J OURNAL OF C RUSTACEAN B IOLOGY, 32(2), 171-179, 2012
PLEOPOD ROWING IS USED TO ACHIEVE HIGH FORWARD SWIMMING SPEEDS
DURING THE ESCAPE RESPONSE OF ODONTODACTYLUS HAVANENSIS
(STOMATOPODA)
Eric Octavio Campos 1,∗ , Daril Vilhena 1 , and Roy L. Caldwell 2
1 Department
2 Department
of Biology, University of Washington, 24 Kincaid Hall, Seattle, Washington 98195, USA
of Integrative Biology, University of California, 1005 VLSB, Berkeley, California 98720-3140, USA
ABSTRACT
The escape responses of twelve individuals of the stomatopod Odontodactylus havanensis (35-64 mm body length) were recorded with
conventional and high-speed (60 and 500 images per second) video cameras. Unlike the typical pattern of escape swimming seen in most
elongate Malacostracan crustaceans in which quick backward swimming is achieved by rapid pleonal flexion (tail-flipping), O. havanensis
always swam forward during its escape response. Rowing of the pleopods provided thrust during swimming. The power phase was
metachronal and the recovery phase was approximately synchronous. The mean stroke frequency, from high-speed video, was 17 Hz.
With this swimming mode, speeds of 1.25 to 1.62 meters per second and 21 to 40 body-lengths per second were attained. The intermittent
nature of the rowing propulsive mode led to temporally unsteady kinematics marked by periodicity. Although forward swimming via
pleopod rowing is a very common form of locomotion employed by elongate crustaceans, it is typically observed only during relatively
slow, “routine” swimming, with escape being driven by tail-flipping. Odontodactylus havanensis breaks this pattern. Further study into how
this species is able to achieve such high speeds via rowing locomotion may yield new insights into our knowledge of animal locomotion
through fluids.
K EY W ORDS: behavior, biomechanics, kinematics, mantis shrimp, Stomatopoda
DOI: 10.1163/193724011X615596
I NTRODUCTION
Many elongate malacostracan crustaceans such as shrimp,
krill, crayfish, and lobsters possess a series of pleopods,
which oscillate during a variety of behaviors, but chiefly during swimming and walking (Kils, 1981; Cowles, 1994; Copp
and Hodes, 2001; Lim and DeMont, 2009). Pleopod oscillation in these groups occurs in an adlocomotory metachronal
coordination pattern in which a “wave” of pleopods progresses anteriorly during the recovery phase of the stroke cycle, followed by sequential power strokes by each appendage
pair beginning with the most posterior (Sleigh and Barlow, 1980). Members of these crustacean groups also have
the ability to flex their abdomens in order to propel themselves powerfully backward during startle escape responses,
and these flexions are often mediated by large axons (‘giant
fibers’ or GFs) (Arnott et al., 1998; Edwards et al., 1999).
Thus, elongate malacostracan crustaceans may be viewed as
having two distinct swimming modes at their disposal. First,
relatively slow forward swimming is mediated, in full or in
part, by metachronal oscillation of the pleopods during routine or non-threatened periods of time. Second, much faster
backward escape swimming is mediated by quick abdominal
flexions during escape situations.
Stomatopods, or mantis shrimp, are a diverse lineage of
carnivorous marine crustaceans comprising the only extant
member of the subclass Hoplocarida within Malacostraca
∗ Corresponding
(Schram, 1986; Ahyong and Harling, 2000). They are
noted for their greatly enlarged second maxillipeds, the
raptorial appendages, which are used during prey capture
and processing and in agonistic interactions. Two broad
functional groups are recognized – spearers and smashers –
depending on raptorial appendage morphology and method
of deployment. They are burrow and cavity dwellers in
shallow tropical and subtropical waters around the globe
(Caldwell and Dingle, 1976).
Like other elongate malacostracans, stomatopods possess
five pairs of pleopods that are used in swimming. Despite
having a benthic habit, some species are highly competent
swimmers as evidenced by observations of surface swarming
in waters sometimes exceeding 1800 meters in depth (Losse
and Merrett, 1971; McCluskey, 1977). Stomatopods also
have a broad tail fan and are capable of flexion. This is
commonly seen in spearers, which are ambush predators
that often sit with their heads looking out of the burrow
entrance. If startled, they withdraw rapidly down their
burrow with the aid of a flexion of the pleon (RLC,
personal observation). However, they do not exhibit the
same giant-fiber mediated escape swimming response seen
in other elongate malacostracans. In the spearer Squilla
mantis (Linnaeus, 1758), pleonal flexion in response to a
threat stimulus results in backward translation of less than
one body-length (Heitler et al., 2000).
author; e-mail: [email protected]
© The Crustacean Society, 2012. Published by Koninklijke Brill NV, Leiden
DOI:10.1163/193724011X615596
172
JOURNAL OF CRUSTACEAN BIOLOGY, VOL. 32, NO. 2, 2012
Species in the genus Odontodactylus are known for
being swift and agile swimmers. One of these species,
Odontodactylus havanensis (Bigelow, 1893), is the subject
of the current study. Belonging to the smasher functional
group, members of this species range in size from 20-70 mm
in body-length and are typically found down to a depth of
50 m in the Caribbean, although maximum depths below
400 m have been reported (Manning, 1967). They construct
U-shaped burrows out of rubble and actively hunt away from
their burrows during the day. When encountered in the open,
they rapidly escape back to the burrow.
In this study we present data on the swimming pattern and
performance of O. havanensis during its escape reaction.
We used a combination of conventional and high speed
video (60 and 500 images per second, respectively) to record
burst escape swimming behavior in the laboratory. This
behavior in O. havanensis is of note because translational
motion during escape swimming is achieved exclusively by
oscillation of the pleopods instead of by pleonal flexion, an
observation that deviates from locomotor patterns typically
observed in other elongate malacostracans.
M ATERIALS AND M ETHODS
Animals
Twelve O. havanensis were purchased from commercial
suppliers of marine invertebrates. Animals were collected in
the Florida Keys and were transported to a laboratory facility
at the University of California, Berkeley. They were kept in
filtered aquaria and were fed twice weekly on a diet of fresh
grass shrimp.
Documenting Animal Movements
Animals were housed individually in glass aquaria (32 ×
32 × 61 cm) containing a 3 cm thick layer of coral rubble
substrate. Because O. havanensis is a burrow or cavitydwelling animal, each filming tank was provided with either
a rock under which the animal could excavate a burrow,
or a PVC pipe into which the animal could find refuge.
Animals were allowed to acclimate to the new surroundings
for 2-3 days prior to video recording. Escape responses were
recorded by first luring the animal from its refuge with a
piece of food (grass shrimp) on the end of a plastic rod and
made to follow the food along the long axis of the aquarium
tank until the animal reached either the left or right side of
the tank. Just before the stomatopod contacted the food, the
investigator triggered the camera and struck the aquarium
glass with his hand to elicit the escape response. Most
of the time, the animal was facing the investigator as the
investigator’s hand traveled toward the tank. In these cases
we refer to the threat stimulus as rostral because the animal
is experiencing the threat stimulus “head-on.” In some cases,
the animal was facing away from the investigator as the
hand traveled toward the tank. In these cases we refer to
the threat stimulus as caudal. Since the animals began their
escape responses before the investigator’s hand made contact
with the aquarium (Fig. 1), the animals must have responded
to the visual cue of seeing a potentially threatening object,
Fig. 1. Still images from standard video camera showing filming set-up and illustrating that animals started their escape reaction in response to visual cue.
Threat stimulus was investigator’s hand moving towards and striking aquarium tank, in this case on left side from perspective of camera’s view. Animal lured
to left side of tank with piece of food on plastic rod. Two video images show animal swimming from left to right toward its burrow refuge in midst of its
escape reaction. Notice investigator’s hand also moving from right to left, and has not yet made contact with aquarium.
CAMPOS ET AL.: FAST SWIMMING VIA PLEOPOD ROWING IN A STOMATOPOD
173
Fig. 2. Line drawing of a swimming stomatopod with pleopods shaded gray. Relative location on animals’ body at which pleopod angle (β) was measured
is indicated for pleopod labeled as number 4. Pleopod angle measured with respect to portion of the pleon immediately posterior to each pleopod. The angled
lines adjacent to pleopod 4 represents angle measured for this pleopod at this video image, translated downward from point at which it was measured for
ease of viewing.
i.e., the hand, rushing toward them. As a result, we deem
the threat stimulus to be sufficiently directed to warrant the
qualifiers of “rostral” and “caudal” depending on whether
an animal was facing toward or away from the stimulus,
respectively, at the time of its application. Only those video
sequences in which the animal traveled parallel to the
camera’s plane of focus were used in the analysis. A mirror
angled at π/4 radians (45 degrees) above the filming tank
and in the camera’s field of view helped the investigator
make these determinations. The animals were filmed with a
digital video camera (Sony DCR-VX2000) at 30 frames per
second and a resolution of 720 by 480 pixels. We recorded
4-8 escape bouts per animal (62 total escape bouts). Video
frames were de-interlaced (separated into pairs of distinct
images containing either odd- or even-numbered pixel rows)
using ProAnalyst© software, providing a temporal resolution
of 1/60 s between consecutive video images.
To verify that the standard video camera was not undersampling the motion, two animals were also filmed with a
high-speed camera (X-PRI, AOS Technologies AG, Baden
Daettwil, Switzerland) at 500 images per second with a
resolution of 1280 by 1024 pixels. The high-speed camera’s
superior sampling rate and spatial resolution also made
it possible to observe the movements of the animals’
appendages during escape responses. All high-speed video
recordings were taken with a distance calibration device
inside the tank, at the same distance from the camera as the
animal.
Pleopod angle through time was measured from highspeed video using the Angle tool in ImageJ (Rasband, 19972010). The ventral portion of the abdomen directly posterior
to each pleopod, as seen from a lateral aspect, served as a
reference for the measurement of pleopod angle (Fig. 2). An
angle of 0 radians corresponds to a pleopod that has its tip
pointing posteriorly while an angle of π radians corresponds
to a pleopod that has its tip pointing anteriorly.
Speed
Position data were acquired from de-interlaced standard
video files by tracking the pixel position of each animal’s
eye (only one eye is visible in a lateral view of the
animal) through each image in each video sequence using
ProAnalyst® software. The length of the animal as it swam
in a straight body posture was used as a distance calibration.
The length of each animal in each video clip was determined
in units of pixels using the Straight Line tool in ImageJ. The
length of each animal used in this study was also measured
to the nearest mm by putting each animal onto a moist
paper towel on a tabletop and using a flexible plastic ruler
to measure the distance from the most anterior extent of the
eyes to the posterior tip of the telson. Having the length of
each animal in mm and pixels allowed us to create distance
conversion factors for each video clip that we recorded.
Speed was calculated by doing the following for each
video file. First, we calculated the change in position between each pair of consecutive video images to find displacement through time. Then the displacement data were
divided by the time elapsed between video images (1/60 s
or approximately 16.7 ms for de-interlaced video) to calculate average speed over the duration of each video image
pair. Plots of speed as a function of time exhibited an obvious peak-trough periodicity, which is not surprising given
the inherent temporal unsteadiness of thrust production in
rowing-based propulsive systems. It is preferable to sample periodic behaviors with the highest temporal resolution
that is practically available in order to make certain to capture the extrema (crests and troughs) of such signals. However, specialized high-speed cameras were not readily available to carry out the bulk of the video recording that this
study required. To make sure that we were adequately capturing the crests and troughs in speed as a function of time
at 60 Hz, we used two interpolations methods to resample
the data sets at higher effective temporal resolutions. Sixthorder b-form smoothing splines were fit to the 60 Hz speed
data with a tolerance of 0, which forced the splines through
each data point. A fast Fourier transform (FFT) interpolation
was also performed on the data, again forcing the interpolation through each data point. The two interpolation methods increased the effective sampling frequencies from 60 Hz
to 300 and 500 Hz, respectively. Both methods were implemented in MATLAB. Thus, for each escape swimming bout
that was filmed with the conventional video camera, we produced three speed-through-time data sets: one at 60 Hz (from
the frame-by-frame data of the video), one at 300 Hz (splineinterpolated data set), and one at 500 Hz (FFT-interpolated
data set).
The decision of which of the three data sets to use to
infer swimming speed was made by examining four escape swimming bouts recorded at 500 images per second.
Position-tracking for these video clips was done with DLTdataviewer2 (Hedrick, 2008). Average speed between all
consecutive image pairs was calculated as for the standard
174
JOURNAL OF CRUSTACEAN BIOLOGY, VOL. 32, NO. 2, 2012
video clips, yielding speed-through-time data sets at 500 Hz.
No interpolation methods were applied to these data sets
since their temporal resolution was naturally very high due
to the high frame rate used. Instead, we took the FFT (implemented in MATLAB) of these four data sets to find the
dominant frequencies. Next, we used a low-pass, 2nd order
Butterworth filter (implemented in MATLAB) to smooth the
data by removing high-frequency noise. The FFTs showed
two prominent peaks at approximately 5 and 20 Hz. As a result, a conservative cutoff frequency of 50 Hz was used for
the low-pass filter. Highest crest and highest trough speed
values from these filtered data sets were taken as the best estimates of the true crest and trough speeds of those animals
during these four escape swimming bouts. Finally, the four
500 Hz position-through-time data sets were down-sampled
to 62.5 Hz (every eighth data point was kept) to simulate
these behaviors as having been recorded at approximately
60 Hz, which is the sampling rate of the standard video camera that we used to collect the majority of our data. There are
eight different ways to down-sample a dataset by a factor of
8, depending on the time step at which down-sampling is
initiated, so each 500 Hz data set was down-sampled eight
times. Thus, we produced eight down-sampled speed data
sets per high-speed video clip. Each of these down-sampled
data sets had the smoothing spline and FFT interpolation
methods applied to them and the highest crest and highest
trough speed values from each were compared to the corresponding values in their parent 500 Hz data set. The standard
deviations of the peak and trough speed values among each
family of down-sampled data sets were taken as measures
of the sensitivity of each interpolation method to the phase
of the swimming cycle at which a 60 Hz video camera begins recording. Low standard deviation values indicate low
sensitivity to starting phase and are desirable.
R ESULTS
Flexion of the Pleon
Escape reactions in O. havanensis frequently began with
abdominal flexion in response to a rostral threat stimulus.
The flexion of O. havanensis seemed to represent an initial
flinch reflex, consisting of an isolated single flexion and not
a series of flexions, and was not used to swim backward
rapidly as seen in many other elongate malacostracans.
Backward translations of less than one body-length were
typical. Flexion was not a necessary part of the escape
response because it was observed only in cases where the
animal was presented with a rostral threat stimulus and it
was facing away from its burrow. In these cases, the flexion
of the pleon seemed to be seamlessly incorporated into a
180-degree reorientation that allowed it to swim forward
toward its burrow. The 180-degree reorientation was carried
out in a variety of ways involving different combinations of
yaw, pitch, and roll maneuvers (Fig. 3). In cases where it
was facing toward its burrow at the time of threat stimulus
application (in which case the threat stimulus would have
been directed toward the caudal end of the animal), the
animal always darted straight toward its burrow with no
abdominal flexion.
Fig. 3. Odontodactylus havanensis always swam forward during escape swimming, necessitating a 180-degree reorientation if it was facing away from its
burrow refuge at time of threat stimulus application. Overlaid images from standard video and line drawing representation showing one way in which O.
havanensis executed 180-degree reorientations during its escape response. Animal initially facing left (A). Threat stimulus (not pictured) coming toward it
from left. Animal responds by initially performing flexion f pleon (B) which causes it to translate slightly up and to the right, away from threat stimulus.
Animal then straightens out of the curl by bringing its head to right. This would normally leave animal facing right in an dorsal-down posture, but it
simultaneously rolls about its longitudinal axis as it straightens out of curl (C), placing it in dorsal-up posture and able to move away from threat stimulus
and toward its burrow refuge by swimming forward via pleopod rowing.
CAMPOS ET AL.: FAST SWIMMING VIA PLEOPOD ROWING IN A STOMATOPOD
175
Fig. 4. Still images from high-speed video (500 images per second) showing pleopod recovery stroke (A-C) and beginning of subsequent power stroke
(D-F). Every fourth frame shown. Synchronous forward sweep of pleopods during recovery causes individual appendages to be indistinguishable from
their neighbors, resulting in appearance of pleopod bulge that sweeps anteriorly. During collective power stroke, most posterior pair of pleopods goes first,
followed by the next most anterior pair, and so on. Resultiing rowing stroke not metachronal throughout entire stroke, being synchronous during recovery
and metachronal during power phase. Elongate crustaceans that swim by rowing of multiple serially arranged appendages typically do so with metachronal
coordination among all appendages throughout entire collective stroke cycle.
Forward Swimming with Pleopods
Following flexion, individuals rapidly swam away from
the threat stimulus and toward their burrow or PVC pipe
refuge. Escape swimming was accomplished by forward
swimming mediated by rowing of the pleopods. In this
description, pleopod pairs are numbered in increasing order
from anterior to posterior and abbreviated as PP. Thus, the
most anterior pair of pleopods will be referred to as PP1
and the most posterior pair as PP5. Since individuals were
often facing away from their refuge at the time of the
threat stimulus, a 180-degree reorientation preceded escape
swimming. The collective recovery stroke of all five pleopod
pairs appeared to happen synchronously, or almost nearly
so, with the five pairs sweeping forward at approximately
the same time. This resulted in the appearance of an
anteriorly sweeping pleopodal bulge, as individual pleopod
pairs were so clumped together during recovery that they
appeared effectively indistinguishable from their neighbors
(Fig. 4A-D). In contrast, the collective power stroke was
not synchronous. PP5 began its power stroke first, and
was followed shortly thereafter by PP4 (Fig. 4E-F), and
so on until all five pairs completed a power stroke and
were folded flush against the body with appendage tips
pointing posteriorly. By plotting pleopod angle through time
176
JOURNAL OF CRUSTACEAN BIOLOGY, VOL. 32, NO. 2, 2012
(down to 62.5 Hz) to verify that the dominant frequencies
are conserved at the lower effective sampling rate. Downsampling was performed on position data. Graphs of the
frequency domains of the data sets confirm that the major
frequencies are indeed captured at a sampling rate just
above 60 Hz (Fig. 7). A prominent frequency peak occurred
between 13.39 and 20 Hz in the data sets of each of
Fig. 5. Pleopod angle (β) through time for all five pleopod pairs during
collective power stroke, from 500 image/second video. Pleopod pairs
numbered in increasing order from anterior to posterior. Pleopods did not
go through their power strokes at same time in accordance with metachronal
coordination pattern. Angle through time data unavailable for collective
recovery stroke because pleopods were practically indistinguishable from
one another during recovery, making it impossible to measure individual
pleopod angles at this level of spatial and temporal resolution.
during the rowing stroke (Fig. 5), it is apparent that the
pleopod power strokes are not synchronous. Instead, the
collective power stroke of O. havanensis during escape
swimming more closely fits the description of a metachronal
coordination pattern.
Swimming Speed
As would be expected of a propulsive system characterized
by distinct power and recovery phases, the kinematics of
pleopod-driven swimming in O. havanensis were unsteady
through time (Fig. 6). Speed oscillated between crests
and troughs throughout the duration of a swimming bout.
To determine whether this periodicity that we observed
at 60 Hz is not the product of aliasing, and whether
recording at 60 Hz would allow us to adequately capture
the dominant frequencies that characterize speed kinematics
through time for these animals, we compared the FFTs
of speed data from four escape bouts recorded at 500 Hz
with the FFTs of down-sampled versions of those data sets
Fig. 6. Graph of swimming speed through time from one of 62 total
escape bouts captured with standard video at 60 images per second. Graph
displays average speed calculated over duration between consecutive video
images (squares with dot in middle) and values of speed interpolated from
smoothing spline. Speeds at highest crest and highest trough in each escape
bout inferred from interpolated data because 60 Hz data points did not
always coincide with crests and troughs. Using 60 Hz data points nearest to
extrema underestimates crests and overestimates troughs to a higher degree
than using spline-interpolated data.
Fig. 7. Recording movements of O. havanensis at 60 images per second
using standard video camera sufficient for capturing major frequencies
present in their translational swimming kinematics. Four escape bouts
recorded at 500 Hz (A) were down-sampled to 62.5 Hz (B), which is as
close to 60 Hz as we could come. Fast Fourier transforms (FFT) of each
data set taken (C, D) to identify major frequencies. Peaks consistently
occurred at same frequencies in both 500 Hz and down-sampled data sets.
Prominent peak occurred for each data set between 13.39 and 20 Hz;
this peak interpreted as pleopod stroke frequency for each escape bout.
Prominent peak in (C), marked by arrow, occurs at 18.52 Hz. Frequency
domain trace in (D) only goes to 31.25 Hz because outputs from Fourier
transforms are only meaningful up to one half of sampling frequency
(Nyquist frequency). Analysis for only one exemplar 500 Hz data set shown
here, but all exhibited same pattern.
CAMPOS ET AL.: FAST SWIMMING VIA PLEOPOD ROWING IN A STOMATOPOD
177
the four escape bouts. Peaks at these frequencies closely
match the average collective pleopod stroke frequencies
of these escape bouts as calculated by inspection of the
video footage, which ranged from 16.13 to 18.18 Hz. Both
measures result in a mean rowing stroke of 17 Hz when
rounded to two significant digits. Although the dominant
peaks remained at the same frequencies in the FFTs of downsampled data, their amplitude was approximately one third
lower than that of the 500 Hz data sets (Fig. 7C-D). This
is not surprising as down-sampling inherently reduces the
amplitude of periodic signals.
Given the unsteadiness of the swimming kinematics,
we decided to focus on the highest crest speed and the
highest trough speed of each escape bout as this would
give an indication of peak performance and the degree of
variation in speed that these animals experience throughout
their stroke cycle. Our comparisons of the use of FFT
interpolation and smoothing splines to augment our 60 Hz
standard video data showed that the splines gave a more
consistent estimate of speed at the crests and troughs in
the kinematic data. Standard deviations were 58 percent
as high among speed estimates from splines of families of
down-sampled 500 Hz data compared to estimates derived
from FFT interpolations of the same down-sampled data
(0.11 vs. 0.19 m/s, respectively). Both interpolation methods
outperformed the use of non-augmented 62.5 Hz data alone
to infer crest and trough swimming speeds (Fig. 8). The
splines of 62.5 Hz data yielded slightly damped speed
estimates that were on average 0.028 m/s below speed
estimates taken from filtered 500 Hz data at the crests
and 0.033 m/s above speed estimates taken from filtered
500 Hz data at the crests. Based on these results, we feel
confident that using smoothing splines to infer swimming
speeds from our 60 Hz standard video data will lead to
accurate, yet conservative, estimates of the highest crest
and highest trough speeds attained during each escape bout.
These data are summarized in Fig. 8. The average highest
crest speed exhibited by specific individuals ranged from
1.24 to 1.62 m/s and from 21.2 to 40.5 body-lengths per
second.
D ISCUSSION
The overall pattern of the escape behavior of O. havanensis is in general agreement with that described for another
stomatopod, S. mantis, by Heitler et al. (2000). Commonalities between the two species include the presence of abdominal flexion as part of an initial flinch reflex and the observation that initial abdominal flexion does not lead to rapid
backward escape swimming, as is more typically the case in
other elongate malacostracans. Flexion of the pleon in O. havanensis in response to rostral threat stimuli was followed by
or incorporated into a non-stereotyped 180-degree reorientation (Fig. 3), followed by forward pleopod-mediated swimming away from the threat stimulus. Again, these observations are in agreement with the escape behavior described for
S. mantis. A 180-degree reorientation followed by forward
swimming has also been described in a thermosbaenacean,
although in that case forward swimming seemed to be accomplished by dorsoventral undulation of the pleon similar
Fig. 8. Mean highest crest and trough speeds with standard error bars for
each individual used in this study. The number of escape bouts recorded per
individual ranged from 4-8, depending on how cooperative each was during
filming process. Swimming speeds given in meters per second (m/s) and
body-lengths per second (L/s). Note horizontal axes not continuous. Each
tick mark represents unique individual (n = 12). Three pairs of individuals
happened to be of equal lengths (at 51, 54, and 63 mm).
to the undulation of a whale’s flukes rather than by rowing
of any of the body appendages (Olesen et al., 2006).
The main difference between the escape behaviors of O.
havanensis and S. mantis is in the speed of the forward
swimming phase. While Heitler et al. (2000) did not report
forward swimming speeds in their study on S. mantis, perhaps because it was regarded as not particularly noteworthy, the high speeds that O. havanensis are able to achieve
(routinely in excess of 1.2 m/s and many times exceeding
30 body-lengths per second) are quite astounding when observed in real time. A blurry streak is usually the best that
we have managed to see with the unaided eye.
178
JOURNAL OF CRUSTACEAN BIOLOGY, VOL. 32, NO. 2, 2012
Table 1. Escape swimming performance of various animals generally regarded as speedy, and that are approximately the
same size as the stomatopods used in this study.
Species
Esox sp. (pike)
Loligo vulgaris (squid)
Acanthephyra exemia (shrimp)
Crangon crangon (shrimp)
Pandalus danae (shrimp)
Swimming mode Length (mm) Speed (L s−1 ) Reference
Fast-start
Jet-propulsion
Tail-flip
Tail-flip
Tail-flip
It is also interesting to note that the forward pleopod
rowing speeds of O. havanensis also exceeds that of other
elongate crustaceans by a substantial margin. Reported values in the literature of similarly sized elongate crustaceans
range from 1.7-8 body-lengths per second for pleopoddriven swimming speed in Sergestes similes Hansen, 1903,
Acanthephyra eximia Smith, 1884, and Euphausia superba
Dana, 1850 (Kils, 1979; Cowles, 1994; Bailey et al., 2005)
and between 4 and 9 Hz for pleopod stroke frequency
in S. similis and Meganyctiphanes norvegica (Sars, 1857)
(Cowles, 1994; Thomasson et al., 2003), whereas the individuals of O. havanensis in this study ranged in speed from
17.4-35.6 L s−1 and, from FFT analysis, displayed average
stroke frequencies of 17 Hz. The range in speeds exhibited by O. havanensis also overlaps with the reported escape
swimming speeds of other similarly sized animals that are
widely acknowledged as speedy and agile swimmers (Table 1).
The occurrence of fast forward pleopod rowing in O. havanensis instead of the rapid backward tail-flip swimming
that characterizes most other elongate malacostracan crustaceans may be understood in light of its biology. Its diurnal
activity pattern, visual acuity, and possession of forwardlydeployed weapons (the raptorial appendages) suggest that
this animal has evolved behavioral and morphological traits
that make it very competent at dealing with most situations
head-on. Also, forward swimming makes it possible for the
animal to take advantage of its visual acuity to locate its
burrow or other shelter during an escape. Backward escape
swimming via abdominal flexion would presumably make
directed flight much more difficult, if not impossible. Key
to this logic is the assumption that the refuge is a very safe
place for the animal to be. In the laboratory, the trajectory
of each individual during its escape response was not haphazard, but highly directed. Animals clearly made an effort
to swim rapidly back to their refuge in the test tank. In the
field, individuals hunt up to 2-3 m away from their burrows
and flee from predators and divers in a direct line back to
their burrow. In fact, individuals often maintain two or three
burrows approximately one meter apart. When evicted from
one burrow, they flee in a direct line to another (RLC, personal observation).
Compared to other elongate crustaceans, stomatopods
have relatively broad pleopods. We are not aware of whether
low aspect-ratio (length divided by width) paddles lend
themselves to more effective “sprint” swimming performance, but such a question is theoretically and empirically
addressable and may yield important insights.
65
65
57
57
70
21.0
19.7
20.1
40.5
38.0-44.3
Webb, 1986
Packard, 1969
Bailey et al.
Arnott et al., 1998
Daniel and Meyhöfer, 1989
In addition, stomatopods, in contrast to true shrimp,
generally have pleons that are more massive than the
cephalothorax, and this arrangement may render backward
tail-flip mediated escape swimming prohibitively ineffective
in stomatopods since the paddle would be more massive than
the “payload.” It may be interesting to investigate what effect
this configuration may have on directional stability during
multiple tail-flips.
This study is the first to document and report escape
swimming ability mediated by rowing of multiple serially
arranged appendages which approaches the performance of
other established fast swimmers in the literature. Swimming mediated by rowing of multiple serially arranged appendages is common in nature, but is typically associated
with low swimming speeds. Among animals that swim in
this fashion, our data indicate that O. havanensis is the
fastest to date. Future studies into the spatial and temporal dynamics at play in this system may contribute to our
growing body of knowledge concerning the importance of
unsteady mechanisms of force generation during animal locomotion through fluids.
ACKNOWLEDGEMENTS
Financial support during data collection was provided by the Biology
Scholars Program and the McNair Scholars Program at the University
of California, Berkeley (UCB). This material was also supported by a
Bank of America Endowed Fellowship from the University of Washington
(UW) Graduate School, Graduate Opportunities & Minority Achievement
Program to E. O. Campos and is based upon work supported by the National
Science Foundation Graduate Research Fellowship under Grant No. DGE0718124 to E. O. Campos. Some equipment and much technical expertise
were provided by Tom Libby and the Center for Integrative Biomechanics
in Education and Research (CiBER) at UCB. The assistance of Eunice
Jong in collecting a portion of the data at UCB is greatly appreciated. The
lab groups of R. L. Caldwell and Sheila Patek, as well as the Berkeley
Biomechanics Group, provided useful advice and guidance during the early
stages of the work. Adam Summers of the UW Friday Harbor Labs provided
valuable guidance on early versions of the figures. Jon Dyer and Tom
Daniel of UW provided valuable guidance on data analysis methods. The
manuscript benefited greatly from comments by two anonymous reviewers.
R EFERENCES
Ahyong, S. T., and C. Harling. 2000. The phylogeny of the stomatopod
Crustacea. Australian Journal of Zoology 48: 607-642.
Arnott, S. A., D. M. Neil, and A. D. Ansell. 1998. Tail-flip mechanism
and size-dependent kinematics of escape swimming in the brown shrimp
Crangon crangon. Journal of Experimental Biology 201: 1771-1784.
Bailey, D. M., P. M. Bagley, A. J. Jamieson, A. Cromarty, M. A. Collins,
A. Tselepidis, and I. G. Priede. 2005. Life in a warm deep sea: routine
activity and burst swimming performance of the shrimp Acanthephyra
eximia in the abyssal Mediterranean. Marine Biology 146: 1199-1206.
Bigelow, R. P. 1893. Preliminary notes on the Stomatopoda of the Albatross
collections and on other specimens in the National Museum. Johns
Hopkins University Circulars 106: 100-102.
CAMPOS ET AL.: FAST SWIMMING VIA PLEOPOD ROWING IN A STOMATOPOD
Caldwell, R. L., and H. Dingle. 1976. Stomatopods. Scientific American
234: 80-89.
Copp, N. H., and S. Hodes. 2001. Asynchronous swimmeret beating
during defense turns in the crayfish, Procambarus clarkii. Journal
of Comparative Physiology A: Neuroethology, Sensory, Neural, and
Behavioral Physiology 187: 737-745.
Cowles, D. L. 1994. Swimming dynamics of the mesopelagic vertically
migrating Penaeid shrimp Sergestes similis: modes and speeds of
swimming. Journal of Crustacean Biology 14: 247-257.
Dana, J. D. 1850. Synopsis generum Crustaceorum ordinis “Schizopoda”.
American Journal of Science 9: 129-133.
Daniel, T. L., and E. Meyhöfer. 1989. Size limits in escape locomotion of
Carridean shrimp. Journal of Experimental Biology 143: 245-265.
Edwards, D. H., W. J. Heitler, and F. B. Krasne. 1999. Fifty years of a
command neuron: the neurobiology of escape behavior in the crayfish.
Trends in Neuroscience 22: 153-161.
Hansen, H. J. 1903. On the crustaceans of the genera Petalidium and
Sergestes from the ‘Challenger,’ with an account of luminous organs in
Sergestes challengeri, n. sp. Proceedings of the Zoological Society of
London 1903: 52-79, pls. 11 and 12.
Hedrick, T. L. 2008. Software techniques for two- and three-dimensional
kinematic measurements of biological and biomimetic systems. Bioinspiration and Biomimetics 3: 1-6.
Heitler, W. J., K. Fraser, and E. A. Ferrero. 2000. Escape behavior in the
stomatopod crustacean Squilla mantis, and the evolution of the caridoid
escape reaction. Journal of Experimental Biology 203: 183-192.
Kils, U. 1979. Swimming speed and escape capacity of Antarctic krill,
Euphasia superba. Meeresforschung 27: 264-266.
. 1982. The swimming behaviour, swimming performance and
energy balance of Antarctic krill, Euphausia superba. Biomass Scientific
Series 3: 1-121.
Lim, J. L., and M. E. DeMont. 2009. Kinematics, hydrodynamics and force
production of pleopods suggest jet-assisted walking in the American
lobster (Homarus americanus). Journal of Experimental Biology 212:
2731-2745.
Linnaeus, C. 1758. Systema Naturae per Regna Tria Naturae, Secundum
Classes, Ordines, Genera, Species, cum Characteribus, Differentiis, Synonymis, Locis (edit. 10). Vol. 1. Laurentii Salvii, Holmiae [Stockholm].
179
Losse, G. F., and N. R. Merrett. 1971. The occurrence of Oratosquilla
investigatoris (Crustacea: Stomatopoda) in the pelagic zone of the Gulf
of Aden and the equatorial western Indian Ocean. Marine Biology 10:
244-253.
Manning, R. B. 1967. Review of the genus Odontodactylus (Crustacea:
Stomatopoda). Proceedings of the United States National Museum 123:
1-35.
McCluskey, W. J. Jr. 1977. Surface swarming of Squilla empusa Say
(Stomatopoda) in Narragansett Bay, Rhode Island, U.S.A. Crustaceana
33: 102-103.
Olesen, J., S. T. Parnas, and J. F. Petersen. 2006. Tail flip and escape
response of Tethysbaena argentarii (Malacostraca: Thermosbaenacea).
Journal of Crustacean Biology 26: 429-432.
Packard, A. 1969. Jet propulsion and the giant fibre response of Loligo.
Nature 221: 875-877.
Rasband, W. S. 1997-2011. ImageJ. U.S. National Institutes of Health,
Bethesda, MD.
Sars, M. 1857. Om 3 nye norske Krebsdyr. Forhandlinger verd de
Skandinaviske Naturforskeres Mote 1856 7: 160-175.
Schram, F. R. 1986. Crustacea. Oxford University Press, New York, 606 pp.
Sleigh, M. A., and D. I. Barlow. 1980. Metachronism and control of
locomotion in animals with many propulsive structures, pp. 49-70. In,
H. Y. Elder and E. R. Trueman (eds.), Aspects of Animal Movement.
Cambridge University Press, Cambridge.
Smith, S. I. 1884. Report on the decapod Crustacea of the Albatross
dredgings off the east coast of the United States in 1883. Report of the
United States Commission of Fish and Fisheries 10[for 1882]: 345-426,
pls. I-X.
Thomasson, M. A., M. L. Johnson, J.-O. Strömberg, and E. Gaten. 2003.
Swimming capacity and pleopod beat rate as a function of sex, size
and moult stage in Northern krill Meganyctiphanes norvegica. Marine
Ecology Progress Series 250: 205-213.
Webb, P. W. 1986. Effect of body form and response threshold on the
vulnerability of four species of teleost prey attacked by largemouth bass
(Micropterus salmoides). Canadian Journal of Fisheries and Aquatic
Sciences 43: 763-771.
R ECEIVED: 30 March 2011.
ACCEPTED: 22 September 2011.