afterward, the entire herd moved away and proceeded on toward the mouth of the e Saguenay River.
Further playback was not effective in drawing the herd back toward the boat.
This reaction, elicited by the playback of natural sound combinations, was a clear demonstration of scouting behavior in the beluga.
Other instances of scouting behavior in this species were observed during the field work and will be fully described in a separate report.
Instances of scouting behavior have also been reported in other cetacean species (Evans and Dreher, 1962;Caldwell and Caldwell, 1964;Caldwell, Caldwell, and Siebenaler, 1965).

B. Playback to three captive animals
Vocal and behavioral reactions.--The playback of sounds from the free-swimming animals elicited no specific vocal or behavioral reaction from the captive animals (Appendix VIII and Table 14, respectively).
This was also the case when aquarium sounds were played back to the freeswimming animals. The overall increase of interest in the sound source during Pb was observed again (Table 15a, b) and was shared by all three whales. Also, the interest decreased to near-PrPb levels during PtPb.
This was the same effect as was seen in reaction to playback of the aquarium sounds to the captive belugas (Tables 3-7). It would thus seem that the normal reaction of captive animals to the presentation of a sound in their tank was an increase of interest directed toward 65 the source of the sound. When such sounds were originally recorded from the same captive population, the strength of the increase was dependent on the specific sound being played back, whereas when the sounds were recorded from a different population, the response was not graded in relation to the sound type.
This latter fact is shown by comparison of Tables 8 and 16, which show the magnitudes of the changes of interest from PrPb to Pb directed toward the sound source by each captive beluga. In response to the sounds from the same population (Table 8), certain sounds are seen to have had a consistently greater effect in all four whales than did others of this series of sounds. In response to sounds from a different population ( Two of the possibilities suggested as reasons for the lack of response shown by the Saguenay animals to the aquarium sounds can be postulated as also having been important in the lack of specific response shown by the captive belugas to the Saguenay ~ounds. These are the possibility of dialects in different populations, and the possibility that sounds are context specific. Also, the sounds may have been rendered 'unreliable' when introduced into the reverberatory concrete tank. It is difficult to place any meaning on the sexual behavior ob-  , then Blanchon may have recognized the Harmonic LLW as a sound emitted by a previous tankmate and competitor, thus becoming aggressively rather than sexually aroused. Kleinenberg, et al (1964) stated that the main mating period of belugas in all seas seemed to be late April to early May, with isolated matings taking place from the end of February until the end of August. Vladykov (1944) Table 17b, which shows that orientation toward the sound source increased during Pb of all seven sounds, and in Table 18, which shows that the number of times Alex circled the tank also increased in response to all seven sounds.
In general then, the two pa~terns found previously were also evident in this playback series to a single animal. First, the overall pattern of reaction to playback was an increase of interest in the sound source during Pb, falling off again during PtPb. Secondly, sounds recorded from one population had less effect when played back to an animal from a second population than when played back to that same population.

III. Playback of Synthetic Sounds to a Single Captive Animal
Vocal react · ii!On. --When synthetic sounds, based on the Harmonic LLW, were played back to Alex, the effectiveness of the playback was dependent upon both frequency and duration (Table 19). At the natural frequencies 68 (2.4 and 4.8 kHz), the most effective duration was 30 seconds, an atypical duration much longer than that of any Harmonic LLW emitted by Alex during the period of observation. At the natural durations (1.7 and 2.7 seconds), the most effective frequency (using the projected results for the 3.3 kHz playbacks) was 3.3 kHz, an atypical frequency never recorded in a Harmonic LLW from Alex. Thus, each of the four combinations that involved either a natural frequency combined with an atypical duration of a natural duration combined with an atypical frequency produced a highly significant increase of Harmonic LLW's in Pb, followed by a highly significant decrease in PtPb in three of the four cases. When both parameters were atypical, there was no significant increase, although the projected data showed a significant decrease in PtPb. In the four cases where both parameters were natural, the 1.7-second duration and the 2.4-kHz frequency were most effective. The only significant reaction in these four cases occurred as an increase of Harmonic LLW's in Pb when the 1. 7-second duration and the 2. 4-kHz frequency were combined.
Thus it seems that the most effective combination of para.meters  (Table 20).
Behavioral reactions.--As was the case when using natural sounds as playbacks to captive animals, the overall reaction to the playback of the synthetic sounds was increased interest in the sound source (Tables 21 a,b) and increased activity (Table 22)  This brings up the question of whether sounds and behavior patterns are produced in reaction to a particular context (are responsive), or whether the sounds themselves are the stimulus for the production of other sounds or behavior patterns (are causal). The results just mentioned in relation to the Harmonic LLW playbacks suggest that at least some sounds may be both responsive and causal. However, not all sounds produced by a species would necessarily be expected to act in this manner. Some would be expected to occur only in response to a particular context and not elicit further reactions from other animals, while others might be produced expressly for the purpose of eliciting a reaction from another animal. This latter type of sound is the type used for communication in the human species, assuming purposive thought and action on the part of the emitting individual, and so far not definitively shown to be used by any non-human species, although such use has been sug-gested in at least two cetacean species, T. truncatus (Lilly, 1963) and T. gilli (Evans and Dreher, 1962).
All sounds of animal origin may thus be c 1assified , as one of three types: (1) Responsive (2) Causal-unintentional (3) Causal-purposive Sounds classified as responsive would not be associated with communication between animals, but would be produced in a manner analagous to that of a human absent-mindedly humming or whistling while he works. In other words, these are sounds produced in response to a certain context that serve no function in transmitting information from one individual to another. This is not to say that the sounds may not serve some f unction for the individual emitting them. The whistling of a youngster walking home in the dark does not constitute a form of communication, but does perform the function of allaying the fears of the youngster. Such outward expression of internal emotions may also be important in non-human species. One must not assume at the outset that all animal sounds are communicative in function.
If we assume, as we must until it is proven otherwise, that nonhuman animals do not produce sounds of the third division ( causal-purposive), then all animal sounds must be regarded as either responsive or causal-unintentional. Causal-unintentional sounds may be defined 72 as those uttered in response to a particular stimulus-context and which, when received by another animal, transmit information regarding that context to the receiving animal, causing it to react vocally and/or behaviorally as it would to the original context. If one looks at the playback of natural sounds to animals in the light of these two divisions it is easy to realize that not all sounds would be expected to elicit a reaction from the animals regardless of the context in which they were played back. Only those sounds which would be classified as causalunintentional would be expected to elicit a reaction.
Context may also influence the effectiveness of sounds in eliciting a response in playback experiments of the nature reported in this paper. The causal-unintentional sound is normally emitted in reaction to a certain set of conditions, or context, which serves as the stimulus for the production of that sound. Would it not be reasonable, then, to assume that some of these sounds would be effective in eliciting a normal reaction only if played back in a context similar to that under which the sounds are normally produced? In other words, all sounds are context-dependent regarding production, but may be either context-dependent or -independent in reference to reception. The results presented herein support this contention. Sounds recorded from the Saguenay herd were played back to the same herd in two contexts: while milling in bays or estuaries along the river, or while moving up or down the river. Of the four sounds found to be effective in eliciting a response from the belugas, three were significantly effective only while the herd was milling, and one was significantly effective only while the herd was moving. In addition, none of the sounds recorded from the belugas at the New York Aquarium elicited a significant response from the Saguenay animals, and none of the Saguenay sounds elicited a response from the captive animals.
On the other hand, one should not assume that all animal sounds have meaning only in a particular context. There are many types of 74 sounds of general meaning that would be expected to convey their meaning in any context of the animal receiving the sound. Examples of such sounds would be those associated with danger, distress, threat, and other states of high emotional arousal, and the signature whistle , which serves for individual recognition .
Since the Harmonic LLW elicited responses from Alex at all times of day or year and when alone or with other belugas, it is concluded that it was one of these types of sounds, being produced in response to some particular context, but having meaning to the receiving animal regardless of the context in which it was received. A suggestion concerning the reason this sound affected only Alex will be discussed later. Thus sounds classified as causal (unintentional or purposive) can be further subdivided as being context-dependent or context-independent with reference to the context in which they are received. Responsive sounds, by definition, are context-dependent only as regards their production.
They are context-independent regarding reception since they are uniformly ignored by the receiving animal in any context.
Syntax may be defined as the ways in wh i ch individual sounds or words are combined to form a code, or message. The results presented herein support the conclusion that syntax was important in the transfer of information from one beluga to another by sound. When played back to the captive belugas, the Buzz-and-Whinny combination was found to be as effective or more effective in regard to increasing interest in the sound source during Pb than were either of the two component sounds alone , and to produce a longer lasting effect, carrying over into PtPb.
The Jaw Clap alone produced no significant changes of interest in the sound source, but when combined with the Buzz-and-Whinny, completely 75 inhibited the increased interest response normally shown in reaction to the Buzz, the Whinny, and the Buzz-and-Whinny. In fact, the Jaw Clap-Buzz-and-Whinny combination showed a tendency to decrease interest in the sound source, as did the Contact Sound-Series, the only other extended series of sounds played back to the captive animals. This decrease of interest may have occurred by wayy of a redirection of interest from sound source to tankmates as a result of the stimulus presented by these two types of sound-series.
The conclusion that syntax was important in information transfer between belugas was further supported by the fact that there was a significant increase in emission of the Contact Sound-Series during playback of the Contact Sound Series, but not during playback of the Jaw Clap-Buzz-and-Whinny combination. The Contact Sound-Series was normally emitted only during instances of major disturbance in the tank such as outside interference, or contact or close proximity between individuals. Since there was no outside interference during these playbacks, the stimulus for the sound series' production must have arisen from contact or near-contact between the belugas. It is concluded that and Ethel were all captured on the East Coast of North America, whereas Alex was obtained from a West Coast population. It is suggested that different populations of belugas may possess different dialects, "in which (the sounds are) similar among most or all individuals living in a particular locality, but are different from one locality to another" (Lemon, 1967). As Lemon states in reference to birds, it seems reasonable that animals exhibiting dialects should respond more to sounds of their own particular dialect. If indeed dialects do occur in cetaceans, one would surely expect to find them in populations as widely separated as those of the belugas of the East and West Coasts of North America.
It has been shown by numerous investigators that sounds differ between cetacean species, and also between individuals of one species with enough regularity that individual recognition is possible , even to the point of recognition of individuals of a different species . It would seem unreasonable, then, not to expect differences between sounds of populations as geographically isolated as the Pacific and Atlantic belugas. Such dialects might even be expected to occur between herds along one coast, in which context they would serve to maintain individual herd-integrity during the times of congregation which are reported to occur for fattening and migration (Kleinenberg, et al;. When working with both captive and free-swimming animals, a general reaction was observed to all sounds played back which, although different between the captive and free contexts, would serve the purpose of putting the animals in a better condition for receiving further information about the sound stimulus. With captive animals this general reaction was an orientation or actual approach toward the sound source (except in the cases of the Jaw Clap and the two sound-series, as discussed above, p. 74). With the free-swimming animals the general response was a decrease in number of sound emissions, a reaction which was also noted upon the approach of a motorboat toward the herd. Whether these reactions were merely those prompted by curiosity, or were a conscious attempt to discover further information about the stimulus cannot be said at this time. The end result however, would have been the same in either case; !·~·' the animals would have been in a more appropriate state for the reception of more information about the 1 stimulus , its source, and its reason for occurrence. The single instance of scouting behavior described would suggest a purposeful attempt to learn more about the sound and its source, but such a teleological explanation can only be advanced as a suggestion at present.
Synthetic playback segments, based on the characteristics of the Harmonic LLW, were constructed by use of a pure tone oscillator .
From the results of these playbacks it is concluded that both frequency and duration were important in the ability of a sound to transfer information from one beluga to another.

79
It is also evident that this association of sound and behavior can not be carried out to best advantage by working with an entire herd in a large and deep river such as the Saguenay. This leaves open but one path: that of enclosing a restricted number of animals in a natural, shallow, relatively large area. Furthermore, the animals to be observed in such a natural enclosure should be of known relationship to one another. It is thus concluded that the next step in selected-sound playbacks should be performed using a mother-calf pair held in one of the natural estuaries along the Saguenay River, or some other enclosable bay within the natural range of the beluga. Such an experimental setup, using the methods described in this paper, would permit the observer to determine the positions of the two animals relative to one another, and, by the use of three or four calibrated hydrophones simultaneously, which was producing the sounds (Watkins and Schevill, 1971). (2) The Harmonic, Long, Loud Whistle elicited a strong stereotyped vocal reaction in captivity from the animal that originally emitted the sound. The response was positive twenty-three out of twentyseven trials. This sound may have occurred normally in association with 'impatience' or 'expectation' contexts.

83
(3) In captivity, contact between individuals or major disturbance led to emission of an extended series of sounds designated the Contact Sound-Series. Playback of this sound-series elicited increased emission of the Contact Sound-Series with the usual inter-individual contact being present in each case. The mechanism by which this response was elicited might have been either direct, by initiating contact, or indirect, by increasing the level of activity in the tank.
Other changes elicited by playback of the Contact Sound-Series are also discussed.
(4) It is suggested that the Jaw Clap or Bang emitted by several 84 cetacean species should be regarded as an 'attention' or 'alerting' call, rather than as a sound of specific meaning. This sound would then be able to serve as either an alarm or a threat, with its particular meaning being determined by the context in which it was emitted and/or received.
(5) The Squeals recorded from the Saguenay herd may have been associated with the presence of beluga calves.
(6) The effectiveness of synthetic-sound playbacks was dependent upon both frequency and duration.
(7) The general, overall response of the captive animals to playback sounds was an increase of interest directed toward the sound source during the playback, with consequent decrease of interest directed toward other objects in the tank. Interest in the sound source then fell off in the three minutes following playback.
(8) The general, overall response of the wild animals to playback sounds was a decrease of vocal emissions during the playback, followed by increasing emissions after playback, but with an overall decrease from the three minutes preceeding playback to the three minutes following playback. Thus belugas may react to extraneous sounds in their environment by decreasimg the mtimber of spmnds that they emit themselves.
(9) Combinations of sounds had different significance for the belugas than did the component sounds by themselves. This suggests that the syntax of beluga sounds was important in the conveyance of the significance of the sounds, probably by placing the sounds in a meaningful context, (10) Behavioral reactions of free-swimming belugas to playback of sounds from their own herd depended on the context in which the sounds were played back. Three sounds were found effective only when the whales were milling in a bay or estuary, and one sound was effective only when the whales were moving up or down the river. * PBS-9. SQUAWK (TYPE-1) (Fig. ld) A raucous-sounding noise with harmonics of alternating intensity suggesting a burst-pulsed sound (Watkins, 1967) with burst-pulse repetition rate of 400/sec. and rising pulse tone from 1.2 kHz at onset to 1.3 kHz at termination. The type-1 squawk shows this simple rising inflection (compare with the type-2 squawk; PBS 11). Duration in this case is 0,60 sec.
The rising inflection of a type-1 squawk may occur smoothly, as in this case, or in stepwise fashion. * PBS-11. SQUAWK (TYPE-2) (Fig. lk) A raucous sound similar to the type-1 squawk (PBS-9), but with a rise-fall inflection rather than the simple rising in-96 flection of the type-1 squawk. Again, an alternating harmonic structure suggests a burst-pulsed sound with burst-pulse repetition rate of 400 bursts/sec. (determined from the harmonic interval; Watkins, 1967) and pulse tone (determined by pulse repetition rate; Watkins, 1967) of 450 Hz at onset, increasing to 1 kHz at the peak of the curve, and falling back to 450 Hz at termination. Duration; 0,41 sec.

PBS-12. CONTROL
Background tank noise. Most energy below 300 Hz, nearly all energy below 1 kHz.
~·< PBS-13. PURE LONG, LOUD WHISTLE (PURE LLW) (Fig, l,t) A single, thin, high, whistling sound somewhat resembling the harmonic LLW, but much lower in intensity, and without strong harmonics. Frequency is constant at 4.5 kHz throughout, with a weak harmonic sometimes apparent at 2.25 kHz. Accompanied by a pulse-train with pulse rate of about 50/sec. Duration; 3.12 sec.