You are on page 1of 23

Biol. Rev. (2001), 76, pp.

449471 " Cambridge Philosophical Society


DOI: 10.1017\S1464793101005759 Printed in the United Kingdom
449
Mechanics and aerodynamics of insect ight
control
GRAHAM K. TAYLOR
Department of Zoology, Oxford University, South Parks Road, Oxford, OX1 3PS, UK
(E-mail : graham.taylor!zoo.ox.ac.uk)
(Received 7 November 2000; revised 14 June 2001
ABSTRACT
Insects have evolved sophisticated ight control mechanisms permitting a remarkable range of manoeuvres.
Here, I present a qualitative analysis of insect ight control from the perspective of ight mechanics, drawing
upon both the neurophysiology and biomechanics literatures. The current literature does not permit a
formal, quantitative analysis of ight control, because the aerodynamic force systems that biologists have
measured have rarely been complete and the position of the centre of gravity has only been recorded in a
few studies. Treating the two best-known insect orders (Diptera and Orthoptera) separately from other
insects, I discuss the control mechanisms of dierent insects in detail. Recent experimental studies suggest
that the helicopter model of ight control proposed for Drosophila spp. may be better thought of as a
facultative strategy for ight control, rather than the xed (albeit selected) constraint that it is usually
interpreted to be. On the other hand, the so-called constant-lift reaction of locusts appears not to be a reex
for maintaining constant lift at varying angles of attack, as is usually assumed, but rather a mechanism to
restore the insect to pitch equilibrium following a disturbance. Dierences in the kinematic control
mechanisms used by the various insect orders are related to dierences in the arrangement of the wings, the
construction of the ight motor and the unsteady mechanisms of lift production that are used. Since the
evolution of insect ight control is likely to have paralleled the evolutionary renement of these unsteady
aerodynamic mechanisms, taxonomic dierences in the kinematics of control could provide an assay of the
relative importance of dierent unsteady mechanisms. Although the control kinematics vary widely between
orders, the number of degrees of freedom that dierent insects can control will always be limited by the
number of independent control inputs that they use. Control of the moments about all three axes (as used
by most conventional aircraft) has only been proven for larger ies and dragonies, but is likely to be
widespread in insects given the number of independent control inputs available to them. Unlike in
conventional aircraft, however, insects control inputs are likely to be highly non-orthogonal, and this will
tend to complicate the neural processing required to separate the various motions.
Key words : Insect ight control, steering, unsteady aerodynamics, kinematics, constant-lift reaction, turning,
manoeuvre.
CONTENTS
I. Introduction ............................................................................................................................ 450
II. Manoeuvre control versus reex stabilisation ........................................................................... 451
III. General principles of control ................................................................................................... 452
IV. Flight control in Diptera (true ies)........................................................................................ 454
(1) Longitudinal control ......................................................................................................... 454
(2) Lateral control .................................................................................................................. 456
450 Graham K. Taylor
V. Flight control in Orthoptera (locusts and crickets) ................................................................. 459
(1) Longitudinal control ......................................................................................................... 459
(2) Lateral control .................................................................................................................. 460
VI. Flight control in other insects .................................................................................................. 461
(1) Longitudinal control ......................................................................................................... 461
(2) Lateral control .................................................................................................................. 463
VII. Discussion ................................................................................................................................ 463
(1) How many degrees of freedom do insects control?............................................................ 463
(2) Evolution of insect ight control systems .......................................................................... 466
VIII. Conclusions .............................................................................................................................. 467
IX. Acknowledgements .................................................................................................................. 467
X. References................................................................................................................................ 467
I. INTRODUCTION
Insect ight control has been studied extensively
from a physiological perspective, but its mechanics
are less well known. Even where the kinematic
changes elicited by a given stimulus have been
dened, their consequences for aerodynamic force
production often remain obscure. Earlier work on
ight control was rmly rooted in the assumptions of
quasi-steady aerodynamics. However, since detailed
quasi-steady analyses and direct force measurements
(e.g. Weis-Fogh, 1973; Norberg, 1975; Cloupeau,
Devillers & Devezeaux, 1979; Ellington, 1984c ;
Ennos, 1989; Dudley & Ellington, 1990b; Wilkin,
1990; Zanker & Go$ tz, 1990; Wilkin & Williams,
1993) indicate that aerodynamic force production in
insects generally relies upon unsteady mechanisms
(Willmott, Ellington & Thomas, 1997; Dickinson,
Lehmann & Sane, 1999), our understanding of
insect ight control must necessarily incorporate
unsteady eects. Although our understanding of
unsteady mechanisms is still very limited, more
recent experimental studies (Go$ tz, 1987; Dickinson,
Lehmann &Go$ tz, 1993; Dickinson, 1999; Dickinson
et al., 1999) have begun to investigate their role in
insect ight control.
An experimental approach integrating physio-
logical and mechanical observations is clearly de-
sirable. For example, although electromyographic
studies have demonstrated that migratory locusts
Locusta migratoria generate dierent motor patterns in
response to roll and yaw stimuli (Zarnack & Mo$ hl,
1977), they cannot tell us whether this is sucient to
allow roll and yaw rotations to be produced
separately. This can only be resolved by directly
measuring the torques and forces that the insects
produce, or by analysing high-speed lm of the
insects in ight. Although the current literature does
permit some synthesis of aerodynamics and kin-
ematics with neurobiology and muscle mechanics
(see Kammer, 1985 for a review of this type), I have
chosen to concentrate in detail upon the aero-
dynamics and kinematics of ight control. This
approach is similar to that of most texts on aircraft
ight mechanics (e.g. Nelson, 1989; Etkin & Reid,
1996; Cook, 1997; Vinh, 1993) in that it does not
complicate the discussion of ight dynamics with
details of the engineering mechanism that adjusts an
aircrafts elevators or varies an insects wingbeat
frequency. Instead, I will concentrate upon how
dierent kinematic inputs aect the dependent
variables of position and velocity, as illustrated
schematically in Fig. 1. The more general question of
how many degrees of freedom insects can control is
one of the central issues of insect ight control and
will form the theme of this review.
The black box between the inputs and outputs
of Fig. 1 is properly replaced by the equations of
motion for a ying body together with a system of
control inputs
e.g. stroke amplitude,
wingbeat frequency
position
attitude
velocity
angular
velocity
angular acceleration
position
attitude
velocity
angular
velocity
acceleration
Fig. 1. Block diagram illustrating the overall approach of
this review. The black box represents the equations of
motion for a ying body, together with a system of
transfer functions providing the mathematical relation-
ship between control inputs and their eects upon the
dependent variables.
451 Mechanics of insect ight control
transfer functions providing the mathematical re-
lationship between the control inputs and their
eects upon the dependent variables. I shall only be
discussing the mechanics of insect ight control in
qualitative terms, however, as the current literature
does not allow for even a formal static, let alone
dynamic, quantitative framework to be applied. The
reasons for this are best seen by considering the two
possible approaches to the quantitative analysis of
insect ight control. The rst is to correlate free-
ight manoeuvres with observed changes in wing
kinematics. For example, Wakeling & Ellington
(1997) correlated velocity and ight force with
several basic kinematic parameters measured from
high-speed lm of free-ying dragonies. However,
the purpose of tting regression lines in that study
was only to identify trends and not to estimate a
functional relationship (Wakeling & Ellington,
1997, p. 566) and since the equations for the
regressions were not recorded, it is dicult to draw
any further quantitative conclusions. In any case,
accurately determining the three-dimensional wing
kinematics of even a stationary insect presents
formidable technical diculties, and to do so during
free-ight manoeuvres is very dicult indeed.
An alternative approach is to measure directly the
forces and moments on tethered insects and to
correlate these with the observed changes in wing
kinematics. Unfortunately, although many studies
have measured selected forces and torques generated
by tethered insects, the resulting force systems are
rarely complete, in the sense of dening the point of
action, as well as the magnitude and direction, of the
resultant ight force. Only two published studies
(Hollick, 1940; Blondeau, 1981) appear to have
made the necessary measurements to determine the
point of action of the resultant ight force, yet this is
essential in determining the eect of a force, as any
child who has ever played on a see-saw will know.
The position of the centre of gravity has likewise
rarely been recorded, but is critical in determining
both the animals responsiveness to control inputs
and its characteristics of stability and equilibrium.
Its role in ight dynamics is analogous to that of the
fulcrum in a game of see-saw. For a full dynamic
analysis of insect ight control, it will be necessary
also to determine the insects moments of inertia.
The importance of the distribution of mass in
determining the envelope of insect ight control has
been emphasised by recent experimental work on
the ight performance of neotropical butteries
(Srygley & Dudley, 1993; Srygley, 1994, 1999;
Srygley & Kingsolver, 2000). A more rigorous
experimental approach is clearly required before we
will make any further inroads towards a quantitative
analysis of insect ight control.
As a nal caveat, it is worth emphasising that the
picture of insect ight mechanics portrayed in this
review is far from complete. For example, although
the open-loop conditions that are normally imposed
by tethering often lead to exaggerated turning
responses (Robert & Rowell, 1992a), laboratory
experiments may not reveal the full gamut of control
responses used by insects ying under natural
conditions. Only where we can identify clear
physiological limits upon performance can we be
sure that we have correctly identied the boundary
of the ight envelope. As the quantity and detail of
experimental work on insect ight control improves,
our knowledge of insects ight envelopes looks
certain to expand.
II. MANOEUVRE CONTROL VERSUS REFLEX
STABILISATION
Insect control systems serve two discrete functions :
to allow controlled manoeuvres and to augment or
provide stability in the face of external (or possibly
internal) perturbations. Since stability will tend to
oppose both accidental and deliberate deviations
from steady ight, these two functions are expected
to be in direct conict unless some mechanism is
present to bypass or desensitise the reex stabilisation
system during manoeuvres. In the absence of such a
mechanism, the command to elicit a manoeuvre
would need to override any signals from the reex
stabilisation system, presumably with a consequent
drop in nervous, if not also mechanical, eciency.
From the viewpoint of enhancing manoeuvrability,
the potential temporarily to disable a reex stabil-
isation system makes active control of stability
preferable to passive stability, which cannot gen-
erally be switched o. The impressive ight
envelopes of ies and other highly manoeuvrable
insects must reect in large part a strong reliance
upon active, rather than passive, maintenance of
stability.
On the other hand, since active control may
augment, rather than completely replace, passive
stability, a brief discussion of the latter may be useful
in setting corrective ight control in context. The
rst point to note is that hovering insects, like
hovering helicopters (Gessow & Amer, 1949;
Johnson, 1980; Padeld, 1996), possess no rst-order
passive stability about any axis. This is because the
452 Graham K. Taylor
resultant ight force vector passes through the centre
of gravity at equilibrium and remains xed in
magnitude, position and direction with respect to
the body axes as the insects orientation changes.
The net turning moment on the insect is therefore
always zero and a stabilising moment can only arise
through second-order changes in the aerodynamic
forces caused by translation of the insect as the ight
force is redirected with respect to gravity. This is the
means by which the dihedral of a xed wing aircraft
provides roll stability, and it is possible that the
mean upward-canted coning angle of the wings of a
apping insect could provide both pitch and roll
stability in a similar way (Dudley, 2000).
The situation may be very dierent in forward
ight, where a nose-up pitching disturbance will
increase the angle of incidence of the wings, thereby
increasing quasi-steady lift production. Provided this
additional lift acts behind the centre of gravity, this
will tend to generate a stabilising nose-down moment
about the centre of gravity. Alternatively, if the
mean ight force acts well above the centre of
gravity, the insect will gain stability much like that
of a ship through a pendulum eect (Lighthill,
1974). It is harder to predict how unsteady mech-
anisms of force production will aect stability. Some,
like Weis-Foghs (1973) clap-and-ing and its
derivatives (Ellington, 1984b) or generation of a
circulation via wing rotation (Ellington, 1984b;
Dickinson et al., 1999), are unlikely to be greatly
aected by changes in the insects orientation, and
are therefore unlikely to provide any passive stab-
ility. On the other hand, translational mechanisms,
such as the maintenance of a leading edge vortex
(Ellington, 1984b; Dickinson, and Go$ tz, 1993;
Ellington et al., 1996; Willmott et al., 1997),
eectively magnify the wing lift coecients and will
therefore tend to enhance any passive stability or
instability that would be present if the ow were
quasi-steady (G. K. Taylor & A. L. R. Thomas, in
preparation).
In the absence of detailed empirical data on the
passive stability of insects, it is dicult to make any
reliable predictions about how reex stabilisation
and passive control will interact. Suce it to say that
active control may be required to provide not just
static stability, in the sense of returning the insect
towards equilibrium immediately following a dis-
turbance, but also dynamic stability, in the sense of
damping out any oscillatory modes that may be
excited by a disturbance. For example, most aircraft
possess some degree of passive static stability, but
pilot or computer control input is generally required
to maintain an acceptable level of dynamic stability.
The same may be expected to be broadly true of
ying insects. With these considerations in mind, I
will go on to consider some more general principles
of control.
III. GENERAL PRINCIPLES OF CONTROL
We will begin by analogy with a more familiar
control system. A car on a skidpan has three degrees
of freedom. This is equivalent to saying that we need
three pieces of information to completely describe its
position and orientation in space (for example, an x
coordinate and a y coordinate for its position, and a
compass bearing for its heading). To describe
completely the motion of the system we would, of
course, need also the rst-order and second-order
derivatives of the three positional degrees of freedom,
but for simplicity we will only consider terms
referring to position and orientation. By Newtons
second law, the application of a force or moment to
a mechanical system produces an acceleration or
angular acceleration of equivalent sense or direction
that can be used to control position or orientation.
Hence, the number of controlled degrees of freedom
cannot exceed the number of independent control
roll pitch
yaw
Fig. 2. A ying insect, like this hummingbird hawkmoth
Macroglossum stellatarum, has six degrees of freedom. The
insect is free to move in translation along or rotate about
three orthogonal axes, centred upon the centre of gravity
(part-lled circle). Rotations about these axes are termed
pitch, roll and yaw.
453 Mechanics of insect ight control
inputs. This is the rst and most fundamental
principle of control.
Our second principle is that it is often unnecessary
(or even undesirable) to control all of the possible
degrees of freedom. For example, a car gives its
driver control of just two of its three degrees of
freedom, making it easier to keep on the road, but
somewhat harder to parallel park. Our third
principle is that one or more redundant control
inputs can often be used to provide enhanced control
of a given degree of freedom. For example, easing o
a cars accelerator has the same qualitative eect as
braking, but most of us would nd it dicult to stop
at trac lights without the aid of brakes.
Such considerations allow us to frame two over-
arching questions that we may ask of insect ight
control. How many independent control inputs does
the insect use? How many degrees of freedom does
this allow the insect to control? Independent control
of all six of an insects degrees of freedom (Fig. 2)
would require a minimum of six independent control
inputs. Full six-degree-of-freedom control is super-
uous in most aerial applications and appears not to
be used by insects, which lack the ability to y
sideways without banking. In fact, for control of
position in space, independent modulation of pitch,
heading and longitudinal acceleration will suce,
requiring just three independent control inputs.
Two-axis control systems like these are a feature of
some high-performance model sailplanes and could
be used by insects with no great need for manoeuvra-
bility.
Three-axis control systems permit independent
modulation of pitch, roll and yaw and thus enable a
far wider range of manoeuvres. Depending upon
whether the direction as well as the magnitude of
longitudinal acceleration can be varied independent
of pitch, three-axis control systems require a mini-
mum of either four or ve independent control
inputs. In xed-wing aircraft and helicopters, the
direction of longitudinal acceleration is normally
dependent upon pitch, although aps may be used
to steepen a descent without upsetting pitch equi-
librium. Amongst operational aircraft, perhaps only
the Harrier jump jet can be considered to possess
complete ve-degree-of-freedom control in the sense
of being able to accelerate at any angle to the ground
without varying pitch. However, whereas ve-
degree-of-freedom control is the exception in aircraft
and four-degree-of-freedom control the rule, the
opposite may be the case in insects with three-axis
control because of important dierences in the
mechanisms by which control inputs are produced.
Fig. 3. Conventional aircraft use mechanically separate
control surfaces (shaded) to separate control of the three
moments. In the simple system shown here, the tail
elevators are used to control pitch, the ailerons to control
roll, and the rudder to control yaw. Insects lack separate
control surfaces and must instead use dierent rotations of
the wings to separate the moments.
Early experiments in wing warping aside, the
conventional aircraft design paradigm has long been
to use separate control surfaces to isolate pitch, roll
and yaw (Fig. 3). This diers markedly from the
control systems of insects in which the same
kinematic parameters are varied symmetrically for
control of the longitudinal forces and moments, and
asymmetrically for control of the lateral forces and
moments. For example, in locusts, pronation (nose-
down rotation of the wing at the top of the stroke) is
increased symmetrically between the forewings in
response to a nose-up pitching disturbance, but is
only increased on the inside forewing during turns
(Gettrup & Wilson, 1964). The result is an eective
doubling of the number of control inputs and a
remarkable economy of control. It follows that
insects that modulate the direction of longitudinal
acceleration by symmetrically varying the balance of
lift and thrust on their wings should also be able to
separate roll and yaw by varying the same kinematic
parameters asymmetrically. Conversely, insects that
modulate roll via asymmetric lift production, and
yaw via asymmetric thrust production, should also
be able to modulate the direction of longitudinal
acceleration to some degree. Hence, we would expect
many insects using three-axis control to modulate
motions in ve, rather than four, of their six degrees
of freedom, although this is not to say that all
combinations of these motions will be possible.
454 Graham K. Taylor
Having considered some general principles of
control, I will now go on to apply these specically
to insects. Although insect ight control has only
been studied in detail in two orders (Diptera and
Orthoptera), there are major systematic dierences
in the mechanics of control. These relate primarily to
dierences in the construction of the ight motor
and the arrangement of the wings and to probable
dierences in the unsteady mechanisms of aero-
dynamic force production used. I will therefore
consider each of the dierent orders in turn, treating
lateral and longitudinal control separately.
IV. FLIGHT CONTROL IN DIPTERA (TRUE
FLIES)
Flies are the most aerobatic of insects, able to turn on
a pinpoint, y sideways and even land upside-down
on a ceiling (Kammer, 1985), but since they possess
just one pair of wings the mechanics of their ight
control should be amongst the simplest to under-
stand. Specically, there can be no question of
interference between the fore- and hindwings, or of a
changing balance of forces between the two. Flies
exhibit two main classes of manoeuvre: fast, vol-
untary, turns, known as saccades because of their
kinematic resemblance to the tracking movements of
vertebrate eyes, and slower, often corrective, turns
under the inuence of the optomotor control system.
Other classes of turn have been also recorded, often
as pursuit responses by courting males. For example,
dolichopodid ies Poecilobothrus nobilitatus have been
lmed executing 180m yaw turns in less than 40 ms
during courtship (Land, 1993). Hovering male
tabanid ies Hybomitra hinei have even been recorded
using a modied form of the Immelmann turn a
classic aerobatic manoeuvre consisting of a half-loop
followed by a half-roll, used to eect rapid reversals
of ight direction (Wilkerson & Butler, 1984).
Throughout this review I will be presenting results
relating to both voluntary and corrective control,
and it should be borne in mind that control inputs
that are suitable for one type of control may not be
suitable for the other, especially where the speed of
the motion is very dierent.
(1) Longitudinal control
Stroke amplitude the angle of the planar arc or
great circle described by the wing leading edge
between the top and bottom of the stroke (Fig.
A B
stroke plane
stroke plane
Fig. 4. (A, B). Stroke amplitude is the planar angle of the
arc described by the wing leading edge between the top
and bottom of the stroke (dotted line in column A).
Column B shows the corresponding points of action of the
mean ight force. Decreasing the stroke amplitude
(bottom row) causes the ight force to shift back,
generating a nose-down pitching moment about the
centre of gravity (part-lled circle).
4A) is normally considered the main control
parameter determining aerodynamic power output
in fruit ies Drosophila spp. Stroke amplitude shows a
strong positive correlation with aerodynamic force
production (Vogel, 1967; Go$ tz, 1968; Lehman &
Dickinson, 1997; Dickinson, Lehmann & Chan,
1998) and is referred to as stroke angle in some
studies. Wingbeat frequency is the other principal
determinant of aerodynamic power output, and
appears to be an important control parameter in ies
(Dickinson et al., 1998). For example, Drosophila
melanogaster increase their wingbeat frequency in
response to an increase in the speed of optic ow
simulating forward ight (Friedrich, Spatz &
Bausenwein, 1994).
In Drosophila melanogaster, wingbeat frequency is
positively correlated with aerodynamic force below
approximately the force required to support body
weight, but is negatively correlated with aerody-
namic force at peak outputs (Lehmann & Dickinson,
1997). Decreases in wingbeat frequency appear to be
compensated by steeper increases in stroke am-
plitude, however, and the product of frequency and
455 Mechanics of insect ight control
amplitude (which determines the translational vel-
ocity of the wing) rises linearly with increasing force
production (Lehmann & Dickinson, 1997). The
drop in wingbeat frequency at peak force production
may result from physiological constraints, so stroke
amplitude and wingbeat frequency are probably not
strictly independent control inputs, even though
they are not coupled in a straightforward fashion. At
peak levels of force production, D. melanogaster
appear to be limited to a unique combination of
stroke amplitude, wingbeat frequency and ight
force coecient, and this in turn constrains their
ability to generate the asymmetries in force pro-
duction necessary for lateral control (Lehmann &
Dickinson, 2001). An analogous degradation of
turning ability with increasing ight speed was
found by Bueltho, Poggio & Werhahn (1980), who
found the maximum angular velocity attained by
free-ying D. melanogaster to decrease with increasing
forward velocity, although in this case the eect is
likely to result from aerodynamic, rather than
physiological, constraints.
In larger calypterate ies (blowies, houseies,
etc.), frequency and amplitude are both positively
correlated with aerodynamic force, but since the two
parameters are likely to be mechanically linked, it is
not necessarily easy to separate cause from eect
(Nachtigall & Wilson, 1967). Symmetrical changes
in stroke amplitude are also correlated with cor-
rective pitch control in larger ies such as Muscina
stabulans and Calliphora erythrocephala, which decrease
the amplitude of both wings in response to an
increase in body angle (Hollick, 1940; Nalbach &
Hengstenberg, 1994). Changes in amplitude are
eected by adjusting the lower turning point of the
wings, which shifts back with a decrease in amplitude
as the wings sweep less far forward on the down-
stroke. This causes the point of action of the mean
ight force to shift back (Fig. 4B), generating a
pitching moment with the ys weight of appropriate
direction to restore it to equilibrium (Hollick, 1940).
Similar, albeit smaller, changes in stroke amplitude
are observed during pitching responses in Drosophila
spp. (Zanker, 1988b, 1990), which may use a similar
mechanism for pitch control (Vogel, 1967). How-
ever, the smaller magnitude of the response is
suggestive of a slightly dierent mechanism. A
prominent feature of unsteady lift-production in
Drosophila spp., at least in tethered ight, is the clap-
and-peel process in which the opening wings draw
additional air into the circulation at the start of the
downstroke (see Ellington, 1984a, b). The extent of
contact between the wings increases with increasing
stroke amplitude, which should increase lift and
thrust production dorsally, inducing a restoring
nose-up torque in response to nose-down stimuli
(Zanker, 1990).
Drosophila melanogaster are also able to vary the
longitudinal position of the stroke plane independent
of amplitude (Zanker, 1988b), which should allow
pitch to be modulated independent of aerodynamic
force output. Similar shifts in stroke plane position
have also been observed in the hoveries Eristalis
tenax and Episyrphus balteatus (Ellington, 1984a).
Stroke plane inclination varies too during pitching
responses in Drosophila spp., but this is tightly
correlated with changes in amplitude, which cause
the lower turning point to move almost horizontally,
rather than parallel to the stroke plane, thereby
altering stroke inclination (Vogel, 1967; Zanker,
1988b). Drosophila spp. do not seem to vary the angle
of attack of the wings for longitudinal control (Vogel,
1967), although the constancy of the angle of attack
suggests that they might actively stabilise it at
dierent speeds (Zanker, 1990). The speed and
timing of wing rotation do appear to be important
lateral control parameters in D. melanogaster (Zanker,
1990; Dickinson et al., 1993), but have not been
investigated in the context of longitudinal control.
In addition to varying their wing kinematics, ies
may control pitch by varying their posture, much
like a hang-glider pilot. Drosophila melanogaster elevate
their abdomen in response to nose-down disturb-
ances (Zanker, 1988a, b), displacing the centre of
gravity dorsal to the line of thrust, which therefore
generates a restoring nose-up moment with the ys
inertia. The concurrent increase in drag on the
dorsal side will further add to this torque (Zanker,
1988b). Similar postural changes have been observed
in Calliphora erythrocephala (Nalbach & Hengstenberg,
1994), so abdominal deection may be a widespread
mechanism of pitch control in ies. Drosophila virilis
have also been observed to elevate their hindlegs
following a nose-down disturbance, which should
increase drag dorsally and generate a nose-up
pitching moment (Vogel, 1967). The raised hindlegs
can hardly fail to meet with interference from the
wake of the wings, and this could enhance their
turning eect in much the same way that a ships
rudder is more eective if placed in the wake of the
propeller (Vogel, 1967).
The various kinematic and postural changes
described above are only sucient to provide two
degrees of freedom for longitudinal control, since
none of them permits control of the direction of the
resultant ight force relative to the body axes. This
456 Graham K. Taylor
would result in essentially the same situation as in
helicopters, in which the direction of the rotor force
is xed with respect to the body. Hence, although a
hovering helicopters vertical displacement can be
controlled directly by varying the aerodynamic force
on the blades, it can only be given horizontal
velocity by changing the body angle to redirect the
ight force forward (e.g. Johnson, 1980). This does
in fact appear to be the mechanism by which
Drosophila spp. normally control their trajectory
(Vogel, 1966; David, 1985; Zanker, 1988b), since
the elevation of the aerodynamic force is apparently
kept xed relative to the body axes (Vogel, 1966;
Go$ tz, 1968; David, 1978; Go$ tz & Wandel, 1984;
but see Ennos, 1989). The relative proportions of lift
and thrust are thus varied primarily by adjusting the
body angle (David, 1985; Zanker, 1988b), which is
negatively correlated with ight speed (Vogel, 1966;
David, 1978) to allow level ight to be maintained as
aerodynamic power output is increased. However,
the fact that Drosophila spp. normally modulate lift
and thrust together need not imply that they are
strictly coupled as in helicopters, and the helicopter
model may perhaps be better thought of as a
facultative strategy for ight control rather than as a
xed constraint.
Similar results have been reported for tethered
houseies Musca domestica ying in still air (Go$ tz &
Wandel, 1984), but the generality of this result is
questionable because M. domestica does not hover
(Wagner, 1986) and a ow of air over the antennae
is necessary for a normal stroke path in Muscina
stabulans (Hollick, 1940). Free-ying M. domestica do
appear to be able to vary the direction of the ight
force vector, but only within a limited range
(Wagner, 1986). Force measurements in tethered
Calliphora spp. have conrmed that they also are able
to modulate thrust independent of lift within a
certain range (Blondeau, 1981, Nachtigall & Roth,
1983), although the relation may be asymmetric
because lift appears never to be modulated in-
dependent of thrust (Blondeau, 1981).
Any ability to alter the direction of the ight force
independent of its magnitude implies that some
other kinematic parameter must be varied inde-
pendent of stroke amplitude and wingbeat fre-
quency, which together provide control of only the
magnitude of the ight force, albeit with an element
of redundancy. Changes in stroke plane inclination
will certainly alter the direction of the resultant
ight force, and would almost certainly aect the
forces generated by wake capture mechanisms in
which the wings gain an additional kick from
vortices shed at the end of the previous half-stroke
(see Dickinson et al., 1999). This is one potential
mechanism by which larger ies could vary the
direction of the ight force. Changes in stroke plane
inclination observed in Drosophila melanogaster are
probably too small to be used for control purposes
(0n5m; Zanker, 1988b), and being linked to
changes in stroke amplitude should probably not be
counted as a separate control input anyway. Other
kinematic parameters have not been investigated in
the context of longitudinal control, but since longi-
tudinal forces and moments can be produced by
symmetrically adjusting parameters used for lateral
control, we will consider other potential candidates
in the context of lateral control.
(2) Lateral control
Although changes in wing kinematics are likely to be
more important for fast ight manoeuvres, postural
adjustments are commonly associated with slow
optomotor steering responses in ies (Zanker,
1988a). Typically, the abdomen is deected into a
turn (Go$ tz, Hengstenberg & Biesinger, 1979;
Zanker, 1988a; Zanker, Egelhaaf & Warzecha,
1991), often together with the inside hindleg
(Hollick, 1940; Faust, 1952; Nachtigall & Roth,
1983; Nalbach, 1989). Occasionally both hindlegs
are extended together into the turn (Go$ tz et al.,
1979). Similar adjustments in locusts have been
interpreted as producing drag-based yaw torques
like a ships rudder (Camhi, 1970) and most workers
have followed the same interpretation for ies (e.g.
Go$ tz et al., 1979). Postural adjustments in other
insects have been suggested to produce lateral
moments by a dierent mechanism, shifting the
centre of gravity sideways from the centre of pressure
(Govind & Burton, 1970; May & Hoy, 1990a). This
would normally be expected to generate a roll rather
than a yaw torque, but abdominal deections in
Drosophila melanogaster do appear to generate yaw
torques gravimetrically (Zanker, 1988a). This is
because the axis about which yawing occurs is tilted
back approximately 30m from the vertical such that
a signicant component of the animals weight acts
normal to the yaw axis and can therefore produce a
yaw couple with the ight force.
Until recently, stroke amplitude was thought to be
the only kinematic parameter consistently correlated
with roll and yaw responses in ies (Drosophila spp. :
Vogel, 1967; Go$ tz, 1968; Go$ tz et al., 1979, Musca
domestica: Srinivasan, 1977, Calliphora spp. : Heide,
1971; Nachtigall & Roth, 1983; Hengstenberg,
457 Mechanics of insect ight control
Sandeman & Hengstenberg, 1986; Nalbach, 1989;
Nalbach & Hengstenberg, 1994). According to the
classical interpretation, the amplitude of the inside
wing is lowered to reduce the aerodynamic force by
shortening the stroke, with opposite changes oc-
curring on the outside wing. This will generate
rolling moments directly through asymmetric lift
production and should also generate yaw moments
directly through asymmetric thrust production.
Since ying insects usually produce much less thrust
than lift (typically an order of magnitude less in
Drosophila virilis ; Vogel, 1966), the resulting torques
about the yaw axis would normally be expected to
be correspondingly small compared to those in roll.
Although this classical view was originally rooted
in the assumptions of quasi-steady aerodynamics, the
outcome could be the same for an unsteady
translational mechanism such as delayed stall, in
which ow remains attached to the wing at higher
angles of attack than would be possible under steady
conditions. Hence, although control torques meas-
ured in Drosophila melanogaster are generally greater
than those calculated from blade-element theory,
which assumes quasi-steady aerodynamics (Zanker,
1990; Zanker & Go$ tz, 1990), stroke amplitude could
still play a direct role in dipteran ight control if
unsteady translational mechanisms were important.
Dickinson et al.s (1993) comment that changes in
amplitude may not be directly responsible for
turning in D. melanogaster (on the grounds that
correlated changes in the timing of stroke reversal
will be more important than stroke length per se if
unsteady rotational mechanisms overwhelm quasi-
steady lift production) implicitly assumes that
unsteady translational mechanisms (which also have
the potential to overwhelm quasi-steady lift pro-
duction) are insignicant.
Changes in amplitude do not permit the direction
of the ight force to be varied directly, unless the
part of the stroke that is curtailed generates a force
in a dierent direction to the remainder of the
stroke. In general, since rolling moments are nor-
mally generated by asymmetric lift and yawing
moments by asymmetric thrust, separate control of
lift and thrust will usually be required for roll and
yaw to be varied independently. Roll and yaw have
therefore been suggested to be unavoidably coupled
in Drosophila spp. (Srinivasan, 1977), which normally
modulate lift and thrust together. To conclude that
Drosophila spp. are unable to uncouple lift and thrust
would seem premature, however, and although
direct evidence that Drosophila spp. modulate roll
and yaw independently is still lacking (Dickinson,
1999), the view that they are strictly coupled has
begun to be challenged (Go$ tz, 1987; Ennos, 1989;
Zanker, 1990; Dickinson et al., 1993). In any case,
varying the relative proportions of lift and thrust
independently on each wing is not the only way to
generate a yaw torque. For example, the clap-and-
peel process does not occur parallel to the sagittal
plane during ctive turns in Drosophila melanogaster,
and the jet of air produced is in the appropriate
direction to contribute to steering (Go$ tz, 1987;
Zanker, 1990). This could allow yaw moments to be
generated independent of roll, although the torques
generated in this way may still be rather small
(Dickinson et al., 1993; Dickinson, 1999).
Larger ies such as Calliphora spp. (Blondeau,
1981; Schilstra & Van Hateren, 1999) and Musca
domestica (Wagner, 1986) can certainly generate roll
and yaw torques separately, though the two are
usually modulated together to produce a banked
turn (Blondeau, 1981). This presumably increases
manoeuvrability in the same way as in a turning
aircraft, by minimising sideslip and directing the
nose into the turn. Sometimes, it may be useful to
swing sideways without changing heading. Heli-
copters achieve this by banking to one side to
redirect a component of the ight force sideways.
This is probably also the mechanism behind Collett
& Lands (1975) observation that the hovery Syritta
pipiens can y sideways without changing its heading.
This implies that hoveries must also possess in-
dependent control of roll and yaw. At least one other
kinematic parameter must be varied in addition to
stroke amplitude for roll and yaw to be controlled
separately, because abdominal deection occurs too
slowly to account for the rapid turns observed in
ies. For example, in Drosophila melanogaster the
abdomen only responds strongly to oscillating visual
stimuli at frequencies below 5 Hz, and the response
is much stronger at lower frequencies (Zanker,
1988a). Peak deection is only attained at 0n05 Hz,
suggesting that the response may be used to correct
for inherent asymmetries in the wings or ight
motor, rather than discrete disturbances (Zanker,
1988a).
Wingbeat frequency can immediately be ruled out
as a lateral control parameter because the con-
struction of the dipteran ight motor makes it im-
possible for the wings to operate at dierent
frequencies (Hollick, 1940). Stroke plane inclination
also appears not to oer an independent control
input because it is closely linked to amplitude in
most studies in which both have been measured
(Vogel, 1967; Nachtigall & Roth, 1983; Zanker,
458 Graham K. Taylor
1988b, see also Chadwick, 1951). For example, in
Calliphora vicina the action of the basalar muscles has
been shown explicitly to determine both stroke plane
inclination and amplitude (Tu & Dickinson, 1996).
However, the precise three-dimensional trajectory of
the wingtip can be varied independently of am-
plitude in Muscina stabulans (Hollick, 1940) and C.
vicina (Tu & Dickinson, 1996), and this might be
used to produce control moments, perhaps by
altering the properties of wake capture. One wing
may even be completely folded over the body during
extremely fast turns (Nachtigall & Wilson, 1967),
although this is unlikely to play any role in
correctional steering.
Increased pronation of the inside wing has been
observed during roll and yaw responses in Calliphora
erythrocephala (Faust, 1952; Hengstenberg et al.,
1986). Similar changes in pronation have been noted
in the craney Tipula oleracea (Faust, 1952) and the
blowy Lucilia serricata (Sandeman, 1980). Unfortun-
ately, the stage of wingbeat recorded was almost
certainly wrongly described in the latter study (D. C.
Sandeman, pers. comm., cited in Kammer, 1985)
and it is not clear whether the magnitude or only the
timing of rotation was varied. Angle of attack
asymmetries are taken to an extreme in the doli-
chopodid y Poecilobothrus nobilitatus, which is able to
execute extraordinarily rapid 180m turns by holding
the inside wing normal to the ow to act as an
airbrake (Land, 1993).
In light of current views of insect ight control
implicating unsteady aerodynamic mechanisms as
the main source of control moments (Zanker, 1990),
the speed and timing of wing rotation are likely to be
at least as important as the degree of pronation
(Ennos, 1989; Dickinson et al., 1993, 1999;
Dickinson, 1994, 1999). Supination (nose-up ro-
tation of the wings towards the bottom of the stroke)
occurs extremely rapidly in ies (Ennos, 1989). Flow
visualisation of tethered Drosophila melanogaster
(Dickinson & Go$ tz, 1996) appears to show the
vorticity produced by the so-called ventral ip
(Dickinson et al., 1993) fusing with the stopping
vortices shed at the end of the downstroke, so rapid
wing rotation might be used for aerodynamic force
production. Experiments with model wings (Dickin-
son, 1994; Dickinson et al., 1999) have demonstrated
that the timing of supination relative to stroke
reversal (the point at which wing translation reverses
direction) is critical in determining the magnitude of
the forces produced. Large positive forces (attribu-
table to both rotational mechanisms and wake
capture) are generated when supination precedes
stroke reversal, whereas negative forces are registered
if supination is delayed (Dickinson et al., 1999).
By asymmetrically varying force production in
this way, ies could readily produce turning
moments for control. For example, the results of
Dickinson et al. (1999) indicate that the instan-
taneous drag force may be extremely high on a wing
using rotational circulation and wake capture, and
well in excess of the instantaneous lift. Hence,
though the instantaneous drag forces largely cancel
over the course of a normal stroke, asymmetric drag
production oers a potent means of generating roll
or yaw torques in insects using rotational circulation
and wake capture. The direction of the axis about
which the resulting torque is generated will depend
upon both the direction of the stroke and the
orientation of the wing chord. For example, in
insects with a horizontal stroke plane, such as
hoveries, the potential to vary the horizontal force
component may be far greater than the potential to
vary the vertical lift. Asymmetric variation of the
forces produced by rotational circulation and wake
capture would then tend to generate a yaw torque,
and this may account for the remarkable ability of
hoveries to turn full circle whilst otherwise re-
maining stationary.
In Calliphora erythrocephala (Nalbach, 1989) and
Drosophila melanogaster (Dickinson et al., 1993), su-
pination is delayed on the inside wing and advanced
on the outside wing during ctive turns. Parallel
changes in the timing of pronation have also been
observed (Nalbach, 1989). Unfortunately, it has not
been possible to resolve changes in the timing of
stroke reversal, so the phase of supination and stroke
reversal remains unknown. However, assuming that
an absolute delay in supination on the inside of a
turn indicates that supination is also delayed relative
to stroke reversal, then the results of Dickinson et al.
(1999) imply that the ight force will be reduced or
reversed on the inside of a turn. Earlier supination is
also correlated with increased rotational velocity in
D. melanogaster (Dickinson et al., 1993, see also
Zanker, 1990), which should increase aerodynamic
force production on the outside of a turn. Both eects
should combine synergistically to produce a torque
in the correct direction to steer into the turn.
The timing of supination seems to be independent
of stroke amplitude in Drosophila melanogaster, even
though the two are usually modulated together
(Dickinson et al., 1993). Changes in stroke amplitude
and the timing of supination could therefore provide
the two inputs necessary to separate control of roll
and yaw. Alternatively, changes in stroke amplitude
459 Mechanics of insect ight control
might not eect steering directly, but could instead
eect steering indirectly through correlated changes
in the timing of stroke reversal (Dickinson et al.,
1993). However, since it is the relative timing of
supination and stroke reversal that primarily deter-
mines aerodynamic force output, changes in stroke
amplitude and the timing of supination would then
combine together to provide a single composite
variable, which would make changes in stroke
amplitude redundant. In this case, some other
independent control input (e.g. the speed of wing
rotation) would still be required to separate control
of roll and yaw. It therefore seems likely that stroke
length per se could be an important control par-
ameter, at least in larger ies in which roll and yaw
are known to be controlled separately.
Returning briey to longitudinal control, sym-
metric changes in the speed or timing of supination
could be used to vary the magnitude of the resultant
ight force (Dickinson et al., 1999). Changes in the
timing of wing rotation might also provide the
additional control input required to separate control
of lift and thrust. Symmetric adjustment of the
ventral ip could certainly enhance the pitching
responses discussed in the previous section, because
the lever arm at the end of the downstroke will
usually be such that even small changes in the
rotational forces will generate a large pitching
moment. Finally, some control over the direction of
vortex shedding would permit an even greater degree
of control of the direction of the resultant ight force
(Ellington, 1984b; Dickinson et al., 1993).
V. FLIGHT CONTROL IN ORTHOPTERA
(LOCUSTS AND CRICKETS)
(1) Longitudinal control
Flight control in locusts is likely to be complicated by
the interaction of ow over the fore- and hindwings,
though some common principles remain. As in ies,
stroke amplitude and wingbeat frequency are posi-
tively correlated with aerodynamic force and ying
speed in tethered Locusta migratoria able to control
their own speed relative to the surrounding air
(Gewecke, 1975; Kutsch & Gewecke, 1979). Teth-
ered L. migratoria also increase their wingbeat
frequency in response to an increase in the speed of
optic ow simulating forward ight (Baader, Scha$ fer
& Rowell, 1992) and the same qualitative re-
lationship between wingbeat frequency and ight
speed has been shown in free-ying L. migratoria
(Baker, Gewecke & Cooter, 1981). Moreover, stroke
amplitude and wingbeat frequency are negatively
correlated in tethered L. migratoria stimulated by air
currents of xed speeds (Gewecke, 1970, 1972), so
the two must also be able to be varied independently.
Locusts appear to regulate lift independent of
thrust and have also been claimed to exhibit a
constant-lift reaction (Wilson & Weis-Fogh, 1962)
in which the vertical component of force is kept more
or less constant following imposed changes of body
angle of up to 20m (Weis-Fogh, 1956a, b; Gewecke,
1975). Electrophysiological (Wilson & Weis-Fogh,
1962; Gettrup, 1966; Zarnack & Mo$ hl, 1977) and
kinematic studies (Gettrup &Wilson, 1964; Gettrup,
1966; Wortmann & Zarnack, 1993) have clearly
demonstrated that tethered locusts increase forewing
pronation to compensate for the increase in their
angle of attack that would otherwise occur following
a nose-up disturbance. The same reaction also
appears to occur in free ight, since climbing ight is
associated with both increased forewing pronation
and increased body angle (Fischer & Kutsch, 2000).
Hindwing pronation is largely independent of body
angle (Gettrup & Wilson, 1964; Wortmann &
Zarnack, 1993), however, so whereas the forewing
lift coecients should remain approximately con-
stant, the hindwing lift should increase with in-
creasing body angle, because the lift on a wing
generally increases with increasing angle of attack.
According to Gettrup & Wilsons (1964) model of
the constant-lift reaction, the drop in ight speed
that accompanies an increase in body angle in
tethered locusts (Weis-Fogh, 1956a; Gewecke, 1975)
combines with these changes to keep the total lift
approximately constant.
The constant-lift reaction has been an inuential
paradigm in the insect ight literature, and is still
referred to frequently in discussions of insect ight
control (e.g. Zanker, 1990). However, quite apart
from the objection that the quasi-steady aerody-
namics implicit in Gettrup & Wilsons (1964) model
have been shown not to apply in locusts (Cloupeau
et al., 1979; Wilkin, 1990), it is not clear why free-
ying locusts should possess a reex maintaining
constant lift at widely dierent body angles. A
stabilising pitch reex would seem to be of far
greater utility. Studies of free-ying locusts (Baker et
al., 1981) are consistent with the existence of a
stabilising pitch reaction, showing no evidence of the
negative correlation between body angle and ight
speed that forms an essential part of Gettrup &
Wilsons (1964) original scheme. Moreover, later
studies (Zarnack & Wortmann, 1989; Wortmann &
460 Graham K. Taylor
Zarnack, 1993) have cast doubt on the existence of a
constant-lift reaction per se, nding no reliable
evidence that individual locusts consistently main-
tain constant lift at dierent body angles. Although
constant lift may occasionally be maintained, there
is usually considerable temporal variation in lift
production (Zarnack & Wortmann, 1989). Weis-
Foghs (1956a, b) original studies neglected temporal
variation in lift and were based upon data averaged
across individuals, so the apparent constancy of lift
in his gures would certainly have been exaggerated
by the averaging process.
On the other hand, given that the angle of attack
of the hindwings increases following an increase in
body angle, whilst the angle of attack of the forewings
remains approximately constant, the changing bal-
ance of lift between fore- and hindwings should
produce a restoring nose-down pitching moment.
Such a torque was in fact detected by Gettrup &
Wilson (1964), but was considered to be an
unpredicted outcome (Gettrup & Wilson, 1964, p.
189) of the constant-lift reaction. This interpretation
almost certainly reverses cause and eect, since a
stabilising pitch response would negate any re-
quirement to maintain constant lift across a range of
body angles by restoring the animal to equilibrium.
Indeed, loss of the sensory input supposed to mediate
pronation in the constant-lift reaction leads to a
complete loss of stability (Gettrup, 1966), conrming
that the reex changes in pronation described above
are used to maintain pitch stability. Constant lift
may occasionally be maintained in tethered locusts
according to Gettrup & Wilsons (1964) original
scheme, but probably as an epiphenomenon of
changes adapted to restore pitch equilibrium in free
ight.
(2) Lateral control
Locusts swing their abdomen laterally and extend
one or both hindlegs like a rudder during yaw turns
(Gettrup & Wilson, 1964; Dugard, 1967; Camhi,
1970; Cooter, 1979; Arbas, 1986; Baader, 1990;
Preiss &Gewecke, 1991; Robert &Rowell, 1992a, b;
Robertson & Reye, 1992; Preiss & Spork, 1993;
Robertson & Johnson, 1993; Lorez, 1995; Robert-
son, Kuhnert & Dawson, 1996; Dawson et al., 1997).
Such postural changes are not always observed
during correctional responses, however, and may be
more important during voluntary turns (Zarnack &
Mo$ hl, 1977; Gewecke & Philippen, 1978). Similar
postural changes have been observed in the bush-
cricket Tettigonia viridissima (Schulze & Schul, 2001)
and the cricket Teleogryllus oceanicus during acoustic
avoidance responses (Moise, Pollack & Hoy, 1978;
Pollack & Hoy, 1981; Pollack, Huber & Weber,
1984; Miles et al., 1992), although abdominal
deection alone is apparently insucient to produce
yawing in T. oceanicus (May & Hoy, 1990b). As well
as increasing drag on the inside of a turn, hindleg
extension may eect a turn by impeding the motion
of the inside hindwing (Camhi, 1970; May & Hoy,
1990a). This mechanism would be expected to cause
signicant wear to the wing and may therefore be
reserved for extreme avoidance manoeuvres.
Lateral movements of the hindlegs and abdomen
have also been observed in response to visual roll
stimuli in locusts (Taylor, 1981a, b), but are incon-
sistently correlated with rolling (Thu$ ring, 1986) and
said to be neither necessary nor sucient to eect a
correctional roll response (Schmidt & Zarnack,
1987; Waldmann & Zarnack, 1988). However,
Zankers (1988a) careful study of abdominal de-
ection in Drosophila melanogaster highlights the need
to ensure that the rotational axis of a roll stimulus is
exactly normal to the axis about which yawing
occurs. If this is not the case, then a supposed roll
stimulus will induce a combined roll and yaw
response because the subject will perceive the
stimulus to contain some component of yaw. This
probably explains why postural changes normally
associated with yaw turns are observed in response to
roll stimuli, although we cannot discount the
possibility that abdominal deection might be used
to generate roll moments by shifting the centre of
gravity. Alternatively, if roll and yaw are modulated
together in free ight, then responses to roll and yaw
stimuli might be reexly coupled to eect a tighter
banked turn.
Asymmetries in stroke amplitude do not appear to
be strongly correlated with turning in locusts (Baker,
1979; Thu$ ring, 1986; Waldmann & Zarnack, 1988;
Zarnack, 1988) or crickets (Wang & Robertson,
1988; but see May, Brodfuehrer & Hoy, 1988),
except during very sharp turns (Robertson & Reye,
1992; Dawson et al., 1997). Likewise, there is no
evidence that dierences in stroke path are cor-
related with turning in locusts (Zarnack, 1988),
although changes in hindwing elevation may be
important for steering in crickets (Wang & Robert-
son, 1988; May et al., 1988) which also bank their
forewings relative to the body axes during turning
(May et al., 1988).
Locusts (Gettrup & Wilson, 1964) and crickets
(Pollack & Hoy, 1981) reliably increase pronation of
the inside forewing during turning responses. In
461 Mechanics of insect ight control
locusts, this is usually correlated with early pronation
of the inside wing (Dugard, 1967; Baker, 1979;
Zarnack, 1988; Robertson & Reye, 1992; Dawson et
al., 1997). Muscle recordings from crickets are
suggestive of an analogous change in the timing of
the hindwing elevators and depressors (Wang &
Robertson, 1988). These changes should reduce lift
and thrust on the inside wing by decreasing its
eective incidence and might allow it to act as a sort
of brake (Zarnack & Mo$ hl, 1977). Antisymmetric
changes in pronation on the outside wing, possibly
combined with increased supination on the upstroke
(Zarnack, 1988), should increase lift and thrust on
the outside of the turn. The net result will be a
combined roll and yaw torque, leading directly into
a banked turn.
Intriguingly, opposite changes in the timing of the
forewing downstroke muscles have been recorded in
tethered locusts during presumed correctional re-
sponses to imposed yaw angles (Mo$ hl & Zarnack,
1975, 1977; Zarnack & Mo$ hl, 1977). These dif-
ferences seem to be real and may result from
dierences in the turning stimuli used (Kammer,
1985), although the oscillating yaw stimulus used by
Zarnack (1988) was qualitatively the same as that
used in the three studies for which atypical time
shifts were observed. Individual locusts may also use
dierent patterns of muscle burst length in response
to the same roll stimulus (Waldron, 1967), so
dierent motor patterns may be common when only
one degree of freedom is considered. The magnitude
of a roll torque does not necessarily depend upon the
motor pattern used (Thu$ ring, 1986), so dierent
motor patterns must either produce the same
kinematics, which seems unlikely, or they must cause
the forces and moments about the other axes to vary
(Kammer, 1985; Thu$ ring, 1986).
This second explanation seems quite likely. For
example, whereas all of the inside forewing de-
pressors contract early relative to some constant
reference during a typical roll response (Mo$ hl &
Zarnack, 1977; Zarnack & Mo$ hl, 1977; Thu$ ring,
1986; Schmidt & Zarnack, 1987; Waldmann &
Zarnack, 1988), individual depressors may be time-
shifted in opposite directions during yaw responses
(Mo$ hl & Zarnack, 1975; Zarnack & Mo$ hl, 1977).
In any case, Bakers comment that roll and yaw
turns in locusts are just two ways of looking at the
same phenomenon (Baker, 1979, p. 57) is certainly
an over-simplication. Although independent modu-
lation of roll and yaw has never been observed in
locusts (Hensler & Robert, 1990), simultaneous
measurements of roll and yaw torques have not been
made and it remains an open question whether
dierences in motor pattern will prove sucient to
separate roll and yaw control in locusts.
Bulk phase shifts between contralateral wings are
bound to induce lateral asymmetries in the phase of
the fore- and hindwings during roll responses,
because the timing of the hindwings changes less
than that of the forewings (Schmidt & Zarnack,
1987) and may even change in the opposite direction
(Thu$ ring, 1986). Lateral asymmetries in the relative
positions of the fore- and hindwings are certainly
correlated with yaw responses in locusts (Dugard,
1967; Baker, 1979; Cooter, 1979; Robertson &
Reye, 1992; Robertson & Johnson, 1993) and will
inevitably alter the aerodynamic coupling of the
wings (Zarnack, 1982, 1983; Schwenne & Zarnack,
1987; Schmidt & Zarnack, 1987; Waldmann &
Zarnack, 1988; Robertson & Reye, 1992). Whilst
there is now little doubt that ipsilateral phase shifts
are important in locust ight control (Rowell, 1988),
the aerodynamic changes involved are unknown.
Contralateral interference remains unquantied in
locusts, and it is not known whether interference
between contralateral wings has any direct eect
upon steering.
VI. FLIGHT CONTROL IN OTHER INSECTS
(1) Longitudinal control
Studies of ight control in other insects have been
rather limited. Negative correlations between body
angle and ight velocity similar to those observed in
Drosophila spp. have been observed in dragonies
(Wakeling & Ellington, 1997), heteropteran bugs
(Betts, 1986), bumblebees (Dudley & Ellington,
1990a), honeybees (Nachtigall, Widmann &
Renner, 1971; Esch, Nachtigall & Kogge, 1975) and
moths (Dudley & DeVries, 1990; Willmott &
Ellington, 1997). Butteries exhibit considerable
variation in the way in which body angle varies with
ight speed, although this is perhaps not surprising
given the extent to which butteries bodies pitch up
and down through the course of the wingbeat cycle
(Betts & Wootton, 1988; Brackenbury, 1995). For
example, whereas free-ying Papilio rumanzovia adopt
higher body angles and shallower stroke planes
during fast ight, free-ying Troides rhadamantus
appear to show the opposite trend (Betts & Wootton,
1988). Such variation is presumably indicative of
dierences in the aerodynamic mechanisms being
used.
462 Graham K. Taylor
It is important to realise that a negative cor-
relation between ight speed and body angle need
not imply a helicopter-like mode of ight control, in
which the direction of the ight force is xed with
respect to the body axes. Such a model may apply in
honeybees Apis mellica, in which the partitioning of
lift and thrust seems to depend directly upon body
angle (Esch et al., 1975), but in heteropteran bugs,
for example, the inclination of the stroke plane to the
body may be varied by as much as 20m (Betts, 1986).
Likewise in hawkmoths Manduca sexta the stroke
becomes steeper relative to the body at higher speeds
(Willmott & Ellington, 1997) and the same trend
has been observed in the dragony Anax parthenope
(Azuma & Watanabe, 1988). No such trend was
detected in the dragony Sympetrum sanguinem in
which the stroke plane was apparently kept xed
with respect to the body (Wakeling & Ellington,
1997). Nevertheless, the direction of the ight force
does not correlate closely with the orientation of
either the body or the stroke plane in this or other
dragonies (Wakeling & Ellington, 1997), and the
orientation of the stroke plane with respect to ight
direction is generally rather variable (Ru$ ppell,
1989). The correlation of body angle with ight
speed and direction in dragonies has therefore been
suggested to be an adaptation more to minimise
body drag than to redirect the ight force (Wakeling
& Ellington, 1997).
In bumblebees Bombus terrestris, the angle of attack
the wings is increased relative to the stroke plane as
the body angle is decreased (Dudley & Ellington,
1990a), with the result that the eective angle of
incidence of the wings remains more or less un-
changed (Dudley & Ellington, 1990b). In contrast
to Drosophila spp., however, the position and am-
plitude of the wingbeat do not appear to vary with
airspeed, and it is not known how changes in body
angle are brought about (Dudley & Ellington,
1990a). Stroke amplitude does appear to be an
important longitudinal control parameter in heter-
opteran bugs, which display decreased stroke ampli-
tudes during climbing ight, together with an
increase in the speed of stroke reversal and a decrease
in the wingbeat frequency (Betts, 1986). No clear
trend has been found with respect to ight speed or
direction and stroke amplitude in butteries. For
example, whereas free-ying Papilio rumanzovia and
Graphium sarpedon have been observed to increase
stroke amplitude during slow ight, free-ying Precis
iphita adopt sharply increased stroke amplitudes
during bursts of fast ight (Betts & Wootton, 1988).
In hawkmoths Manduca sexta, stroke amplitude
appears to decrease slightly with increasing speed
(Willmott & Ellington, 1997). This is due to a less
ventral excursion of the wings on the downstroke,
which would tend to shift the resultant ight force
rearward as in ies and presumably provides the
nose-down pitching moment required to decrease
the body angle for fast ight.
Opposite trends have been noted in the dragony
Sympetrum sanguinem and the damsely Calopteryx
splendens. In these species, hindwing stroke amplitude
increases with increasing speed (Wakeling & Elling-
ton, 1997), although the regressions were not
signicant at the 95% condence level and there
was no apparent change in the mean positional
angle of the wings. This does not appear to reect a
general trend in dragonies. For example, whereas
no correlation was found between stroke amplitude
and ight speed in free-ying Anax parthenope (Azuma
& Watanabe, 1988), a strong negative correlation
was observed in tethered Orthetrum cancellatum
(Gewecke, Heinzel & Philippen, 1974). Decreased
stroke amplitudes in the latter were associated with
reductions in only the dorsal excursions of the wings,
which would tend to shift the resultant ight force
forward, rather than rearward as in ies and
hawkmoths. If nothing else, these results caution
against extrapolating control mechanisms from one
group of insects, or even one species, to another.
Wingbeat frequency appears not to be an im-
portant longitudinal control parameter in either
hawkmoths (Willmott & Ellington, 1997) or dragon-
ies (Azuma & Watanabe, 1988; Wakeling &
Ellington, 1997), and has also been found to be
uncorrelated with ight speed in damselies (Sato &
Azuma, 1997; Wakeling & Ellington, 1997), and
bumblebees (Dudley & Ellington, 1990a). Wakeling
& Ellington (1997) did, however, note a trend for
wingbeat frequency to increase with increasing force
production in the damsely Calopteryx splendens
(P0n1). Although trends at the 90% signicance
level were deemed good in that study (Wakeling &
Ellington, 1997, p. 566), the signicance of trends
reaching even the 95% level should be viewed with
caution given that some 52 independent regressions
were performed, grossly inating the risk of making
a Type I error. Wingbeat frequency is decreased
during rising ight in heteropteran bugs (Betts,
1986), and there is reasonably strong evidence that
wingbeat frequency is positively correlated with
ight speed in butteries (Betts & Wootton, 1988).
However, given the high degree of interspecic
variation revealed by the latter study in the
correlations between other kinematic parameters
463 Mechanics of insect ight control
and ight velocity, it would seem premature to
interpret data from a few species as indicative of a
more general trend.
(2) Lateral control
Postural adjustments like those observed in ies and
locusts appear to be ubiquitous steering responses in
insects, having also been observed in mantids
(Brackenbury, 1995), heteropteran bugs (Govind &
Burton, 1970; Govind, 1972), strepsipterans (Pix,
Nalbach & Zeil, 1993) and moths (Roeder, 1967;
Kammer, 1971; Kammer & Nachtigall, 1973).
Asymmetries in stroke amplitude like those observed
in turning ies have also been observed in dragonies
(Alexander, 1986), beetles (Burton, 1964, 1971;
Schneider &Kramer, 1974), moths (Kammer, 1971)
and heteropteran bugs (Govind & Burton, 1970;
Govind, 1972). In bugs, the stroke path of the
outside wing may also be shifted dorsal to the inside
wing (Govind & Burton, 1970), which is similar to a
steering mechanism identied by Stellwaag (1916)
in bees and similar to an atypical steering pattern
observed during yaw turns in locusts (Cooter, 1979).
In moths and butteries, the wings on the inside of
a turn are typically held more posteriorly than the
outside wings, and may be slightly exed (Roeder,
1967). These changes probably correlate with
changes in the timing of muscle ring (Kammer,
1971, 1985; Kammer & Nachtigall, 1973; Obara,
1975) and presumably reduce lift and thrust on the
inside wing.
In dragonies, amplitude asymmetries lead to
banking with a strong component of sideslip within
one or two strokes, usually followed by a component
of yaw (Alexander, 1986). This time lag suggests
that yawing may arise indirectly through aero-
dynamic coupling of the moments, rather than
directly through asymmetric thrust production.
Specically, the component of sideslip resulting from
banking means that the oncoming ow is no longer
parallel to the longitudinal axis, which could lead to
the production of a coupled yaw moment. The inside
wing may also be more strongly pronated than the
outside wing during banked turns, although this is
usually secondary to changes in amplitude and may
only be used to ne-tune the response (Alexander,
1986). Very sharp yaw turns involving much more
extreme rotation of the wings have also been
observed in dragonies (Alexander, 1986), sug-
gesting three-axis control of the moments. Phase
shifts between the fore- and hindwings analogous to
those observed in locusts have also been implicated
in free-ight manoeuvres in dragonies (Alexander,
1984, 1986; Ru$ ppell, 1989) and hawkmoths (A. L. R.
Thomas, personal communication). In hawkmoths,
the wings (which are normally physically coupled)
may be completely uncoupled so as to operate in
antiphase. Such phase shifts are likely to be a com-
mon feature of ight control in four-winged insects
and need not necessarily be restricted to insects
in which the wings are normally uncoupled.
VII. DISCUSSION
(1) How many degrees of freedom do insects
control?
To date, three-axis control of the moments has only
been proven in larger ies and dragonies. Exper-
iments to determine whether other insects can
modulate roll and yaw separately have simply not
been performed. However, since the most common
mode of turning is the banked turn, in which roll and
yaw are advantageously coupled to reduce the
turning circle, two-axis control of the lateral and
longitudinal moments could also suce (Dickinson,
1999). Roll and yaw would then be modulated
together about an axis lying somewhere between the
morphological roll and yaw axes. Tables 1 and 2
provide a summary of the various kinematic par-
ameters available for lateral and longitudinal control
in ying insects. In the case of locusts, it is not
obvious whether independent control of forewing
pronation and ipsilateral coupling would permit
separate control of roll and yaw, or whether the
dierent motor patterns observed during roll and
yaw responses result in further kinematic changes
that have so far gone undetected. In Drosophila spp.,
on the other hand, there are more than enough
independent kinematic inputs to permit separate
control of roll and yaw, although it is equally
possible that one or more of these inputs is redundant
and is simply used to provide ner control of the
moments. For example, Zanker (1988a) has found
that abdominal deection and changes in stroke
amplitude are modulated together in Drosophila
melanogaster to produce a common torque about the
yaw axis. Three-axis control must therefore be
considered an unproven possibility in Drosophila spp.
Even where three-axis control of the moments can
be proven, this does not imply that the control axes
must be identical with the main body axes or indeed
that the control axes must be strictly orthogonal
(that is, perpendicular to one another). For example,
the nominal yaw axis is tilted some 30m back from the
4
6
4
G
r
a
h
a
m
K
.
T
a
y
l
o
r
Table 1. Independent kinematic parameters (bold) available for longitudinal control in dierent insects
Stroke
trajectory
Stroke
amplitude
Wingbeat
frequency
Clap-and-peel
or clap-and ing
Supination:
speed and
timing of
rotation
Degree of
pronation
Phase of
ipsilateral
fore- and
hindwings
Diptera:
calypterate
ies
Independent
parameter;
could be used
for
longitudinal
control
Amplitude
decreased in
response to
nose-up
disturbances
Could be used
for longitudinal
control, but
likely to be
tightly linked to
amplitude
? Independent
parameter;
could be
used for
longitudinal
control
Independent
parameter;
could be
used for
longitudinal
control
Not
applicable
Diptera:
drosophilid
ies
Longitudinal
position of
stroke plane
shifted
independent
of amplitude;
stroke
inclination
apparently
linked to
amplitude
Amplitude
decreased in
response to
nose-up
disturbances;
amplitude
increases with
increasing
force
production
Frequency
increases with
increasing force
production, but
decreases again
at peak outputs,
probably
because of
physiological
constraints
Wings
approach
more closely
in response to
nose-up
disturbances;
may be
directly
linked to
amplitude?
Independent
parameter;
could be
used for
longitudinal
control
Degree of
pronation
appears not to
be varied
Not
applicable
Orthoptera:
locusts
Probably not an
important
control
parameter
Amplitude
increases with
increasing
force
production
Frequency
normally
increases with
increasing
force
production,
but may be
varied in the
opposite
direction
under
open-loop
conditions
Generally not
applicable
Probably not an
important
control
parameter
Pronation
increased in
response to
nose-up
disturbances
Likely to be
important;
eects
unknown
4
6
5
M
e
c
h
a
n
i
c
s
o
f
i
n
s
e
c
t

i
g
h
t
c
o
n
t
r
o
l
Table 2. Independent kinematic parameters (bold) available for lateral control in dierent insects
Stroke trajectory Stroke amplitude
Clap-and-peel or
clap-and ing
Supination: speed
and timing of
rotation
Degree of
pronation
Phase of ipsilateral
fore- and hindwings
Diptera:
calypterate
ies
Independent
parameter; could
be used for lateral
control.
Amplitude
decreased on
inside wing
? Supination delayed
on inside wing
Pronation
sometimes
increased on
inside wing
Not applicable
Diptera:
drosophilid
ies
Independent
parameter; could
be used for lateral
control ; stroke
inclination
apparently linked
to amplitude
Amplitude
decreased on
inside wing
Wings meet at an
angle to the
sagittal plane
during turns
Supination delayed
on inside wing;
speed of
supination
increased on
outside wing
Degree of pronation
probably not an
important control
parameter
Not applicable
Orthoptera:
locusts
Probably not an
important control
parameter
Probably not an
important control
parameter
Not usually
applicable
Probably not an
important control
parameter
Pronation
increased on
inside wing
Likely to be
important. Eects
unknown
Odonata:
dragonies
? Amplitude
decreased on
inside wing
during banked
turns
Not applicable to
dragonies
? Pronation
increased on
inside wing,
especially during
yaw turns
Likely to be
important; eects
unknown
466 Graham K. Taylor
dorso-ventral body axis in Drosophila melanogaster
(Zanker, 1988a), though whether an independent
roll axis exists or whether this represents a combined
roll-yaw axis in a two-axis system is unknown.
Orthogonality of the lateral control axes is not
essential for three-axis control, but any non-ortho-
gonality that may exist will complicate the neural
processing required to separate the various motions.
As a useful but rather imperfect analogy, consider
how dicult it is to predict the path of a bouncing
rugby ball compared to a bouncing football (it is
dicult to nd a better analogy, precisely because
manmade control systems are usually designed to act
orthogonally). The lack of separate control surfaces
to isolate the three moments actually makes it rather
likely that insect control systems will be non-
orthogonal. For example, there is no particular
reason to suppose that changes in stroke amplitude
will result in a torque about an axis orthogonal to
that due to asymmetric changes in the degree of
pronation or supination. Hence, although the con-
trol inputs of aircraft are not always strictly
orthogonal, non-orthogonality is likely to be much
more pronounced in insects and this will tend to
complicate their ight control. This is likely to be
one of the most signicant dierences between the
ight control systems of insects and conventional
aircraft. The evolved interface between mechanics
and processing in a non-orthogonal control system
should prove an especially fruitful eld for inter-
disciplinary research.
(2) Evolution of insect ight control systems
The ubiquity of postural ight control suggests that
it may be a primitive character amongst pterygote
insects. The necessary neuromuscular mechanisms
for postural control would have been in place long
before the evolution of the stroke cycle, so it would
not be surprising if this were the main means of
control during the early evolution of ight. Indeed,
if apping ight evolved via an intermediate gliding
stage, then postural control could even have been
used to stabilise the descent of smaller insects prior to
the evolution of wings (Flower, 1964). Current
developmental, neurological and morphological evi-
dence suggests that the wings themselves evolved
from gills or associated epipodal structures on the
limbs of an aquatic ancestor (Wigglesworth, 1973,
1976; Kukalova! -Peck, 1978, 1983; Wootton, 1981;
Robertson, Pearson & Reichert, 1982; Kingsolver &
Koehl, 1994; Thomas & Norberg, 1996; Averof &
Cohen, 1997). Under this scenario, the incipient
wings would already have been articulated and
muscularised at their base. However, they would
probably not have possessed the degree of control
required for successful modulation of the apping
cycle, unless their gill-like precursors were used for
swimming. Hence, the evolution of improved wing
control whether for gliding or for some other
putative intermediate stage such as surface-skim-
ming (Marden & Kramer, 1994; Marden et al.,
2000) would almost certainly have been prerequi-
site for the evolution of apping ight. Indeed, a
wing that has evolved rotations about three axes for
control already has at least a rudimentary version of
each of the fundamental motions of the stroke cycle,
oering one possible route to the evolution of the
stroke cycle (Hinton, 1963; Wigglesworth, 1963,
1976; Wootton, 1976).
It is dicult to draw any rm conclusions as to the
stage of evolution at which dierent kinematic
control mechanisms would have been acquired. It
seems likely that the earliest apping insects would
have had some control of stroke amplitude, allowing
direct modulation of the aerodynamic power output
and at least a rudimentary degree of lateral control.
Changes in the degree of wing pronation are also
likely to have been important from the outset, being
a fundamental motion of any apping wing. Changes
in stroke amplitude or the degree of wing pronation
would be expected to aect both steady and
unsteady mechanisms of force production, though
their eects could be qualitatively dierent de-
pending upon the mechanism used. For example,
increasing the stroke amplitude on one wing should
increase quasi-steady lift production, but might also
lower unsteady lift production by destabilising the
leading edge vortex associated with delayed stall
(C. P. Ellington, personal communication, cited in
Betts, 1986).
Many of the other control inputs that insects use
can only be understood in the context of unsteady
aerodynamics. This is especially true of those relating
to the speed and timing of wing rotation or to the
phase of the fore- and hindwings. To a large degree,
then, the evolution of ight control in insects must
parallel their evolutionary renement of unsteady
aerodynamic mechanisms, and dierences in the
kinematics of ight control between insect orders will
therefore reect dierences in the unsteady mech-
anisms used. For example, the reason that changes in
the speed and timing of supination are an important
control input in Drosophila spp., whereas changes in
the degree of pronation are not, is presumably that
rotational and wake-capture mechanisms are the
467 Mechanics of insect ight control
major source of unsteady force production in
Drosophila spp. On the other hand, changes in the
degree of pronation do appear to be important in
locusts, which presumably reects the importance to
locusts of translational mechanisms of aerodynamic
force production.
A corollary of this is that as our knowledge of
unsteady mechanisms and their relationship to wing
kinematics improves, it may become possible to
recognise the aerodynamic mechanisms being used
by a given insect on the basis of how the insect varies
its wing kinematics during real or ctive man-
oeuvres. This is potentially more powerful than
looking merely at the kinematics of steady ight,
since it will always be dicult to assess the
importance of translational mechanisms such as
delayed stall, which, in contrast to rotational
mechanisms, may have no clear kinematic mani-
festation. For example, the counter-intuitive re-
duction in stroke amplitude that accompanies
elevated force production in heteropteran bugs
(Betts, 1986) and hawkmoths (Willmott &Ellington,
1997) may indicate that delayed stall accounts for a
signicant portion of peak force production, es-
pecially if the wingbeat frequency is reduced
simultaneously as in bugs. Future studies of the
mechanics of insect ight control will have to take
unsteady aerodynamics explicitly into account.
Experiments correlating changes in the wing kin-
ematics with instantaneous force measurements and
ow visualisation oer one way in which this might
be achieved.
VIII. CONCLUSIONS
(1) The current literature does not permit a
formal, quantitative analysis of insect ight control,
because the aerodynamic force systems that biol-
ogists have measured have rarely been complete,
and because the position of the centre of gravity has
rarely been recorded in studies of insect ight
control. Future experimental studies will need to pay
close attention to both these points. For a full
dynamic analysis, it will be necessary also to
determine the insects moments of inertia.
(2) Two inuential paradigms in the insect ight
control literature appear either to have been widely
misinterpreted or to be a misinterpretation of the
empirical data. Specically, although Drosophila spp.
normally modulate lift and thrust together, the
helicopter model of their ight control is probably
better thought of as a facultative strategy, rather
than as a xed constraint. On the other hand, the so-
called constant-lift reaction of locusts (Wilson &
Weis-Fogh, 1962) appears not to be a reex for
maintaining constant lift at varying angles of attack,
but rather a mechanism to restore the insect to pitch
equilibrium following a disturbance.
(3) The number of degrees of freedom that insects
are able to control cannot exceed the number of
independent control inputs. Since the latter is usually
quite high, it seems likely that some of the control
inputs will be redundant, providing ner control of
the various degrees of freedom.
(4) Although most insects are likely to generate
control moments about all three axes, full three-axis
control has only been proven for larger ies and
dragonies.
(5) Control inputs are unlikely to operate even
approximately orthogonally in insects. This will tend
to complicate the neural processing required to
separate out the various motions.
(6) Taxonomic dierences in insect ight control
kinematics probably reect dierences in the un-
steady aerodynamic mechanisms being used. The
evolution of insect ight control is therefore likely to
have paralleled the evolutionary renement of
unsteady aerodynamic mechanisms.
(7) Classication of an insects control kinematics
during real or ctive turns could provide a relatively
simple assay for determining the dominant unsteady
aerodynamic mechanisms being used.
IX. ACKNOWLEDGEMENTS
I am grateful to Adrian Thomas and Robert Nudds
for their comments upon the manuscript and thank
Bob Srygley for the image of the hawkmoth. The
comments of an anonymous referee were extremely
helpful in improving the manuscript. This work was
funded by a Christopher Welch Scholarship from
the University of Oxford.
VIII. REFERENCES
Aiix.xiin, D. E. (1984). Unusual phase relationships between
the forewings and hindwings in ying dragonies. Journal of
Experimental Biology 109, 379383.
Aiix.xiin, D. E. (1986). Wind tunnel studies of turns by ying
dragonies. Journal of Experimental Biology 122, 8198.
Ann.s, E. A. (1986). Control of hindlimb posture by wind-
sensitive hairs and antennae during locust ight. Journal of
Comparative Physiology A 159, 849857.
A\inoi, M. & Conix, S. M. (1997). Evolutionary origin of
insect wings from ancestral gills. Nature 385, 627630.
468 Graham K. Taylor
Azix., A. & W.1.x.ni, T. (1988). Flight performance of a
dragony. Journal of Experimental Biology 137, 221252.
B..iin, A. (1990). The posture of the abdomen during locust
ight : regulation by steering and ventilatory interneurones.
Journal of Experimental Biology 151, 109131.
B..iin, A., Sn.$ iin, M. & Rowiii, C. H. F. (1992). The
perception of the visual ow eld by ying locusts : a
behavioural and neuronal analysis. Journal of Experimental
Biology 165, 137160.
B.iin, P. S. (1979). The wing movements of ying locusts
during steering behaviour. Journal of Comparative Physiology A
131, 4958.
B.iin, P. S., Giwiii, M. & Coo1in, R. J. (1981). The
natural ight of the migratory locust Locusta migratoria L. III.
Wing-beat frequency, ight speed and attitude. Journal of
Comparative Physiology A 141, 233237.
Bi11s, C. R. (1986). The kinematics of Heteroptera in free
ight. Journal of Zoology, London. Series B 1, 303315..
Bi11s, C. R. & Woo11ox, R. J. (1988). Wing shape and ight
behaviour in butteries (Lepidoptera: Papilionoidea and
Hesperioidea): a preliminary analysis. Journal of Experimental
Biology 138, 271288.
Bioxii.i, J. (1981). Aerodynamic capabilities of ies, as
revealed by a new technique. Journal of Experimental Biology 92,
155163.
Bn.iixninv, J. (1995). Insects in ight, London: Cassell
Biii1noii, H., Poooro, T. & Winnn.nx, C. (1980). 3-D
analysis of the ight trajectories of ies (Drosophila melano-
gaster). Zeitschrift fuWr Naturforschung C 35, 811815.
Bin1ox, A. J. (1964). Nervous control of ight orientation in a
beetle. Nature 204, 1333.
Bin1ox, A. J. (1971). Directional change in a ying beetle.
Journal of Experimental Biology 54, 575585.
C.xnr, J. M. (1970). Yaw-correcting postural changes in
locusts. Journal of Experimental Biology 52, 519531.
Cn.iwri, L. E. (1951). Stroke amplitude as a function of air
density in the ight of Drosophila. The Biological Bulletin of the
Marine Biological Laboratory, Woods Hole, Massachusetts 100,
1527.
Cioivi.i, M., Di\riiins, J.-F. & Di\izi.ix, D. (1979).
Direct measurements of instantaneous lift in desert locust ;
comparison with Jensens experiments on detached wings.
Journal of Experimental Biology 80, 115.
Coiii11, T. S. & L.xi, M. F. (1975). Visual control of ight
behaviour in the hovery, Syritta pipiens L. Journal of
Comparative Physiology 99, 166.
Cooi, M. V. (1997). Flight Dynamics : Principles, London: Arnold.
Coo1in, R. J. (1979). Visually induced yaw movements in the
ying locust, Schistocerca gregaria (Forsk.). Journal of Comparative
Physiology A 131, 6778.
D.\ri, C. T. (1978). The relationship between body angle and
ight speed in free-ying Drosophila. Physiological Entomology 3,
191195.
D.\ri, C. T. (1985). Visual control of the partition of ight
force between lift and thrust in free-ying Drosophila. Nature
313, 4850.
D.wsox, J. W., D.wsox-Siiiv, K., Ronin1, D. & Ronin1-
sox, R. M. (1997). Forewing asymmetries during auditory
avoidance in ying locusts. Journal of Experimental Biology 200,
23232335.
Drirxsox, M. H. (1994). The eects of wing rotation on
unsteady aerodynamic performance at low Reynolds
numbers. Journal of Experimental Biology 192, 179206.
Drirxsox, M. H. (1999). Haltere-mediated equilibriumreexes
of the fruit y, Drosophila melanogaster. Philosophical Transactions
of the Royal Society of London. Series B 354, 903916.
Drirxsox, M. H. & Go$ 1z, K. G. (1993). Unsteady aerody-
namic performance of model wings at low Reynolds numbers.
Journal of Experimental Biology 174, 4564.
Drirxsox, M. H. & Go$ 1z, K. G. (1996). The wake dynamics
and ight forces of the fruit y Drosophila melanogaster. Journal
of Experimental Biology 199, 20852104.
Drirxsox, M. H., Linx.xx, F. & Go$ 1z, K. G. (1993). The
active control of wing rotation by Drosophila. Journal of
Experimental Biology 182, 173189.
Drirxsox, M. H., Linx.xx, F. & Cn.x, W. P. (1998). The
control of mechanical power in insect ight. American Zoologist
38, 718728.
Drirxsox, M. H., Linx.xx, F. & S.xi, S. P. (1999). Wing
rotation and the aerodynamic basis of insect ight. Science 284,
19541960.
Diiiiv, R. (2000). The Biomechanics of Insect Flight. Princeton
University Press.
Diiiiv, R. & Eiirxo1ox, C. P. (1990a). Mechanics of forward
ight in bumblebees I. Kinematics and morphology. Journal
of Experimental Biology 148, 1952.
Diiiiv, R. & Eiirxo1ox, C. P. (1990b). Mechanics of forward
ight in bumblebees II. Quasi-steady lift and power require-
ments. Journal of Experimental Biology 148, 5388.
Diiiiv, R. & DiVnris, P. J. (1990). Flight physiology of
migrating Urania fulgens (Uraniidae) moths : kinematics and
aerodynamics of natural free ight. Journal of Comparative
Physiology A 167, 145154.
Dio.ni, J. J. (1967). Directional change in ying locusts.
Journal of Insect Physiology 13, 10551063.
Eiirxo1ox, C. P. (1984a). The aerodynamics of hovering ight.
III. Kinematics. Philosophical Transactions of the Royal Society of
London. Series B 305, 4178.
Eiirxo1ox, C. P. (1984b). The aerodynamics of hovering ight.
IV. Aerodynamic mechanisms. Philosophical Transactions of the
Royal Society of London. Series B 305, 79113.
Eiirxo1ox, C. P. (1984c). The aerodynamics of hovering ight.
VI. Lift and power requirements. Philosophical Transactions of
the Royal Society of London. Series B 305, 145181.
Eiirxo1ox, C. P., V.x iix Bino, C., Wriixo11, A. P. &
Tnox.s, A. L. R. (1996). Leading-edge vortices in insect
ight. Nature 384, 626630.
Exxos, A. R. (1989). The kinematics and aerodynamics of some
Diptera. Journal of Experimental Biology 142, 4985.
Esn, H., N.n1ro.ii, W. & Koooi, S. N. (1975). Cor-
relations between aerodynamic output, electrical activity in
the indirect ight muscles and wing positions of bees ying in
a servomechanically controlled wind tunnel. Journal of
Comparative Physiology 100, 147159.
E1irx, B. & Riri, L. D. (1996). Dynamics of Flight : Stability and
Control, New York: John Wiley & Sons, Inc.
F.is1, R. (1952). Unterschungen zum Halterenproblem.
Zoologische JahrbuWcher. Abteilung fuWr allgemeine Zoologie und
Physiologie der Tiere 63, 326366.
Frsnin, H. & Ki1sn, W. (2000). Relationships between body
mass, motor output and ight variables during free ight of
juvenile and mature adult locusts, Schistocerca gregaria. Journal
of Experimental Biology 203, 27232735.
Fiowin, J. W. (1964). On the origin of ight in insects. Journal
of Insect Physiology 10, 8188.
469 Mechanics of insect ight control
Fnriinrn, R. W., Sv.1z, H.-C. & B.isixwirx, B. (1994).
Visual control of wingbeat frequency in Drosophila. Journal of
Comparative Physiology A 175, 587596.
Gissow, A. & Axin, K. B. (1949). An introduction to the physical
aspects of helicopter stability. NACA TR 993.
Gi11niv, E. (1966). Sensory regulation of wing twisting in
locusts. Journal of Experimental Biology 44, 116.
Gi11niv, E. & Wrisox, D. M. (1964). The lift control reaction
of ying locusts. Journal of Experimental Biology 41, 183190.
Giwiii, M. (1970). Antennae: another wind-sensitive re-
ceptor in locusts. Nature 225, 12631264.
Giwiii, M. (1972). Antennen und Stirn-Scheitelhaare von
Locusta migratoria L. als Luftstro$ mungs-Sinnesorgane bei der
Flugsteuerung. Journal of Comparative Physiology 80, 5794.
Giwiii, M. (1975). The inuence of the air current sense
organs on the ight behaviour of Locusta migratoria. Journal of
Comparative Physiology 103, 7995.
Giwiii, M. & Pnrirvvix, J. (1978). Control of the horizontal
ight-course by air-current sense organs in Locusta migratoria.
Physiological Entomology 3, 4352.
Giwiii, M., Hirxzii, H.-G. & Pnrirvvix, J. (1974). Role of
antennae of the dragony Orthetrum cancellatum in ight
control. Nature 249, 584585.
Go$ 1z, K. G. (1968). Flight control in Drosophila by visual
perception of motion. Kybernetik 4, 199208.
Go$ 1z, K. G. (1987). Course-control, metabolism and wing
interference during ultralong tethered ight in Drosophila
melanogaster. Journal of Experimental Biology 128, 3546.
Go$ 1z, K. G. & W.xiii, U. (1984). Optomotor control of the
force of ight in Drosophila and Musca II. Covariance of lift
and thrust in still air. Biological Cybernetics 51, 135139.
Go$ 1z, K. G., Hixos1ixnino, B. & Brisrxoin, R. (1979).
Optomotor control of wing beat and body posture in
Drosophila. Biological Cybernetics 35, 101112.
Go\rxi, C. K. (1972). Dierential activity in the coxo-subalar
muscle during directional ight in the milkweed bug,
Oncopeltus. Canadian Journal of Zoology 50, 901905.
Go\rxi, C. K. & Bin1ox, A. J. (1970). Flight orientation in a
coreid squash bug (Heteroptera). Canadian Entomologist 102,
10021007.
Hirii, G. (1971). Die Funktion der nicht-brilla$ re Flugmuskeln
von Calliphora. Teil II: Muskula$ re Mechanismen der Flugs-
teuerung und ihre nervo$ se Kontrolle. Zoologische JahrbuWcher.
Abteilung fuWr allgemeine Zoologie und Physiologie der Tiere 76,
99137.
Hixos1ixnino, R., S.xiix.x, D. C. & Hixos1ixnino, B.
(1986). Compensatory head roll in the blowy Calliphora
during ight. Proceedings of the Royal Society of London. Series B
227, 455482.
Hixsiin, K. & Ronin1, D. (1990). Compensatory head rolling
during corrective ight steering in locusts. Journal of Com-
parative Physiology A 166, 685693.
Hrx1ox, H. E. (1963). The origin of ight in insects. Proceedings
of the Royal Entomological Society of London. Series C 28, 2425.
Hoiiri, F. S. J. (1940). The ight of the dipterous y Muscina
stabulans Falle! n. Philosophical Transactions of the Royal Society of
London. Series B 230, 357390.
Jonxsox, W. (1980). Helicopter theory. pp. 768769, New York.
Dover Publications.
K.xxin, A. E. (1971). The motor output during turning ight
in a hawkmoth Manduca sexta. Journal of Insect Physiology 17,
10731086.
K.xxin, A. E. (1985). Flying. In Comprehensive Insect Physiology,
Biochemistry and Pharmacology, Vol. V (eds. G. A. Kerkut and
L. I. Gilbert), pp. 491552. New York: Pergamon.
K.xxin, A. E. & N.n1ro.ii, W. (1973). Changing phase
relationships among motor units during ight in a saturniid
moth. Journal of Comparative Physiology 83, 1724.
Krxosoi\in, J. G. & Koini, M. A. R. (1994). Selective factors
in the evolution of insect wings. Annual Review of Entomology 39,
425451.
Kii.io\.! -Pii, J. (1978). Origin and evolution of insect
wings, and their relation to metamorphosis as documented by
the fossil record. Journal of Morphology 156, 53125.
Kii.io\.! -Pii, J. (1983). Origin of the insect wing and wing
articulation from the arthropodan leg. Canadian Journal of
Zoology 61, 16181669.
Ki1sn, W. & Giwiii, M. (1979). Development of ight
behaviour in maturing adults of Locusta migratoria: II.
Aerodynamic parameters. Journal of Insect Physiology 25,
299304.
L.xi, M. F. (1993). The visual control of courtship behaviour
in the y Poecilobothrus nobilitatus. Journal of Comparative
Physiology A 173, 595603.
Linx.xx, F. & Drirxsox, M. H. (1997). The changes in
power requirements and muscle eciency during elevated
force production in the fruit y Drosophila melanogaster. Journal
of Experimental Biology 200, 11331143.
Linx.xx, F. & Drirxsox, M. H. (2001). The production of
elevated ight force compromises manoeuvrability in the fruit
y Drosophila melanogaster. Journal of Experimental Biology 204,
627635.
Lron1nrii, J. (1974). Aerodynamic aspects of animal ight. In
Swimming and Flying in Nature, Vol. 2 (eds. T. Y. Wu, C. J.
Brokaw and C. Brennen), pp. 423491. New York: Plenum
Press.
Loniz, M. (1995). Neural control of hindleg steering in ight in
the locust. Journal of Experimental Biology 198, 869875.
M.niix, J. H. & Kn.xin, M. G. (1994). Surface-skimming
stoneies a possible intermediate stage in insect ight
evolution? Science 266, 427430.
M.niix, J. H., ODoxxiii, B. C., Tnox.s, M. A. & Bvi,
J. Y. (2000). Surface-skimming stoneies and mayies : the
taxonomic and mechanical diversity of two-dimensional
aerodynamic locomotion. Physiological and Biochemical Zoology
73, 751764.
M.v, M. L., Bnoiiiinnin, P. D. & Hov, R. R. (1988). Kine-
matic and aerodynamic aspects of ultrasound-induced nega-
tive phonotaxis in ying australian eld crickets (Teleogryllus
oceanicus). Journal of Comparative Physiology A 164, 243249.
M.v, M. L. & Hov, R. R. (1990a). Leg-induced steering in
ying crickets. Journal of Experimental Biology 151, 485488.
M.v, M. L. & Hov, R. R. (1990b). Ultrasound-induced yaw
movements in the ying Australian eld cricket (Teleogryllus
oceanicus). Journal of Experimental Biology 149, 177189.
Mriis, C. I., M.v, M. L., Hoinnooi, E. H. & Hov, R. R.
(1992). Multisegmental analysis of acoustic startle in the
ying cricket (Teleogryllus oceanicus): kinematics and electro-
myography. Journal of Experimental Biology 169, 1936.
Mo$ ni, B. & Z.nx.i, W. (1975). Flugsteuerung der Wander-
heuschrecke durch Verschiebung der Muskelaktivita$ t. Natur-
wissenschaften 62, 441442.
Mo$ ni, B. & Z.nx.i, W. (1977). Activity of the direct
downstroke ight muscles of Locusta migratoria (L.) during
470 Graham K. Taylor
steering behaviour in ight. II. Dynamics of the time shift and
changes in the burst length. Journal of Comparative Physiology A
118, 235247.
Morsiii, A., Poii.i, G. S. & Hov, R. R. (1978). Steering
responses of ying crickets to sound and ultrasound. Proceedings
of the National Academy of Sciences of the United States of America
75, 40524056.
N.n1ro.ii, W. & Ro1n, W. (1983). Correlations between
stationary measurable parameters of wing movement and
aerodynamic force production in the blowy (Calliphora vicina
R.-D.). Journal of Comparative Physiology A 150, 251260.
N.n1ro.ii, W. & Wrisox, D. M. (1967). Neuro-muscular
control of dipteran ight. Journal of Experimental Biology 47,
7797.
N.n1ro.ii, W., Wrix.xx, R. & Rixxin, M. (1971). U= ber
den ortsfesten freien Flug von Bienen in einem Saugkanal.
Apidologie 2, 271282.
N.in.n, G. (1989). The gear change mechanism of the blowy
(Calliphora erythrocephala) in tethered ight. Journal of Com-
parative Physiology A 165, 321331.
N.in.n, G. & Hixos1ixnino, R. (1994). The halteres of the
blowy Calliphora. II. Three-dimensional organization of
compensatory reactions to real and simulated rotations.
Journal of Comparative Physiology A 175, 695708.
Niisox, R. C. (1989). Flight Stability and Automatic Control,
Singapore: McGraw-Hill.
Nonnino, R. AH . (1975). Hovering ight of the dragony Aeschna
juncea L. Kinematics and aerodynamics. In Swimming and
Flying in Nature. Vol. 2 (eds. T. Y. Wu, C. T. Brokaw and C.
Brennen), pp. 763781, New York: Plenum Press.
On.n., Y. (1975). Mating behaviour of the cabbage white
buttery Pieris rapae crucivora. VI. Electrophysiological de-
cision of muscle functions in wing and abdomen movements
and muscle output patterns during ight. Journal of Comparative
Physiology 102, 189200.
P.iiriii, G. D. (1996). Helicopter Flight Dynamics, Oxford:
Blackwell Science Ltd.
Prx, W., N.in.n, G. & Ziri, J. (1993). Strepsipteran fore-
wings are haltere-like organs of equilibrium. Naturwissen-
schaften 80, 371374.
Poii.i, G. S. & Hov, R. R. (1981). Phonotaxis in ying
crickets : neural correlates. Journal of Insect Physiology 27,
4145.
Poii.i, G. S., Hinin, F. & Winin, T. (1984). Frequency
and temporal pattern-dependent phonotaxis of crickets
(Teleogryllus oceanicus) during tethered ight and compensated
walking. Journal of Comparative Physiology A 154, 1326.
Pnirss, R. & Giwiii, M. (1991). Compensation of visually
simulated wind drift in the swarming ight of the desert locust
(Schistocerca gregaria). Journal of Experimental Biology 157,
461481.
Pnirss, R. & Svoni, P. (1993). Flight-phase and visual-eld
related optomotor yaw responses in gregarious desert locusts
during tethered ight. Journal of Comparative Physiology A 172,
733740.
Ronin1, D. & Rowiii, C. H. F. (1992a). Locust ight steer-
ing. I. Head movements and the organization of correctional
manoeuvres. Journal of Comparative Physiology A 171, 4151.
Ronin1, D. & Rowiii, C. H. F. (1992b). Locust ight steer-
ing. II. Acoustic manoeuvres and associated head movements
compared with correctional steering. Journal of Comparative
Physiology A 171, 5362.
Ronin1sox, R. M. & Jonxsox, A. G. (1993). Collision avoid-
ance of ying locusts : steering torques and behaviour. Journal
of Experimental Biology 183, 3560.
Ronin1sox, R. M. & Rivi, D. N. (1992). Wing movements
associated with collision-avoidance manoeuvres during ight
in the locust Locusta migratoria. Journal of Experimental Biology
163, 231258.
Ronin1sox, R. M., Kinxin1, C. T. & D.wsox, J. W. (1996).
Thermal avoidance during ight in the locust Locusta
migratoria. Journal of Experimental Biology 199, 13831393.
Ronin1sox, R. M., Pi.nsox, K. G. & Rirnin1, H. (1982).
Flight interneurons in the locust and the origin of insect wings.
Science 217, 177179.
Roiiin, K. D. (1967). Turning tendency of moths exposed to
ultrasound while in stationary ight. Journal of Insect Physiology
13, 873888.
Rowiii, C. H. F. (1988). Mechanisms of ight steering in
locusts. Experientia 44, 389395.
Ri$ vviii, G. (1989). Kinematic analysis of symmetrical ight
manoeuvres of Odonata. Journal of Experimental Biology 144,
1342.
S.xiix.x, D. C. (1980). Angular acceleration, compensatory
head movements and the halteres of ies (Lucilia serricata).
Journal of Comparative Physiology A 136, 361367.
S.1o, M. & Azix., A. (1997). The ight performance of a
damsely Ceriagrion melanurum Selys. Journal of Experimental
Biology 200, 17651779.
Snris1n., C. & V.x H.1inix, J. H. (1999). Blowy ight
and optic ow. I. Thorax kinematics and ight dynamics.
Journal of Experimental Biology 202, 14811490.
Snxri1, J. & Z.nx.i, W. (1987). The motor pattern of
locusts during visually induced rolling in long-term ight.
Biological Cybernetics 56, 397410.
Snxiriin, P. & Kn.$ xin, B. (1974). Die steuerung des Fluges
beim Sandlaufka$ fer (Cicindela) und beim Maika$ fer (Melo-
lontha). Journal of Comparative Physiology 91, 377386.
Sniizi, W. & Snii, J. (2001). Ultrasound avoidance be-
haviour in the bushcricket Tettigonia viridissima (Orthoptera:
Tettigoniidae). Journal of Experimental Biology 204, 733740.
Snwixxi, T. & Z.nx.i, W. (1987). Movements of the
hindwings of Locusta migratoria, measured with miniature coils.
Journal of Comparative Physiology A 160, 657666.
Snrxr\.s.x, M. V. (1977). A visually-evoked roll response in the
housey. Open-loop and closed-loop studies. Journal of Com-
parative Physiology A 119, 114.
Snvoiiv, R. B. (1994). Locomotor mimicry in butteries the
associations of positions of centers of mass among groups of
mimetic, unprotable prey. Philosophical Transactions of the
Royal Society of London. Series B 343, 145155.
Snvoiiv, R. B. (1999). Locomotor mimicry in Heliconius
butteries : contrast analyses of ight morphology and
kinematics. Philosophical Transactions of the Royal Society of
London. Series B 354, 203214.
Snvoiiv, R. B. & Diiiiv, R. (1993). Correlations of the
position of center of body mass with buttery escape tactics.
Journal of Experimental Biology 174, 155166.
Snvoiiv, R. B. & Krxosoi\in, J. G. (2000). Eects of weight
loading on ight performance and survival of palatable
Neotropical Anartia fatima butteries. Biological Journal of the
Linnean Society 70, 707725.
S1iiiw..o, F. (1916). Wie steuern die Insekten wuahrend des
Fluges. Biologisches Zentralblatt 36, 3044.
471 Mechanics of insect ight control
T.vion, C. P. (1981a). Contribution of compound eyes and
ocelli to steering of locusts in ight. I. Behavioural analysis.
Journal of Experimental Biology 93, 118.
T.vion, C. P. (1981b). Contribution of compound eyes and
ocelli to steering of locusts in ight. II. Timing changes in
ight motor units. Journal of Experimental Biology 93, 1931.
Tnox.s, A. L. R. & Nonnino, R. AH . (1996). Skimming the
surface The origin of ight in insects? Trends in Ecology and
Evolution 11, 187188.
Tni$ nrxo, D. A. (1986). Variability of motor output during
ight steering in locusts. Journal of Comparative Physiology A 158,
653664.
Ti, M. S. & Drirxsox, M. H. (1996). The control of wing
kinematics by two steering muscles of the blowy (Calliphora
vicina). Journal of Comparative Physiology A 178, 813830.
Vrxn, N. X. (1993). Flight Mechanics of High-performance Aircraft,
Cambridge: Cambridge University Press.
Vooii, S. (1966). Flight in Drosophila. I. Flight performance of
tethered ies. Journal of Experimental Biology 44, 567578.
Vooii, S. (1967). Flight in Drosophila. II. Variations in stroke
parameters and wing contour. Journal of Experimental Biology
46, 383392.
W.oxin, H. (1986). Flight performance and visual control of
the free-ying housey (Musca domestica L.) I. Organization of
the ight motor. Philosophical Transactions of the Royal Society of
London. Series B 312, 527551.
W.iix.xx, B. & Z.nx.i, W. (1988). Forewing movements
and motor activity during roll manouevers in ying desert
locusts. Biological Cybernetics 59, 325335.
W.iinox, I. (1967). Neural mechanism by which controlling
inputs inuence motor outputs in the ying locust. Journal of
Experimental Biology 47, 213228.
W.xo, S. & Ronin1sox, R. M. (1988). Changes of the hind-
wing motor pattern associated with phonotactic steering
during ight in the cricket, Teleogryllus oceanicus. Journal of
Comparative Physiology A 164, 219229.
W.iiirxo, J. M. & Eiirxo1ox, C. P. (1997). Dragony ight.
II. Velocities, accelerations and kinematics of apping ight.
Journal of Experimental Biology 200, 557582.
Wirs-Foon, T. (1956a). Biology and physics of locust ight. II.
Flight performance of the desert locust (Schistocerca gregaria).
Philosophical Transactions of the Royal Society of London. Series B
239, 459510.
Wirs-Foon, T. (1956b). Biology and physics of locust ight. IV.
Notes on sensory mechanisms in locust ight. Philosophical
Transactions of the Royal Society of London. Series B 239, 553584.
Wirs-Foon, T. (1973). Quick estimates of ight tness in
hovering animals, including novel mechanisms for lift pro-
duction. Journal of Experimental Biology 59, 169230.
Wrooiiswon1n, V. B. (1963). The origin of ight in insects.
Proceedings of the Royal Entomological Society of London. Series C 28,
2324.
Wrooiiswon1n, V. B. (1973). Evolution of insect wings and
ight. Nature 246, 127129.
Wrooiiswon1n, V. B. (1976). The evolution of insect ight.
Symposia of the Royal Entomological Society of London 7, 255269.
Wriiinsox, R. C. & Bi1iin, J. F. (1984). The Immelmann
turn: a pursuit maneuver used by hovering male Hybomitra
hinei wrighti (Diptera: Tabanidae). Annals of the Entomological
Society of America 77, 293295.
Wriirx, P. J. (1990). The instantaneous force on a desert locust,
Schistocerca gregaria (Orthoptera: Acrididae), ying in a wind
tunnel. Journal of the Kansas Entomological Society 63, 316328.
Wriirx, P. J. & Wriir.xs, M. H. (1993). Comparison of the
aerodynamic forces on a ying sphingid moth with those
predicted by quasi-steady theory. Physiological Zoology 66,
10151044.
Wriixo11, A. P. & Eiirxo1ox, C. P. (1997). The mechanics of
ight in the hawkmoth Manduca sexta. I. Kinematics of
hovering and forward ight. Journal of Experimental Biology
200, 27052722.
Wriixo11, A. P., Eiirxo1ox, C. P. & Tnox.s, A. L. R.
(1997). Flow visualization and unsteady aerodynamics in the
ight of the hawkmoth, Manduca sexta. Philosophical Transactions
of the Royal Society of London. Series B 352, 303316.
Wrisox, D. M. & Wirs-Foon, T. (1962). Patterned activity of
co-ordinated motor units, studied in ying locusts. Journal of
Experimental Biology 39, 643667.
Wrooiiswon1n, V. B. (1976). The fossil record and insect ight.
Symposia of the Royal Entomological Society of London 7, 235254.
Woo11ox, R. J. (1981). Palaeozoic insects. Annual Review of
Entomology 26, 319344.
Won1x.xx, M. & Z.nx.i, W. (1993). Wing movements and
lift regulation in the ight of desert locusts. Journal of
Experimental Biology 182, 5769.
Z.xiin, J. M. (1988a). How does lateral abdomen deection
contribute to ight control of Drosophila melanogaster? Journal of
Comparative Physiology A 162, 581588.
Z.xiin, J. M. (1988b). On the mechanism of speed and
altitude control in Drosophila melanogaster. Physiological En-
tomology 13, 351361.
Z.xiin, J. M. (1990). The wing beat of Drosophila melanogaster.
III. Control. Philosophical Transactions of the Royal Society of
London. Series B 327, 4564.
Z.xiin, J. M. & Go$ 1z, K. G. (1990). The wing beat of
Drosophila melanogaster. II. Dynamics. Philosophical Transactions
of the Royal Society of London. Series B 327, 1944.
Z.xiin, J. M., Eoiin..i, M. & W.nzin., A. K. (1991). On
the coordination of motor output during visual ight control
of ies. Journal of Comparative Physiology A 169, 127134.
Z.nx.i, W. (1982). Kinematische, aerodynamische und neuro-
physiologisch-morphologische Untersuchungen des Heuschreckenugs,
Habilitationschrift. Universita$ t Go$ ttingen.
Z.nx.i, W. (1983). Untersuchungen zum Flug von Wander-
heuschrecken. Die Bewegungen, ra$ umlichen Lagebezie-
hungen sowie Formen und Prole von Vorder- und Hinter-
u$ geln. In BIONA Report 1, pp. 79102 (ed. W. Nachtigall).
Stuttgart : Fischer.
Z.nx.i, W. (1988). The eect of forewing depressor activity
on wing movement during locust ight. Biological Cybernetics
59, 5570.
Z.nx.i, W. & Mo$ ni, B. (1977). Activity of the direct
downstroke ight muscles of Locusta migratoria (L.) during
steering behaviour in ight. I. Patterns of time shift. Journal of
Comparative Physiology A 118, 215233.
Z.nx.i, W. & Won1x.xx, M. (1989). On the so-called
constant-lift reaction of migratory locusts. Journal of Ex-
perimental Biology 147, 111124.

You might also like