Blowing bubbles: An aquatic adaptation that risks protection of the respiratory tract in humpback whales (Megaptera novaeangliae)


  • Joy S. Reidenberg,

    Corresponding author
    1. Center for Anatomy and Functional Morphology, Mount Sinai School of Medicine, New York, New York
    • Box 1007, Center for Anatomy and Functional Morphology, Mount Sinai School of Medicine, 1 Gustave L. Levy Place, New York, NY 10029
    Search for more papers by this author
  • Jeffrey T. Laitman

    1. Center for Anatomy and Functional Morphology, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author


Cetaceans (whales, dolphins, and porpoises) have developed extensive protective barriers to exclude water or food from the respiratory tract, including valvular nostrils, an intranarial elongated larynx, and a sphincteric soft palate. A barrier breach can be lethal, as asphyxiation may occur from incursions of water (drowning) or food (choking). Humpback whales (Megaptera novaeangliae), however, exhibit a possibly unique and paradoxical behavior concerning respiratory protection: they release a “bubble cloud” (a cluster of tiny bubbles) underwater from the mouth. How they do this remains unclear. This study tests the hypothesis that the larynx plays a role in enabling bubble cloud emission. The anatomy and position of the larynx was examined in seven specimens of Megaptera novaeangliae. Results indicate that the epiglottis can be manually removed from behind the soft palate and placed in the oral cavity during dissection. Unlike that of toothed whales (odontocetes), the humpback whale larynx does not appear to be permanently intranarial. The elongated and trough-shaped epiglottis may function as a tube when placed against the undersurface of the soft palate and, thus, facilitate channeling air from the larynx to the oral cavity. The pointed tip and lateral edges of the epiglottis fit tightly against the undersurface of the soft palate, perhaps functioning as a one-way valve that lets air out but prevents water from entering. Bubble cloud generation likely involves air passing directly from the larynx into the oral cavity, and then expulsion through the mesh of the baleen plates. A laryngeal–oral connection, however, compromises the anatomical aquatic adaptations that normally protect the respiratory tract. A potential for drowning exists during the critical interval in which the larynx is intraoral and during re-insertion back to the normal intranarial position. The retention of this risky behavior indicates the importance of bubble clouds in predator avoidance, prey capture, and/or social signaling. Anat Rec, 290:569–580, 2007. © 2007 Wiley-Liss, Inc.

Aquatic adaptations allow mammals to function in a liquid environment. While these adaptations confer necessary survival advantages, there are also functional limitations. Perhaps nowhere are these constraints as obvious as in the respiratory tract, for this system bridges an interface between two environments of very different media. Cetaceans (whales, dolphins, and porpoises) have developed extensive protective mechanisms to exclude water from the respiratory tract. These anatomical adaptations include: opposable nostril alae, nasal plugs covering the anterior choanae, a rostrally positioned larynx with an intranarial aditus, elongated laryngeal cartilages designed to interlock with the nasal region, and a sphincteric soft palate (palatopharyngeal sphincter) that encircles the anterior larynx (Reidenberg and Laitman, 1987, 1994, 1999). The valvular nasal barriers ensure the upper respiratory tract is protected from incursions of water into the blowholes, while the laryngeal and soft palate barriers work in concert to separate and protect the respiratory tract from water or food in the digestive tract. A breach of any of these barriers can be lethal, as asphyxiation can occur from accidental invasions of water (drowning) or food (choking). Humpback whales (Megaptera novaeangliae), however, exhibit a unique and paradoxical behavior concerning respiratory protection. They have been noted to release air from the mouth while submerged. We hypothesize that oral air release involves modifying laryngeal protective anatomy and/or positioning, potentially placing the airway at risk for incursions of water. We further hypothesize that oral air release forced to exit through the mesh of the baleen plates would result in a bubble cloud (an underwater gaseous formation composed of a large cluster of very tiny bubbles). This study examines the respiratory tract in seven specimens of Megaptera novaeangliae, focusing on the anatomy and position of the larynx to determine whether there are any unique adaptations related to underwater air release, particularly bubble cloud formation. While it is recognized that the gaseous composition within the respiratory tract is not technically “air,” this term (as well as “airflow”) will be used to describe respiratory gases as this is the more familiar and classically used descriptor.


Humpback whales exhibit a distinctive behavior compared with other mysticete (baleen) whales: they release large quantities of air while underwater. This behavior creates large collections of bubbles and appears to subserve several diverse functions, including feeding (Jurasz and Jurasz, 1979; Watkins and Schevill, 1979; Hain et al., 1982; Croll, 2002), play (Hain et al., 1982; Weinrich et al., 1992a), and aggression (Tyack and Whitehead, 1983; Baker and Herman, 1984; Silber, 1986). These authors have categorized observed bubble releases into two general groups based upon the gross pattern of the bubble clusters: (1) bubble walls (nets, streams, columns), and (2) bubble clouds. Bubble walls contain relatively large bubbles usually greater than 0.5 m in diameter, while bubble clouds consist of aggregations of smaller bubbles less than 1 cm in diameter. Bubble walls generally involve air release from the paired blowholes, and appear as two narrow trails of bubbles. As the whale swims forward, the parallel trailing streams merge into one large wall of rising bubbles. Bubble walls can also be released from the mouth of a swimming whale, forming two parallel trails emerging from the lateral (caudal) edges of the gape. Bubble walls released from a stationary whale are called bubble columns due to the vertical orientation of the wide bubble stream. Intermittent, sequential releases from a moving whale can also produce an interrupted bubble wall comprised of a series of bubble columns. Bubble walls released as a whale swims in a vertical spiral may be used to create a cylindrical wall composed of rising bubbles. These cylindrical walls are called bubble nets, because they are used to encircle and concentrate prey for feeding (Sharpe and Dill, 1997). The bubble net may also buffer the prey from sounds emitted by the humpback whales swimming just outside the net, thus enticing prey to concentrate within the confines of the bubble net (Leighton et al., 2004, 2005). Bubble walls are also used as visual displays for conspecifics in breeding competition (Tyack and Whitehead, 1983; Baker and Herman, 1984). While these behaviors certainly create a dramatic visual display, the sound accompanying such a release may also be an important component of the signal.

Unlike the other bubble formations described above, a bubble cloud involves tiny bubbles each less than 1 cm in diameter, which collectively comprise a large mass that can cover up to 10 m in diameter (Hain et al., 1982; Weinrich et al., 1992b). A bubble cloud may be used as a dramatic visual, and perhaps auditory, display. The exact mechanism of bubble cloud generation, however, remains unknown.

In order to investigate the production of bubble clouds, it is first necessary to understand the nature of the air containing passages in humpback whales and all possible portals of exit underwater. Gas/air is contained only in the digestive or respiratory tracts, and may thus emerge from only three openings: the anus (Fig. 1a), the paired blowholes (Figs. 1b, 2), and the mouth (Figs. 1c, 2). The nature of these openings will determine to a large extent the patterns of bubbles that may be released. Each of the paired slot-like blowhole openings can be widened or narrowed by movements of the lateral wall against the medial wall. Likewise, the anus can be dilated or constricted to vary its diameter. Regulating the size of these openings should limit the maximum diameter of each emerging bubble, as well as the total volume of air comprising a bubble cluster that can be released over a brief period. As the gape of a humpback whale is relatively large (its head is approximately one quarter to one third of its length), air released orally could result in extremely large diameter bubbles and a large total volume of expelled air per unit time. The blowholes and the anus are relatively small openings, and thus the maximum amount of air (gas) released from these openings per unit time should be dramatically less compared with releases from the large mouth.

Figure 1.

a: Underwater video footage of air bubbles released from the anus of a sperm whale. Photo courtesy of Nan Hauser. b: Underwater video footage of air bubbles released from the blowholes of a humpback whale. Photo courtesy of Dan Salden and Harrison Stubbs, Hawaii Whale Research Foundation. c: Underwater video footage of air bubbles released from the mouth of a humpback whale. Photo courtesy of Tom Kieckhefer, Pacific Cetacean Group.

Figure 2.

a: Aerial view of bubble streaming humpback whale (still frame taken from video footage). Note double bubble stream indicating air release from the lateral aspects of the head (i.e., the lateral gape of the mouth) as the whale swims forward. b: Same photograph as a, but the outline of the whale's head and flippers is traced in red to help visualize the sites of air bubble emergence from the lateral aspects of whale's head. c: Aerial view of same whale approximately 2 seconds later. Note appearance of air release in between the two oral bubble streams. This midline burst of air was released from the blowholes while simultaneously still producing the two lateral oral bubble streams. d: Same photograph as c, but the outline of the whale's head and flippers is traced in red to help visualize the sites of bubble emergence from both the lateral aspects and the midline of the whale's head. e: Underwater still from video footage of a swimming humpback whale releasing a small stream of bubbles from the left side of the mouth while simultaneously releasing a nasal bubble stream. Photos courtesy of Dan Salden and Harrison Stubbs, Hawaii Whale Research Foundation.

Bubble cloud production differs from other air releases in having the smallest diameter bubbles collected into the largest bubble formation. This requires that the orifice of exit must be capable of dividing the emerging air into many small bubbles to be released simultaneously. This may be done by fractionation of the air into many small units by a mesh or screen-like structure. Bubble cloud composition appears consistent despite echosounder data that shows a variety of fish school sizes, densities, or relative locations (Weinrich et al., 1997), thus indicating that bubble releases are not fractionated into clouds simply by passing through schools of small fish. The mechanism of bubble fractionation is most likely to be a structure located at the air expulsion orifice. Although small bubbles could be produced by expulsion through extremely rapid serial openings and closings of a small orifice, such as the blowholes, such releases would likely form only a stream of bubbles rather than a cloud. Oral release has the advantage over nasal or anal release in that the total volume of air expelled per unit time is much greater due to the larger size of the oral aperture.

How does air reach the whale's oral cavity? Although whales do not breathe through the mouth, air could reach the mouth through either “gulping” air at the surface to be released at depth, or releasing air into the oral cavity from the lungs. Both of these mechanisms are supported by behavioral observations. Silber (1986) noted that a humpback whale could lift its head above the water, engulf air to distend its throat pleats in an aggressive display, and then expel the air from the mouth while underwater. Baker and Herman (1984) suggest that air may be directed from the lungs to the mouth, based upon their observations of nasal inspiration at the surface followed by underwater oral air expulsion. It is the latter mechanism that will be the focus of this study, as it may involve sacrificing the usual mechanisms of mammalian airway protection.


Our anatomical observations are based upon dissections of seven humpback whales (Megaptera novaeangliae) found dead along the northeastern United States: two adult females, two juvenile (subadult) females, two male calves, and one female calf (see Table 1). These specimens were dissected and compared with odontocete larynges described in previous studies (Reidenberg and Laitman, 1988, 1994; Laitman and Reidenberg, 1999). This is a very large sample, considering humpback whale specimens are difficult to obtain as they are protected and listed as an endangered species in the United States. Furthermore, they are very difficult to dissect due to their large size, making access to laryngeal specimens limited in certain field necropsy situations. All cetacean specimens were donated through the National Marine Mammal Stranding Network, and were collected post mortem from carcasses found along the coast of the United States (northwest Atlantic Ocean). The epicranial nasal tissues were excised from two male humpback whale calves and each narial passageway was dissected open. Whole larynges were excised from each carcass with the hyoid apparatus and as much of the surrounding soft palate and pharyngeal wall as possible. Laryngeal position relative to soft palate, hyoid, tongue, and skull base was noted during the excision. Larynges were then opened along the dorsal aspect to view the internal structure of the larynx. The cricoid cartilage was incised in the dorsal midline, and the cut edges were reflected laterally to view the internal arrangement of the larynx. The shape of the epiglottis was noted. Some larynges were then cut into left and right halves along the midsagittal plane. Larynges were immersed in formalin fixative or stored frozen for future examination.

Table 1. Humpback whale specimens dissected
Specimen IDField necropsy dateStateSexAgeLength (rostrum to fluke notch) (cm)
NY-766-91June 1, 1991NYFemaleAdult1386
NY-814-91September 5, 1991NYFemaleJuvenile962
NY-881-92February 24, 1992NYFemaleCalf794
MH-96-479-MnApril 19, 1996MAMaleCalf845
MH-98-629-MnMarch 1998MAMaleCalfNot available
NY-2411-00March 24, 2000NYFemaleCalf548
NY-2818-2002April 19, 2002NYFemaleAdult1535


Nasal Region

The epicranial portion of the humpback whale nasal passageways is separated into two nasal openings (blowholes). There are no air-containing nasal diverticulae or sinuses associated with the airway. The lateral surface can be opposed against the medial surface of each blowhole nostril to close the opening. During opposition, the nostril becomes slit-like in shape. When opened, the passageway is triangular. Just deep to the opposable nostrils, the lumen of each nasal passage is occluded by a large, round soft tissue mass called the “nasal plug.” The muscle within each plug attaches to the dorsal surface of the skull rostral to the blowhole openings. The free edge of this soft tissue plug is covered by black epithelium, and faces caudally to occlude the anterior choanae (dorsal narial opening in the skull). No screen-like structure capable of dividing an emerging air stream was found projecting into the nasal passageways.

Laryngeal Cartilages

The humpback larynx skeleton is composed of the unpaired epiglottic, cricoid, and thyroid cartilages, and the paired arytenoid and corniculate cartilages (Reidenberg and Laitman, 2007, this issue). There are no cuneiform cartilages. The caudal edge of the cricoid cartilage is fused dorsally to the first four to eight tracheal cartilage rings. The cricoid cartilage is incomplete ventrally, and this gap is occupied by an unpaired diverticulum in the midline called the laryngeal sac (see below). The thyroid cartilage is often broken into several pieces that are partially attached to each other in the ventral midline. Laterally and dorsally, the thyroid cartilage extends into a long, narrow process that is homologous to the inferior (caudal) cornua. Each cornua forms an arch that is dorsally convex and attaches by means of a joint on the lateral aspect of the cricoid cartilage. It is not clear from our dissections whether this joint is fibrous or synovial.

Each corniculate cartilage is fused to the dorsal aspect of an arytenoid cartilage, and thus is affected by movements of the arytenoids (see below). The corniculates curl caudally forming a C-shape that abuts the nasopharyngeal wall (Fig. 3). A flap of soft tissue extends rostrally from the convex edge of the corniculate. The medial aspects of both corniculates, including the flap extensions, are flattened. These flat surfaces appear able to oppose one another to narrow or seal shut the caudal-most aspect of the laryngeal lumen, or perhaps repeatedly contact and release to produce vibrations (pulsed sounds).

Figure 3.

a: Left lateral view of an odontocete larynx (white-sided dolphin, Lagenorhynchus acutus). The laryngeal aditus, between the epiglottic (E) and corniculate (C) cartilages, has been stretched to its maximum opening or approximately 2 cm, indicated by the ruler. b: Left lateral aspect of larynx of humpback whale calf (Megaptera novaeangliae). Note the epiglottis (E) overlapping above the soft palate. The large laryngeal inlet is located between the epiglottis and the pair of fused corniculate/arytenoid cartilages (C). Centimeter scale is placed on tissue of the nasopharynx. White square = 1 cm. c: Dorsal–caudal view of an odontocete larynx (harbor porpoise, Phocoena phocoena) with the palatopharyngeal sphincter removed. Note the curved tip of the epligottis (E), narrow laryngeal aditus, and the very tall corniculate cartilages (C) covering the dorsal–caudal aspect of the larynx. d: Dorsal–caudal view of the same humpback whale larynx as in b. Note the very large gap (laryngeal aditus) between the epiglottis (E), which points rostrally, and the paired corniculates (C), which point caudally. The palatopharyngeal fold has been cut in the dorsal midline, but can be seen in the lateral aspects of the photograph as a ridge on either side pointing toward the midline of the larynx (centimeter scale is on left cut palatopharyngeal fold). White square = 1 cm. e: Dorsal–caudal view of the same harbor porpoise larynx as in c. The rostral aspect of the larynx, composed of the epiglottis (E) and paired corniculate cartilages (C), is circumferentially encircled by the palatopharyngeal sphincter of the soft palate (P). f: Frontal view (from the nasopharynx) of the same humpback whale larynx as in b and d. Note the epiglottis (E) overlapping the dorsal surface of the soft palate (P), which is being pulled rostrally by a metal retractor. Further retraction of the soft palate enabled the epiglottis to move ventrally and be inserted into the oral cavity.

The arytenoid cartilage supports the corniculate cartilage dorsally, exhibits a raised muscular process laterally, and extends ventrocaudally into the laryngeal lumen as the vocal process. The terminal edge of the vocal process is tapered to a point that curves toward the midline, where it is connected to the other arytenoid's vocal process by a ligament. These vocal process extensions support a U-shaped ridge bordering the opening into a ventral diverticulum called the laryngeal sac (Reidenberg and Laitman, 2007, this issue). The muscular process of the arytenoid cartilage supports the attachments of several muscles, including the posterior and lateral cricoarytenoid muscles and the interarytenoid muscle. There is a large synovial joint between the arytenoid cartilage and the leading edge of the cricoid cartilage. The arytenoids can slide medially and laterally along this joint, and can pivot rostrally and caudally. Internal and external rotation, however, appears to be very limited. Movements at the cricoarytenoid joints affect the position of the arytenoids' vocal processes: medial and lateral sliding results in adduction/abduction, and rostral/caudal pivoting causes dorsal/ ventral displacement at the tips. As the vocal processes surround the opening of the laryngeal sac, their movements regulate air flow into and out of the sac.

The epiglottic cartilage is narrow and trough-shaped (Fig. 3). The overall shape resembles the bow of a boat. It has a deeply concave dorsal surface and high lateral margins, with a cross section resembling a “V.” The epiglottis is angled rostrally toward the posterior choanae, and elongated to a pointed tip that is placed dorsally over the soft palate. The epiglottic cartilage is not fused to the thyroid cartilage, and there is no synovial joint at its base. However, there is movement at the base of the epiglottic cartilage where connective tissue attaches it to the thyroid cartilage. The dorsal aspect of the humpback epiglottis is unopposed due to a great height difference between the long epiglottis and the short corniculate cartilages.

Soft Palate and Laryngeal Position

The soft palate extends laterally and caudally around the larynx where the left and right sides unite to form a single, horizontally positioned palatopharyngeal fold. This fold projects rostrally from the caudal pharyngeal wall at the junction between the nasopharynx and the oro-laryngopharynx. The musculature of the soft palate and the palatopharyngeal fold together form a muscular ring called the palatopharyngeal sphincter. The sphincter encircles the laryngeal opening at the base of the epiglottic and corniculate cartilages. The dorsal aspect of the sphincter is overlapped rostrally by the epiglottis and caudally by the tips of the corniculates (Fig. 3). The concave aspect of the corniculate faces the projecting palatopharyngeal fold. Together, these cartilages and sphincter form an interlocking barrier that circumferentially protects the laryngeal inlet from incursions of the digestive tract, effectively separating the upper respiratory tract (laryngeal aditus and nasopharynx) from the upper digestive tract (oro-laryngopharynx and esophagus).

In our postmortem dissections, the humpback's epiglottis is always found in situ positioned overlying the soft palate and inserted into the nasopharynx (i.e., an intranarial position; Fig. 3). In fresh specimens, the cartilage is rigid and the epiglottis cannot be manually folded caudally over the laryngeal inlet. (The epiglottis is deceptively flexible with progressive tissue autolysis in decomposing specimens.) However, the epiglottis in fresh specimens can be dislodged from the nasal region and inserted into the oral cavity by manually retracting the whole larynx caudally or elevating the soft palate above the epiglottic tip. In this intraoral position, the epiglottis and soft palate together form a tubular channel, with the epiglottis resembling a trough and the soft palate serving as a roof. Manual pressure applied against the ventral aspect of the epiglottis forces it against the soft palate, effectively sealing the oral opening of the epiglottis–soft palate tube.

The humpback larynx has a very large aditus between the epiglottic and corniculate cartilages that leads into the laryngeal lumen (Fig. 3). The lumen extends ventrally into the laryngeal sac. The laryngeal sac has many small, randomly placed folds that indicate the ability for expansion and contraction of this space. There are no prominent tissue folds raised off either the lumen of the larynx or the laryngeal sac. Several parallel tissue folds were noted on the ventral aspect of the tracheal lumen surrounding the opening into the laryngeal sac. These folds appear to direct air flow from the lungs into the laryngeal sac.


Comparisons With Odontocetes

The upper respiratory tract of humpback whales differs from odontocetes in many respects, including the configuration of the nasal region. There are two blowhole openings (as opposed to one blowhole opening in odontocetes), and the nasal region does not contain any diverticulae. The humpback nasal region is thus relatively simple in construction compared with the complex anatomy of the odontocete nasal passageways (e.g., Mead, 1975; Cranford, 1996). Air released from the humpback's blowholes is, therefore, unlikely to produce the tiny bubbles characteristic of a bubble cloud. The humpback oral cavity, however, is well suited for dividing the emerging air into multiple, smaller units as its opening is modified by the presence of baleen plates suspended from the upper jaw (Fig. 4). When the mouth is partially opened, baleen form a sieve-like barrier through which exiting air would have to pass, in turn breaking it into many very small units that could be released simultaneously to produce a mist-like bubble cloud.

Figure 4.

a: Ventral–oblique view of a humpback calf head being lifted by a crane during necropsy at the Caven Point Army Corps of Engineers Station in NJ. Note the separate plates of baleen that comprise the baleen racks (particularly visible where the hoisting rope has broken or parted the baleen plates). The baleen has the appearance of a brush on the lingual (medial) aspect, and the appearance of a comb on the labial (lateral) aspect. b: Close-up of a rack of baleen plates photographed from the lingual aspect, showing the parallel alignment of the plates and the gaps through which bubbles may exit the baleen.

The humpback larynx is largely composed of the same muscles and cartilages as the larynx of a typical terrestrial mammal. Differences include a lack of cuneiform cartilages, and the caudal edge of the cricoid cartilage was usually fused dorsally to the first four to eight tracheal cartilage rings (Reidenberg and Laitman, 2007, this issue). It also shares some unique characteristics with odontocetes, most notably the ventrally incomplete cricoid cartilage and ventrocaudally elongated processes of the arytenoid cartilages. There were several gross structural differences found which distinguish the humpback larynx from that of odontocetes. Whereas the odontocete larynx has a narrow aditus (Fig. 3a), the humpback larynx has a very large aditus between the epiglottic and corniculate cartilages that leads into the laryngeal lumen (Fig. 3b). The ventral luminal surface of the odontocete larynx usually has a trabeculated appearance with numerous small ventral diverticulae and a midline laryngeal fold, whereas the humpback larynx has only one large ventral diverticulum (laryngeal sac) and no prominent midline laryngeal fold.

The humpback's larynx is not closed dorsally to form a complete tube, as in odontocetes (Fig. 3a,c,e). Rather, the dorsal aspect of the humpback epiglottis is unopposed due to a great height difference between the long epiglottis and the short corniculate cartilages (Fig. 3b,d,f). In terrestrial mammals, the epiglottis is flexible and can come into opposition with the laryngeal inlet as the larynx is raised cranially toward the tongue during a swallow. The odontocete epiglottis is rigid and fused to the thyroid cartilage, and thus has little or no independent movement. The humpback's epiglottis is also rigid, but it is not fused to the thyroid cartilage as in odontocetes. Although it cannot be easily folded caudally over the laryngeal inlet, it does have some movement at the base and, thus, could be dislodged from the nasal region and inserted into the oral cavity while manually elevating the soft palate above the epiglottic tip (Fig. 3f). The tip of the humpback's epiglottis is pointed (Fig. 3b,d,f), while the tip of the odontocete epiglottis is rounded, curled rostrally, and enlarged laterally for interlocking with the palatopharyngeal sphincter (Fig. 3a,c,e). The muscles of the soft palate and palatopharyngeal fold of the humpback collectively form the palatopharyngeal sphincter, but it is not as well developed as the muscular palatopharyngeal sphincter of odontocetes (Fig. 3e).

Intranarial vs. Intraoral Larynx

Channeling air to the mouth from the lungs entails redirection of air from the nasolaryngeal portion of the respiratory tract to the oral cavity. This overlap between the respiratory and digestive tracts is counterintuitive, particularly in mammals adapted to protect the airway. Many mammals (excluding adult humans) have effectively separated the respiratory and digestive tracts (Laitman and Reidenberg, 1988). In the typical mammalian configuration, the epiglottis of the larynx interlocks behind the soft palate (Negus, 1949; Harrison, 1995). This arrangement helps channel air from the nose directly to the trachea and lungs, while food usually passes lateral to the larynx in the food channels (piriform sinuses) en route to the esophagus (Laitman and Reidenberg, 1993). For example, herbivores do not appear to unlock the larynx during swallowing, but rather, pass their liquid and semisolid foods around it in the above manner (Laitman and Reidenberg, 1998). Although baleen whales (mysticetes) are not herbivorous, the consistency of their swallowed food is probably somewhat similar, as it is composed of relatively small organisms (e.g., zooplankton, small fish). Thus, mysticetes probably ingest their food in much the same way as terrestrial mammals, maintaining an intranarial larynx during deglutition. The capability to unlock the larynx from the nasal region for a brief period of time, however, can occur among some terrestrial mammals. For example, a momentary unlocking of the larynx can enlarge the food passageway when swallowing a large piece of meat, or enable communication with the oral cavity as occurs during panting (Laitman and Reidenberg, 1993). While laryngeal unlocking does occur briefly in some terrestrial mammals, it has not been documented in aquatic mammals.

In order to channel air into the mouth for production of a bubble cloud, the ability to unlock the larynx in humpback whales would have to differ from the pattern in other cetaceans. Toothed whales (odontocetes) have a larynx that is held in a permanently intranarial position, which isolates and protects the respiratory tract from the digestive tract (Reidenberg and Laitman, 1987). The elongated, rigid, and arrowhead shape of the rostral laryngeal cartilages facilitates interlocking with the encircling palatopharyngeal sphincter (Reidenberg and Laitman, 1994). Humpback whales also have an elongated larynx and a palatopharyngeal sphincter, although the musculature is not as prominent as that of odontocetes. The humpback's corniculate cartilages curve caudally, and appear to facilitate interlocking with the palatopharyngeal fold. The epiglottis, however, is not as rigid nor is it swollen into an arrowhead shape as it is in odontocetes. As these laryngeal features differ from odontocetes, it appears that the humpback whale's larynx may unlock from behind the soft palate. This unlocking of the epiglottis places the laryngeal aditus intraorally, creating a laryngeal–oral connection that facilitates oral release of air to generate bubble clouds.

An oral mechanism for air release is a dangerous behavior for a marine mammal because is increases risk of drowning. Most terrestrial mammals (except adult humans) have a larynx that is positioned intranarially (Laitman and Reidenberg, 1998). This protects it from accidental incursions of food or liquid from the digestive tract, and even enables simultaneous breathing or vocalizing while swallowing. Aquatic mammals have a particular need to protect the larynx from such incursions due to the added danger of feeding underwater. Odontocetes have exaggerated the terrestrial pattern and have a larynx that is permanently intranarial (Reidenberg and Laitman, 1994). Modifications of the odontocete's rostral laryngeal cartilages and soft palate create an interlock that ensures that the laryngeal aditus remains connected to the nasal region (for breathing and/or vocalizing) and is sealed off from the digestive tract even while pursuing prey, open mouthed, underwater, upside down (MacLeod et al, 2007, this issue; Werth, 2007, this issue).

Although the humpback larynx is normally positioned intranarially (Fig. 5a), we suggest that the epiglottis may be unlocked to be transiently intraoral. Elevation of the soft palate could expose the laryngeal inlet to the oral cavity, as the epiglottis is directed rostrally rather than dorsally. In this arrangement, the soft palate would lie over the deeply concave, dorsal surface of the epiglottis. A tubular channel is thus formed between these structures, allowing air stored in either the lungs or the ventrally located laryngeal sac to be channeled directly into the oral cavity (Fig. 5b). It should be noted that the intraoral position may be maintained only very briefly for the purpose of expelling air, and the larynx then immediately reinserted above the soft palate to re-establish nasolaryngeal patency. Any prolonged exposure of the laryngeal inlet into the oral cavity, particularly after air expulsion, could be risky as water or remnant food could inadvertently enter the unprotected larynx.

Figure 5.

a: Schematic representation of the head and neck of a humpback whale viewed in the midsagittal plane during normal respiration. The respiratory tract is shown in red, and the digestive tract is shown in blue. The larynx is depicted in its usual intranarial position with the epiglottis (E) overlapping above the soft palate (SP). This nasolaryngeal connection directs air between the blowholes and the lungs. There is a large laryngeal inlet between the epiglottis and the fused corniculate/arytenoid cartilages (C). The corniculates do not oppose the epiglottis, but rather curl caudally over the cricoid cartilage (Cr). A laryngeal sac (S) is located ventrally between the thyroid cartilage (T) and the trachea (Tr). During swallowing, food would be directed around the larynx in lateral food channels (not shown), which connect the oral cavity with the esophagus (Es). The fused corniculate/arytenoid cartilages are not shown in white, as they are paired and parasagittally positioned. Only cartilages cut in the midsagittal plane are shown in white. b: Hypothesized anatomical position of the larynx during bubble cloud production. The respiratory tract is shown in red, and the digestive tract is shown in blue. Yellow arrows indicate the path of air from the trachea, through the larynx, into the mouth, and through the baleen. The baleen of the right side, although not a midsagittal structure, is included in this figure and diagrammatically represented lateral to the tongue. The larynx is depicted with the epiglottis in a transient intraoral position. The nasal cavity is closed dorsally at the blowhole, and ventrally by the elevated soft palate. In this situation, air may pass either rostrally from the trachea directly over the epiglottis, or ventrally from the trachea into the laryngeal sac and then rostrally over the epiglottis. The tubular channel formed between the epiglottis and the oral surface of the soft palate would then conduct air into the oral cavity. As shown here, the mouth is closed thereby trapping air in the expandable oral cavity. Once a sufficient amount of air is accumulated, the whale could return the epiglottis back to the intranarial position above the soft palate. The whale could then open the mouth slightly to expose only the baleen and elevate the tongue. This action would force the air to exit through the mesh created by the baleen, thereby breaking the air mass into many small bubbles to produce a bubble cloud.

Air entering the oral cavity through the intraoral epiglottic connection may be stored there due to the nature of the expanding throat pleats. If the mouth is opened only slightly, so that the gape is completely blocked by the exposed baleen plates, air would be forced to exit laterally through the mat of bristly fibers that comprise the lingual surface of the stacked baleen plates (Fig. 4). The exiting air would be disrupted by passage through these bristles, breaking it into many small bubbles. The above-described mechanisms of laryngeal transport of air and its subsequent oral emission may, thus, produce the large volume of very small bubbles in one location underwater that comprise a bubble cloud.

Field observations and video analysis of humpback whales surface feeding in waters off New England confirm the transport of air from the lungs into the mouth as a probable source for the production of bubble clouds (F. Sharpe, personal communication). The “gulping” of air at the surface is unlikely, because feeding whales appear to dive with a streamlined, as opposed to distended, throat before bubble cloud release. Furthermore, the throat decreases in size throughout the surfacing and filtering is completed with the jaws nearly closed (water exits through the lateral and caudal aspects of the mouth). Air accidentally captured during prey engulfment has been observed forcefully exiting the posterior corners of the mouth after prey capture. This behavior, seen repeatedly in several recognized animals that have been studied over a long period of time, always occurred before diving, and without any previous “air gulping” (M. Weinrich, personal communication).

Underwater still photographs (T. Kieckhefer, personal communication) and underwater video footage (D. Salden, personal communication) also confirm oral air release by humpback whales in Hawaiian waters in a nonfeeding context (Figs. 6, 7). These whales were not noted to have gulped air at the surface before these underwater air releases, thus supporting a laryngeal–oral air transport mechanism.

Figure 6.

a: Underwater video footage of a humpback whale (on right) bubbling while in the presence of another humpback whale (on left). The whale on the right has previously released a small puff of air from the blowholes (seen rising above whale's head), and is now releasing air from the mouth. Note 3 puffs of bubbles on each side of the head, forming a U-shape that matches the opening of the mouth dorsally. b: This frame is taken 2 seconds later in the same video sequence, and shows how the orally released bubbles form a cloud large enough to hide all of the whale's head and most of the whale's body. Only the whale's white pectoral flippers are visible on either side of the cloud. Both frames are courtesy of Dan Salden and Harrison Stubbs, Hawaii Whale Research Foundation.

Figure 7.

a: Close-up underwater video footage of a dorsal view of a humpback whale's head. The whale is releasing air bubbles from near the tip of its mouth. b: This frame is approximately 2 seconds later than the frame shown in a, and shows the whale releasing more air from the mouth while swimming through the orally released bubble cloud. c: This frame is approximately 2 seconds after the frame shown in b, and shows the whale continuing to swim through the bubble cloud. d: This frame is approximately 2 seconds after the frame shown in c, and shows the whale lifting its tail into the bubble cloud. e: This frame is approximately 2 seconds after the frame shown in d, and shows the expansion of the bubble cloud after the tail has dispersed it. The whale is right behind the bubble cloud, but is no longer discernable to the videographer (except for its right pectoral flipper barely visible at the lower right corner of the image). All frames of this sequence are courtesy of Dan Salden and Harrison Stubbs, Hawaii Whale Research Foundation.

Humpback whales have developed a unique epiglottic shape that may facilitate laryngeal–oral air transport behavior without risking excessive compromise. The trough-shape may enable the humpback whale to channel air through the lumen of the epiglottis as it is positioned against the undersurface of the soft palate (Fig. 5b). This arrangement creates a tunnel connecting the laryngeal aditus with the oral cavity. The pointed and relatively narrow tip of the epiglottis fits tightly against the soft palate, but if it is slightly depressed ventrally, it can allow air to easily escape over its surface in an oral “exhale.” It can be quickly re-sealed against the soft palate to prevent water incursion once air release is accomplished. Pressure from water or prey entering the oral cavity would force the epiglottis up against the soft palate, thus closing the aditus. The critical period for the whale would thus be limited to the interval during which the larynx is re-inserted back over the hard palate to its normal position for respiration or deglutition.

Purpose of Bubble Clouds

The retention of the risky behavior of bubble cloud generation indicates its survival importance. Although bubble clouds have been observed in conjunction with prey capture and feeding, they have also been observed during nonfeeding behaviors. Bubble clouds might serve an important function in conspecific social signaling including aggression, mate attraction, or play. Humpbacks have been noted to release large quantities of bubbles from the mouth, particularly while in the presence of another humpback whale. Oral air releases form a bubble cloud large enough to hide the whale's body (Fig. 6). This cloud can be enlarged by swimming through it and dispersing it with the tail (Fig. 7). It is not clear whether the visual signal is the most important characteristic, or whether bubble clouds have associated sonic characteristics that serve an important role in communication. The most important use of bubbles is probably in self preservation. Bubbles can be used to confound predators or aggressive conspecifics. Just as the fleeing squid uses ink to distract a predator, a bubble cloud can serve as an underwater “smoke screen” to hide the humpback whale or confuse the predator. Bubbles may function as both a visual screen as well as an acoustic screen. For example, schools of herring have been observed to generate extensive gas bubble releases from the anus (flatulence) while pursued by orcas (Nottestad, 1998). The herring gas bubbles create a barrier that, due to the density difference from water, disrupts orca echolocation (Miller et al., 2006). Humpback whales are similarly vulnerable to attacks by orcas in the open ocean, where there are no obstacles to camouflage their large bodies. Thus, a humpback may defend against predation by hiding behind its bubble clouds, screening it from both visual and auditory detection.

Other species of mysticetes have also been documented to produce bubble emissions. It is not clear, however, whether these air emissions can be orally released. Further research on the laryngeal anatomy other mysticete species is necessary to establish whether the ability for laryngeal–oral air transport is unique to humpback whales, or can be generalized to other mysticetes. Future studies linking the anatomy of the mysticete respiratory tract with growing behavioral data should help clarify the mechanism(s) of mysticete air releases and the various situations in which they are produced and used.


We thank the Northeast Regional Stranding Network for permission to dissect cetacean specimens, particularly the Riverhead Aquarium and Research Foundation for providing access to the five humpback whales specimens from New York and the New England Aquarium for access to the two specimens from Massachusetts; Tom Kieckhefer of Pacific Cetacean Group, Dan Salden and Harrison Stubbs of Hawaii Whale Research Foundation for allowing us to use slides and still frames from video footage of humpback whales emitting bubbles underwater; Fred Sharpe of Alaska Whale Foundation for allowing us to view video footage of bubbling whales; Charles Mayo of the Provincetown Center for Coastal Studies in Cape Cod, MA; and Mason Weinrich of the Cetacean Research Unit in Glocester, ME, for thought provoking conversations regarding bubble cloud behavior; The skilled crew of the Army Corps of Engineers, Point Caven, NJ, for their invaluable assistance in recovering and facilitating dissections of several of the whales; The United States Coast Guard for their assistance in recovering specimens at sea; Drs. Samuel Márquez, Douglas Broadfield, Armand Balboni, and Michael Lipan, Mr. Calvin Keys, Mr. Duane Mims, and many student members of the Mount Sinai School of Medicine for technical assistance; and Drs. Ann Pabst, the late John Heyning, Mason Weinrich, and Lori Mario for helpful comments on a previous draft of this manuscript. We also thank Dr. Robert Gisiner for his ongoing encouragement and the Office of Naval Research for support of our research.