Content uploaded by David Ellerby
Author content
All content in this area was uploaded by David Ellerby on May 17, 2018
Content may be subject to copyright.
1 3
J Comp Physiol A (2016) 202:425–433
DOI 10.1007/s00359-016-1090-3
ORIGINAL PAPER
The effects of steady swimming on fish escape performance
Sanam B. Anwar1 · Kelsey Cathcart1 · Karin Darakananda1 · Ashley N. Gaing1 ·
Seo Yim Shin1 · Xena Vronay1 · Dania N. Wright1 · David J. Ellerby1
Received: 11 March 2016 / Revised: 29 April 2016 / Accepted: 30 April 2016 / Published online: 9 May 2016
© Springer-Verlag Berlin Heidelberg 2016
Introduction
The ability to execute rapid maneuvers is essential to
the survival and fitness of many animals (Watkins 1996;
Miles 2004; Walker et al. 2005; Husak 2006; Domenici
et al. 2014). Successful predator avoidance and prey cap-
ture depend on achieving high translational and rotational
velocities (Daniel 1984) stemming from coordinated
movements that effectively transfer force to a substrate or
momentum to a fluid (Weihs 1973; Kullberg et al. 1998;
Nauen and Shadwick 2001; Card 2012). In nature, many
factors have the potential to constrain or disrupt perfor-
mance and, therefore, organismal fitness (Domenici 2010;
Domenici et al. 2011a, b). One important consideration
is that escape movements are frequently initiated when
an animal is already in motion (Jayne and Lauder 1993;
Hedenström and Rosén 2001). Under these circumstances,
the escaping animal may be interacting with its environ-
ment, recruiting locomotor muscles, and positioning its
limb and body segments in ways that reduce performance
while transitioning to an escape maneuver. Despite this, the
majority of the available data for performance, kinemat-
ics, and neuromuscular control of escape behaviors were
obtained with the animal starting at rest (Webb 1976; Eaton
and Emberley 1991; Domenici and Blake 1997; Gerry et al.
2012), and may not fully predict performance and its impli-
cations for fitness under field conditions.
Fish escapes are among the best characterized of verte-
brate escape maneuvers, typically consisting of a sequence
of axial bends (Weihs 1973). The stage 1 bend is generated
by near simultaneous activation of the fast muscle along
one side of the body (Jayne and Lauder 1993; Goldbogen
et al. 2005), and the antagonistic, stage 2 bend is driven by
a wave of fast muscle activation proceeding axially from
anterior-to-posterior (Jayne and Lauder 1993; Goldbogen
Abstract Escape maneuvers are essential to the survival
and fitness of many animals. Escapes are frequently initi-
ated when an animal is already in motion. This may intro-
duce constraints that alter the escape performance. In
fish, escape maneuvers and steady, body caudal fin (BCF)
swimming are driven by distinct patterns of curvature of
the body axis. Pre-existing muscle activity may therefore
delay or diminish a response. To quantify the performance
consequences of escaping in flow, escape behavior was
examined in bluegill sunfish (Lepomis macrochirus) in both
still-water and during steady swimming. Escapes executed
during swimming were kinematically less variable than
those made in still-water. Swimming escapes also had
increased response latencies and lower peak velocities and
accelerations than those made in still-water. Performance
was also lower for escapes made up rather than down-
stream, and a preference for down-stream escapes may be
associated with maximizing performance. The constraints
imposed by pre-existing motion and flow, therefore, have
the potential to shape predator–prey interactions under field
conditions by shifting the optimal strategies for both preda-
tors and prey.
Keywords Escape · Swimming · Kinematics · Predation
* David J. Ellerby
dellerby@wellesley.edu
1 Department of Biological Sciences, Wellesley College,
Wellesley, MA 02481, USA
426 J Comp Physiol A (2016) 202:425–433
1 3
et al. 2005). These axial bends transfer momentum to
the water, generating rapid translation and rotation of the
body (Tytell and Lauder 2008). The movements are not
stereotyped; as there is kinematic variation in still-water,
and the behavior can be modulated with respect to chang-
ing environmental factors (Domenici et al. 2011a, b). Rapid
acceleration and behavioral variation interact to determine
escape success (Howland 1974), and both may be affected
by pre-existing motion and water flow.
If already engaged in sustained body caudal fin (BCF)
swimming driven by the waves of slow myotomal muscle
contraction passing from anterior-to-posterior along the
body axis (Altringham and Ellerby 1999; Shadwick and
Gemballa 2006), the fish must switch between distinct
motor control patterns. This may create a mechanical con-
flict between pre-existing patterns of motor activity, mus-
cle contraction, and body curvature that limits mechani-
cal performance and the ability to evade predators. Water
flows may also shape patterns of escape variation. In flow-
ing water, escape trajectories parallel to the flow may be
favored, as they would minimize drag forces, and poten-
tially enhance performance if a fish allowed itself to be
carried with the flow. Such constraints on variation could;
however, increase prey vulnerability by allowing the pre-
diction of escape behavior by predators.
Given the high likelihood of fish escape behaviors being
initiated under dynamic conditions of pre-existing motion
and flow, an understanding of the effects these conditions
have on escape performance is essential for understand-
ing behavior and fitness in the field (Rubin et al. 2016).
To determine the performance consequences of a transi-
tion between steady locomotion and unsteady escapes, we
have examined fish escape maneuvers both in still-water
and while swimming in a steady flow. Bluegill sunfish
were used for the experiment, as they swim in two modes:
median paired fin (MPF) swimming using the pectoral fins
at low speeds, and BCF swimming at higher speeds. This
should allow us to determine if a change in escape perfor-
mance during swimming is associated with exposure to
flow per se (MPF swimming), or at least partially associ-
ated with mechanical resistance due to pre-existing axial
muscle activity (BCF swimming). We hypothesized that:
(1) escapes executed in flow would be kinematically dis-
tinct from and less variable than those executed in still-
water, and (2) there would be a performance cost associ-
ated with transitioning from BCF swimming to an escape
maneuver due to the mechanical resistance offered by the
axial muscle powering BCF swimming. To test these pre-
dictions, fish were exposed to acoustic stimuli from up-
and down-stream locations, and their response latency,
escape direction, and center of mass motion relative to the
both the flow, and an external view point was quantified.
This external frame of reference allowed an assessment of
performance relative to a potential predator external to the
flow.
Methods
Bluegill sunfish (Lepomis macrochirus, Rafinesque)
were obtained from a commercial supplier (Carolina Bio-
logical, Burlington, NC, USA). Fish were maintained in
20-gallon aquaria at 21 °C and fed on frozen blood worms
ad libitum. Escape performance and response latency
data were obtained from six fish (body mass 3.6 ± 0.2 g,
mean ± SD).
All data were collected from fish in the working section
(20 × 20 × 70 cm) of a 90-L recirculating flume (Loligo
Systems, Denmark). Escapes were triggered by dropping a
weight, consisting of 1-L PVC bottle part filled with water,
into the flume at either the up-stream or down-stream
end of the working section to create an acoustic stimulus
(Eaton et al. 1981). To avoid visual cues, the weight was
dropped down an opaque, 10-cm internal diameter PVC
pipe with its opening positioned 2 mm above the water sur-
face (Fig. 1). Weights were suspended from a quick release
shackle and released by pulling the trigger pin via a length
of monofilament fishing line that placed the operator out
of the field of view of the fish. The falls were arrested by
a length of parachute cord that restricted their penetration
below the water surface to 3-cm depth. This minimized
water displacement by the weight and prevented a second
stimulus arising from the impact of the weight on the bot-
tom of the working section. This stimulus setup ensured
that escape responses were triggered by the pressure wave
initiated by weight contact with the water surface, rather
than any visual cues from the falling weight or an opera-
tor. During swimming, fish oriented themselves directly
Fig. 1 Experimental setup for quantifying escape performance dur-
ing steady swimming. Escape responses were triggered by releasing a
weight down an opaque tube onto the water surface at either the front
or back of the working section
427J Comp Physiol A (2016) 202:425–433
1 3
into the flume flow. This meant that the two stimulus loca-
tions were definitively up-stream or down-stream relative
to the fish. For still-water escapes with no orienting flow in
the working section, this no longer applied. To minimize
the confounding effects from differences in stimulus direc-
tion, analyses of still-water performance were confined
to escapes, where the long axis of the fish was not more
than 15° out of alignment with the long axis of the flume
working section. To minimize hydrodynamic wall effects,
constraints on escape direction, and confounding effects
of stimulus distance, fish escapes were excluded from the
analysis if a fish was within 1 body length of the walls of
the working section or 2 body lengths of the up- or down-
stream stimuli.
Responses to stimuli during swimming were consist-
ently away from the stimulus source. The stimulus arrange-
ment therefore prevented a test of whether fish preferred
up-stream or down-stream escape trajectories in response
to stimuli that were not unambiguously located either up-
or down-stream. Additional data were therefore collected
for a lateral stimulus. This consisted of an object swung
inward as a pendulum to give a direct lateral impact on the
flume working section at the approximate location of the
fish during swimming. Responses where the weight did not
strike the working section perpendicular to the fish location
were excluded from the analysis. Measurements of pres-
sure waves associated with each stimulus type established
a peak sound level of 158 dB (reference pressure 1μPa) for
the lateral stimulus, and 165 dB (reference pressure 1μPa)
for the up- and down-stream stimuli.
Fish were video recorded from above using a high-
speed digital video camera (AOS X-PRI camera, AOS
Technologies AG, Baden Daettwil, Switzerland) at a frame
rate of 1000 Hz and resolution of 1024 × 800 pixels (1
pixel = 0.6 mm). A mirror placed along the side of the
flume, angled at 45° provided a simultaneous lateral view.
This allowed the trajectory of the fish to be reconstructed
in three dimensions and the impact of the stimulus weight
with the water to be viewed.
Escape responses were recorded under three flume
conditions, still-water, 9.4 cm s−1 flow and 23.1 cm s−1
flow, and with stimuli either from the front or back of
the working section. The order in which a given fish was
exposed to a particular flow regime was determined using
a random number generator. A given fish was exposed to
up to ten stimuli per flow regime with stimulus direction
for a given trial determined at random. Fish were allowed
3-min rest between stimuli, and if exposed to a flow, the
flow speed reduced to 5.0 cm s−1 during the rest period.
This spacing and number of stimuli have previously been
found sufficient to avoid reduced performance or respon-
siveness in this species (Gerry et al. 2012; Hitchcock
et al. 2015).
Response latency was defined as the time from weight
impact on the water surface until initial lateral dis-
placement of the fish snout during stage 1 of the escape
response. The speed of an acoustic pressure wave in fresh
water is approximately 1500 ms−1 (Marczak 1997). The
pressure disturbance from the stimulus impact should
therefore take approximately 0.1 ms to reach the fish in the
center of the working section. The center of mass (COM)
of bluegill sunfish is located approximately 40 % of total
body length from the snout when the fish is in a straight
position (Tytell and Lauder 2008; Gerry et al. 2012). This
location on the midline and the snout of each fish were
tracked using Image J. Although typically taken as an indi-
cator of COM position during fast-starts for tracking pur-
poses (Webb 1978; Domenici and Blake 1997), the true
COM does shift from the straight body COM location
during body bending (Wakeling 2006). Position data were
smoothed using a smoothing spline interpolation in the
application Igor Pro (v. 6.2, Wavemetrics, Lake Oswego,
OR, USA). This method is similar to the cubic spline
algorithm recommended by Walker (1998) for calculating
velocities and accelerations from position data. The level
of smoothing was dictated by the standard deviation of the
data. Smoothed COM position-time data were integrated to
obtain displacement, differentiated to obtain COM veloc-
ity, and velocity differentiated to obtain COM acceleration.
We recorded the peak velocity and accelerations reached
during each escape, as well as the displacement and aver-
age velocity achieved over the first 20 ms of the escape.
This approach approximates that used by Ghalambor et al.
(2004), where the fixed time period approximates the dura-
tion of a predator–prey interaction. Performance was deter-
mined in both absolute terms, based on the motion of the
fish relative to the mean velocity of the water flow, and
apparent terms, based on the motion of the fish relative
to an observer external to the flow. Absolute performance
fully characterizes the mechanical performance of the fish,
as it accounts for the motion of the fish relative to the water
flow, whereas apparent performance has fitness implica-
tions, as it defines the motion of fish relative to a potential
predator located external to the flow.
COM and snout position data were also used to calculate
the change in heading of the fish between initiation of the
escape and the end of stage 2, defined from the completion
of the second, contralateral axial curve. Angle data were
analyzed for uniformity and left:right symmetry using a
circular statistics package (Oriana, ver. 3.21, Kovach Com-
puting Services, Pentraeth,UK). Rayleigh’s test established
that angle distributions were non-uniform for both left and
right turns in all fish (Rayleigh, p < 0.05). Circular vari-
ance, equivalent to a coefficient of variation for non-direc-
tional data, was used as a relative indicator of the disper-
sion of the distributions, with 0 indicating concentration
428 J Comp Physiol A (2016) 202:425–433
1 3
at a single direction. Frequency distributions for left and
right turns for each fish were compared using a Mardia-
Watson-Wheeler test (Mardia 1972), a nonparametric test
for differences between samples of circularly distributed
data; where no differences in the angle distributions were
detected between left and right turns, data were combined
for further analysis. Multiple pairwise comparisons based
on the Mardia-Watson-Wheeler test were also used to test
for inter-individual differences escape angle. To account for
the use of multiple comparisons, the experiment-wise error
rate was adjusted using a sequentially rejective multiple
test procedure applying Ryan’s Q (Ryan 1960).
Data were tested for normality using a Kolmogorov–
Smirnov test (p < 0.05) and Levene’s equality of error vari-
ances test (p < 0.05). Analysis of variance (ANOVA) was
used to test for the effects of swimming speed and stimu-
lus direction on mechanical performance. Both absolute
and apparent peak velocity (Vabs and Vapp) and acceleration
(aabs and aapp), displacement (Dabs and Dapp), and average
velocities (
¯
Vabs
and
¯
Vapp
) were used as dependent variables.
Swimming speed (still water, 9.4 and 23.1 cm s−1) and
stimulus direction (up-stream/anterior, down-stream/poste-
rior) were included as fixed factors in the statistical models.
An identifier for each fish was also included as a random
factor. Partial-eta squared (
η2
p
) is reported as a measure of
effect size. A planned post-hoc comparisons procedure
was used to test for differences between performance dur-
ing swimming and performance in still-water as the con-
trol condition. A binomial test was used to determine
whether the proportion of up-stream:down-stream escapes
in response to a lateral stimulus deviated from a 1:1 ratio.
All linear statistical analyses were carried out using the
application PASW Statistics (Version 18, SPSS, Chicago,
IL, USA).
Results
Up-stream escapes during swimming showed an initial
S-shaped stage 1 curve (Fig. 2b). Down-stream escapes
during swimming had a pronounced C-bend down-stream
turn (Fig. 2c). For escapes executed in still-water, the mean
escape angles were 83.9° and 58.1° for anterior and poste-
rior stimuli, respectively (Fig. 3a). There was no difference
between the escape angle distributions for anterior and
posterior stimuli in still-water (Mardia-Watson-Wheeler,
W = 4.76, p < 0.05). In contrast, the mean combined
escape angle during swimming was 15.1° for a down-
stream stimulus and 156.6° for an up-stream stimulus
(Fig. 3b). These were different from one another and from
the escape angles distributions for still-water (Mardia-Wat-
son-Wheeler, p < 0.05). There was little variation in escape
angle during swimming compared to still-water, particu-
larly for responses to down-stream stimuli (Fig. 3). Cir-
cular variance for swimming escapes with a down-stream
stimulus was 0.007 compared to 0.059 for an up-stream
stimulus. For escapes executed in still-water, the circular
variances were 0.067 and 0.144 for anterior and posterior
stimuli, respectively. In response to a lateral stimulus, the
ratio of up-stream to down-stream escapes was 16:68, a
significant deviation from a 1:1 ratio of up-stream:down-
stream escape responses (Binomial test, p < 0.05).
Response latency increased significantly with flow speed
(F = 38.0 (2,255), p < 0.05,
η2
p
= 0.70), but did not change
with respect to stimulus direction (F = 2.26 (1,255),
p > 0.05,
η2
p
= 0.15). There was no interaction between
speed and stimulus direction (F = 0.34 (2,255), p > 0.05,
η2
p
= 0.03). Planned contrasts indicated that latency was
longer at the fastest swimming speed (p < 0.05), but not dif-
ferent between the zero- and low-speed conditions (Fig. 4).
Performance was determined in both absolute terms,
based on the motion of the fish relative to the mean velocity
of the water flow, and apparent terms, based on the motion
of the fish relative to an observer external to the flow. Vabs
and aabs changed with both stimulus location and flow
speed (Fig. 5a, c; Table 1), declining overall with speed
(Post-hoc planned contrasts, p < 0.05). Dabs and
¯
Vabs
did
not change with speed, but did change with stimulus loca-
tion (Fig. 5e, g; Table 1). Performance was greater for up-
stream than down-stream stimuli (Fig. 5). No interaction
effects were detected for stimulus location and flume speed
for absolute performance measures.
As with absolute performance, all apparent perfor-
mance measures changed with stimulus direction (Table 2).
There were significant interaction effects between swim-
ming speed and stimulus direction:
Vapp
(F = 4.93 (2,145),
p < 0.05,
η2
p
= 0.49), aapp (F = 3.38 (2,145), p < 0.05
η2
p
= 0.40),
Dapp
(F = 9.75 (2,145), p < 0.05,
η2
p
= 0.65),
ab c
Fig. 2 Sequential fish outlines taken from still video images of rep-
resentative escape responses. a Still-water escape. b Up-stream
escape during BCF swimming in response to a down-stream stimu-
lus. c Down-stream escape during BCF swimming in response to an
up-stream stimulus. Outlines are numbered sequentially in ms from
initiating the escape movement. The starting position of the fish has a
dotted outline, and the approximate end of the stage 1 bend, defined
from the reversal of direction of the snout indicating onset of the con-
tralateral stage 2 axial curve, is indicated by the shaded outline
429J Comp Physiol A (2016) 202:425–433
1 3
and
¯
Vapp
(F = 6.56 (2,145), p < 0.05,
η2
p
= 0.56). Given
the difficulties in interpreting main effects associated with
an interaction between factors, separate one-way ANOVAs
were performed for up-stream and down-stream stimuli
with planned contrasts to compare the performances dur-
ing steady swimming to those in still-water. Detected dif-
ferences are indicated in Fig. 5. All apparent performance
measures in response to a down-stream stimulus declined
with increasing flow speed. For an up-stream stimulus, Vapp
did not change with speed, aapp declined, and both Dapp and
¯
Vapp
increased with speed (Fig. 5).
Discussion
Changes in mechanical performance in the presence
of flow
Escape behavior and performance in flowing water were
significantly altered relative to those in still-water. There
was a major kinematic shift relative to still-water, with
escapes being directed either up- or down-stream (Figs. 2,
3). In general, the effects of flow on absolute performance
were negative. Peak velocities and accelerations were
reduced, and response latency was increased (Figs. 4, 5).
The extent and nature of the effect of flow were highly
dependent on the direction of the stimulus and the result-
ing escape trajectory. Unsurprisingly perhaps, escapes per-
formed against the flow had lower performance than those
with a down-stream turn. This was most marked when per-
formance was considered from a point of view external to
the flow. In this case, the performance impairment associ-
ated with attempting to escape up-stream was much more
apparent, and there was a potential performance increase
associated with turning down-stream and being carried by
the current (Figs. 2, 5). A similar down-stream performance
enhancement was noted by Diamond et al. (2016).
The delayed response at the highest flow speed likely
resides in the need to switch motor patterns during the
steady swimming to escape transition. Steady BCF swim-
ming is driven by waves of slow myotomal muscle contrac-
tion passing from anterior to posterior along the body axis
(Altringham and Ellerby 1999; Shadwick and Gemballa
ab
Fig. 3 Angular frequency distributions for escape responses a in
still-water and b during flume swimming. 0° initial orientation of the
fish. Low- and high-speed swimming distributions were not different
and are combined. There were no left:right differences in angle dis-
tributions, and both are combined to show single distributions. Ante-
rior/up-stream and posterior/down-stream stimuli were delivered at
approximately 0 and 180° to the initial fish orientation, as indicated
by closed and open arrows, respectively. Black bars show responses
to anterior/up-stream stimuli, and open bars posterior/down-stream
stimuli
40
30
20
10
0
Response latency (ms)
0.20.10.0
Flume flow speed (ms
-1
)
*
ns
Fig. 4 Response latency of fish to stimuli delivered from the front or
back of the working section at a range of flume flow speeds. Open
symbols indicate responses to a down-stream or posterior stimulus,
and closed symbols to an up-stream or anterior stimulus. Latency was
greater at the highest swimming speed than during slow swimming or
in still-water. N = 6, mean ± sem
430 J Comp Physiol A (2016) 202:425–433
1 3
2006). The steady swimming to escape transition requires
a switch to fast muscle recruitment (Jayne and Lauder
1993), and the motor output that drives the escape can
override the steady swimming pattern (Svoboda and Fet-
cho 1996). For bluegill of the size used in the present study,
peak force development by the fast myotomal muscle takes
approximately 5 ms (Carroll et al. 2009). Relaxation by the
slow muscle takes approximately 100 ms (DE, unpublished
data). There must therefore be resistance by active slow
muscle during the contraction of the fast myotomal muscle
driving the escape response. The extent to which this ham-
pers the escape will depend on the relative physiological
cross section of the slow and fast myotomal muscle. Slow
muscle represents approximately 5 % of the total myotomal
muscle mass in bluegill (DE, unpublished data); therefore,
overall resistance should be relatively low. Relative cross-
sectional areas of slow and fast muscle do; however, change
with axial location (Ellerby et al. 2000), and particularly in
the caudal region, the physiological cross section of slow
muscle and the resistance to bending is likely to be much
greater than the average. This may explain a reduction in
body curvature during escapes initiated from BCF swim-
ming in adult bluegill (Jayne and Lauder 1993). Given how
dependent escape success is on rapid acceleration to high
velocity (Katzir and Camhi 1993), even a modest effect on
response time and axial kinematics may have a significant
impact on fitness.
Effects of flow on escape variability
Responsiveness and mechanical performance are not the
only factors contributing to escape success. Variation and
unpredictability are also important, as predators can poten-
tially exploit stereotyped prey behaviors (Jablonski and
Strausfeld 2001; Catania 2009). Unpredictability would
be maximized by random escape angles (Humphries and
Driver 1970). In practice, still-water fish escape maneu-
vers are not random (Wöhl and Schuster 2007; Domenici
2010; Domenici et al. 2011a, b; Marras et al. 2011; Hitch-
cock et al. 2015), although still highly variable (Fig. 3a).
In contrast, escapes made in flowing water are much less
variable (Fig. 3b). Escape angles shift, so that the fish
escapes on a final trajectory aligned relatively closely to
the flow direction (Figs. 2, 3). As fish are initially oriented
into the flow, escape angles are, therefore, either small to
retain up-stream orientation, or much larger than in still-
water to reverse the direction of movement (Fig. 3). This is
a major kinematic shift from still-water, where the majority
of escape angles fall between these two extremes (Figs. 2,
3). This intermediate angle range may not be compat-
ible with high performance in flowing water, as it would
expose the up-stream lateral body surface to the flow creat-
ing large drag forces that destabilize the fish and push it
down-stream. Down-stream escapes were kinematically
much more similar to a typical still-water C-start than up-
stream escapes, although executed with a greater turn angle
during stage 1 (Fig. 2). The stage 1 curve was held until the
head of the fish was oriented down-stream, and the stage
2 contralateral wave of the curvature started, propelling
1.6
1.4
1.2
1.0
Velocity (ms
-1
)
0.20.10.0
1.6
1.4
1.2
1.0
0.20.10.0
240
200
160
120
0.20.10.0
240
200
160
120
Acceleration (ms
-2
)
0.20.10.0
Flume flow velocity (ms
-1
)
0.022
0.020
0.018
0.016
0.014
0.20.10.0
0.022
0.020
0.018
0.016
0.014
Displacement (m)
0.20.10.0
*
1.0
0.9
0.8
0.7
0.6
1.0
0.9
0.8
0.7
0.6
Average velocity (m
s
-1
)
0.20.10.0 0.20.10.0
*
*
*
*
*
*
*
*
*
*
*
a Peak velocity b Peak velocity
**
*
Relative to flow
Relative to external
reference point
c
Peak acceleration
d
Peak acceleration
f
Displacement,
20ms
h
Average velocity,
20ms
g
Average velocity,
20ms
e
Displacement,
20ms
Fig. 5 Escape performance of juvenile bluegill sunfish at a range of
swimming speeds. Performance is expressed either in absolute terms
based on center of mass (COM) motion relative to the water flow (left
graphs) or in apparent terms based on COM motion relative to a point
external to the flow (right graphs). Graphs show peak COM veloci-
ties (a, b). Peak COM accelerations (c, d). COM displacements after
20 ms (e, f). Average velocity over the first 20 ms of the escape (g,
h). Closed symbols show the performance in response to an up-stream
stimulus (resulting in a down-stream turn), and the open symbols per-
formance in response to a down-stream stimulus (resulting in an up-
stream escape). *Indicates a difference from performance at the zero
flume speed level as detected by a post-hoc planned contrasts analysis
(p < 0.05). Mean ± sem, N = 6
431J Comp Physiol A (2016) 202:425–433
1 3
the fish down-stream. In contrast, up-stream escapes were
kinematically similar to the S-start described during still-
water escapes (Webb 1976; Spierts and van Leeuwen 1999;
Hale 2002) and prey capture lunges (Hoogland et al. 1956;
Harper and Blake 1991) with low turn angles. Rather than
the pronounced C-bend typical of escapes in still-water, the
body axis was thrown into a double, S-bend, and a wave
of undulation similar to that seen in steady BCF swimming
passed along the body axis (Fig. 2b). S-starts may allow
greater control of the final heading than in C-starts (Spi-
erts and van Leeuwen 1999), a requirement for maintain-
ing a stable trajectory against the water flow. The shift to
kinematically less variable escapes in flowing water could
increase the vulnerability of fish to predation, as escapes
are much more likely to occur parallel to the current, nar-
rowing the range of possible escape trajectories that need to
be anticipated by a predator. The up-stream escape may be
a particularly poor option for predator avoidance, as it com-
bines low variability with reduced mechanical performance
relative to a down-stream escape. On that basis, it might
be expected that fish should prefer down-stream escapes.
When exposed to the lateral stimulus, there was a marked
preference for down-stream turns, suggesting that if possi-
ble the fish maximize performance and trajectory variation
by making a down-stream turn.
The observed shifts in trajectory and performance rela-
tive to still-water escapes are likely to affect escape success
when exposed to predatory fish, or avian plunge-diving or
bill-striking predators. Rapid acceleration to high veloc-
ity can increase escape success (Katzir and Camhi 1993;
Walker et al. 2005), but modulation of escape behavior is
also an important factor. Prey can potentially modulate their
escape trajectory to maximize distance from an approach-
ing predator (Howland 1974; Arnott et al. 1999; Soto et al.
2015). The importance of optimizing an escape trajectory
for predator avoidance is dependent on the relative veloci-
ties of predator and prey (Soto et al. 2015). Where veloc-
ities of predator and prey are broadly similar, as is often
the case for interactions between predatory fish and their
smaller prey species, optimizing escape trajectory relative
to the predator is the dominant factor in determining escape
success (Soto et al. 2015). The constraints of escaping in
flow and the low variation in escape trajectory (Fig. 3) may,
however, limit the adoption of optimal trajectories.
Where predators are fast relative to their prey, preda-
tor accuracy rather than prey response becomes the domi-
nant factor, determining the outcome of the predator prey
interaction (Soto et al. 2015). Avian plunge-diving or bill-
striking predators fall into this category, as the dive veloc-
ity for successful kingfishers is 7.9 ms−1 (Katzir and Camhi
1993), compared to average escape velocities in the present
study in the 0.6–1.0 ms−1 range (Fig. 5). Predator accuracy
could be aided by the relatively narrow range of trajecto-
ries adopted in flow and likelihood of down-stream turns
increasing prey predictability. Accurate prey strikes are
particularly important to avian plunge or strike predators,
as they must compensate for air/water light refraction that
displaces the apparent location of underwater prey (Katzir
et al. 1999), and have limited opportunity to modulate
their rapid prey capture movements once they are initi-
ated (Katzir and Intrator 1987; Katzir and Camhi 1993).
The relative predictability of escape responses in flow may
Table 1 ANOVA results for absolute escape performance relative to water flow
F-statistics are from a two way ANOVA model with flow speed (0, 9.4 and 23.1 cm s−1) and stimulus direction (up-stream or down-stream) as
fixed effects. Partial-eta squared (
η2
p
) is reported as a measure of effect size
Performance measure Speed effect Stimulus direction effect
Peak absolute velocity, Vabs, (ms−1)F(2,145) = 5.44, p < 0.05,
η2
p
= 0.51 F(2,145) = 23.87, p < 0.05,
η2
p
= 0.82
Peak absolute acceleration, aabs, (ms−2)F(2,145) = 5.55, p < 0.05,
η2
p
= 0.55 F(2,145) = 12.55, p < 0.05,
η2
p
= 0.67
Absolute displacement, Dabs, (m) F(2,145) = 0.05, p > 0.05,
η2
p
= 0.01 F(2,145) = 8.32, p < 0.05,
η2
p
= 0.62
Mean absolute velocity,
¯
Vabs
, (ms−1)F(2,145) = 0.72, p > 0.05,
η2
p
= 0.13 F(2,145) = 16.61, p < 0.05,
η2
p
= 0.77
Table 2 ANOVA results for apparent performance relative to a viewpoint external to the flow
F-statistics are from a two way ANOVA model with flow speed (0, 9.4 and 23.1 cm s−1) and stimulus direction (up-stream or down-stream) as
fixed effects. Partial-eta squared (
η2
p
) is reported as a measure of effect size
Performance measure Speed effect Stimulus direction effect
Peak apparent velocity, Vapp, (ms−1)F(2,145) = 0.07, p > 0.05,
η2
p
= 0.03 F(2,145) = 55.98, p < 0.05,
η2
p
= 0.92
Peak apparent acceleration, aapp, (ms−2)F(2,145) = 3.98, p < 0.05,
η2
p
= 0.44 F(2,145) = 26.60, p < 0.05,
η2
p
= 0.83
Apparent displacement, Dapp, (m) F(2,145) = 0.02, p > 0.05,
η2
p
= 0.004 F(2,145) = 16.17, p < 0.05,
η2
p
= 0.76
Mean apparent velocity,
¯
Vapp
, (ms−1)F(2,145) = 0.31, p > 0.05,
η2
p
= 0.06 F(2,145) = 35.38, p < 0.05,
η2
p
= 0.88
432 J Comp Physiol A (2016) 202:425–433
1 3
therefore explain the higher capture rates of small fish by
egrets in small streams relative to still-water (Lotem et al.
1991), but direct observations of predator prey interactions
are required to confirm whether predators can exploit the
effects of flow (Rubin et al. 2016).
Acknowledgments The work was supported by the Department of
Biological Sciences at Wellesley College, a Wellesley College Faculty
Research Grant, and National Science Foundation grant 1354274 to
DE. All procedures were approved by the Institutional Animal Care
and Use Committee at Wellesley College.
References
Altringham JD, Ellerby DJ (1999) Fish swimming: patterns in muscle
function. J Exp Biol 202:3397–3403
Arnott SA, Neil DM, Ansell AD (1999) Escape trajectories of the
brown shrimp Crangon crangon, and a theoretical consideration
of initial escape angles from predators. J Exp Biol 202:193–209
Card GM (2012) Escape behaviors in insects. Curr Opin Neurobiol
22:180–186
Carroll AM, Ambrose AM, Anderson TA, Coughlin DJ (2009) Feed-
ing muscles scale differently from swimming muscles in sunfish
(Centrarchidae). Biol Lett 5:274–277
Catania KC (2009) Tentacled snakes turn C-starts to their advantage
and predict future prey behaviour. Proc Natl Acad Sci USA
107:11183–11187
Daniel TL (1984) Unsteady aspects of aquatic locomotion. Am Zool
24:121–134
Diamond KM, Schoenfuss HL, Walker JA, Blob RW (2016) Is fish escape
performance influenced by flow? Tests in the Hawaiian stream goby
Sicyopterus simpsoni. Integr Comp Biol 56(suppl 1):e53
Domenici P (2010) Context-dependent variability in the components
of fish escape response: integrating locomotor performance and
behaviour. J Exp Zool Part A 313:59–79
Domenici P, Blake RW (1997) The kinematics and performance of
fish fast-start swimming. J Exp Biol 200:1165–1178
Domenici P, Blagburn JM, Bacon JP (2011a) Animal escapology I:
theoretical issues and emerging trends in escape trajectories. J
Exp Biol 214:2463–2473
Domenici P, Blagburn JM, Bacon JP (2011b) Animal escapology II:
escape trajectory case studies. J Exp Biol 214:2474–2494
Domenici P, Wilson ADM, Kurvers RHJM, Marras S, Herbert-Read
JE, Steffensen JF, Krause S, Viblanc PE, Couillaud P, Krause
J (2014) How sailfish use their bills to capture schooling prey.
Proc R Soc B 281:20140444. doi:10.1098/rspb.2014.0444
Eaton RC, Emberley DS (1991) How stimulus direction determines
the trajectory of the mauthner-initiated escape response in a tel-
eost fish. J Exp Biol 161:469–487
Eaton RC, Lavender WA, Wieland CM (1981) Identification of
Mauthner-initiated response patterns in goldish: evidence from
simultaneous cinematography and electrophysiology. J Comp
Physiol 144:521–531
Ellerby DJ, Altringham JD, Williams T, Block BA (2000) Slow mus-
cle function of Pacific bonito (Sarda chiliensis) during steady
swimming. J Exp Biol 203:2001–2013
Gerry SP, Robbins A, Ellerby D (2012) Variation in fast-start per-
formance within a polyphenic population of bluegill sunfish
(Lepomis macrochirus). Physiol Biochem Zool 85:694–703
Ghalambor CK, Reznick DN, Walker JA (2004) Constraints on adap-
tive evolution: the functional trade-off between reproduction and
fast-start swimming performance in the Trinidadian guppy (Poe-
cilia reticulata). Am Nat 164:38–50
Goldbogen JA, Shadwick RE, Fudge DS, Gosline JM (2005) Fast-
start muscle dynamics in the rainbow trout Oncorhynchus
mykiss: phase relationship of white muscle shortening and mid-
line curvature. J Exp Biol 208:929–938
Hale ME (2002) S- and C-start escape responses of the muskellunge
(Esox masquinongy) require alternative neuromotor mechanisms.
J Exp Biol 205:2005–2016
Harper DG, Blake RW (1991) Prey capture and the fast-start perfor-
mance of Northern Pike Esox lucius. J Exp Biol 155:175–192
Hedenström A, Rosén M (2001) Predator versus prey: on aerial hunt-
ing and escape strategies in birds. Behav Ecol 12:150–156
Hitchcock AC, Chen T, Connolly E, Darakananda K, Jeong J, Quist
A, Robbins A, Ellerby DJ (2015) Trade-offs between perfor-
mance and variability in the escape responses of bluegill sunfish
(Lepomis macrochirus). Biol Open. doi:10.1242/bio.201511577
Hoogland R, Morris D, Tinbergen N (1956) The spines of stickle-
backs (Gasterosteus and Pygosteus) as means of defence against
predators (Perca and Esox). Behaviour 10:205–236
Howland HC (1974) Optimal strategies for predator avoidance: the
relative importance of speed and manoeuverability. J Theor Biol
47:333–350
Humphries DA, Driver PM (1970) Protean defence by prey animals.
Oecologia 5:285–302
Husak JF (2006) Does survival depend on how fast you can run or
how fast you do run? Funct Ecol 20:1080–1086
Jablonski PG, Strausfeld NJ (2001) Exploitation of an ancient escape
circuit by an avian predator: relationships between taxon-specific
prey escape circuits and the sensitivity to visual cues by a preda-
tor. Brain Behav Evol 58:218–240
Jayne BC, Lauder GV (1993) Red and white muscle activity and
kinematics of the escape response of the bluegill sunfish during
swimming. J Comp Physiol A 173:495–508
Katzir G, Camhi JM (1993) Escape response of black mollies (Poe -
cilia sphenops) to predatory dives of a pied kingfisher (Ceryle
rudis). Copeia 1993:549–553
Katzir G, Intrator N (1987) Striking of underwater prey by reef her-
ons, Egretta gularis schistaeea. J Comp Physiol A 60:517–523
Katzir G, Tamir S, Schechtman E, Hareli S, Arad Z (1999) Cattle
egrets are less able to cope with light refraction than are other
herons. Anim Behav 57:687–694
Kullberg C, Jakobsson S, Fransson T (1998) Predator induced take-
off strategy in great tits (Parus major). Proc R Soc Lond B
265:1659–1664
Lotem A, Schechtman E, Katir G (1991) Capture of submerged prey
by little egrets, Egretta garzetta garzetta: strike depth, strike angle
and the problem of light refraction. Anim Behav 42:341–346
Marczak W (1997) Water as a standard in the measurements of speed
of sound in liquids. J Acoust Soc Am 102:2776–2779
Mardia KV (1972) A multi-sample uniform scores test on a circle and
its parametric competitor. J R Stat Soc Ser B 34:102–111
Marras S, Killen SS, Claireaux G, Domenici P, Domenici P, McKen-
zie DJ (2011) Behavioural and kinematic components of the fast-
start escape response in fish: individual variation and temporal
repeatability. J Exp Biol 214:3102–3110
Miles DB (2004) The race goes to the swift: fitness consequences of
variation in sprint performance in juvenile lizards. Evol Ecol Res
6:63–75
Nauen JC, Shadwick RE (2001) The dynamics and scaling of force
production during the tail-flip escape response of the California
spiny lobster Panulirus interruptus. J Exp Biol 204:1817–1830
Rubin AM, Diamond KM, Schoenfuss HL, Blob RW (2016) Field
observation of intraspecific and predatory attack behaviors of the
Hawaiian sleeper fish, Eleotris sandwicensis. Integr Comp Biol
56(suppl 1):e53
Shadwick RE, Gemballa S (2006) Structure, kinematics, and muscle
dynamics in undulatory swimming. In: Shadwick RE, Lauder
433J Comp Physiol A (2016) 202:425–433
1 3
GV (eds) Fish Biomechanics. Fish Physiology vol 23. Academic
Press, San Diego, pp 241–280
Soto A, Stewart WJ, McHenry MJ (2015) When optimal strategy mat-
ters to prey fish. Int Comp Biol 55:110–120
Spierts IL, van Leeuwen JL (1999) Kinematics and muscle dynam-
ics of C- and S-starts of carp (Cyprinus carpio L.). J Exp Biol
202:393–406
Svoboda KR, Fetcho JR (1996) Interactions between the neural networks
for escape and swimming in goldfish. J Neurosci 16:843–852
Tytell ED, Lauder GV (2008) Hydrodynamics of the escape response in
bluegill sunfish, Lepomis macrochirus. J Exp Biol 211:3359–3369
Wakeling JM (2006) Fast-start mechanics. In: Shadwick RE, Lauder
GV (eds) Fish biomechanics, fish physiology, vol 23. Academic
Press, San Diego, pp 333–368
Walker JA (1998) Estimating velocities and accelerations of animal
locomotion: a simulation experiment comparing numerical dif-
ferentiation algorithms. J Exp Biol 201:981–995
Walker JA, Ghalambor CK, Griset OL, McKenney D, Reznick DN
(2005) Do faster starts increase the probability of evading preda-
tors? Funct Ecol 19:808–815
Watkins TB (1996) Predator-mediated selection on burst swimming
performance in tadpoles of the Pacific tree frog, Pseudacris
regilla. Physiol Zool 69:154–167
Webb PW (1976) The effect of size on the fast-start performance of
rainbow trout Salmo gairdneri, and a consideration of piscivo-
rous predator-prey interactions. J Exp Biol 65:157–177
Webb PW (1978) Fast-start performance and body form in seven spe-
cies of teleost fish. J Exp Biol 74:211–226
Weihs D (1973) The mechanism of rapid starting in slender fish. Bior-
heology 10:343–350
Wöhl S, Schuster S (2007) The predictive start of hunting archer fish:
a flexible and precise motor pattern performed with the kinemat-
ics of an escape C-start. J Exp Biol 210:311–324
A preview of this full-text is provided by Springer Nature.
Content available from Journal of Comparative Physiology A
This content is subject to copyright. Terms and conditions apply.