The Biomechanics of Insect Flight: Form, Function, Evolution 9780691186344, 0691094918, 0691044309

Citation preview

THE

BIOMECHANICS

OF I N S E C T

FLIGHT

D R A G O N F L Y T H O R A C I C A N A T O M Y A N D W I N G CROSS SECTIONS, F R O M T H E FIRST B O O K (1822) O N INSECT F L I G H T (J. C H A B R I E R : ESSAI SUR LE VOL INSECTES

ET OBSERVATIONS

SUR QUELQUES

PARUES

DE LA MÉCANIQUE

DES MOUVEMENS

PROGRESSIFS

DE L'HOMME

ET DES ANIMAUX

DES

VERTÉBRÉS).

THE BIOMECHANICS OF INSECT FLIGHT FORM,

FUNCTION,

EVOLUTION

Robert Dudley

P R I N C E T O N

P R I N C E T O N ,

U N I V E R S I T Y

NEW

J E R S E Y

PRESS

Copyright © 2000 by Princeton University Press Published by Princeton University Press, 41 William Street, Princeton, New Jersey 08540 In the United Kingdom: Princeton University Press, 3 Market Place, Woodstock, Oxfordshire OX20 1SY A l l Rights Reserved Second printing, and first paperback printing, 2002 Paperback ISBN 0-691-09491-8 The Library of Congress has cataloged the cloth edition of this book as follows Dudley, Robert. The biomechanics of insect flight : form, function, evolution / Robert Dudley p.

cm.

Includes bibliographical references ( p . ) and index. ISBN 0-691-04430-9 (cloth : alk. paper) 1. Insects—Flight. QL496.7.D83

I. Title.

1999

573.7'98157—dc21

99-29653

British Library Cataloging-in-Publication Data is available This book has been composed in Palatino Printed on acid-free paper. ° o www.pupress.princeton.edu Printed in the United States of America 10

9 8 7 6 5 4 3 2

For my wife, Lu Min

CONTENTS

ACKNOWLEDGMENTS

ix

SYMBOLS

xi

CHAPTER ONE

Flight and the Pterygote Insecta 1.1 1.2 1.3 1.4

Insect Diversity Basic Aerodynamics Why Study Flight Biomechanics? Summary

3 5 15 29 34

CHAPTER TWO

Morphology of the Flight Apparatus 2.1 2.2 2.3 2.4

Thoracic Design Wings Ancillary Structures Summary

36 36 52 72 74

CHAPTER THREE

Kinematics and Aerodynamics of Flight 3.1 3.2 3.3 3.4

Wing and Body Motions Aerodynamics Mechanical Power Requirements Summary

75 75 105 144 157

CHAPTER FOUR

Energetics and Flight Physiology 4.1 4.2 4.3 4.4

Oxygen Consumption Muscle Physiology Thermoregulation in Flight Summary

159 159 172 196 202

CHAPTER FIVE

Stability, Maneuverability, and M a x i m u m Flight Performance 5.1 5.2 5.3 5.4 5.5

Stability Maneuverability Three-Dimensional Flight Behavior Limits to Insect Flight Performance Summary

203 203 222 233 242 259

Viii

CONTENTS

C H A P T E R SIX

Evolution of Flight and Flightlessness 6.1 Origin of Flight in Hexapods 6.2 Evolutionary Consequences of Flight and Flightlessness 6.3 Summary

261 261 291 300

CHAPTER SEVEN

Flight and Insect Diversification 7.1 7.2 7.3 7A 7.5 7.6

Miniaturization Pollination Prédation Long-Range Dispersal and Migration Comparison of Insect and Vertebrate Flight Summary

302 302 309 313 322 331 336

CHAPTER EIGHT

Future Directions i n Insect Flight Biomechanics 8.1 Aerodynamic Mechanisms 8.2 Insect Flight Biomechanics in Nature 8.3 Exploring Insect Diversity

338 338 341 347

GLOSSARY

353

REFERENCES

361

INDEX

465

ACKNOWLEDGMENTS

T

H E PRACTICE of science, like other human endeavors, relies upon cultural transmission of information. I am particularly i n ­ debted to m y parents, Bettina Dudley and the late Ted Dudley (botaniste extraordinaire), for regular exposure to diverse features of natural history and geographical exploration. I thank Geerat Vermeij for an early introduction to concepts of biogeography and evolution­ ary biology, and Pierre Sprey for his inspirational roles as aerodynamicist and connoisseur. Steve Vogel provided wonderfully riotous i n ­ struction and incitement d u r i n g m y undergraduate tenure at Duke University. Peter Klopfer, Horst Meyer, and Steve Wainwright at the same institution were also generous w i t h their advice and expertise. I thank Charlie Ellington at the University of Cambridge for his tute­ lage d u r i n g the course of m y graduate studies. A t the Smithsonian Tropical Research Institute i n Panama, I have benefited substantially from collaborations w i t h Greg Adler and Stan Rand; numerous dis­ cussions w i t h Egbert Leigh and the late A l a n Smith have also been i n ­ formative. I am deeply indebted to Carl Gans for teaching me electro­ myography and for providing scholarly i f not rabbinical perspectives on diverse issues; our continuing interactions invariably convince me of the primacy of the organism i n evolutionary biology. Ongoing sci­ entific collaborations w i t h Peng Chai, Phil DeVries, Evandro Oliveira, and Bob Srygley encourage me to consider animal flight performance i n natural environments, the ultimate testing ground for biomechanical design.

Numerous colleagues have commented either on particular chap­ ters (chapter 3: Charlie Ellington, Hao L i u , Sandy Willmott; chapter 4: Tim Casey, Bob Full, Bob Josephson, Raul Suarez; chapter 5: Jim Marden; chapter 6: Conrad Labandeira, Jim Marden, Riley Nelson, Stan Rand, Geerat Vermeij, M a r y Jane West-Eberhard; chapter 7: Phil DeVries, Bill Eberhard, Larry Gilbert, Evandro Oliveira, Bob Srygley, Geerat Vermeij, M a r y Jane West-Eberhard), or on the entire book (Greg Adler, D o u g Altshuler, Claire Balint, Peng Chai, Wai Pang Chan, M a r k Denny, Michael Dickinson, Erica Feuerbacher, Carl Gans, D m i t r y Grodnitsky, Jon Harrison, Rebecca Johnston, Kate Loudon, Don Pick, Steve Roberts, Sanjay Sane, Jocelyn Staunton, Swifty Steven­ son, Steve Vogel, Robin Wootton, Diana W u , and Lijiang Zeng). I am grateful for their constructive advice and criticism. Gwen Gage and Kristina Schlegel helped extensively w i t h illustrations; D m i t r y Grod­ nitsky and Vadim Rossman have kindly assisted w i t h the procure-

X

ACKNOWLEDGMENTS

ment of Russian articles and w i t h their translation. Publication of the color plates was enabled by a University Cooperative Society Subven­ tion Grant awarded b y the University of Texas at Austin. I n an era of unprecedented ecosystem destruction and species ex­ tinction, the scarcity of research funds for organismal biology stands as a powerful indictment of governmental funding priorities. I am, however, extremely grateful to diverse institutions that have enabled m y research i n flight biomechanics. The British Marshall Commission k i n d l y provided financial support for doctoral studies on insect flight, whereas the Smithsonian Tropical Research Institute and its d i ­ rector, Ira Rubinoff, have generously supported m y postdoctoral and ongoing research i n the Republic of Panama. I feel particularly fortu­ nate to have had the unusual and interesting opportunity to live on Barro Colorado Island for five years. Throughout m y tenure at the University of Texas, Larry Gilbert, together w i t h intramural Reeder fellowships, have provided substantial logistical assistance. Financial support from the National Geographic Society has enabled various flight projects that lie beyond the pale of traditional funding. It can only be hoped that comparable research opportunities, particularly i n rapidly disappearing tropical ecosystems, w i l l not be diminished for future students.

SYMBOLS

a Ao c c(r)

C

D

CD, pro Q CL, max D F Fvert g i ) I L ^/^max ^span m n Pw P Pacc aero cm Pind ^muscle P p par per P Ppro zero R Re ^body wing 1

1 1

L

1

1

1

1VC

acceleration area of actuator disk aspect ratio w i n g chord w i n g chord at radial distance r coefficient of drag profile drag coefficient coefficient of lift m a x i m u m lift coefficient drag force drag force on body aerodynamic force net vertical forces produced by w i n g flapping gravitational acceleration moment of inertia of the w i n g mass and virtual mass advance ratio characteristic dimension of w i n g or body lift force m a x i m u m lift : drag ratio lift force on body lift force per unit w i n g span body mass; median sclerite wingbeat frequency w i n g loading inertial power i n first half of a half-stroke aerodynamic power requirements mechanical power to oscillate the center of body mass induced power power output of the flight muscle parasite power power requirements assuming perfect storage of inertial energy profile power power requirements assuming zero storage of inertial energy w i n g length Reynolds number Reynolds number for the body Reynolds number for the w i n g chord

XU

SYMBOLS

s V V ^down v¡ V (r) v Vvert Vx, Vy, V

reference area of w i n g or body; total w i n g area virtual w i n g mass relative fluid velocity; insect airspeed mean w i n g velocity during downstroke induced velocity w i n g relative velocity at radial distance r mean w i n g velocity d u r i n g upstroke vertical component of the body velocity velocity component i n the (x), (y) or (z) dimension, respectively

a ß r e 0 ^horizontal ^vertical

angle of attack stroke plane angle circulation muscle strain w i n g elevational angle; glide angle horizontal error angle vertical error angle dynamic viscosity kinematic viscosity air density; density of muscle myofibrillar stress positional angle of the w i n g m a x i m u m positional angle m i n i m u m positional angle mean positional angle m a x i m u m angular velocity of w i n g during a half-stroke stroke amplitude body angle

R

up

z

jU V

p o

10 ) exhibit highly tur­ bulent and inertially driven flows; viscous effects are much reduced i n importance. The product of characteristic dimension and velocity 3

4

6

CHAPTER ONE

18 TABLE

1.2

Representative Values of the Reynolds Number for the Body (Re^dy) and for the Wing Chord (Re in ) for Species of Body Mass m during Flight at a Forward Velocity V, or While Hovering W

g

Order

Genus and Species

Odonata Orthoptera Hymenoptera

Anax junius Locusta migratoria Bombus terrestris

Hymenoptera

Encarsia formosa Encarsia formosa Urania fulgens Drosophila virilism

a

0

0

0

m (mg) 900 1500 175

0.025

6

Lepidoptera Diptera

í

350 2

V (m/s)

Rebody

Rewing

7.5 4.6 (hovering) 1 2.5 4.5 (hovering) (hovering) 3.4 2

33,000 12,300 n/a 1240 3100 5580 n/a n/a 4480 300

10,500 2500 1210 1360 1940 2990 15-18 0.07 5270 350

Maximum speed in nature (Rüppell, 1989; additional data from May, 1991). Value for Re m is based on the forewing chord and assumes that the stroke amplitude and the stroke plane angle equal 90°. Flight in migratory swarm (Baker et al., 1981; additional data from Faure, 1932, and Baker and Cooter, 1979a). Value for Re ing is based on the forewing chord and assumes a stroke amplitude of 100° and a stroke plane angle of 70°. Bumblebee worker flying freely in jet of w i n d tunnel (Dudley and Ellington, 1990a). Hovering flight in laboratory (Weis-Fogh, 1973). Re ing based on wing setal diameter (Ellington, 1975). Migratory flight in nature (Dudley and DeVries, 1990). § Tethered in jet of wind tunnel (Vogel, 1966). a

W

b

W

c

d

e

W

f

determines Re, however, so that even large structures m o v i n g at very l o w speeds (e.g., an airplane m o v i n g at 10~ m / s ) experience almost exclusively viscous forces i n the boundary layer surrounding their ex­ ternal surface area. Most importantly, different situations of fluid flow are physically equivalent i f the corresponding Re's are approximately the same. That is, patterns of fluid flow at the same Re axe dynamically equivalent and are independent of the physical context of motion. This important conclusion enables scaled physical models (attained b y varying /), physically variable fluids (varying p and / / ) , and different flow velocities to be employed i n dynamic reconstructions of other­ wise experimentally intractable flow situations (Vogel, 1994). Operationally, flying insects can be characterized by t w o distinct but correlated estimates of Re, one for the wings and one for the body (table 1.2). Selection of the characteristic dimension used i n calculation of the Re is somewhat arbitrary, but this choice is more readily under­ stood i n light of the comparative and not absolute nature of the Re. Body length serves as the characteristic dimension for the Re of the 9

g

FLIGHT AND THE PTERYGOTE

19

INSECTA

body i n forward flight, whereas the mean w i n g chord (the mean dis­ tance between the leading and trailing edge of the w i n g , calculated as the w i n g area d i v i d e d b y the w i n g length) serves as the relevant quan­ tity for the Re of wings. I n general, both wings and bodies of larger insects operate at higher Re and are positively correlated, but this rela­ tionship is typically nonlinear and varies w i t h relative w i n g size, wingbeat kinematics, and flight speed. M a n y studies of insect flight have w o r k e d w i t h fairly large taxa operating i n an Re range of 10 -10 (table 1.2). This range is generally understood to be characterized mostly by inertially driven flows (Vogel, 1994). Because most insect species are only 4-5 m m i n body length (section 7.1), however, the Re for flying insects most typically falls w i t h i n the less-studied range of 10°-10 (section 7.1.3). A t such l o w Re, flow is essentially viscous i n character, and locomotor mechanisms used at higher Re (e.g., circu­ lation-based lift; see below) are much less effective. Finally, flight i n variable air temperatures may, i n addition to potential thermal effects on muscle performance (section 4.3.1), also influence flight aerody­ namics through correlated variation not only i n air density and viscos­ ity but also, b y implication, i n the Re of the wings and body. The total fluid drag D acting on an object represents the summed effects of pressure and viscous drag, and can be measured empirically for different shapes and sizes using force transducers mounted i n ei­ ther flow tanks or w i n d tunnels. Comparison of the drag of different objects (i.e., the relative extent of streamlining) is facilitated through definition of a nondimensional drag coefficient C such that: 2

4

2

D

c

° = ^ '

( i

-

2 )

where p is the fluid density, S a reference area of the object (most t y p i ­ cally the cross-sectional area), and V the velocity of fluid relative to object (Vogel, 1994). Drag coefficients are not, i n general, constant for any given object, but vary substantially both w i t h Re and w i t h object orientation relative to flow (section 3.2.1). This variation arises i n part from the interaction between turbulent and viscous flows around the object as flow speed changes, and i n part from the differential depen­ dence of inertial and viscous drag on object dimensions. Accordingly, extrapolation of drag coefficient data beyond the Re range of empirical validation is not warranted for most biological structures. Drag coeffi­ cients of both insect wings and bodies do, however, tend to decline at higher Re (section 3.2.1). Also, equation (1.2) refers only to motion at constant velocity. For objects accelerating w i t h i n fluids, a force supple­ mental to body drag is also required to accelerate the added or virtual mass of surrounding air entrained by the object's motion (Denny,

20

CHAPTER ONE

1993). Because of the l o w density of air and the small size of most i n ­ sects, however, forces associated w i t h acceleration of this added mass are likely to be insignificant i n comparison w i t h viscous and pressure drag. Drag forces on the bodies of flying animals have been classically termed parasite drag, signifying the drag supplemental to that associ­ ated w i t h force production b y the flapping wings. For bilaterally sym­ metric insect bodies oriented directly into flow, the dextral (right) and sinistral (left) components of parasite drag offset one another and the total drag imposes a net backward force. Insects flying w i t h the body not oriented i n the direction of motion, however, experience a supple­ mental component of drag force that is not parallel to the insect's body and that w i l l impose a lateral deceleration i n absence of compensation by the wings. Similarly, the drag force on a w i n g i n two-dimensional flow (i.e., w i t h no spanwise movement of air toward the w i n g tip) is termed the profile drag and acts parallel to the relative w i n d velocity, tending to decelerate the w i n g . Profile drag can be further represented as the sum of pressure drag (associated w i t h flow separation and mo­ mentum losses i n the wake of the wing) and of skin friction (derived from shear forces w i t h i n the boundary layer on both dorsal and ven­ tral w i n g surfaces). The mechanical power required to overcome both profile drag on the wings and parasite drag on the body is a major component of the total power expenditure i n flapping flight. Drag co­ efficients for both insect wings and bodies have correspondingly been the subject of empirical investigation (section 3.2.1). 2.2.3 Lift Forces Just as drag is the force parallel to flow, lift is defined as the compo­ nent of force orthogonal to flow and thus perpendicular to drag i n a two-dimensional perspective. A l l objects positioned asymmetrically i n m o v i n g fluids experience lift as w e l l as drag, but the magnitude of lift forces thus generated varies considerably w i t h object geometry. For most objects i n biology, lift forces are small relative to drag. Structures that, on the contrary, produce h i g h lift relative to drag are termed air­ foils i f they are technological i n character; wings are simply biotic air­ foils. High-lift structures such as wings represent one extreme of the lift-drag continuum, whereas the other extreme is exemplified b y highdrag and low-lift objects such as insect bodies. Note, however, that the distinction between lift and drag is purely conceptual, and that aero­ dynamic forces can be equivalently viewed as a single net force vector of variable magnitude and direction acting on insect wings and bodies (see Dickinson, 1996). Force components perpendicular to airflow are

FLIGHT AND THE PTERYGOTE INSECTA

21

FIG. 1.2. (A) Airflow over a wing moving at relative velocity V. The angle of attack a indicates the orientation of the wing chord (line segment connecting the leading and trailing edges) with respect to the relative air velocity. (B) Two-dimensional perspective of circulation r around a translating wing. Cir­ culation of magnitude equal to that of the bound vortex but of opposite sign is shed at the start of wing translation. somewhat counterintuitive, however, and a brief description of the physical mechanisms underlying lift production is necessary to ap­ preciate the nature of the aerodynamic forces generated b y flapping wings. A simplified representation of a flapping insect w i n g is to consider a cambered w i n g momentarily m o v i n g at a fixed orientation and con­ stant speed relative to the surrounding air (fig. 1.2A). This condition can be termed steady-state flow, as opposed to unsteady conditions characterized by changes over short time intervals i n w i n g speed, geometry, and orientation. A t l o w angles of attack under steady-state conditions, air flows smoothly over both dorsal and ventral w i n g sur­ faces. Because pressure drag of well-designed airfoils is low, momen­ t u m extraction from the m o v i n g fluid is minimized and the dorsal airstream merges w i t h that over the ventral surface near the trailing

22

CHAPTER ONE

edge of the airfoil. Shear stress w i t h i n the boundary layer at the trail­ ing edge prevents the dorsal airstream from flowing directly to the trailing edge (with a necessarily ventral component of velocity) and from instantaneously rejoining the mainstream flow that is orthogonal to the dorso-ventral axis of the w i n g . Instead, the dorsal airflow sepa­ rates from the w i n g slightly anterior to the trailing edge, whereas the ventral airstream diverges from the w i n g precisely at the trailing edge. A small turbulent wake reflecting the slightly premature departure of the dorsal airstream is thus present even at very l o w angles of attack relative to oncoming flow. The positive camber of the airfoil and the merging of dorsal and ventral airstreams near the trailing w i n g edge yield slightly different translational velocities for the t w o airstreams—airflow is slightly faster above and slower beneath the airfoil. Bernoulli's Principle i n d i ­ cates that this difference i n dorsal and ventral airstream velocities re­ sults i n a pressure gradient and net lift directed across the surface area of the w i n g . The difference i n translational velocities of the dorsal and ventral airstreams is physically equivalent to net movement of air from the ventral w i n g surface around the leading edge and then over the dorsal surface of the w i n g . I n a two-dimensional perspective, then, air appears to move anteriorly from the ventral to dorsal w i n g sur­ face and to yield a net rotational movement of air around the w i n g (fig. 1.3B). This motion of air is equivalent to a rotating flow field (or vortex) that circulates around the w i n g and that is centered about the w i n g itself (usually at a point 25% along the chord length from the leading edge). A n y such b o u n d vortex is an irrotational vortex w i t h the quantitative characteristic that, at any point w i t h i n the vortex, the product of the local tangential velocity and the distance between the point and the vortex core (i.e., the radius) is a constant (see Vogel, 1994). The rotational intensity of the vortex is simply all tangential components of velocity summed around the circumference of the vor­ tex, a quantity termed the circulation (r). The physical origin of lift thus lies w i t h i n creation of net air circulation about a m o v i n g w i n g . Circulation, as a form of angular momentum, requires kinetic en­ ergy i n order to be initiated. For a w i n g section that is initially at rest and then begins to translate, circulation increases continuously i n ac­ cordance w i t h that required to satisfy the equilibrium value associated w i t h flow separation from the dorsal w i n g surface. Conservation of energy for a rotating system (i.e., the conservation law of circulation k n o w n as Kelvin's theorem) dictates that circulation of comparable magnitude but of opposite sense must be generated i n the surround­ ing fluid. Thus, a bound vortex on a w i n g must always be associated w i t h a so-called starting vortex of circulation -f that remains i n the

FLIGHT AND THE PTERYGOTE INSECTA

23

FIG. 1.3. Three-dimensional perspective of the vortex structure generated dur­ ing the downstroke. The starting, tip, and stopping vortices are linked to form a complete vortex structure; vortex structures of each wing pair likely combine to produce a single vortex ring (section 3.2.2). Vorticity shed consecu­ tively at the end of down- and upstrokes generate the vortex wake of a flying insect, although vortical interactions between down- and upstroke are not well resolved. region of initial acceleration of the airfoil (fig. 1.2B). The presence of this vortex i n the airfoil's wake results i n a mathematical field of vor­ ticity that, at any point i n space, is proportional to the angular velocity of the local air particles. The angular momentum represented by this vorticity field must be supplied by the w i n g and represents a transient energetic input associated w i t h the initiation of circulation about the wing. Once a bound vortex is i n place, additional w o r k is necessarily asso­ ciated w i t h sustained lift production. Circulation must be maintained i n the face of viscous dissipation that characterizes on all vortices. More importantly, the ventral momentum flux associated w i t h a bound vortex represents a continuous transfer of energy from the w i n g to the surrounding air. For three-dimensional airfoils, this mo­ mentum flux derives from yet another vortex i n addition to the bound and starting vortices. As a w i n g translates i n space, the pressure gradi­ ent underlying lift production yields airflow not only around spanwise w i n g sections (the bound circulation), but also around the w i n g tip. This pattern of airflow creates an additional vortex, the t i p vortex (fig. 1.3), that is unavoidably associated w i t h lift production by threedimensional wings. The t i p vortex produces a net dorsoventral air flow (the induced velocity) that yields a useful flux of air but also that imposes an additional force on the airfoil, termed the induced drag.

24

CHAPTER ONE

For lift generation to persist, the w i n g must actively supply energy to the surrounding air i n order to overcome this induced drag. The total energetic costs of lift production are thus positively correlated w i t h the circulation of the w i n g t i p vortex and w i t h the magnitude of the i n ­ duced velocity. Wings that are long relative to the mean w i n g chord (i.e., of h i g h aspect ratio) generate t i p vortices that are small i n magni­ tude relative to the bound circulation. The induced velocity is rela­ tively l o w i n such cases, and the associated momentum flux of air de­ rives primarily from a greater mass of air m o v i n g ventrally at lower velocities. Relative to wings of lower aspect ratio, h i g h aspect ratio wings are associated w i t h a reduced power expenditure to create com­ parable lift forces. Given that a three-dimensional vortex is generated by a translating w i n g , the associated reactive lift force on the w i n g can be estimated. The t i p vortex is linked to the starting and to the bound w i n g vortices (fig. 1.3) and creates a closed vortex loop that exerts a momentum flux on the surrounding air. Intuitively, the momentum induced by the presence of the vortex should be proportional to the mass of the air moved ventrally (i.e., to the air density) and to the velocity at w h i c h the air moves (i.e., the circulation). The Kutta-Joukowski law re­ lates the magnitude of lift produced per unit w i n g span ( L ) to the air density, the translational velocity of the w i n g , and the b o u n d circulation: span

Lspan =

pVr.

(1.3)

The dependence of lift production on w i n g velocity can best be under­ stood by realizing that the area of the closed vortex loop increases linearly w i t h the translational velocity, i n essence sweeping the bound vortex through space over a greater area at higher velocities (fig. 1.4). Circulatory lift is thus inextricably associated w i t h the momentum flux produced b y any vortex structure (see Dickinson, 1996). Furthermore, the instantaneous force that any vortex can produce is directly propor­ tional to the rate at w h i c h it can be moved through space. More rapid flapping motions by wings, for example, w i l l yield greater lift. As w i t h drag, the total lift L on a structure can be nondimensionalized through definition of a lift coefficient Q ; 2L

pSV

(1.4)

where p, S, and V are as i n equation (1.2). For comparative purposes, the reference area used for airfoils is typically the plan area of the w i n g (i.e., the horizontal projection of w i n g area). Inertial characteristics of

FLIGHT AND THE PTERYGOTE INSECTA

25

lift (=pVT)

bound vortex

translational velocity (V)

^

starting vortex

drag

^^^^N.

circulation (r)

FIG. 1.4. Two-dimensional cross section of the vortex wake generated by a moving wing. The instantaneous translational velocity (V) of the wing inter­ acts with the bound circulation r to produce a downward momentum flux in the wake and an instantaneous force on the airfoil. flow dorsally and ventrally over wings are directly influenced by air density, whereas airflow separation from the trailing edge of the w i n g is constrained b y viscous effects w i t h i n the wing's boundary layer. The Re at w h i c h a w i n g is operating thus exerts a strong influence on lift production. Under highly viscous circumstances, vortex generation becomes more difficult as intermolecular stickiness progressively i m ­ pedes rotational motions of airflow (section 3.2.1). Circulatory lift be­ comes increasingly more costly to maintain at l o w Re, and very small insects may rely on noncirculatory mechanisms of force production to stay aloft (section 7.1.3). Bound vortices are, however, probably characteristic of most insects several millimeters and larger i n w i n g length. A t the higher Re charac­ teristic of such wings, cambered profiles characterize effective airfoils as such shapes facilitate smooth merging of dorsal and ventral airflows downstream of the airfoil at l o w angles of attack. The b o u n d circula­ tion is of a magnitude comparable to the local airspeed, and lift forces increase i n approximate proportion to the square of the translational velocity of the w i n g (see eq. 1.3). As the angle of attack increases, lift forces increase linearly but then peak at angles typically between 20° and 30°. Separation of the dorsal airstream from the w i n g then be­ comes more pronounced and occurs more anteriorly at higher angles of attack, ultimately leading to disruption of circulatory flow and f i ­ nally to stall, a sharp decline i n lift production (section 3.2.1).

26

CHAPTER ONE

Lift is a tremendously useful force because the lift : drag ratios of well-designed airfoils, including insect wings, are w e l l i n excess of unity. Lift forces over much of the Re range relevant to insect flight also increase i n approximate proportion to the square of w i n g velocity. I n addition to the induced drag and power expenditure associated w i t h maintenance of the t i p vortex, however, profile drag is always present on wings and incurs energetic cost. I n return, however, the resultant force is substantially higher than w o u l d otherwise be avail­ able through use of drag-based propulsion alone. Large aerodynamic forces can thus be generated w i t h m i n i m a l energetic expenditure. W i n g lift i n fact predominates the overall force balance of flapping i n ­ sect wings; only i n the takeoff flight of pierid butterflies (Ellington, 1980a, 1984a; Sunada et al., 1993b) has pressure drag been found to substantially exceed w i n g lift. Furthermore, the directionality of lift is easily altered on a flapping airfoil. Simply b y altering w i n g orienta­ tion relative to flow a n d / o r the wing's relative airspeed through vari­ ation i n wingbeat kinematics, rapid change i n the magnitude and direction of resultant forces is possible. Wing lift is not necessarily d i ­ rected vertically; lateral forces as w e l l as horizontal thrust can also be generated by a m o v i n g w i n g . As w i t h drag, net lateral component of lift on bilaterally symmetric bodies and on bilaterally paired wings or w i n g pairs is minimal. D u r i n g maneuvers, however, asymmetric w i n g kinematics can result i n substantial imbalance between opposite (con­ tralateral) wings, generating sideways displacement and body rotation (see section 5.2.2). Insect bodies can themselves, somewhat surpris­ ingly, generate useful lift forces at positive angles of attack (section 3.2.1), but the magnitude of these forces is small relative to lift on wings. 1.2.4 Force Production through Wing Flapping The preceding discussion has referred to aerodynamic forces acting on an airfoil operating at constant velocity and angle of attack (i.e., under steady-state nonaccelerating conditions). Insect wings, by contrast, are flapped continuously about the w i n g base. This reciprocating m o t i o n generates a spanwise velocity gradient along the wings—flapping ve­ locity is greatest at the w i n g t i p and declines monotonically to zero at the w i n g base. Because wings are reciprocated at continuously varying angular velocities (in contrast, for example, to a continuously rotating helicopter blade), the translational velocity, acceleration and higher positional derivatives of any point along the wingspan vary continu­ ously. Moreover, periodic rotation of wings about their longitudinal

FLIGHT AND THE PTERYGOTE INSECTA

27

axis results i n time-dependent angles of orientation relative to the stroke plane. W i n g rotation occurs mostly at the beginning and at the end of a wing's half-stroke, and these rotations are thus approximately 90° out of phase w i t h respect to w i n g oscillation about the w i n g base. The angle of attack may also vary through the wingbeat via smallerscale rotational adjustments. Both t i m i n g and the angular extent of w i n g rotation at ends of half-strokes can be actively controlled and may exhibit substantial variation according to aerodynamic demand (section 3.1.2). These kinematic deviations from the above steady-state portrayal of lift generation render impossible the accurate prediction of the aerody­ namic forces on flapping wings. A t the very least, w i n g orientation and relative velocity must be assumed to vary continuously and to yield correlated variation i n both the magnitude and direction of w i n g lift and drag. The unsteady aerodynamic effects associated w i t h w i n g rotation and w i n g acceleration also yield instantaneous forces that dif­ fer substantially from those predicted by equations (1.2) and (1.4). I n sum, aerodynamic forces on flapping wings are not w e l l understood, although unsteady lift appears to be a fundamental motive force underlying insect flight and maneuverability (see section 3.2.2). Further complexities arise from consideration of the vortex flows as­ sociated w i t h multiple reciprocating wingbeats. As a w i n g accelerates and then decelerates i n a half-cycle of a full wingbeat, vorticity is i n i ­ tially shed spanwise across the full length of the wing's trailing edge (the starting vortex), a t i p vortex is shed continuously as the w i n g translates, and finally termination of motion results i n release of the bound circulation from the w i n g . A complete vortex loop is formed that then translates i n space away from the point of formation (fig. 1.3). Depending on the extent of lift generation, a similar vortex structure may be produced d u r i n g the upstroke (section 3.2.2.5). Vortex loops of the d o w n - and upstroke may link together or move i n close proximity, each influencing the other's motion. More importantly, preexisting vorticity i n the fluid surrounding a w i n g may enhance or detract from the generation of new bound circulation. Consecutive wingbeats gen­ erate series of such vortex structures that translate and interact w i t h one another and w i t h the bound circulation of the wings i n as yet undescribed ways; the three-dimensional geometry of the vortex wake is i n general u n k n o w n for arbitrary configurations of continuously oscil­ lating and rotating wings. For insects w i t h t w o pairs of wings (the ap­ parent ancestral condition), this situation is further complicated b y interaction of the shed vortex sheets between ipsilateral (same-side) and potentially between contralateral wings. A l t h o u g h little is k n o w n

28

CHAPTER ONE F

vert

+

L

b

(A)

thrust < body drag aerodynamic pitch

V wing base i center of mass gravitational pitch

mg

(B)

FIG. 1.5. Force and moment balance for a free-flying insect: (A) lateral perspec­ tive, and (B) dorsal perspective. Forces and torques through and about the center of body mass are balanced in nonaccelerated horizontal flight. F t : net vertical forces produced by wing flapping; L&: body lift. Body weight mg is usually much larger in magnitude than the body drag. ver

FLIGHT AND THE PTERYGOTE INSECTA

29

about the vortex wake of free-flying insects, extensive experimental w o r k has been carried out on vortex production by tethered insects (see section 3.2.2.5). Independent of the mechanisms of force generation, an overall bal­ ance of forces and torques characterizes the steady nonaccelerating forward flight of insects (fig. 1.5). Because there are three rotational and three translational axes of movement for any object i n space, a total of six degrees of freedom characterizes body motion of a freeflying insect. Most experimental w o r k on insect flight mechanics has evaluated only a subset of these kinematic quantities (section 5.2). For stable forward flight, net vertical force production must balance the body weight less any body lift forces, whereas thrust (net horizontal force) must offset parasite drag forces on the body. Equilibrium of ro­ tational moments also applies i n steady flight: the net pitching mo­ ment about the center of mass must equal zero, whereas left and right w i n g motions must be bilaterally symmetric such that yaw and roll moments are balanced (see fig. 1.5). Stability i n flight therefore i n ­ volves a wingbeat-to-wingbeat maintenance of this force and moment balance. Such dynamic precision clearly requires continuous sensory transduction and kinematic compensation i n the face of unexpected turbulence and intrinsic instability of underlying aerodynamic mecha­ nisms. Transient disruption of the force and moment balance, by con­ trast, provides for the rapid body rotations and accelerations that are the essence of insect flight maneuverability (section 5.2).

1.3 W H Y STUDY FLIGHT BIOMECHANICS? A central theme of this book is that diverse features of insect biology are united by the commonality of biomechanical performance during flight. This thesis can be briefly illustrated by examining morphologi­ cal diversity of the extant fauna, flight performance i n relation to par­ ticular features of insect ecology, and the largely unexplored realm of unsteady aerodynamic force production associated w i t h the flapping and rotation of flexible wings. 1.3.1 Evolution of Morphological

Diversity

The extraordinary morphological diversity of the w i n g e d insects is best evaluated i n the context of functional utility. Genetic and abiotic constructional constraints notwithstanding, w i n g and body morpholo­ gies often appear to correlate w i t h various taxon-specific aspects of flight performance and locomotor capacity. Many scenarios of physio-

30

CHAPTER ONE

logical adaptation have been criticized for assumptions of optimized matching between structure and function (see Garland and Huey, 1987; Dudley and Gans, 1991; Weibel et a l , 1998). The biomechanical evaluation of extant structures does not, however, i m p l y either partic­ ular evolutionary trajectories or current selective advantage. Rather, discussions of aerodynamic performance and biomechanical feasibility can be used to frame hypotheses of adaptation that can then be tested rigorously using a phylogenetic framework. This approach is specifi­ cally used i n chapters 6 and 7 to test various hypotheses relating flight and insect diversification. Throughout the book, correlational state­ ments relating functional utility to performance w i t h i n particular be­ havioral and ecological contexts are intended primarily as hypotheses for evaluation w i t h i n the context of modern comparative biology. Pterygote insects are defined taxonomically b y the presence of wings, and w i n g morphology i n particular exhibits m y r i a d functional modifications (section 2.2). The phylogenetically ancestral w i n g condi­ tion is generally presumed to be that of homonomy, namely of iteratively homologous w i n g pairs that share equivalent ontogenetic ori­ gin, shape, and function. Homonomous wings are best illustrated by damselflies (Odonata: Zygoptera; see plate 2). I n this taxon, fore- and hindwings are morphologically and kinematically highly similar, op­ erating at similar beat frequencies and serving comparable aerody­ namic functions. Homonomous w i n g pairs can be found i n a number of insect orders (section 2.2.3), but i n general fore- and hindwings have diverged historically both i n terms of aerodynamic contributions to flight and i n their nonaerodynamic biological roles. One major trend i n insect evolution has been relative reduction i n area of one w i n g pair and functional or actual coupling of the t w o ipsilateral wings to yield one continuous aerodynamic surface. Relative w i n g reduction is t y p i ­ cally mirrored by a reduction i n the size of the corresponding thoracic segment. Major orders exhibiting this trend of coupling and transfor­ mation to one effective w i n g pair include the Hymenoptera and the Lepidoptera, together almost 30% of the extant insect fauna (table 1.1). I n addition to functioning aerodynamically, the overlapping and i n most cases physically linked ipsilateral wings of the Lepidoptera per­ form numerous nonaerodynamic roles i n such diverse contexts as crypsis, sexual selection, thermoregulation, and mimicry. A second major theme of w i n g transformation at the ordinal level is conversion of forewings to protective devices w i t h reduced and usu­ ally m i n i m a l aerodynamic function. The forewings of beetles, for ex­ ample, are termed elytra and are heavily thickened and sclerotized, serving primarily to protect the wings and abdomen when the insect is not i n flight. Forewings i n the order Hemiptera are termed hemely-

FLIGHT AND THE PTERYGOTE INSECTA

31

tra and are also partially thickened to serve, at least i n part, a protective function. Forewings of various other orders are slightly sclerotized and leathery, and are k n o w n as tegmina (section 2.2.3). A general tendency toward tegminization and elytrization of the forewings is thus widespread among the winged insects. Dramatic reduction i n the size of one w i n g pair has also occurred i n various taxa. Most notably, the hindwings of flies (the haltères) are miniaturized and function as sensory rather than aerodynamic structures; analogous miniaturization of one w i n g pair can be found i n at least t w o other orders (see section 2.2.3). M a n y insect taxa are rendered secondarily flightless through substantial reduction and even through complete loss of both w i n g pairs (section 6.3.1). Although functional correlates of w i n g differentiation have not been systematically investigated among i n sect orders, major differences i n flight performance and ecology are clearly associated w i t h interordinal patterns of w i n g modification and reduction. 1.3.2 Ecological Contexts of Flight: Specific Examples Numerous features of insect ecology depend either directly or indirectly on the ability of insects to fly. Selection on flight performance has apparently facilitated adaptive radiation of insects into such varied ecological roles as herbivores, pollinators, blood feeders, and long-range dispersers. Brief consideration of each of these ecological habits is therefore instructive from the unifying perspective of aerial locomotion. A l l nonaquatic insects live w i t h i n a structural context defined by the presence primarily of angiosperms and secondarily of gymnosperms. Roots, stems, seeds, fruits, shoots, flowers, and leaves provide a grand diversity of habitats for insect eggs, larvae, and adults. Herbivory and pollination are the t w o dominant forms of ecological interaction between insects and plants. Most insects are phytophagous (plant eating) for at least some portion of their life cycle, and many insects are i n volved i n pollen transfer between conspecific angiosperm flowers. Both herbivory and pollination are at least i n part influenced by pterygote flight capacity. M a n y plant viruses are also vectored by insects. Spatial mobility is essential for location of widely dispersed plant resources, either for purposes of immediate consumption or for o v i position i f herbivory is confined to the larval stages (as is characteristic of most Lepidoptera). Host specificity of many herbivore-plant interactions indicates that aerial search for the appropriate species of host plant is commonplace. Similarly, pollination of angiosperms has often involved reciprocal coevolutionary interactions between insect

32

CHAPTER ONE

locomotor capacity and floral structure and location. Specific coevolutionary interactions between pollination and flight performance are explored i n greater detail i n section 7.2. Hematophagy, or the ingestion of blood, is a highly specialized feeding mode that superficially appears to be of little relevance i n a discussion of insect flight biomechanics. After all, hematophagy has been a major factor associated w i t h w i n g loss i n t w o insect orders that are obligate ectoparasites on vertebrates (Phthiraptera and Siphonaptera, or the lice and fleas, respectively; see Waage, 1979). Also found among hematophagous insect species, however, are numerous d i p teran vectors involved i n transmission of bacterial, viral, and proto­ zoan diseases. Malaria, dengue, and leishmaniasis are some of the major diseases that are transmitted to humans by flies. Such detrimen­ tal behavior is not confined to the Diptera: various disease-transmit­ ting hemipteran Reduviidae have acted as vectors of Chagas disease (a degenerative cardiac condition of bacterial origin) to millions of h u ­ mans throughout Central and South America. The advent of global w a r m i n g may serve to hasten the spread of many of these pathogens and of their vectors; malaria i n particular is increasing rapidly i n fre­ quency and is thought conservatively to infect hundreds of millions of people w o r l d w i d e . Selection on flight performance plays a major role i n disease trans­ mission by pathogen-bearing insects i n at least t w o distinct ways. I n i ­ tial host location often involves extended dispersal to locate hosts, as w e l l as implementation of aerial search strategies typically based on olfactory cues (Lehane, 1991). Maneuverability, once the potential host is found, is essential to avoid aversive efforts that interfere w i t h para­ sitism. Humans w o r l d w i d e spend considerable amounts of their time both outdoors and indoors swatting at flies. Hematophagous insects are w e l l equipped aerodynamically to evade their vertebrate hosts, and the rapidity of such flight is i n part the product of millions of years of unintentional selection by vertebrates for increased flight perfor­ mance. After feeding, temporary ectoparasites must fly away from the host w i t h a greatly increased body mass. Tsetse flies, for example, i n ­ crease their body mass by a factor of 2-3 d u r i n g feeding (Langley, 1970), and the flight apparatus must be capable of offsetting such heavy blood loads (section 5.4.1). Flight metabolism of these hema­ tophagous insects is correspondingly specialized: water from the blood meal is rapidly excreted, whereas amino acids derived from i n ­ gested blood are used as metabolic fuels (Bursell, 1981). This latter fea­ ture distinguishes hematophagous insects from other w i n g e d insects, w h i c h generally power flight using carbohydrates and lipids (section 4.1.2). Hematophagy has thus imposed specific performance demands

FLIGHT AND THE PTERYGOTE INSECTA

33

on the insect flight apparatus that are of biomechanical, physiological, and medically anthropocentric interest. A final flight-related theme of insect ecology is long-distance disper­ sal, a behavior characteristic of many insects and most typically associ­ ated w i t h resource location. O n timescales ranging from seconds to months, insects implement spatial movements typically to relocate w i t h i n or among appropriate habitats. Wings are not strictly essential for such behavior (most arthropods disperse to varying degrees, and some insects migrate via walking or hopping), but flight increases the geographical range available for relocation b y orders of magnitude. Many insects can utilize ambient air motions to aid long-distance dis­ persal, the most extreme form of w h i c h is intercontinental transport. Flying insects have, for example, been routinely trapped over all major oceans (Bowden and Johnson, 1976). As a corollary of widespread i n ­ sect dispersal, a continuous and substantial fallout of the aerial insect fauna occurs over most terrestrial habitats, facilitating colonization of such remote regions as mountaintops and oceanic islands. The aerody­ namics of insect dispersal and associated flight behaviors are treated i n chapter 7. L 3 . 3 Insect Mastery of Unsteady Aerodynamics As discussed above i n section 1.2.4, generation of lift forces on flap­ ping wings involves continuous shedding of vorticity into the sur­ rounding air and generation of a complex flow field termed the vortex wake. Production and manipulation of these unsteady vortex flows forms the basis for extraordinary maneuverability. I n contrast to the staid flight paths typical of airplanes, most insects are capable of ex­ ceedingly rapid changes i n flight speed and direction. Transient gener­ ation of forces and torques results i n near-instantaneous turns, acceler­ ations, and directional changes, the dynamic characteristics of w h i c h are i n most cases unmatched by technology. Small body size facilitates rapid accelerations for allometric reasons (see section 5.4.1), but the maneuvers attained by some insects are nonetheless remarkable. For example, a 180° reversal of flight direction can be carried out by vari­ ous flies i n 40-100 ms, whereas backwards and even sideways flight is at least transiently implemented by many insects (see chapter 5). Tabanid flies can rapidly reverse flight direction using an Immelmann t u r n (a vertical half-loop followed by a half-roll; see Wilkerson and Butler, 1984). M a n y of these amazing maneuvers occur during events of aerial prédation or mate selection, contexts for w h i c h successful eva­ sion or capture has direct consequences for either mortality or repro­ ductive fitness, respectively.

34

CHAPTER ONE

The kinematic correlates of insect maneuverability have not been fully elucidated, although controlled instability is likely a general fea­ ture not only of such maneuvers but also of apparently steady forward flight (chapter 5). Unsteady forces on flapping wings can readily be manipulated through subtle but significant changes i n w i n g contour and orientation, and asymmetric w i n g motions typically form the basis of maneuverability i n insects. Leg and abdominal motions can also contribute to the generation of advantageous torques d u r i n g ma­ neuvering flight; additional flexibility is provided b y aerodynamic i n ­ teractions between opposite w i n g pairs. Neither the complete range of maneuverability nor the diverse kinematic means of force regulation has been fully described for any insect. Moreover, precise regulation and control of unsteady forces may be of more than zoological interest. Transient application of aerodynamic forces supplemental to weight support and thrust generation is otherwise k n o w n as vectored maneu­ verability, currently the subject of intense development for technologi­ cal applications (e.g., Gal-Or, 1990). A l t h o u g h the Re range relevant to insect flight is orders of magnitude below that of conventional aero­ nautics, ongoing interest i n the construction of microair vehicles may soon render the study of insect flight of more than biological interest.

1.4 SUMMARY The ability to fly has been central to the evolutionary diversification of insects. Initial w i n g evolution i n the late Paleozoic led to exten­ sive adaptive radiations throughout the Carboniferous and Permian, establishing most of the ordinal-level diversity i n flight-related mor­ phology that exists today. Flight has been a key enabling character underlying major adaptive trends i n contemporary insect diversity, i n ­ cluding miniaturization, aerial prédation, dispersal, and mating sys­ tems. Winged insects are both quantitatively and qualitatively impor­ tant components of terrestrial ecosystems. Pollinating and herbivorous insects impose intense selective regimes o n plants; flying insects are also preyed u p o n by a variety of animals and the occasional insectivo­ rous plant. Forces of both natural and sexual selection have contrib­ uted synergistically to the evolution of insect flight performance and maneuverability. I n spite of the apparent advantages of flight, second­ ary flightlessness is also widespread i n the pterygote Insecta. Insects fly i n regimes of fluid flow that are neither purely inertial nor purely viscous i n character, but rather that represent the combined ef­ fects of these t w o forces. Use of the Reynolds number provides a nondimensional parametrization to evaluate the relative influence of fluid

FLIGHT AND THE PTERYGOTE INSECTA

35

inertia and viscosity; the Re for flying insects ranges from 10° to 10 . Vortices and circulation-based lift about airfoils can be effectively gen­ erated b y many flying insects, but equally the viscous features of air impose substantial resistive forces on wings and bodies. The tremen­ dous diversity i n insect shape, size, and wingbeat kinematics indicates comparable variation i n Re and i n the aerodynamic mechanisms used to effect flight. Unsteady aerodynamic flows invariably result, how­ ever, from the flapping and rotating wings characteristic of most vo­ lant animals. 4

Chapter Two MORPHOLOGY O F T H E FLIGHT APPARATUS

I

NSECTS C A N FLY only i f the necessary aerodynamic machinery is expressed morphologically Anatomical structures required for flight include wings to generate aerodynamic force, thoracic mus­ culature, and an articulated juncture between wings and thorax that permits the transmission of muscular force and active control of w i n g orientation. This chapter treats first the structure of the thorax, then the functional morphology of the wings, and finally the ancillary mor­ phological structures used d u r i n g flight. M o d e r n reviews of the func­ tional morphology of wings are provided b y Brodsky and Ivanov (1983a), Wootton (1979, 1981b, 1992), and Grodnitsky and Morozov (1994). The flight apparatus is evaluated here i n a context of bio­ mechanical design rather than phylogenetic diversity. Because vari­ ation i n flight-related morphology among the orders is so extreme, I here use primary descriptive sources for i n d i v i d u a l taxa rather than secondary ones. Comprehensive ordinal-level sources on w i n g and thoracic morphology include works by Comstock (1918), M a k i (1938), Rohdendorf (1949), Séguy (1959,1973), Matsuda (1970,1979), Brodsky (1994), and Grodnitsky (1999). I n general, the partial or complete modification of one set of wings relative to the other has dramatically defined new patterns of morphological evolution i n the Insecta.

2.1 THORACIC DESIGN The structural components required for flight derive from the morpho­ logical underpinnings of insect body design. Intrinsic to biomechanical performance of the insect flight apparatus are the flexible cuticular properties of the arthropod exoskeleton. The material characteristics of arthropod cuticle derive from the presence of polysaccharide chitin microfibers embedded i n a protein matrix. Such material design is termed a fibrous composite, a structure that combines the intrinsic strength of the embedded fibers w i t h a h i g h toughness (the ability to absorb energy) derived from the b i n d i n g forces between fibers and matrix (see Wainwright et al., 1976; Neville, 1993). Fibrous composites are therefore both strong and elastically flexible, and arthropod cuticle is the finest zoological example of this mechanical design. Flexibility is essential for the insect thorax, as internal muscular forces cause the

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

37

cuticle to deform and either directly or indirectly transmit forces to the wings. Elastic return of stored energy i n the thorax is also essential to minimize the total energetic cost d u r i n g w i n g flapping. The wings themselves are very thin cuticular structures that, particularly i n larger insects, act as semiflexible airfoils that often substantially bend and change shape d u r i n g flight. For biomechanical purposes, wings can be considered to be inanimate structures that are activated b y the insect only when it applies muscular forces at the axillary apparatus. 2.1.1 Segmentation and Cuticular Anatomy M u c h of arthropod biology derives from the functional specialization of sets of adjacent body segments. Insects by definition possess three thoracic segments, each of which i n both larval and adult stages bears a pair of legs. I n adult insects, the meso- and metathoracic segments are collectively termed the pterothorax, upon w h i c h the wings are joined and w i t h i n w h i c h the flight muscle is found (fig. 2.1 A ) . Seg­ ments are composed externally of sclerites, well-defined cuticular plates w i t h distinct boundaries that can either be membranous or su­ tural i n character. I n arthropods generally, the dorsal sclerite of any given segment is termed the tergum. For thoracic segments of insects, the tergum (or tergal sclerite) is specifically termed the n o t u m (or notai sclerite). The notum, furthermore, is subdivided i n many insect taxa into an anterior scutum and a posterior scutellum. Ventrally, each tho­ racic segment is bounded by the sternum; t w o pleural sclerites or pleura (singular: pleuron) define the lateral limits of the thoracic seg­ ment (fig. 2.1B). A dorsoventral pleural suture divides each pleuron into t w o distinct regions, an anterior episternum and a posterior epimeron. A n internal pleural ridge runs parallel to the external pleural suture and strengthens the pleural sclerite dorsoventrally. Dorsally, the pleural suture ends i n the pleural w i n g process that acts as a ful­ crum directly beneath the w i n g base (fig. 2.1B). Two small sclerites, the basalar and the subalar, are located directly anterior and posterior to the pleural w i n g process, respectively. These sclerites serve as inser­ tion points for muscles that act to flap the w i n g dorsoventrally as w e l l as to rotate the w i n g about its longitudinal axis. Legs originate at the ventral base of the pleuron. The leg is an integral component of the flight apparatus i n that some dorsoventral flight muscles terminate ventrally w i t h i n the first and second leg segments (the coxa and tro­ chanter, respectively). I n contrast to the ventral location of insect legs, wings of insects con­ nect to the thorax i n the dorsal region of the pleuron. Two processes (small projections) extend laterally from the notum, one anterior to and one posterior to the pleural w i n g process (fig. 2.2A). The base of

38

CHAPTER TWO

(B)

FIG. 2.1. (A) Pterygote tagmosis and thoracic segmental identity in dorsal per­ spective (Zoraptera: Zorotypus brasilensis; modified from Choe, 1992). The ab­ dominal cercus is a small appendage that bears mechanosensory hairs. (B) Generalized anatomy of a pterothoracic segment in lateral oblique perspective.

M O R P H O L O G Y OF T H E F L I G H T APPARATUS

39

FIG. 2.2. Generalized anatomy of a pterothoracic segment and wing articula­ tion in dorsal (A) and ventral (B) perspective (modified from Snodgrass, 1935). The first, second, and third axillary sclerites are indicated by their correspond­ ing numerals; a median sclerite is indicated with an m.

40

C H A P T E R TWO

the w i n g inserts on the thorax among these three processes. A median w i n g process is present i n some orders and projects between the ante­ rior and posterior notai processes. (Alternative nomenclature for notai processes was introduced by La Greca 1947 and was used by Matsuda 1970 i n his comprehensive anatomical monograph.) The w i n g articula­ tion is comprised of a set of small axillary sclerites lying between the w i n g and the notum. A d d i t i o n a l preaxillary sclerites termed the h u ­ meral plate and the tegula may be contained w i t h i n the w i n g articu­ lation anterior to the anterior notai process (see fig. 2.2A); these t w o sclerites and the axillary sclerites are collectively termed the pteralia or the axillary apparatus. W i n g articulation and the number of axil­ lary sclerites exhibit substantial variation even at the basal levels of pterygote phylogeny. This observation is consistent w i t h ancestral diversity i n the anatomical structures that underlie flight i n w i n g e d insects. Specific evolutionary transformation of the ancestral insect w i n g base is still the subject of considerable speculation. For example, the precise identity of axillary sclerites i n the Paleoptera is not w e l l estab­ lished. The Odonata have only a humeral plate (lying at the base of and supporting the costal w i n g vein; see below) and one axillary scler­ ite; the homology (or common phylogenetic origin) of this latter scler­ ite w i t h that i n other orders is uncertain (Tannert, 1958; Matsuda, 1970; Pfau, 1986, 1991). The three axillary sclerites of the Ephemeroptera, b y contrast, appear to be homologous w i t h those of the Neoptera (Ma­ tsuda, 1970; Kukolavá-Peck, 1985). I n the Neoptera, the first axillary sclerite lies directly lateral to the anterior notai process, whereas the second axillary sclerite is joined laterally w i t h the first axillary sclerite and lies directly above the pleural w i n g process (fig. 2.2B). The second axillary sclerite and the subalar sclerite are mechanically coupled b y internal cuticular connections, whereas the basalar sclerite attaches d i ­ rectly to the axillary apparatus at the base of the most anterior vein, the costa. Median sclerites may lie laterally to the second axillary scler­ ite and the base of w i n g veins. The third axillary sclerite is supported by the posterior notai process (see fig. 2.2A) and forms the attachment point for muscles that connect to the pleuron. These pleuroalar mus­ cles flex the w i n g against the abdomen w h e n the insect is not flying, but may also influence w i n g orientation d u r i n g flight (e.g., Heide, 1971a, b). The first, second, and t h i r d axillary sclerites are joined distally w i t h the bases of the subcostal, radial, and anal w i n g veins, re­ spectively (see below). A fourth axillary sclerite, w h i c h arises from the posterior notai process, is found i n the Hymenoptera and Orthoptera (Matsuda, 1970). The secondary musculature associated w i t h this sclerite may subtly influence the kinematic characteristics of the w i n g beat (see Pringle, 1968).

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

41

2.1.2 Muscle Configuration and Function Contraction of intrinsic flight musculature generates forces that are transmitted to the surrounding cuticle of the exoskeleton and to the the base of insect wings. Two different sets of thoracic muscles are p r i ­ marily responsible for w i n g motions. Wing depressors and w i n g ele­ vators effect the downstroke and the upstroke, respectively. Muscles may insert either directly on the w i n g base and sclerites of the axillary apparatus, or may act indirectly to move the wings via indirect tho­ racic deformation. I n either case, muscles connect to the cuticle directly via tonofibrillae (cuticular microfibrils) without intervening tendons (see Korschelt, 1932; Boettiger, 1960; Auber, 1963; Lai-Fook, 1967). Elastic tendon-like structures are, however, widespread i n flight mus­ cles of odonates of the suborder Anisoptera (Clark, 1940). The rota­ tional axis for w i n g depression and elevation is predominantly defined by dorsoventral movement of the first axillary sclerite against the notum, and particularly against the anterior notai process (fig. 2.2). Thoracic muscles that effect dorsoventral w i n g motion act either d i ­ rectly on sclerites at the w i n g base, or activate the w i n g indirectly via displacement of the n o t u m and transmission of force to the wings through the axillary apparatus. I n some insect taxa, both modes of force transmission may be present depending on the particular muscle under consideration. Direct action of flight muscles on the w i n g base is phylogenetically ancestral. A m o n g extant taxa, the Odonata (Pfau, 1986, 1991) and the Blattaria (Tiegs, 1955) use primarily direct muscles to generate the downstroke. I n these taxa, the dorsolongitudinal muscles are small rel­ ative to the direct basalar and subalar muscles as w e l l as to the i n ­ direct dorsoventral muscles that effect the upstroke. Action of direct depressor muscles lowers the wings through direct transmission of ap­ plied force along axillary sclerites to the basal w i n g venation. The d i ­ rect w i n g depressors of Odonata (i.e., the basalar and subalar muscles) insert via tendons lateral to the pleural ridge and rotate the w i n g downwards about the pleural w i n g base (fig. 2.3A). These w i n g de­ pressors originate on a cuticular brace (the furca) located at the base of each thoracic segment and above the leg musculature (Sargent, 1917; fig. 2.3A). Direct flight muscles also are prominent i n the Ephemeroptera (Matsuda, 1970), Orthoptera (Tiegs, 1955; Pfau, 1978) and some Coleóptera (Pringle, 1957; Schneider, 1987). Such flight muscles gener­ ally insert on the basalar and subalar sclerites and terminate w i t h i n the coxa, although some trochanteral muscles can also act directly on the w i n g base. M a n y direct dorsoventral muscles i n the Orthoptera (Wilson, 1962) and Blattaria (Fourtner and Randall, 1982) have been suggested to contribute additionally to leg motions during cursorial

42

C H A P T E R TWO

FIG. 2.3. Direct and indirect flight muscles in cross-sectional perspective. ( A ) Generalized odonate pterothorax (modified from Sargent, 1917; Pfau, 1991). (B) Generalized pterothoracic segment with both direct (basalar) muscles and indirect flight muscles (modified from Snodgrass, 1935). In both ( A ) and (B), the subalar muscle and corresponding subalar sclerite are directly behind the basalar muscle and basalar sclerite, respectively. locomotion, blurring the functional distinction between muscle origin and insertion. Leg position d u r i n g flight may also be influenced by tonic (constant force) activity i n direct flight muscles (see Kutsch and Usherwood, 1970). Such bifunctionality (use i n flight and i n running) may be confined to only one of such muscles (Burrows, 1996), h o w ­ ever, and different patterns of central neural control characterize acti­ vation of these bifunctional muscles (Ramirez and Pearson, 1988). I n addition to originating i n the leg, muscles that act to displace the ba­ salar and subalar sclerites may be pleural i n character, originating on the episternum and the epimeron, respectively. I n addition to their role i n the downstroke, direct flight muscles also influence rotation of the w i n g about its longitudinal axis. Because the basalar and subalar sclerites are not coincident w i t h the pleural w i n g process (see fig. 2.2B), contraction of the corresponding basalar and subalar muscles generates not only downwards displacement of the w i n g but also torque and concomitant w i n g rotation. Pronation refers to a nose-downward rotation of the leading edge of the w i n g , whereas supination conversely indicates w i n g rotation of the opposite sense. I n the locust, near-simultaneous contraction of the basalar and subalar muscles produces the downstroke, whereas the extent of w i n g prona­ tion is correlated w i t h variation i n contraction of the basalar muscle (Pfau, 1977, 1978; Wolf, 1990). Differential activity of the basalar and

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

43

subalar muscles also effects the fore-aft (anteroposterior) w i n g mo­ tions k n o w n as promotion and remotion. I n beetles, for example, subalar muscles serve to regulate w i n g angle of attack as w e l l as more subtle kinematic features involved i n flight control (Kammer, 1971; Govind, 1972; Pfau and Honomichl, 1979). Fore-aft motions as w e l l as the extent of w i n g rotation effected by the direct flight muscles are largely defined by rotational mobility of the second axillary sclerite about the pleural w i n g process (fig. 2.2B). Activity of additional direct flight muscles can also influence w i n g orientation. I n locusts, pleuroalar muscles inserting on the third axil­ lary sclerite act antagonistically to the basalar and subalar muscles during the upstroke. Particularly during the upstroke, the pleuroalar muscles regulate w i n g supination and angle of attack relative to the dorsoventral plane of motion (Nachtigall, 1981a; Pfau and Nachtigall, 1981; Heukamp, 1984). Similar functions characterize action of the third axillary muscle i n the lepidopteran Manduca sexta, w h i c h func­ tions not only i n w i n g folding but also contracts periodically at the wingbeat frequency (i.e., phasically) to regulate w i n g promotion and remotion (Rheuben and Kammer, 1987). Bilateral asymmetry i n activa­ tion of this muscle also influences axial and rotational maneuverability (Kammer, 1971; Kammer and Nachtigall, 1973; Wendler et al., 1993; see section 5.2.2). Thus, the third axillary muscle provides not only for w i n g folding but also for subtle yet important features of flight control. I n contrast to the phylogenetically ancestral case of direct muscle action at the w i n g base, indirect flight musculature effects w i n g mo­ tions via notai displacement. Some phylogenetically basal insect or­ ders (e.g., Ephemeroptera, Plecoptera) nonetheless possess a highly derived flight apparatus i n that the ventral w i n g motions appear to be generated indirectly via thoracic deformation (see Matsuda, 1970; Brodsky, 1994). I n many groups w i t h indirect flight muscles, contrac­ tion of the dorsolongitudinal muscles produces the downstroke through indirect arching and vertical displacement of the n o t u m be­ tween opposite w i n g bases of a segment (fig. 2.3B). This motion causes the lateral edge of the n o t u m and the adjacent axillary apparatus to rise above and rotate about the pleural w i n g process, thereby depress­ ing the wings. The Odonata, although generally considered to utilize only direct muscles i n the downstroke, do possess limited dorsolongi­ tudinal musculature that may have ancestrally arched the n o t u m dorsally to assist w i t h w i n g depression (Pfau, 1986; see fig. 2.3A). I n present-day odonates, these muscles n o w function i n promotion or remotion and exhibit different roles i n the meso- and metathoracic seg­ ments (Pfau, 1991). Muscle function during both the d o w n - and the upstroke is thus highly variable among extant winged insects. Rather

44

C H A P T E R TWO

than a strict dichotomy between direct and indirect musculature as has been classically assumed, a more complete description w o u l d entail evaluation of the relative contribution of different muscles to w i n g de­ pression, elevation, rotation, and fore-aft movements. Most evalua­ tions of muscle action have been made strictly on the anatomical bases of insertion point and relative muscle volume. Patterns of neuromus­ cular activation are much less studied, and the experimental manipu­ lation of muscles (e.g., denervation) to investigate subsequent effects on w i n g motions has rarely been carried out i n a systematic compara­ tive sense. The small size of most insects unfortunately precludes most such investigations at the present time. W i n g elevation i n all insect orders is primarily attained through ac­ tion of indirect dorsoventral muscles that connect the n o t u m to the sternum as w e l l as to proximal leg segments (fig. 2.3B). Dorsal oblique muscles that r u n from the n o t u m to the base of the cuticular phragmata (partitions) between segments may also function as w i n g eleva­ tors i n some cases (e.g., Barber and Pringle, 1966). Indirect elevation of wings is attained b y muscle action that lowers the n o t u m relative to the pleural w i n g process, thereby rotating the w i n g upwards about the axillary hinge. Muscles that elevate the wings thus act antagonistically to the dorsolongitudinal muscles (and, i n some cases, to direct dorso­ ventral muscles) that function i n w i n g depression. The phylogenetic diversity of direct and indirect muscle action has not been systematically explored, but the major differences even among pterygote orders i n muscle number and configuration (see Matsuda, 1970) must be reflected i n alternative strategies of thoracic deformation and active w i n g control. For any given taxon, only a m i ­ nority of flight muscles has been studied electrophysiologically on tethered experimental preparations (i.e., stationary insects flapping their wings i n fictive flight). One general result that emerges, however, is that phylogenetically more derived orders decouple more effectively the powerful muscles involved i n dorsoventral w i n g movement from those muscles controlling w i n g orientation. I n dipterans, hymenopterans, and coleopterans, for example, the indirect flight muscles that effect notai bending and thoracic deformation are physiologically and mechanically distinct from the w i n g control muscles (see Nachtigall and Wilson, 1967; Kammer, 1985; Dickinson and Tu, 1997; Wisser, 1997; Nachtigall et al., 1998). Such a decoupling may be a necessary feature of any evolutionary increase i n wingbeat frequency, as contrac­ tile dynamics become more difficult to regulate actively for a rapidly oscillating muscle. For the asynchronous flight muscles characterized by multiple contractions for a single nervous impulse (see section 4.2.2), wingbeat-by-wingbeat control is impossible to attain neuro-

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

45

nally. Instead, w i n g orientation and other kinematic features not de­ r i v i n g from dorsoventral movements are under the active control of additional muscles that act at the w i n g base. Dipterans, hymenopterans, and coleopterans are i n fact all char­ acterized by asynchronous indirect flight muscles (section 4.2.2), whereas the direct control muscles remain synchronous, contracting phasically at a usually high frequency but producing little power. A remarkable total of eighteen such control muscles can be found i n the dipteran mesothoracic segment (Dickinson and Tu, 1997). These direct muscles (particularly the basalar and pleurosternal muscles) receive neural activation at the wingbeat frequency and impinge both tonically (at constant contraction strength) and phasically ( w i t h tempo­ rally fluctuating contraction strength) on w i n g motions (Heide, 1968, 1971a,b; Dickinson et al., 1993). The first basalar muscle i n particular appears to influence w i n g base mechanics and thus w i n g orienta­ tion through changes i n stiffness induced by variable t i m i n g of neu­ ral activation (Tu and Dickinson, 1994). Action of the first and second basalars together w i t h that of additional axillary muscles influences both stroke amplitude and wing-tip paths (Heide and Götz, 1996; Leh­ mann and Götz, 1996; Tu and Dickinson, 1996). Flies thus exemplify the physiological and functional divergence between indirect powerproducing muscles and direct control muscles (Dickinson and Tu, 1997). Evolution of this functional dichotomy must presumably paral­ lel the evolution of asynchronous muscle i n many taxa, but the relative contributions of indirect and direct flight muscles remain to be deter­ mined i n comparative contexts. 2.1.3 Diversity of Thoracic Design A general theme of thoracic evolution is that segment size can be dra­ matically modified or reduced i n accordance w i t h evolution of w i n g function. The prothoracic segment of winged insects, bearing a pair of legs but no wings, is much smaller i n volume than either pterothoracic segment. Schwanwitsch (1943, 1958) used the terminology of anteromotorism, posteromotorism, and bimotorism to describe relatively en­ larged forewings, enlarged hindwings, or equivalent w i n g size, re­ spectively (see table 2.1). Here these terms are used to indicate not only relative size of wings on pterothoracic segments but also the likely ex­ tent of aerodynamic contributions d u r i n g flight. For most taxa, the rel­ ative roles of wings i n force production has not been assessed quanti­ tatively, although the relative sizes of the meso- and metathorax tend to follow trends of w i n g specialization and thus locomotor dedication of the segment i n question. For example, the coleopteran mesothorax

46

C H A P T E R TWO

TABLE 2.1

Wing Number and Pterothoracic Specialization among Extant Pterygote Insect Orders Order Ephemeroptera Odonata Plecoptera Embioptera Orthoptera Phasmatodea Grylloblattodea Dermaptera Isoptera Mantodea Blattaria Zoraptera Thysanoptera Hemiptera Homoptera Psocoptera Phthiraptera Coleóptera Neuroptera Megaloptera Raphidioptera Hymenoptera Trichoptera Lepidoptera Strepsiptera Diptera Siphonaptera Mecoptera

Alary Mode (wing number) Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Aptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Aptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Tetraptery Diptery Diptery Aptery Tetraptery 3

Locomotor Mode Anteromotorism Bimotorism Posteromotorism Bimotorism Posteromotorism Posteromotorism n/a Posteromotorism Bimotorism Posteromotorism Posteromotorism Anteromotorism Anteromotorism Anteromotorism Anteromotorism Anteromotorism n/a Posteromotorism Bimotorism Bimotorism Bimotorism Anteromotorism Anteromotorism Anteromotorism Posteromotorism Anteromotorism n/a Bimotorism b

c

d

e

1

j

k m

1

11

Tegminization/ Elytrization None None None None Tegmina Tegmina n/a Tegmina None Tegmina Tegmina None None Hemelytra None None n/a Elytra None None None None None None None None n/a None f

Wing Miniaturization None None None None None None n/a None None None None None None None None None n/a None None None None None None None Haltères (forewing) Haltères (hindwing) n/a None 8

h

0

Notes: Alary mode - the number of wings (tetra-, di-, or aptery indicate 4, 2, and 0 wings, respectively). Locomotor mode - anteromotorism (enlarged fore wings), posteromotorism (enlarged hindwings), bimotorism (equivalent aerodynamic use of homonomous wing pairs). Details of wing tegminization/elytrization and wing miniaturization are given in the text. Because flightlessness has evolved in approximately half of the extant orders (fig. 6.2), these characterizations represent the general trend only of winged taxa within each order. Hindwings are reduced in Ephemeroptera and are absent in Caenidae and in certain Baetidae and Tricorythidae. Hindwings are generally larger than forewings in the suborder Anisoptera. Wings are homonomous in some Plecoptera. The family Mastotermitidae exhibits enlarged hindwings. Hindwings are vestigial in some species. Forewings are tegminized in some Auchenorrhyncha (e.g., Fulgoridae). Miniaturized hindwings of Coccidae may act as haltères. a

b

c

d

e

f

g

M O R P H O L O G Y OF T H E F L I G H T APPARATUS

47

is substantially smaller than the metathorax (Matsuda, 1970; Schnei­ der, 1978a; Crowson, 1981), consistent w i t h reduced aerodynamic roles of the forewing (section 2.2.3). By contrast, the mesothorax pre­ dominates i n both the Diptera (Young, 1921) and the Hymenoptera (Snodgrass, 1910), as w e l l as i n many Hemiptera and Homoptera. Other orders w i t h alary (wing) specialization show a similar tendency (fig. 2.4). Mesothoracic musculature i n the Orthoptera is, for example, reduced relative to that of the metathorax (Tiegs, 1955), mirroring the relative aerodynamic roles of wings on the t w o segments (see section 3.2.1). I n insects w i t h high wingbeat frequencies (e.g., various Diptera and Hymenoptera), reduction i n the number of indirect muscles w i t h i n the mesothoracic segment is evident, suggesting functional con­ solidation of muscle action (see Brodsky, 1994). More generally, evo­ lutionary trends i n thoracic musculature have been described by M a ­ tsuda (1963a, b, 1970), w h o also emphasized homology of pterygote flight muscles w i t h locomotor muscles i n apterygote taxa. One recurrent biomechanical theme i n thoracic design is that use of indirect flight muscle to effect w i n g motions necessitates substantial cuticular bending. Dorsoventral muscles p u l l and deform the n o t u m ventrally, whereas action of the opposing dorsolongitudinal muscles reverses this action at frequencies often exceeding 100 H z (see chap­ ter 3). These muscle-induced deformations of the thorax are usually constrained morphologically. Various notai grooves (particularly the suture between scutum and scutellum; see fig. 2.1B) facilitate bending along prefixed lines, whereas the internally strengthened pleural su­ ture prevents excessive segmental deformation i n the transverse axis. Notai grooves are most evident i n orders w i t h pronounced notai flex­ ion during flight (e.g., Diptera and Hymenoptera; see Janet, 1899; Brodsky, 1994). Thoracic deformations are, b y contrast, limited by the phragmata that occur at intersegmental boundaries of adjacent notai sclerites. These planar structures r u n dorsoventrally, strengthening the pterothoracic segment i n the transverse plane as w e l l as serving as attachment points for dorsolongitudinal and dorsal oblique muscles. TABLE 2.1 (cont.) Elytra of some beetles are miniaturized and may function gyroscopically Anteromotorism characterizes the family Nemopteridae. J Hindwings may be absent in mymarid and mymarommatid Hymenoptera. Hindwings are vestigial in some Hydroptilidae. Wings are homonomous in some Trichoptera. Hindwings may be absent in ctenuchid (= syntomid) moths. Forewings are generally enlarged, but wings are homonomous in some Lepidoptera; hindwings can be enlarged relative to forewings, particularly among butterflies. ° Male strepsipterans have miniaturized forewings; females are wingless. h

1

k

1

m

n

48

C H A P T E R TWO

Ephemeroptera °

Odonata Plecoptera

°

Embioptera Orthoptera Phasmatodea

' ° Grylloblattodea Dermaptera °

Isoptera Mantodea Blattaria Zoraptera Thysanoptera Hemiptera

•

Homoptera Psocoptera

°

Phthiraptera

• Coleóptera n

Neuroptera

n

Megaloptera

°

Raphidioptera Hymenoptera Trichoptera

Locomotor Mode

3

aptery

''^x

^

jßL,

K

^

^

jf

equivocal

Strepsiptera

U

E w i l bimotorism anteromotorism posteromotorism r

Lepidoptera

T

^^^^^^km >k

^_

Diptera .

,

> ^ y ? ° Siphonaptera N¿\

^

n

n

Mecoptera

FIG. 2.4. Phylogenetic distribution of locomotor modes among pterygote orders. Note that the most parsimonious reconstruction using the pterygote phylogeny of fig. 1.1 identifies the ancestral locomotor mode as anteromotorism rather than as homonomous bimotorism (MacClade 3.0; see Maddison and Maddison, 1992).

M O R P H O L O G Y OF T H E F L I G H T APPARATUS

49

Segmental rigidity i n all taxa may be augmented by internal apodemes (chitinous projections) that also act as attachment points for muscles. Such apodemes, particularly pleural and sternal projections (e.g., the ventral furca of fig. 2.3B), are often joined by pleurosternal muscles. Contraction of these muscles can tighten the pleuron and w i n g base relative to the sternum (e.g., Nachtigall and Wilson, 1967; Kutsch and H u g , 1981), thereby altering the stiffness of the mechanically driven pterothoracic system. I n odonates, the use of dorsoventral muscles to act as both w i n g elevators and depressors necessitates additional mea­ sures to prevent thoracic compression, for w h i c h purpose a pro­ nounced scalariform (ladder-like) apódeme strengthens the pleuron dorsoventrally (Sargent, 1937; Russenberger and Russenberger, 1959, 1960; see fig. 2.3A). Mechanically driven systems of given stiffness and damping (inter­ nal resistance to motion) exhibit resonant oscillatory frequencies at w h i c h advantageous energetic expenditure is highest. A n y mechanical system can be driven at a frequency different from the resonant fre­ quency, but the d r i v i n g force is then out of phase w i t h motion of the structure or object i n question. The amplitude of oscillation is reduced considerably and substantial energy is dissipated i n nonuseful work. The presence of elastic recoil w i t h i n the thorax and antagonistic action between w i n g depressors and elevators suggest that the insect flight thorax is a mechanically resonant system driven periodically through muscular contraction (Pringle, 1949; Roeder, 1951; Danzer, 1956; Rus­ senberger and Russenberger, 1959, 1960; Greenewalt, 1960b). Muscles contract against the inertial load of the aerodynamically active wings and the mass of the muscles themselves. Intrinsic stiffness of the axil­ lary apparatus derives from the geometry of the w i n g articulation, the pterothoracic cuticle, and activity of direct flight muscles at the w i n g base. Tonic sternal and tergal (i.e., originating on the notum) muscles that insert on the pleural ridge are particularly w e l l suited for this lat­ ter purpose. Restorative elastic elements i n the flight apparatus i n ­ clude the thoracic cuticle as well as specific anatomical components of the flight muscle (section 3.3.2). The pterothorax thus contains those elements necessary for effective mechanical resonance. Wingbeat fre­ quencies of free-flying insects are characterized by fairly l o w coeffi­ cients of variation, and both interspecific and experimentally induced variation i n wingbeat frequency is consistent w i t h the hypothesis of mechanical resonance w i t h i n the thorax (see section 3.1.2). The presence of elastic elements w i t h i n the pterothorax is of particu­ lar importance for flight energetics. Such elements stretch at the end of a d o w n - or upstroke (i.e., at the end of any given half-stroke) and store w i n g kinetic energy as strain energy. This energy is subsequently

50

C H A P T E R TWO

released and contributes to the generation of subsequent half-strokes (Weis-Fogh, 1959, 1965, 1972). Such action is energetically important because a major component of power expenditure d u r i n g flight is the inertial energy required to accelerate the wings d u r i n g each half-stroke (section 3.3.2). The actual magnitude of elastic storage d u r i n g free flight is u n k n o w n for any insect, but elastic components w i t h i n the locust thorax have been systematically investigated (see section 3.3.2). Elastic storage b y biological materials is generally frequency depen­ dent (i.e., viscoelastic). A l t h o u g h time-dependent measurements of elastic storage by the integral flight apparatus have not been made, some energetic recovery of w i n g inertial energy (particularly b y elastic components inherent to the flight musculature) is likely i n all w i n g e d insects. The inertial energy of oscillating appendages increases w i t h the square of oscillation frequency, and insects w i t h h i g h wingbeat fre­ quencies likely exhibit substantial elastic storage w i t h i n the pterothorax (section 3.3.2). A l t h o u g h the insect flight apparatus might superficially seem less complicated than comparable vertebrate musculoskeletal systems, me­ chanical operation of the pterothorax is not understood i n detail for any insect group. The basic action of w i n g elevators (indirect muscles) and depressors (direct and indirect muscles) is not disputed, but de­ tailed action of muscles and their interaction w i t h the axillary sclerites to produce w i n g motions are much less clear. Complex pterothoracic mechanics arise from a h i g h number of interacting elements, as exem­ plified b y functional intricacies of the dipteran thorax. Sixteen sclerites articulate at the w i n g base between the n o t u m and the w i n g , w i t h as many direct control muscles (in addition to the indirect flight muscula­ ture) influencing sclerite position and axillary deformation (Ritter, 1911; Heide, 1971a,b; Dickinson and Tu, 1997). The dipteran thorax has been the target of classic investigations i n insect flight mechanics. Using anesthetized flies, Boettiger and Furshpan (1952) described a click mechanism whereby w i n g articulation and pterothoracic muscu­ lature could stably position the wings only at either dorsal or ventral extremes of the wingbeat, w i t h intermediate positions being elastically unstable due to the specific configuration of axillary sclerites. Wings were accelerated and then decelerated rapidly between these t w o ex­ treme positions, w i t h a waveform very different from the sinusoidal motion generally characteristic of simple resonant systems. This expla­ nation of w i n g movement, w i t h t w o stable extreme positions, per­ sisted for many years i n the specialized flight literature as w e l l as i n entomology textbooks. Recent w o r k suggests, however, that the click mechanism is likely an experimental artifact. M i y a n and E w i n g (1985b) showed that w i n g

M O R P H O L O G Y OF T H E F L I G H T APPARATUS

51

motions of tethered flies were inconsistent w i t h the click mechanism. Carbon tetrachloride was used on flies by Boettiger and Furshpan (1952) as an anesthetic, and action of this chemical may have i n ­ duced abnormal contraction of the pleurosternal muscles, altering the normal axillary configuration. The issue of normal operating config­ uration is i n fact central to the resolution of conflicting views on the functional action of the dipteran thorax. For example, diverse anatom­ ical studies of the dipteran w i n g base have suggested the presence of a mechanical stop on the pleural w i n g process that constrains ventral motion of the w i n g through physical contact w i t h a projection at the base of the radial vein (e.g., Pfau, 1973, 1987; Wisser and Nachtigall, 1984,1997a; M i y a n and Ewing, 1985a, 1988; Ennos, 1987; Wisser, 1987, 1988). I n stroboscopie video studies of tethered flies, however, N a l ­ bach (1989) rarely saw contact between the base of the radial vein and the proposed endstop of the pleural w i n g process. The kinematic re­ sults of Ennos (1987) also implicated rotational mobility of the lateral edge of the scutum i n control of w i n g movements, although no evi­ dence for such rotation was found i n tethered flies (Miyan and Ewing, 1988). Such contrasting explanations of pterothoracic operation cannot be systematically compared for at least t w o general reasons. Different fly taxa under variable tethering and experimental regimes evidently acti­ vate their wings differently, and wingbeat kinematics d u r i n g free flight may differ substantially from those i n tethered flight (see sec­ tion 3.1.2). For example, wing-tip motions of flies i n free flight closely approximate simple harmonic motion (Ellington, 1984c; Ennos, 1989c), a waveform not found i n aforementioned studies of tethered flies. Unitary explanations of pterothoracic operation are also unlikely be­ cause force output required from the flight motor can vary dramati­ cally w i t h airspeed and other dynamic demands, necessitating changes i n kinematic parameters such as wingbeat frequency and amplitude (see chapter 3). Because axillary sclerites are three-dimen­ sional (albeit small and flattened) structures, quantitative analysis of w i n g base motions requires image reconstruction comparable to that n o w available for w i n g - t i p motions (section 3.1.2). The best approach to evaluating pterothoracic and axillary movements d u r i n g flight w o u l d be to implement high-speed studies of wing-base motions i n free-flying insects; such a goal w o u l d be attainable using a h i g h degree of optical magnification and remote optomotor control of insect posi­ tion i n space (see chapter 5). Unfortunately, the intense illumination required for such studies can substantially disrupt the optomotor re­ sponses and natural flight behavior of subject insects under such circumstances.

52

C H A P T E R TWO

2.2 WINGS The thoracic structures responsible for generating useful aerodynamic forces are the wings, the often cambered surfaces that project laterally from pterothoracic segments. Venation serves to strengthen these sur­ faces against deleterious deformations and to facilitate advantageous w i n g geometries. W i n g modification and functional differentiation be­ tween pterothoracic segments define major features of ordinal-level i n ­ sect diversity. 2.2.1 Descriptive

Anatomy

Insect wings are thin cuticular structures that arise ontogenetically from specialized patches of cells (in many taxa, the imaginai disks) on developing pterothoracic segments. Alary buds, or external structures of differentiated pre-wing tissue, are evident i n larval stages of many exo- and endopterygote taxa. These winglike but nonarticulated struc­ tures are retained through consecutive larval instars (molts) u n t i l the final molt (for hemimetabolous insects) or emergence from the pupa (in holometabolous insects). A t this point, the incipient wings w i t h i n the alary buds expand into their final configuration. Expansion i n ­ volves hydrostatic use of the hemolymph (insect blood) w i t h i n the veins, i m p l y i n g involvement of the insect's circulatory system. Physio­ logical processes endogenous to the w i n g may also be involved, as even isolated w i n g buds expand autonomously (see Glaser and Vin­ cent, 1979). The wings remain soft and flexible d u r i n g expansion, but d r y out upon reaching their final geometry over a period of up to sev­ eral hours. A fully expanded w i n g consists of membranous regions of epider­ mal bilayers supported by veins. Extracellular cuticle layers expressed dorsally and ventrally from the epidermis determine the structural characteristics of the w i n g membrane per se. Wing veins are typically hollow and circular i n cross section, providing a conduit for nerves and hemolymph. Wootton (1992) emphasizes, however, that there are a large number of exceptions to this characterization; veins and other functional equivalents (e.g., cuticular thickenings on the wing) are best interpreted i n terms of their structural roles rather than morphological origins. Some veins contain no nerves and others contain no hemo­ l y m p h . Venational cross sections range from circular to oval and cam­ panulate (bell shaped). Some veins are flattened and consist merely of thickened regions of cuticle w i t h no intervening lumen between dorsal and ventral layers, whereas others are annulate structures of h i g h flex-

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

53

ibility. Substantial phylogenetic variation is also evident i n vein mor­ phology. I n Hemiptera (Heteroptera), for example, so-called channel veins consist simply of valley-like basins expressed either convexly or concavely on the w i n g surface (Betts, 1986a). These channel veins may or may not be delineated by ridgelike edges. Veinlike thickened struc­ tures of nontracheal origin may also serve to strengthen wings i n d i ­ verse orders (e.g., Wootton and Betts, 1986). Given such morphological diversity of veins, associated mechanical properties are likely to be highly variable. The flexibility of wings also relies on cuticular hydration of the wings; wings of desiccated insect specimens are notoriously fragile. The mechanical properties of veins and of wings generally are sustained i n part by circulating hemol y m p h (Arnold, 1964; Wasserthal, 1982; Wootton, 1992). Anteriorly w i t h i n the w i n g , veins are connected to the hemocoel of the body and are exposed to l o w positive pressure generated b y the insect heart. Apical motion of hemolymph w i t h i n the wings may also be facilitated by centrifugal pressures induced by w i n g flapping (see Larimer and Dudley, 1994). Posteriorly w i t h i n the w i n g , circulation is maintained by action of accessory pulsatile organs, structures of the insect circula­ tory system that are located at the w i n g base of all w i n g e d insects (see Krenn and Pass, 1994, 1994/95). These muscular pumps create suction that pulls hemolymph basally w i t h i n the posterior w i n g veins. Such action b y the accessory pulsatile organs may be physiologically neces­ sary i f the w i n g cuticle is to remain adequately hydrated. Patterns of w i n g venation are often highly complex and divergent even at the ordinal level; the phylogenetically most ancestral pattern of venation must remain speculative i n the absence of modern cladistic analysis (see Kukalová-Peck, 1991). A general scheme for w i n g vena­ tion is that of Wootton (1979; see fig. 2.5A), w h o emphasized the de­ scriptive character of the scheme w i t h o u t i m p l y i n g historical direc­ tionality. The major veins originate at the axillary apparatus, running distally and, i n some cases, toward the trailing edge of the w i n g (fig. 2.5A). Ordinal-level differences i n w i n g venation are largely defined by repeated bifurcations, anastomoses (merging of veins), and often loss of one or more of the major veins. Cross-venation and cutic­ ular thickenings between major veins and associated minor branches are widespread and morphologically diverse (see H a m i l t o n , 1972; Wootton, 1992). Most anteriorly, the leading edge of the w i n g is defined b y the costa and subcosta, thickened veins that provide structural rigidity (fig. 2.5A). The radius, a vein typically thicker than the costa and sub­ costa, extends and branches behind the subcosta, defining terminally the w i n g apex. The medial, cubitus, and anal veins project from the

54

C H A P T E R TWO

furrow

FIG. 2.5. ( A ) Generalized wing venation (modified from Wootton, 1979). Note that the medial vein bifurcates into the anterior media and the posterior media near the wing base, as does analogously the cubitus. (B) Major flexion lines and functional regions of the forewing.

w i n g base more posteriorly and largely define the trailing edge of the w i n g . The median flexion line represents a radial groove or region of increased flexibility along w h i c h the w i n g can deform and yield vari­ able camber (see fig. 2.5B). The median flexion line generally runs ante­ rior to the median vein, variably crossing secondary branches of the radial vein. A similar line of flexion, the claval furrow, lies radially between the cubitus and anal veins. Longitudinal bending about the claval furrow demonstrates functional partitioning between the claval region (or clavus) defined b y the posterior anal veins, and the much

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

55

larger anterior region termed the remigium (fig. 2.5B). A jugal fold separates the most posterior and basal region of the w i n g , the j u g u m , from the claval region. Some insect taxa, most notably the orthopterans, blattarians, mantodeans, and some plecopterans, exhibit basally expanded hindwings. This region is termed the vannus and an addi­ tional folding line, the vannai fold, then separates this region from the w i n g anteriorly. The jugal and vannai fold lines generally facilitate h i n d w i n g folding against the body. W i n g mass arises most immediately from the mass of w i n g vena­ tion; contributions of the membrane i n nontegminized wings is negli­ gible by comparison. The ratio of w i n g mass to body mass is highly variable i n insects, ranging from 0.5^1% i n typical dipterans and hymenopterans (Ellington, 1984b) to 3-10% i n butterflies (Betts and Wootton, 1988; Dudley, 1990). A l t h o u g h relatively light, wings and particularly the longitudinal distribution of w i n g mass are of mechani­ cal significance. Spanwise mass distribution determines the wing's moment of inertia and influences the potentially h i g h inertial power expended d u r i n g w i n g flapping (section 3.3.2). Mass tends thus to be concentrated near the w i n g base rather than at the w i n g tip. The center of w i n g mass i n dipterans and hymenopterans typically lies at 0.30.4R, where R is the w i n g length (see Ellington, 1984b), but can be apically shifted as h i g h as 0.44.R i n some butterflies (Dudley, 1990). Both span- and chordwise distributions of w i n g mass arise primarily from patterns i n venational distribution. Wing mass, vein diameter, and particularly endocuticular thickness of veins (see Banerjee, 1988b) de­ crease from base to t i p , indicating a gradient of w i n g stiffness. The con­ centration of veins at the w i n g base (fig. 2.5A) is primarily responsible for the greatest stiffness of the w i n g i n this region. The spanwise stiff­ ness gradient of wings is paralleled i n chordwise transects, w i t h the trailing edge of the w i n g being significantly more flexible than the leading edge. Chordwise mass distributions are much less studied than are spanwise mass distributions, but the greater number of larger veins at the leading edge of the w i n g (see fig. 2.5A) is primarily re­ sponsible for such variation i n stiffness. Parallel w i t h venational patterns, w i n g shape (usually considered to be the two-dimensional projection of w i n g area, or planform) exhibits considerable geometrical diversity (plate 2). The total w i n g area S i n ­ fluences the magnitude of aerodynamic forces produced by the flap­ p i n g wings (see eqs. 1.2 and 1.4); increased w i n g area yields increased forces i n most cases of steady-state as w e l l as unsteady flow. I n nonaccelerating flight, the average pressure exerted on the surrounding air b y the wings is given b y the w i n g loading (p ), the ratio of body weight (mg) to total w i n g area. W i n g area tends to increase w i t h the w

56

C H A P T E R TWO

square of linear body dimensions, whereas body mass is a general function of volume and increases w i t h the cube of linear dimensions. The ratio of mass (or body weight) to w i n g area thus tends to increase linearly w i t h body dimensions. W i n g loading is therefore generally higher i n insects of greater body mass, w i t h consequent implications for w i n g aerodynamic pressures and airspeeds d u r i n g flight (section 3.1.1). Also, insects of equivalent body mass can vary dramatically i n w i n g loading because of differences i n total w i n g area. For example, w i n g loadings i n bumblebee workers are typically 15 N / m (Dudley and Ellington, 1990a), whereas butterflies of comparable body mass exhibit w i n g loadings an order of magnitude lower (see Dudley, 1990). W i n g loading is thus a general comparative parameter (much as is the Re) that is of greatest significance w h e n used i n conjunction w i t h other measures of morphology. A complete description of w i n g shape necessitates, i n addition to knowledge of w i n g length R and w i n g area, t w o - and even threedimensional analyses of w i n g geometry. The w i n g aspect ratio AL (= 4JR /S) provides the simplest means of describing w i n g shape; a higher aspect ratio indicates a relatively more narrow w i n g . Aspect ratios of insect wings range from l o w values near 2 for a coupled w i n g pair of some butterflies (Dudley, 1990) to values of 10 or higher for certain odonate wings (see plate 2A). Aspect ratios of individual wings from otherwise coupled w i n g pairs (e.g., as i n many Lepidoptera) can be near unity. The aspect ratio indicates the ratio of the w i n g length to the mean w i n g chord but does not address spanwise distribution of w i n g area (i.e., area could be concentrated basally or distally for t w o wings of equal aspect ratio). This parameter must therefore be used i n conjunction w i t h additional analyses of the distribution of w i n g area i n both spanwise and chordwise dimensions. For a flapping w i n g , the local velocity of any given w i n g section varies linearly w i t h the radial distance from the w i n g base. More dis­ tal w i n g sections experience higher relative air velocities and thus generate greater aerodynamic force per unit area. The spanwise distri­ butions of w i n g area have been w e l l studied through the efforts of Ellington (1984b; see also Magnan, 1934; Weis-Fogh, 1972, 1973), w h o determined such distributions on diverse insect wings. The center of w i n g area (the first moment of distribution for the w i n g area) is t y p i ­ cally located at 0.4-0.6R (e.g., Ellington, 1984b; Betts, 1986a; Ennos, 1989; Dudley, 1990). A l t h o u g h the mass of w i n g veins is h i g h relative to that of the intervening membrane, distributions of w i n g mass and w i n g area are tightly coupled i n many insects (see Ellington, 1984b). A n additional morphological parameter derived from the w i n g area distribution is the virtual mass of the w i n g . As a w i n g section acceler2

2

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

57

ates, a volume of air around the w i n g is simultaneously accelerated. The mass of this air volume, the virtual mass (v), is proportional to the chord length of the w i n g section and the air density (Ellington, 1984b). Wing area distributions determine the spanwise location of the mean w i n g chord and thus influence virtual mass distributions. The virtual w i n g mass can be comparable i n magnitude to the w i n g mass itself at high wingbeat frequencies, and thus represents an additional compo­ nent of w i n g inertia resisting acceleration by the flight muscle (see sec­ tion 3.2.2.1). Diverse ultrastructural features can be found on the surface of many wings (Wootton, 1992). Major veins may be supported b y small mem­ branous brackets projecting vertically from the w i n g surface. Sensory structures, particularly the mechanoreceptive campaniform sensillae (section 5.1.1), are found along vein surfaces. Various spines and t r i chiae (hairs) similarly relieve the smoothness of venation. Surface texture of some heteropteran hindwings resembles minute pointed cones, and is of no k n o w n functional significance (Betts, 1986a). Many insects, particularly Lepidoptera (Downey and A l l y n , 1975) but also some Trichoptera (Huxley and Barnard, 1988) and to a lesser degree several other orders, possess flattened epidermal cells called w i n g scales. Most scales are utilized primarily i n nonaerodynamic roles, particularly for coloration, but may subtly influence flow patterns and boundary layer structure over wings (see below). Finally, many insects possess discrete pigmented regions (pterostigmata) near the leading edge of the w i n g that are heavier than the w i n g membrane and the veins i n the immediate vicinity of pigmentation. Pterostigmata are found on both fore- and hindwings i n Odonata and Mecoptera, but also on the forewings of certain Homoptera, Hymenoptera, Neuroptera, and Psocoptera. 2.2.2 Functional Morphology Wings are subjected to aerodynamic and inertial forces d u r i n g flap­ ping that induce deformation, bending, and torsion. Venational pat­ terns correspondingly appear to reduce and control the extent of w i n g bending i n both transverse and longitudinal planes (Wootton, 1992). Venational mass and vein density is highest at the base (see fig. 2.5A) where w i n g bending moments d u r i n g flapping are greatest. This proximal concentration of w i n g mass also reduces the wing's moment of inertia and the associated energy required for angular acceleration (see section 3.3.2). The three-dimensional configuration of veins pro­ jecting from the membranous surface also serves a mechanical func­ tion. I n a chordwise transect from leading to trailing edge of the w i n g ,

58

C H A P T E R TWO

corrugation i n the form of alternating furrows and ridges is evident as veins. Corrugation tends to be more pronounced toward the w i n g base and i n anterior regions of the w i n g (Rees, 1975a, b). Veins alter­ nately project dorsally and ventrally from the surface, yielding a pat­ tern of convex and concave veins that transects from leading to trailing edge. The net result is a corrugated three-dimensional structure that is of higher flexural stiffness than a flat beam of equivalent mass (Rees, 1975b). Such corrugation, however, is effective, only i f chordwise flat­ tening and bending can be prevented b y the presence of cross-veins and similar mechanical connections that r u n chordwise (see N e w m a n and Wootton, 1986). Mechanically effective venation thus requires both spanwise and chordwise components. The presence of both com­ ponents is most evident i n wings w i t h reticulate (meshlike) venation (e.g., wings of Odonata and Neuroptera). Venational patterns may also mitigate damage d u r i n g collisions of wings w i t h their contralateral counterparts or w i t h external objects. Typically, wings flex and yield i n such collisions. For example, flatten­ ing of corrugations and chordwise bending characterizes impacts of dragonfly wings w i t h enclosure walls (Newman and Wootton, 1986). Quantitative description of the particular strengthening afforded b y vein geometry has recently been made possible through application of finite element analysis (FEM) to w i n g structures. This engineering method spatially decomposes three-dimensional structures into a large network of interacting elements, and enables predictions of w i n g deformation i n response to applied force w h e n elastic features of con­ stituent elements are specified (e.g., Kesel, 1997). This method can be used not only to predict effects of particular veins and their geometry, but also to explore biomechanical consequences of venational designs not realized i n nature. I n addition to the veins, the w i n g membrane itself functions i n mechanical stiffening, as evidenced b y Wagner ten­ sion fields w i t h i n cells of stressed dragonfly wings (Newman and Wootton, 1986). These fields are manifested as folds induced i n the membrane that are parallel to the major axis of imposed tension. Such folding indicates force transmission through the w i n g membrane, demonstrating service i n structural as w e l l as aerodynamic roles. The functional design of wings illustrates design based on structural hierarchy. The flapping w i n g itself is comprised of a venational net­ w o r k and associated membrane, whereas the particulars of vein geom­ etry and cross-connections w i t h other veins influence patterns of de­ formation on smaller spatial scales. This deformation i n t u r n derives from the local stiffness of the constructional material, a function that ultimately emerges from molecular composition and organization. Relative to apterygote cuticle, a feature specific to pterygote w i n g

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

59

veins is the presence of both helicoidal and parallel layers of cuticle (Neville, 1993). W i t h i n helicoidal layers, microfibrils exhibit continu­ ous change i n orientation across a cuticular transect, whereas parallel layers of cuticle are characterized by unidirectional orientation of chit i n microfibrils. Such varying orientation of chitin fibrils i n insect w i n g veins is attributed to the need to resist both bending and torsion of venation d u r i n g flight (Neville, 1984; see also Banerjee, 1988a). Such microstructural arrangements are probably widespread i n w i n g design. The claval furrow of locusts, an axis of extensive bending during the upstroke, possesses larger chitin microfibrils than are found elsewhere on the w i n g , presumably i n response to increased mechani­ cal demands d u r i n g the upstroke (Banerjee, 1988a). Biochemical fea­ tures of cuticular construction may influence w i n g durability, as Jacobs (1985) showed that injection of a key cuticular amino acid (betaalanine) into newly eclosed Drosophila increased the puncture resis­ tance of wings. One of the less-appreciated points i n flight mechanics is the number of cycles over w h i c h wings reciprocate i n the course of an insect's lifetime. Mean forces comparable i n magnitude to the body weight are applied w i t h each cycle, and the h i g h wingbeat frequencies of most insects ensure at least some alary degradation. I f only 5% of a typical five-week life span of an adult Drosophila is utilized for flight at a wingbeat frequency of 200 H z , then the wings undergo i n excess of 30 million reciprocations. Interestingly, about 23 m i l l i o n wingbeats were obtained i n a tethered simulation of long-duration flight using a single Drosophila melanogaster (Götz, 1987). Mechanical changes i n w i n g structure under such intensive cycling are not k n o w n , but w i n g damage among insects, not surprisingly, does increase w i t h age. The more flexible trailing margin is particularly susceptible to tearing. Flight performance of insects also declines w i t h age, although this ef­ fect is matched i n part b y parallel deterioration of the flight muscle (see section 4.2.4). I n general, effective airfoils are moderately cambered structures that operate at l o w angles of attack relative to oncoming airflow (see sec­ tion 3.2). The otherwise well-cambered design of most wings w o u l d superficially seem to be compromised b y the extensive venational cor­ rugation along chordwise transects. A i r flowing around the w i n g can, however, be trapped i n the venational folds, forming recirculating bubbles that alter the effective aerodynamic contour of the w i n g (see section 3.2.1.2). A n irregular w i n g surface can thus be as functionally effective as is a smooth airfoil. I f viewed end-on, wings from larger insects often exhibit a geometrical twist along the w i n g span, w i t h the w i n g chord becoming more pitched ventrally toward the w i n g tip. Be­ cause the relative air velocity of flapping wings increases linearly w i t h

60

C H A P T E R TWO

distance from the w i n g base, the direction of this vector also changes to become more coincident w i t h the plane of the beating wings further along the wingspan. Intrinsic w i n g twist maintains an approximately constant chord orientation and angle of attack along the wingspan. Be­ cause lift and drag on wings can vary considerably w i t h angle of attack (section 3.2.1.2), a constant w i n g chord relative to oncoming air results i n advantageous aerodynamic characteristics along the length of the w i n g . I n contrast to large wings, spanwise w i n g twisting is less pro­ nounced i n smaller wings that operate at fairly l o w Re, presumably because of a reduced orientational dependence of force production i n more viscous flows (see section 3.2.1.2). W i n g corrugation, camber, and spanwise twist are not morphologi­ cally invariant but rather vary w i t h inertial and aerodynamic loading. Wootton (1981b) has argued that w i n g structural features l i m i t associ­ ated three-dimensional deformation and help to regulate aerodynamic shape; insect wings can thus act as adaptive airfoils. For example, cam­ ber is automatically induced i n the vannus of the orthopteran h i n d w i n g as the downstroke proceeds (Wootton, 1995). The precise aerody­ namic consequences of such camber induction are not k n o w n , but an increase i n force production is likely, at least at higher Re. The corru­ gated design of w i n g venation appears predisposed to effect such shape changes w i t h o u t substantial bending. Using physical and theo­ retical models of wings, Ennos (1988a) showed that branching of cor­ rugated spars from a thickened leading edge results i n h i g h resistance to bending but substantial torsional flexibility. Externally imposed aerodynamic and inertial forces may, via this mechanism, induce flex­ ion about the w i n g base and facilitate advantageous changes i n an­ gular orientation of the w i n g . The extent of such flexion depends on the torsional stiffness of the w i n g and the location of the wing's rotational axis relative to the point of an externally applied force. The rotational axis usually lies near the leading edge of the w i n g , whereas aerodynamic forces typically act behind both this axis and the chordwise center of w i n g mass. Aero­ dynamic forces are thus not coincident w i t h the wing's rotational axis and tend not only to alter w i n g camber but also to twist the w i n g (Ennos, 1988a). I f steady-state conditions apply, then aerodynamic torque is greatest i n the middle of a half-stroke when the wing's rela­ tive velocity is highest. This torque w i l l have the effect of inducing w i n g pronation d u r i n g the downstroke, and provides for indirect aerodynamic regulation of w i n g orientation (Ennos, 1989b). Further­ more, a force on the concave side of a cambered plate w i l l , i n addition to bending the plate, increase camber i f the point where the force is applied is not along the longitudinal axis of the structure (Ennos,

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

61

1995). Aerodynamic forces on flapping wings are imposed ventrally (on the concave w i n g surface) d u r i n g the downstroke, and w i l l thus enhance w i n g camber. This increased camber w i l l at the same time re­ sist further w i n g torsion and deformation, yielding an equilibrium (and perhaps aerodynamically advantageous) profile for a deformable wing. By contrast, forces applied to the convex surface of a cambered w i n g can enhance both longitudinal bending and w i n g twisting (Ennos, 1995). Such forces are applied aerodynamically through the upstroke and d u r i n g supinatory rotation. Wings should therefore exhibit more flexion during supination than d u r i n g pronation, as has been con­ firmed by static bending tests on butterfly wings (Wootton, 1993). Both camber and angular orientation of flapping wings can thus be influ­ enced by the force and direction of the instantaneous aerodynamic force vector. This method of aerodynamic control arises passively from the interaction of airflow w i t h w i n g geometry, and is distinct from active muscular torques applied at the w i n g base. N o t all insect wings conform to this model, however. The torsional resistance of dragonfly wings is higher than stiffness i n bending (Zeng, Matsumoto, et al., 1996), and passively induced aerodynamic changes i n w i n g pro­ file are not likely to be pronounced. Wing profile i n odonates is instead influenced b y the action of a small muscle (the fulcroalar muscle) that is integral to the axillary complex at the w i n g base. This muscle acts as the base of the cubitus vein and is used to control w i n g profile (New­ man and Wootton, 1988) i n addition to the changes i n w i n g contour caused by the direct downstroke muscles. Whereas steady-state aerodynamic forces predominate i n the m i d ­ dle of a half-stroke, inertial forces are greatest at the ends of halfstrokes w h e n the w i n g mass and virtual mass first decelerate and then reaccelerate i n the opposite direction. Transverse inertial bending of wings is most likely at either end of the wingbeat w h e n first angular deceleration and then acceleration are highest. A t the ventral extremes of wingbeats, transverse bending associated w i t h inertial deceleration is often pronounced and may be facilitated b y a transverse flexion line. Flexion i n the posterior region of the w i n g and bending at the w i n g t i p may mitigate inertial bending at the w i n g base. I n dipterans and hymenopterans, the extent of bending at the w i n g t i p is generally less than 5% of total w i n g length (e.g., Ellington, 1984c), although other taxa (e.g., Panorpa) often exhibit a much more pronounced t i p deflec­ tion (see Dalton, 1975; Brackenbury, 1992). Substantial spanwise de­ flection is uncommon at the end of the upstroke, at w h i c h point the positive camber and corrugated venation of the w i n g resist ventro­ dorsal bending.

62

C H A P T E R TWO

Not only w i n g bending but also twisting and rotation may be i n ­ duced b y w i n g inertial forces at the ends of each half-stroke. Norberg (1972c) first noted that the chordwise center of w i n g mass typically lies behind the rotational axis of the w i n g , and that w i n g deceleration at the end of half-strokes w o u l d tend to swing the w i n g mass chordwise about the longitudinal axis. Inertially induced rotation of different spanwise w i n g regions w i l l vary according to the local section mass and to the distance between the rotational and inertial axes. Such rota­ tion w i l l be most pronounced for those regions closest to the torsional axis of the w i n g (e.g., the distal regions of fly wings; Ennos, 1988b). To alleviate potentially excessive inertial torques, the center of w i n g mass must be located more anteriorly and closer to the wing's rota­ tional axis. This requirement is met b y a concentrated mass along the leading edge of the w i n g that is located distally near the w i n g tip; the pterostigmata of wings from diverse taxa is of sufficient mass and i n the appropriate location to passively regulate w i n g orientation (Nor­ berg, 1972c). The relative size of hymenopteran pterostigmata i n ­ creases w i t h decreased body size, possibly i n response to the greater torsional inertia associated w i t h the higher wingbeat frequencies that come w i t h small body size (Danforth, 1989; see also chapter 3). I n bee­ tles w i t h folding h i n d w i n g apices, the pterostigma at the base of the fold may inertially augment w i n g supination (Brackenbury, 1994b). M a n y insect taxa have no obvious pterostigmata. For all w i n g e d i n ­ sects, however, location of the center of w i n g mass behind the wing's rotational axis indicates that inertia w i l l promote supination at the end of the downstroke. Similarly, w i n g inertia w i l l advantageously induce pronation at the end of the upstroke, although the location of the relevant mechanical axes is more difficult to specify for a w i n g already variably twisted about its base i n supination. W i n g orientation through the wingbeat and particularly t i m i n g and extent of rotation are also influenced b y action of the pterothoracic musculature. Inertial regulation of w i n g ori­ entation is, however, probably more pronounced i n those insects w i t h h i g h wingbeat frequencies. Ennos (1988b, 1989b) suggested that w i n g inertia at stroke reversal was sufficient by itself to generate the rota­ tions observed i n flies. These calculations of w i n g inertia, however, i n ­ volved formulations for the virtual w i n g mass that may not ade­ quately incorporate the complex accelerations associated w i t h rotating wings (see section 3.2.2). Forward flapping flight necessitates force production that is asym­ metric between d o w n - and upstroke (section 3.1.2), and a variety of kinematic strategies are employed toward this goal. Direct modifica­ tions of w i n g geometry include changes i n the angle of attack (i.e.,

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

63

w i n g twisting) and variation i n w i n g camber and profile. Differences i n w i n g orientation between the d o w n - and upstroke are particularly marked i n the flight of taxa w i t h only one actual or functional w i n g pair (e.g., Diptera, Hymenoptera). One indirect consequence of w i n g supination is the reduction of the effective w i n g surface area d u r i n g the upstroke. Twisting and reversal of w i n g orientation indicate that basal regions of the w i n g w i l l be less effective aerodynamically during the upstroke. I n hindwings, this effect can be enhanced by the pres­ ence of the vannus, w h i c h mechanically folds d u r i n g the upstroke to reduce the total effective w i n g area (Wootton, 1992). Similarly, the sense of w i n g camber (i.e., convex or concave) often reverses from d o w n - to upstroke. Bending along flexible cross-veins as w e l l as along the median flexion line facilitates camber reversal at the transition be­ tween half-strokes. Because of the dorsoventral asymmetry i n the de­ sign of biological wings and their basal connection, no animal w i n g is fully reversible between half-strokes i n terms either of camber or effec­ tive surface area. As a consequence, the upstroke is inevitably less ef­ fective aerodynamically than the downstroke. Variation i n w i n g profile between half-strokes is largely effected via the radial flexion lines, most importantly the claval furrow and the me­ dian flexion line (Rohdendorf, 1958/59b; Wootton, 1979; Grodnitsky and Morozov, 1994; see fig. 2.5B). Radial bending along these lines at stroke reversal generates a "Z"-shaped cross-sectional profile during the upstroke, first described i n the forewing of the locust (Jensen, 1956; see also Pfau, 1977,1978; Nachtigall, 1981a) but characteristic of many other insects (Wootton, 1992). Note, however, that a diverse cross sec­ tion of profiles is subsumed w i t h i n the concept of a " Z " shape, and upstroke profiles likely vary according to spanwise location and the particular force output required under different flight conditions. For example, claval flexion can be extensive d u r i n g w i n g supination i n hovering hymenopterans (see Brackenbury, 1994a) but decreases at higher airspeeds as rotational angles decrease. Aerodynamic charac­ teristics of the "Z"-shaped upstroke profile have been investigated only i n locusts (Jensen, 1956; Nachtigall, 1981a,b), but i n general such flexion and associated profile change render the w i n g less effective i n lift production. Bending along the median flexion line may also result i n advantageous vortex shedding from the w i n g i n an unsteady flex mechanism (Ellington, 1984d; see section 3.2.2.3). Evolutionary change i n patterns of upstroke flexion may be associated w i t h enhanced ma­ neuverability and flight performance. I n a mecopteran, for example, the w i n g base assumes a "Z"-shaped profile through a longitudinal flexion about the claval furrow (Ennos and Wootton, 1989). By con­ trast, the more agile dipterans (the sister taxon of Mecoptera; Wootton

64

C H A P T E R TWO

and Ennos, 1989; see fig. 1.1) extensively supinate the wings and re­ verse w i n g camber i n the upstroke, exhibiting neither claval flexion nor an irregular w i n g profile. I n contrast to forward flight, hovering demands a nearly complete reversal of w i n g camber and orientation i f consecutive half-strokes are to be aerodynamically symmetric w i t h no net thrust production. The magnitude of torsional reversal is constrained b y the size of the w i n g articulation relative to the w i n g chord. Particularly for insects w i t h linked fore- and hindwings, reversal of w i n g orientation between halfstrokes is structurally impeded, as exemplified by many butterflies. N o t surprisingly, hovering is usually limited to insects w i t h highly flexible wings that effectively reverse camber and exhibit neither van­ nai folding of hindwings nor a pronounced claval region. Even so, re­ duced aerodynamic output i n the upstroke is likely. I n support of this hypothesis, w i n g elevator muscles tend to be smaller than depressor muscles i n insects, although systematic data relating to this observa­ tion have not been collected i n insects. I t is noteworthy, however, that even i n hummingbirds the main elevator muscle (the supracoracoideus) is approximately one-third the mass of depressor muscles (Greenewalt, 1962) i n spite of the approximately symmetric w i n g mo­ tions and near-perfect reversal of w i n g camber and orientation exhib­ ited b y these birds d u r i n g hovering (see section 7.5.2). W h e n at rest, paleopterous insects h o l d their wings either laterally outspread (Odonata: suborder Anisoptera) or obliquely above the body (Odonata: suborders Anisozygoptera and Zygoptera, Ephemeroptera). Neopterous insects at rest typically fold their wings on or over the abdomen, i n some cases utilizing radial plication i n the jugal and vannai regions (i.e., fanwise folding of the membrane along radial lines) to reduce exposed w i n g surface area against the body. The ex­ panded vannus of the orthopteran h i n d w i n g provides the best exam­ ple of such plication (Wootton, 1995; see fig. 2.6B), and such folding strategies are generally characteristic of orders w i t h tegminized fore­ wings. Radial pleating is also used b y some Hymenoptera to fold the forewings w h e n the insect is at rest (Danforth and Michener, 1988). The longitudinal w i n g span is unaffected by such folding, and i n many neopterous insects the wings correspondingly lie above and extend substantially beyond the abdomen. I n some orders, however, transverse folding is used to dramatically reduce w i n g length. Elytrization i n part demands such transverse fold­ i n g i f the hindwings are to be effectively concealed beneath the elytra. The Coleóptera, for example, have elaborate means for both radial and transverse w i n g folding that position the hindwings beneath the smaller elytral forewings (fig. 2.6E). Similar transverse folding char-

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

65

acterizes the Blattaria and Dermaptera (Scudder, 1876; Kleinow, 1966; Haas and Wootton, 1996). Because wings lack the intrinsic muscula­ ture necessary to effect such transverse bending, indirect forces must instead be used to carry out such unfolding and subsequent refolding. The well-defined creases i n the w i n g membrane act together w i t h the inherent elasticity of the w i n g to move the w i n g tip outward when suitable torsion is applied at the w i n g base (Haas and Wootton, 1996). Some fully unfolded wings remain stably unfolded w i t h o u t active maintenance of tension, whereas others are unstable and must remain forced open. Some of the necessary tension i n this latter mode may be provided by the aerodynamic and inertial forces of w i n g flapping (Haas and Wootton, 1996). Thus, elytrization of forewings is necessar­ ily associated w i t h evolution of sophisticated structural mechanisms i n the h i n d w i n g , as exemplified by elaborate folding mechanisms i n the beetles (see section 8.3.2). One apparently unique structural modification for w i n g folding is found w i t h i n the pterophorid (plume) moths, many of which, when at rest, roll ipsilateral fore- and hindwings together into a tubular struc­ ture of small diameter (Wasserthal, 1974). This peculiar manner of w i n g folding about the longitudinal axis may have evolved to escape detection by visually oriented predators. Interestingly, some ptero­ p h o r i d species as w e l l as the unrelated orneodid moths exhibit highly lobed fore- and hindwings that often are fringed w i t h fine hairs. N o kinematic adaptations particular to the extraordinary w i n g structure of pterophorids are evident, and adjacent w i n g lobes appear to form a continuous surface (Norberg, 1972b). Wing fringing is also characteris­ tic of many small dipterans, coleopterans, and hymenopterans (as w e l l as thrips and microlepidopterans), and has the effect of creating an aerodynamically continuous w i n g (see section 7.1.3). The aerodynamic implications of ornamentation on w i n g structure are largely uninvestigated. A tremendous diversity of scales, spines, protrusions, and varied surface textures can be found on wings of dif­ ferent taxa. Whatever their ultimate function, most of these structures may simply yield viscous drag on the w i n g surface. Some microstruc­ tural features may, however, subtly interact w i t h a turbulent boundary layer to reduce drag. Lift characteristics may also be affected, and an unresolved issue i n this field concerns the potentially advantageous aerodynamic presence of scales on wings of butterflies and other Lepi­ doptera (see section 3.2.1.2). The presence of scales is clearly not re­ quired to effect flight i n Lepidoptera, as many moths (e.g., various Sesiidae; Kristensen, 1974) and some butterflies (e.g., the clearwing ithomiine genera, the satyrine genera Cithaerias, Dulcedo, and Haetera) exhibit only m i n i m a l scalation along the w i n g margins. Longitudinally

66

C H A P T E R TWO

striated w i n g scales can also be found i n various Megaloptera, Psocoptera, Trichoptera (the sister taxon to Lepidoptera), and even i n some Coleóptera and Diptera. Scales are also abundant on the bodies of both Trichoptera and Lepidoptera, but their aerodynamic significance, i f any, is u n k n o w n . 2.2.3 Phylogenetic Diversity The phylogenetically ancestral pterygote condition is usually assumed to have been meso- and metathoracic homonomous wings of approxi­ mately equivalent size, shape, and function. Using the present-day dis­ tribution of pterothoracic character states, however, the most parsi­ monious reconstruction of the ancestral w i n g configuration shows i t to be anteromotoric i.e., flight derives predominantly from action of the forewings (fig. 2.4). H o m o n o m y evidently characterized a number of Paleozoic orders (e.g., Paleodictyoptera, Protodonata), but homon­ omy of pterothoracic segments and associated bimotorism (equiva­ lent action of fore- and hindwings) is infrequent among extant insects. Instead, modification of one pterothoracic segment results i n either antero- or posteromotorism, and i n some cases a transformation from tetraptery (expression of four wings) to either functional or actual diptery (table 2.1). A n y presumed ancestral condition of homonomy has clearly been modified many times d u r i n g the evolution of w i n g e d insects. Antero- and posteromotorism occur at approximately equal fre­ quencies among the insect orders, although posteromotorism is ( w i t h the exception of the Coleóptera and Strepsiptera) confined to the exopterygotes (fig. 2.4). I n most extant insect orders, one w i n g pair is ac­ cordingly either reduced i n relative size or is transformed morpholog­ ically to serve i n various nonaerodynamic roles. Evolution of alary heteronomy (wings of different function) is thus a general trend i n i n ­ sect biology. Such transformations are associated w i t h changes i n the relative size and shape of ipsilateral wings i f both w i n g pairs retain aerodynamic function. Brodsky (1994) noted that posteromotorism tends to be associated w i t h broadening of the hindwings at the base, whereas those insects characterized by bimotorism tend to have wings w i t h elliptical planform. Anteromotoric insects, by contrast, often dis­ play h i n d w i n g reduction (see below). A l t h o u g h a dominant theme i n pterygote evolution, such trends i n w i n g shape and size have not been rigorously quantified, nor have patterns of intraordinal w i n g heter­ onomy been systematically evaluated. The most common such transformation is the use of a strengthened mesothoracic w i n g pair i n defensive actions. Forewings of most insects

M O R P H O L O G Y OF T H E F L I G H T APPARATUS

67

tend to be slightly thickened relative to the hindwings (Rohdendorf, 1949), but this effect is most pronounced i n tegminization and elytrization, and typically involves a reduction i n area but an increase i n cutic­ ular strength and rigidity of the forewings. Tegmina (characteristic of the Orthoptera but also found i n other orders; see table 2.1) tend to be thickened leathery structures w i t h a reduced aerodynamic role rela­ tive to the hindwings (plate 2C, D). Elytra of Coleóptera are an autapom o r p h y (a unique derived character) of the taxon; elytra are generally much tougher than tegmina and i n many cases are r i g i d (plate 2E). W i n g venation tends to be reduced or essentially eliminated from both tegmina and elytra. Elytrization of the forewings pair is also associated w i t h venational modifications of the metathoracic pair to enable fold­ ing either radially or transversely beneath the protective elytra. Posteromotorism and locomotor specialization of the hindwings is a necessary consequence of such mesothoracic specialization (table 2.1). I n the Coleóptera, the elytral forewings play a small aero­ dynamic role whereas the hindwings are the major force producers (see section 3.2.1.2). Elytra are typically w e l l sclerotized and func­ tion primarily as sheathlike devices to protect the hindwings and abdomen. When the insect is at rest, the paired elytra meet dorsally at the midline and typically lock together at an elytral suture or fit into notai grooves. A similar modification characterizes some Mecoptera (Hlavac, 1974) as w e l l as various Dermaptera and Hemiptera. Such linkage contributes to the overall mechanical rigidity of the pterotho­ racic complex w h e n the insect is not i n flight. Ventral surfaces of most beetles are also heavily sclerotized, rendering the entire body resistant to crushing. Few data exist on the mechanical protection afforded by elytra, although Hough-Goldstein et al. (1993) found that a chrysomel i d beetle w i t h these structures removed was less likely to survive pecking b y domestic fowl (see also Thiele, 1977). Variation i n microstructural design undoubtedly contributes to the w i d e range of elytral rigidity and toughness evident i n different beetle taxa. Krzelj (1969) found that elytral flexibility varied primarily w i t h thickness and rela­ tive size of the exo-to endocuticular layers, although variation i n mate­ rial stiffness (i.e., Young's modulus) may also play a role (Krzelj and Jeuniaux, 1968; see also Hepburn and Ball, 1973). The elytral conver­ sion of mesothoracic beetle wings has been correlated w i t h an anterior shift of the h i n d w i n g base, perhaps to compensate for reduction i n torque about the center of body mass w h e n the forewings no longer have a major aerodynamic function (Crowson, 1981). Also, the de­ mands of pterothoracic cuticularization probably operate at crosspurposes to thoracic deformations induced by indirect flight muscles. The indirect dorsolongitudinal muscles are correspondingly reduced

68

C H A P T E R TWO

i n many beetle families, and direct basalar muscles are the primary means of effecting w i n g depression (Larsen, 1966). Elytriform wings of other orders also serve nonaerodynamic pur­ poses. The hemelytra of Hemiptera (Heteroptera) represent intermedi­ ate conversion of the forewing from an aerodynamic to a protective structure. Interestingly, the membranous and coriaceous (leathery) re­ gions of the hemelytron do not merge smoothly into one another but rather meet at a distinct junction (see plate 2B). Elytrization of hemipteran forewings probably represents a compromise between mechani­ cal protection and aerodynamic force production, as hemelytra are linked i n flight to the hindwings and remain aerodynamically func­ tional (Puchkova, 1971; Betts, 1986a; Wootton and Betts, 1986; Woot­ ton, 1996). I n many heteropteran lineages w i t h scutella that extend the full length of the abdomen (e.g., Scutelleridae, Thyreocoridae, some Pentatomidae), the hemelytra are essentially elytral i n character and interface w i t h the scutellum to result i n a nearly continuous dorsal protective surface. Also contributing to mechanical protection are the partially or totally sclerotized forewings of Blattaria, Dermaptera, Orthoptera, Phasmatodea, and some Homoptera. Forewings are also slightly coriaceous i n many Neuroptera, Mecoptera, and Megaloptera. I n tetrigid orthopterans, the tegmina are vestigial but the p r o n o t u m (the n o t u m of the prothoracic segment) extends posteriorly to cover the hindwings as w e l l as much of the abdomen. The extinct orders Glosselytrodea and Protelytroptera were also characterized by robust elytra (Carpenter, 1992). I n addition to elytrization and tegminization, morphological strate­ gies have evolved that enhance cuticular defense of the thorax and abdomen, most notably the various posterior extensions of the meta­ thorax i n the membracid homopterans, and a remarkable shieldlike pro­ jection of the scutellum i n the dipteran family Celyphidae (the beetleflies). Use of such static morphological defense against predators also has behavioral correlates. M a n y insects w i t h tegmina or elytra are behaviorally more reluctant to fly than are other insects, or they tend to use w i n g flapping only as an adjunct to a j u m p (e.g., the saltatorial, or jumping, Orthoptera). Flight may i n fact be energetically more costly i n taxa w i t h thickened forewings. The added cuticular weight of elytri­ form wings and associated body sclerotization (cuticular hardening) may increase substantially the mechanical power expended d u r i n g flight, but the general theme of pterothoracic modification for defense has nonetheless been widespread i n insect evolution (table 2.1). Anteromotorism (i.e., h i n d w i n g reduction) represents the second form of pterothoracic differentiation. Flight using only one w i n g pair is true diptery, and at the ordinal level is characteristic only of the

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

69

Diptera and Strepsiptera (see table 2.1; plate 2G). More typically, the h i n d w i n g is reduced i n size and is mechanically coupled to the forew i n g , a functional form of diptery (Chadwick, 1940; Grodnitsky, 1995) first evident historically i n the Permian paleodictyopteran taxon Permothemistida (Brodsky, 1981). A variety of insect orders display functional diptery (see Table 2.1), as exemplified by the linked wings of contemporary Hymenoptera (Gauld and Bolton, 1988; see plate 2F). Quantitative data on relative masses and areas of hymenopteran wings are unfortunately lacking. Some Ephemeroptera (e.g., the Caenidae) exhibit extreme h i n d w i n g reduction (although the fore- and hindwings remain uncoupled), and i n some cases the hindwings are completely lost. A n extreme case of h i n d w i n g reduction is miniaturization of the w i n g pair, as i n the dipteran haltères used for stabilization (section 5.1.1). The general trend i n anteromotorism, however, tends to be overlapping or physical coupling of the relatively smaller h i n d w i n g to the forewing to yield one continuous aerodynamic surface (e.g., Lepidoptera). Conversely, some insect orders demonstrate an increase i n h i n d w i n g dimensions relative to the forewing, although this trend is n u merically less significant than is relative expansion of the forewing (table 2.1). The phylogenetically basal order Odonata exhibits limited expansion of the hindwings w i t h i n the suborder Anisoptera. H i n d wings are expanded relative to forewings i n the Orthoptera (see plate 2B) and the related Phasmida, i n Blattaria and its sister taxon Mantodea, i n some Plecoptera, and i n the mastotermitid termites (Wootton, 1992). I n general, the expanded basal region of such h i n d wings precludes substantial rotation about the longitudinal w i n g axis, and correspondingly prevents a reversal of w i n g orientation between down-and upstroke. Such insects are generally unable to hover and display only limited maneuverability i n the air. Diversity of w i n g coupling mechanisms parallels phylogenetic pathways i n differentiation of pterothoracic segments. Nachtigall (1974a) discusses the general physical principles behind biological connectors and illustrates basic mechanisms for wings. I f wings overlap and operate i n phase (but w i t h no specific mechanical connection), the coupling is termed amplexiform. Amplexiform wings are characteristic of mayflies (Ephemeroptera) and butterflies (Bourgogne, 1951). Moths t y p i cally possess a frenulum (set of fused setae or hairs) projecting from the h i n d w i n g that terminates i n a small hook on the forewing, joining the t w o wings (Tillyard, 1918). Some m o t h taxa alternatively have a cuticular lobe on the forewing (the jugum) that hooks under the h i n d w i n g ; these mechanisms of w i n g coupling render the majority of moths functionally dipterous. Some phylogenetically derived moths,

70

C H A P T E R TWO

particularly arctiids, sesiids, and sphingids, have evolved h i g h w i n g beat frequencies and hindwings that are much reduced i n relative area. However, phylogenetically basal moths have an ineffective fren­ u l u m and are functionally tetrapterous w i t h fore- and hindwings of comparable size (Grodnitsky and Kozlov, 1985). Mechanical cou­ p l i n g between wings thus appears to be correlated w i t h relative w i n g reduction. I n many butterflies, wingbeat frequencies and concomitant inertial forces are l o w enough to obviate the need for extensive mechanical connections between the large and similarly sized fore- and hindwings (Grodnitsky and Kozlov, 1985, 1990, 1991). Similarly, l o w wingbeat frequencies and reduced inertial forces characterize such orders as Ephemeroptera, Mecoptera, Psocoptera, Raphidioptera, and Trichoptera, all of w h i c h have simple hooklike mechanisms connecting foreand hindwings (Edmunds and Traver, 1954; Ivanov, 1985, 1990; Lawson and Chu, 1974; New, 1974). Fore- and hindwings of auchenorrhynchous Homoptera are connected b y a claval fold on the forewing that links w i t h a smaller fold, lobe, or hook (New, 1974; D'Urso and Ippolito, 1994). The claval fold and the h i n d w i n g do not lock together but rather permit the t w o wings to slide relative to each other both d u r i n g flapping and w h e n wings are being deployed for flight (Ossiannilsson, 1950). Mechanically constrained sliding between fore- and h i n d w i n g is evidently unnecessary given the reduced extent of relative w i n g motion that occurs d u r i n g flight. When such a w i n g coupling device is absent, the taxon i n question tends to be either brachypterous ( w i t h reduced wings) or saltatorial (D'Urso and Ippolito, 1994). N e w (1974) described additional projections near the pterostigma of the homopteran forewing that are used to lock fore- and hindwings w h e n the insect is at rest. By contrast, tetrapterous insect taxa w i t h h i g h wingbeat frequencies tend to have sophisticated w i n g connectors that link fore- and h i n d ­ wings either tightly at one point or across a finite length of the contig­ uous wings (see Schneider and Schill, 1978; D'Urso, 1993; Brodsky, 1994). Hymenopteran wings are tightly linked b y a line of small hooks, the hamuli, that vary i n location along the anterior edge of the h i n d ­ w i n g (Basibuyuk and Quicke, 1997). Similar structures characterize w i n g coupling mechanisms i n the true Hemiptera (Heteroptera). The sternorrhynchous Homoptera (psyllids, aleyrodids, aphids and coccids) either possess hooks that couple fore- and h i n d w i n g , or alterna­ tively exhibit h i n d w i n g reduction. Such tight connections between fore- and h i n d w i n g , together w i t h substantial h i n d w i n g reduction, are almost certainly necessitated b y the h i g h wingbeat frequencies found i n these taxa (see section 4.2.2).

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

71

Insect wings are used i n a variety of functions supplemental to their principle role of force production d u r i n g flight. Various hymenopterans use w i n g motions to create convection currents used i n thermo­ regulation (see Neuhaus and Wohlgemuth, 1960; Herbst and Freund, 1962; Wohlgemuth, 1962; Stern and Dudley, 1991). The approximately planar structure of most nonelytral wings presents a convenient loca­ tion upon w h i c h to display both two-dimensional patterns as w e l l as coloration that may function i n either natural or sexual selection. Such patterned wings appeared early i n pterygote evolution and are w e l l developed i n some taxa b y the Upper Carboniferous (Carpenter, 1971). M a n y extant insects are cryptic by virtue of pattern matching be­ tween the dorsal w i n g surface and the natural background. Cryptic patterns may be inherent to the cuticle itself or may be exogenous i n character—certain weevil species (Curculionidae: Coleóptera) i n N e w Guinea carry moss gardens on their elytra that apparently function i n camouflage (Gressitt et al., 1968). Contrariwise, aposematic insects can use wings to display warning coloration. Deimatic (startle) displays on hindwings are also common among insects and are presumably used to frighten would-be predators (Edmunds, 1974). Parallel w i t h natural selection, sexual selection is a force of comparable significance for the evolution of w i n g coloration. Intraspecific communication via w i n g patterns is particularly widespread among the butterflies (Silberglied, 1984; plate 2H), but can also be found i n insect taxa as disparate as dragonflies, drosophilid flies, and euglossine bees (Thornhill and A l cock, 1983). Because w i n g appearance and aerodynamic function can essentially be decoupled, the superficial alary morphology is particu­ larly w e l l suited to modification for use i n visual communication to predators and conspecifics. Wings are also used by some insects to generate acoustic signals. The sounds of w i n g vibration, for example, are used for communica­ tion i n honeybees and i n drosophilid flies (Bennet-Clark and Ewing, 1968; Ewing, 1979). Beetles are particularly w e l l k n o w n for sound pro­ duction via stridulation between the elytra and the abdomen or h i n d ­ wings (Dumortier, 1963). A m o n g the Orthoptera, leg motion against the wings or stridulation of overlapping opposite wings are among the mechanisms used to produce the diverse sounds characteristic of this order. It is noteworthy that sound production i n both the Coleóptera and Orthoptera is associated w i t h forewing thickening (elytrization and tegminization, respectively). Sclerotization of the forewings i n evolutionary time may thus interact synergistically w i t h the evolution of sound production (see also Toms, 1986; Desutter-Grandcolas, 1995). Interestingly, stridulation between opposite wings has apparently evolved at least once i n the Diptera (Petrunkevitch, 1956), whereas

72

C H A P T E R TWO

clicking of the forewing against the pronotum to produce sound has been recorded i n an O l d World cicada (Popov, 1981). I n a noncommunicational context, collection of pollination b y bees grasping flowers often involves high-frequency but low-amplitude w i n g motions that dislodge pollen from the anthers b y the production of high-frequency sound (Buchmann, 1983; K i n g et al., 1996). Thus, a variety of insect taxa use w i n g vibration to produce sound, but this behavior usually occurs only when the insect is at rest. Intentional sound production d u r i n g flight is much less common among insects. I n the Neotropical butterfly genus Hamadryas, dorsal impact of opposite wings d u r i n g flight generates pulses of audible clicking sounds that are used i n territorial interactions (Otero, 1990; Monge-Nájera and Hernández, 1991; Monge-Nájera, 1992). Percussive use of wings d u r i n g flight has also been described i n agaristid (Bailey, 1982; Alcock et al., 1989) and noctuid moths (Dumortier, 1963; see also Hampson, 1892; McCrae, 1975). Sound production through opposite w i n g impact during flight may occur i n some satyrid butterflies (Kane, 1982) and i n some temperate-zone grasshoppers both d u r i n g escape from predators and d u r i n g apparent advertisement to conspecifics (pers. obs.). Finally, a novel use of wings has been described i n the African cricket Phaeophilacris spectrum. Using w i n g motions, these cave crickets generate vortices that subsequently travel to conspecifics and perhaps communicate information (Heinzel and Dambach, 1987; see also Heidelbach et al., 1991). Such behavioral use of wing-generated vortices is perhaps unique i n the Insecta. Loss of wings is widespread on evolutionary (section 6.2), ecological (section 7.4), and behavioral timescales. Alary polymorphisms and fac­ ultative w i n g expression range from fully winged individuals to brachypterous and apterous morphs. Production of different w i n g morphs depends on local ecological circumstances and is usually corre­ lated w i t h a behavioral tendency to migrate. Some insects can also autotomize (voluntarily break off) their wings. Winged female reproduc­ tives of ants and termites similarly undergo dealation following mat­ i n g flights, and at least one dipteran species (Lipotena) engages i n this drastic behavior. Zorapterans lose their wings at sexual maturity, as do various Thysanoptera and even some Blattaria.

2.3 ANCILLARY STRUCTURES Sensory functions such as vision and mechanoreception act p r o m i ­ nently i n the control of flight. The morphology and physiology of flight-related sensory structures (including eyes, antennae, and vari-

M O R P H O L O G Y OF T H E F L I G H T A P P A R A T U S

73

ous mechanoreceptors) are discussed i n chapter 5. More directly, i n ­ sect legs and the insect abdomen play various minor but nonetheless significant roles i n flight. The most direct contribution of insect legs is to enable the obligatory j u m p i n g takeoff subsequent to w h i c h the tarsal reflex (absence of contact w i t h ground) initiates w i n g flapping (see section 5.2.1). Jumping may also have been an evolutionary pre­ requisite to the initial evolution of insect flight (section 6.1.4). I n some saltatorial taxa (e.g., alucine chrysomelids, various homopterans, orthopterans), legs are hypertrophied and flight becomes an adjunct to locomotion by jumping, although wings are usually retained. W i t h i n the Diptera, Rohdendorf (1958/59a,c) has correlated taxonomic pat­ terns of w i n g venation w i t h leg structure and cursorial performance. I n general, a phylogenetic trend toward heavier and more muscular legs was observed i n the more derived families and super families; this correlation presumably indicates more rapid j u m p i n g and takeoff abil­ ities i n the aerodynamically more agile flies. Insect legs are typically retracted during steady flight to reduce drag forces on the body. I n some cases, laterally asymmetric leg movements d u r i n g flight may contribute to maneuverability (section 5.2.2). Aerodynamic drag of the insect abdomen and associated expendi­ ture of power exert a major influence on performance during flight (section 3.3.2). Little w o r k has been done on the nature of abdominal design toward reduced drag, but effective streamlining of the abdomi­ nal and the overall body profile is likely (see section 3.2.1). Abdominal cerei may ancestrally have contributed to stability i n pitch and yaw (section 6.1.3), whereas abdominal motions are used by some insects to generate advantageous aerodynamic torques d u r i n g maneuvers (sec­ tion 5.2.2). Contrariwise, disadvantageous aerodynamic torques may ensue i f the abdomen and the longitudinal body axis are not aligned parallel to the flight path. For insects w i t h relatively short abdomina (e.g., Diptera and Hymenoptera), these torques are likely to be small i n magnitude. By contrast, aerodynamic torque on the elongate abdomen of Odonata (required perhaps by the secondary genital pores of males and associated aerial copulatory habits) may preclude effective use of yawing and pitching movements d u r i n g flight. Odonates therefore typically initiate maneuvers by rolling about the longitudinal body axis. I n general, the location of the center of body mass must influence rotational maneuverability during flight (section 5.2.2). The abdomen may not strictly be necessary for flight i n dragonflies: one l i v i n g odonate specimen that had lost its abdomen escaped from its captors and flew off into the forest on Barro Colorado Island (Zotz and D u d ­ ley, pers. obs.)! Finally, differential allocation of mass to reproductive organs i n the abdomen may account for substantial differences i n w i n g

74

C H A P T E R TWO

morphology and flight performance between male and female insects, although such sexual d i m o r p h i s m has not been systematically investigated from the perspective of flight biomechanics. Egg loads of gravid females, endogenous l i p i d reserves of migrants, and food stored i n the crop can dramatically increase the power requirements of flight (section 3.3.3). A secondary reduction i n maneuverability and an increased vulnerability to prédation may also ensue.

2.4 SUMMARY The pterothorax of insects consists of flight muscles, cuticular sclerites, a complex axillary articulation, and wings. Induction of w i n g movements through cuticular transmission of muscular force applied to the w i n g base is widespread among insects. Both direct and indirect muscles contribute to dorsoventral w i n g motions as w e l l as to orientational control of the w i n g . Wings are cambered surfaces supported b y networks of veins that act to control and l i m i t deformations induced by aerodynamic and inertial forces. Wings can also function as adaptive airfoils that deform advantageously i n response to applied forces (e.g., alteration or reversal of camber). Thoracic configuration is broadly similar i n all insects, but w i n g modification and reduction i n relative size have substantially altered the homonomous condition of equivalent w i n g size and function. Dominant evolutionary themes i n pterothoracic modification include tegminization, elytrization, a n d / o r relative reduction i n size of one w i n g pair. Bimotorism is rare among extant w i n g e d insects relative to antero- and posteromotorism. Cooption of wings for secondary or supplemental nonaerodynamic purposes (e.g., elytra, haltères) largely defines patterns of interordinal insect diversity. Body shape, legs, and various sensory structures, although not directly contributing to flight aerodynamics, exert secondary influences on aerial performance.

Chapter Three KINEMATICS A N D AERODYNAMICS OF FLIGHT

E

FFECTIVE FLIGHT i n insects cannot emerge solely from the morphological expression of wings. Instead, flight i n animals derives from rhythmic wingbeat motions that generate aerody­ namic forces. Vertical forces are necessary to support the body weight and to control altitude; a net thrust is used to effect forward propul­ sion. Subtle alterations to the motions of flapping wings transiently disrupt this balance and permit flying insects to alter flight speed and trajectory. Flight is energetically costly, and associated expenditure of mechanical power is influenced b y w i n g and body morphology, w i n g beat kinematics, and patterns of body movement i n three-dimensional space. Morphological diversity of w i n g e d insects is matched by a cor­ responding kinematic diversity i n w i n g and body motions. A l t h o u g h less w e l l described than trends of morphological diversification, these motions largely determine the temporal patterns of force production and aerodynamic power expenditure d u r i n g flight.

3.1 W I N G AND BODY MOTIONS Fundamental to a biomechanical analysis of flight performance is the description of w i n g and body kinematics. Insect flight speeds are most generally referenced to the surrounding air volume (the airspeed of a flying insect), but can also refer to the speed relative to an external coordinate system fixed i n the earth (the so-called groundspeed). A i r ­ speed and groundspeed are not equivalent because of the effects of ambient winds. Of all w i n g and body motions, speed of the wings w i t h respect to the surrounding air is the most important aerodynamically and reflects the combined influence of the body's velocity and the mo­ tions of the wings relative to the body. The latter kinematics of the wingbeat, including wingbeat frequency, stroke amplitude, and w i n g orientation, directly influence the magnitude of and temporal vari­ ation i n aerodynamic forces and mechanical power expenditure. Meth­ odology specific to the kinematic analysis of flying insects was re­ viewed b y Dudley (1992).

76

CHAPTER

THREE

3.1.1 Speed of Flight A l t h o u g h one of the most fundamental of kinematic variables, insect airspeeds are also one of the least k n o w n features of flight performance. This unfortunate situation arises because flight velocities i n nature reflect the vector combination of the ambient w i n d velocity and the insect's velocity w i t h respect to the surrounding air (see fig. 3.1). Insects fly through air that is also potentially m o v i n g relative to the earth itself (e.g., w i n d ) . Groundspeed measurements thus combine insect air velocity w i t h the motions of the air itself relative to the terrestrial coordinate system, and these latter motions are nontrivial w i t h respect to insect flight speeds. Typical diurnal values for ambient w i n d speeds both w i t h i n and above vegetational cover range from 0.1 to 10 m / s , and are often higher (Rumney, 1968; see section 7.4.1). W i n d speeds can substantially exceed insect airspeeds and are, moreover, difficult i f not impossible to measure i n the immediate vicinity of flying insects. One approach to dealing w i t h the effects of ambient w i n d motions is simply to study flight indoors. Flight speeds w i t h i n large wind-free enclosures or rooms, for example, can provide at least m i n i m u m estimates for insect airspeeds i n forward flight. Demolì (1918), for example, used a stopwatch to time various insects i n their flight across a room to a brightly l i t w i n d o w . A hawkmoth attained the highest speed (15 m / s ) and a tabanid fly and a dragonfly (Agrión sp.) reached 14 m / s ; most other insects flew i n the range of 1-4 m / s . The values i n excess of 10 m / s seem inordinately h i g h and may be associated w i t h t i m i n g errors for rapid flight over short distances. By contrast, Stevenson et al. (1995) obtained m a x i m u m airspeeds of 5.3 m / s for sphingid moths (Manduca sexta) flying freely i n a large arena, whereas syrphid flies could attain airspeeds of 10 m / s i n chases (Collett and Land, 1978). I n a much-cited study, Lewis and Taylor (1967) determined flight speeds of 0.4-8 m / s for a group of mostly neopterous insects flying across a room. Body size of individual insects was presented as a product of w i n g and body length, and body mass was not reported. I n general, detailed morphological data have not been taken on insects for w h i c h flight speed measurements have been made, although most such studies have been made on fairly large insects. By contrast, remarkably few data exist on the flight speeds of small insects (e.g., 1-3 m m body length), i n spite of their majority i n the contemporary fauna (section 7.1). David (1978) recorded m a x i m u m speeds of 0.9 m / s for Drosophila melanogaster i n experimental settings (see also Craig, 1986; Marden et al., 1997). Comparably small aphids

KINEMATICS AND AERODYNAMICS

77

(A)

wind direction

FIG. 3.1. Insect flight velocity measured with respect to a terrestrial coordinate system is the vector summation of the ambient wind velocity and the air veloc­ ity of the insect. (A) When ambient wind speed is small relative to the insect airspeed, insect drift with wind is small and the airspeed is similar to the groundspeed. (B) When wind speed is high and oriented against the insect's heading, groundspeed measurements can give misleading estimates of the true airspeed during flight.

78

CHAPTER

THREE

fly at airspeeds less than 0.8 m / s (Kennedy and Thomas, 1974; David and Hardie, 1988; Hardie and Young, 1997). Tiny diaspidid homopterans are essentially incapable of flying at airspeeds greater than 0.5 m / s (Rice and Moreno, 1970). Airspeeds of small insects thus rarely ex­ ceed 1 m / s , a l o w value relative to the speed of most ambient w i n d s (see section 7.4.2). Badly needed are further airspeed and morphologi­ cal data for diverse insect orders flying under controlled experimental conditions (particularly w i t h control of insect thermal regime). Infor­ mation on variance i n airspeeds w i t h i n and among individuals w o u l d also be desirable. Such data should be analyzed using a comparative method to control for potentially nonrandom phylogenetic associa­ tions among species (see Harvey and Pagel, 1991). Unfortunately, chamber dimensions influence choice of airspeed, at least i n butterflies (Dudley and Srygley, 1994) and i n the sphingid m o t h Manduca sexta (Stevenson et al., 1995), so particular attention must be paid to the experimental context during such measurements. The effects of air temperature and illumination on insect flight speed are potentially substantial. For example, indoor flight speeds of the fly Calliphora i n ­ crease w i t h ambient temperature, whereas flight speeds of a hymenopteran and various dipterans and lepidopterans toward an i l l u m i ­ nated w i n d o w increase w i t h the intensity of illumination (Schneider, 1965). Relative to flight i n confined spaces, insect airspeeds i n nature are even less w e l l known. Most such estimates are conjectural or are com­ promised b y experimental artifacts such as entrainment i n the flow field of moving vehicles (see Johnson, 1969). Generations of entomolo­ gists hanging out the w i n d o w s of speeding cars or even trains have done little more than to confirm that substantial air volumes (and sometimes insects) are transported alongside m o v i n g vehicles. I n a similar vein, suggestions of supersonic flight i n deerflies (Townsend, 1927) are entertaining but clearly implausible on physical grounds alone (Langmuir, 1938). Townsend (1939,1942) responded to this latter criticism b y stating that the flies i n question had "never been cap­ tured" and were "entirely invisible at top speed." Most estimates of insect airspeeds i n natural contexts have decoupled airspeed from ground speed by assuming an average ambient w i n d speed and direc­ tion. This method has given airspeed values generally higher than those estimated for indoor settings (table 3.1). Because w i n d s exhibit considerable spatial and temporal heterogeneity, assumptions of con­ stant flight speed and direction may not be appropriate, particularly for insects flying at heights different from the height at w h i c h w i n d velocity was measured. None of the airspeed studies presented i n table 3.1 measured ambient w i n d motions directly i n the vicinity of a

KINEMATICS AND AERODYNAMICS

79

TABLE 3.1

Insect Airspeeds Calculated Indirectly from Measurements of Ground Speed, Flight Direction, and the Wind Velocity (see fig. 3.1) Order Odonata Orthoptera

Hymenoptera

Diptera

Species

Airspeed (m/s)

Reference

2-5 10 (maximum) Schistocerca gregaria 3-8 Nomadacris septemfasciata 13 (maximum) (Migratory locusts) Apis mellifera 6-8

Rüppell (1989) May (1991) Rainey et al. (1957) Waloff (1972a) Baker (1981) von Frisch and Lindauer (1955) Wenner (1963)

Glossina spp. (Tsetse flies)

Brady (1991)

(Various)

5

Note: Wind velocity is typically measured in the vicinity of the investigator and not of the flying insect.

flying insect, and logistical methods preclude such measurements under most circumstances. Sayter (1965) presented an ingenious pho­ tographic method that relies on multiple images of the same insect i n different orientations to deduce locally the ambient w i n d velocity and insect airspeed. Major reorientation of the body axis i n a short period of time is required to make such calculations, and this method is there­ fore not generally applicable to free-flying insects i n nature. However, direct airspeed measurements have been recently made on insects i n natural free flight. Flight speeds of insects crossing a body of water can be measured b y holding a unidirectional anemometer lat­ erally from the b o w of a boat m o v i n g parallel to and at the same speed (relative to a terrestrial coordinate system) as the flying insect (see Dudley, 1992). This approach is analogous to flying alongside the i n ­ sect and obtaining a measurement of local airspeed. Because the ane­ mometer is held outside of the flow field around the boat's hull, this speed measurement directly corresponds to the speed of the insect rel­ ative to the surrounding air. Flight direction of the insect is determined from a compass reading of boat orientation d u r i n g such airspeed measurements. Wingbeat kinematics can also be obtained simultane­ ous w i t h airspeed measurements using cameras mounted on the boat. Following capture of the insect (plate 3), the boat can immediately be stopped to permit measurements of ambient w i n d speed and direc­ tion, thus uniquely resolving the vector triangle of insect air velocity, velocity of the insect w i t h respect to the ground, and the w i n d velocity (see fig. 3.1).

80

CHAPTER

THREE

To date, this free-flight method has been used only on butterflies and moths that are fairly large and thus easily tracked d u r i n g their flights over a large body of water. For example, an average airspeed of 3.9 m / s was determined for the diurnal uraniid m o t h Urania fulgens d u r i n g migratory flight (DeVries and Dudley, 1990; Dudley and DeVries, 1990). Similarly, airspeeds for sixty-two Neotropical butterfly species i n natural free flight ranged from 0.7 to 7.5 m / s (Srygley and Dudley, 1993; Dudley and Srygley, 1994). Direct measurements of airspeed can potentially be implemented on diverse pterygote taxa other than the aforementioned Lepidoptera, as insects can also be taken to the center of bodies of water, released to elicit flight, and then subsequently followed. Aforementioned data for both constrained and natural free flight airspeeds indicate, however, that insects generally fly at airspeeds between 0.5 and 10 m / s . Insect airspeeds i n excess of 10 m / s are rare and require careful documentation. M u c h of the variation i n insect flight speeds is a consequence of body size, as larger insects tend to fly faster (e.g., Lewis and Taylor, 1967 ). I f larger insects are similar i n shape to smaller insects and vary only i n size, then both w i n g length R and w i n g area S w i l l scale isometrically w i t h body mass m (i.e., R oc m and S «= m for geometrically similar organisms). The general dependence of steady-state aerodynamic forces on w i n g area and on the square of the relative velocity of airfoils (eqs. 1.2 and 1.4) permits the body weight of a flying insect to be equated w i t h the mean aerodynamic lift on the flapping wings (see Lighthill, 1977; Norberg, 1990). Variation of forward airspeed w i t h w i n g loading p and w i t h body mass can then be estimated for isometrically scaled organisms: j_ V~p¿ (3.1) 1/3

2/3

w

V

oc

m .

(3.2)

6

These equations strictly refer to conditions i n w h i c h the forward airspeed is h i g h relative to the mean flapping speed of the wings, but provide a useful initial basis for comparison w i t h empirical results. Because insect airspeed data are so sparse, the generality of equations (3.1) and (3.2) has not been evaluated for a w i d e diversity of taxa. Johnson (1969) compiled data relating insect airspeed to body length for a variety of orders; these data refer to some chosen airspeed i n a laboratory context, and are derived primarily from results of Lewis and Taylor (1967). I f isometry i n body length is assumed, the interspecific analysis of Johnson (1969) indicates a dependence of airspeed on m° (see Dudley, 1994). This exponent differs significantly from the value of 29

KINEMATICS AND AERODYNAMICS

81

0.167 indicated b y equation (3.2), although the underlying assumption of isometric body design is unlikely to be met i n such interordinal comparisons. Allometric studies of insect flight speeds i n nature are limited i n terms of taxonomic coverage to the butterflies (Lepidoptera: Rhopalocera). Butterfly airspeeds i n natural free flight exhibit interspecifically a stronger dependence on w i n g loading (V p ) and on body mass (V oc m ) than is predicted by equations (3.1) and (3.2) (see Srygley and Dudley, 1993; Dudley and Srygley, 1994). Such relatively i n ­ creased flight speeds at higher body masses may reflect the positive allometry of w i n g loading i n butterflies and the corresponding devia­ tion from isometric design (see Chai and Srygley, 1990; Dudley, 1990; Srygley and Dudley, 1993). N o specific prediction exists on aerody­ namic grounds for the relationship between flight speed and w i n g as­ pect ratio (Norberg and Rayner, 1987; Norberg, 1990), and only a weak inverse relationship exists between aspect ratio and natural airspeeds of butterflies (Dudley and Srygley, 1994). I n contrast to butterflies, bats demonstrate a positive correlation between flight speed and w i n g as­ pect ratio (Norberg and Rayner, 1987). The underlying reasons for such a difference between these taxa are not k n o w n . Because average insect body size is l o w by anthropomorphic stan­ dards (4-5 m m ; section 7.1), only relatively few insect species (e.g., those w i t h body masses approximately >250 mg) are likely to fly at airspeeds exceeding 5 m / s (Dudley, 1994; see section 7.4.2). Exactly such insects, however, have been the primary focus of biomechanical as w e l l as ecological investigation. Natural airspeeds as w e l l as aero­ dynamic mechanisms of the smaller but far more numerous insect taxa are little k n o w n (section 7.1). Also, consequences of intraspecific mor­ phological variation for airspeed selection have not been comprehen­ sively examined. M u c h of such variation is, as w i t h interspecific vari­ ation, influenced directly by body size. However, the implications of sexual dimorphism for choice of airspeed may be more subtle and likely derive from the varying ecological roles of flight between the sexes (e.g., McLachlan, 1986a; Wickman, 1992). Intraspecific variation i n flight speeds is also evident on evolution­ ary timescales. Flies of the genus Drosophila respond positively to arti­ ficial selection for increased flight speeds (Weber, 1988, 1996; Marden et al., 1997), indicating substantial genetic variation i n flight perfor­ mance. When coupled w i t h biomechanical assays of m a x i m u m per­ formance (section 5.4.1), such selection experiments could provide substantial insight into functional constraints on flight capacity as w e l l as evolutionary responsiveness of the flight apparatus to variable se­ lective regimes. Particularly w i t h Drosophila, genetic variation i n flight in

w

036

82

CHAPTER

THREE

performance is consistent w i t h the widespread suggestion that longterm laboratory culture indirectly selects on fly strains for reduced flight capacity. For example, cultures of Drosophila melanogaster re­ spond rapidly to a selection regime for enhanced u p w i n d flight per­ formance, and also decline substantially i n performance once selection is relaxed (see Weber, 1996). Evolutionary lability i n flight capacity is therefore considerable, and phenotypic interactions between flight ability and fecundity (see section 6.2.4) may be particularly pro­ nounced w i t h i n the artificially constrained environment of a labora­ tory culture bottle. This hypothesis has not been systematically investi­ gated, but is clearly relevant for assays of flight performance i n this important genus. Airspeed is but one of many features of performance that delineate the flight envelope of insects. Flight behaviors that entail curved trajec­ tories and often complex three-dimensional maneuvers are described i n section 5.3. Rectilinear flight may involve components of vertical ascent and descent as w e l l as horizontal motion. Few data are avail­ able on the kinematic correlates of ascending or descending flight. Larsen (1934) evaluated the flight of honeybees through an experimen­ tal tube that could be tilted horizontally; wingbeat frequency increased whereas upwards flight speed decreased as the tube was progressively tilted toward vertical. Also, foraging honeybees fly more quickly to d o w n h i l l feeders than to u p h i l l feeders (Heran, 1956). These changes i n flight speed are consistent w i t h increased flight costs associated w i t h gain i n potential energy (see section 5.4.2), but further investigation can be envisioned. For example, controlled flight i n a variable-tilt w i n d tunnel could be used to evaluate interactions between horizontal flight speed and flight angle over the entire range of flight speeds, not just at the preferred cruising speed. If natural insect airspeeds are little k n o w n , capacities for accelera­ tion are even less w e l l described. Success i n aerial chasing and evasion may be critically dependent on acceleration and directional changes (section 5.3), but few comparative data exist on such transient features of flight performance. I n general, estimates of acceleration from posi­ tional data must be treated cautiously because the results are sensitive to t i m i n g and digitization errors; such studies must be carried out w i t h h i g h filming frequencies and adequate spatial resolution of film i m ­ ages i f sufficient accuracy is to be attained (see Harper and Blake, 1989; Walker, 1998). Termier (1970a) suggested that hoverflies could attain accelerations of up to 18 g w h e n startled, whereas accelerations up to 9 g have been proposed i n corduliid dragonflies (Ryazanova, 1966). More realistically, male hoverflies i n chase attain accelerations of 3 g (Collett and Land, 1978). Lacewings escaping from bats accelerate up

KINEMATICS AND AERODYNAMICS

83

to 2.6 g (Miller and Olsen, 1979), whereas odonates i n the suborder Anisoptera have greater capacity for acceleration (up to 2.5 g) than do those i n the suborder Zygoptera (Rüppell, 1989; May, 1991; Wakeling and Ellington, 1997b). Higher-order temporal derivatives of body posi­ tion (e.g., the rate of change of acceleration) are uninvestigated i n i n ­ sects but may be important w h e n capturing prey, w h e n escaping from predators, and d u r i n g mating encounters. Gliding flight characterizes very few insects, most notably some dragonflies (Hankin, 1921a, b; Wakeling and Ellington, 1997a) and but­ terflies. Slow glides w i t h intermittent flapping are, for example, char­ acteristic of many unpalatable ithomiine and heliconiine butterflies (e.g., Benson et al., 1989). D u r i n g such glides, the fore- and hindwings of gliding butterflies are often nonoverlapping (see Betts and Wootton, 1988). Nemopterid neuropterans also glide occasionally (Picker, 1987), and some mayflies engage i n transient parachuting w i t h rapid vertical descent (Brodsky, 1994). M a n y insects engaged i n long-distance migra­ tion may glide intermittently, albeit for periods usually less than a sec­ ond (e.g., dragonflies: Corbet, 1962; locusts: Baker and Cooter, 1979b; the m o t h Urania fulgens: Dudley and DeVries, 1990). Similarly, the sphingid m o t h Manduca sexta glides for periods of several wingbeats w h e n flying i n large enclosures (Stevenson et al., 1995). Best studied of insect gliders is the monarch butterfly Danaus plexippus, w h i c h utilizes u p w a r d convection and prevailing winds to facilitate long-distance migration (Gibo and Pallett, 1979; Gibo, 1981a-c). Glides of several centimeters have been recorded i n secondarily flightless beetles; wings are held outstretched i n such glides but are not flapped (Bilton, 1994). Similarly, parachuting and i n some cases gliding behavior has been recorded i n both larvae and adults of several Southeast Asian leafmimicking species of stick insects that are flattened dorsoventrally (Siedlecki, 1917). Selection of a forward flight speed near or equal to zero is termed either stationary flight (Magnan and Sainte-Laguë, 1933) or hovering flight. Hovering is aerodynamically challenging because the absence of a forward velocity vector demands that all requisite forces must be produced solely b y w i n g flapping. Because w i n g area relative to body mass generally increases for smaller animals, the ability to hover i n ­ creases w i t h reduced body size. Hovering may have facilitated the diversification of small dipteran and hymenopteran parasitoids (sec­ tion 7.1), and is also an important characteristic of many insect pollina­ tors (section 7.2). Most insect hoverers can also fly backward, i f only for transient periods, and i n general are highly maneuverable. For ex­ ample, many insect hoverers are capable of flying sideways (e.g., Collett and Land, 1975). Hovering is also an important feature of many

84

CHAPTER

THREE

dipteran and odonate mating systems. Although most common among the neopterous insects, the ability to hover also characterizes termites i n their nuptial flights, Thysanoptera, and many Homoptera. A t least one orthopteran (Dissosteira carolina; see Pierce, 1948) hovers d u r i n g a courtship display. A m o n g flying vertebrates, only h u m m i n g ­ birds and some nectarivorous bats can engage i n hovering for other than transient intervals. 3.1.2 Wing Kinematics Relative to the study of translational body motions, quantitative analy­ sis of the rapid movements of beating insect wings requires substan­ tially higher temporal resolution from filming or other experimental methodology. Moreover, most evaluations of wingbeat kinematics have been carried out i n the laboratory because of the difficulties of studying insect flight performance i n nature. Free flight i n enclosed chambers and insectaries as w e l l as controlled free flight i n w i n d tun­ nels have been widely used to this end. Unfortunately, insects are psy­ chologically resistant to anthropogenic suggestions of directed flight performance at a constant velocity. The study of free flight thus usu­ ally relies on analysis of volitional flight rather than of a performance regime determined by the experimenter, and behavioral variation i n kinematics may accordingly be high. Studies of free flight at l o w or zero airspeed alleviate many methodological problems associated w i t h the evaluation of forward flight, and much of our knowledge and parameter standardization of wingbeat kinematics derives from the comprehensive hovering study of Ellington (1984c). M u c h experimental w o r k on w i n g kinematics has involved attach­ ing insects by the thorax or abdomen to rigid mounts or tethers. Wing beating can then be elicited using a variety of sensory cues; this exper­ imental preparation is best termed a tethered simulation of flight as the phrase "tethered flight" is an oxymoron. Classic examples of this approach include w o r k on locusts (Weis-Fogh, 1956a; Zarnack, 1972), calliphorid flies (Hollick, 1940; Nachtigall, 1966, 1973), and Drosophila (Vogel, 1966, 1967a; Götz, 1968). Potential artifactual complications of tethering are not w e l l understood, although kinematics of free flight and of tethered preparations can differ substantially. I t was, i n fact, first suggested by Marey (1891) that tethering could induce abnormal w i n g movements (see also Voss, 1914; Schmidt, 1938). To cite Magnan and Planiol (1933b) on tethered preparations: "En effet la plupart des insectes sont doués d ' u n tempérament q u i semble à l'opérateur essen­ tiellement capricieux, leurs actions étant généralement imprévues, fan­ tasques et décourageantes" [In effect, the majority of insects are en-

KINEMATICS AND AERODYNAMICS

85

dowed w i t h a temperament that seems to the operator to be essentially capricious, their actions being generally unforeseen, w e i r d and dis­ couraging]. This observation has rarely been evaluated quantitatively, although Baker et al. (1981) determined that wingbeat frequencies of free-flying locusts are significantly greater than those of tethered animals (see also Kutsch and Stevenson, 1981; Schneider, 1981a). Wingbeat kinematics of tethered insects can also vary substantially w i t h elapsed flapping time and w i t h the particular sensory regime experienced by the re­ strained animal (e.g., Weis-Fogh, 1956a; Vogel, 1967a; Gewecke and Kutsch, 1979; Curtsinger and Laurie-Ahlberg, 1981; Kutsch and H u g , 1981; Ward and Baker, 1982; Grodnitsky and Kozlov, 1987; Grodnitsky and Morozov, 1993; Lehmann and Dickinson, 1997; Nachtigall, 1997b). Furthermore, the force balance of tethered insects may differ substan­ tially from that i n free flight (section 3.2.2). Tethered simulations of flight are thus heuristically informative but cannot evaluate quantita­ tively the extent to w h i c h associated kinematics and aerodynamic mechanisms are those actually used by free-flying insects. O n the other hand, a major limitation of free-flight studies is the often h i g h i l l u m i ­ nation required to visualize wingbeat kinematics—intense lights likely disrupt normal optomotor responses and may substantially alter natu­ ral flight behavior. Also, most investigations of the muscle actions and neurophysiology underlying w i n g motions (section 4.2) are logistically impossible unless the insect can be fixed to an experimental mount. Progress i n understanding insect flight mechanics has thus derived historically from a combination of tethered and free-flight approaches. The longitudinal axis of insect wings moves i n an approximate plane largely constrained by the geometry of the axillary apparatus and the line of action of the dorsoventral musculature (figs. 2.IB, 2.3). This plane is oriented at an angle ß (the stroke plane angle) relative to horizontal (see fig. 3.2A). The orientation of the longitudinal body axis relative to the line of action of the dorsoventral musculature is con­ strained anatomically, whereas the body axis oriented relative to hori­ zontal defines the body angle % w h e n the insect is viewed laterally (fig. 3.2A). A wingbeat consists of t w o consecutive half-strokes i n the stroke plane that are termed the d o w n - and upstroke. I n fact, these terms are misnomers as the wingbeat is most appropriately referenced to the body coordinate system and not to gravity; dorsoventral and ventrodorsal strokes, respectively, w o u l d be more accurate, although I w i l l follow the conventional usage here. Insects i n flight move their wings rhythmically w i t h i n the stroke plane (fig. 3.2B). The mean number of oscillatory cycles per second (i.e., Hz) is termed the wingbeat frequency (n). Because w i n g motions

86

CHAPTER

THREE

/

-K/2 FIG. 3.2. Wingbeat kinematics and body orientation for a flying insect. ( A ) The longitudinal axis of the body forms an angle % with respect to horizontal when the insect is viewed laterally. Wing motions define a stroke plane oriented at the stroke plane angle ß . The wing tip can be elevated above or below the stroke plane angle (positive and negative 6, respectively). (B) The angular po­ sition cp describes the location of the wing tip projected onto the stroke plane. The wing moves through an arc of amplitude