Content uploaded by Mason Weinrich
Author content
All content in this area was uploaded by Mason Weinrich on Mar 27, 2019
Content may be subject to copyright.
Underwater components of humpback whale
bubble-net feeding behaviour
David Wiley1,6),Colin Ware2),Alessandro Bocconcelli3),Danielle
Cholewiak1),Ari Friedlaender4),Michael Thompson1)
& Mason Weinrich5)
(1Stellwagen Bank National Marine Sanctuary, NOAA National Ocean Service, 175
Edward Foster Road, Scituate, MA 02066, USA; 2Centre for Coastal and Ocean Mapping,
University of New Hampshire, 24 Colovos Road, Durham, NH 03824, USA; 3Woods Hole
Oceanographic Institution, 266 Woods Hole Road, Woods Hole, MA 02543, USA; 4Duke
University Marine Laboratory, 135 Pivers Island Road, Beaufort, NC 28516, USA; 5Whale
Centre of New England, 24 Harbour Loop Road, Gloucester, MA 01931, USA)
(Accepted: 24 March 2011)
Summary
Humpback whales (Megaptera novaeangliae) employ a unique and complex foraging be-
haviour — bubble-netting — that involves expelling air underwater to form a vertical
cylinder-ring of bubbles around prey. We used digital suction cup tags (DTAGs) that con-
currently measure pitch, roll, heading, depth and sound (96 kHz sampling rate), to provide
the first depiction of the underwater behaviours in which humpback whales engage during
bubble-net feeding. Body mechanics and swim paths were analysed using custom visual-
ization software that animates the underwater track of the whale and quantifies tag sensor
values. Bubble production was identified aurally and through spectrographic analysis of tag
audio records. We identified two classes of behaviour (upward-spiral; 6 animals, 118 events
and double-loop; 3 animals, 182 events) that whales used to create bubble nets. Specifically,
we show the actual swim path of the whales (e.g., number of revolutions, turning rate, depth
interval of spiral), when and where in the process bubbles were expelled and the pattern of
bubble expulsion used by the animals. Relative to other baleanopterids, bubble-netting hump-
backs demonstrate increased manoeuvrability probably aided by a unique hydrodynamicly
enhanced body form. We identified an approximately 20 m depth or depth interval limit to
the use of bubble nets and suggest that this limit is due to the physics of bubble dispersal
to which humpback whales have behaviourally adapted. All animals were feeding with at
6) Corresponding author’s e-mail address: David.Wiley@noaa.gov
©Koninklijke Brill NV, Leiden, 2011 Behaviour 148, 575-602
DOI:10.1163/000579511X570893 Also available online - www.brill.nl/beh
576 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
least one untagged animal and we use our data to speculate that reciprocity or by-product
mutualism best explain coordinated feeding behaviour in humpbacks.
Keywords: humpback whale, feeding, bubble net, kinematic, spiral-loop, double-loop.
1. Introduction
Humpback whales (Megaptera novaeangliae) are large baleen whales (8.5 m
at 0.5 years to 14.3 m at 17 years of age; Stevick, 1999) that feed on a variety
of relatively small prey species, each of which aggregate in dense concentra-
tions. Common prey include krill (euphausiid spp.), and schooling fish such
as herring (Clupea spp.), capelin (Mallotus villosus) and sand lance (Am-
modytes spp.) (e.g., Matthews, 1937; Tomilin, 1967; Overholtz & Nicolas,
1979; Ichii & Kato, 1991). In the Gulf of Maine, humpback whales typically
target small fish, primarily herring (Clupea harengus) and offshore American
sand lance (Ammodytes dubius; Hain et al., 1982; Kenney et al., 1985; Payne
et al., 1986, 1990). American sand lance, the preferred prey for whales in the
southern Gulf of Maine and the only prey identified during our study, live
in relatively shallow water, school in large aggregations and are relatively
weak swimmers (Overholtz & Nichols, 1979; Hain et al., 1982; Weinrich
et al., 1997). In particular, their tendency to school near the surface during
daylight hours, often in ‘chimney-like’ vertical columns, enables efficient
feeding by predatory humpback whales (Hain et al., 1982; Friedlaender et
al., 2009; Hazen et al., 2009).
Like all balaenopterids, humpback whales feed by engulfing a large vol-
ume of water containing prey and separating food and water using sieve-
like baleen plates (Slijper, 1962; Mackintosh, 1965). However, humpback
whales have unique behavioural and morphological adaptations that distin-
guish them from other baleen whales.
Behaviourally, humpback whales capture prey by engaging in complex
feeding manoeuvres that are often accompanied by the apparently directed
use of air bubbles. The ability of bubble barriers to corral or herd fish has
been reported by a number of authors (e.g., Smith, 1961; Blaxter & Batty,
1985; Sharpe & Dill, 1997). Bubble use by humpback whales has been
observed in many of their feeding habitats and is reported to occur in a
variety of configurations. These bubble-feeding behaviours appear to vary
in nature among both individuals and regions; for example, bubble clouds
Humpback whale bubble-net feeding behaviour 577
(the production of a single or multiple bursts of seltzer-sized bubbles) are
commonly observed from humpback whales in the Gulf of Maine, but never
in Alaskan waters.
Of the various bubble configurations reported, the most complex appears
to be the bubble net (Jurasz & Jurasz, 1978; Watkins & Schevill, 1979; Hain
et al., 1982). Existing descriptions of this unique and complex behaviour
are currently derived only from surface observations, predominately Jurasz
& Jurasz (1979) and Hain et al. (1982). As described by Jurasz & Jurasz
(1979), bubble nets are rings of distinctive bubbles that appear at the surface
in a closed circle or figure ‘9’. In the Gulf of Maine, bubble nets have been
further described by Hain et al. (1982) as a ring formed by a series of discrete
bubble columns, blown at 3–5 m depth, by a whale that is rotated inward
with the flippers in a vertical plane. The nets were described as incorporating
1.25–2 revolutions with smaller bubbles grading into larger bubbles as the net
was closed. In both descriptions, whales fed in the centre of the completed
bubble net at or near the surface.
Morphologically, as compared to other baleen whales, humpbacks whales
are adapted for manoeuvrability. The species is unique in the greater length
and higher aspect ratio of its flippers and the existence of a series of protuber-
ances (tubercles) along the leading edge of the flippers (Fish & Battle, 1995;
Fish, 2002; Miklosovic et al., 2004). These features have been hypothesized
to aid manoeuvrability by increasing lift and decreasing drag, allowing an-
imals to accomplish greater turning at lower speeds (Fish & Battle, 1995;
Fish, 2002; Miklosovic et al., 2004). In addition, humpbacks have large
flukes relative to their body size providing greater thrust for quick manoeu-
vres (Woodward et al., 2006). While other balaenopterid whales typically
feed by swimming rapidly forward in a relatively straight line and lunging in
a narrow plane to engulf prey (Ridgeway & Harrison, 1985; Goldbogen et al.,
2006), the morphologic adaptations favouring manoeuvrability are thought
to allow humpbacks to undertake the fine-scale movements needed to create
bubble nets (Fish & Battle, 1995; Fish, 2002; Miklosovic et al., 2004; Wood-
ward et al., 2006). However, the movements thought to be used by humpback
whales to create bubble nets are based only on surface observations and no
information exists regarding the actual kinematics of the sub-surface ma-
noeuvres used during feeding events. Therefore, the degree to which and/or
how humpback whales would need to manoeuvre when creating bubble nets
is unknown.
578 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
In this paper, we use data from short-term (<24 h) deployments of syn-
chronous motion, acoustic recording tags (Johnson & Tyack, 2003) to pro-
vide the first quantitative descriptions of the subsurface behaviours used by
humpback whales engaged in bubble-net feeding. We provide both detailed
kinematic descriptions and quantitative analyses of the behaviour patterns
accompanying bubble-net feeding, including the onset, pattern and duration
of bubble emission. Additionally, we examine the hypothesis that swim pat-
terns and bubble production occur in a way that act to aggregate prey rather
than simply surrounding it. We, therefore, provide novel information regard-
ing behaviour unique to humpback whales; the creation of bubble nets to
capture prey. In doing so, we provide data demonstrating the complex suite
of behaviours needed to create nets, which would be facilitated by the in-
creased manoeuvrability thought to result from the specialized morpholo-
gical adaptations (flipper size and shape) unique to the species. In addition,
we use our data to examine a possible vertical limit to bubble net creation and
use (approximately 20 m) and speculate on the apparent coordinated nature
of bubble net feeding in humpback whales.
2. Material and methods
2.1. Field methods
2.1.1. Study area and population
The study took place in the southern Gulf of Maine, primarily within the
Stellwagen Bank National Marine Sanctuary (42◦1843N, 70◦1853W) in
July 2006 and July 2007. In addition, one animal in our study was tagged in
the Great South Channel (41◦279N, 69◦1812W) in 2004; its tag was re-
covered near the Northeast Peak of Georges Bank (41◦4230N, 68◦223W)
several days after detachment. Because the records of surface feeding came
late in the tag record, we assume that the feeding behaviour took place near
the spot of recovery.
2.1.2. Tagging
We used digital acoustic recording, synchronous motion tags (DTAGs; John-
son & Tyack, 2003) to record the orientation, movements and acoustic be-
haviour of feeding whales. DTAGs are small, non-invasive, archival tags at-
tached via suction cups that contain a pressure sensor (depth) and 3-axis
Humpback whale bubble-net feeding behaviour 579
magnetometer and accelerometers to determine heading, pitch and roll at a
sampling rate of 50 Hz. The tags used two embedded hydrophones (Fs-64
and 96 kHz) to record acoustic information concurrent with the other sen-
sors. Tags had a memory-limited data collection duration of approximately
20 h. The tags also contained a VHF transmitter allowing the tracking of
whales independent of visual observation and to aid in the retrieval of tags.
Once recovered, data were downloaded for analysis.
Tags were placed on humpback whales that were approachable, but not
pre-selected. Attachment used a 7 m rigid-hulled inflatable boat (RHIB) with
a 15 m, bow-mounted, cantilevered pole or a 4 m RHIB and 7 m hand held
pole. Tagged whales were individually identified using naturally distinctive
markings on their dorsal fin and tail flukes (Katona & Whitehead, 1981;
Blackmer et al., 2000). This allowed us to know if an animal was tagged
more than once, a situation that occurred twice during the study. Tags were
placed as high on the back of the animal as possible to facilitate tracking
the VHF signal emitted by the tags. Tags were set to release at a pre-defined
time, which was determined by a series of factors including memory ca-
pacity, weather conditions and programmed release for other simultaneously
deployed tags.
Responses of the whales to the tagging event varied from none to indica-
tions of short-term disturbance such as diving, trumpet blowing, or acceler-
ating (Weinrich et al., 1991). The first 10 min of behavioural data (2–4 dives)
from all tags were discarded because of this potential response period.
2.1.3. Focal animal follows
Tagged animals were followed at a distance of 100–400 m by the RHIBs or,
if necessary, at greater distances by larger support vessels (either the 16 m
R/V Auk or the 70 m R/V Nancy Foster). During daylight hours (approxi-
mately 0600–2000 h) and when weather permitted, surface behaviours were
selected from an ethogram of >80 humpback whale behaviours and the times
(to the second) at which they occurred were recorded (e.g., Weinrich, 1991;
Weinrich et al., 1992). These data were synchronized using time and GPS po-
sitions to directly associate tag-derived data with their surface counterparts.
Bubble-net feeding events were initially identified by observing a circle of
bubbles on the surface followed by the whale’s emerging though the ring
with its mouth gaped. Because the swim track signature of these bubble-net
behaviours in TrackPlot (see below) was so distinct from other portions of
580 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
the whale swim tracks, we were able to identify additional bubble-net feed-
ing events during analysis. This allowed us to use data from periods when
inclement weather or other environmental conditions precluded focal fol-
lows.
2.2. Data analysis
2.2.1. Visualization of tag data
The DTAG data provided a continuous record of the tagged animal’s ‘atti-
tude’ (azimuth, pitch and roll) and depth. We converted this to a pseudo-track
(see Johnson & Tyack, 2003) by assuming that the animal was travelling at
a constant speed of one meter per second. Although this speed assumption
is likely not precisely accurate, the pseudo track provides a valuable tool
for understanding different kinematic behaviour patterns (i.e., Friedlaender
et al., 2009; Hazen et al., 2009), and has been shown to accurately depict
the movement patterns (but not geographic location) of the tagged animals
(Schmidt et al., 2010).
Data were visualized and analyzed using TrackPlot, a custom software
application designed for the project (Ware et al., 2006). TrackPlot uses a
ribbon to represent the 3D swim path (track) of the whale, with the ribbon’s
centre being the pseudotrack centre (Figure 1). In scale, the ribbon is four
meters wide and is twisted around the along-track direction to show roll
behaviour. A pattern of chevrons on the top surface of the ribbon provides
travel direction, segments the ribbon into 1-s intervals and gives an additional
orientation cue. Loops, turns and twists in the ribbon correspond to the same
orientations in the DTAG and, therefore, the whale. TrackPlot also features
rapid track traversal and an active zooming, rotational interface allowing us
to rapidly visualize different sections of a pseudo-track at different scales and
from different angles (Figure 1). This assured that we could obtain the best
visualization of the data and investigate a behavioural event in the context of
its pre and post behaviours.
TrackPlot also supports the generation of basic dive statistics, such as
duration, maximum depth, rate of descent, rate of ascent and rate of turn
(change in heading). Dive duration of a feeding behaviour was calculated
from the tag record by manually selecting the point of tag submergence for
a terminal dive and calculating the interval to the tag’s subsequent return to
the surface (depth reading of <1 m). Maximum depth for a dive/behaviour
Humpback whale bubble-net feeding behaviour 581
(a) (b)
Figure 1. DTAG derived data were visualized and analysed using TrackPlot, a custom
software application. TrackPlot creates a ribbon that represents the temporally accurate, 3D
swim path of the tagged whale. Chevrons on the ribbon’s top surface reveal travel direction
and segment the ribbon into 1-s time intervals. Loops, turns and rolls in the ribbon correspond
to the same orientations in the DTAG and, therefore, the whale. Panel (a) is a visualization
of several hours of data, while (b) shows a zoomed-in and rotated portion of the initial
dive in panel (a). TrackPlot features rapid track traversal and an active zooming, rotational
interface that allows the users to move through the track and view behaviour from different
angles. This figure is published in colour in the online edition, which can be accessed via
http://www.brill.nl/beh
was determined by identifying the deepest record during a dive interval.
Additionally, selecting any point in the track provided a read-out of the depth
and time for that location. This feature was used to correlate timed surface
behaviour observations with subsurface kinematic patterns. This feature was
also used for determining the depth and time at which we first heard bubbles
released by the whale and the duration of bubble emission (see below).
To investigate variation in turn rate (change in heading) and body orienta-
tion (roll angle) during the creation of a bubble net, we manually identified
the beginning and end of a bubble-producing swim track (dive). TrackPlot
would then integrate the turn rate or roll angle over that interval (e.g., total
degrees turned/time interval). This information was saved to a file along with
the turn angle, start depth, end depth, start time and end time. Time series
averages (and their standard deviations) were calculated by first determin-
ing mean segment duration; the segment’s durations were then normalized
to match this mean. The mean turn rate and roll angle and their standard de-
582 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
viation was then computed for each time step in the normalized segment to
determine if it changed throughout the course of the dive sequence. We hy-
pothesized that if animals exhibited increasing rates of turn the spirals were
becoming tighter and prey could be compacted in the upper portions of the
bubble net. If rate of turn remained constant or decreased through the bub-
ble net sequence we hypothesized that the animal did not spiral inwards and
bubbles were likely used to corral rather than concentrate prey. Similarly,
we hypothesized that if animals demonstrated increasing roll angle through
the spiral sequence this would be indicative of a constricting spiral and the
concentration of prey in the upper portions of the net.
2.2.2. Determination of bubble production
We identified bubble production by a whale by listening to the tagged ani-
mal’s audio record for clearly defined bubbles and visualizing a spectrogram
of the tag’s acoustic record using Raven, a bio-acoustic analysis program
(Charif et al., 2006). Bubbling behaviour produced a signature sound, which
was further connected to whale bubbling behaviour by matching sounds re-
corded from humpback whales seen to simultaneously emit bubbles in clear
Hawaiian waters (although this was not a feeding context). Hawaiian acous-
tic data were provided to us by Alison Stimpert, Biology Department, Uni-
versity of Hawaii, Honolulu HI, USA. We identified a continuous emission
of bubbles as a ‘stream’ and pulsed expulsions as ‘bursts’. No attempt was
made to determine whether the placement of tags on the whales affected the
recording of bubbling sounds. However, the placement of tags had relatively
low variability, typically located along the dorsal surface near or anterior to
the dorsal fin.
Since audio and sensor files are time-synchronized in the DTAG record,
we were able to locate the depth at which bubble production was first re-
corded in each behavioural event. In some cases, the number of dives anal-
ysed for bubble production was less than the total number of dives recorded
because the sound of passing ships interfered with bubble sounds. For one
animal (192a_06) we used a random numbers generator to sub-sample 20
events for analysis from the 109 available in the record.
2.2.3. Relationship of bubble feeding to water depth
To understand the relationship between a whale’s bubble-producing dive be-
haviour and the bottom depth over which it was feeding, we used ArcGIS
Humpback whale bubble-net feeding behaviour 583
9.2 Geographic Information System software (ESRI, Redlands, CA, USA).
We combined whale position data, using surface position fixes that combined
Leica laser-range finder binoculars with the RHIB’s GPS position during fo-
cal follows, with multibeam bathymetry (Valentine et al., 2001) to determine
the approximate water depth over which animals were feeding. We used a
linear regression model to test for a relationship between ocean depth and
the maximum depth of the foraging dive where a humpback whale produced
a bubble net.
3. Results
We recorded 300 tag-derived bubble-net feeding events from 9 individual
humpback whales; of the 300 events, 180 were complemented by surface ob-
servations. We found two distinct kinematic techniques associated with ani-
mals observed feeding via bubble nets; ‘upward-spirals’ and ‘double-loops’
(Figure 2). Because the swim track signature of these bubble net behaviours
(a) (b)
Figure 2. TrackPlot visualizations of the two main kinematic behaviours used by hump-
back whales to create bubble-nets as an aid to capturing prey; (a) an upward-spiral net and
(b) a double-loop net. Upward-spirals were produced as a single, continuous step, while
the double-loop technique was produced using 3 separate steps: (1) the deep corral-loop,
(2) a lobtail at the surface and (3) the capture-loop. At the end of each behaviour the whale
appeared in the net with its mouth gaped. This figure is published in colour in the online
edition, which can be accessed via http://www.brill.nl/beh
584 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
in TrackPlot was so distinct from other portions of the whale swim tracks, we
used all tag-derived events for kinematic analysis. We also identified two ad-
ditional behaviours of note, each recorded from a single animal: a combined
spiral-net/double-loop technique (one of the 9 individuals) and a free-form
technique where the animal, while surfacing in a bubble net with its mouth
open (feeding), swam neither a swim path that could produce a bubble net
nor expelled bubbles. Our analysis does not include numeric data on this
tenth tagged animal as it neither expelled bubbles nor swam a curvilinear
path that could be described. An example of its swim path is in Figure 8.
3.1. Upward-spiral bubble-net feeding
We analysed 118 upward-spiral bubble-net feeding events from six animals
(Table 1; seven animals are contained in Table 1: animals 198d_07 and
199a_07 are the same animal tagged on different days). The kinematic be-
haviour consisted of a clockwise upward spiral (Figure 3), mean ±SD =
2.1±0.3 revolutions (individualized mean range 1.5±0.3to2.3±0.4).
The combined mean duration of the dive segment associated with spiralling
was 70.6±15.2 s (individualized mean range 38.7±4.0to102.7±21.2s
(Table 1)). The combined mean rate of turn during spirals was 11.1±1.7◦/s,
(individualized mean range 7.9±0.8to14.2±1.2◦/s (Table 1)).
Body orientation (roll angle) during the spiralling event varied among
animals. Three animals tended to increase their body roll angle from the
beginning to the end of the spiral; two animals showed an initial increase
in body roll angle that then diminished through the spiral; and one animal,
tagged on two different days, exhibited increased roll angle during the initial
part of the spiral, followed by decreasing roll angle in the mid portion of
the spiral and increasing roll angle again during the terminal portions of the
spiral (Figure 4a–d). The rate of turn for four of the six animals showed an
increased turning rate in the final portion of the spiral (Figure 4e). Taken in
aggregate, there was a tendency for turning rate to increase through the spiral
duration (Figure 4f). The combination of increasing turn-rate and change in
body-roll angle tended to form a constricting spiral (Figure 5).
Upward-spiralling behaviour occurred in many parts of the water column
(Table 1). The deepest point of initiation was 41.1 m and the deepest point of
termination was 35.4 m. The shallowest point of initiation was 17.8 m and
the shallowest termination was 4.4 m. The mean depth interval (initiation to
Humpback whale bubble-net feeding behaviour 585
Table 1 . Kinematics of upward-spiral bubble-net feeding behaviour in humpback whales feeding on schools of small
fish (Ammodytes dubius). Data derived from synchronous motion, acoustic recording tags (DTAGs) attached to feeding
whales.
Animal Number Start depth of End depth of Depth interval of Spiral Number of Spiral turn
of events spiral (m) (σ)spiral(m)(σ)spiral(m)(σ) duration (m) (σ) revolutions rate (◦/s) (σ)
189b_04 4 34.9 (1.8) 28.3 (3.0) 6.6 (2.0) 38.7 (4.0) 1.37 (0.2) 7.9 (0.8)
195b_06 26 34.0 (4.7) 10.2 (1.4) 23.8 (5.2) 102.7 (21.2) 2.25 (0.4) 7.9 (0.8)
198c_07 7 23.8 (1.4) 6.7 (0.8) 17.1 (1.5) 50.5 (10.8) 1.48 (0.27) 10.7 (1.8)
198d_07* 10 20.0 (2.1) 5.3 (.83) 14.9 (2.1) 49.7 (8.1) 1.88 (0.2) 13.8 (1.25)
199a_07* 50 24.4 (1.53) 7.3 (2.0) 17.1 (3.0) 76.8 (18.0) 2.25 (0.35) 10.9 (1.97)
200b_07 3 19.6 (2.0) 12.0 (1.8) 7.7 (1.6) 62.2 (6.6) 1.97 (0.31) 11.41 (1.34)
202a_07 18 32.3 (3.4) 20.5 (4.6) 11.8 (4.1) 61.4 (14.4) 1.7 (0.3) 10.2 (1.3)
*Same animal tagged on different days.
586 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
Figure 3. Sample TrackPlot visualizations of the swim path used by individual study an-
imals to create upward-spiral bubble nets to capture prey. Data derived from synchronous
motion, acoustic recording tags (DTAGs) attached to the whales. This figure is published in
colour in the online edition, which can be accessed via http://www.brill.nl/beh
termination of spiralling) was 14.2 m (individual mean range 6.6–23.8 m).
For two animals (mn189b_04 and mn202a_07) the entire spiral sequence
occurred at depth, with termination deeper than 20 m. Concurrent surface
behaviour data on mn202a_07 showed that this animal surfaced with its
mouth closed and at distances >50 m from where the bubble net reached
the surface.
We were able to acoustically identify 70 complete bubble production
events during the creation of bubble nets using the upward-spiral method
(Table 2). Bubble onset typically occurred at the deepest portion of the ani-
mal’s dive, when the animal initiated its first turn (5 animals, 61 events). Bub-
ble production consisted of a continuous, long duration (approximately 50–
60 s) stream (4 animals, 30 events) or stream-to-burst sequence (2 animals,
31 events) that was emitted throughout the spiral (Table 2, Figure 6). Two
animals, 198c_07 and 198d_07/199a_07, departed from this pattern. Ani-
mal 198c_07 produced short duration (approximately 4 second) bursts only
at the top of the spiral. Animal 198d_07 produced streams-to-bursts only in
the bottom section of its dive during one tag application, while producing
streams throughout the spiral during its second tag application (199a_07).
Humpback whale bubble-net feeding behaviour 587
(a) (b)
(c) (d)
(e) (f)
Figure 4. Animals changed their body roll angle and turn rate (change in heading) as they
proceeded through the creation of an upward-spiral bubble net. Change in body roll angle
varied by animal (a). Three animals increased their roll angle in the latter portions of the
spiral (b), two animals showed initial increase roll angle in the early portions of the spiral
(c) and one animal, tagged on two different days, exhibited a bimodal pattern of increased
roll angle during the initial portions of the spiral, decreased roll angle in the mid-section
and increased roll angle again in the upper portions near the terminus of the spiral (d). Rate
of turn increased through the spiral duration for 4 of the 6 animals (e). In aggregate, there
was a tendency for animals to increase their rate of turn as they proceeded through spiral
formation (f). This figure is published in colour in the online edition, which can be accessed
via http://www.brill.nl/beh
588 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
(a) (b)
(c) (d)
Figure 5. Sample TrackPlot visualizations of turn rate and body roll contributing to spiral
formation. Constricting spirals formed by: (a) increasing turn rate and increased body roll
through the spiral (mn198c_07), (b) increasing turn rate and bimodal body roll through
the spiral (mn198d_07) and (c) increasing turn rate and decreased body roll through the
spiral (mn202b_07). A non-constricting spiral (d) formed by relatively constant turn rate and
bimodal body roll (mn199a_07). Body roll angle of >40◦are shown in yellow. Data derived
from synchronous motion, acoustic recording tags (DTAGs) attached to feeding humpback
whales. This figure is published in colour in the online edition, which can be accessed via
http://www.brill.nl/beh
3.2. Double-loop bubble-net feeding
In 2006, we recorded a total of 182 double-loop bubble nets produced by
three whales (animals 189b_06 and 192_06 are the same individual tagged
on different days). Double-loop bubble-net feeding behaviour consisted of
two individual dive loops separated by a surfacing that included one or more
‘lobtails’ (using the flukes to forcefully strike the water’s surface). The se-
quence’s initial loop was termed the ‘corral-loop’, while the second dive,
which terminated in the lunging behaviour and the consumption of prey, was
termed the ‘capture-loop’ (Figure 2). One animal was tagged on two differ-
ent days (189b_06 and 192a_06) and exhibited similar behaviour on both
days (Table 3, Figure 7).
Humpback whale bubble-net feeding behaviour 589
Table 2 . Bubble production by humpback whales during upward-spiral technique used for bubble-net feeding directed
at schools of small fish (Ammodytes dubius).
Animal Number of Bubble style Bubble location Depth bubble Depth bubble Bubble
events analysed onset (m) (σ) termination (m) (σ) duration (s) (σ)
189b_04 4 Stream Throughout spiral 34.9 (1.8) 28.3 (3.0) 63.6 (6.72)
195b_06 24 Stream to burst Throughout spiral 25.4 (4.93) 7.7 (2.29) 61.4 (19.43)
198c_07 9 Burst Top of spiral 10.5 (1.3) 7.6 (1.6) 4.4 (1.7)
198d_07* 7 Stream to burst Bottom of spiral 20.3 (1.41) 13.2 (1.29) 22.9 (1.98)
199a_07* 6 Stream Throughout spiral 24.0 (0.72) 12.6 (1.56) 52.9 (3.94)
200b_07 3 Stream Throughout spiral 18.9 (2.85) 12.5 (1.55) 55.7 (18.55)
202a_07 17 Stream Throughout spiral 32.9 (5.87) 19.8 (10.45) 54 (29.06)
Data derived from synchronous motion, acoustic recording tags (DTAGs) attached to feeding whales. Bubble production was identified aurally
and through spectrographic analysis using the acoustic software package Raven.
*Same animal tagged on different days.
590 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
(a)
(b) (c)
Figure 6. Raven generated spectrogram showing an example of stream-to-burst bubble
production used to create a bubble net (a), TrackPlot visualization of the spiral swim path
used to create the bubble net with portion of the track during which bubbles were expelled
coloured orange, portion of the track in which body roll angle exceeded 40◦coloured yellow
and chevron indicating the direction of travel (b) and an aerial photograph of a humpback
showing surface manifestation of stream-to-burst bubble net production (c). Data for (a) and
(b) derived from synchronous motion, acoustic recording tags (DTAGs) attached to feeding
humpback whale number198d_07. This figure is published in colour in the online edition,
which can be accessed via http://www.brill.nl/beh
The mean maximum depth for all corral-loops was 21.6±3.0 m (indi-
vidualized mean range 20.6±2.8to22.2±3.1 m). The mean dive duration
for the corral-loop was 62.4±SD 14.0 seconds (individualized mean range
54.9±9.8to66.0±13.5 s). The mean rate of turn for all corral-loops was
5.8±1.3◦/s (individualized mean range 5.5±0.3to6.5±1.17◦/s (Table 3)).
Humpback whale bubble-net feeding behaviour 591
Table 3 . Kinematics of double-loop bubble-net feeding behaviour in humpback whales directed at schools of small
fish (Ammodytes dubius).
Animal Number of Corral loop Corral loop Corral loop turn Capture-loop Capture-loop Capture-loop
events depth (m) (σ) duration (s) (σ) rate (deg/s) (σ)depth(m)(σ) duration (s) (σ)turnrate(
◦/s) (σ)
189b_06* 13 21.7 (3.5) 66.0 (13.5) 5.4 (1.1) 12.4 (1.2) 34.7 (3.5) 10.4 (1.0)
192a_06* 109 21.6 (2.9) 65.9 (16.0) 5.5 (1.3) 12.2 (1.1) 36.2 (4.0) 9.9 (1.1)
189c_06 33 22.2 (3.1) 54.9 (9.8) 6.5 (1.2) 13.6 (1.0) 34.2 (2.6) 10.5 (0.8)
196a_06 27 20.6 (2.8) 55.5 (9.8) 6.5 (1.1) 11.8 (1.1) 27.3 (2.1) 13.2 (1.0)
Data derived from synchronous motion, acoustic recording tags (DTAGs) attached to feeding humpback whales.
*Same animal tagged on different days.
592 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
Figure 7. Sample TrackPlot visualizations of double-loop swim path used by individual
study animals during the creation of a bubble net to capture prey. Animals 189b_06 and
192a_06 are the same animal that exhibited the same behaviour on different days. Data
derived from synchronous motion, acoustic recording tags (DTAGs). This figure is published
in colour in the online edition, which can be accessed via http://www.brill.nl/beh
The mean maximum depth for all capture-loops was 12.5±1.0 m (individ-
ualized mean range 11.8±1.1to13.6±1.0 m). The mean dive duration
for all capture-loops was 32.6±3.5 s (individualized mean range 27.3±2.1
to 36.2±4.0 s). The mean rate of turn for all capture-loops was 10.6◦/s
(individualized mean range 9.9±1.1to13.2±1.0◦/s; Table 3).
We chose 60 double-loop feeding events for which we had complete sound
files (e.g., sound from passing boats or ships did not interrupt the acous-
tic record) to examine parameters of bubble production during double-loop
feeding (Tables 4 and 5).
During the corral-loop, one animal, tagged on two different days (189b_06
and 192a_06), used a stream (189b_06) and burst or stream (192a_06) bubble
expulsion emitted from the bottom of the loop through most of its ascent on
all of its dives (N=10 and N=20, respectively), one animal (189c_06)
used a burst expulsion during its descent, but expelled bubbles during a
minority (7/19) of its corral-loop dives and one animal (196a_06) did not
expel bubbles during its swimming of the corral-loop (N=11).
During the capture-loop, one animal, tagged on two different days
(189b_06 and 192a_06), expelled bursts of bubbles during its descent, but
Humpback whale bubble-net feeding behaviour 593
Table 4 . Bubble production by humpback whales during the corral-loop portion of double-loop bubble-net feeding
directed at schools of small fish (Ammodytes dubius).
Animal Number of events Bubble style Number of Bubble location Depth of Depth of bubble Bubble
analysed/containing expulsions bubble onset termination duration
bubbles (N)(σ)(m)(σ)(m)(σ)(s)(σ)
189b_06* 10/10 Stream 1.4 (0.97) Bottom of loop–most of ascent 18.4 (2.80) 10.0 (3.33) 29.9 (5.52)
192a_06* 20/20 Stream (N=9)1.4 (0.84) Bottom of loop–most of ascent 18.3 (3.75) 4.4 (3.33) 29.5 (8.07)
Bursts (N=11)
189c_06 19/7 Burst 4.7 (3.15) Descent–bottom of loop 15.3 (3.72) 19.9 (2.34) 10.1 (3.97)
196a_06 11/0 NA 0 NA NA NA NA
Data derived from synchronous motion, acoustic recording tags (DTAGs) attached to feeding humpback whales. Bubble production was
identified aurally and through spectrographic analysis using the acoustic software package Raven.
*Same animal tagged on different days.
594 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
Table 5 . Bubble production by humpback whales during the capture-loop portion of double-loop bubble-net feeding
directed at schools of small fish (Ammodytes dubius).
Animal Number of events Bubble style Number of Bubble location Depth of Depth of bubble Bubble
analysed/containing expulsions bubble onset termination duration
bubbles (N)(σ)(m)(σ)(m)(σ)(s)(σ)
189b_06* 10/3 Burst 2.5 (0.89) Descent–bottom of loop 5.1 (1.21) 11.6 (0.67) 7.0 (2.03)
192a_06* 20/9 Burst 2.4 (0.92) Descent–bottom of loop 7.2 (2.19) 9.1 (2.18) 10.9 (7.21)
189c_06 19/19 Burst 4.1 (0.97) Bottom of loop 10.9 (1.81) 11.5 (1.53) 8.7 (2.64)
196a_06 11/11 Burst 2.7 (0.65) Bottom of loop 10.6 (2.18) 10.4 (1.27) 5.4 (1.43)
Data derived from synchronous motion, acoustic recording tags (DTAGs) attached to feeding humpback whales. Bubble production was
identified aurally and through spectrographic analysis using the acoustic software package Raven.
*Same animal tagged on different days.
Humpback whale bubble-net feeding behaviour 595
used bubbles in only a minority of its capture-loop dives (3/10 and 9/20).
Animals 189c_06 and 196a_06 expelled bursts at the bottom of all capture-
loops (N=19 and N=11, respectively).
Combining the corral-loops and capture-loops into the double-loop se-
quence, animals tended to show a preference for expelling bubbles in one or
the other. Animal 189b_06/192a_06 emitted bubbles during all of its corral-
loops (10/10 and 20/20), but in a minority of its capture loops (3/10 and
9/20). Animals 189c_06 and 196a_06 expelled bubbles in a minority of their
corral-loops (7/19 and 0/11, respectively), but in all of their capture-loops
(19/19 and 11/11, respectively). Animals emitted streams or bursts in the
corral-loop, but only bursts in the capture-loop.
3.3. Anomalous techniques
While most animals exhibited only a single bubble-feeding strategy, one
whale combined the two techniques. Animal 192a_06 engaged primarily in
double-loop feeding as described above, but on 11 occasions used an upward
spiral to create the corral-loop. We also recorded one animal (192b_06) that,
while surfacing in bubble nets with its mouth gaped, showed no indication
of behaviours capable of forming a bubble net. In the 10 events we recorded
from this animal, its more free-form swim track was variable, but relatively
linear (not spiralled or looped) and no bubble expulsion could be identified
from the acoustic record (Figure 8).
3.4. Dive-depth vs. bottom depth
We found no significant relationship between bottom depth and the maxi-
mum depth of a bubble-producing foraging dive (R2=0.13, F=2.84,
N=104, p<0.0001).
4. Discussion
We combined tag-derived and time-synchronized audio and kinematic data,
focal surface observations, and novel visualization software to provide the
first detailed descriptions of the underwater behaviours employed by hump-
back whales as they created bubble nets as an aid to capturing prey. We
identified two general classes of behaviour (upward-spiral and double-loop)
596 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
Figure 8. Sample TrackPlot visualization of the swim path used by animal 192b_06, which
surfaced through the centre of 10 bubble nets with its mouth gaped (feeding), but swam
neither a swim path that could create or mimic a net nor produced bubbles during the dive
preceding surfacing in the net. This figure is published in colour in the online edition, which
can be accessed via http://www.brill.nl/beh
that animals used to create bubble nets, each of which would require substan-
tial manoeuvrability and be aided by the unique hydrodynamicly enhanced
morphology of humpbacks.
Our data substantially expand upon existing descriptions, which are de-
rived only from surface observations summarized by Ingebrigsten (1929),
Jurasz & Jurasz (1979) and Hain et al. (1982). Specifically, we show the ac-
tual swim path of the animals (e.g., number of revolutions, turning rate, depth
interval of spiral), when and where in the process bubbles were expelled and
the pattern of bubble expulsion used by the animals. In the upward-spiral
technique, the onset of bubble production was generally consistent, begin-
ning at the deepest point of the dive and at or just before the initiation of a
turn that then became the start of the spiral. Continuous expulsion of bubbles
(presumably forming a bubble-stream curtain) was most common, but some
individuals also formed nets from individual bursts of bubbles (presumably
forming discrete columns) or a sequential gradation of the two techniques
(stream to bursts). Individual animals tended to be consistent in their strategy,
with most of the variation occurring among, not within, individuals. How-
ever, individuals did show variation, as demonstrated by animal 192a_06 that
used a spiral-net to create a corral-loop in 11 of its 109 double-loop events.
Humpback whale bubble-net feeding behaviour 597
While emitting bubbles during spiral-net formation, animals oriented their
bodies in a variety of ways. Some exhibited increased roll angle in the initial
stages of the spiral, others in the later stages; one whale, tagged on two sepa-
rate days, showed a bimodal trend with increased roll angle during the initial
and final portions. Some animals increased their turn rate towards the end
of the spiral (four of six animals), which supports the conjecture that spiral-
nets function to compact the whales’ prey prior to capture. However, not all
animals did so, which suggests that some animals use nets to contain, not
concentrate, prey or that some prey patch conditions are conducive to con-
tainment and others to compaction. This supports our depiction of the highly
plastic nature of humpback whale behaviour, with different animals accom-
plishing a similar task in varying ways or responding to different conditions
with altered behaviours.
While spiral-net behaviour has been partially described from surface ob-
servations in previous studies of humpback whale feeding behaviour, double-
loop behaviour used to create a bubble net has not been previously described.
The typical behaviour pattern consists of a three-step process: (1) the corral-
loop (the deeper first dive often containing an initial bubble event), (2) a brief
surfacing with 1–3 lobtails (using the flukes to forcefully strike the water’s
surface) and (3) the capture-loop (the shallower second dive where additional
bubbles can be released and when the actual feeding occurs). Since swim
loops and bubbles occur prior to the lobtails, it is unlikely that the lobtail is
used to mark a location for the net’s creation, as was suggested by Weinrich
et al. (1992).
The dive aspects of corral and capture-loop formation were consistent
across animals and relative to one another. For all animals, the deepest por-
tion of the corral-loop was approximately 21 m, the dive duration for the
loop was approximately 60 s and the turn rate to form the loop was approx-
imately 6◦/s. For the capture-loop, the deepest portion was approximately
13 m (slightly less than the body length for an adult Gulf of Maine hump-
back whale (True, 1904; Stevick et al., 1999), the dive duration was approx-
imately 33 s and the rate of turn to form the loop was approximately 11◦/s.
Hence, the capture-loop was a quick, shallow dive of approximately half the
depth, dive duration, and twice the turning rate relative to the corral-loop.
We remain unable to determine how the lobtail phase of this sequence aids
the whale in prey capture. However, since the corral-loop consists of a sin-
gle circle that would contain rather than concentrate prey, we speculate that
598 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
the lobtail action might serve to mass the fish more tightly within the net,
thereby increasing feeding efficiency of the whale(s). Clustering behaviour is
a common response of fish to predators or other frightening stimuli (Pitcher
& Parrish, 1993) and the percussive sound created when the whale’s flukes
strike the water’s surface during a lobtail could elicit such a response in sand
lance.
The consistent dive depth (approximately 13 m) and associated small
standard deviations of the capture-loop across all animals is most likely a
result of an animal diving to a distance equivalent to its body length before
turning up and into the net. For spiral-net and corral-loops, dive depth of a
bubble-producing dive was independent of water depth and most frequently
started at 20–25 m with a vertical span of <20 m. Hazen et al. (2009) used
dive data from our whales tagged in 2006 and concurrent SIMRAD EK-60
echosounder measures of prey fields to determine that maximum dive depth
of bubble-feeding whales was independent of maximum prey field depth.
Sharpe (2001) used echosounder tracking of bubble-feeding humpbacks
in Alaska to identify the same approximately 20 m limit to bubble use and
conducted tank experiments to observe the rise of simulated bubble nets to
the surface. He concluded that the differential rise speed of the different sized
bubbles comprising a net resulted in substantial gaps emerging after a rise
distance of approximately 20 m. That different sized bubbles move through
viscous mediums at different speed is a common tenant of fluid mechanics
(Hassan et al., 2008). Thus, the association of bubble releases within an
approximately 20 m depth interval observed in two separate oceans might be
related to the physics of bubble dispersion over depth, to which humpback
whales have adapted their behaviour. As such, it could be a universal aspect
of bubble-net feeding in humpbacks.
While for most animals bubble production and swim tracks matched what
might be expected to unilaterally create a net, for some animals it did not. For
instance, whale 198c_07 produced only single-burst bubble expulsions of
short (approximately 4-second) duration towards the top of the spiral. Sim-
ilar mismatches occurred during double-loop feeding; one whale (196a_06)
swam, but did not produce any bubbles during the corral-loop. It did pro-
duce a number of short (approximately 5-second) bursts at the bottom of all
capture-loops, but these would seem unlikely to form the net observed at the
surface.
Humpback whale bubble-net feeding behaviour 599
It is possible that these cases can be resolved by considering the behaviour
of associates of the tagged whale. All of the tagged animals were feeding in
groups that contained at least one associated animal and coordination among
feeding humpback whales has been noted numerous times (Whitehead, 1983;
Baker, 1985; D’Vincent et al., 1985; Weinrich, 1991; Weinrich & Kuhlberg,
1991), with cooperation (D’Vincent et al., 1985; Ramp et al., 2010) and role
specialization (Sharpe, 2001) hypothesized. Hence, bubbles produced during
only portions of a spiral might add to the bubbles produced by associates and
increase the capture success of the net. It is also possible that swimming spi-
rals without producing bubbles might synchronize movements of the group,
or that the body might be used as a herding device (Brodie, 1977).
While cooperative feeding by humpbacks has been hypothesized, the evo-
lution of cooperative strategies is most likely tooccur under conditions where
close kin relationships are maintained (Hamilton, 1964) and the social sys-
tem of humpback whales (e.g., promiscuous breeding, short mother–calf
bond, single birth offspring, wide dispersal of juveniles (Weinrich, 1991;
Clapham, 1994, 2000) is unlikely to promote such relationships and strate-
gies. In addition, any theoretical basis for humpback cooperation must also
account for the many instances in which the behaviour of the tagged whale
was capable of unilaterally creating the net, but our behavioural sequencing
data showed that other animals also fed in the net. Additionally, cases such
as animal 192b_06 that repeatedly surfaced in the centre of a net with its
mouth gaped, but neither swam a path that would produce or mimic a net nor
expelled bubbles must be included. While kin selected cooperation seems
unlikely, reciprocity or by-product mutualism might be occurring, with po-
tential cheaters acting as net robbers (Sachs et al., 2004).
Our findings demonstrate that the creation of bubble nets require hump-
back whales to perform complex body manoeuvres that are not used by
other baleanopterids, which employ a more linear feeding method (e.g.,
Goldbogen et al., 2006). Such manoeuvrability would require adaptations
favouring increased hydrodynamic performance, such as that provided by the
humpback’s unique flipper morphology (Fish, 1995; Fish & Battle, 2004).
Whether or not the evolution of the humpback flipper was caused by the ma-
noeuvrability required during complex feeding movements, their presence
has certainly contributed to the development of unique behavioural traits
(such as bubble-netting) that allow humpbacks to feed in a manner differ-
ent from other balaenopterids. This might allow humpbacks to exploit prey
600 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
patches with increased efficiency or to access highly mobile prey that would
otherwise be unavailable.
Acknowledgements
We thank the officers and crew of the NOAA research vessels Nancy Foster and Auk for
their capable assistance during field operations. We also thank the various members of our
field team over the years, including Roland Arsenault, Pat Halpin, Elliot Hazen, Tom Hurst,
Just Moller, Susan Parks, Cara Pecarcik, Allison Rosner, Kate Sardi, Jamison Smith, Alison
Stimpert, Jennifer Tackaberry, Becky Woodward and Jeremy Winn. Funding was provided
by the Stellwagen Bank National Marine Sanctuary, Office of National Marine Sanctuaries,
and the National Oceanographic Partnership Program. Whale tag data were collected under
permit Nos 775-185 (Northeast Fisheries Science Centre) and 605-1904 (Whale Centre of
New England) issued by the United States National Marine Fisheries Service. We thank
Jim Hain and Robert D. Kenney for providing the photos for the cover of this issue. The
manuscript benefitted from the comments of Phil Clapham, Bruce Alexander Schulte and
two anonymous reviewers.
References
Baker, C.S. (1985). The population structure and social organization of humpback whales
Megaptera novaeangliae in the central and eastern North Pacific. — PhD thesis, Uni-
versity of Hawaii, Honolulu, HI, 306 pp.
Blackmer, A.L., Anderson, S.K. & Weinrich, M.T. (2000). Temporal variability in features
used to photo-identify humpback whales (Megaptera novaeangliae). — Mar. Mamm.
Sci. 16: 338-354.
Blaxter, J.H. & Batty, R.S. (1985). Herring behaviour in the dark: responses to stationary and
continuously vibrating obstacles. — J. Mar. Biol. 65: 1031-1049.
Brodie, P.F. (1977). Form, function and energetics of Cetacea: a discussion. — In: Functional
anatomy of marine mammals, Vol. 3 (Harrison, R.J., ed.). Academic Press, New York,
NY, p. 45-58.
Charif, R.A., Clark, C.W. & Fristrup, K.M. (2006). Raven 1.3 user’s manual. — Cornell
Laboratory of Ornithology, Ithaca, NY.
Clapham, P.J. (1994). Maturational changes in patterns of association in male and female
humpback whales, Megaptera novaeangliae. — Can. J. Zool. 234: 265-274.
Clapham, P.J. (2000). The humpback whale: seasonal feeding and breeding in a baleen whale.
— In: Cetacean societies (Mann, L.M., Connor, R.C., Tyack, P.L. & Whitehead, H.,
eds). University of Chicago Press, Chicago, IL, p. 173-218.
D’Vincent, C.G., Nilson, R.M. & Hanna, R.H. (1985). Vocalization and coordinated feeding
behaviour of the humpback whale in southeastern Alaska. — Sci. Rep. Whales Res.
Inst. 36: 41-48.
Fish, F.E. (2002). Balancing requirements for stability and maneuverability in cetaceans. —
Integ. Comp. Biol. 42: 85-93.
Fish, F.E. & Battle, J.M. (1995). Hydrodynamic design of the humpback whale flipper. —
J. Morphol. 225: 51-60.
Friedlaender, A.S., Hazen, E.L., Nowacek, D.P., Ware, C., Weinrich, M.T., Hurst, T. & Wi-
ley, D.N. (2009). Changes in humpback whale (Megaptera novaeangliae) feeding be-
haviour in response to sand lance (Ammodytes spp.) behaviour and distribution. — Mar.
Ecol. Progr. Ser. 395: 91-100.
Humpback whale bubble-net feeding behaviour 601
Goldbogen, J.A., Calambokidis, J., Shadwick, R.E., Oleson, E.M., McDonald, M.A. & Hilbe-
brand, J.A. (2006). Kinematics of foraging dives and lunge feeding in fin whales. —
J. Exp. Biol. 209: 1231-1244.
Hain, J.H.W., Carter, G.R., Kraus, S.D., Mayo, C.A. & Winn, H.E. (1982). Feeding behavior
of the humpback whale, Megaptera novaeangliae, in the Western North Atlantic. —
Fish. Bull. 80: 259-268.
Hamilton, W.D. (1964). The genetical evolution of social behaviour. — J. Theor. Biol. 7:
1-52.
Hassan, N.M.S., Khaqn, M.M.K. & Rasul, M.G. (2008). A study of bubble trajectory and
drag co-efficient in water and non-newtonian fluids. — WSEAS Trans. Fluid Mech. 3:
261-270.
Hazen, E., Friedlaender, A., Thompson, M., Ware, C., Weinrich, M.T., Halpin, P. & Wi-
ley, D.N. (2009). Fine-scale prey aggregations and foraging ecology of humpback
whales Megaptera novaeangliae. — Mar. Ecol. Progr. Ser. 395: 75-89.
Ichii, T. & Kato, H. (1991). Food and daily food consumption of southern minke whales in
the Antarctic. — Polar Biol. 11: 479-487.
Ingebrigtsen, A. (1929). Whales caught in the North Atlantic and other seas. — Rapp. P.-V.
Reun. Int. Counc. Explor. Mer. 56: 1-26.
Johnson, M. & Tyack, P. (2003). A digital acoustic recording tag for measuring the response
of wild marine mammals to sound: marine mammals and noise. — IEEE J. Ocean. Eng.
28: 3-12.
Jurasz, C.M. & Jurasz, V.P. (1979). Feeding modes of the humpback whale (Megaptera
novaeangliae) in southeast Alaska. — Sci. Rep. Whales Res. Inst. 31: 69-83.
Katona, S.K. & Whitehead, H. (1981). Identifying humpback whales using their natural
markings. — Polar Rec. 20: 439-444.
Kenney, R.D., Hyman, M.A.M. & Winn, H.E. (1985). Calculation of standing stocks and
energetic requirements of the cetaceans of the Northeast United States outer continental
shelf. — NOAA Technical Memorandum NMFS-F/NEC-41, National Marine Fisheries
Service, Woods Hole, MA.
Mackintosh, N.A. (1965). The stocks of whales. — Fishing News, London.
Matthews, L.H. (1937). The humpback whale, Megaptera nodosa. — Discov. Rep. 17: 7-92.
Miklosovic, D.S., Murray, M.M., Howie, L.E. & Fish, F.E. (2004). Leading-edge tubercles
delay stall on humpback whale (Megaptera novaeangliae) flippers. — Phys. Fluids 16:
39-42.
Overholtz, W.J. & Nicolas, J.R. (1979). Apparent feeding by the fin whale, Balaenoptera
physalus, and humpback whale, Megaptera novaeangliae, on the American sand lance,
Ammodytes americanus, in the Northwest Atlantic. — Fish. Bull. 77: 285-287.
Payne, P.M., Nicolas, J.R., O’Brien, L. & Powers, K.D. (1986). The distribution of the hump-
back whale, Megaptera novaeangliae, on Georges Bank and in the Gulf of Maine in
relation to densities of the sand eel, Ammodytes americanus. — Fish. Bull. 84: 271-277.
Payne, P.M., Wiley, D.N., Young, S.B., Pittman, S., Clapham, P.J. & Jossi, J.W. (1990).
Recent fluctuations in the abundance of baleen whales in the southern Gulf of Maine in
relation to changes in selected prey. — Fish. Bull. 88: 687-696.
Pitcher, T.J. & Parrish, J.K. (1993). Function of shoaling behaviour in teleosts. — In: Behav-
ior of teleost fishes, 2nd edn. (Pitcher, T.J., ed.). Chapman & Hall, London, p. 363-439.
Ramp, C., Hagen, W., Palsboll, P., Berobe, M. & Sears, R. (2010). Age related multi-year
associations in female humpback whales (Megaptera novaeangliae). — Behav. Ecol.
Sociobiol. 64: 1563-1576.
602 Wiley, Ware, Bocconcelli, Cholewiak, Friedlaender, Thompson & Weinrich
Ridgeway, S.H. & Harrison, R.J. (1985). Handbook of marine mammals. Volume 3: The
sirinians and baleen whales. — Academic Press, New York, NY.
Sachs, J.L., Mueller, U.G., Wilcox, T.P. & Bull, J.J. (2004). The evolution of cooperation. —
Q. Rev. Biol. 79: 135-160.
Schmidt, V.E., Weber, T.C., Wiley, D. & Johnson, M.P. (2010). Underwater tracking of hump-
back whales (Megaptera novaeangliae) with HF pingers and acoustic recording tags. —
IEEE J. Ocean. Eng. 35: 821-836.
Sharpe, F.A. (2001). Social foraging of the Southeast Alaskan humpback whale, Megaptera
novaeangliae. — Dissertation, Simon Fraser University, Burnaby, BC, 129 pp.
Sharpe, F.A. & Dill, L.M. (1997). The behaviour of Pacific herring schools in response to
artificial whale bubbles. — Can. J. Zool. 75: 725-730.
Slijper, E.J. (1962). Whales. — Hutchinson & Co., London.
Smith, K.A. (1961). Air-curtain fishing for Maine sardines. — Fish. Rev. 23: 1-14.
Stevick, P.T. (1999). Age-length relationships in humpback whales: a comparison of strand-
ings in the western North Atlantic with commercial catches. — Mar. Mamm. Sci. 15:
725-737.
Tomilin, A.D. (1967). Mammals of the USSR and adjacent countries. — Cetacea 9: 1-717
(Transl. Isr. Prog. Sci., Jerusalem).
True, F.W. (1904). The Whalebone whales of the Western North Atlantic, compared with
those occurring in European waters, with some observations on the species of the North
Pacific. — Smithson. Contrib. Knowl. 33: 1-332.
Valentine, P.C., Middleton, T.J. & Fuller, S.J. (2001). Sun-illuminated topography, and
backscatter intensity of the Stellwagen Bank National Marine Sanctuary region off
Boston, Massachusetts. — United States Geological Survey Open-File Report 00-410,
scale 1:60 000, 1 CD-ROM.
Ware, C., Arsenault, R., Plumlee, M. & Wiley, D. (2006). Visualizing the underwater be-
haviour of humpback whales. — IEEE Comput. Graph. 26: 14-18.
Watkins, W.A. & Schevill, W.E. (1979). Aerial observation of feeding behaviour in four
baleen whales: Eubalaena glacialis,Balaenoptera borealis,Megaptera novaeangliae,
and Balaenoptera physalus. — J. Mamm. 60: 155-163.
Weinrich, M., Martin, M., Griffiths, R., Bove, J. & Schilling, M. (1997). A shift in distribution
of humpback whales, Megaptera novaeangliae, in response to prey in the southern Gulf
of Maine. — Fish. Bull. 95: 826-836.
Weinrich, M.T. (1991). Stable social associations among humpback whales (Megaptera no-
vaeangliae) in the southern Gulf of Maine. — Can. J. Zool. 69: 3012-3019.
Weinrich, M.T. & Kuhlberg, A.E. (1991). Short-term association patterns of humpback whale
(Megaptera novaeangliae) groups on their feeding grounds in the southern Gulf of
Maine. — Can. J. Zool. 69: 3005-3011.
Weinrich, M.T., Schilling, M.R. & Belt, C.R. (1992). Evidence for acquisition of a novel
feeding behaviour: lobtail feeding in humpback whales, Megaptera novaeangliae. —
Anim. Behav. 44: 1059-1072.
Whitehead, H. (1983). Structure and stability of humpback whale groups off Newfoundland.
— Can. J. Zool. 61: 1391-1397.
Woodward, B.L., Winn, J.P. & Fish, F.E. (2006). Morphological specializations of baleen
whales associated with hydrodynamic performance and ecological niche. — J. Morphol.
267: 1284-1294.
