On the Wing

On the Wing
Insects, Pterosaurs, Birds, Bats
and the Evolution of Animal Flight

David E. Alexander
Illustrations by Sara L. Taliaferro

1
1
Oxford University Press is a department of the University of Oxford. It furthers
the University’s objective of excellence in research, scholarship, and education by
publishing worldwide. Oxford is a registered trade mark of Oxford University Press
in the UK and in certain other countries

Published in the United States of America by Oxford University Press

198 Madison Avenue, New York, NY 10016, United States of America

© Oxford University Press 2015

All rights reserved. No part of this publication may be reproduced, stored

in a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by license, or under terms agreed with the appropriate reproduction rights
organization. Inquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press,
at the address above.

You must not circulate this work in any other form

and you must impose this same condition on any acquirer

Cataloging-in-Publication data is on file at the Library of Congress

9780199996773

9 8 7 6 5 4 3 2 1

Printed in the United States of America on acid-free paper

To the late Larry Martin
gone too soon
CONTENTS

Acknowledgments  ix
Note on Sources   xi

1. Can’t Tell the Players without a Scorecard  1

2. Theme and Variations: Similarities and Differences among
Nature’s Flyers  21
3. How to Fly?  39
4. Gliding Animals: Flight without Power  60
5. Insects: First to Fly  74
6. Birds: The Feathered Flyers  103
7. Bats: Wings in the Dark  130
8. Pterosaurs: Bygone Dragons  147
9. Pedestrians Descended from Flyers: Loss of Flight  164
10. Unifying Themes?  173

Bibliography  183
Index  197
ACKNOWLEDGMENTS

This book had an amazingly long gestation. I wrote the first draft of the
first chapter in 2002. Between then and now, I talked to many people
about the evolution of flight, and while I may have forgotten some of the
conversations or people, they nevertheless helped shape my thinking on
the topic.
Many of these conversations were with my colleagues at the University
of Kansas, including Ron Barrett-Gonzalez, Dave Burnham, George Byers,
Amanda Falk, Rudolf Jander, Matt Jones, and Bob Timm, as well as the
many students in my Animal Flight Seminar over the years. Ed Wiley,
Bruce Lieberman, and Michael Engel all helped this non-systematist un-
derstand (and keep from embarrassing myself with) phylogenetic system-
atics. Mark Robbins showed me museum specimens when I needed to see
details of bird anatomy and feather structure and answered many orni-
thological questions. I learned a lot about paleontology, dinosaurs, and
birds from the many hours I spent in the lab of the late Larry Martin; no
one enjoyed a good argument or took it less personally than Larry. Larry’s
former student, Chris Bennett, patiently and thoroughly answered my
many e-mails with questions about pterosaurs. I also discussed animal
flight or flight evolution at various times with Roy Beckemeyer, Kristin
Bishop, Sankar Chatterjee, Jeff Dawson and his grad student Ryan Chle-
bak, Robert Dudley, Jimmy McGuire, the late John McMasters, Jake
Socha, Sharon Swartz, Jim Usherwood, and Steve Yanoviak, and there
were probably others whom I am sorry to say I have forgotten. Chris Ben-
nett, Robert Dudley, Nick Longrich, and Jake Socha each read and com-
mented on a chapter, and comments from two anonymous reviewers who
heroically read the entire manuscript significantly improved the final ver-
sion. Any mistakes or inaccuracies that may remain are, of course, my re-
sponsibility. Sara Taliaferro drew most of the illustrations; hers say
“(courtesy of S. T.)” in the figure legend. Also, Roy Beckemeyer and my son,
Kevin Alexander, each drew a figure. (If no credit is listed with a figure, it

  [ ix ]
is my own original drawing.) The staff at the University of Kansas Writing
Center (especially Amanda Hemmingsen) were the first to read the drafts
of my chapters, and I thank them for their useful feedback.
My biggest thanks go to my wife, Helen Alexander, for all her support
and encouragement; she remains my most enthusiastic and tireless booster.

[ x ] Acknowledgments
NOTE ON SOURCES

Scientists traditionally use the “author, year” method of citing sources,

which is fine for technical literature, but I have always felt that it breaks
up the flow of text for non-scientists. I have chosen to use a simplified
form of endnotes to cite my sources: the superscript numbers in the text
refer to a set of abbreviated bibliographic entries placed at the end of
each chapter under “References.” The reader can then refer to the Bibliog-
raphy at the end of the text for full bibliographic information for all the
sources cited.

  [ xi ]
On the Wing
CH A P TER  1

Can’t Tell the Players without

a Scorecard

I was inspecting my tomato plants when I heard a familiar buzz. Looking

up, I expected to see the fuzzy yellow-and-black of a bumblebee; instead,
I caught a glimpse of bright metallic green. My buzzing visitor was actu-
ally a beetle, a singularly handsome scarab in a group called the bumble
flower beetles. I was able to get a very nice look at my little visitor as it
came to a near-standstill to inspect the foliage. The beetle hovered near a
plant, probably looking for flowers for a quick bite of pollen, effortlessly
matching the swaying of leaves and stems in the breeze. This beetle had a
flattened, green metallic body with a coppery stripe down each side,
making it Cotinus nitida, the green June beetle.* I could only see the bee-
tle’s wings at certain angles where the barely visible blur told me they were
a good bit longer than the beetle’s body. Although I know that the beetle
was beating its wings at about 80 beats per second and that it uses ingen-
ious origami-like folds to get its big wings stowed under their covers after
landing, my real delight came in watching the little beast delicately in-
spect a couple of plants while giving me an expert demonstration of its
hummingbird-like hovering abilities. After a few seconds, apparently de-
ciding my garden offered too few rewards, the beetle rose up and shot
across the lawn so fast that I lost sight of it after 3 or 4 meters (10 or
12 feet).

*  These flashy scarabs are not all that closely related to the more common, brown
June beetle. Green June beetles are bigger, more agile, and fly during the day, unlike
common June beetles, which tend to be nocturnal. Scarabs make up a huge family of
beetles, and biologists put green June beetles and common June beetles in entirely
separate subfamilies.
 What does a brief encounter with a garden insect have to do with the
evolution of flight? That green June beetle represents the beetle lineage,
the most successful lineage of all living animals, and its success is due in
no small part to beetles’ highly developed flight ability. That beetle could
hover, match its movements to plants dancing in a breeze, and fly very
fast. These abilities require a sophisticated and specialized flight control
system—brain and nervous system, reflexes, senses—powerful muscles;
exceptionally light, strong wings; and a body strong enough to withstand
flight forces yet light enough to minimize the efforts of the flight muscles.
Beetles arrived at their modern form through a 300-million-year period
of evolutionary trial and error and continuous refinement. Beetles or
robins or bats seem perfectly at home on the wing, yet their flying ability
was produced over millions of generations and is still being fine-tuned
today, by natural selection.
Animals have evolved flapping (powered) flight only four times in the
400-million-year history of land animals. Although flying animals cover a
huge size range, wing flapping looks surprisingly similar in small insects
and large birds. The basic flapping pattern does not change much because
these animals all operate under the same constraints. First, they depend
on the physical properties of the air; air properties do change a bit with
size but are fundamentally the same for all animals. Second, all animals
power their wings with muscles. Muscle properties do not change much
from one animal to another, even between animals as different as dragon-
flies and swans, so the properties of muscles themselves determine how
muscles can be employed.

FOCUS ON POWERED FLIGHT

In the strict technical sense, flight means using a wing-like surface to pro-
duce enough lift to move through the air. A careful reader might notice
that this definition includes gliding, or unpowered flight. We will certainly
look at gliding because many scientists think that gliding may have been a
step in the evolution of flight in one or more animal groups. A startling
variety of animals can glide, even some species of frogs and snakes. Glid-
ing is inherently limiting, however, so our main focus is on animals that
use powered flight. These are easy to distinguish from gliders because they
flap their wings and can stay aloft as long as their muscles permit.
The difference between gliding and powered flight is exactly analogous
to coasting and pedaling on a bicycle. Coasting—like gliding—is great for

[ 2 ]  On the Wing

going downhill but not much use for moving uphill or on long level
stretches. Pedaling—like powered flight—gives true versatility. Just as a
bicyclist can pedal pretty much anywhere wheels can go, a flyer can go
pretty much any direction through the air by flapping. Gliding is an im-
portant part of the story, but most of my emphasis in this book is on pow-
ered, or flapping, flight.

BENEFITS OF POWERED FLIGHT

Flying animals enjoy a number of benefits over non-flying, ground-bound

(terrestrial) animals of similar size. Some are obvious. Flight gives an
animal a very potent predator-escape route, one that cannot be followed
by a running predator. The converse is equally true: a flying predator can
attack prey from the third dimension whereas its terrestrial prey can only
seek to escape in the other two dimensions. These advantages have huge
potential payoffs. A flying predator might get a meal or a flying prey
animal might live another day, where a non-flying animal might not.
Another clear advantage is that flying animals can reach remote or in-
accessible locations that may be difficult or impossible for non-flyers to
reach. The vast rookeries of seabirds on seaside cliffs of remote islands are
well-known examples of this advantage. Flyers tend to reach islands or
newly opened (disturbed) habitat before other animals. For example,
when researchers first ventured into the devastated area around Mt. St.
Helens within days after the eruption, they observed that flies were al-
ready there, long before any other new immigrants.1
The physics of flight requires flying animals to fly on the order of 10
times faster than a similarly sized runner can run. This speed gives flying
animals yet another advantage: they can search much more effectively.
Not only can a flyer cover distance 10 times faster than a runner, but the
elevated vantage point of flight means the flyer can usually see much far-
ther to each side than a searcher restricted to the ground. Thus, whether
searching for food, a mate, or shelter, a flyer can search much more rapidly
and cover much more ground than a runner, all else being equal.
A final advantage enjoyed by flying animals is economical travel. A
flying animal uses substantially less energy (food, fuel) to travel a given
distance than a runner of similar size.2 This difference has enormous eco-
logical consequences, for both short-term and long-term journeys. In the
short term, flight makes practical daily commutes for foraging of a couple
kilometers (a mile or more) for a honeybee, or many kilometers (several

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 3 ]
miles) for a great blue heron. In the long term, flight makes long-distance
migration practical. Because of the way metabolic power scales with size,
only large terrestrial mammals (such as caribou or African antelope) mi-
grate significant distances, and those 300- to 500-kilometer (200- to
­300-mile) migrations on foot pale next to migrations of flyers. Animals as
small as butterflies and as large as whooping cranes routinely make sea-
sonal migrations of more than 3,200 kilometers (2,000 miles). Without
flight, animals could not migrate these huge distances so we would no
longer see such classic seasonal indicators as wedges of geese flying south
in the winter or the return of the first robins of spring.
Insects include more known species than all other animals combined.
Among vertebrates, birds include almost twice as many known species as
either mammals, reptiles, or amphibians. Given that most insects and
birds fly, many biologists have assumed that flight played a major role in
the success and diversity of these groups. Demonstrating a causal rela-
tionship between flight and evolutionary success is, however, extremely
difficult. To test the question experimentally, researchers would have to
go back 400 million years and re-start insect or bird evolution without
flight, an obvious impossibility. Moreover, we have the example of the
pterosaurs to prove that the ability to fly is not enough of an advantage to
prevent extinction. Nevertheless, flight has such clear and conspicuous
benefits that the ability to fly was probably an important ingredient in the
success of the earliest flying species in each of the lineages that evolved
powered flight.

THE “BIG FOUR” POWERED FLYERS

Only four animal groups have evolved the ability to fly under power, and
each of these groups evolved powered flight only once. These four groups
are, in the order that they took to the air, insects, pterosaurs (also called
pterodactyls), birds, and bats. Although they all fly by flapping a set (or
two) of wings, their basic body plan, and especially their wing structure,
is quite varied. Insects are arthropods, meaning that they have a rigid
exoskeleton that they must periodically shed or “molt” in order to grow.
They have no backbone—nor, indeed, bone of any kind; hence, they are
invertebrates. Exoskeletons are rather handy, combining the supporting
framework of a skeleton with the protection of a suit of armor. They afford
better protection when small, and since exoskeletons become dispropor-
tionately heavy and unwieldy if they get too large, most insects are small.

[ 4 ]  On the Wing

In contrast, vertebrates have endoskeletons made of bone, which are more
effective for supporting a body at moderate and large sizes, so the verte-
brate flyers are generally larger than insects. Insects evolved flight over
350 million years ago, and pterosaurs, the first vertebrate flyers, took to
the skies over 200 million years ago, then went extinct with the dinosaurs
about 65 million years ago. Birds arose around 160 million years ago, and
bats were the last to evolve powered flight, more than 55 million years
ago.3 These four groups each achieved flight independently; none gave rise
to any of the other groups.

FEATS OF FLYERS, FROM HUMDRUM TO HEROIC

Flying animals are nearly everywhere, from the pigeons of many big cities
to the mosquitoes of the high Arctic; from the albatrosses of the open
ocean to the parrots and damselflies of the rain forest. Flying animals
have discovered and made themselves at home pretty much everywhere
that has air to breathe and food to eat. Mostly, they go about their busi-
ness doing what all animals do: looking for food and shelter, searching for
mates, and, if they are fortunate, reproducing. Flying animals have one
huge advantage over those of us stuck on the ground, though. As I men-
tioned, flyers can fly much faster than terrestrial animals can walk or run.
The very physics of flight requires flyers to move five to twenty times
faster than a walker or runner of the same size can travel. For example, a
small dog might be able to trot for long distances at 6 or 8 kilometers per
hour (4 to 5 miles per hour), whereas a goose of the same weight can easily
cruise at over 80 kilometers (50 miles) per hour. A migrating lemming
would be hard-pressed to average over 2 or 3 kilometers per hour (1 to
2  miles per hour), but a pigeon typically flies at over 40 kilometers per
hour (25 miles per hour). Even small flyers—a dragonfly or a sparrow, for
instance­—can fly much faster than I can run, and most can keep it up for
hours. So a flyer can cover much more territory, and cover it faster, in the
never-ending search for food or mates.

How Fast?

Now that even the humblest flyers have left us in the dust, what about the
extremes? Size plays a major role in flight speeds, as we shall see in Chap-
ter 3, so most of the fast cruisers are big; pelicans and golden eagles clock
in at around 50 kilometers (31 miles) per hour, albatrosses and swans at

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 5 ]
60 to 70 kilometers per hour (37 to 44 miles per hour), and Canada geese
at over 80 kilometers per hour (50 miles per hour).4 The true royalty of
high-speed flight, however, are the falcons. Although their cruising speeds
are only 40 to 50 kilometers per hour (25 to 30 miles per hour), peregrine
falcons attack other birds in flight by diving on them at high speed. These
steep dives, or “stoops,” give the prey less time to escape, and the falcon
usually hits its hapless victim so hard that it breaks the prey’s neck.
Trained falcons reached speeds of 130 to 140 kilometers per hour (81 to
86 miles per hour) in dives carefully monitored by scientists,5 and the
most reliable estimates for wild falcons give diving speeds in excess of
200 kilometers per hour (125 miles per hour). At the other end of the scale,
tiny fruit flies or mosquitoes can still cruise at the equivalent of a brisk
walk for us, 3 or 4 kilometers per hour (2 or 3 miles per hour).6 The ulti-
mate low-speed extreme, of course, is hovering, something most insects
and small birds can do.

How High?

How high do animals fly? For cruising in search of food, most fly just high
enough to get a good view. Worker honeybees, for instance, when search-
ing for flowers or returning to the hive, have no need to fly higher than 3
or 4 meters above the terrain. A great blue heron commuting from its nest
in a woodlot to a nearby river might fly 100 meters (330 feet) above the
ground, and a nighthawk chasing insects in warm, rising air might cruise
100 to 200 meters (330 to 650 feet) above the ground. A researcher fol-
lowed a flock of migrating cranes in a small airplane and saw them fly at
altitudes ranging from 150 to 1,000 meters (500 to 3,300 feet).7 Migrating
birds often fly much higher, both to take advantage of favorable winds at
different altitudes and to ensure that they clear any obstacles (hills,
mountains) in their paths. Typical migratory altitudes for birds are any-
where from 200 to 1,000 meters (660 to 3,300 feet) above the terrain, but
some rise much higher. Radar operators have tracked flocks of migrating
warblers and plovers over the Atlantic at between 1,000 and 5,000 meters
(3,300 and 16,000 feet), and pilots have reported eagles at 3,000 meters
(9,800 feet) and sandpipers at 4,000 meters (13,000 feet). Condors soar at
6,000 meters (20,000 feet) in the Andes, and bar-headed geese and curlew
sandpipers migrate over the Himalayas at 7,000 to 9,000 meters (23,000
to 30,000 feet) above sea level.8,9 Finally, airline pilots have reported
flocks of migrating swans at over 8,000 meters—over 5 miles high. Al-
though we humans have built vehicles that fly to the outer edges of the

[ 6 ]  On the Wing

atmosphere, consider this: for eight years after their first flight, the Wright
brothers only flew as high as they felt they could safely fall, typically 10 or
12 meters (30 or 40 feet) above the ground. About eight years after that
first flight, the Wrights and others finally started working to set altitude
records. Although a mere eyeblink in evolutionary time, for a significant
chunk of our century of human flight, people flew no higher than a three-
or four-story building.

How Far?

Pick a distance and some flyer probably forages or migrates that far. At the
low end, male scale insects may just fly from the branch where they
hatched to another branch on the same tree or on the nearest neighboring
tree to find a mate. Certain milkweed bugs on islands in the Baltic Sea lit-
erally migrate a few hundred meters from one side of the island to the
other. Most flyers, however, make more impressive trips. Foraging worker
honeybees may fly 3- or 4-kilometer (2- to 2½-mile) round trips collecting
pollen and nectar. That distance may not sound terribly remarkable, but
scaled to body size, it would be like me making a trip (with no map, com-
pass, or vehicle) of over 150 kilometers (90 miles), or about the distance
from Chicago, Illinois, to Milwaukee, Wisconsin. And the honeybee makes
several such trips each day.
On a more dramatic absolute scale, migrations include the pinnacles of
animal distance records. Many birds migrate thousands of kilometers,
and a substantial portion of them do so with few or no stops along the way.
Songbirds and shorebirds migrate nonstop over the North Atlantic from
eastern Canada and the northeastern United States to Caribbean islands
and Venezuela. Their journeys range from 3,000 to 4,000 kilometers
(1,800 to 2,500 miles), depending on their destination. Brant geese mi-
grate from Alaska to Hawaii nonstop, a distance of about 4,000 kilometers
(2,500 miles). One species of forest-roosting bat migrates about 1,400 ki-
lometers (870 miles) between upstate New York and Georgia. Even insects
migrate long distances. The best-known insect migrant is the Monarch
butterfly, which migrates up to 3,000 kilometers (1,900 miles) from the
northern United States to mountains in Mexico. Some seabirds win the
long-distance prize, however. The gull-like Manx shearwater migrates
10,000 kilometers (6,000 miles) between northern European islands and
Brazil. Several waders, including the curlew sandpiper, also migrate along
routes of approximately 10,000 kilometers (6,000 miles) each way. The un-
disputed champion of long-distance travel is the diminutive Arctic tern.

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 7 ]
This pigeon-sized bird breeds above the Arctic Circle and spends its non-
breeding time on the coast of Antarctica, for a one-way distance of over
20,000 kilometers (12,500 miles). Some terns make the whole trip in a
year, but others may spend two years in the Antarctic before returning
north.3

How Small or Large?

Several kinds of nearly microscopic insects make up the smallest flyers.

These include thrips, certain parasitic wasps, and a family of minuscule
beetles. Because thrips are so tiny—from 3 or 4 millimeters (less than
2⁄10  inch) down to ½ millimeter (2⁄100 inch) in length—few people other
than entomologists ever notice them. Thrips, however, are actually quite
common; they can often be found in large numbers on a single dandelion
flower. All the tiny insects in this size range have in common a rather
strange wing structure. Their wings look more like bristly rods or loose
feathers than insect wings (Fig. 1.1).10 Because air seems more viscous to
tiny creatures, these bristle-wings are just as effective at very small sizes
and low speeds as more conventional wings are at larger scales.
You might wonder how big flying animals can get. This question actu-
ally has several answers, depending on exactly how you ask it. Among
living animals, wandering albatrosses have the longest wingspan, almost

Figure 1.1:
Wings of very tiny insects. A. Thrips. B. Featherwing beetle. (Courtesy of S. T.)

[ 8 ]  On the Wing

3½ meters (12 feet). These are large birds, with a body mass of about 9 ki-
lograms or 20 pounds (compared with 5 kilograms [11 pounds] for a bald
eagle or 6 kilograms [13 pounds] for a Canada goose). Andean condors are
heavier than albatrosses, at about 12 to 15 kilograms (26 to 33 pounds).
Condors have slightly shorter but much broader wings. So albatrosses
have the greatest wingspans, but condors are the heaviest flying birds
alive today (Fig. 1.2).* If the question is restated as, “What is the largest
bird that ever flew?” the prize goes to an extinct group of vulture-like
birds called teratorns. Although they apparently looked and flew like vul-
tures, teratorns were probably predators rather than scavengers. They
ranged from slightly larger than condors (from fossils found in the La Brea
tar pits in California) to the enormous South American species, Argentavis
magnificens. This last species was stunningly huge, with an estimated body
mass of over 75 kilograms (165 pounds; more than me!), and a wingspan
of approximately 6 to 8 meters (20 to 26 feet).11
As big as it was, Argentavis was not the largest flying animal ever. The
largest flying animals were pterosaurs. Pteranodon ingens, which lived in

A B

Figure 1.2:
Very large birds. A. Condor. B. Albatross. (Courtesy of S. T.)

*  The kory bustard, a large bird of the African plains, is often described as the larg-
est flying bird, with typical weights of 13 kilograms (33 pounds), and a reputed max-
imum weight of 20 kilograms (44 pounds). The kory bustard much prefers to run and
is a marginal flyer at best. It is only able to take off from level ground with difficulty
and is unable to fly for more than a few seconds. It is probably less aerial than a
domestic chicken, and in my opinion, not a true, fully powered flyer.

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 9 ]
Figure 1.3:
Quetzalcoatlus northropi, the largest known flying creature. (Human figure for scale.)
(Courtesy of S. T.)

what is now the central United States, was similar in size to Argentavis,
with a wingspan of about 7 meters (23 feet). At one time Pteranodon was
considered the largest pterosaur, but in the 1970s, it was bumped from
this perch by the discovery in Texas of a partial wing skeleton of its huge
relative, Quetzalcoatlus northorpi (see Box 1.1. Species Names). Based on
size relationships of other pterosaurs, the estimated wingspan of Quetzal-
coatlus was 11 to 12 meters (over 39 feet; see Fig. 1.3), substantially greater
than a Piper Cub airplane! Weight estimates put Quetzalcoatlus just slightly
heavier than Argentavis, around 80 to 90 kilograms (176 to 200 pounds).12
So from smallest to largest, flying animals span an enormous size range,
from thrips with a wingspan of less than a millimeter to pterosaurs with
wingspans more than 10,000 times longer.

Vertical Takeof f and Hovering

In the normal course of going about their daily lives, all insects and most
birds can take off vertically or near-vertically from a standing start. More-
over, as amply demonstrated by the green June beetle with whom I opened
this chapter, insects and small birds can hover in one place when neces-
sary. In contrast, relatively few of our flying machines can hover or take off

[ 10 ]  On the Wing

Box 1.1:  SPECIES NAMES

In spite of a very long, complex set of rules governing the naming of

animal species—the International Rules of Zoological Nomenclature­—
scientists occasionally manage to sneak something amusing into scien-
tific names. For instance, Quetzalcoatlus northropi, the largest known
pterosaur, has a bit of whimsy in its name. Before we look into that, how-
ever, we first need some background on species naming conventions.
The scientific name of a species has two parts. The first part is also
the name of the genus containing that species. A “genus” is a group of
closely related species. The second part is the term that identifies the
particular species from among those in the genus. Apis mellifera, for ex-
ample, is the species name of the domesticated honeybee, just one of
several honeybee species in the genus Apis. The second term in the spe-
cies name (“mellifera” in this case) is not capitalized and is never used
alone: the species name is Apis mellifera, not mellifera. Note that genus
and species names are italicized, and genus names are capitalized
whether alone or part of a species name; genus and species names are
traditionally based on Latin or latinized Greek words, but they can also
be latinized words from other languages. Sometimes they are latinized
forms of the common name—Falco peregrinus for the peregrine falcon­—
but this is rare. They can be based on any words the scientist chooses
when he or she first describes the species.
The genus of Quetzalcoatlus northropi is named after an important
deity of pre-Columbian Mexico, the feathered serpent god Quetzalcoatl
(presumably because the fossils were from a flying creature found in
southern Texas near Mexico). The second term, northropi, refers to the
airplane manufacturer Northrop Company, which was well known in
the mid-20th century for designing all-wing, tail-less airplanes. This
pterosaur’s wing was so big and its tail so small, it undoubtedly would
have looked as if it were all wing when in flight.
Some species names are just fun to say, such as Saurus soarus, a spe-
cies of gliding lizard, and Aha ha, a small wasp. Then there is the extinct
genus of tiny, gnat-like flies, Iyaiyai (pronounced eye-yi-yi) and the
moth species Eubetia bigaulea (pronounced you-betcha bygolly). Pitohui
(pronounced pi-tooey) is the name of a genus of tropical birds with poi-
sonous flesh and supposedly represents the spitting sound made by the
scientist who discovered and tasted it. Species are often named in honor
of people, such as the beetle genus Garylarsonus, named after Far Side
cartoonist Gary Larson. (Larson is a favorite of biologists for his off-
beat biology humor, and his name appears in several other scientific
names.)

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 11 ]
 Box 1.1:  Continued

My final example is the clever but cumbersome Arthurdactylus conan-

doylensis, a pterosaur whose fossils were found in a South American
jungle. The name refers to the author Arthur Conan Doyle, writer of the
Sherlock Holmes mysteries. The pterosaur name refers, not to the
Holmes stories, but to a story of his called The Lost Worlds, in which ex-
plorers discover a living pterosaur in a jungle, capture it, and return it to
London.

vertically. Those that can—helicopters, jump-jets—tend to be exception-

ally complex, voraciously fuel-hungry, and much more challenging to pilot.
Engineers had created useful conventional airplanes several decades before
they were able to design practical helicopters, strong evidence of the great
technical challenge posed by hovering and vertical takeoff (and landing).
For both vertical takeoff and hovering, a flyer needs to produce lift
equal to or greater than its own weight without moving horizontally
through the air. In effect, the flyer must move its wings fast enough to
make up for the fact that its body is not moving horizontally. Geometry
works against big flyers because as they get bigger, their weight increases
with their length cubed, but their wing area and muscle force (related to
muscle cross sectional area) only go up with length squared. In other
words, if you double a flyer’s length, its wing area and muscle force increase
by fourfold but its weight increases by eightfold. Moreover, increasing
weight requires wings to move faster to produce enough lift to get the flyer
aloft (Chapter 3), so to hover, large flyers need to move relatively smaller
wings much faster with muscles that are effectively less forceful. When
animals get much bigger than small birds, this becomes a losing battle. A
house fly weighs so little that it can easily move its wings fast enough to
hover, a sparrow can barely move its wings fast enough to hover for short
periods, and an eagle’s muscles can flap fast enough for horizontal flight
but nowhere near fast enough for hovering. Humans are far outside the
muscle-powered hovering boundaries; we simply weigh too much relative
to our muscles’ abilities. A half-century of research on human-powered
aircraft showed that very fit humans can barely power an airplane in level
flight; hovering and vertical takeoff are effectively out of the question.
If machines that take off vertically and hover are more challenging to
operate, do we see any parallels in animals? Through the action of natural
selection, birds and insects have evolved an extremely sophisticated and

[ 12 ]  On the Wing

effective flight control system. Their nervous systems and senses are finely
tuned to control their flight with little or no practice or training. In fact,
vertical takeoff is an important escape response for many flyers and is a
hard-wired reflex in most small- to medium-sized flyers. Thus, animals
don’t necessarily have to learn these skills, but the specializations of their
nervous systems show the level of difficulty of the task.

Flying Both in Air and Underwater: “Amphibian” Flyers

Lots of swimming animals flap a wing-like flipper or fin to swim underwa-

ter; the basic principles of wings work just as well in the water as in air. Sea
lion and turtle flippers, whale tails, and the wing-like fins of stingrays are
all examples of underwater wings, or hydrofoils. Very few animals, how-
ever, can actually use the same wings to fly in air and swim under water.
For example, living birds of the auk family—murres, guillemots, puffins,
auklets—can “fly” in both air and water. These birds personify the “jack of
all trades, master of none” approach. In order for their wings to be strong
enough and small enough to work effectively under water, they are smaller
and heavier than wings of other seabirds of similar size. Their wings and
bodies tend to be “overbuilt” for flight so they can withstand the rigors of
swimming, but compared to totally aquatic swimmers like penguins and
seals, auks are light and fragile underwater.13 Small auks tend to be better
flyers and slower swimmers; in contrast, the largest auks tend to be fast,
powerful swimmers, but they have difficulty getting airborne unless they
can leap from a tall cliff or make a long takeoff run on calm water. (Murres,
the largest living auks, sometimes migrate by swimming when the water
is too rough for them to make their long takeoff run.) As we will see in
Chapter 9, the recently extinct Great Auks became so large that they com-
pletely lost the power of flight in favor of strong swimming.

From Air Combat to Aerial Sex

One of the most difficult skills for an aviator to master is air combat: in-
tercepting and attacking other aircraft in flight. In the Second World War,
the majority of fighter pilots never managed to shoot down another air-
plane, and roughly 10% of all fighter pilots made over half of all air-to-air
kills. Even with modern tools like radar, computerized gunsights, and
heat-seeking missiles, a pilot needs a great deal of skill and training to
successfully intercept and engage another aircraft.14 Contrast this with

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 13 ]
aerial predators: if a dragonfly cannot catch and eat a mosquito in the air,
it starves. If a bat or a swift cannot intercept a moth on the wing, or if a
falcon cannot pounce on a pigeon in flight, it goes hungry. For these aerial
predators, accurately intercepting and attacking other flying animals is a
matter of survival. If a fighter pilot misses too many interceptions, he or
she may be assigned to some other kind of flying. If a bat misses too many
interceptions, he or she will not be able to rear young and may ultimately
die. The stakes are high for these aerial predators, so natural selection
drove these animals to evolve the skills and instincts necessary to per-
form their aerial tasks.
Flying animals do an astonishing variety of things while in flight. Al-
though I don’t know of any flyers that give live birth in flight, a number of
insects lay eggs in flight. For instance, some dragonflies drop them on the
water surface in flight, and bee flies (whose larvae are parasites of bee
larvae) eject eggs into bee burrows while hovering over them. Many beach-
combers have discovered, to their chagrin, that seagulls defecate in flight.
Based on radar tracking, biologists think that swifts may actually sleep in
flight. I have often watched swifts and swallows drinking in flight by
swooping down close to the surface of a calm pond and briefly dipping
their beaks in the water to snatch a gulp, and swifts are reputed to bathe
on the wing, by dragging their belly and breast feathers in the water as
they skim the surface.
The most accomplished flyers even perform part or all of their mating
on the wing. Many male flyers chase rivals and potential mates in the air,
and they need to be able to tell which is which! Many species also have
elaborate courtship rituals amounting to an aerial dance that one or both
of a courting pair perform on the wing. Hawks and dragonflies are re-
nowned for the dazzling aerobatics that males perform to attract females.
Once the female shows interest, an aerial ballet involving both partners
ensues. If they both manage to do the dives, climbs, turns, and twists in
the proper sequence, they will copulate, and even this may occur on the
wing. Damselflies, mayflies, honeybees, hover flies, and possibly some
hawks all copulate in flight!

VARIET Y (AND EATING) IS THE SPICE OF LIFE

Flying animals acquire food in an unbelievable variety of ways, many of

which would be difficult or impossible without flight. Osprey and fishing
bats snatch fish from just below the surface while flying low over the

[ 14 ]  On the Wing

water. Brown pelicans and terns dive into the water to catch fish, while
cormorants and puffins chase fish underwater. Swifts, nighthawks, and
small bats catch and eat flying insects in the air. Thrushes and starlings
pluck insects from the ground and off low plants. Sandpipers and plovers
use their beaks to probe into sand on beaches just above the waves for tiny
invertebrates. Herons wade in shallow water and spear and eat any animal
that moves. Red-tailed hawks pounce on mice, voles, and rabbits from
high perches or while soaring whereas harriers do the same from a low
glide over open meadows. Bluebirds, wrens, and warblers glean insects
from tree branches, and woodpeckers probe under bark for insects. Tiny
fig wasps actually grow up inside a developing fig, living on nourishment
the fig plant provides. Adult fig wasps fly off to find another fig in which to
lay eggs, and, in the process, pollinate the fig. Adult bees, wasps, and
moths drink nectar from flowers (often while hovering) and bees feed
their larvae pollen. Moth larvae—­caterpillars—eat leaves, and in turn,
many wasps feed caterpillars to their larvae. Aphids, cicadas, and some
woodpeckers suck sap from plant stems and tree trunks and twigs. Ter-
mites and a variety of wood-boring beetles chew into and eat wood. Turkey
vultures, burying beetles, and flesh flies eat carrion. Several kinds of bee-
tles and flies eat dung or feed it to their larvae. Mayflies spend a year or
more as aquatic, predatory larvae, and then don’t eat at all as adults; they
spend a day or two flying about as adults, searching for mates, mating,
laying eggs, and then dying. Parasites feed on their hosts, and parasites of
various stripes are not uncommon among flyers. The larvae of the human
bot fly enter the skin at the site of a mosquito bite, feed on body fluids and
grow just under the skin, and break through the skin to pupate as horse
fly–sized adults. Frigate birds may have the most unbelievable lifestyle of
all. Frigate birds are seabirds that cannot swim, so they have great diffi-
culty fishing! A type of behavioral parasite, frigates often feed by stealing
food from other birds on the wing. They attack other birds such as gan-
nets, boobies, or gulls when the others are on their way home from a hard
day’s fishing. The frigate bird harasses its victim until the other bird regur-
gitates its fish, which the frigate catches and eats in midair. Yum!

RECONSTRUCTING EVOLUTIONARY HISTORY: PHYLOGENETIC TREES

One of the most potent tools for studying evolution, phylogenetic system-
atics, was developed by biologists over the latter half of the 20th century.
It has gone from being a niche technique used by a subset of taxonomists

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 15 ]
(when I was in graduate school in the early 1980s) to being a pervasive
method used throughout organismal biology.15 A modern biologist would
no more try to discuss the evolutionary history of a lineage without a phy-
logenetic tree than she would try to measure temperature without a ther-
mometer. So if we want to understand research on flight evolution, we’ll
need to know something about the products of phylogenetic systematics:
phylogenies.
Phylogenies or phylogenetic trees are tree-like diagrams that scientists
use to illustrate evolutionary relationships among species. In everyday
terms, phylogenies represent family trees of species (or groups of species),
and they illustrate ancestor-descendent relationships among species just as
a family tree illustrates ancestor-descendent relationships among people in
a family.
Scholars have been using tree diagrams to show various types of rela-
tionships among organisms since before Charles Darwin, and Darwin
himself used what amounts to a hypothetical phylogeny to help explain
natural selection in On the Origin of Species.16 Evolutionary biologists and
paleontologists continued to use trees to show evolutionary relationships
from the late 19th century into the mid-20th century, although exactly
what relationship was being illustrated and exactly how the tree was built
could vary. Indeed, many such trees were based as much on the author’s
opinion and intuition as on hard data.
In the mid-20th century, phylogenetic systematics was introduced and
it has gradually become the dominant approach to the study of evolution-
ary relationships. The goal of phylogenetic systematics is to study and
attempt to trace patterns of evolutionary relationships, based on the fun-
damental assumption that all life is descended from earlier life, and
modern species have thus descended from earlier species. Systematists
thus seek to tease out which species are most closely related to which by
comparing shared derived characteristics—features or “characters” that
are shared by all descendants of a common ancestor and have been inher-
ited from that ancestor. Traditionally, biologists have used characters
from anatomy, embryology, and paleontology to assemble phylogenetic
trees. Unfortunately, the vast majority of species that have ever lived are
now extinct, and we only have fossils from a tiny (and biased) sample of
extinct species, so paleontology is of limited help in most cases. Scien-
tists also face the problem of determining whether characters shared by
species are truly derived (inherited), versus having been evolved inde-
pendently in response to similar selection pressures. For example, do
swifts and swallows have long, thin, sickle-shaped wings because they
inherited them from a common ancestor or because they independently

[ 16 ]  On the Wing

evolved similar structures in response to similar selection pressures and
physical constraints? The latter would be an example of convergent evolu-
tion, which happens to be more likely in the swift-swallow case. When
biologists are limited to anatomical characteristics for building phyloge-
nies, convergences and shared, derived characters can be difficult to tell
apart,* which in turn can lead to ambiguities in the resulting phyloge-
netic trees.
Over the last decade or two, systematists have benefited from advances
in both computer and molecular biological technology. The great increases
in computer power coupled with huge decreases in the cost of computer
memory led to quantum leaps in the number of species and the number of
characters scientists could analyze on one tree. This allowed systematists
to build more detailed trees and to look at ever more inclusive groups. Sys-
tematists also gained a completely new source of characters as techniques
of molecular biology allowed researchers to compare similarities in DNA—
the actual genetic material—among extant organisms. Such gene-based
or molecular trees have gone a long way toward sorting out convergences
that confounded earlier anatomically based trees. Both the traditional and
the molecular approaches are in widespread use and in many cases are
combined in the same study. For extinct species known only from fossils,
normally the only available characters are anatomical (Jurassic Park not
withstanding).
With four or five species, using a small handful of characters, a person
may be able to assemble a phylogenetic tree in a reasonable amount of time
by trial and error. This approach quickly becomes impractical as the num-
bers of species and characters increase. Evolutionary biologists have devel-
oped several computational (computer-based) approaches for assembling
trees for large groups of species using dozens or hundreds of characters.
Regardless of the complexity of the method, any systematist will always
remind you that a phylogeny is a hypothesis: it is the researcher’s best esti-
mate of the patterns of descent and relationships among species. In the
real world, scientists face problems that challenge even the most sophisti-
cated tree-assembling programs. Lineages can lose characters inherited
from ancestors, or characters that appear to be inherited from a common
ancestor may have evolved independently in multiple lineages, or informa-
tion on characters may simply be missing for various practical reasons.
Moreover, when large data sets produce large, complex trees, the data often

*  Systematists usually call characters that are similar due to convergence “homo-
plasies,” to distinguish them from “homologies,” which are characters that are simi-
lar due to inheritance from a shared ancestor that had the character.

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 17 ]
fit equally well on more than one tree, so scientists have developed statisti-
cal methods to merge trees or choose trees most likely to be closest to the
actual pattern. Lacking some way to look back in time, however, we can
never know the actual patterns. Thus, even trees based on huge numbers of
characters are still only estimates of the actual evolutionary pattern.
Figure 1.4 shows some simplified phylogenetic trees to illustrate some
of their important features. First, the branch points represent ancestral
species that diverged to form new ones; they may be known (as from fos-
sils), but more typically they are hypothetical or assumed. Thus, the fewer
branch points between two species on the tree, the more closely they share

A
le
di an n
ou
t
og
rd ke oc
o oo f ar
za na ird um b ol
Tr Fr Li S B Cr H Ba W Be

B Frog C Salamander
Salamander Frog
Deer Human
Cow Chimpanzee

Chimpanzee Cow
Human Deer

Figure 1.4:
Sample phylogenies or phylogenetic trees. A. Tree oriented vertically, so downward rep-
resents going back in time. Each fork or “node” represents a species (usually implied or
hypothetical) ancestral to the species or nodes above. The “root” of the tree at the very
base represents the common ancestor of all the species on the tree. Species (or groups of
species) that are each other’s closest relatives are called “sisters”; on this tree wolves and
bears would be sister groups. B. Phylogenies are often rotated sideways to save space; in
this case, leftward represents back in time. C. Because phylogenetic trees can rotate freely
about any node, this tree is exactly equivalent to tree B. (Courtesy of S. T.)

[ 18 ]  On the Wing

a common ancestor. Wolves and bears share a closer common ancestor
than humans and bears. Lizards share a fairly recent common ancestor
with snakes, a more distant one with crocodiles, and yet a more distant
one with baboons. One subtle feature is that branches are free to swivel
about the branch points or nodes. The trees shown in Figures  1.4B and
1.4C both show the exact same evolutionary relationships. Humans are
traditionally shown on the far right of vertical trees, subtly yet incorrectly
suggesting that we represent some sort of evolutionary pinnacle; note
how Figure 1.4A avoids that by rotating humans in toward the center.
Fossils play a role in phylogenies in a couple of ways. If fossils can be
included on phylogenies with extant species, they can give information
about when lineages arose or split in actual geological history rather than
simply relative to when other lineages split. This also helps scientists de-
termine which characters are truly ancestral. More important for our pur-
poses, phylogenies can be constructed entirely using fossil species, which
can help explain how characters evolve. For example, if we can look at a
phylogeny of possible bird ancestors and see where feathers appear on the
tree, we can answer questions about whether feathers evolved as a flight
adaptation or for some other function in non-flying bird ancestors.

LESSONS FROM THE BEETLE

The green June beetle from the beginning of this chapter nicely represents
the end result of evolution acting on the beetle’s ancestors to improve
flight. The beetle’s powerful flight muscles, under the control of its special-
ized nervous system, allow it to hover, take off and land vertically, and fly
very fast. Its wings are light enough to flap without undue effort but strong
enough to carry its body weight. Its exoskeleton is strong enough to anchor
those powerful flight muscles as well as acting as protective armor, yet not
too heavy to carry in flight. Its small size makes hovering relatively easier
and fast flight relatively harder than for a larger flyer like a goose. The ev-
olution of these specializations—flight muscles, control systems, me-
chanical properties, and structures—and how they are affected by overall
size are some of the major threads we will follow looking at the evolution
of flight in other animals.

REFERENCES
1. J. S. Edwards (1986) Northwest Environmental Journal.
2. K. Schmidt-Nielsen (1972) Science.

C a n ’ t T e l l t h e P l a y e r s w i t h o u t a   Sc o r e c a r d   [ 19 ]
 3. D. E. Alexander (2002) Nature’s Flyers: Birds, Insects, and the Biomechanics of
Flight.
4. H. Tennekes (1996) The Simple Science of Flight: From Insects to Jumbo Jets.
5. V. A. Tucker, T. J. Cade, and A. E. Tucker (1998) Journal of Experimental Biology.
6. J. H. Marden, M. R. Wolf, and K. E. Weber (1997) Journal of Experimental
­Biology.
7. C. J. Pennycuick, T. Alerstam, and B. Larsson (1979) Ornis Scandinavica.
8. L. W. Swan (1970) Natural History.
9. I. Newton (2008) The Migration Ecology of Birds.
10. R. F. Chapman (1982) The Insects: Structure and Function.
11. K. E. Campbell and E. P. Tonni (1983) Auk.
12. S. Chatterjee, and R. J. Templin (2004) Geological Society of America Special
Papers.
13. D. Boag and M. Alexander (1986) The Atlantic Puffin.
14. J. F. Dunnigan (2003) How to Make War: A Comprehensive Guide to Modern War-
fare in the Twenty-first Century.
15. E. O. Wiley and B. S. Lieberman (2011) Phylogenetics: The Theory of Phylogenetic
Systematics.
16. C. Darwin (1860) On the Origin of Species.

[ 20 ]  On the Wing

CH A P TER  2

Theme and Variations

Similarities and Differences among Nature’s Flyers

P igeons and dragonflies are both skilled flyers, and they both fly by flap-
ping their wings. The flapping process is essentially the same in these
two flyers (Chapter 3), but you need only a quick look to see that the struc-
ture of their wings is completely different. The pigeon’s wing surface is
made mostly of an array of dozens of partly overlapping feathers, sup-
ported by a highly modified front leg; the dragonfly wing uses a complex
arrangement of rigid struts—called veins—supporting a thin, flexible
membrane. Birds and insects followed very different evolutionary routes
to achieve flight. Although their wings do the same thing, natural selec-
tion worked with different raw material in evolving birds’ and insects’
wings. What structures evolved into wings, and what did the animal use
those structures for before evolving wings? These are among the most im-
portant questions in the evolution of animal flight. They have inspired
many vociferous arguments, and even today we have only partial answers
to them.
Natural selection, the process that leads to most evolutionary change,
can work only with the materials and structures that an organism already
possesses. As a potential flying animal, I might have more effective wings
if I could make them out of fiberglass or aluminum, but if I am a vertebrate
animal, I am stuck with using (and modifying) things like skin, bone,
muscle, and tendons. Likewise, if I were an insect, I might be able to make
stronger wing veins if I could use a graphite-epoxy composite, but I am
limited to using the material that makes up my exoskeleton. Natural se-
lection works by gradually modifying what an animal already has, not by
providing totally novel materials or structures. Biologists call these limi-
tations “historical constraints.” If I am an insect, I am unlikely to evolve
feathers because my outer body covering, an exoskeleton, is non-living
and does not grow. Moreover, my exoskeleton is mostly made up of chitin,
a very different substance from keratin, the material that makes up a
bird’s feathers (keratin is a major component of vertebrate skin but not
found in insects). These are historical constraints; my ancestry limits how
much natural selection can change my lineage.

THE KEY: FUNCTION OF PROTOWINGS

In order to evolve a wing in the face of historical constraints, natural se-

lection must have acted on some structure that had at least slightly wing-
like properties, perhaps a large surface area or something extending out
on each side of the body. Scientists sometimes call such structures “pro-
towings,” but this name can be misleading because the original structure
most likely had no aerodynamic function at all. Bat wings, for example,
are modified front legs, and the ancestors of bats undoubtedly used them
for climbing trees and walking. If natural selection, in improving the orig-
inal function of the protowing, also by coincidence gave it a bit of aerody-
namic ability, the protowing would have become a structure with two
functions. Perhaps long arms and big hands for climbing also helped sta-
bilize the creature’s leaps between branches; if this were a big enough ben-
efit, further increase in surface area for even better stabilizing might
occur, allowing natural selection to act on the aerodynamic function and
beginning the process of evolving true wings. The protowing would go
through a transition period, when it would retain the original function
while improving its flight functions, such as a front leg being used for both
climbing and gliding. If gliding proved to be sufficiently beneficial, natural
selection would favor improving the aerodynamics of the protowing to the
point that it becomes a fully functional wing.1 This process of evolving a
new use for an existing structure and eventually losing the original func-
tion is called “exaptation.” Even quite complex structures can evolve this
way. (See Box 2.1. Evolution of Complex Structures.)

MUSCLE-POWERED FLIGHT

All flying animals use the same “engine” to power their flight: muscles and
muscle tissue. Muscle tissue works the same way in all animals, and

[ 22 ]  On the Wing

 Box 2.1:  EVOLUTION OF COMPLEX STRUCTURES

Can natural selection produce a structure as finely tuned and special-

ized as a flapping wing? Since the time of Charles Darwin, skeptics of
evolution have raised the same objection: since a complex organ like a
vertebrate eye or a mammalian kidney works only if it has all its parts,
how can it have evolved gradually? If my eye is missing any of its major
parts—lens, iris, retina, cornea—it doesn’t work. Darwin himself re-
sponded to this objection and explained how such complex organs
could have evolved gradually, and his explanation still suffices today.
His key point is that the function must change as the structure becomes
more complex. For instance, an eye does not have to form a sharp image
to be useful (even though I would be considered “blind” if my eye does
not form an image): animals as simple as jellyfish and flatworms have
“eyes” that really just detect the presence or absence of light. Imagine
that the skin over such an eye becomes thickened to form a lens. Now
this eye is directional, so the animal can tell more precisely where the
light is coming from. Such an eye might be very useful for avoiding
predators, even though it does not form an image. In addition to being
directional, however, lenses focus an image whether or not the light
sensors are capable of distinguishing that image. Initially, light sensors
would not have been able to resolve an image, but any increase in the
size or resolution of the light sensors would form a retina and allow the
eye to start detecting a crude image. Further elaboration of the retina
to sharpen and refine the image would lead to a fully functional eye.
So, yes, natural selection is up to the task of forming complex struc-
tures such as eyes, kidneys, and wings. Both the function and the struc-
ture start out simple—and possibly unrelated to later functions—and
as they both are refined by natural selection, the function becomes
more sophisticated right along with the structure.

muscle tissue from an eagle and from a mosquito are quite similar, down
to the arrangement of protein filaments inside the cells that do the work.
The key operating principle of muscle is that it works by shortening or pull-
ing.2 Muscles can’t push or lengthen forcefully, and they can’t rotate. An
engineer would say that they are linear motors that work in only one di-
rection. Muscles are usually arranged in antagonistic pairs, where each
member of the pair reverses the action of the other member—in humans,
the biceps muscle bends the elbow, and the triceps straightens the elbow,
making them an antagonistic pair. So all flying animals have one (or more)
downstroke muscle and one (or more) antagonistic upstroke muscle.

T h e m e a n d Va r i a t i o n s   [ 23 ]
 Knowing how muscles work, we can see that flapping to power flight is
a natural arrangement. Antagonistic muscles pretty much require some
sort of to-and-fro motion rather than a continuous rotation. All four
animal groups that use powered flight independently evolved the up-and-
down flapping movements because they all use muscles with the same op-
erating properties.
Muscles provide both power and force for flapping flight. People use
“power” and “force” interchangeably in everyday speaking, but in science
they are quite distinct. Force is a push or pull, something that can acceler-
ate or decelerate an object. Power is the rate of doing work, that is, work
performed divided by the time to perform it. Force and power are con-
nected by work, because work is the force applied to an object times the
distance moved. (If the object does not move, then no work is done and no
energy is consumed: as my former graduate advisor liked to point out, the
chain holding a chandelier provides the force for free, no fuel required.) If
I use a rope to lift a 1-kilogram (2.2-pound) brick 10 meters (33 feet), I
must pull on the rope with a force of 9.8 newtons (2.2 pounds-force), re-
gardless of how fast I raise the brick. If I raise it quickly, however, I use
more power than if I raise it slowly. I use 10 times as much power raising
it at 10 meters per second as raising it at 1 meter per second, even though
I pull it with approximately the same force. In both cases, I have done the
same amount of work—9.8 newtons times 10 meters, or 98 joules of
work—but in the first case I have done the work 10 times as fast, so I have
used 10 times as much power.
Wing physics dictates that animals must move their wings very fast for
them to be effective, so power is important. As a general rule, flight re-
quires more power than walking because the flapping movements must be
much faster.3 Curiously, the amount of force the muscles must provide
need not be much different from walking or running, so the muscles don’t
need to be unusually strong, but they do need to be fast. So, to supply the
needed force while moving very rapidly, wing muscles must be powerful
due to speed rather than force.
The high power requirements of flight run up against another feature
of the way muscles operate. Muscle tissue is specialized for either long
periods of continuous use or brief bursts of intense activity. The “contin-
uous use” type is called “aerobic” because it requires a continuous supply
of oxygen while it is active. The “burst” type is “anaerobic,” meaning
that it does not require oxygen during its bursts of activity.4 Why the
difference? Anaerobic muscle tissue can react faster and it is usually
stronger—it can lift a heavier load than the same amount of aerobic
muscle—but it produces lactic acid as a byproduct during activity. 5

[ 24 ]  On the Wing

Anaerobic muscles can react faster partly because they run on their own
stored fuel (carbohydrates), but once this fuel is used up, they must rest
while the body replenishes their fuel. Moreover, when lactic acid builds
up to high levels, the muscles cannot contract any more until the lactic
acid is recycled or removed. Aerobic muscles, in contrast, are not as
strong, but they can continue to contract as long as the body supplies
them with oxygen and fuel (usually fats). In vertebrates, the difference
in muscle tissue is easily visible: aerobic muscles contain an oxygen
storage pigment, myoglobin, that gives them a deep red color. Anaerobic
muscles lack myoglobin and tend to be pale pink or white. This differ-
ence is the basis for “dark” and “white” meat in chicken: dark meat is
aerobic muscle and white meat is anaerobic. Chickens spend almost all
their time on the ground and fly only in emergencies, so their leg mus-
cles are aerobic and their flight (breast) muscles are anaerobic. In con-
trast, most ducks, which are powerful long-distance flyers, have dark
breast meat—their flight muscles are aerobic.* To see the tradeoff and
understand the consequences for an animal’s body, think of aerobic
muscles working on a “pay as you go” basis. They must receive a contin-
uous supply of oxygen and fuel, but they produce little or no lactic acid.
Anaerobic muscles, however, work on credit. They can work quickly and
forcefully, but they produce a toxic byproduct, lactic acid, that the body
recycles via oxygen-consuming reactions; biologists even call the lactic
acid buildup the “oxygen debt.” Clearly, flying any great distance or for
any substantial duration will depend on aerobic flight muscles. Some of
the controversy over birds’ ancestors (Chapter  6) revolves around
whether the ancestors of birds (or the earliest flying birds themselves)
had aerobic flight muscles.

EVOLVING EXPERTISE
Brains and Sense Organs

The flapping movements of powered flight are much more than a simple
up-and-down motion (as we will see in Chapter 3). To fly effectively, ani-
mals had to evolve nervous systems that could produce the appropriate
movement patterns and reflexes, not only to fly straight and level but to

*  Our muscles, like those of other mammals, usually contain both types of cells,
although most muscles tend to have substantially more of one type or the other. Just
to add confusion, we mammals even have a couple of intermediate cell types that
have properties in between aerobic and anaerobic muscle cells.

T h e m e a n d Va r i a t i o n s   [ 25 ]
maneuver as well. Flying crows or bumblebees don’t use their tails like
boat rudders to turn; they change direction by modifying their wingbeat
pattern. These wing movements would have been quite different from
walking or climbing, and they were not automatic in the beginning. Natu-
ral selection would have modified the nervous system’s “walking” control
patterns to allow them at first to control both walking and flying and later
to primarily control flight.
Sense organs would have changed as well. Flyers tend to fly as much
as 10 or 15 times faster than a terrestrial animal of similar size can run.
So flyers need faster and more precise sensing to avoid obstacles than
their ground-bound relatives. Flyers tend to have better vision than sim-
ilar non-flying animals, for example. Moreover, flyers need to sense new
things: is the air flowing smoothly over my wings? Am I rising or sink-
ing? Clearly, to get the most out of flight, flying animals need to evolve
different sensing abilities. Early on, such sensing (and the associated
control systems) would have been crude and rudimentary, but selection
for more effective flight mechanisms produced the highly specialized
flyers present today.

Stability and Maneuverability

Modern airplanes tend to be either “stability configured” or “control con-

figured.”6 Stability-configured airplanes tend to fly straight and level by
themselves; if knocked off kilter by a gust or turbulence, they tend to
right themselves automatically, just due to their physical arrangement.
­Control-configured airplanes don’t have built-in stability; they tend to
change direction or orientation with the slightest gust. Strongly control-
configured airplanes can be so unstable that they require some sort of
automatic-stabilizing control system to actively tame their lack of stabil-
ity and make them flyable.
Why have one versus the other? Stability-configured airplanes are
easier and safer to fly, are simpler to build, and may be more economical to
operate (because they tend to fly straighter, a straight line being the short-
est distance between two points). Training and most transport and pas-
senger airplanes are normally stability configured. Because they are so
stable, these craft tend to resist steering efforts, so they are not very agile.
Conversely, the main advantage of control-configured airplanes is that
they are extremely agile and maneuverable. Military combat aircraft, es-
pecially fighters, are often control configured. In combat, maneuverability
outweighs economical operation.

[ 26 ]  On the Wing

 Modern flying animals are, by and large, control configured. They must
continuously monitor and adjust their flight path, because they have lim-
ited built-in stability.7 They have what amounts to an automatic stabilizing
system of sensors and reflexes built into their nervous systems. This ar-
rangement gives flying animals unparalleled maneuverability. Humans
have yet to build a flying machine with half the aerial agility of a robin or
house fly.
Over half a century ago, aeronautical-engineer-turned-evolutionary-
biologist John Maynard Smith noticed the extreme maneuverability of
modern flying animals and realized that it represents a potent specializa-
tion. He pointed out that the earliest member of any lineage to fly could
not have had such specialized abilities; it would have needed a physical
design with more built-in stability.8 He went further and pointed out that
we should expect primitive flyers to be inherently stable; conversely,
highly maneuverable, inherently unstable flyers with active stabilizing
mechanisms must therefore be more derived (less primitive). In a sense,
Smith’s theory is hard to demonstrate, because all modern flying animals
are descendants of lineages that have been flying for tens of millions of
years. His concept turns out to be helpful, however, in explaining poten-
tially stabilizing features of very primitive flyers like Archaeopteryx and
early pterosaurs (as we will see in Chapter 6).

OTHER DESIGN TR ADE- OFFS

Modern flying animals have largely evolved into control-configured flyers

with active stabilization. Other design trade-offs have not been so clear-
cut. (See Box 2.2. Concept of “Design” in Evolution.) The trade-off between
structural strength and lightness, for example, depends on the animal’s
size, among many other aspects of its biology. Sometimes anatomical
structures don’t scale up or down very well. Moths and mosquitoes, like
all insects, have exquisitely light, somewhat umbrella-like wings that are
surprisingly strong for their weight. Scaled up to the size of a crow’s wing,
however, they would be heavy and too weak to carry the crow’s weight
(Chapter 3). The crow needs a lighter, stronger wing, and, like all birds, its
wings are of a completely different design from an insect’s, based on a
modified front leg and feathers. Even among birds, some have wings that
are more lightly built while others have more robust wings. Gulls have
long, lightly built wings suitable for flying and soaring in wide-open
spaces, whereas puffins have short, strong, heavy wings because they use
their wings to swim underwater as well as to fly.3,9

T h e m e a n d Va r i a t i o n s   [ 27 ]
 Box 2.2:  CONCEPT OF “DESIGN” IN EVOLUTION

Biologists, as a rule, tend to avoid using the term “design” when ap-
plied to living organisms. This reluctance is not because the concept of
an “animal’s design” is somehow invalid but because they fear that
some people may take it to imply the work of a conscious designer.
Such self-censorship is unfortunate because when we compare dif-
ferent structural features of, say, a bird wing and a bat wing, which
have important and obvious functional consequences, we are really
comparing two different wing designs. The overwhelming majority of
biologists take it as given that these two different wing forms resulted
from evolution operating through natural selection. When I, as a biolo-
gist, mention the “design” of some feature of an animal, I am simply
referring to the overall structural arrangement, usually implying vari-
ous functional advantages and disadvantages. I am not implying that a
deity or supernatural designer is necessary, just that the animal has a
particular structural arrangement different from other animals. The
“designer” is natural selection.

As specialized as flying animals are, sometimes processes other than

flight affect their weight. Fat storage for migration or eggs about to be laid
can greatly increase a flyer’s weight. This weight increase requires stronger
(and heavier) wings in order to carry flight loads. If animals evolve wings
strong enough for periods when their bodies are unusually heavy, their
wings might appear to be “overbuilt” at other times.

Speed versus Distance

Fast flight speeds trade off against flight distances (or duration, which
amounts to nearly the same thing). High flight speeds are clearly useful in
some situations, such as catching prey or avoiding predators, for example,
or covering lots of territory quickly while searching for food or mates.
High flight speeds, however, require disproportionately more power,
meaning that a fast flyer runs out of fuel sooner and covers less total dis-
tance. Lower speeds are more economical, so animals flying long distances
tend to fly a good bit slower than their maximum speed. Moreover, flight
forces are greater when an animal is flying fast, so it needs a stronger body
structure to fly very fast. A more leisurely flyer can get by with a lighter

[ 28 ]  On the Wing

structure. The choice depends on the animal’s lifestyle. For their size,
ducks and falcons and hover flies fly fast, butterflies and gulls and vul-
tures fly slowly.10

Drag Reduction

Any object moving through the air experiences drag, a slowing force or a
force that tends to decelerate the object. Most flyers would benefit aerody-
namically from reducing drag as much as possible. For example, fast flyers
could fly faster on the same power, or long-distance flyers could fly farther
on the same fuel load, if their drag was lower. What can natural selection
do to reduce flyers’ drag?
For larger animals—most birds and bats, and the largest insects­—
streamlining and smoothing the body surface reduces drag. For small- and
medium-sized insects (mosquitoes or house flies, for example) streamlin-
ing is less effective, and being more spherical to reduce surface area is more
effective (Chapter 3).
For some animals, drag reduction is of paramount importance. Fast-
flying falcons are highly streamlined; so are gulls and terns, which often
fly very long distances. Turkeys, on the other hand, are not nearly so
streamlined. Turkeys walk a lot more than they fly, so drag reduction is of
negligible benefit to them. Or consider mosquitoes: neither streamlined
nor spherical, they have big feathery antennae, long legs, and roughly cy-
lindrical abdomens. The ability to locate and walk on hairy prey—big an-
tennae and long legs—and swell up with a big blood meal—cylindrical
abdomens—clearly take precedence over drag reduction in these crea-
tures. In a similar way, the male pheasant’s long tail or the shape modifica-
tions that allow a leafhopper bug to look like a thorn or a katydid to look
like a leaf surely incur drag penalties. The benefits, however—the pheas-
ant attracting mates, the insects hiding from predators—in these cases
outweigh the drag costs.

HOW TO BUILD A WING: FOUR VARIATIONS ON A THEME

In the next chapter, we will look in detail at some of the aerodynamic

characteristics of effective wings. For now, let us assume that a wing needs
to be a large, light, more-or-less flat surface with a good bit more surface
area than a typical leg. Its performance will improve some if it is stream-
lined in cross section, that is, a bit thicker in front than at the back. It will

T h e m e a n d Va r i a t i o n s   [ 29 ]
also perform better if it is a bit “cambered,” meaning the top is convex or
slightly bowed upward. The wing needs to be able to flap, which involves
rather complicated up/down, fore/aft, and twisting movements. So it
needs a joint or articulation with the body that allows it to make these
complex movements while still supporting the body’s weight in flight.
Some ability to change the wing’s shape or area is also handy, because dif-
ferent circumstances may require adjusting the wing. For instance, birds
greatly reduce their wing area during steep dives compared with level
flight. An animal’s wing has other, non-aerodynamic requirements as
well. It usually needs to fold up in some compact way so that the animal
can get around when not flying, for example. These are the primary re-
quirements of a flapping wing, although particular animals may have
other, more specialized requirements.

Bird Wings

Of all the flying animals, the ones people are probably most familiar
with are birds. Bird wings are highly modified front legs, but surpris-
ingly little of the wing is skin, flesh, or bone. In a typical bird wing, at
least 75% of the wing area is just feathers. The part of a bird’s wing cor-
responding to our upper arm, forearm, hand, and fingers is much short-
ened and extends little more than half the length of the wing. The
upper arm bone—­humerus—and forearm bones—radius and ulna—
though small, don’t look all that different from ours (Fig. 2.1). The wrist
and hand, however, are massively reduced, from more than a dozen
bones in mine (not counting finger bones) to a mere three in a pigeon.
As for fingers, birds only have three: a semi-mobile thumb with one or
two bones, and a large (two bones) and a small (one bone) finger func-
tionally fused into a single unit. The thumb carries a small group of
feathers, the alula or bastard-wing, which the bird probably uses to
adjust the airflow over the main wing. Although they are relatively
shorter and fewer in number, the hand and finger bones of birds look
quite stout compared to ours.11
The main flight feathers (primary and secondary feathers) are attached
by ligaments directly to bones. The large primaries that make up the
wingtip attach to the hand and finger bones. The secondaries, which make
up the back half of most of the wing, are attached to one of the forearm
bones (Fig. 2.1).
We can move the joints in our arm and hand independently, but a bird’s
joints, muscles, and ligaments all work together. When a bird straightens

[ 30 ]  On the Wing

 Thumb Hand
Forearm bone bone
bones Finger
bones

Secondary Primary
feathers feathers

Figure 2.1:
Bird wing skeleton and feathers, with bones shaded. (Human arm shown for comparison,
not to scale.)

(extends) its elbow, its wrist and fingers also extend; when it bends (flexes)
its elbow, the other joints flex automatically as well.11,12 Moreover, the pri-
mary and secondary flight feathers are attached to the skeleton and to
each other at their bases so that when the arm and hand extend, the feath-
ers fan out like a hand of playing cards. When the arm and hand flex, the
feathers slide together in a neat stack automatically.13 This movement il-
lustrates the elegance of the bird wing’s design: the bird can fully extend
its wing or it can partly flex it to reduce its area, and the wing always stays
smooth and stiff because the feathers automatically overlap each other as
the wing flexes. So birds can change the size of their wings in flight much
more than other flying animals. Also, losing a few flight feathers doesn’t
keep a bird from flying, the way a damaged wing might ground other
flyers. In fact, birds replace their flight feathers annually or “molt,” and
most birds molt their flight feathers a few at a time and never lose the abil-
ity to fly.
Feathers are remarkable structures. They are very light for their sur-
face area, yet stiff enough to carry flight loads. Contour feathers—the
flat, stiff ones, not the fluffy down feathers—can have their side branches
or “barbs” thoroughly mussed and be preened back into a smooth, stiff
surface. The Polynesians’ armor made from bird feathers was not just
for ornamentation. Several layers of contour feathers can provide a
surprising amount of protection from impacts and penetration (as
­
birds, no doubt, discovered quite a long while before the Polynesians).
Each primary feather has an airfoil-shaped cross section (Chapter  3)
and in many birds, these primaries can each function as a tiny, individ-
ual wing.

T h e m e a n d Va r i a t i o n s   [ 31 ]
Bat Wings

The flight surface of a bat’s wing, unlike a bird’s wing, is directly supported
by the skeleton (Fig. 2.2). Bat and bird wing skeletons, in fact, have spe-
cialized in almost opposite directions. Where bird hand and wing bones
have become shorter and stouter, with lots of fusion and loss of bones, bat
hand and finger bones have become greatly elongated and narrowed, with
little fusion or loss of bones. The bat wing consists of a large, stretchy,
multi-layered membrane supported by the greatly elongated arm, hand,
and finger bones. The membrane is anchored to the body wall all along the
bat’s trunk and extends back along the legs to the ankles, as well as from
the legs to the tail.14
Biologists divide the wing membrane into four regions, two small and
two large ones. The propatagium or prowing is a small triangle in front of
the elbow, from the shoulder to the wrist. The handwing is the membrane
between the bones of the hand and fingers, which makes up approximately
the outer half of the wing. The inner half is the armwing, which runs from
the handwing (last finger) to the body wall and leg and makes up the inner
half of the wing. Finally, the uropatagium or tailwing runs between the
two hind legs and usually incorporates the tail.15 The membrane is much
more than just a top and bottom layer of skin. It contains an elaborate net-
work of fibers, some tough and others stretchy, plus several of its own
highly specialized muscles. Bats use these fibers and muscles to help con-
trol the shape of the wing in flight.16
Imagine taking your arm skeleton, lengthening and thinning the upper
arm and forearm bones, and greatly lengthening your hand and most
finger bones, while leaving your thumb and wrist bones about the same
size. That is pretty much the arrangement of the bat wing skeleton. The

Wrist bones
Forearm Thumb bone
bone Hand
bones

Finger bones

Wing membrane

Figure 2.2:
Bat wing structure, showing bones and wing membrane or patagium (shading and human
arm as in Figure 2.1).

[ 32 ]  On the Wing

clawed thumb is free of the wing membrane and relatively mobile. Bats
use their thumbs for grasping and climbing. The second digit, the bat’s
index finger, is less than half as long as the other wing fingers and acts
mainly as a stiffener of the wing’s leading edge, reinforcing the third digit.
The third, fourth, and fifth digits, corresponding to our middle, ring, and
pinkie fingers, support most of the handwing (Fig. 2.2); their joints give
the bat great ability to modify the wing’s shape, particularly its curvature
or camber (Chapter 3).
Bats have evolved a striking functional similarity to bird wings: bat
wings also have an automatic flexing and extending mechanism. As with
birds, when the bat extends its elbow, that movement also causes the
wrist and hand to extend.17 The structure is different in bats, however. In
birds, the shape of the elbow joint and the parallelogram-like arrange-
ment of the forearm bones are as important as the muscles and other soft
tissue, but in bats the mechanism relies only on muscles.15
Although a number of books and reviews state that the armwing sup-
ports the body’s weight and the handwing produces thrust, this descrip-
tion is a great oversimplification.15,17 In fact, the whole wing produces
upward and forward forces. The proportion of upward force is high at the
base and decreases toward the tip, while the proportion of thrust is high
at the tip and decreases toward the base. The change in upward and for-
ward force along the wing is gradual, and there is no sharp division be-
tween lift- and thrust-producing regions.3 This distribution of upward
and thrust forces applies to any flapping wing, not just those of bats.

Pterosaur Wings

Pterosaur wings look superficially more like bat wings than bird wings.
The basic pterosaur wing arrangement had a single, long, supporting skel-
etal arc at the front bearing a membrane stretching from the wingtip to
the body wall and back to the legs (Fig. 2.3). The skeleton that formed the
front of the wing had unremarkable arm bones, hand bones that were
short in some species and elongated in others, and one tremendously elon-
gated finger.* The wing finger typically was longer than the rest of the
wing skeleton combined. Pterosaurs also had three normal-sized, clawed
fingers separate from the wing membrane, making the elongated wing

*  “Pterodactyl,” the alternate common name for the group, literally means “wing
finger.” Scientists tend to prefer “pterosaur” (“winged lizard”) as the common name
because “pterodactyl” technically refers to one subgroup of pterosaurs (see Box 8.1).

T h e m e a n d Va r i a t i o n s   [ 33 ]
 Hand bones
Forearm
bones Digit 1, 2, 3

Digit 4

Wing membrane

Figure 2.3:
Pterosaur wing structure (details as in Figure 2.2).

finger the fourth digit.18 From fossils preserving impressions of the wing
membrane, the membrane was apparently flexible enough to fold up when
the animal was not flying.19
Scientists long thought that pterosaur wings were sort of simplified bat
wings; they now think pterosaur wings were actually quite different from
bat wings. For over a century, scientists envisioned the pterosaur wing
membrane as a flexible sheet that would have billowed freely, with no
means of controlling its shape other than the tension of the fingertip
stretching the membrane away from the body. A couple of decades ago,
when paleontologists first looked seriously at the mechanics of pterosaur
wings, they realized that a floppy membrane would not allow the ptero-
saur to flex its wings to reduce area (it would go slack if flexed), and keep-
ing the membrane stretched taut would place improbably high loads on
the wing finger’s slender skeleton. Back in the late 1800s, paleontologists
noticed what appeared to be some sort of fiber in impressions of the wing
membrane. When modern scientists looked more closely at those fibers,
they concluded that the fibers were probably membrane stiffeners.20 The
fibers run in the same direction as the shafts of primary and secondary
feathers of bird wings,21 and if they were stiff like cartilage or fingernails,
they would have been ideally suited to keep the membrane spread out:
they would have kept the membrane from getting narrower as the ptero-
saur stretched out its wings. Stiff fibers would take some of the load off the
finger bones and also would have allowed the pterosaur to flex (shorten)
its wing in flight, to reduce the wing’s surface area without allowing it to
billow or go slack.21,22
So far, we cannot tell from the fossils if pterosaurs had an automatic
wing extension-flexion mechanism like birds and bats—and because we
only know them from fossils, we may never know. A close study of the
joints may give some hints, but without the muscles, we won’t be able to say
for sure. Since the muscles themselves don’t normally fossilize, scientists

[ 34 ]  On the Wing

must reconstruct muscles based entirely on markings on the bones, and
this technique is not likely to be accurate enough to answer this question.

Insect Wings

Unlike the wings of all other flying animals, insect wings are not modified
legs; indeed, insects are the only flapping flyers that did not give up any
legs in the process of evolving wings. (Just what was the precursor for
insect wings is the subject of a spirited debate; see Chapter 5.) Insect wings
are attached to the middle body region, or thorax, on the sides near the
top; the legs are attached to the bottom of the thorax, and adult insects
normally have six legs, regardless of whether they have wings.
The insect wing structure is completely different from the others we
have seen. Insects are arthropods, meaning they have a rigid exoskeleton
outside their body rather than an internal skeleton like us. The exoskele-
ton is a secreted, non-living material, in somewhat the same way that our
fingernails or hair are non-living. Unlike nails or hair, however, insect
exoskeletons don’t grow once they have formed, and they can’t be repaired
or modified. In fact, insects must periodically shed their exoskeletons and
secrete and harden a new one in order to grow. Insect wings form from
modified regions of exoskeleton. The material of exoskeleton, called cuti-
cle, can be hard, where it forms rigid plates or tubes, or soft, where it forms
hinges or joints.23
The insect wing is essentially a sandwich of two layers of cuticle. Over
most of the wing, these two layers are extremely thin and joined tightly
together to form the wing membrane. The membrane is supported by rod-
like “veins,” which are also formed from the same two layers of cuticle; the
veins, however, are hollow, with much thicker walls and a blood-filled core.
The membrane is delicate and flexible, so the veins are the main load-­
bearing structures (Fig. 2.4). The veins radiate out from the wing base to
the tip like spokes, and the main lengthwise veins are often connected by
cross-veins.24 The exact arrangement of the veins varies tremendously
from group to group and is so specific that biologists often use the vein
pattern to identify insect families and sometimes even species.
Most insect wings are not simply flat arrays of veins and membranes.
The wings tend to be slightly pleated, so that some veins are on top of low
ridges and others are at the bottom of shallow valleys (Fig.  2.4). These
pleats are usually well developed at the front of the wing and dwindle away
toward the rear. This combination of veins and pleating adds a lot of stiff-
ness and resistance to lengthwise bending at almost no weight penalty.25

T h e m e a n d Va r i a t i o n s   [ 35 ]
 A

epidermis

cuticle

Figure 2.4:
Insect wing structure, showing the strengthening struts or “veins” supporting the deli-
cate membrane. A. Top view. B. Cross section. (Courtesy of S. T.)

Moreover, scientists were surprised to discover that the pleats had little or
no adverse aerodynamic effects.26 Because of the small size of insect wings,
air tends to flow smoothly over the pleats as if they formed a smooth sur-
face from peak to peak.26,27
Another huge difference between insect wings and those of verte-
brate flyers is that insect wings contain no muscle or actively powered
joints­—nothing equivalent to elbows or wrists. Insects move their wings
entirely by muscles acting on the wing base at its joint or articulation
with the thorax.24 Without any muscles or joints out in the wings, in-
sects have no way to flex the wing to reduce its area while flapping. Most
insects can fold their wings over their bodies to keep them out of the
way when not in use, and many insects can fold up their wings like fans
or accordions to stow them between flights, but they generally cannot
use these folding mechanisms to reduce wing area in flight. Moreover,
because insect wings contain so little living tissue, insects cannot nor-
mally shed them when molting, which means only fully grown adults
have functional wings.
Finally, where birds, bats, and pterosaurs have a single pair of wings,
the basic insect pattern is to have two pairs of wings. Some insects, such
as dragonflies and katydids, have more-or-less equal, independent pairs of
front and hindwings. Some insects have reduced hindwings or they have
mechanisms to couple the hindwings to the forewings so they function as
a single pair; many, such as butterflies, bees, wasps, cicadas, and aphids,
have both. Flies have modified the hindwings into sense organs, and

[ 36 ]  On the Wing

beetles have turned the front wings into protective covers, so they each
have a single functional pair of wings. Insects thus have more flexibility in
number of wings than other flying animals.

THE BIG FOUR

Over the course of animal evolution, dozens of species have evolved un-
powered flight or gliding (as we will see in Chapter 4). In all that time,
only four lineages have evolved powered or flapping flight: insects, ptero-
saurs, birds, and bats. Flapping is a much more sophisticated process than
gliding, so evolving flapping seems to have required overcoming a much
higher evolutionary hurdle than gliding. In later chapters, we will cover
each of the Big Four groups in turn. I will describe the circumstances that
scientists think may have favored flight and the features of the animals
involved that may have predisposed them to evolve wings. Some of these
topics remain open questions, but we will see that researchers have used
indirect evidence and clever experiments to narrow down the range of
possibilities.

REFERENCES
1. S. J. Gould (1985) Natural History.
2. T. A. McMahon (1984) Muscles, Reflexes, and Locomotion.
3. D. E. Alexander (2002) Nature’s Flyers: Birds, Insects, and the Biomechanics of
Flight.
4. S. Vogel (2001) Prime Mover: A Natural History of Muscle.
5. K. Schmidt-Nielsen (1990) Animal Physiology: Adaptation and Environment.
6. M. J. Abzug and E. E. Larrabee (1997) Airplane Stability and Control: A History of
the Technologies That Made Aviation Possible.
7. D. R. Warrick and K. P. Dial (1998) Journal of Experimental Biology.
8. J. Maynard Smith (1952) Evolution.
9. H. Tennekes (1996) The Simple Science of Flight: From Insects to Jumbo Jets.
10. J. M. V. Rayner (1988) Current Ornithology.
11. J. J. Videler (2005) Avian Flight.
12. H. I. Fisher (1957) Science.
13. A. S. King and D. Z. King (1979) in Form and Function in Birds.
14. C. J. Pennycuick (1972) Animal Flight.
15. J. D. Altringham (1996) Bats: Biology and Behaviour.
16. S. M. Swartz, M. S. Groves, H. D. Kim, et al. (1996) Journal of Zoology.
17. G. Neuweiler (2000) The Biology of Bats.
18. P. J. Currie (1991) The Flying Dinosaurs.
19. D. M. Unwin and N. N. Bakhurina (1994) Nature.
20. K. Padian (1991) in Biomechanics in Evolution.

T h e m e a n d Va r i a t i o n s   [ 37 ]
 1. K. Padian and J. M. V. Rayner (1993) American Journal of Science.
2
22. S. C. Bennett (2000) Historical Biology.
23. P. J. Gullan and P. S. Cranston (2010) The Insects: An Outline of Entomology.
24. R. F. Chapman (1982) The Insects: Structure and Function.
25. R. J. Wootton (1986) Journal of Experimental Biology.
26. C. J. C. Rees (1975) Nature.
27. B. G. Newman, S. B. Savage, and D. Schoulla (1977) in Scale Effects in Animal
Locomotion.

[ 38 ]  On the Wing

CH A P TER 3

How to Fly?

E agles fly, ducks fly, bumblebees fly, house flies fly. They all fly with
wings, so before we can look at the evolution of flight we first need to
know what wings are and at least a little about how they work. In the pre-
vious chapter, we looked at the wing anatomy of the flying lineages. Now
we will turn to the physical properties of wings in general and see what
properties make wings more or less effective. All wings, from those of
house flies to those of airliners, are subject to the same general aerody-
namic and physical rules. In later chapters, we will see that the physical
properties required by wings are a major constraint on the evolution of
animals’ wings.
Scientists who study flight and engineers who design airplanes treat
wings as if all wings are very complex, specialized structures. Yes, effective
and efficient wings are rather specialized, but any more-or-less flat surface
can act like a wing. Take a human hand, for example. When I am a passenger
in a car with the windows down, I sometimes “fly” my hand out the window.
I hold my hand flat, fingers together, palm down, in the onrushing air. If I
tilt the front of my hand up, my hand is pulled up. If I tilt the front of my
hand down, my hand drops. My hand is acting as a wing; not a very efficient
one but one that generates the same basic categories of force as an eagle’s
wing or a bumblebee’s wing. Tilting the front edge up represents what an
engineer would call “increasing the angle of attack,” which is one of the ways
that lift is increased on any wing. My hand feels more lift and is pulled up.

PHYSICS OF WINGS

Lift is one of the two important forces on a wing. In aerodynamics, lift is

a force at right angles (perpendicular) to the direction of motion. So a
wing moving horizontally would experience lift directed vertically upward.
(If a wing moves at some angle other than horizontal, the lift will be tilted
away from the vertical, which we will see is critical to the function of flap-
ping and thrust production.) The other important force is the intuitively
named “drag.” Drag is any force that tends to slow the object as it moves
through the air; it is a parallel force in the opposite direction from the
motion. In this chapter, we will see that drag can be considered the “cost” of
producing lift.

Producing Lift

Perhaps you have read or been taught that wings are curved on top and flat
on the bottom, forcing air to move faster over the top; when that happens,
Bernoulli’s equation tells us that the faster air causes low pressure on top.
This low pressure zone produces an upward force that we call lift. This tra-
ditional explanation is partly true, but it is incomplete and rather mis-
leading. If it were true, wings would not work upside down (which would
make most air shows rather boring), and several types of perfectly func-
tional wings that are not flat on the bottom would not work. Yes, most
wings are more convex on top than on the bottom, but this shape is not
essential for lift production.
Rather than going into Bernoulli’s equation or other technical aspects
of aerodynamics, we will take a more intuitive approach. Imagine a flat
plate such as, say, a sheet of plywood. If I lay the plywood sheet on a small
cart so that it is parallel to the ground and I move it horizontally through
the air, nothing much happens. But if I prop up the plywood so the front or
“leading” edge is raised up a bit and then push it through the air, the ply-
wood feels an upward force that we call “lift” (Fig. 3.1). The strength of the
lift depends on how much I prop up the plywood: the lift gets stronger the
higher I tilt the leading edge, at least up to about 15 degrees. If I raise the
leading edge beyond that, lift suddenly gets much weaker. Finally (perhaps
this is obvious—or perhaps not), the force on the plywood is the same
whether I push the cart through the air or blow air over a stationary cart.
When we try to visualize how wings operate, imagining the wing standing
still with air blowing over it is often easier than imagining the air flowing
around the wing as it moves through the air. Just as in sailing, the “relative
wind” is the wind the plywood feels as it moves through the air, and in
flapping, the wing’s orientation to the relative wind can be crucial.
When I tilt the leading edge of the plywood sheet up, the plywood de-
flects air downward. Newton’s third law says that for every action there is

[ 40 ]  On the Wing

 A
No lift
Direction of motion

Lift
B

α
Direction of motion

Figure 3.1:
A. A flat surface moving through the air without any inclination does not experience
lift. B. If the same surface is tilted up by some angle α (the “angle of attack”) and moved
through the air, it will now experience an upward force, lift. (The angle of attack has been
exaggerated slightly for clarity.) (Courtesy of Kevin Alexander, used by permission.)

an equal and opposite reaction, so the air the plywood sheet deflects down
exerts an upward force on the sheet. This upward force is called “lift,” and
it is at right angles (perpendicular) to the movement of the plywood. Intu-
itively, we might think that the air is deflected downward by hitting the
bottom of the plywood sheet and bouncing downward, like billiard balls
hitting the side cushion of a billiard table; but this is not what happens. In
fact, for any surface to act like a wing, the air flowing off the top and
bottom surface must flow smoothly off the back or “trailing” edge (and for
this to work properly, the flow over the top matters more than the air “hit-
ting” the bottom). If the orientation of the trailing edge causes the air to
flow downward as well as backward off the trailing edge, that downward
component leads to lift production.
Remember, I said earlier that the lift increases as I tilt up the plywood
(as I increase the “angle of attack,” Fig. 3.1), but only up to about 15 de-
grees. If I increase the plywood’s angle above some critical angle, the air
stops flowing smoothly over the upper surface and peels away. Rather
than following the upper surface, the air flows straight back, leaving a
large turbulent region right above the upper surface (Fig. 3.2). This phe-
nomenon is rather confusingly called “stall” even though it has nothing to
do with engines. A stalled wing feels a sharp loss of lift and a simultaneous
increase in drag (a slowing force); the turbulent air on top of the surface
produces a broad, turbulent wake that actually causes a slight suction
pulling back.

H o w t o F ly ?   [ 41 ]
 Relative wind Direction of motion
Lift, no stall

Direction
of motion

Relative wind
Stall

Figure 3.2:
At low angles of attack, the air flows smoothly over the surface (top), but if the angle of
attack becomes too large, air flow over the top is disrupted, and lift drops off sharply
(bottom).

The actual stall angle for any wing—plywood plate, bumblebee wing,
jet airliner wing—depends on many factors, including size, speed, and
shape. Larger, faster wings tend to stall at lower angles. The airliner’s wing
probably stalls at about 13 degrees, and wings of large birds like eagles or
swans might stall in the 15-degree to 20-degree range. Small, slow wings,
in contrast, stall at much higher angles. A bumblebee wing may reach 35
degrees or 40 degrees before stalling, and a fruit flight may not stall till its
angle of attack exceeds 50 degrees.1,2
Although our plywood sheet acts like a wing in many ways, it is not
very efficient: it does not produce nearly as much lift as—and it produces
more drag than—a well-designed wing of a similar size. So what needs to
change to allow a wing to produce more lift and less drag? One useful
feature is streamlining. Streamlining greatly reduces drag (except at
very small size scales). A streamlined shape looks something like a
stretched-out teardrop: blunt and rounded in front, widest about one-
third of the way along its length (Fig. 3.3A). If we could slice an airplane
or bird wing from front to back and look at the cut edge, we would see
that this cross section, the “airfoil” shape, is clearly streamlined (Fig. 3.3).
Depending on size and speed, streamlining can cut the drag by up to a
factor of 10!2
Some wings have a symmetrical, stretched-teardrop cross section
(Fig. 3.3A), but most have airfoils with camber. A cambered airfoil is a bit
more convex on top and a bit less convex (Fig. 3.3B) or even concave
(Fig. 3.3C) on the bottom. A wing with no camber, such as the symmetrical

[ 42 ]  On the Wing

 A

Streamlined, symmetrical

Typical airplane

C
Typical bird

Figure 3.3:
Cross sectional shapes of three wings to show streamlining, with front or leading edge
to the left.

airfoil of Figure 3.3A, does not produce any lift when it is directly aligned
with the oncoming airflow (in other words, when the angle of attack equals
zero); it must be tilted up to some angle above zero to generate lift. A cam-
bered wing, however, produces lift even when directly aligned with the
airflow—at an angle of attack of zero—and tends to produce more lift at
any given angle than a wing with no camber. Although a few specialized
airplanes have uncambered wings, most airplanes and all animals other
than very small insects have wings with some camber.3
If a little camber is good, is a lot of camber better? Not necessarily. Im-
agine grabbing the leading and trailing edges of a wing with a symmetrical
airfoil and bending them down (so the middle is forced up). That adds
camber to the wing, and a typical wing might have “10%” camber, that is,
the middle might be bent up above the no-camber shape by about 10% of
the distance from leading to trailing edge. Too much camber increases
drag and may even cause the wing to stall at low angles of attack. Fifteen
percent camber is a lot for a wing (too much for many); a little camber goes
a long way.

Adjusting the Amount of Lift

Several things affect how much lift and drag a wing produces. Increasing
angle of attack and camber (up to a point) increases lift, albeit at the cost
of increasing drag. Increasing speed also increases lift. In fact, lift is pro-
portional to the square of the speed, which means that doubling the speed

H o w t o F ly ?   [ 43 ]
would cause a fourfold increase in lift. So one way for a given wing to lift a
heavier load is to fly faster, and a modest increase in speed can produce a
healthy increase in lift.
One final wing property that affects lift production is wing area. This
makes sense: a larger wing should provide more lift, and it does. Whereas
the area of an airplane wing is essentially constant, birds are masters of
adjusting wing area to change the amount of lift for a given situation. For
instance, to descend rapidly, a bird can partly or completely fold its wings
and drop sharply. Birds’ ability to change wing area, as well as speed and
angle of attack, gives birds a degree of maneuverability far beyond any-
thing possible for airplanes.

Drag and the Lift-to-Drag Ratio

The “cost” of producing lift on a wing is an increase in drag, the slowing

force. Even when a wing produces little or no lift—say, an uncambered
wing moving at an angle of attack of zero—it still experiences drag, just
as I experience drag when riding my bicycle fast. Lift production, how-
ever, adds on extra drag, and as lift increases, this extra, so-called in-
duced drag also increases.2 Induced drag thus represents the main cost of
generating lift. The amazing thing about wings is that even with induced
drag, wings can produce a lot more lift than drag. Under typical flight
conditions, a run-of-the-mill wing might produce 10 times more lift than
drag. To put this in concrete terms, if I push on this wing with 20 pounds
of force to overcome its drag, it rewards me with 200 pounds of lift. In
fact, the “lift-to-drag ratio,” or what scientists call “L/D,” is more impor-
tant than the amount of lift itself produced by a wing.1 A wing may pro-
duce huge amounts of lift, but if that wing requires overcoming huge
amounts of drag, then it is not a very efficient or effective wing. The lift-
to-drag ratio is thus a key measure of the effectiveness of a wing.* For
most purposes, a high lift-to-drag ratio is a Good Thing. What kinds of
wings have high lift-to-drag ratios? Induced drag is generated by aerody-
namic processes that converge at the wingtip, so a shape that somehow
minimizes the amount of wingtip should reduce induced drag. In fact,
long, narrow wings tend to have higher lift-to-drag ratios than shorter,
broader wings. All else being equal, a long, skinny wing will have a lower

*  The L/D is actually not constant on a wing; it changes with angle of attack. What
scientists and engineers actually use to characterize a particular wing is the wing’s
maximum L/D.

[ 44 ]  On the Wing

induced drag—and so a higher lift-to-drag ratio—than a short, broad
wing of the same area. Scientists use “aspect ratio” to indicate the narrow-
ness of a wing, where the aspect ratio is the wingspan—distance from tip
to tip—divided by the average wing chord, or distance from front to back
of the wing. Long, narrow wings are called high aspect ratio wings. They
effectively have less tip for the amount of wing area than stubby, low
aspect ratio wings, so high aspect ratio wings have less induced drag.4 Of
course, any wing—bird, bug, or airplane—is the result of design trade-
offs. When other factors—structural strength, say, or ease of folding—
override sheer aerodynamic performance, a low aspect ratio wing might
be the best overall choice. Scientists can infer a surprising amount about
the biology of extinct flyers like pterosaurs from the aspect ratios of their
wings (Chapter 8).

Scale Ef fects

Humans are near the large end of the animal-size scale; yet even at a
sprint we are not very fast compared to a running deer or a flying robin,
so we do not experience noticeable air drag. In contrast, drag—and
­particularly the way it changes with size—can be of paramount impor-
tance for small flying animals like insects. As animals get smaller and
move slower, the air feels more viscous. The air is no different for me or
a mosquito, but I am so big that pressure-related processes swamp
­v iscosity-related processes, and I hardly notice air’s viscosity. For the
mosquito, in contrast, forces that depend on pressures decline in im-
portance so the viscosity becomes quite apparent.* (See Box 3.1. The
“Reynolds Number.”) This effect of size also affects wing performance.
The main consequence is that for tiny flyers, streamlining is less effec-
tive and wings have much lower lift-to-drag ratios. In a nutshell, at
small scales, wings are simply less effective. For example, a turkey vul-
ture might have a lift-to-drag ratio of 15:1 (on the high side for birds); a
dragonfly’s lift-to-drag ratio probably would not exceed 5:1, which,
even though quite high for an insect, barely makes it into the low end of
the bird range. 5

*  Viscosity is actually a form of friction and should not be confused with density:
honey is hundreds of times more viscous than water in spite of being just slightly
denser whereas mercury’s viscosity is not too different from the viscosity of water,
but it is about 14 times denser.

H o w t o F ly ?   [ 45 ]
 Box 3.1:  THE “REYNOLDS NUMBER”

A person can’t study animal aerodynamics or hydrodynamics very long

without running into the concept of the “Reynolds number.” The Reyn-
olds number, given the symbol Re, is a dimensionless index that
­indicates whether pressure drag or viscous drag is dominant. Viscous
drag comes from viscosity. Viscosity is the resistance of air (or any gas
or liquid) to flowing past itself or a solid surface, a sort of three-­
dimensional friction. Pressure drag is dominant at high Reynolds
numbers, and viscous drag is dominant at low Reynolds numbers.
Streamlining can dramatically reduce drag at high Reynolds numbers,
but at low Reynolds numbers, reducing surface area (being spherical)
cuts drag more effectively. The Reynolds number equals air density
times speed times body (or wing) length divided by viscosity; as size
and speed (or both) decrease, the Reynolds number decreases, and vice
versa. For some technical uses, the exact value of the Reynolds number
matters, but most of the time the general order of magnitude (powers
of 10) of the Reynolds number is all that counts. Insect wings have
Reynolds numbers of tens to thousands; bird wings, tens of thousands
to hundreds of thousands; and airplanes, millions to hundreds of mil-
lions. Airplane designers can get away with ignoring viscous drag, but
animal aerodynamicists usually cannot.
Osborne Reynolds, the man who gave the Reynolds number its
name, was not actually studying drag directly when he developed the
original concept. He was looking at water flows in pipes. He discovered
that with slow flows or narrow pipes, water moved smoothly through
the pipes, but with fast flows and larger pipes, the flow became turbu-
lent and lost its smooth, laminar character. Reynolds proposed what
became the Reynolds number as a way to predict whether a given flow
would be laminar (smooth) or turbulent (full of random eddies). These
flows turned out to correspond to viscous-dominated versus pressure-
dominated conditions, respectively, and so the Reynolds number ap-
plies to any kinds of flows, not just in pipes.
Nowadays, the Reynolds number is most often used as an aerody-
namic scale factor. For example, two wings with the same Reynolds
number will have the same air flow patterns and the same ratios of forces.
Conversely, two geometrically similar wings with very different Reyn-
olds numbers will experience very different airflows, and force ratios
such as the lift-to-drag ratio will be very different. At low Reynolds num-
bers, the flow patterns that produce lift are weaker, so lift and lift-to-
drag ratios are lower at very low Reynolds numbers. In contrast, wings
tend to be more effective and efficient at high Reynolds numbers.

[ 46 ]  On the Wing

GLIDING

Gliding is unpowered flight. When a heron or a dragonfly drifts along

through the air without flapping its wings, it is gliding. Gliding sometimes
gets short shrift in animal flight discussions, but gliding was probably a
stage in the evolution of flight for some animals, so we need to know some-
thing about gliding to look at the evolution of powered flight. Indeed,
whether birds went through a gliding stage or not has been a source of much
argument for decades, and at least some of the controversy may be related
to imperfect understanding of flight mechanics, including gliding.

Gliding Basics

Gliding, to continue my earlier analogy, shares many similarities with

coasting downhill on a bicycle. A bicycle can coast downhill continuously,
but if the road levels out or turns uphill, the bike slows to a stop unless the
rider begins to pedal. Gliding works the same way: as long as a glider de-
scends through the air, it can keep flying; but if the glider levels out or
tries to climb, it will slow down and eventually stall. Perhaps, rather than
“unpowered,” we should think of gliders as “powered by gravity.” Just as
gravity pulls a coasting bicycle down a ramp, gravity pulls a glider down as
well. One difference between a glider and a coasting bike, however, is that
the characteristics of the glider’s wings determine how steep a ramp it
glides down. Curiously, it turns out that a gliding wing’s lift-to-drag ratio
sets its glide angle. A bit of fiddling with the geometry shows that the lift-
to-drag ratio must be the same as the glide ratio, or the distance the glider
moves forward for every meter it descends.5 So a glider with a lift-to-drag
ratio of 10:1 also has a glide ratio of 10:1, meaning it moves forward 10
meters (33 feet) for every 1 meter (3.3 feet) it descends.*
One peculiar and surprising consequence of the glider’s lift-to-drag
ratio setting its glide angle is the fact that changing a glider’s weight
does not have much effect on its glide path. Changing a glider’s weight
(without changing anything else) changes its speed but not its glide angle.
If I don’t change the glider’s geometry in any way, I can add lead weights
and it will fly faster but land just as far away as before I added weights. In
other words, it will glide just as far, but it will get there faster! This effect

*  The mathematically inclined reader will no doubt see that the glider’s angle of
descent or “glide angle” can easily be calculated using trigonometry; for a glide ratio
of 10:1, the cotangent of the glide angle is 10, giving a glide angle of 5.7 degrees.

H o w t o F ly ?   [ 47 ]
actually has more general consequences: as a rule of thumb, heavier
flyers (whether gliding or powered) tend to fly faster than similar but
lighter flyers.
If a glider’s goal is to glide long distances, then a high lift-to-drag ratio
is clearly beneficial because it produces a shallow glide angle. Both a wing’s
shape and its size play a big role here. High aspect ratio wings have high
lift-to-drag ratios, but even high aspect ratio wings have diminishing lift-
to-drag ratios as they get smaller. Few insects glide, and the ones that
do—dragonflies and damselflies, for example—are large insects and have
wings with high aspect ratios. A number of animals have evolved to glide
with wings of low aspect ratio, such as flying squirrels. These are all much
larger than insects, however, and their stubby wings come nowhere near
the effectiveness of bird or bat wings (a colleague calls them “glide-assisted
jumpers”). In short, if you are big, then even stubby wings may be effective
enough for some limited gliding, but if you are tiny, even long, skinny,
high aspect ratio wings may not be efficient enough for extensive gliding.

Soaring

“Hold on,” you say, “I’ve seen buzzards and hawks gliding on motionless
wings, and they can go up. I thought you said gliders have to go down.”
These birds really are gliding, but it is a special form of gliding called “soar-
ing.” The birds are still going down relative to the air, but they are taking
advantage of rising air. If the air rises faster than the bird descends, the
bird can actually ascend while gliding. Returning to our coasting bicycle
analogy, if, instead of coasting down a ramp, our bicycle coasts down an
inclined conveyer belt, imagine what happens if the conveyer belt runs
uphill. When the conveyer belt runs uphill faster than the bicycle coasts
down, the bike will actually be carried up the belt. This is exactly what a
soaring bird does: it seeks rising air—warm air heated by the ground or
wind blowing up a steep slope—and climbs by riding the rising air upward.

Wingless Gliders?

About a decade ago, then-graduate student Stephen Yanoviak was study-

ing ants by observing them from catwalks built high in the rain forest
canopy in Peru. He made the rather startling discovery that if he knocked
an ant off a branch, the ant somehow steered its fall toward the nearest
tree trunk, often landing on the trunk. These were worker ants with no

[ 48 ]  On the Wing

trace of wings, so their ability to “fall” back to the trunk was astonishing.
By coincidence, one of the faculty members in Yanoviak’s department was
Robert Dudley, an animal flight specialist who just happened to do much
of his research in the tropics. Dudley was amazed by Yanoviak’s story, and
together they designed a series of experiments to see if ants really could
guide their fall back to tree trunks, and if so, how they did it.
In their original experiments, 85% of the ants landed on the trunks of
trees they fell from. Several studies later, Yanoviak and his colleagues
have established that in addition to the original ant species they studied,
a variety of other animals living in rain forest canopies have this ability.
These include the workers of several unrelated species of arboreal ants
from Africa and Central and South America as well as a variety of other
wingless insects and spiders.6–8 Curiously, the first species of ant they
studied steered its falls backward (tail first), and seems mainly to use its
hind legs for steering;9 almost all the species they looked at later steered
reassuringly head first.
Aerodynamically, worker ants are lousy gliders, with a glide angle of
75 degrees from the horizontal, compared with winged gliders, with glide
angles of 10 degrees or 20 degrees. In fact, traditionally, biologists have
called glide angles of greater than 45 degrees “parachuting” and reserved
“gliding” for angles of less than 45 degrees. This is an arbitrary distinction
with no basis in aerodynamics because any angle less than 90 degrees re-
quires some lift production, so we will consider these ants to be gliding,
albeit at a very steep glide angle.
The researchers dubbed this ability “directed aerial descent” to refer to
animals with no apparent aerodynamic specializations that nevertheless
perform a very crude, inefficient, but useful form of gliding. They suggest
two important implications for their discovery. First, given the variety of
wingless animals that display this ability, the behavior must have signifi-
cant survival benefits. Indeed, small animals adapted to life in the tops of
rain forest canopy trees are poorly adapted to life on the ground and tend
to be easy prey for terrestrial predators. Moreover, worker ants generally
do not survive unless they can return to their own colony’s nest, so land-
ing back on the trunk of their original tree is at a premium.
Second, if wingless animals with no noticeable aerodynamic adapta-
tions can consistently and reliably adjust the direction of their descent
during a fall, then they already possess most of the sensory and neural
mechanisms they would need to control a more efficient glide. No one
would think of a cat as a glider, yet cats reliably land on their feet after
even short falls. That kind of orienting-during-a-fall behavior is directed
aerial descent at its most rudimentary, and many arboreal animals have

H o w t o F ly ?   [ 49 ]
taken it a step further and evolved the ability to steer during falls. This
ability could be a key exaptation* in the evolution of gliding. If many arbo-
real animals have this ability, as now seems likely, and if some of those
animals experienced selection pressure to extend falls into glides, they
would have a head start in evolving more effective gliding. Ironically, bi-
ologists have long considered the evolution of flight control ability to be
one of the major hurdles to be overcome during the evolution of flight,10
but this “hurdle” may already be behind many arboreal animals.

APPLYING POWER

Flyers need to fly under power to take full advantage of flight and to over-
come the limitations of gliding. Birds in powered flight flap their wings; if
they are not flapping their wings, they are not powered and are gliding. Air-
plane wings may flex a bit in heavy turbulence, but airplane wings certainly
do not flap the way birds flap their wings (at least, not more than once!). So
airplanes can fly perfectly well without flapping their wings. Why, then, do
flying animals need to flap their wings in order to fly under power?

The Function of Flapping

Flapping is all about thrust production. “Thrust” is the force that moves a
flyer forward by overcoming the force of drag on the wings and body. An
airplane uses its wings to produce lift but it has separate devices—­engines,
jet, or propeller—to generate thrust. Birds do not have separate engines,
so they have to use their wings to produce both the upward force and the
forward force. Simply moving a wing through the air causes it to produce
lift (that’s how gliding works), so logically, flapping must be entirely for
producing thrust.
In powered flight, the wings of flying animals do not work all that much
like airplane wings. A bird’s or an insect’s wings actually operate much
more like a helicopter’s rotor blades. Rotor blades are long, narrow, highly
specialized wings, and the helicopter’s rotor generates both lift and thrust,
much like animal wings. Moreover (and unlike airplanes), helicopters can

*  Recall from Chapter 2 that an exaptation is something evolved for one function
that can also perform or be co-opted for a new, second function. Biologists originally
called these “preadaptations,” but most have abandoned that term due to its teleo-
logical overtones.

[ 50 ]  On the Wing

use their rotors to hover, just as many flying animals can use their wings to
hover. Helicopter rotor blades are not an exact model for animal wings,
however: rotor blades rotate continuously, whereas a flapping bird wing
cannot rotate continuously in one direction, so it periodically reverses di-
rection in a cycle of upstrokes and downstrokes. Flapping wings operate
differently during the upstroke and the downstroke parts of the cycle.
Actually, flying animals do not just move their wings up and down when
flapping. During the downstroke, the wing moves down and forward with
the leading edge tilted slightly down. On the upstroke, the leading edge of
the wing tilts up and the wing moves up and back relative to the animal’s
body (Fig. 3.4). The wingtip thus follows an inclined path, from up and back
at the top of the upstroke to down and forward at the bottom of the down-
stroke (Fig. 3.4). In general, almost all of the lift (including the upward force)
and some or all of the thrust are produced during the downstroke. Larger
flyers—herons, geese—may have a “passive” upstroke where they produce
little or no useful force. Smaller flyers—hummingbirds, honeybees—may

Top of
upstroke

Wing cross
section
(airfoil)
Bottom of downstroke

Mid-downstroke

Mid-upstroke

Figure 3.4:
Wing movements during flapping. A. Wing moves down and forward during the down-
stroke. B. Wing is inclined downward during the downstroke. C. Wing is inclined upward
during the upstroke. (Solid arrows show direction of wing movement relative to body.)

H o w t o F ly ?   [ 51 ]
have an “active” upstroke that produces a significant amount of thrust.5 (See
Figure 3.5 and Box 3.2. Flapping and Forces.)

Power and Speed

Powered flight has a major built-in difference from terrestrial locomotion

in how speed affects power requirements. Whether I am on foot or driving

Box 3.2:  FLAPPING AND FORCES

DOWNSTROKE
As the wing moves down and forward, it is tilted so that the lift is tilted
forward (Fig. 3.5A). The lift, remember, is perpendicular to the direc-
tion of movement of the wing, so it is no longer straight up on a flapping
wing. This forward tilt of the lift is where the thrust comes from. We
can separate the lift into its upward component (the upward force bal-
ancing the animal’s weight) and a forward component, the thrust. (I am
ignoring the wing’s drag for simplicity, but the drag is so much smaller
than the lift that it does not significantly change the general pattern.)
Most of the useful flight forces—the upward force and the thrust—are
produced during the downstroke, so it tends to last noticeably longer
than the upstroke.

UPSTROKE
Figure 3.5B shows the passive upstroke, typical of a large bird. This up-
stroke produces little useful force because the wing is aligned directly
with the air flow. A passive upstroke is mainly a return stroke: it is just
the movement necessary to move the wing into position for the next
downstroke with as little effort as possible. It may or may not produce a
bit of upward force but it does not contribute any thrust.
Active upstrokes vary in how much useful force they generate, but
the extreme is shown in Figure 3.5C. If an animal can tilt its leading
edge straight up and move its wing up and back very rapidly, it can actu-
ally produce a significant amount of thrust. This process is only really
effective if the flyer can move its wings backward faster than its body
flies forward through the air (see wingtip path, Fig. 3.5C). Insects, hum-
mingbirds, and very small bats are probably the only flyers that can take
full advantage of active upstrokes in normal flight—although some
larger flyers may be able to perform such a stroke when flying slowly—
or take advantage of less extreme (and less effective) variations.

[ 52 ]  On the Wing

 upward
direction
lift force
of motion

A thrust α
relative
wind
path of wingtip
relative
lift
wind
B

direction
of motion
relative
wind path of wingtip
α
thrust

C lift

direction
of motion

Figure 3.5:
Forces on a flapping wing. A. Downstroke; lift has upward and forward (thrust) compo-
nents. B. Passive upstroke, used by medium-sized and large birds in fast forward flight;
forces minimized. C. Active upstroke, used by smaller flyers, especially in slow flight; neg-
ative angle of attack generates forward-tilted lift on bottom of wing. Insets: path of the
wingtip through the air.

a car, if I want to go faster, I need to put more power into moving. Going
slower requires less power, and stopping requires essentially no power.
(Yes, a bit of power goes into keeping the engine running in my car or
keeping my body alive even at rest, but this is a drop in the bucket com-
pared to the power needed to move.) In a nutshell, standing still uses no
power, and moving requires more power the faster I go.
Flight has fundamentally different power requirements. At higher
speeds, as I fly faster I need more power, just like running. Strangely, as I
slow down, the power requirement at first drops, but then as I fly slower
and slower, I begin to require more and more power. If I am in an airplane
and I slow down below a certain speed, I actually have to increase the throt-
tle to fly level at slower speeds.* If I am a bird, I have to flap my wings faster
and harder as I slow down below a certain speed, if I want to fly level.

*  Test pilots call this “being on the back side of the power curve,” and it is a very bad
place to be in most airplanes.

H o w t o F ly ?   [ 53 ]
 To us pedestrian creatures, the peculiar relationship between flight
power and speed is quite foreign. Scientists call it a “U-shaped power
curve” because a graph of power against speed is high at both low and
high speeds and low in the middle, thus making a U-shape.11 This rela-
tionship has several practical consequences for flyers. One is that any
flyer has some specific intermediate speed that requires the least effort to
fly. For example, the minimum power speed for a swallow is about 3½
meters per second (just under 8 miles per hour), and for a kestrel (small
falcon) it is about 5 meters per second (11 miles per hour). These mini-
mum power speeds—flight speeds that require the least effort—are as
fast as or faster than the maximum running speeds of similarly sized ter-
restrial animals. Curiously, the speed of minimum power (least effort) is
not the speed that gives best “fuel economy,” that is, that uses the least
food energy per distance traveled. The best fuel economy speed is the
speed that allows the flyer to fly the greatest distance, and the speed for
best fuel economy is a bit higher than the speed of minimum power. Al-
though this sounds contradictory, at speeds a bit higher than the mini-
mum power speed, the reduction in travel time more than compensates
for the increased energy consumption. Scientists have observed birds in
the middle of long-­d istance migrations on radar, and these migrants
mostly fly at speeds quite close to their estimated best-economy speeds,
even though the flyers would feel that they are exerting more effort than
flying at the speed of minimum power.
If flying slower and slower requires more power, what about coming to
a complete stop in flight? In other words, what are the power requirements
of hovering? As you probably guessed, hovering requires a lot of power.
Hovering may be the most power-hungry activity in the animal kingdom;
it is certainly the most power-hungry activity in animal (and mechanical)
locomotion. The main reason that hovering requires so much power is that
the wings get no air moving over them due to the body’s movement
through the air; the animal’s body is not moving. Instead, the animal
must use its wing muscles to move the wings fast enough for them to pro-
duce enough lift. In other words, in forward flight, a bird needs only
enough power to overcome its drag so the wings can move through the air,
but when hovering, the bird needs enough power to move the wings very
rapidly to make up for the lack of air flow from forward flight.
Hovering requires some rather contradictory structural properties.
Think about how an animal’s weight is suspended in flight: the animal’s
body literally hangs from its wing joints. A flying bird’s weight is carried
entirely by its shoulder joints, so these joints must be strong, robust struc-
tures. A loose, highly flexible joint may not stand up to the loads of flight

[ 54 ]  On the Wing

because its flexibility limits how rigidly it can be supported. Yet hovering
requires a highly flexible wing joint so that the wings can make the neces-
sary, exaggerated flapping movements—hovering normally requires both
a larger wing stroke and a lot more wing twisting than forward flight. So
in order to hover, the hoverer must have a strong enough wing joint to
carry its whole body weight, but the joint must also be flexible enough to
allow the greater degree of movement that hovering requires. As we will
see shortly, both power and structural constraints make hovering easier if
you are small. Insects and hummingbirds are masters of hovering and can
do so for long periods. Sparrows and finches can hover for a few seconds,
but geese and eagles are simply too large to hover.
One subtle consequence of the speed-power relationship of flying is that
the minimum-power flight speed is so high that it can make flight very eco-
nomical, much more so than walking or running. Even though flight re-
quires lots of power, flyers travel so much faster that they actually use less
total energy than a runner. A sparrow may use 5 times more power to fly
than a mouse needs to trot, but if the sparrow flies 10 times as fast as the
mouse trots, the sparrow will consume half as much energy (fuel) as the
mouse to go the same distance. In other words, a flyer typically consumes
much less energy to go a given distance than a similarly sized runner.

GLIDING VERSUS FL APPING: FALSE DICHOTOMY?

So far I have described flapping flight and gliding flight as two completely
separate activities. This division seems reasonable given the way I have
defined the two modes of flight; it has even been used as an argument
against gliding as a possible precursor of flapping. Some scientists have
argued that weak or poorly developed flapping would be so ineffective that
it would provide no benefit to gliding animals (or might even be less effec-
tive than gliding), so gliding could not have led directly to flapping.12, 13
In fact, both theoretical modeling and experiments using flapping
robots show that even low-amplitude, weak flapping can produce enough
thrust to be useful, even when such flapping is too weak to maintain level
flight.14–16 These results mean that we must think of gliding versus fully
powered flight—flapping flight as used by living birds, bats, and insects—
as two extremes on a continuum. Between these extremes, animals could
use flapping with a range of effectiveness, from weak flapping to slightly
extend a glide to stronger flapping that might increase glide distance by
five- or tenfold. Although living animals appear to use either pure gliding
or fully powered flapping flight, the evidence suggests that at least some

H o w t o F ly ?   [ 55 ]
(possibly all) of the living flapping flyers went through a partially powered
stage. For clarity, from this point on, when I use the terms “flapping flight”
or “powered flight,” I am referring to the fully powered extreme of the
continuum, and I will use “weak flapping” or “partially powered flight” to
refer to intermediate behavior between gliding and fully powered flight.

SIZE MAT TERS

Just as size affects the hovering ability of flyers, size affects other facets of
flight in important ways. Size will figure prominently as we look at the
evolution of flight. Nature’s flyers range from insects with wingspans of
little more than a millimeter (1⁄10 inch or so) to the largest animals that
have ever flown, such as pterosaurs and the vulture-like teratorns, with
wingspans in the 8- to 12-meter (25- to 40-foot) range. The differences
between being really small and being really big matter a lot more for flyers
than for walkers. The physics of a cockroach’s legs are not all that different
from the physics of a horse’s legs. Flying is different. Although their wings
work basically the same way, small flyers face a rather different suite of
constraints and opportunities than do large flyers. Bees can do things that
buzzards can’t, and vice versa.2
Hovering, for example, strongly favors small flyers for at least a couple
of reasons. For example, the surface-to-volume ratio greatly favors small
animals, as we saw in Chapter 1. As animals get smaller, their weight goes
down much faster than their surface area. For animals of the same general
shape, cutting the body length in half will reduce the wing area to one-
fourth but the weight to one-eighth of the original values. Smaller flyers
are effectively supporting lighter bodies with larger, lighter wings, so they
don’t have to move those wings nearly as fast in order to hover. Insects
take this a step further: with their unusual, corrugated structure, insect
wings are remarkably light even for their small size while still being ex-
ceptionally stiff and strong. In addition to overall geometry, changes in
muscle cross sectional area (Chapter 1) and muscle mass with changing
body size may play a role in hovering ability as well. In concrete terms,
insects and hummingbirds can hover easily, robins and pigeons may be
able to hover briefly, but crows probably cannot hover at all.
Large animals suffer a double whammy with hovering, because they are
also at the wrong end of the structural scale. Big wings are disproportion-
ately heavy: double their length and their surface area will increase by
four times (good) but their weight will increase by eight times (very bad).
(See Box 3.3. Scaling of Size.) Heavy wings have two drawbacks for

[ 56 ]  On the Wing

 Box 3.3:  SCALING OF BODY SIZE

If animals retain the same basic shape but change in size, different
measures of size change at different rates. To take the simplest shape,
basic geometry says that if I double the diameter of a sphere, the
sphere’s surface area increases by a factor of 2-squared or 4 and its
volume increases by a factor of 2-cubed or 8. The same would apply to
two animals of the same shape if one were twice as long as the other. An
animal’s weight is normally directly proportional to its volume, so if an
animal’s length doubles, its weight goes up by eight times. If one animal
were three times longer than another of the same shape, the big animal
would weigh 3-cubed or 27 times more than the small one.
Flying animals span a stunning size range, from tiny insects like
thrips and parasitic wasps with wingspans of less than 1 millimeter
(1⁄25 inch) to giant pterosaurs with wingspans of over 11 meters (more
than 35 feet). This is a 10,000-fold range on the length scale; if thrips
and pterosaurs were the same shape, they would differ in weight by a
factor of about one trillion-fold. Because pterosaurs and thrips are not
the same shape (and also because wingspan, as opposed to body length,
may exaggerate linear dimensions) pterosaurs are “only” about a billion
(1,000,000,000) times heavier than thrips. So we cannot predict ex-
actly how much disproportionately heavier flyers get with increasing
size; clearly, both surface area and weight increase much faster than
linear dimensions as animals get bigger.
Because of the way weight increases as flying animals become larger,
big flyers face conflicting priorities: they need to be as light as possible
to make flight easier, but they need disproportionately stronger wing
structures to carry their disproportionately heavier body weight. The
structural requirement leads to wings becoming even more dispropor-
tionately heavy. Very heavy flyers thus tend to have “overbuilt,” heavy
wings compared to their smaller relatives; by the same token, very
small flyers tend to have amazingly simple, lightweight wing struc-
tures. This pattern shows up even within insects: a large wasp like a
hornet will have four or five main supporting wing veins (providing the
main structural support for the wing), whereas a tiny parasitic wasp
will only have one or two wing veins.

hovering. First, they require a stronger, stouter joint, which works against
the flexibility required by hovering. Second, heavy wings have a lot of in-
ertia, so the animal must work harder to move them. Hovering requires
faster and larger wing movements than forward flight, so heavy wings
hinder hovering. Hovering is clearly the province of small flyers.

H o w t o F ly ?   [ 57 ]
 Gliding, in contrast, is where big flyers come into their own. Because of
the way viscosity affects flyers of different sizes, small wings produce
more drag for a given amount of lift, or to turn it around, big wings pro-
duce less drag for a given amount of lift (see Box 3.1). All else being equal,
a big wing—say, a wing with a 200-centimeter (79-inch) span like a
stork—will have a higher lift-to-drag ratio than a small wing—say, one
with a 6-centimeter (2⅓-inch) span, such as a dragonfly. Fruit flies have
miserable lift-to-drag ratios of around 2:1, and larger insects probably do
not get much higher than 4:1 or 5:1. (Remember, a lift-to-drag ratio of 2:1
means that if the fruit fly tried to glide, it would move forward only
2 meters for every 1 meter it descends.) Birds range from about 4:1 on the
low end for small, stubby-winged birds like sparrows, to around 10:1 for
small birds with high aspect ratio wings like swallows, and large birds
with moderate aspect ratios like herons and hawks. Big birds specialized
for gliding run even higher: turkey vultures (buzzards), for instance, at
about 15:1, and the largest albatross species at 19:1.5
The most effective soaring animals must be good gliders, so the real
masters of soaring, the ones that depend on soaring as much as or more
than on flapping, are all big. Vultures, hawks, and eagles are probably the
most familiar of these specialized soarers, which also include the biggest
modern flying birds—condors (heaviest) and albatrosses (longest wings).
These very large birds must get by with proportionately less powerful
muscles—they are barely powerful enough for forward flapping flight—
so they soar and avoid flapping as much as possible.
Knowing how size affects the physiology and aerodynamics of modern
flyers allows us to understand the flight of ancient animals. A 300-million-
year-old mayfly with a 3-centimeter (1-inch or so) wingspan could proba-
bly hover but certainly did not soar routinely (Chapter 5). In contrast, the
70-million-year-old pterosaur Pteranodon, with a 7-meter (23-foot) wing-
span could not have hovered but probably spent most of its flight time
soaring (Chapter 8). This “biomechanics” approach has, in recent decades,
greatly expanded our understanding of how extinct animals lived, and has
helped us understand what behaviors would have been probable, possible,
or impossible for these ancient animals.

REFERENCES
1. J. J. Bertin and M. L. Smith (1979) Aerodynamics for Engineers.
2. S. Vogel (1994) Life in Moving Fluids: The Physical Biology of Flow.
3. D. E. Alexander (2009) Why Don’t Jumbo Jets Flap Their Wings? Flying Animals,
Flying Machines, and How They Are Different.

[ 58 ]  On the Wing

 4. J. D. Anderson (2007) Fundamentals of Aerodynamics.
5. D. E. Alexander (2002) Nature’s Flyers: Birds, Insects, and the Biomechanics of
Flight.
6. S. P. Yanoviak and R. Dudley (2006) Journal of Experimental Biology.
7. S. P. Yanoviak, M. Kaspari, and R. Dudley (2009) Biology Letters.
8. S. P. Yanoviak, Y. Munk, and R. Dudley (2011) Integrative and Comparative
Biology.
9. S. P. Yanoviak, R. Dudley, and M. Kaspari (2005) Nature.
10. J. Maynard Smith (1952) Evolution.
11. J. M. V. Rayner (1999) Journal of Experimental Biology.
12. G. R. Caple, R. T. Balda, and W. R. Willis (1983) American Naturalist.
13. P. Burgers and L. M. Chiappe (1999) Nature.
14. U. M. Norberg (1985) American Naturalist.
15. R. L. Nudds and G. J. Dyke (2009) Evolution.
16. K. Peterson, P. Birkmeyer, R. Dudley, et al. (2011) Bioinspiration & Biomimetics.

H o w t o F ly ?   [ 59 ]
CH A P TER  4

Gliding Animals
Flight without Power

A small, mottled-brown lizard scurries up the trunk and out on a limb

 of a very tall tree in a rain forest in Malaysia. As the lizard noses
around on the limb searching for insects to eat, it is startled by a bird land-
ing nearby. The lizard leaps off the branch away from the tree, in spite of
being over 30 meters (100 feet) off the ground. As if by magic, a stubby
wing sprouts from each side of the lizard’s body. After a brief dive to gain
airspeed, the lizard flattens its glide and steers toward another nearby
tree. Having picked out a likely landing spot on a large limb, the lizard
rears up into a short, steep climb to slow down just before contact. Judg-
ing speeds and distances with exquisite precision, the lizard alights gently
and scurries off in search of dinner.
This lizard is a specimen of Draco formosus, one of the dozen or so spe-
cies in the genus Draco, commonly called “flying lizards” or “flying drag-
ons.” (The genus name, “Draco,” is Latin for “dragon.”) All the lizards in
this genus have wings formed by elastic skin stretched over a set of
springy, elastic, retractable ribs along the sides of their bodies (Fig. 4.1).
When folded up, the wing is barely noticeable and appears not to interfere
with the lizard’s running or climbing. When extended, however, these
wings generate enough lift to give the lizard about a 2:1 glide ratio—it
glides forward two feet for every one foot it descends in a steady-state
glide. Moreover, it can flatten its glide considerably by gradually increas-
ing its angle of attack and decelerating.
These lizards are gliders. They cannot flap their wings so they cannot
sustain level flight. To say that these animals are not “true” flyers, how-
ever, is to seriously underestimate their abilities. A purely gliding animal
Figure 4.1:
Draco gliding, looking down on it from above. (Courtesy of S. T.)

needs to be able to do most of what a bird can do: steer, recover from gusts,
avoid obstacles, detect safe landing spots, and perform soft landings.
Unlike a bird, a glider’s flights are always brief and always finish lower
than they began.
A surprising diversity of animals (including even fish!) have evolved
the ability to glide, and the extinct gliding kuehneosaurs, which looked
much like modern Draco, were some of the earliest vertebrates to take to
the skies. Moreover, gliding was most likely a step on the way to powered
flight for at least some of the lineages of flapping flyers.

GLIDERS VERSUS FL APPERS

Gliding animals such as Draco have several features in common with flap-
ping flyers like birds. Most obviously, both have wings. The wings of glid-
ers and flappers are very different in structure, but both have large,
flattened, cambered surfaces that produce lift using the same mechanism.
Although not as mobile and flexible as a bird wing, the wings of almost all
gliders are adjustable enough to use for steering, just the way a gliding
bird or bat steers.
A more subtle similarity involves the “control system.” Gliders need to
be able to judge whether they can safely glide to a particular target, which
requires both sharp vision and sophisticated mental processing. Once
aloft, a glider needs to use that sharp vision along with quick reflexes to
avoid objects and stay on course. The glider needs the brainpower to accu-
rately control the glide and to determine just the right time for the pull-up
to produce a soft landing. These are all basically the same as the sensory
and mental tasks of a flapping flyer like a robin or a house fly, with the
added necessity for the glider to always get it right the first time: if the
robin arrives too low or too slow, it can always flap its wings to compensate,

Gliding A nim a l s  [ 61 ]

but Draco does not have that option. The glider thus has an even greater
need for high precision and accuracy.
The fundamental difference between gliding animals and flapping
flyers is that pure gliders, when flying, must always descend. In physical
terms, gliders are propelled by gravity; they can’t flap, so they must always
have a downward component to their flight—gliders cannot continuously
fly level or climb. Just as a coasting bicycle will eventually slow to a stop
on a level road, a glider attempting to maintain level flight will decelerate
and eventually stall. When a glider adjusts its flight so that all the forces
(lift, weight, drag, thrust) are balanced, it will be in an equilibrium or non-
accelerating glide. Typically, gliding animals in equilibrium glides have
fairly steep glide angles, descending at approximately 30 degrees or more
below the horizontal.1 Gliding animals actually spend little of their time
in equilibrium glides, however.
When Draco jumps off a tree limb, at first it falls steeply to accelerate to
a high enough air speed for its wings to become effective. Scientists call
this the “ballistic phase” of the flight. Once the lizard is going fast enough,
it levels out, and then it begins to gradually increase its angle of attack and
decrease its speed to maintain a flat glide, the “aerodynamic phase.” In
typical glides, Draco spends most or all of the aerodynamic phase deceler-
ating, giving it a much flatter glide than when in an equilibrium glide
(Fig. 4.2). When executed properly, the glider stalls just as it reaches its
intended landing spot. For very long glides, or glides without a well-­
defined landing site, Draco may enter an equilibrium glide and just pull up
into a brief, decelerating climb to slow down before alighting.
Many birds use rising air currents to stay aloft while gliding, staying up
for hours and covering many miles in the process. This is called soaring
(Chapter  3), and in principle, gliding animals could soar. Most gliders,
however, have wings that appear almost square or circular from above,
meaning they have very low aspect ratios. These wings are aerodynami-
cally inefficient (they have low lift-to-drag ratio values), so gliding animals
would need very high-speed updrafts—much higher than they would nor-
mally encounter and much higher than the updraft speeds used by soar-
ing birds.
The difference between the aspect ratios (AR) of flapping flyers—such
as birds and bats—and gliding animals—such as Draco or flying squirrels­—
are quite striking. Birds typically have aspect ratios of 6 or 8 (their wings
are 6 or 8 times wider from tip to tip than from front to back). For in-
stance, robins have aspect ratios of about 6 and mallard ducks have aspect
ratios of about 8. Even birds with fairly short, broad wings, like sparrows
(AR = 5.3), have considerably higher aspect ratios than gliding animals.2

[ 62 ]  On the Wing

Figure 4.2:
Gliding trajectories. Width of the path arrow represents speed. (Courtesy of S. T.)

The aspect ratios of the wings of several species of Draco range from 1.7 to
2.3,1 and flying squirrels have aspect ratios in the range of 1.2 to 2.2.3
Thus, pure gliding animals generally have wings with aspect ratios that are
too low for them to soar effectively.
Why do most gliders seem to be limited to wings of low aspect ratio?
Although scientists have debated this question without reaching a clear
consensus, the most reasonable explanation has to do with the body parts
that make up the wing. In Draco, the trunk (torso) ribs have hinged exten-
sions that support the wing. If these rib extensions were much longer,
they would not have room to fold flat along the animal’s flanks when not
in use and would also get entangled with the hind legs. Flying squirrels,
like all other mammalian gliders, form their wings from a stretchy fold of
skin, the patagium, that runs from their front legs along their flanks to
their hind legs. That arrangement limits their wingspan to the lengths of
their outstretched legs. The legs of gliding squirrels are slightly longer
than the legs of non-gliding squirrels, but not dramatically longer. Why
haven’t flying squirrels evolved much longer legs so they could have longer,
narrower wings? Flying squirrels spend much more time climbing around
on tree branches than gliding, and biologists suspect that if their legs
became much longer, the length might interfere with climbing. Similar
arguments apply to other gliders, such as frogs and geckos.
These low aspect ratio wings do have at least one virtue: they can operate
at very high angles of attack without stalling. Functionally, this means that
such a wing should be able to fly slower than a longer, narrower wing with

Gliding A nim a l s  [ 63 ]

the same area. In other words, as a glider like Draco slows down, it can keep
increasing the angle of attack to maintain lift as it flies slower and slower,
which should allow it to land at a lower speed than would be possible with
a high aspect ratio wing. This feature is a distinct advantage for a flyer that
cannot flap its wings in a braking action to slow down for landing.
In Chapter 3, I described wingless worker ants that can direct their falls
so that they land on the trunk of the tree they fell from. Should we con-
sider these ants to be gliders? They have no apparent wing-like aerody-
namic surface and, at best, their “glide” is a steep 75 degrees or 80 degrees
below the horizontal. Yet they do not fall vertically, and they can, without
a doubt, actively steer their course. These ants use aerodynamic forces to
direct and probably slow their fall. They may not be very efficient gliders,
but what they do fits the aerodynamic definition of a glider with a very low
lift-to-drag ratio (L/D) value. Moreover, ants aren’t the only ones doing
this. Other wingless arboreal arthropods—some spiders, bristletails, and
immature insects—also have this ability.4

WHO’S WHO? EX TINCT AND LIVING GLIDERS

Extinct Gliders

Lizards and lizard-like reptiles have evolved a Draco-like gliding body

form at least four different times. (Although some of them are not techni-
cally lizards, they are on the lineage leading to modern lizards and they
had body proportions and postures generally similar to present-day liz-
ards.) The earliest one we know about so far was Coelurosauravus elivensis
from the late Permian period, about 252 million years ago (Fig.  4.3).
Unlike the other lizard-like gliders, Coelurosauravus supported its wings
with struts of dermal bone separate from its ribs, but externally its wing
would have looked very much like Draco’s.5 The late Triassic kuehneosau-
rids from about 220 million years ago include the largest—several species
in the genus Kuehneosaurus—and the smallest—Icarosaurus siefkeri—of
the lizard-like gliders.6–8 The youngest (and most recently discovered) of
these extinct gliding reptiles is Xianglong zhaoi, from the early Cretaceous
period, about 127 million years ago.9 Other than Coelurosauravus, all had
wing membranes supported by modified ribs.
Intriguingly, all these fossil gliders appear to have had substantially
longer wings for their size than Draco. Given that Draco’s wing length seems
to be limited by the need for its wings to fold neatly along its flanks and not
interfere with its hind legs, we are left to wonder exactly how these fossil
gliders stowed their wings. Presumably these extinct animals received

[ 64 ]  On the Wing

 Kuehneosaurus

Icarosaurus

Coelurosauravus

Figure 4.3:
Extinct gliding lizards and lizard-like reptiles, reconstructed to approximately the same
scale; Kuehneosaurus had a wingspan of approximately 40 centimeters or 16 inches. Color
patterns are entirely speculative but are inspired by patterns on wings of living species of
Draco. (Courtesy of Lori Messenger, used by permission.)
some benefit from their long wings. These longer wings give them higher
aspect ratios, which in turn give them higher L/Ds, so at first glance, these
early gliders would seem to have more aerodynamically effective wings
than Draco. In fact, the ancient gliders show very little aerodynamic im-
provement over their modern counterpart. Partly because all gliders start
every glide with a ballistic fall to build up airspeed—where wings don’t
matter—and partly because the extinct gliders’ wings were still short by
bird standards, they appear not to have been especially efficient compared
to Draco. Researchers Jimmy McGuire and Robert Dudley used models of
the extinct gliders in wind tunnels to show that their slightly higher aspect
ratios did not seem to confer any particular advantage other than simply
increasing the wing area.8 All else being equal, more wing area means lower
wing loading, and the researchers found that lowering wing loading im-
proved glider performance much more than increasing aspect ratio.
A couple of early reptiles evolved gliding with less Draco-like bodies. For
example, Mecistotrachelos apeoros, of similar age to the kuehneosaurids,
had a similar wing membrane to the kuehneosaurids but had a startlingly
long neck—almost as long as its trunk or torso length.10 Birds with long
necks usually fly with the neck pulled into an S-curve to shorten it, but the
neck of Mecistotrachelos was not that flexible, so it must have glided with
its neck extended (Fig.  4.4). Although this posture might be unstable,
slight head movements would have been very effective for steering.
Sharovipteryx mirabilis was another, even more unusual gliding reptile.
This animal had short front legs and very long hind legs, and it had a large
gliding membrane supported mainly by the hind legs. Scientists have puz-
zled over how Sharovipteryx held its legs in flight ever since it was first de-
scribed by a Russian scientist back in 1971. The most recent suggestion is
that Sharovipteryx held its hind legs out so that they formed a delta-shaped
wing,11 reminiscent of the supersonic Concorde airliner (Fig. 4.4)!
Very few fossils have been found of gliding mammals, perhaps because
they tend to be small and delicately built and also because they may not be
recognized as gliders if the wing membrane is not preserved. One spec-
tacular exception is Volaticotherium antiquum, a petite, 13-centimeter (5-
inch) primitive mammal from the Jurassic of China (the exact age is
uncertain, but it is well over 130 million years old).12 It appears to have
been an arboreal animal well adapted to gliding between scattered trees in
open woodlands. This animal is so ancient that it is not closely related to
any modern mammals. The ancient age of Volaticotherium, plus the
common occurrence of gliding among modern mammals, suggests that
gliding evolves fairly readily in arboreal animals, so scientists suspect that
a number of extinct gliding mammal species await discovery.

[ 66 ]  On the Wing

 A

Figure 4.4:
Extinct gliders with strange anatomy. A. Mecistotrachelos. B. Sharovipteryx. (Courtesy
of S. T.)

Living Gliders

Given Draco’s rather small-looking wing, these creatures are surprisingly

capable flyers. One biologist described a territorial male chasing an inter-
loping male along tree branches and through the air, with the aerial part
of the chase involving as much turning and twisting as the part along
branches.1 People have seen gliding Draco lizards perform sharp turns, U-
turns, and even barrel rolls. Although their long tails undoubtedly help to
stabilize their flight, these lizards seem to steer mainly using wing adjust-
ments, just like a bird.
Several species of gecko also glide, and they also use a flap of skin along-
side their flanks. In geckos, however, these skin flaps have no internal
support, are not connected to the legs, and appear to be extended pas-
sively, entirely by air pressure.13 They are much smaller than the wings of
a similarly sized Draco, but geckos compensate at least partly with webbed
feet. These webbed front and hind feet not only provide additional wing

Gliding A nim a l s  [ 67 ]

area, but geckos must also use them for steering, since they have no way
to actively adjust the wing.
Snakes must seem like the least likely of all animals to evolve gliding,
but snakes in the Southeast Asian genus Chrysopelea have done just that.
Biologist John “Jake” Socha has studied these amazing snakes for several
years and he was the first to try to analyze their flight.14 He found that
these snakes don’t just fall off a branch to start gliding; they actually jump
up, which may be unique among snakes. To prepare to jump, a snake allows
the front half of its body to droop below the branch, then quickly flings
the front of its body up, which lifts the rest of its body off the branch. The
snake then forms its body into two or three big sideways loops, looking
like a sort of squashed S-shape when viewed from above. The snake spreads
out its ribs so its body becomes somewhat flattened, partly for streamlin-
ing and partly to increase its “wing” area. Perhaps the most disconcerting
feature of Chrysopelea gliding is that the body is not static; it undulates
rapidly, almost as if it were swimming through the air. Although they are
not stellar gliders—their glide angles are between 20 degrees and 45
degrees­—these snakes are capable of making sharp turns in flight, and
they much prefer to land in other trees or shrubs rather than on the
ground.

Mammals

Gliding has evolved at least six times among living mammals: twice in ro-
dents, three times in marsupials, and once in the colugos or “flying lemurs.”
Among rodents, the 40 or so species of flying squirrels are extremely
widespread, from North America to northern Europe and across central,
south, and southeast Asia. They can be fairly common in mature forests,
but most people rarely see them because many species are nocturnal.
The southeast Asian colugos are among the largest modern gliders and
are also probably the most aerodynamically sophisticated. These ferret-
sized, secretive gliders live in the canopies of some of the world’s tallest
rain forests in Malaysia, Indonesia, and the Philippines, so they have es-
caped attention from most scientists. Indeed, they are so poorly studied
that they have long been known as “flying lemurs,” even though they are
certainly not lemurs and are not particularly closely related to lemurs.
Colugos glide on a large fold of skin that stretches from their front
limbs to their hindlimbs and continues between the hindlimbs to include
the tail; it also continues forward to the side of the neck (Fig. 4.5). Unlike
other mammalian gliders, the colugo’s aerodynamic surface includes the

[ 68 ]  On the Wing

Figure 4.5:
Colugo gliding, showing the flight membrane (patagium) and webbed fingers. (Cour-
tesy of S. T.)

hands and feet; they have highly elongated, webbed fingers and toes. Al-
though we know very little about how colugos glide, for the hand to be
incorporated into the flying surface hints at unusual maneuverability.
Changing the shape of the tip of the wing can be a very powerful turn
mechanism, and dexterous fingers would give colugos very fine, precise
control of such a potent mechanism.
Colugos also have low wing loading for gliding animals (that is, large
wings for their body weight), which allows them to glide more slowly for
their size than other gliding mammals.15 Adjustable wingtips and slow air
speeds would make colugos extremely maneuverable. Thus, although we
know very little about their behavior or ecology, colugos’ anatomy sug-
gests they are probably the most accomplished of gliders.

Other Gliders

At least two different lineages of frogs have evolved gliding. Gliding frogs
may sound odd, but in many moist temperate and especially tropical re-
gions, frogs live in trees. Moreover, frogs typically have webbed hands and
feet, which make handy steering devices for directed aerial descent if they
should happen to fall out of a tree. Their webbed feet are a nice example of
an exaptation (Chapter 2): they originally evolved for swimming, but are
quite useful for directed aerial descent with little modification. They thus

Gliding A nim a l s  [ 69 ]

provide the raw material for gliding adaptations involving greater surface
area. Extant gliding frogs have elongated fingers and toes, so their webbed
hands and feet form aerodynamic surfaces on which they glide. Such
“flying frogs” are found in both South American and southeast Asian
rain forests.
People once thought that oceanic flying fish (Exocoetidae) and freshwa-
ter hatchetfish (Gastropelecidae) could fly by flapping their enlarged pec-
toral fins as wings. We now know that exocoetid flying fish can only glide
when airborne—although they do have the remarkable ability to dip their
tails in the water while airborne and beat them to accelerate for another
bout of gliding.16 These oceanic flying fish are thus the only known ani-
mals that evolved gliding without living in trees. In contrast, after de-
cades of anecdotes about tropical hatchetfish flapping their fins to fly in
air, careful research showed that they are excellent leapers but get little or
no benefit from their fins while traveling in air.17 Not only do they fail to
flap but they are not even any good at gliding.
Finally, as noted above, ants as well as several other kinds of wingless
arthropods perform a rudimentary kind of gliding when they “skydive”
back to the tree trunk after being dislodged or jumping from a branch.

WHY GLIDE?

Given the enormous variety of animals that have evolved gliding ability,
gliding must confer some advantage significant enough to outweigh the
cost of growing the wing surface and carrying it around when not gliding.
Probably the biggest advantage is escaping from predators. A glider can
leap out of a tree and safely reach another tree trunk or branch that is
much too far away for a non-gliding leaper to reach. So if a glider can detect
a non-gliding predator in time, the glider will normally escape from such a
predator with ease. This ability has an extremely strong evolutionary ben-
efit, because being killed by a predator removes the prey individual’s genes
from the gene pool; an individual that can escape from a whole category of
predators is much more likely to contribute genes to the next generation.
Thus, many biologists suspect that escape from predators was the selec-
tive factor that initiated gliding adaptations in most gliding animals.18
Escaping from predators is not the only advantage of gliding, however.
Arboreal animals commonly risk falling out of trees, and the risks of
falling­—and benefits of gliding—are different depending on the animal’s
size. For instance, as animals get larger, they are increasingly likely to be
injured by falls from great heights. When animals get much larger than,

[ 70 ]  On the Wing

say, a rat, both their terminal velocity (maximum falling speed) and their
momentum are relatively high, so the bigger they get, the more likely they
are to be injured on impact with the ground. Gliding obviously reduces
this risk and allows a falling animal to convert vertical speed into hori-
zontal speed and achieve a safe landing.
Smaller animals, in contrast, have high aerodynamic drag due to their
high surface-to-volume ratios, and also low body weight. These features in
turn give them low terminal velocities and low momentum: a thumb-sized
shrew or a pencil-sized millipede is unlikely to be injured by a fall from
even a very tall tree. (A colleague who shall remain nameless dropped
mice from the roof of a five-story building to demonstrate their low termi-
nal velocity, and the mice seemed largely unaffected.) Scientists discov-
ered the phenomenon of directed aerial descent (Chapter  3) in worker
ants, who have almost no risk of being injured in a fall.19 If you are arbo-
real and very small, the advantage of gliding is that you can avoid landing
on the ground and can stay up in the tree. Most specialized tree dwellers
are poorly adapted for getting around on the ground, and at least one
study has shown that arboreal ants are much more likely to be eaten by
predators on the ground than ants that normally live on the ground.20 In
short, gliders can better avoid landing on the ground where they are ex-
posed to unfamiliar predators. This advantage does not just apply to small
gliders; videos of captured colugos released on the ground show that they
invariably hop as quickly as they can to the nearest tree, and they find it
distressing to be on the ground at all.
Once a species evolves some gliding ability, another factor comes into
play: energy conservation. A flying squirrel can glide from one tree to an-
other distant tree using much less energy than a non-gliding squirrel
would use climbing down the first tree, running across the intervening
ground, and climbing back up the second tree. This difference is a great
benefit when an arboreal animal needs to travel long distances or search
many trees to find food. Not only does gliding require less energy but it is
inherently faster. So a glider can search more territory faster and more
economically than a similar non-glider.

STEPPINGSTONE TO FL APPING VERSUS THE SUCCESSFUL END STATE

Biologists are prone to refer to flapping flight as “true flight,” thus relegat-
ing gliding to some other state of less-than-true flight. As I have said
before, this distinction is aerodynamically absurd because a flying squir-
rel’s wing membrane is doing the same thing as the wing of a jet airliner,

Gliding A nim a l s  [ 71 ]

and no one would argue that the airliner is not truly flying. Nevertheless,
given that prevailing attitude, biologists and paleontologists in the past
have often seemed to treat gliding as a sort of transitional state, a tempo-
rary steppingstone on the way to evolving powered flight. Indeed, bats
probably evolved from arboreal gliders, and at least a few scientists think
that may be true for birds and insects as well. Treating gliders as if they
are in transition on the way to some more “advanced” state, however, mis-
reads both the evolutionary process and the evidence; today most scien-
tists would consider such an attitude to be obsolete.
I hope this chapter has shown that gliding is actually a sophisticated
adaptation that involves changes in anatomy, sensing, behavior, and judg-
ment. In addition to the obvious structural modifications to produce some
sort of wing, the glider needs acute vision to accurately see distant land-
ing spots and rapidly approaching obstacles. A gliding animal also needs
the mental ability to judge whether a potential landing site is within safe
gliding distance and must have the behavioral adaptations that allow ef-
fective aerial steering. Effective gliding is thus a complex adaptation, one
that provides substantial benefits even without being actively powered.
Gliding has evolved dozens of times in the animal kingdom, and we
have no evidence that these gliders were short-lived transitional forms.
Although the fossil record is not very good for modern gliders, evidence
from molecular phylogenetics suggests that Draco split off from other liz-
ards at least 35 million years ago and has probably been gliding most of
that time.21 The great diversity of fossil and living gliders, as opposed to
only four lineages of powered flyers, suggests that gliding is a smaller ev-
olutionary step than flapping flight.
Yet even though gliding is a perfectly viable lifestyle on its own, at least
some powered flyers evolved from arboreal gliders. In the debates sur-
rounding the evolution of bird flight, some scientists have expressed the
view that “gliders don’t flap, flappers don’t glide.”22 Leaving aside the fact
that many flappers do glide routinely, the phrase illustrates the misappre-
hension that gliding and flapping are two qualitatively different, mutually
exclusive processes. As we saw in Chapter 3, gliding and flapping are in
fact two ends of a continuum, and nothing physical or aerodynamic pre-
vents a glider from taking advantage of minimal attempts at flapping.

REFERENCES
1. E. H. Colbert (1967) American Museum Novitates.
2. H. Tennekes (1996) The Simple Science of Flight: From Insects to Jumbo Jets.
3. R. W. Thorington Jr. and L. R. Heaney (1981) Journal of Mammalogy.

[ 72 ]  On the Wing

 4. S. P. Yanoviak, M. Kaspari, and R. Dudley (2009) Biology Letters.
5. E. Frey, H.-D. Sues, and W. Munk (1997) Science.
6. P. L. Robinson (1967) Science & Culture.
7. E. H. Colbert (1970) Bulletin of the American Museum of Natural History.
8. J. A. McGuire and R. Dudley (2011) Integrative and Comparative Biology.
9. P.-P. Li, K.-Q. Gao, L.-H. Hou, et al. (2007) Proceedings of the National Academy of
Sciences of the United States of America.
10. N. C. Fraser, P. E. Olsen, A. C. Dooley Jr., et al. (2007) Journal of Vertebrate
Paleontology.
11. G. J. Dyke, R. L. Nudds, and J. M. V. Rayner (2006) Journal of Evolutionary
Biology.
12. J. Meng, Y. Hu, Y. Wang, et al. (2006) Nature.
13. A. P. Russell, L. D. Dijkstra, and G. L. Powell (2001) Journal of Morphology.
14. J. J. Socha and M. LaBarbera (2005) Journal of Experimental Biology.
15. B. J. Stafford (1999) [Dissertation].
16. J. Davenport (1994) Reviews in Fish Biology and Fisheries.
17. F. C. Wiest (1995) Journal of Zoology.
18. U. M. Norberg (1990) Vertebrate Flight: Mechanics, Physiology, Morphology, Ecol-
ogy and Evolution.
19. S. P. Yanoviak, R. Dudley, and M. Kaspari (2005) Nature.
20. S. P. Yanoviak (2010) in Ant Ecology.
21. M. P. Heinicke, E. Greenbaum, T. R. Jackman, et al. (2012) Biology Letters.
22. K. L. Bishop (2008) Quarterly Review of Biology.

Gliding A nim a l s  [ 73 ]

CH A P TER  5

Insects
First to Fly

L ong before amphibious vertebrates first tentatively squirmed onto

land, insects had already colonized land and evolved fully powered
flight. Indeed, insects had already achieved flight more than 150 million
years before the first, most primitive dinosaurs appeared, and at least 160
million years before pterosaurs followed them into the air as the second
group of powered flyers. We have ample fossil evidence that flying insects
were well established and diverse over 300 million years ago, but the de-
tails of how they got there are more than a bit murky.

OLDEST INSECTS

Scientists have discovered an abundance of winged insect fossils from as

far back as 318 million years ago, but prior to that, the insect fossil record
is exceedingly sparse. Rocks from about 407 million years ago have pro-
duced fossils of a single species of springtail (order Collembola).1 Spring-
tails are actually fairly common today, but they are all tiny—the largest is
about half a centimeter (1⁄5 inch) long and most are very much smaller­—
and they live in soil and leaf litter, so hardly anyone other than entomolo-
gists encounter them. They are called springtails because of a tail-like
appendage that they use like a catapult for leaping, but no springtail has
anything remotely like wings. In the past, biologists considered spring-
tails to be extremely primitive insects, but nowadays most biologists con-
sider them not to be true insects but the closest relatives of true insects.
In any case, those 407-million-year-old fossil springtails represent the
oldest known fossils on the lineage leading to insects. Other rocks from
about 380 million years ago yielded fragments of a bristletail (order Ar-
chaeognatha),2 a very primitive type of insect that looks superficially like
a silverfish.* Both the springtail and the bristletail diverged from the lin-
eage leading to winged insects well before flight evolved; scientists call
such animals “primitively wingless” as opposed to insects descended from
flying insects that have secondarily lost wings. Until recently, scientists
had nothing to link these most ancient insect and related fossils with
100-million-year-younger flying insects.
Sometime around 2000, entomologist David Grimaldi and my Univer-
sity of Kansas colleague, Michael Engel, were on a trip to visit museums
around the world. They were writing a book on the evolution of insects,
and they were visiting museums to take original photos of fossil insect
specimens to illustrate their book. As Michael recalls, they were on the
way back from Moscow and decided to make a brief stop in London to
look at the original fossils of those 407-million-year-old springtails at
the British Museum of Natural History. While they were looking at the
microscope slides of the collembolans, Michael happened to notice an-
other slide in the same drawer of a different specimen from the same
fossil formation. (It had been given a name, Rhyniognatha, more or less in
passing when it was first prepared in the 1920s, but it was never really
described.) Deciding he might as well take a quick look at it since they
were there anyway, Michael put it on a microscope—and was stunned.
The fossil appeared to be a set of mandibles or jaws with a double hinge
(“dicondylic”) that is utterly diagnostic of flying insects. In other words,
if it had been from a living insect, it would have unquestionably come
from an insect with wings. Trying not to show too much excitement,
Grimaldi and Engel casually asked if they could borrow this “nondescript”
specimen to take it back to the University of Kansas for more careful
study. When they published their study, scientists realized that this fossil
could mean that insect flight had already evolved by 407 million years
ago.3 Unfortunately, all that survived were the mandibles; no trace of the
thorax fossilized, so we don’t know for sure that this insect had wings.
Moreover, as Michael has explained many times, the conditions (boiling
mineral springs) where the fossils formed would most likely have de-
stroyed wings if they had been present, so he says looking for wings in the
flint-like rocks where the mandibles were found is “a fool’s errand.” We
thus have a tantalizing hint that insect flight may have evolved nearly

*  Despite the name, silverfish are insects, not fish. They have a silvery appearance
due to a covering of very fine scales, and their bodies do have a vaguely fishlike shape.

Insec ts  [ 75 ]

100 million years earlier than previously assumed, which would make the
gap between the time of the origin of flight and the oldest known fossils
of winged insects even longer.
Scientists face a major puzzle in trying to understand the evolution of
wings and flight in insects. Starting about 318 million years ago, the
fossil record becomes rich in winged insects (including many isolated
wings). Because insect wings usually have distinctive vein patterns that
identify their lineage, these early insect wing fossils clearly demonstrate
that a lot of evolution and diversification had already occurred among
flying insects. For all that diversification to have already happened, flight
must have first evolved very much earlier—many tens of millions of
years at a minimum. This observation, combined with Engel’s new look at
those tantalizing mandibles, led Engel and his colleagues to suggest that
insects must have evolved flight at least 400 million years before the
present (and possibly considerably earlier), but so far we have found no
insect fossils of any kind earlier than those frustratingly enigmatic jaws
from 407 million years ago, nor any trace of insect wings older than 320
million years ago. This gap in the fossil record might seem implausible,
but due to tectonic changes of the continents, terrestrial and freshwater
fossils from that period are extremely hard to find. Early insects were
most likely terrestrial,4 and since the main source of fossils of terrestrial
animals is from those that fall into swamps and get buried in silt, the
absence of non-marine fossils from that time means that we have no
useful source of insect fossils from the relevant periods (late Silurian
through mid-Carboniferous, 420 to 320 million years ago). We thus have
a weak suggestion of when insects may have evolved flight, but so far, we
have no direct evidence of any of the early stages of wing evolution in
insects.
Just to add a bit to the puzzle, in 2014 a German-Czech team led by
Arnold Staniczek re-described a 309-million-year-old fossil, originally de-
scribed as a juvenile mayfly, as a wingless adult. Carbotriplurida kukalovae
was a large insect (10 centimeters or 4 inches long) with long legs, big eyes,
and long tail filaments. It also had flat, side extensions of the roof of the
thorax and abdomen just as predicted by the paranotal lobe theory de-
scribed later in this chapter. These researchers suggest that Carbotriplur-
ida was evolutionarily intermediate between the wingless Zygentoma
(silverfish and relatives) and the winged insects.5 Carbotriplurida, how-
ever, lived many tens of millions of years after insects had already evolved
flight, so if it does represent the direct ancestor of winged insects, a wing-
less part of the lineage must have split off early and persisted remarkably
long into the age of winged insects.

[ 76 ]  On the Wing

TR ADITIONAL WING ORIGIN THEORIES

A lack of informative fossils has not prevented scientists from speculat-

ing about the evolutionary beginnings of insect flight. The arguments
about possible wing precursors go back well over a century. In fact, the
two traditional, widely accepted theories were both first described in an
evolutionary context in the mid-1800s. In 1870, Carl Gegenbaur con-
verted a 60-year-old opinion that insect wings were modified from
larval gills into a more modern scientific theory consistent with Dar-
win’s just-published theory of evolution.6,7 Almost immediately, in 1873,
Fritz Müller proposed the alternative paranotal lobe theory (see below)
as a counterargument because he thought some developmental evidence
in termites was inconsistent with the gill theory.8,9 These two theories
were the only widely accepted theories to explain the evolution of insect
flight throughout the 20th century. Even after all that time, variations
and updates of these theories are still accepted by many scientists to
this day.

The Gill Theory

The gill theory (also known as the pleural appendage theory) is based on
the fact that several lineages of insects have juvenile stages that possess
several pairs of external, movable gills, usually on the abdomen
(Fig. 5.1). Not only are these gills movable but they often contain tra-
cheae (air tubes) in a branching pattern reminiscent of wing veins. Pro-
ponents of the gill theory suggest that the ancestors of winged insects
had juveniles with gills on the thorax as well as the abdomen. Moreover,
some modern aquatic insect larvae (technically naiads; see Box 5.1 for
terminology used for immature insects) use these gill plates more like
fans to drive water over additional respiratory surfaces, and some even
use them as paddles for swimming. The main idea of the gill theory is
that some ancestral insect evolved to use the gill plates on the thorax
more as paddles than for respiration. Then if this insect spent part of its
time out of water, the thoracic paddles might have allowed steering
during leaps or falls, or a little gliding to extend leaps, or even weak
flapping to aid steering and gliding. As the thoracic paddles became
useful in air, natural selection would have led to bigger paddles to pro-
duce more lift and thrust and more effective steering, and eventually to
stronger muscles and more refined flapping movements to achieve fully
powered flight.

Insec ts  [ 77 ]

Figure 5.1:
An immature mayfly (“naiad”), showing the rows of gills along its abdomen. (Courtesy 
of S. T.)

Box 5.1:  IMMATURE INSECTS

The terminology for juvenile or immature insects can be challenging.

Entomologists call an immature individual of a species that goes
through a pupal stage (those with “complete” metamorphosis) a “larva,”
plural “larvae.” Caterpillars, fly maggots, and beetle grubs are familiar
examples. Outside of entomology, a “larva” can be any immature stage,
although it usually refers to a juvenile that looks different from the
adult; non-entomologist biologists thus tend to call all immature in-
sects “larvae.” Juveniles of insects without a pupal stage (“incomplete”
metamorphosis) tend to look rather like miniature adults although
without wings or external reproductive structures, so they are properly
called “nymphs.” Aquatic juveniles of insects with incomplete
metamorphosis­—for example, dragonflies, mayflies, or stoneflies—
can be considered nymphs, but since they look noticeably different
from the adult stage, entomologists often refer to them as “naiads.”

[ 78 ]  On the Wing

 Box 5.1:  Continued

­ uriously, “nymph” and “naiad” are both borrowed from Greek mythol-
C
ogy, Nymphs being minor female deities associated with specific places
and Naiads being a subset of Nymphs associated with streams and
lakes. Using “naiad” has a certain logic to it, but using “nymph” makes
no sense to me, especially given that Nymphs are often associated with
liberated sexuality. Insect nymphs are, by definition, incapable of re-
productive behavior!

The gill theory has the advantage of starting with a movable appendage
in more or less the right place on the body, a useful exaptation for steering
and incipient flapping. This theory has some drawbacks, however. First, if
these gill-paddles served a partly respiratory function underwater, they
would be unable to do so effectively in air. In order to have a thin enough
covering to allow sufficient gas exchange under water, gills are usually too
floppy to support themselves out of water, typically collapsing into inef-
fective, low-surface-area clumps in air. Most gills also require some venti-
lating mechanism to drive a flow of water over them, and a pump evolved
to move water would be useless in air. Second, simply passing through the
air-water interface can be quite difficult for an insect-sized animal. Many
insects depend on the surface tension of the air-water interface to support
their entire body weight, water striders being a familiar example. The fact
that an insect’s entire body weight is not enough to push through the sur-
face shows how much effort tiny animals need just to pass through that
interface. The large surface area of thoracic gill-paddles big enough to be
useful in air would make exiting water even more difficult if not impossi-
ble. The biggest problem, however, is that most entomologists think that
early insects evolved on land, and those with aquatic larvae evolved sec-
ondarily from terrestrial ancestors.4 The reason is that the insect respira-
tory system consists of a branching network of air tubes throughout the
body called tracheae. Tracheae can be quite effective at exchanging oxygen
and carbon dioxide in air, but this system simply does not work well un-
derwater and certainly would not have evolved there. The tracheal system
is a fundamental feature of insects and it unquestionably evolved in air.
Bristletails and silverfish are members of two ancient lineages that split
off from other insects before wings evolved and they both have tracheae,
so most entomologists think that the earliest insects must have been
terrestrial.10

Insec ts  [ 79 ]

The Paranotal Lobe Theory

The essence of the paranotal lobe theory is that the ancestors of winged
insects evolved flat plates extending from the top of the thorax out to
each side. These are the paranotal lobes, “para-” meaning “alongside”
and “notum” being one of the plates of the exoskeleton that make up
the roof of the thorax. The main idea here is that this ancestral insect
could have evolved these lateral extensions for some non-aerodynamic
function—for example, camouflage, side protection, or strengthened
anchoring for leg muscles—although an early aerodynamic function
cannot be ruled out. Whatever the original use, these plates or “lobes”
eventually became large enough to have a significant aerodynamic
effect. They might then have been used for parachuting, or for gliding
to extend leaps from foliage or other elevated perches. These lobes
would initially have been fixed or immobile. If, however, they could be
adjusted for steering, they would have been more versatile for gliding
or just steering during leaps. They would then have evolved a joint or
hinge, which in turn would require co-opting some muscles to control
their movements. Once the lobes became movable for steering, the
basic building blocks were in place. The lobes would get larger to pro-
duce more lift, and the steering movements would evolve from weak
flapping to fully powered flight.11
The main objection to the paranotal lobe theory is that it requires
the incipient wing to evolve a new joint or articulation from scratch.
Although not impossible, evolution of completely new joints in rigid
skeletons is surprisingly rare; fusion of skeletal elements (bones, or
plates of exoskeletons) or loss of joints seems fairly common, but de-
veloping an entirely new joint within a single plate or bone, as opposed
to modifying an existing joint, is extremely unusual. Most books and
articles that discuss the paranotal theory, however, mention one
thought-provoking example of a new joint: that in oribatid mites.
Mites are relatives of spiders, not closely related to insects, and the
oribatid mite lineage has evolved what amount to paranotal lobes, in-
cluding a new hinge, although these paranotal lobes have no known
flight-related function.10
For most of the 20th century, the paranotal lobe theory was more
widely accepted than the gill theory. For example, Richard Alexander re-
viewed previous work and argued that the paranotal lobe theory was more
likely. Noting that only adults have wings, he suggested that paranotal
lobes may have been used originally in courtship displays or mating be-
havior.12 If lateral thoracic extensions had originally evolved as courtship

[ 80 ]  On the Wing

displays to attract members of the opposite sex, sexual selection* can
indeed be a potent agent to increase elaboration, as amply demonstrated
by peacock tails. Sexual selection could have driven selection for ever-
larger lobes and even conceivably a new joint allowing movement to in-
crease the effectiveness of the display. Alexander contended that if an
insect with such advertising devices were also an active, vegetation-­
inhabiting species, its thoracic display structures could also help guide
and extend leaps within and between plants so that the insect could avoid
landing on the ground.
Although the paranotal lobe theory probably spent more time as the
leading contender, proponents of the gill theory never disappeared. In
1973, eminent insect physiologist Vincent Wigglesworth revised and up-
dated the gill theory by suggesting that wings evolved from side branches
off the bases of legs called “exites” or “exopodites.” He pointed out that
very primitive insects like silverfish have tiny, peg-like appendages called
“styli” on the thorax and abdomen, which he suggested were vestigial rem-
nants of larger, mobile exites. (See Box 5.2. Insect and Crustacean Limbs.)
If exites had evolved into gills in aquatic species, they would have pro-
vided the basis for mobile protowings,13 although they would have had to
shift up higher on the side of the body somewhere along the line.
Jarmila Kukalova-Peck, a prolific insect paleontologist, took Wig-
glesworth’s suggestion a step further. She suggested, based on fossils,
that the primitive insects had miniature legs on all the abdominal
segments as well as large legs on the three thoracic segments. She
argued that the primitive insect leg had an extra basal segment that
evolved into the side walls of the thorax and abdomen, and the exites
of this hypothetical segment formed flattened, mobile plates all along
the upper sides of the thorax and abdomen. These exites all evolved
into gills in an early aquatic lineage, and from there, her suggestion
follows the standard gill theory. Although she assumed that an aquatic
stage was part of the process, her exites could in fact fit equally well
into the terrestrial scenario as either fixed or mobile paranotal lobes.
Kukalova-Peck seems to see more detail in fossils than other research-
ers, and few other scientists accept her hypothetical leg segment; so
until better fossils come along, her version of the gill theory remains
controversial. 5,14,15

*  Sexual selection is a form of natural selection (first described by Charles Darwin)

where one sex chooses mates based on some attribute not otherwise related to sur-
vival, which then causes that attribute to become exaggerated. The bright red color
of the male cardinal is another example.

Insec ts  [ 81 ]

 Box 5.2:  INSECT AND CRUSTACEAN LIMBS

Crustacean limbs are surprisingly complex compared to insect limbs.

Crustacean limbs are based on a forked or Y-shaped main axis of several
segments, often with smaller side branches (exites and endites) as well.
Biologists call such limbs “biramous.” For instance, on the lobster
thorax, the inner or medial main branch (endopod) develops into a
walking leg and the outer or lateral main branch (exopod) becomes a
feathery gill hidden under the carapace. An exite is a small, lateral side-
branch from near the base of the limb.
Insect legs, in contrast, are unbranched, forming a single column
made of several segments. Insects share these so-called uniramous
limbs with myriopods (centipedes and millipedes); for a long time, bi-
ologists thought that this trait indicated a close relationship between
insects and myriopods. “Uniramia” was even used as a name for the
group uniting the two. In the last few years, however, the evidence from
molecular phylogenies has become quite convincing that crustaceans
are the closest relatives of insects. In fact, insects are nested within the
crustacean phylogeny and so are technically a subgroup of crustaceans.
This means that insects and myriopods evolved their uniramous legs
independently. Wigglesworth’s suggestion was controversial at the
time because insect legs were then considered to be fundamentally un-
branched. Given the evolutionary relationship between crustaceans
and insects, Wigglesworth’s contention that styli of silverfish may be
vestigial exites cannot be discounted. Whether that has anything to do
with the evolution of wings is a separate question.

MODERN THEORIES: EXPERIMENTS AND OBSERVATIONS

Model Tests

Frustrated by the lack of fossil evidence for wing origins in insects, a

number of researchers have taken a different tack. Some have used experi-
ments to try to tease apart the physical features and constraints that
might have favored acquisition of wings. Others have looked at living in-
sects to find behaviors that might have a bearing on the evolution of flight.
For example, evolutionary biologist Joel Kingsolver and biomechanics
specialist Mimi Koehl built a series of models of hypothetical insects to test
their thermodynamic and aerodynamic properties. Earlier scientists had
looked at insect wings as possible solar collectors—warm muscles contract
faster, which might be favored by natural selection—or at aerodynamics of

[ 82 ]  On the Wing

cylinders as simplified stand-ins for wingless insect bodies.16,17 But King-
solver and Koehl took a more comprehensive approach. They wanted to see
if small, flat thoracic projections or “protowings” could have provided any
non-aerodynamic benefits and to figure out the conditions where such pro-
towings might provide aerodynamic advantages. They used models over a
range of sizes from 2 to 10 centimeters (about 1 to 4 inches) in length and
either wingless or with wings over a range of lengths (Fig.  5.2A). They
looked at thermal properties by measuring the models’ core temperature
under flood lamps and their aerodynamic properties in a wind tunnel.18
Their results show intriguing interrelationships.

Figure 5.2:
Simplified models used to test theories of insect wing origins (not to scale). A. Models
used to compare thermal and aerodynamic properties, redrawn from Kingsolver and
Koehl (by permission of Cambridge University Press).18 B. Models used to test gliding
stability, redrawn from Wootton and Ellington (by permission of John Wiley & Sons).20

Insec ts  [ 83 ]

 They found that stubby wings are aerodynamically effective—­
increasing glide distance—on big insects, but small insects need rather
long wings to get similar benefits. Conversely, short, stubby winglets
make very effective solar collectors, but longer wings provide little or no
additional solar warming. The effect was most pronounced in the largest
insect model; here stubby winglets of less than a ½ centimeter (1⁄5 inch)
greatly improved solar heating over no winglets, whereas the smallest
model gained little or no extra heating from short winglets. Kingsolver
and Koehl concluded that if natural selection favored solar heating, per-
haps to warm up faster than your predators or prey on cool mornings,
stubby thoracic projections on medium- to large-sized insects could be ad-
vantageous. If such projections could be adjusted to orient toward or away
from the sun, they might have been even more effective, and this would
have favored movable plates.
Kingsolver and Koehl showed that if such a hypothetical ancestral
insect were large, its solar-collectors-cum-protowings would not have had
to get very big to start having some aerodynamic effects. Small insects, in
contrast, would need quite long protowings to get much aerodynamic ben-
efit. With big enough protowings, all the advantages of gliding (Chapter 4)
would come into play, particularly if the solar collectors were movable to
allow steering. If this insect benefited from increasing glide distances,
then increasing wing size and developing flight muscles would have been
favored—and this would have put such an insect well on the road to flap-
ping flight. Some scientists have called this analysis oversimplified, saying
that such solar collectors would be quite cumbersome when not in use and
that not all situations would necessarily favor heated muscles. Neverthe-
less, this study showed how thermal biology and aerodynamics could in-
teract to favor wing evolution.
In a later review article, Kingsolver and Koehl referred to this approach
as “bounded ignorance,”19 which is one of my favorite technical terms.
What they meant is that we can use experiments to show what is physi-
cally possible (or impossible) and what factors could be advantageous­—
placing boundaries—even when the gaps in the fossil record mean that
we remain ignorant of what actually happened. A number of other studies
have taken this bounded-ignorance approach to the study of flight
evolution.
British zoologist Robin Wootton and his colleague Charles Ellington (a
top expert in insect aerodynamics) performed one of these studies. They
also built models, which they tossed as gliders to study stabilizing effects
of a variety of possible anatomical embellishments. Much earlier, aero-
nautical engineer J. W. Flower had shown that insect-sized cylinders could

[ 84 ]  On the Wing

achieve surprisingly good glide angles (up to 45 degrees) if they could be
stabilized at the proper falling angle.17 Wootton and Ellington tested
model cylinders with either a series of nine pairs of small adjustable wing-
lets down each side or a single pair of larger, fixed winglets near the front,
each version with and without long tail filaments (Fig. 5.2B). Their main
goal was to see whether these structures could give the cylinders the nec-
essary stability to fall at an angle that generated lift and, hence, horizon-
tal motion.20 In other words, could they achieve a decent glide angle with
a basically cylindrical body? Their answer was a qualified “yes,” depending
on the particular combination and orientation of the structures. For ex-
ample, the glide angle and stability of the model with several pairs of
small winglets was largely set by the angle of the hindmost pair of wing-
lets, and the model became unstable if that pair was removed. By giving
the model paper “tail” filaments like those of many primitive insects,
however, the model regained stability. These scientists thus showed that a
hypothetical ancient insect could have converted a vertical fall into a 30-
degree to 45-degree glide with small, stubby winglets and tail filaments.
As with other researchers, they found that the aerodynamic benefits were
greatest for models representing larger insects.
Wootton and Ellington thus showed that if the ancestor of winged in-
sects had a series of little plates down the side of the thorax and abdomen
(as in the gill theory), and if the plates were adjustable, the plates did not
have to be very big to provide some gliding. Alternatively, they showed
that a single, immobile, stubby pair of winglets (as in the paranotal lobe
theory) along with a set of tail filaments was not quite as effective but still
gave the cylinder the stability needed to convert a fall to a glide. So even
very short, fixed wing surfaces could have provided some aerodynamic
benefit.
All of these model studies can help explain what is physically possible
and what is improbable, but they cannot produce a definite answer to the
question of how wings evolved. Moreover, their results are quite size-­
dependent. The effectiveness of wings can change quite a bit over the
range of size from small insects to the largest of these models (16 centi-
meters, about 6 inches) because wing performance is quite sensitive to a
scale factor called the Reynolds number (see Box 3.1). If the direct ances-
tors of flying insects were smaller than 1 or 2 centimeters (½ to 1 inch) in
length, most of these model studies would not have accurately described
the aerodynamics. If the pre-flight insects were moderately large (in the
6- to 10-centimeter or 2- to 4-inch range or bigger), as some scientists sug-
gest, then the model studies could be relevant. So far, however, we have
little or no direct evidence of the size of these early insects.

Insec ts  [ 85 ]

Modern Analogs

A very different method from building physical models is to look at struc-

tures, processes, and behaviors in living animals and imagine how similar
features might have assisted the evolution of flight. This is still a form of
bounded ignorance: practitioners don’t say that living animals inherited
the relevant features directly from the earliest flyers; rather, they propose
that if modern animals could evolve such features, so might have the an-
cient insects who first evolved flight. James Marden’s surface-skimming
theory, stimulated by his observations of living stoneflies, illustrates this
approach.
Stoneflies are rather run-of-the-mill, unspecialized insects with aquatic
nymphs (larvae) and air-breathing adults. Even though most adult stone-
flies can fly, they are not particularly strong flyers so they rarely stray far
from water. Evolutionary biologist James Marden noticed that on cold
days, some stoneflies could not summon the power to fly up in the air, but
they could use their wings as propellers to skim quickly over the surface of
the water like an Everglades airboat. He and his then-student Melissa
Kramer studied the mechanics of surface skimming. They found that they
could clip off up to 80% of a stonefly’s wing area, and the stonefly could
still locomote across the water surface. Skimming speed, however, de-
creased as more and more of the wing was removed. Marden and Kramer
concluded that surface skimming might be an intermediate stage in the
evolution of flight.21 If an aquatic insect with gill plates (or gill-like pad-
dles) evolved an air-breathing adult stage, it might be able to skim across
the water surface even with small, weak winglets. Selection for faster
skimming would lead to longer wings and faster muscles. These improve-
ments might then allow hops off the water at first, leading eventually to
powered flight. In later work, Marden pointed out that some stoneflies
don’t even flap their wings, they just raise them up and angle them to catch
the wind, exactly like a sailboat.22 That behavior pushes the theory even
further back, to a semi-aquatic insect that used gills or paddles as sails
when on the water surface.
Marden has suggested that surface skimming may be a relict behavior
directly inherited from the ancestral insects that first evolved flight.23
Most scientists, however, think skimming stonefly behavior evolved long
after the origin of flight rather than having been directly inherited from
distant ancestors. Yes, stoneflies are fairly primitive, but dragonflies and
mayflies are even more primitive, so the ancestors of stoneflies must
surely have been capable of effective, fully powered flight; surface skim-
ming evolved secondarily from powered flight in stoneflies (as well as in

[ 86 ]  On the Wing

some flies and caddisflies). Moreover, as I have said, the preponderance of
evidence is that ancient insects were terrestrial, not aquatic.4,10,24
In my view, the most significant description of a feature of modern an-
imals relevant to the evolution of insect flight, perhaps even animal flight
in general, is work by Steven Yanoviak, Robert Dudley, and Michael Kas-
pari on directed aerial descent in ants (as discussed in Chapter 4).25–27 The
possibility that many arboreal animals can direct their descent without
any overt aerodynamic modifications means that we need to reconsider
many of the arguments about the origins of gliding and flying animals.
Yanoviak and colleagues’ work was actually foreshadowed in an article by
German entomologist I. Hasenfuss. Hasenfuss started from the premise
that the closest non-flying relatives of the winged insects are those in the
order Zygentoma (silverfish are common members of this group). He as-
sumed that the ancestors of flying insects must have had anatomy and
lifestyles similar to silverfish and their kin (Fig.  5.3): fast runners with
flattened bodies to fit easily into cracks and crevices. Hasenfuss dropped
silverfish from various heights and observed that they have a cat-like abil-
ity to always land on their feet, even if dropped from just a few body-
lengths above the surface. They spread their legs, antennae, and tail
filaments to slow their fall and bend their abdomens to roll over so that
they always land right-side-up, ready to scuttle off.
Hasenfuss then made simplified models of silverfish and dropped them
to see if they could glide.28 He found that the models fell with the head
aimed slightly down in a stable, steep glide. He then added short, thoracic
winglets to his models. He found that even with very short winglets, he
could more or less double glide distances of larger models; his small
models, however, did not glide noticeably better with short winglets, par-
alleling what Kingsolver and Koehl had found.
Hasenfuss hypothesized that the ancestral, silverfish-like insect ate at
the tips of growing plant shoots. If so, this insect could save time and
energy by jumping off the plant after it finished eating the most tender

Figure 5.3:
A silverfish, a primitively wingless insect. (Courtesy of S. T.)

Insec ts  [ 87 ]

part, landing safely, and running to climb up the nearest uneaten plant.
Some modern spiders can also reorient in falls to land on their feet, and
spiders were probably common predators of these ancient insects. If so,
then the insects would have benefited if they could angle their fall a bit so
they could land a little farther from the original plant (and from a pursu-
ing spider) and closer to a new plant. Indeed, parachuting to slow the fall
could even be detrimental, so a fast, steep glide would be most useful in
this setting. Natural selection would favor any plate-like extensions of the
thorax to extend glides and eventually reach distant plants without
spending any time on the ground at all. Hasenfuss even suggested how
flappable wings might have evolved based on quite sophisticated anatomi-
cal comparisons and described how initial weak flapping would have aided
escape from predators.28
If, as Yanoviak and company suggest, many animals prone to jumping
or falling from substantial heights can direct their descent toward the
nearest tree, then the behavioral exaptation to steer a glide might well be
common in arboreal animals. Ants are neither primitive nor ancestral to
other flying insects, but they are wingless* and nothing about their anat-
omy hints at any aerodynamic function. The ability of some ants to pre-
dictably steer a fall demonstrates that any arboreal animal, whether it has
obvious gliding surfaces or not, may have the sensory and behavioral abil-
ity to steer a glide. Moreover, additional studies have shown that a wide
variety of flightless animals use some aerodynamic behaviors (righting,
steering).29 If natural selection strongly favored longer glides, such an
animal would tend to evolve enlarged surfaces to improve lift production.
This selective force could come from predator evasion, quicker movement
to new trees, avoidance of time on the ground, or some combination.
“Parachuting” for soft landings or wind dispersal would probably not have
been one of these selective forces (contrary to arguments of earlier re-
searchers)30,31 because parachuting favors very small body size and high
drag, which work against the evolution of both gliding and flapping.10,18

A Working Hypothesis

Due to the unfortunate lack of fossils from the time that insects were
evolving flight, we have no direct evidence for the process; but thanks to

*  Worker ants, who make up the overwhelming majority of colony members, are
entirely wingless. Reproductive individuals (drones and queens) have wings before
mating; males die and females shed their wings after mating.

[ 88 ]  On the Wing

the bounded-ignorance approach and based on processes in living insects,
we can make some fairly sophisticated guesses about how insect flight
probably evolved. The direct ancestors of flying insects were probably
rather silverfish-like and probably at least a centimeter (half inch) or so
long, perhaps much bigger. They probably fed on the reproductive struc-
tures or growing tips at the tops of stalk-like plants and routinely jumped
to the ground to escape predators or to speed up travel to nearby plants.
They probably had fairly good vision (unlike modern silverfish, but like
some fossil relatives) to see approaching predators and find new plants to
feed on.
This ancient species surely had the ability to reorient and land right-
side-up during a jump or fall and may well have already evolved the ability
to steer its fall back toward its original plant or toward a new one. It may
have had lateral extensions—either fixed or movable—along its thorax
and abdomen for some non-flight function, perhaps like those of Carbo-
triplurida. If these extensions produced even a little bit of lift, they could
extend the jump into a glide; and under the right conditions, they could
evolve larger surface area and eventually become wings. If the extensions
were already hinged, flapping may have evolved almost simultaneously; if
not, leg muscles might initially have deformed the exoskeleton of the
thorax enough to move the wings a bit, even without a flexible hinge. A
hinge would have evolved eventually to allow greater wing movement for
steering and flapping. Flapping movements would have increased in
strength and complexity until the insect achieved fully powered flight.

FOSSIL EVIDENCE: EARLIEST KNOWN FLYERS

As we have seen, scientists discovered a few fossils from very primitive

insects that lived over 400 million years ago, then nothing for the next
100 million years.* Rocks from just over 300 million years ago have yielded
many fossils of insects and especially insect wings. Frustratingly, these
fossils are mostly from insects clearly capable of fully powered flight.
Thus, while primitive in many ways, they were well beyond the transition
from winglessness to powered flight.

*  Researchers recently described a fossil from an ancient swamp in what is now

Belgium that appears to be a 360-million-year-old juvenile (hence, wingless) insect
from the winged insect lineage.32 This fossil could help pin down the age of the lin-
eage but is no help in clarifying wing evolution. Other scientists, however, see more
crustacean than insect characteristics,33 and the lone specimen is not well enough
preserved to convincingly resolve this dispute.

Insec ts  [ 89 ]

 These ancient insects include a number of extinct groups that were
once widespread and common, and they also offer some clues about a
major wing innovation that separates most modern insects from their
more primitive ancestors and living sister groups (see phylogeny,
Fig. 5.4). This innovation is the neopterous wing articulation, a modifica-
tion that allows most living insects to fold their wings flat over their
abdomens at rest by flexing them backward, as in modern house flies and
bees. More primitive flying insects have a simpler wing hinge, called a
paleopterous articulation, that allows the wings to flap up and down but
not to fold flat over the abdomen. Dragonflies, for example, have paleop-
terous hinges, so they perch with the wings outstretched when at rest.
Scientists used to consider insects with the paleopterous hinge to be a
single lineage (called Paleoptera), but now generally consider them to
represent several different lineages. Insects with the neopteran wing
hinge do form a monophyletic group, according to both anatomical and
genetic evidence, so they are placed together in a taxonomic group called
Neoptera.4

Ectognatha
(including Collembola)
Ancestral Archaeognatha
crustacean (bristletails)
Zygentoma
(including silverfish)
Ephemeroptera (mayflies)
† Rhyniognatha
† Palaeodictyopterida
Pterygota
Odonatoptera
(wings evolved)
(† griffenflies,
dragonflies, damselflies)
Plecoptera (stoneflies)
Phasmatodea (stick insects)
Orthoptera
(grasshoppers, crickets)
Dermaptera (earwigs)

Blatteria (cockroaches)
Neoptera All other flying (and secondarily
(neopteran flightless) insects
wing hinge)

Figure 5.4:
Simplified phylogeny based on Grimaldi and Engel.4 (Several lineages omitted for clarity.)
A vertical bar represents the base of the indicated lineage (or first appearance of a trait
characteristic of that lineage); all subsequent branches beyond—to the right—of the bar
are part of that lineage. †: Extinct.

[ 90 ]  On the Wing

Palaeodictyopterida

The Palaeodictyopterida was a major lineage of insects with paleopter-

ous wing articulations that flourished early and then went extinct at the
end of the Permian (about 250 million years ago). 34 From the time they
first appeared, well over 300 million years ago, until the great End-
Permian Extinction, Palaeodictyopterida made up about half of all
insect fossils.4 These insects had prominent beaks that they apparently
used to pierce and feed on plant tissues. Some fossils even preserve dis-
tinct banding and other color patterns on the wings. Many of them had
a curious little third pair of wings (“prothoracic lobes” or “winglets”),
complete with veins, in front of the two normal pairs of wings (Fig. 5.5).
Scientists are divided on whether these prothoracic winglets were

Figure 5.5:
Reconstruction of paleodictyopteridan, from Kukalova-Peck, used by permission of John
Wiley & Sons.32

Insec ts  [ 91 ]

actually mobile, for steering or flapping, or fixed, perhaps as some sort
of stabilizer or protective cover.4 Scientists have identified juveniles of
palaeodictyopteridans and these all appear to be terrestrial. 35–37 Most
researchers consider the fact that these nymphs were air-breathers to be
convincing evidence that the ancestors of winged insects were terres-
trial and not aquatic.
Two other lineages of insects share the paleopterous wing articula-
tion with palaeodictyopteridans: Odonatoptera, including modern dam-
selflies and dragonflies and their extinct relatives, and Ephemeroptera,
consisting of fossil and modern mayflies. Because these three lineages
possess a primitive wing hinge and are present among the earliest
winged insect fossils, they were traditionally considered to be three
branches of a single primitive lineage called Palaeoptera. 38,39 Recently,
biologists and paleontologists have uncovered mounting evidence that
these groups represent three separate lineages that may not be particu-
larly closely related. A consensus is emerging that the ephemeropterans
are the most primitive or basal of the flying insects, the palaeodicty-
opteridan lineage split off later, and the odonatopterans were the last
group to split off from the rest of the insects before the neopteran wing
hinge evolved (Fig. 5.4).

Ephemeroptera and the Subimago

Mayflies, in the order Ephemeroptera, have long been considered among

the most primitive of flying insects. Their paleopterous wing hinge obvi-
ously places them near the base of the flying insect phylogeny, but they
have another, uniquely primitive trait: they molt once after they develop
full-sized wings. In all other winged insects, the last molt an insect ever
makes is from a flightless immature stage to a fully winged adult. In may-
flies, the last aquatic juvenile stage fills with air and floats to the surface
where it molts to the so-called subimago (subadult) stage. This stage has
full-sized wings, weak flight muscles, and undeveloped reproductive
structures. Barely capable of flight, it flutters quickly to shore, finds a
secure perch, and molts almost immediately to a fully flight-capable, fully
mature adult.
Many scientists consider the mayfly subimago stage to be primitive be-
cause insects that split off the main insect lineage before wings evolved­—
primitively wingless insects like silverfish—continue molting and growing
after reaching sexual maturity. In contrast, no flying insects other than
mayflies molt after developing functional wings. There is a very good

[ 92 ]  On the Wing

reason for this difference: in all other winged insects, the living tissue
needed to form a new wing exoskeleton dies and is reabsorbed in the wing
membrane, remaining only in the wing veins (the epidermis, see Fig. 2.4).
Without living cells in the wing membrane, the insect can’t form a new
membrane exoskeleton and thus can’t molt. In contrast, the mayfly
subimago retains living tissue in its wings; this allows molting but makes
subimago wings heavier and less rigid than a normal adult wing. Few sci-
entists were surprised when molecular genetic analyses placed mayflies at
the very base of the flying insect phylogeny. These insects apparently sepa-
rated from the rest of the winged insects before flying insects lost the abil-
ity to molt as adults.
Mayflies have a rich fossil history, with many specimens of juveniles
and adults described, including several extinct families from the Perm-
ian (over 260 million years ago). Modern mayflies have small, sometimes
nearly vestigial, hindwings, and the adults have incomplete, nonfunc-
tional digestive systems. The adult stage rarely lasts more than 24 hours,
so adults do not eat. Fossils show that some of the early mayflies had
hindwings as big as forewings and apparently functional jaws. A few of
these ancient mayflies were huge (see Box 5.3. Carboniferous Giants);40
most, however, were of modest size. By the end of the Jurassic (approxi-
mately 150 million years ago), mayflies had evolved to their modern
form, and some fossil species can even be placed into families of living
mayflies.

Odonatoptera and the Grif fenf lies

The remaining insect group with the paleopterous wing hinge is the
Odonatoptera, the group containing dragonflies and damselflies as well
as the fascinating, extinct griffenflies (formerly called “giant dragon-
flies,” a misleading name because griffenflies were a separate lineage
from true dragonflies). This lineage includes the largest known insect of
all time, the griffenfly Meganeuropsis permiana (see Box 5.3. Carbonifer-
ous Giants), with a wingspan of nearly 70 centimeters (28 inches)!
Griffenflies, particularly the big ones, are known mostly from fossils of
isolated wings or wing fragments (for example, see Fig. 5.6).41 Based on a
few partial body fossils of smaller species, griffenflies seem to have been
aerial predators, much like modern dragonflies, but their wings lack
some diagnostic features of true dragonfly wings. Dragonfly and damsel-
fly fossils first appear in the early Permian, some 10 or 20 million years
after the griffenflies.

Insec ts  [ 93 ]

 Box 5.3:  CARBONIFEROUS GIANTS

In the late Carboniferous, roughly 300 million years ago, a variety of

terrestrial animal groups evolved enormous body sizes. For example,
some amphibians reach a length of over 6 meters (20 feet). The largest
terrestrial arthropods of all time lived then, including Arthropleura, a
centipede that reached 2.6 meters (over 8 feet) in length, as well as
1-meter (3-foot) millipedes and 70-centimeter (28-inch) scorpions. The
largest known flying insects also lived during this period.
Scientists think gigantism evolved because the proportion of oxygen
in the atmosphere was higher during much of that time. 38 Some geo-
physical studies suggest that atmospheric oxygen rose as high as 35%
(compared with our current level of 21%) of the total atmosphere.
More atmospheric oxygen would have let animals distribute oxygen
to tissues over greater distances in their bodies, allowing larger body
size. This improvement would have been especially significant for in-
sects, with their diffusion-powered tracheal respiratory system.
Although most carboniferous insects were no larger than modern-
day insects, many were quite large and a few were immense. Curiously,
many of these giant flyers are known only from fossils of isolated wings
or wing fragments (for example, see Fig.  5.6), and many of these are
from the Elmo fossil bed in central Kansas, a couple of hours drive from
my office at the University of Kansas. Because researchers from Yale
and Harvard did most of the original collecting at Elmo, however, the
vast majority of the fossil insect specimens from Elmo are now housed
in museums at those universities.
These Carboniferous giants dwarf any flying insects alive today.
Some mayflies had wingspans of over 43 centimeters (17 inches) and
several species of palaeodictyopteridans had wingspans of over 50 cen-
timeters (20 inches). Griffenflies, members of the Odonatoptera, in-
clude the largest known insect of all time, the griffenfly Meganeuropsis
permiana with a wingspan of 66 to 70 centimeters (26 to 28 inches).
Some of the “smaller” Kansas griffenflies include Megatypus schucherti
with a wingspan of a mere 40 centimeters (16 inches) or Typus gracilis at
33 centimeters (13 inches), still impressive by modern standards.39

WING NUMBER

The four-winged arrangement seems to have been the original pattern

from which all other insect wing arrangements are derived. (We will
ignore the prothoracic winglets of some fossil species because they are

[ 94 ]  On the Wing

 1 cm

Figure 5.6:
Griffenfly wing traced from a fossil (courtesy of Roy J. Beckemeyer, used with permis-
sion). Inset: Tracing of a wing of Anax junius, one of the largest living dragonflies, for scale
(courtesy of S. T.).

quite small relative to the main wings, and we don’t even know for sure
whether they were mobile.) Ancient extinct mayflies, most palaeodicty-
opteridans, and both ancient and modern odonatopterans all have two
pairs of very similar or “homonomous” wings. Many neopterans,* in con-
trast, now have one pair of wings dominant and the other pair reduced
(flies, for instance), although some of the less specialized neopterans still
have two pairs of large, independently flapped wings.
Grasshoppers, for instance, are fairly unspecialized neopterans with
large front wings and hindwings. They flap the front wings and hindwings
out of phase, with the hindwings about a quarter stroke earlier than the
forewings. In other words, when the forewings are just starting their
downstroke, the hindwings are already about one-fourth of the way into
their downstroke. We have good evidence that wing muscles evolved from
leg muscles in insects—some modern insects have a pair or two of mus-
cles that they use for both walking and flying—and the three pairs of legs
also step out of phase during walking to avoid interfering with each other.
So the grasshopper flapping pattern may be close to the original primitive
pattern.
Dragonflies and damselflies have taken this pattern and amplified it. In
the simpler grasshopper pattern, the front wings and hindwings always
stay about one-quarter of a stroke out of phase, and the forewing makes
most of the stroke adjustments the grasshopper needs for steering or ac-
celeration. Not so in dragonflies. Dragonflies can shift from their more
common out-of-phase pattern to the less common simultaneous or in-
phase stroke pattern. In the out-of-phase pattern, when the forewings
start their downstroke, the hindwings are just about finished with their

* Members of Neoptera, with the neopteran (folding) wing hinge; most flying
insects are in this lineage.

Insec ts  [ 95 ]

downstroke. Moreover, forewings and hindwings are separately adjust-
able so that a dragonfly can adjust only the forewings, only the hindwings,
or both for maneuvers like steering, climbing, or diving.42 This great in-
crease in adjustability of their wingbeat pattern clearly contributes to the
spectacular maneuverability displayed by flying dragonflies. As a rule,
when aerial predators are much larger than their prey, their higher flight
speeds prevent them from turning as tightly as their prey. Dragonflies’
extreme maneuverability seems to at least partly offset this size problem
because dragonflies are highly successful aerial predators of tiny flying
insects.
This adjustability comes with a cost: the wing articulations of the fore-
wing and hindwing are basically right next to each other, and when the
forewings are doing something very different from the hindwings, their
sheer physical proximity means they tend to interfere with each other.
Dragonflies avoid at least a little of this cost by having separate forewing
and hindwing muscles attached directly to the wing articulations; most
other insects have “indirect” flight muscles that actually deform the
thorax to move the wings rather than pulling directly on each wing, and
this reinforces the mechanical linkage of the front and back wings. To
make matters worse, the wake of the air flowing over the front wing can
actually interfere with lift production of the hindwing when it passes
through the wake (which is why biplane airplanes have wings stacked one
above the other rather than one on the same level close behind the other).
Obviously, dragonflies have evolved the necessary strength and coordina-
tion to overcome this interference.
Most neopterans have evolved in a different direction where one pair
of wings becomes dominant and the other pair is either reduced or takes
on a new function (as we will see shortly). The vast majority of insects
are neopterans, and they include the insects most familiar to nonspe-
cialists, such as grasshoppers and crickets, praying mantises, cock-
roaches, termites, aphids, cicadas, bees, ants, wasps, beetles, butterflies,
mosquitoes, and house flies. The neopteran wing hinge allows insects to
flex their wings back so that the wings lie flat over the abdomen. Being
able to stow the wings out of the way means these insects can live in
nooks, crannies, and burrows that would be too cramped to inhabit with
wings held out to the side like a dragonfly or held straight up over the
back like a mayfly. We must be careful not to confuse correlation with
causation, but logic dictates that neopterans should be capable of moving
easily in spaces closed to more primitive flying insects, and neopterans
are far and away the most diverse and abundant insects. Indeed, scien-
tists have described more species of neopterans than all other species of

[ 96 ]  On the Wing

animals combined. Ecologically and evolutionarily, Neoptera includes
the most successful animals on the planet.
Flying insects have evolved along a fundamentally different trajectory
than flying vertebrates like birds or bats. First, insects did not sacrifice
any legs to gain wings, so insects have not given up any terrestrial mobil-
ity to achieve flight. Second, the original insect pattern is based on two
pairs of wings rather than the single pair used by pterosaurs, living
birds,* and bats. And finally, insects do not develop large, movable wings
until they are fully adult, that is, full-grown and sexually mature. Birds
and bats possess semi-functional wings as juveniles,43 but the young
grow very rapidly and are near adult size when they begin to fly compe-
tently, even if they are not sexually mature. Pterosaurs apparently grew
more slowly and seem to have had functional wings over a wide range of
body sizes (Chapter  8) although we have little evidence of their flight
capabilities and not all researchers are convinced that the young ones
could actually fly.

JUVENILES WITH WINGS?

Thanks to that frustrating gap in the insect fossil record, we have no direct
evidence about whether the earliest insects to achieve flapping flight had
functional wings before adulthood. The indirect evidence is, unfortu-
nately, somewhat conflicting. I have already described how mayflies, the
living representatives of the most ancient known flying insects, have a
pre-adult, fully winged stage; this could suggest that the earliest flying
insects had fully winged juveniles. On the other hand, the wings of all
other flying insects are physiologically and anatomically incapable of
being molted due to that missing epidermis in the wing membranes. And
since molting is a fundamental part of growth and development in in-
sects, if the earliest flying insects had wings like most living insects, they
could not have had functional wings before the adult stage. Moreover, at
least one scientist thinks that the mayfly subimago may actually be a spe-
cialized adaptation to ease the mayfly’s transition between an aquatic
stage and an aerial stage rather than a primitive holdover from non-flying
ancestors.44

*  Paleontologists have now discovered several species of apparently four-winged

birdlike dinosaurs (see Chapter 6), so birds may have also started out with two pairs
of wings. Unlike insects, however, these four wings incorporated the animal’s entire
inventory of legs.

Insec ts  [ 97 ]

 The nymphs of modern insects with incomplete metamorphosis have
wing pads in their later stages; these are tiny winglets immovably glued
down onto the thorax and completely nonfunctional. Some researchers
have pointed out that fossils of nymphs of ancient, extinct mayflies have
much larger wing pads, perhaps large enough to have been used for swim-
ming or perhaps even very rudimentary flapping (see Fig.  5.7).45 Some
have even argued that these nymphs are evidence for some variation on
the gill/exite theory of wing evolution. A closer look suggests otherwise.
First, the fossils do not appear to have a well-developed, flexible articu-
lation. Second, the wing pads are bent sharply backward, as is obvious in
Figure 5.7. They are not flexed backward at the hinge, as in a modern house

Figure 5.7:
Reconstruction of fossil mayfly nymph, from Kukalova-Peck, used by permission of John
Wiley & Sons.32

[ 98 ]  On the Wing

fly; instead the whole wing pad itself is curved backward. In this orienta-
tion, the wing pad would be quite useless as either a wing or a swimming
paddle. These wing pads clearly were not being used for locomotion. Most
likely, these are fossils of older nymphs nearly ready to molt to adults, and
these primitive insects had larger wing pads because the developmental
programming to allow more dramatic wing size changes between molts
had not yet evolved. In other words, large wing pads in juveniles might
have been a developmental necessity early on, and later insects then
evolved more sophisticated development so nymphs did not need to carry
around large, nonfunctional wing pads just so adults could have big wings.

FOUR WINGS VERSUS T WO

Although the primitive insect pattern is based on two pairs of wings and
we have seen how dragonflies have made a virtue of that necessity, more
recently evolved insect groups have repeatedly evolved a single domi-
nant pair of wings that do most of the aerodynamic work. We can’t go
back in time and do controlled experiments on evolution so we cannot
clearly demonstrate cause-and-effect, but we do know that independ-
ently flapping fore- and hindwings can interfere with each other aerody-
namically and that the articulations of fore- and hindwings flapping out
of phase can interfere with each other mechanically. In the majority of
flying insects, the forewings are dominant and the hindwings are re-
duced or vestigial. In beetles and earwigs, however, the forewings have
evolved into hard, protective wing covers too small to fly with, so the
hindwings are the main flight surfaces and must be folded origami-style
to fit under the much smaller wing covers. In other insects, where the
forewings rather than hindwings are dominant, wing muscles hold the
front of the smaller hindwing in contact with the back of the forewing;
the fore- and hindwings are flapped simultaneously, that is, in phase.
They function, in fact, like a single pair of wings. This pattern is used by
butterflies and a few moths. Many such insects have gone even further
and evolved attachment or locking mechanisms: hooks or bristles or
Velcro­-like structures to hold the hindwings to the front wings to form a
single surface. Bees and wasps, aphids, cicadas, and stink bugs are ex-
amples of insects that use locking mechanisms to hold front and back
wings together while flapping.
Some insects have carried dominance of the forewings to an extreme.
Dipterans or true flies—including house flies, horse flies, blow flies, bee
flies, mosquitoes, and midges—appear to have but a single pair of wings,

Insec ts  [ 99 ]

corresponding to the forewings of other insects. Upon close inspection,
we can see that flies have a tiny club-shaped appendage behind each wing,
not much bigger than a large bristle. These minute structures, called “hal-
teres,” are actually hindwings that have evolved into sensory structures.
The halteres swing up and down when the wings are flapping and have
properties similar to a gyroscope. They can detect unintentional heading
changes, as when the fly is deflected by a gust, and they control a set of
reflexes that adjusts the wing stroke pattern to correct any unintended
course deviations.46

LOSS OF FLIGHT

Scientists have named approximately 1 million species of insects, and es-

timates of the number of those yet to be named range from 2 million to 30
million. The overwhelming majority of insects can fly; yet given this vast
diversity, we should not be surprised if here and there across the insect
phylogeny we find scattered pockets of insects that have abandoned flight
in their evolution. In fact, possibly as many as 10% of the members of the
flying insect lineage have lost the ability to fly.
Some insects, such as silverfish, have no wings because they branched
off the main lineage of insects before wings evolved. Other insects, fleas
for example, evolved from winged ancestors. We have seen evidence that
tells us the silverfish are primitively wingless, but how can we tell that
fleas and many other insects are secondarily wingless, that is, descended
from winged ancestors? Some are easy: fleas have complete metamorpho-
sis (distinct larval, pupal, and adult stages) and scientists have abundant
evidence that complete metamorphosis evolved long after insects evolved
flight. For others, phylogenetic analysis leaves little doubt. Lice, which are
all wingless, have incomplete metamorphosis, but unlike silverfish, they
never molt as adults. Moreover, lice are closely related to the neopteran
order Psocoptera (book lice, most of which are winged), and so they must
clearly be neopteran. Or consider bedbugs. Bedbugs are in the family
Cimicidae, a family of a few dozen blood-feeding species, only two of
which prefer humans. The entire family is wingless, but they are clearly
nested within the order Hemiptera or true bugs, most of which are
winged. Bedbugs’ closest relatives can fly, so bedbugs clearly evolved from
flying bugs.
Given the obvious advantages of flight, why have some insects lost this
ability? Many factors could be involved, and scientists have begun to tease
out some significant patterns. We will take a more detailed look at loss of

[ 100 ]  On the Wing

flight in insects, as well as other groups, in Chapter 9. These secondarily
flightless insects, however, are little more than a drop in the bucket com-
pared to the overwhelming majority of insect species that can fly.

REFERENCES
1. P. Whalley and E. A. Jarzembowski (1981) Nature.
2. W. A. Shear, P. M. Bonamo, J. D. Grierson, et al. (1984) Science.
3. M. S. Engel and D. A. Grimaldi (2004) Nature.
4. D. A. Grimaldi and M. S. Engel (2005) Evolution of the Insects.
5. A. H. Staniczek, P. Sroka, and G. Bechly (2014) Systematic Entomology.
6. C. Gegenbaur (1870) Grundzüge der vergleichenden Anatomie.
7. C. Gegenbaur (1878) Elements of Comparative Anatomy.
8. F. Müller (1873) Jenaische Zeitschrift für Naturwissenschaft.
9. F. Müller (1875) Jenaische Zeitschrift für Naturwissenschaft.
10. R. Dudley (2000) The Biomechanics of Insect Flight: Form, Function, Evolution.
11. W. T. M. Forbes (1943) American Midland Naturalist.
12. R. D. Alexander and W. L. Brown (1963) Occasional Papers of the Museum of Zool-
ogy of the University of Michigan.
13. V. B. Wigglesworth (1973) Nature.
14. O. Béthoux and D. E. G. Briggs (2008) Systematic Entomology.
15. D. A. Grimaldi (2010) Arthropod Structure & Development.
16. M. M. Douglas (1981) Science.
17. J. W. Flower (1964) Journal of Insect Physiology.
18. J. G. Kingsolver and M. A. R. Koehl (1985) Evolution.
19. J. G. Kingsolver and M. A. R. Koehl (1994) Annual Review of Entomology.
20. R. J. Wootton and C. P. Ellington (1991) in Biomechanics in Evolution.
21. J. H. Marden and M. G. Kramer (1994) Science.
22. J. H. Marden and M. G. Kramer (1995) Nature.
23. J. H. Marden (2003) Acta Zoologica Cracoviensia.
24. M. S. Engel, S. R. Davis, and J. Prokop (2013) in Arthropod Biology and Evolution:
Molecules, Development, Morphology
25. S. P. Yanoviak and R. Dudley (2006) Journal of Experimental Biology.
26. S. P. Yanoviak, R. Dudley, and M. Kaspari (2005) Nature.
27. S. P. Yanoviak, Y. Munk, M. Kaspari, et al. (2010) Proceedings of the Royal Society
Biological Sciences Series B.
28. I. Hasenfuss (2002) Journal of Zoological Systematics and Evolutionary Research.
29. A. Jusufi, Y. Zeng, R. J. Full, et al. (2011) Integrative and Comparative Biology.
30. V. B. Wigglesworth (1963) Nature.
31. V. B. Wigglesworth (1976) in Insect Flight.
32. R. Garrouste, G. Clement, P. Nel, et al. (2012) Nature.
33. T. Hornschemeyer, J. T. Haug, O. Béthoux, et al. (2013) Nature.
34. J. Kukalova-Peck (1978) Journal of Morphology.
35. R. J. Wootton (1972) Palaeontology.
36. F. M. Carpenter (1992) in Treatise on Invertebrate Paleontology, Part R, Arthop-
oda 3–4.
37. A. J. Ross (2010) Scottish Journal of Geology.
38. W. Hennig (1981) Insect Phylogeny.

Insec ts  [ 101 ]

 9. J. Kukalova-Peck (1997) in Arthropod Relationships.
3
40. J. Kukalova-Peck (1985) Canadian Journal of Zoology.
41. R. J. Beckemeyer (2000) Kansas School Naturalist.
42. D. E. Alexander (1986) Journal of Experimental Biology.
43. K. P. Dial (2003) Science.
4 4. V. C. Maiorana (1979) Biological Journal of the Linnean Society.
45. J. Kukalova-Peck (1983) Canadian Journal of Zoology.
46. D. E. Alexander (2002) Nature’s Flyers: Birds, Insects, and the Biomechanics of
Flight.

[ 102 ]  On the Wing

CH A P TER  6

Birds
The Feathered Flyers

S ay “flying animal” and most people will probably think “bird.” Birds are
certainly the most conspicuous and probably the most familiar of
flying animals. Birds communicate with sounds, like us, and they are
warm-blooded, like us, and many familiar birds (especially males) have
showy, visually striking plumage. Humans often seem to feel a stronger
affinity to birds than to other mammals, even though we are much more
distantly related to the birds. Birdwatching is both a popular individual
hobby and the focus of numerous regional and national organizations, but
I have yet to hear of anyone going “rodent-watching” or joining a “mammal-­
watching” society. This affinity for birds may explain why more effort has
gone into studying the evolution of bird flight than the flight of all other
animals combined.

FIRST FOSSILS

Any discussion of the fossil history of birds logically begins with Archaeop-
teryx, the classic “missing link” fossil between birds and dinosaurs. (Indeed,
researchers who study insect or bat evolution often lament the lack of an
Archaeopteryx-like fossil for those groups.)1,2 Several magnificent speci-
mens of Archaeopteryx were found in the Solnhofen limestone of Germany
as it was being quarried for use in making lithographic prints. The smooth,
fine-grained texture of Solnhofen limestone that makes it so highly prized
for lithography also makes it an excellent medium for preserving impres-
sions of delicate soft tissue structures of entombed organisms. Two of the
Figure 6.1:
The “Berlin” Archaeopteryx fossil, housed in the Humboldt Museum für Naturkunde in
Berlin. These animals are smaller than most people expect; this specimen was just slightly
bigger than a pigeon, with a body length of about 23 centimeters (9 inches) not counting
the long tail. (Photo courtesy of David A. Burnham, used by permission.)

original Archaeopteryx specimens, the “London” specimen discovered in

1861 and the “Berlin” specimen found in 1877, clearly preserved extensive
birdlike plumage, including unmistakably aerodynamic wing feathers (see
Fig. 6.1).3,4 Yet the body was clearly reptilian, with teeth, long separate fin-
gers with claws, and a long, whip-like bony tail (albeit covered with feath-
ers). Coming hard on the heels of Darwin’s publication of his theory of
evolution in 1860,5 these fossils must have seemed tailor-made to support
his new concept. Over the years, several more specimens of Archaeopteryx
have been found, and although most have feather impressions, none show
such clear and detailed feathers as those first two.6
For nearly a century and a half, biologists and paleontologists ac-
cepted Archaeopteryx as the oldest and most primitive bird known, having
lived about 160 million years ago. Its feathers were so birdlike and its
wings were so clearly aerodynamic that to suggest that Archaeopteryx was

[ 104 ]  On the Wing

not at the base of the bird family tree was to invite ridicule. Scientists
generally agreed that if Archaeopteryx itself was not ancestral to modern
birds, it was certainly a close relative to that ancestor. In the past, some
scientists have questioned whether Archaeopteryx was a powered flyer or
whether it could even glide,7 although most now view it as at least a weak
powered flyer. Recent fossil finds in China have complicated our view of
the relationships among ancient birds and other dinosaurs, sometimes
displacing Archaeopteryx, as we’ll see in this chapter. The iconic status of
Archaeopteryx, however, makes researchers who move it anywhere out-
side the bird lineage on a phylogeny very careful to fully justify that
placement.

DESCENDED FROM DINOSAURS

Almost from the beginning of the modern scientific study of evolution,

scientists linked birds with dinosaurs. After the spectacular London and
Berlin specimens of Archaeopteryx were discovered and widely publi-
cized, scientists quickly noticed that this species had birdlike feathers on
a body with many dinosaurian features. For decades, most scientists
took Archaeopteryx as clear evidence that birds evolved from dinosaurs.
In the 1920s, this belief took a curious detour. Danish artist Gerhard
Heilmann wrote some articles on bird evolution that were published in
book form (in Danish) in 1916 and largely ignored. In 1926, he published
an English translation, The Origin of Birds, which became extremely influ-
ential. In it, he said that, yes, Archaeopteryx does seem to have a lot of
dinosaur-like characteristics, but birds all have a furcula (“wishbone”)
formed from modified clavicles or collarbones, and dinosaurs have no
clavicles. He thus reluctantly concluded that birds could not have de-
scended directly from dinosaurs but must instead share with them some
more-ancient common ancestor.8 Heilmann seems to have stated his
view so carefully and forcefully that “birds descended from dinosaurs”
quickly became a minority view.
Heilmann’s book prompted researchers to propose possible ancestors
for birds other than dinosaurs. Usually these were “thecodonts,” a collec-
tion of ancient, primitive, dinosaur-like reptiles. At the time, paleontolo-
gists considered thecodonts to be a group of closely related animals
ancestral to dinosaurs, although we now view them as a miscellaneous
collection of not-so-closely-related animals, of which one or two may be
related to dinosaur ancestors. Since Archaeopteryx was so much like a dino-
saur, looking for the common ancestor of birds and dinosaurs among

Bir ds  [ 105 ]

thecodonts largely amounted to looking for the closest relatives of dino-
saurs in that group.
Other researchers proposed that birds arose from the crocodilian lin-
eage. Modern crocodiles and their relatives might seem an unlikely
source for bird ancestors, but proponents made reasonable arguments.
First, crocodiles, birds, dinosaurs, and pterosaurs make up the lineage
Archosauria, and at the time almost nothing definitive was known
about who was related to whom among the four archosaurian groups.
Moreover, not all of the early crocodilians were the long, sprawling,
semi-aquatic beasts of today. The most ancient crocodilians were much
more diverse and included bipedal runners that some scientists thought
may even have been able to climb trees.9 Archaeopteryx teeth also seemed
to be more crocodilian than dinosaurian, at least based on the dino-
saurs known in the mid-20th century. Although never a majority view,
the crocodilian-bird-ancestry theory was advocated by some respected
scientists.10
In the 1970s and 1980s, several events shifted views of bird ancestry
back toward dinosaurs. First, paleontologists started using modern phylo-
genetic methods (Chapter 1), which uniformly showed a close relationship
among dinosaurs, Archaeopteryx, and birds. Second, eminent Yale paleon-
tologist John Ostrom had his epiphany linking Archaeopteryx to small
theropod (meat-eating) dinosaurs (we will look in detail at Ostrom’s im-
portant insight later). And, not least, researchers found clavicles (and
even furculas) on more complete specimens of dinosaur fossils.11,12 The
evidence has become quite strong that birds descended from theropod di-
nosaurs. Specifically, birds descended from a diverse lineage of theropod
dinosaurs called Maniraptora. Figure 6.2 illustrates how Maniraptora and
other groups important to bird evolution form subgroups within dino-
saurs. Maniraptora includes predatory dinosaurs like the mollusk-eating
Oviraptor and the weird, short-armed Mononykus, as well as those with the
enlarged, retractable toe-claw such as Velociraptor and Deinonychus (see
Box 6.1).
Scientists argue about whether Archaeopteryx should be considered a
bird or a birdlike dinosaur, but most recent phylogenies place Archaeop-
teryx on a side branch off the very base of the bird lineage (see Fig. 6.2), so
we will call Archaeopteryx a basal bird. Nowadays, paleontologists call
this bird lineage of the maniraptoran family tree “Avialae” (“bird wing”).
It was originally defined as all feathered animals capable of flapping
flight13 so as to distinguish it from Aves, the traditional classification for
the living birds. (Avialae thus includes all living birds and extinct birds
back to Archaeopteryx­—and arguably a couple of species even older than

[ 106 ]  On the Wing

 dinosaurs

theropods
maniraptorans

dromaeosaurids
avialans

troodontids

Ornithischia
non-theropod Saurischia
Ancestral dinosaur Tyrannosaurus
Compsognathus
Dromaeosauridae
Dienonychus
Microraptor
Theropoda Troodontidae
Maniraptora Archaeopteryx
Enantiornithes (extinct
“opposite birds”)
Avialae
Ornithurae (including
modern birds)

Figure 6.2:
(Top) This Venn diagram names many of the dinosaur subgroups close to the bird lineage
and shows how they are nested. (Bottom) A phylogeny demonstrating the same relation-
ships. (Courtesy of S. T.)
 Box 6.1:  THE “REAL” VELOCIRAPTOR

Scientists know of Velociraptor from several fossils found in and near

Mongolia starting in the 1920s. It is very similar to Deinonychus, found
in North America and described in the 1960s. Both were lightly built,
agile predators with the famous, enlarged “killing” toe claw; Deinony-
chus actually means “terrible claw.” (Scientists disagree on whether the
enlarged claw was actually for killing prey; biomechanical studies sug-
gest it may have been used to hold down smaller prey or in defense
rather than in slashing attacks.) The public knows of Velociraptor from
the “Jurassic Park” book and movies, which, however, were not entirely
accurate. At less than 1 meter (about 2 or 3 feet) tall, Velociraptor would
not have been a very menacing villain. The movie “velociraptors” seem
to have been mostly based on the much larger (and scarier) Deinonychus.
Thanks to the movies’ terminology, however, “velociraptor” has entered
the general lexicon whereas few people other than scientists know of
Deinonychus.

Archaeopteryx.) Since Avialae is a lineage that evolved within theropod

dinosaurs, then phylogenetically birds are dinosaurs; this is the reason
we must say “non-bird dinosaurs” or “non-avialan theropods” to refer to
groups we traditionally consider dinosaurs. For the same reason, when
paleontologists are asked why dinosaurs went extinct, they tend to
answer that dinosaurs are not extinct: today’s dinosaurs are just a profu-
sion of small, feathered animals we know as birds.

THE ORIGIN OF FL APPING FLIGHT IN BIRDS

Two Traditional Theories

In an odd parallel to the situation in insects, scientists have been arguing

for more than a century over two theories that explain how bird flight
evolved. One theory starts with an arboreal (tree-climbing) protobird
that then evolves from leaping, through a gliding stage, then through
steering and weak flapping, to finally achieving fully powered flight. This
is essentially the same theory that Darwin originally proposed for the
evolution of flight in bats, and scientists call it the “arboreal” or “trees-
down” theory. The other theory starts with a fast, bipedal runner. This

[ 108 ]  On the Wing

runner evolves aerodynamic surfaces on its arms, perhaps to aid in bal-
ance or to stabilize turns while running. These surfaces enlarge and allow
a bit of gliding during leaps. Flapping would have evolved initially for
thrust only—to increase running speed. Flapping in time became strong
enough to turn leaps into short flights and eventually evolved into fully
powered flight. Scientists call this the “cursorial” or “ground-up” theory.
Some proponents of the cursorial theory see no need for a gliding stage at
all: they see flapping for thrust (to increase running speed) as sufficiently
beneficial by itself that flapping could have evolved directly. In that form,
the cursorial theory achieves powered flight in almost the same way as
Marden’s theory of surface-­skimming stoneflies (Chapter 5).
One problem with the arguments over these theories is that research-
ers kept trying to use evidence from Archaeopteryx to support one or the
other. Supporters of the cursorial theory would say, “Archaeopteryx has
features a and b, making it a runner, so flight must have evolved from the
ground up.” Then supporters of the arboreal theory would say, “Archaeop-
teryx has features x and y, making it a climber, so flight must have evolved
from the trees down.” Both sides missed the point that Archaeopteryx was
past the evolutionary stage where being arboreal or cursorial mattered.
Archaeopteryx wasn’t in the process of evolving flight, it was already a flyer.
It may not have been the most powerful, agile flyer, but its wing shape and
wing loading are well within the range of modern birds,* and it appears to
have been capable of at least some powered flight.14 The bottom line is that
Archaeopteryx is very important for unraveling the phylogenetic relation-
ships of birds and their relatives (which can help researchers figure out
which characteristics were inherited from ancestors), but its flight abili-
ties are advanced enough that it may not tell us much about the begin-
nings of flight in birds.

A Misleading Linkage

Another problem was with researchers’ tendency to mix up ancestry with

flight evolution. Yale paleontologist John Ostrom made discoveries that
clarified the phylogenetic position of Archaeopteryx but that also misdi-
rected flight evolution research. Ostrom had discovered and described
Deinonychus, so he was intimately familiar with the dinosaurs that would
later be called maniraptorans. He then turned his attention to pterosaurs.

*  Wing loading is the lift per unit surface area of a wing and relates wing size to
body weight.

Bir ds  [ 109 ]

He traveled to the Netherlands to visit the Teyler Museum to see a ptero-
saur fossil that had been described a few years before Archaeopteryx in the
1800s. As soon as he looked at it, he knew the fossil was not a pterosaur,
and on closer inspection he discovered faint feather impressions: it was
actually an unrecognized fossil of Archaeopteryx!6 Ostrom shifted his re-
search focus toward Archaeopteryx, and the more he looked at Archaeop-
teryx fossils, the more he saw similarities to Deinonychus. In the 1970s,
Ostrom pointed out these similarities,15,16 and other paleontologists began
building phylogenies that clearly placed Archaeopteryx in or near the Dei-
nonychus lineage.13 To these paleontologists, Archaeopteryx—and by ex-
tension all birds—had now become a subgroup of dinosaurs.
The misdirection came from combining the new phylogenies with a
view of Archaeopteryx as if it were just beginning to evolve flight and an
incomplete knowledge of Maniraptora. These factors led most paleontolo-
gists to argue that Archaeopteryx is clearly a close relative of theropod di-
nosaurs, and since theropod dinosaurs are bipedal runners and not tree
climbers, flight must have evolved from the ground up.
For several decades before Ostrom looked at Archaeopteryx, the arbo-
real theory seems to have been more widely accepted.17,18 After Ostrom
revived the cursorial view, two groups of scientists remained uncon-
vinced. One group included many ornithologists, who considered Archae-
opteryx to be a bird and who were skeptical of claims of close kinship
between Archaeopteryx and theropod dinosaurs.19 The opinion of these
researchers was essentially a mirror image of the view put forth by
Ostrom and the dinosaur paleontologists: birds evolved from some line-
age much more ancient than Maniraptora; Archaeopteryx was not a runner
so it could not have descended from theropods; and since bird ancestors
were climbers (presumably arboreal), flight must have evolved from the
trees down.
The arguments between the cursorial supporters and the arboreal sup-
porters,* although couched in dispassionate scientific terms, tended to
sound a bit like “whatever they are for, I’m against.” In hindsight, the “not
from theropods/trees down” proponents were also misdirected. Because
they favored an arboreal origin for flight (and they viewed Archaeopteryx as
a bird and not a dinosaur), they felt compelled to oppose a close theropod-­
bird relationship.

*  In addition to Ostrom, prominent supporters of the “theropod ancestry/ground

up” position include UC-Berkeley paleontologist Kevin Padian and his former stu-
dent, Jaques Gautier. On the other side, University of North Carolina ornithologist
Alan Feduccia and my University of Kansas colleague, the late Larry Martin, champi-
oned the “more ancient ancestry/trees down” view.

[ 110 ]  On the Wing

Does Physics Matter?

Yet a third group of researchers was drawn into this debate. These were the
flight biomechanists, scientists studying the aerodynamics and func-
tional anatomy of bird flight (full disclosure: my own research falls closest
to this group). Prominent members of this group included British biologist
Jeremy Rayner and Swedish biologist Ulla Norberg. These researchers fo-
cused on the physics of flight and were much less concerned about phylo-
genetic relationships. They pointed out (repeatedly) that from a physical
standpoint, the arboreal model makes much more sense than the curso-
rial model.14,20–23 The fundamental point is that if you start out gliding
from an elevated perch, you are working with gravity and you gain all the
advantages of gliding (Chapter 4). In contrast, Rayner analyzed the phys-
ics and showed that Archaeopteryx-sized animals cannot run fast enough
to gain any benefit from leaping and gliding.24
Proponents of the ground-up model responded that flapping could have
evolved directly, skipping a gliding stage, to produce thrust for faster run-
ning.25 Using flapping wings to run faster is certainly possible—chickens
do it all the time—but evolving wings just for that purpose seems unlikely.
First, protowings would have to get fairly big and flapping muscles would
have to get pretty strong before they could produce enough extra thrust to
help creatures run faster. This is an especially large hurdle to overcome be-
cause simply evolving longer legs can do the same thing. Second, not a
single animal, living or extinct, is known that uses flapping appendages to
run faster on the ground except those descended from flapping flyers.
Indeed, flightless cursorial birds like ostriches and emus are excellent run-
ners and are even descended from flapping flyers, yet they do not use their
wings to increase running speed; their wings are essentially vestigial. In
the early 1990s, these three groups of scientists seemed to be at an im-
passe. No one could marshal enough evidence to sway members of the
other groups. Then the first of what became a flood of new fossils appeared.
These new fossils have forced researchers to re-evaluate their arguments.

EXPLOSION OF FOSSILS FROM CHINA

The fossil beds in Liaoning Province of northern China have produced a

wealth of fossil mammals, insects, plants, extinct birds, and pterosaurs
from the early Cretaceous, about 150 million years ago. They are best
known to both scientists and the general public, however, for the abun-
dant dinosaur fossils they have produced.

Bir ds  [ 111 ]

 As political conditions for scientific research and collaboration in China
improved in the 1980s and 1990s, paleontologists began describing a va-
riety of very well-preserved dinosaurs from the Liaoning beds. Among
these were theropod dinosaurs that showed a close relationship to Dei-
nonychus and Velociraptor (including the famous enlarged toe-claw), plac-
ing them in the Maniraptora. These dinosaurs gave researchers more data
to refine their phylogenetic trees, and the more detailed the trees became,
the stronger the evidence looked for placing Archaeopteryx in that group
as well.
Like the Solnhofen limestone, the fine-grained texture of the rock of
these Chinese fossil beds often preserves extremely fine detail, including
what appear to be hair-like or feathery external structures. Some of the
first fossils to show such structures were specimens of Confuciusornis sanc-
tus, a very primitive beaked bird. These sparrow-sized birds are extremely
common among Liaoning fossils, often including preserved feathers. They
are remarkable because a few have a pair of very long tail feathers whereas
most have only very short tail feathers. Although the feathers were pre-
served in very fine detail, no one was surprised by the presence of feath-
ers; these were birds, after all.

Feathered Dinosaurs?

In 1998, a team of Chinese and North American scientists reported a

startling find: two different dinosaur fossils that both showed what ap-
peared to be feathers.26 Protarchaeopteryx and Caudipteryx were both
small (goose-sized) maniraptoran theropod dinosaurs.* Both fossils had
clear impressions of what appeared to be different types of feathers:
short, fluffy, filamentous or down-like feathers and longer “pennaceous”
feathers. Pennaceous feathers are what we usually envision when we
think of feathers: a central shaft with a flat array of interlocking barbs on
each side forming a vane or blade. Protarchaeopteryx had a fringe of fluffy
feathers along its chest, legs, and tail, plus several pennaceous feathers
on its tail. The Caudipteryx fossil, on the other hand, showed a well-­
developed row of obvious pennaceous feathers on each hand and on its
tail (in addition to a fringe of downy feathers surrounding its body).
Other than their small size—and feathers—these dinosaurs were fairly
typical bipedal, carnivorous, maniraptoran dinosaurs. We’ll see that their

*  Recent phylogenies place them in the ovoraptorids, one of several subgroups of

Maniraptora.

[ 112 ]  On the Wing

feathers, though pennaceous, did not have the right shape or arrange-
ment to form a wing, so they certainly could not fly (or even glide effec-
tively). Aside from their feathers, they were no more birdlike than many
other maniraptoran dinosaurs.
When earlier researchers had described hair-like fibers fringing the
skeleton of a Liaoning dinosaur as protofeathers,27 other scientists dis-
missed them as tendons or skin fibers that only appeared to be external
due to fossilization processes.28,29 With the discovery of unmistakable
pennaceous feathers on Caudipteryx, a close relationship between mani-
raptoran dinosaurs and birds became unmistakable. Researchers devel-
oped phylogenies showing relationships among maniraptoran dinosaurs,
Archaeopteryx, and other birds, and these required constant updating as
new fossil discoveries from Liaoning steadily increased. The paleontolo-
gists who described Caudipteryx nearly two decades ago would probably be
amazed at the huge increase in known maniraptoran dinosaur species as
well as the number of those dinosaur species that show signs of feathers.
For example, with only minimal effort, I tracked down information on
over two dozen non-bird, maniraptoran dinosaurs described after Caudip-
teryx. Guided by a recent phylogenetic analysis of over 100 avialans (bird
lineage animals) and non-avialan theropods,30 I counted at least 10 with
apparent down-like feathers and at least 10 more with unmistakable pen-
naceous feathers. We now know of a number of theropods that had downy
filaments and are only distantly related to maniraptorans, including two
species closely related to Tyrannosaurus rex!

Four Wings?

Among the many startling fossils from Liaonning, perhaps the most re-
markable is Microraptor gui (Fig. 6.3). Microraptor was a small (1-kilogram
or 2-pound) maniraptoran dinosaur with long, asymmetrical pennaceous
flight feathers on both its arms and legs. This peculiar beast appears to
have been a four-winged flyer. Modern flying birds have asymmetrical pri-
mary (pinion) feathers making up the wingtip, with the shaft closer to the
front edge of the blade rather than in the center (Fig. 6.4). In flight, this
asymmetry stabilizes the wingtip and also lets each primary feather act as
a tiny separate wing when the primaries are spread out during parts of the
wing stroke.31 In contrast, Caudipteryx’s arm feathers were symmetrical,
so they certainly were not acting as pinion feathers (Fig. 6.4). The asym-
metrical feathers of Microraptor, however, are convincing evidence that
both its arms and its legs functioned as wings. Although possessing four

Bir ds  [ 113 ]

Figure 6.3:
Microraptor gui reconstructed as a four-winged glider. (Courtesy of S. T.)

A B C

Figure 6.4:
Primary (pinion) feathers from different birds. A. Archaeopteryx. B. Modern flying bird
(pigeon). C. Emu, a flightless bird. The primary feathers from flying birds are always
asymmetrical, with the central shaft of the feather closer to the front of the wing. When
birds lose the ability to fly, the feathers tend to become more symmetrical, like the sec-
ondary feathers on the back of the wing and the contour feathers that cover the body.
(From Alexander.)29
well-developed wings, Microraptor was apparently a glider and not a fully
powered flyer. It did not have the shoulder joint or chest muscles for well-
developed flapping, although it may have used very weak flapping to
extend glides.
Microraptor is significant in several ways. It is clearly a maniraptoran
dinosaur, and it is just as clearly arboreal. In fact, the feathers on its feet
were about 18 centimeters (7 inches) long, on a leg of only 26 or 28 centi-
meters (10 or 11 inches) from hip to toes, so I have a hard time imagining
how it could walk on the ground at all; it was clearly not a bipedal runner.
The existence of Microraptor means that the proponents of the cursorial
model of flight evolution can no longer say “birds evolved from theropods,
and all theropods are bipedal runners, so birds must have evolved flight
from the ground up.” Microraptor provides solid evidence that theropods
could be arboreal and could evolve gliding.
Microraptor lived a few million years after Archaeopteryx, so cannot be
directly ancestral to birds. Nevertheless, Microraptor shows that thero-
pods could evolve an arboreal lifestyle and could evolve gliding from the
trees down. Some scientists consider Microraptor’s front wings to be so
birdlike that birds and Microraptor must have inherited these wings from
a common ancestor. Others doubt that birds and Microraptor share a
common gliding ancestor, but they see Microraptor as establishing a possi-
bility: the possibility that a dinosaur similar to Microraptor evolved gliding
much earlier, and that gliding dinosaur gave rise to Archaeopteryx and the
rest of the bird lineage.
One recent phylogenetic study actually supports a gliding, four-winged
common ancestor for Microraptor and birds.30 Figure 6.5 shows a simplified

Tyrannosaurus
Ancestral Compsognathus
theropod
Epidendrosaurus
Dromaeosauridae
Troodontidae
Maniraptora Aurornis
Anchiornis
Archaeopteryx
Avialae Confuciusornis
Enantiornithes
Ornithurae

Figure 6.5:
Part of the phylogeny of Godefroit et al., redrawn and with some groups omitted for clar-
ity. 30 All these groups are extinct except for one lineage of Ornithurae leading to modern
birds.

Bir ds  [ 115 ]

version of their phylogeny. Many of the more primitive members of the
Avialae, including Anchiornis, Archaeopteryx, Sapeornis, Confuciusornis, and
some enantiornithines, had long leg feathers, again demonstrating a link
to a Microraptor-like body plan.32

ARCHAEOPTERYX DETHRONED

Quite a number of feathered theropod dinosaurs are now known, includ-

ing both distant and close relatives of avialans. Paleontologists have de-
scribed several feathered maniraptorans in recent years that would
unquestionably have been considered typical dinosaurs if they had not
possessed feathers. How do these feathered and birdlike dinosaurs affect
our view of bird ancestry? These finds have forced scientists to keep
modifying their phylogenies. Sometimes these modified phylogenies
move Archaeopteryx out of Avialae33 while other studies put it back into
the avialan lineage, 34 usually at the base. Moving Archaeopteryx out of
Avialae has profound implications: either flapping flight evolved twice
(once in Archaeopteryx and once in Avialae), or all the maniraptoran di-
nosaurs close to Avialae (Velociraptor, Deinonychus, other dromaeosau-
rids and troodontids) are secondarily flightless. Most scientists would
consider either of these possibilities as very unlikely (although at least a
couple support the latter), 35 so the strong consensus is that flight evolved
only once in the bird lineage, a view that keeps Archaeopteryx within
Avialae.
Recent studies challenge the First Bird status of Archaeopteryx, how-
ever. Pascal Godefroit and colleagues published a description of another
feathered maniraptoran, Aurornis xui. They combined this description
with by far the most detailed phylogeny ever developed for theropods,
including dozens of birdlike dinosaurs and primitive extinct birds—
more than 100 total species. 30 In their most likely phylogeny, both Au-
rornis and another well-feathered maniraptoran, Anchiornis huxleyi,
appeared at the base of Avialae, even further back in time than Archae-
opteryx (Fig. 6.5 shows part of their phylogeny). Both these newcomers
had long, flat feathers on their arms and legs, although the feathers are
not as aerodynamically specialized as Archaeopteryx wing feathers,* so
they were probably gliders or at best very weak flappers. Anchiornis was

*  Auronis and Anchiornis wing feathers are symmetrical rather than asymmetrical
like those of Archaeopteryx.

[ 116 ]  On the Wing

originally described as a troodontid, a group of large-brained, feathered
maniraptorans generally considered to be the closest non-bird relatives
of Avialae (Troodontidae, Fig. 6.5), 36 so even from the beginning, it was
linked with birds. If Godefroit and company’s analysis withstands the
test of time, Aurornis would displace Archaeopteryx at the base of the
lineage leading to birds. The large number of species and enormous
number of anatomical features used in the study by Godefroit and col-
leagues give it good credibility, but only time will tell whether other sci-
entists accept it and whether it remains consistent with future fossil
discoveries.
Although a handful of scientists are still skeptical that the fluffy fila-
ments on many Liaoning dinosaur fossils are really feathers, 37 the fos-
sils showing pennaceous feathers seem to me to be unassailable
evidence that a wide variety of non-bird theropods had feathers. The
presence of unmistakable feathers on non-flying species like Protarchae-
opteryx, Caudipteryx, and several other maniraptorans decisively an-
swers one long-standing question in the evolution of bird flight: did
feathers evolve specifically as an aerodynamic adaptation, or did they
evolve initially for some other function and later get co-opted for flight?
We now know that feathers evolved long before flight. Moreover, if
those filamentous tufts on a wide diversity of theropods turn out to be
protofeathers, that implies that most theropods may have been feath-
ered. Theropods seem to have been active carnivores, and many, espe-
cially the later ones, may have been partly or completely endothermic
(warm-blooded), so the filamentous protofeathers may have evolved as
insulation to help retain heat.
The temperature-regulation question is still rather murky, however, be-
cause body size matters greatly, and theropods include dinosaurs with a
huge range of body sizes, from the 6-ton Tyrannosaurus and 1½-ton Gigan-
toraptor to the 2- to 4 ½-kilogram (5- to 10-pound) Bambiraptor and ½-kil-
ogram (1-pound) Eosinopteryx. We will return to this question when we
look at physiological adaptations for flight. For now, suffice it to say that
insulation can be very beneficial for small endothermic animals; as ani-
mals get big, however, its value decreases, and for really big animals like
Tyrannosaurus, insulation has no physiological benefit at all. In other
words, Tyrannosaurus would not have needed insulation to maintain a
high body temperature. Yet two of its closest relatives, Dilong and Yutyran-
nus show traces of filamentous feathers, and the latter was also huge. If all
these filaments on theropod dinosaurs really are protofeathers, either
they have some function other than insulation or their function is differ-
ent in really small and really big dinosaurs.

Bir ds  [ 117 ]

NEW EVIDENCE, NEW VIEWS
Pouncing Proavis

In 1999, British researchers Joseph Garner, Graham Taylor, and Adrian

Thomas published one of the more innovative attempts to combine phy-
logenies and biomechanics. Their study set out to match the phylogeny of
then-known feathered dinosaurs and birds with the order they expected
to see various features evolve in either the arboreal or the cursorial
models.38 At the time, these scientists knew about Protarchaeopteryx and
Caudipteryx, but discovery of small or feathered theropods like Bambirap-
tor, Microraptor, and Eosinopteryx lay in the future. The phylogeny of Garner
and company led them to the conclusion that neither the arboreal nor the
cursorial models exactly fit the order in which features appeared on their
phylogenetic tree. For example, they described how Caudipteryx’s feath-
ered hand fit better with a version of the cursorial theory, whereas Archae-
opteryx’s long bony tail and dinosaurian hips fit better with the arboreal
model.
They proposed instead that the ancestor of birds was a predator that
specialized in pouncing on prey from elevated perches. If such predators
mainly used their feet to attack prey (with that enlarged claw), their fore-
limbs could be used for balance and to help steer during the leap. At first,
aerodynamic enhancement of the hands would have helped steer during
leaps; then it could provide a bit of lift for “turning a pounce into a
swoop,”38 when lift production increased as a byproduct of more effective
steering. Caudipteryx fits nicely into this scenario because the feathers on
its hands are not well shaped or correctly oriented for flapping or gliding,*
but they could be quite effective for steering in mid-air. Once these feath-
ered “control surfaces” started to provide lift and extend leaps, selection
for longer swoops would lead to more effective wings, then glides, and
eventually flapping. They called their model the “pouncing proavis.”
When I first read the article by Garner and colleagues, I thought, “At last,
here is a fresh idea that can help bridge the gap between the arboreal and
cursorial camps.” In fact, the pouncing proavis theory has been largely ig-
nored, and I was chagrinned to see essentially the same old cursorial and
arboreal arguments being rehashed in the scientific literature long after the
Garner study was published. Apparently, the more outspoken members of

*  Caudipteryx’s hand feathers are symmetrical so they would not have been effec-
tive as primary (pinion) feathers extending to the side. The only aerodynamically
stable orientation for them would have been straight back, parallel to the air flow.
This orientation would make them ineffective as wings but very effective as control
flaps, exactly like airplane ailerons.

[ 118 ]  On the Wing

both sides have such entrenched opinions that they are not very open to
conflicting views (a distinctly unscientific attitude).

Wing-Assisted Incline Running

Ornithologist and flight researcher Kenneth Dial has also tried to bring a
fresh idea to the debate. Like James Marden and his stoneflies, Dial ob-
served a behavior in living birds that he decided might be a useful model
for the evolution of flapping flight. He noticed that young partridge chicks
flap their poorly developed wings when they run up steep slopes. More-
over, they can climb steeper slopes when flapping than when not flapping,
and they can climb near-vertical slopes by flapping, even before their
wings are big enough for powered flight. Dial dubbed this ability “wing-
assisted incline running” or WAIR. He studied WAIR in partridges as they
developed from the time of hatching through to fledging, and he also did
feather-clipping experiments to adjust wing sizes.39 He showed that birds
could ascend increasingly steeper slopes as their wings got bigger, and he
also showed in other experiments that much of the flapping force was di-
rected toward the inclined surface. In other words, WAIR pushes the bird’s
body against the surface to increase its traction; this requires the flapping
wings to produce a downward force component rather than the net upward
force they produce during flight.
Dial suggested that flapping flight could have arisen if a small theropod
with “incipiently feathered forelimbs”39 (similar to Caudipteryx) gained an
advantage avoiding predators or pursuing prey if it could better run up
steep slopes. Flapping would thus have evolved at first for improving trac-
tion rather than for faster running or for flight. If the protobird also
jumped to get down from whatever it ran up, flapping could have also ex-
tended its leaps. Once such flapping evolved, converting it to full-fledged
flight would just be a matter of enlarging the wings and refining the stroke
pattern. Dial and his colleagues have shown how animals can benefit from
flapping a small protowing40 and have analyzed the physics of WAIR in
detail.41,42
One strong point of WAIR is that even animals with small, poorly
feathered protowings can get some advantage from flapping. Also, WAIR
starts directly with flapping and bypasses a gliding stage, which some re-
searchers see as a persuasive simplification. Moreover, it sidesteps the tra-
ditional controversy because it would work just as well for arboreal or
cursorial animals, although it might fit bipedal cursors a bit better. A sig-
nificant weakness is that WAIR is based on the behavior of animals with

Bir ds  [ 119 ]

shoulders already highly specialized for flapping. Yes, hatchling partridges
have small wings with poorly developed feathers, and yes, they can still do
something useful by flapping, but their flapping movements use the
highly specialized shoulder joint and control system (brain, senses, re-
flexes) of a fully aerial species. Dial’s research group showed that in less
aerially skilled birds, like quail, the wingbeat pattern does not change
very much between WAIR, level flight, and aerial descent.43 They suggest
that if such a “fundamental stroke” evolved first for WAIR, it could have
led to powered flight just by increasing wing size and changing body
orientation.
Whether a protobird with an unspecialized shoulder joint and no flap-
ping reflexes could have evolved such a fundamental stroke remains an
open question. As with the cursorial flapping-for-thrust scenario, we
know of no animals that use WAIR except those that are already flapping
flyers, which does not disprove the theory but does show that WAIR is
much more difficult to evolve than gliding.

Directed Aerial Descent

Although originally described in ants, the concept of directed aerial de-

scent that we saw in Chapters 3 and 4 may have a bearing on flight evolu-
tion in vertebrates as well.44 What if arboreal vertebrates evolved skydiving
behavior to steer descents or land upright before obvious aerodynamic
adaptations appeared? Some arboreal lizards—anoles, for example—use
a characteristic spread-eagled posture in falls and make soft, controlled
landings even in falls from great heights.45 (Although he has not published
a formal study, a colleague of mine has seen mice do the same thing when
falling from a great height.) Given that Microraptor was certainly arboreal,
and even smaller, feathered, non-avialan theropods like Eosinopteryx
probably were as well, a trees-down origin for bird flight is no longer in-
compatible with phylogenies showing birds derived from theropods. Al-
though this idea is not yet mainstream among dinosaur paleontologists,
the flight mechanics researchers among my colleagues are all relieved that
the more biomechanically reasonable arboreal model can be reconciled
with the phylogeny of birds.
Some dinosaur researchers might disagree, but I think that objections
to an arboreal origin of bird flight have been laid to rest. The objection that
theropods were too big to have climbed trees can no longer be taken seri-
ously. Compsognathus, also found in Solnhofen limestone, is only slightly
bigger than Archaeopteryx, and we now know of several maniraptoran

[ 120 ]  On the Wing

theropods that were even smaller (for example, Microraptor and Eosinop-
teryx). The more serious objection that all theropods were bipedal runners
and not tree climbers is no longer true, now that we know about Microrap-
tor and other similar dinosaurs. I am quite satisfied with the idea of an ar-
boreal evolution of flapping flight occurring in maniraptoran dinosaurs.

PHYSIOLOGICAL ADAPTATIONS FOR FLIGHT

Birds, like us, are endothermic or “warm-blooded.”* Endotherms use a

high metabolic rate—essentially a greatly speeded-up physiology—and
insulating mechanisms to retain heat and produce an elevated body tem-
perature, even in cold environments. Endothermy comes with a high
price; an endotherm typically needs to eat about 10 times as much as an
ectothermic (“cold-blooded”) animal of the same size. What is the payoff?
Physiologists have studied and debated this question for many decades,
but most now agree that the main advantage is a great increase in aerobic
scope, which means endotherms have much greater endurance, stamina,
and continuous muscle power output.46,47 Living reptiles like lizards are
ectothermic and tend to be sprinters. Their muscles can produce short
bursts of activity anaerobically (without oxygen), but soon they must
stop and endure a period of aerobic recuperation. Birds and mammals,
being endotherms, can maintain continuous activity for much longer pe-
riods, and their muscles are much better at continuous, long-term power
production.
Flight is a very power-hungry activity (Chapter 3), and many biologists
think that the increased aerobic scope produced by endothermy may be a
requirement for flapping flight in birds. If so, then the question becomes,
did protobirds evolve endothermy along with (or soon after) powered flight?
Or did certain dinosaurs evolve endothermy first, which then became an
exaptation as avialans began to evolve powered flight?

Endothermic Dinosaurs

When John Ostrom first proposed a close connection between birds and
theropod dinosaurs, that linkage led some paleontologists to speculate

*  Physiologists dislike the term “warm-blooded” because the blood of a basking

lizard or dragonfly can be every bit as warm as yours or mine. These ectothermic
(“cold-blooded”) animals are warmed by the environment, however, not by the high
metabolic rate that gives us endotherms an internal furnace.

Bir ds  [ 121 ]

about the possibility of endothermy in dinosaurs.48 Prior to that time,
people thought of dinosaurs as giant lizards, lumbering and sluggish. The
possibility that dinosaurs were endothermic dramatically changed both
scientific and public perception, leading to a more dynamic view of dino-
saurs being active and birdlike, as exemplified by the “Jurassic Park” book
and movies. Although not all scientists are convinced that dinosaurs were
endothermic,49 a consensus has grown that if any dinosaurs were endo-
thermic, theropods are the most likely candidates.
Very large dinosaurs would not have needed endothermy to give them
a high and constant body temperature,* but endothermy could have been
a great advantage for small theropods such as Compsognathus or Velocirap-
tor.50 Small animals lose heat quickly, and if they were not endothermic,
these small dinosaurs would not have been able to hunt or run from preda-
tors during cold seasons or when temperatures dropped at night. Tiny di-
nosaurs like Bambiraptor and Eosinopteryx would have been particularly
limited.
Some of the evidence for endothermy in small theropod dinosaurs in-
cludes (1) apparent down-like feathers scattered throughout the theropod
family tree, even on primitive members such as Sinosauropteryx; (2) une-
quivocal feathers, both downy and pennaceous, on many maniraptorans;
and (3) the extremely small size of several recently discovered species,
such as Eosinopteryx and Epidendrosaurus—both are pigeon-sized or
smaller. The most obvious function of filamentous or downy feathers is
for insulation, and only endothermic animals need insulation. These facts
all suggest that endothermy evolved well before flight. The high activity
levels, endurance, and strong, continuous muscle power of small thero-
pods would have been a key exaptation in the evolution of bird flight.

Origin of the Unique Bird Lung

Modern birds have the most sophisticated, efficient respiratory systems of

any air-breathing vertebrates. Bird lungs have a unique, one-way, flow-
through arrangement. Rather than using in-and-out or tidal air move-
ments into blind-ended pouches or sacs (like us), air flows through bird
lungs from back to front in a nearly continuous stream. Again unlike other

*  Due to their very low surface-to-volume ratio, large dinosaurs would lose heat
very slowly and would have had a high and constant body temperature from size
alone, with no need for elevated metabolic rates. Scientists call this “gigantothermy”
and it would have made Tyrannosaurus a fast, active predator even without the phys-
iological modifications for endothermy.

[ 122 ]  On the Wing

vertebrates, the bird lung does not expand and contract during breathing.
Instead, air is driven through the lungs by the bellows-like action of a set of
internal air sacs. Bird breathing inflates and deflates these sacs in a coordi-
nated sequence that keeps air flowing smoothly and almost continuously
through the lungs. This arrangement is both physically and physiologically
more efficient than mammal lungs. It is physically more efficient because
much less energy is wasted reversing the direction of air flow, and it is
physiologically more efficient because oxygen is easier to extract from a
one-way, continuous flow of air than from air that is essentially standing
still in the deepest air sacs of mammalian lungs.
And birds do seem better able to obtain oxygen than mammals. In a
famous experiment, physiologist Vance Tucker put a mouse and a sparrow
in a pressure chamber and lowered the air pressure. When the pressure got
so low the mouse was near comatose from lack of oxygen, the sparrow
could still fly!51 Scientists thus tend to assume that the bird respiratory
system is an adaptation for flying at high altitudes where the oxygen con-
centration is low, but the real advantage is probably that this unique res-
piratory system can provide the very high oxygen uptake needed to fuel
the high power demands of flapping flight. The high-altitude performance
is most likely a beneficial byproduct.
When did the avian respiratory system evolve? Did dinosaurs have one-
way lungs and air sacs? Some of the early dinosaur-endothermy propo-
nents suggested that dinosaurs had one-way lungs well before flight
evolved, although other scientists see the size of the air passages or the
lack of a nasal heat exchanger as evidence against this theory.52 One
thought-provoking study recently showed that living crocodilians have
lungs with small regions that seem to have one-way flow.53 Crocodiles are
archosaurs, which makes them both the closest living relatives of birds
and fairly close relatives of dinosaurs. One-way lungs in crocodilians
raises the surprising possibility that at least partial one-way, flow-through
lungs may be primitive in dinosaurs, and the fully developed bird-type
system may have evolved in early theropods, long before flight evolved.
Indeed, an ectothermic dinosaur would gain little benefit from a highly
efficient respiratory system, so if early theropods did have one-way lungs,
logic dictates that they would also have been endothermic.

AVIAN POWERED FLYERS ARRIVE

Living birds have several anatomical modifications to aid flight. I have al-
ready mentioned the specialized shoulder joint. This joint has the socket

Bir ds  [ 123 ]

reoriented upward so birds can flap their wings well above the horizontal­—
many birds can raise their wings so high they nearly or actually touch at
the top of the upstroke. (Try lying face-down on a bench and raising your
arms to the side; if you can get them much above horizontal, you are doing
better than me.) The bird shoulder also includes a unique pulley-like ar-
rangement that allows the tendon of a breast muscle to cross over the top
of the shoulder joint and pull up on the humerus (upper arm bone).* This
arrangement puts the primary upstroke muscle low on the chest rather
than above the shoulder, which keeps the center of gravity of the body in
a lower, more stable location. Moreover, this upstroke muscle also twists
the wing leading-edge-up, which automatically changes the wing from the
downstroke to the upstroke orientation. This motion is especially impor-
tant in slow flight and during takeoff and landing.

Pneumatic Bones

Many people probably have heard that birds have air-filled or pneumatic
bones, which are usually described as a flight adaptation to lighten the
skeleton. While true, that description is an oversimplification. (For ex-
ample, we have sinuses—air-filled spaces in our skulls—that have noth-
ing to do with flight.) The air spaces in modern bird bones are tied into the
air sac system that supplies the lungs. These air spaces are not respiratory
since air spaces in bones can’t be used as bellows for pumping or to absorb
useful amounts of oxygen. In larger birds, the lightening of the skeleton
may be vital for flight, but the benefit is less in smaller birds. Indeed, some
small birds as well as diving birds that swim underwater have completely
lost skeletal air spaces.54
As for the evolution of pneumatic bones, several theropod skeletons
have holes that paleontologists see as evidence of air spaces in neck or
sometimes back vertebrae; since this feature is more common in larger
dinosaurs, it might serve as a weight-reduction adaptation for very large
body size.55 Some researchers see pneumatic bones as evidence that respi-
ratory air sacs and possibly one-way lungs were widespread throughout
theropods,56 although not everyone is convinced that those bones were
really air-filled. Even if some or all of those theropod bones really did have
air spaces, they do not necessarily require a birdlike respiratory system;
they could be connected to other airways, like our sinuses are. Moreover,

*  The triosseal canal through the shoulder bones serves as the pulley for the tendon
from the supracoracoideus muscle.

[ 124 ]  On the Wing

if pneumatic bones in large dinosaurs are an adaptation for large size,
then the ancestors of birds (probably quite small theropods) may have
evolved them independently. At this stage, the evidence for when the bird-
like, one-way respiratory system evolved is mixed: almost surely by the
time fully powered flight evolved and possibly quite early in the evolution
of theropods. The evidence of limited one-way flow in crocodilians is in-
triguing, but that lineage shows no evidence of air sacs or pneumatic
bones, so all it really tells us is that dinosaurs may have started out with
partial one-way flow in their lungs. A fully developed one-way system
must have evolved sometime after the lineage leading to dinosaurs split
from that of the crocodilians.

Other Skeletal Specializations

The backbone of birds is heavily modified for flight, even compared to non-
avialan theropods. It is shortened and stiffened in the trunk region to
keep the body’s center of gravity close to the wings so the wings can better
support the weight of the hindquarters. This shortening is even more ex-
treme for the tail. Whereas non-avialan theropods typically have long,
whip-like tails with over 30 vertebrae, living birds have a stubby little
stump, irreverently called the “parson’s nose.” This is supported by a single
bone formed from a handful of fused vertebrae, called the pygostyle.
When we think of a bird’s tail, almost all of what we see as tail is made up
of feathers. The flesh and blood part makes up very little of the tail sur-
face. This arrangement both lightens and shortens the tail. Lightness is
obviously beneficial for flight, but length is more complex. Long tails can
add stability, possibly beneficial for primitive flyers, but living birds have
mostly traded stability for maneuverability, so they evolved short tails for
enhanced maneuvering.57
The hand and wrist skeleton of living birds is heavily modified for flight.
Most non-avialan theropod dinosaurs had three or more wrist bones and
three (sometimes four) long, clawed fingers of several bones each. Recall
from Chapter 2 that living birds have very reduced wrist and hand skele-
tons, including a single wrist bone, a thumb with only one bone, and one
large and one small finger bound tightly together with soft tissue forming
a single functional digit (and no claws). Although reduced in number, the
bones are relatively stout so they serve as solid anchors for the feathers
that bear the body weight in flight (see Fig. 2.1).
Finally, modern birds have a lightweight beak of horn-like material in-
stead of teeth. Beaks appear to be yet another weight-reducing adaptation,

Bir ds  [ 125 ]

presumably for flight. Oddly enough, many extinct birds had teeth whereas
beaks apparently evolved independently several times among extinct birds,
as well as multiple times among dinosaurs and other reptiles.

Archaeopter yx: A Mosaic

The skeleton of Archaeopteryx actually does not show some of the flight
adaptations we have discussed. The shoulder socket is only partly reori-
ented and does not have the pulley arrangement for an upstroke muscle
on the chest. Archaeopteryx probably could raise its wings only slightly
above the horizontal, which suggests it may have had difficulty taking off
from level ground or flying slowly. Its backbone is not noticeably short-
ened and it had a long, bony tail instead of a pygostyle. It apparently had
some pneumatic neck vertebrae, but scientists disagree about whether any
other bones had air spaces;55,58 the fossils just have too little detail to say
for sure whether it had air sacs and one-way lungs. And, of course, it had
teeth. The basal avialans, such as Archaeopteryx and Aurornis, had penna-
ceous wing feathers and more or less limited flight abilities, but they also
had teeth, long tails, separate fingers with claws, and only partially modi-
fied shoulder joints. They were clearly not as specialized for flight as
modern birds. So when did these flight-related modifications appear?

Other Ancient Birds

One of the most primitive avialans to show modern flight adaptations is

Confuciusornis sanctus (encountered earlier when I described Liaoning fos-
sils with feather traces). Confuciusornis is amazing in several ways. It is
probably the most abundant species from Liaoning—museums in China
apparently house nearly 1,000 specimens, and local farmers have exca-
vated and sold many more to collectors. Many specimens have complete
skeletons and soft tissue including well-preserved feathers, and whereas
most have very short tail feathers, a few individuals have a pair of long,
ribbon-like tail feathers longer than the rest of the body. Confuciusornis
had very long, asymmetrical primary feathers, giving it unusually high
aspect ratio wings and indicating economical flight. It had a horny beak, a
pygostyle, and a reoriented shoulder socket, but probably not a triosseal
pulley. The skull was heavily built compared to more modern bird skulls,
and the breastbone (where the main flight muscles are anchored) was
bigger than that of Archaeopteryx but not quite as big as those of modern

[ 126 ]  On the Wing

flying birds. It was probably a stronger and more agile flyer than Archaeop-
teryx but probably not as strong or agile as most living birds.
Although Confuciusornis would have looked quite birdlike in life, due to
its beak and modern-looking feathers, recent phylogenetic studies place it
on a primitive side branch near the base of the avialan lineage (Fig. 6.5).
After the Confuciusornis side branch split off, the two major bird lineages
evolved: first the Enantiornithes, or “opposite birds,” and then the Or-
nithurae, the lineage that leads to modern living birds (Neornithes).
Fully powered flight was largely perfected by the time the Enantior-
nithes split off from the Ornithurae in the very early Cretaceous (over 130
million years ago), although lots of evolution and diversification took place
in both lineages after the split. The Enantiornithes seem to have gotten
the earlier start. Their fossils have been found all over the world,59 sug-
gesting they were powerful, effective flyers. Many have perching adapta-
tions, suggesting they were forest dwellers. The Ornithurae were also
present in the Cretaceous, but ornithurine fossils are much less common
and they seem to have been restricted to oceanic or shoreline habitats.
Ironically, the most common Cretaceous ornithurine fossils are of a large,
flightless, toothed diving bird called Hesperornis, although a few flying
species are also known from Cretaceous fossils.60
At the end of the Cretaceous, enantiornithine birds disappeared along
with the dinosaurs, and the ornithurines survived to give rise to the Neor-
nithes, the modern living birds. The fossil record is not complete enough
to show whether the Enantiornithes were gradually being replaced by the
Ornithurae before the end of the Cretaceous or whether they disappeared
abruptly like the non-avian dinosaurs. In any case, only the ornithurines
survived and they have since undergone a stunning diversification. Scien-
tists have described more species of living birds—over 10,000—than all
other air-breathing vertebrates—amphibians, non-bird reptiles, and
mammals—combined. Birds are clearly the most diverse, widespread, and
abundant of the living land vertebrates. Flight, aiding the ability to spread
into new environments and exploit new habitats, must surely have con-
tributed to the great ecological success of birds.

REFERENCES
1. R. J. Wootton and C. P. Ellington (1991) in Biomechanics in Evolution.
2. D. A. Grimaldi and M. S. Engel (2005) Evolution of the Insects.
3. J. Evans (1865) Natural History Review, series 2.
4. C. Giebel (1877) Zeitschrift für die Gesammten Naturwissenschaften.
5. C. Darwin (1860) On the Origin of Species.

Bir ds  [ 127 ]

 6. P. Shipman (1998) Taking Wing: Archaeopteryx and the Evolution of Bird Flight.
7. J. R. Speakman (1993) Evolution.
8. G. Heilmann (1927) The Origin of Birds.
9. A. D. Walker (1972) Nature.
10. L. D. Martin, J. D. Stewart, and K. N. Whetstone (1980) Auk.
11. H. N. Bryant and A. P. Russell (1993) Journal of Vertebrate Paleontology.
12. P. J. Makovicky and P. J. Currie (1998) Journal of Vertebrate Paleontology.
13. J. A. Gauthier (1986) in The Origin of Birds and the Evolution of Flight.
14. U. M. Norberg (1990) Vertebrate Flight: Mechanics, Physiology, Morphology,
­Ecology and Evolution.
15. J. H. Ostrom (1975) Annual Review of Earth and Planetary Sciences.
16. J. H. Ostrom (1976) Biological Journal of the Linnean Society.
17. W. J. Bock (1965) Systematic Zoology.
18. W. J. Bock (1969) Annals of the New York Academy of Sciences.
19. A. Feduccia (1996) The Origin and Evolution of Birds.
20. U. M. Norberg (1985) in The Beginnings of Birds.
21. J. M. V. Rayner (1985a) in The Beginnings of Birds.
22. J. M. V. Rayner (1988) Biological Journal of the Linnean Society.
23. J. M. V. Rayner (1989) in Evolution and the Fossil Record.
24. J. M. V. Rayner (1985b) in The Beginnings of Birds.
25. P. Burgers and L. M. Chiappe (1999) Nature.
26. Q. Ji, P. J. Currie, M. A. Norell, et al. (1998) Nature.
27. P. J. Chen, Z. M. Dong, and S. N. Zhen (1998) Nature.
28. T. Lingham-Soliar (2003) Naturwissenschaften.
29. A. Feduccia, T. Lingham-Soliar, and J. R. Hinchliffe (2005) Journal of
­Morphology.
30. P. Godefroit, A. Cau, H. Dong-Yu, et al. (2013) Nature.
31. D. E. Alexander (2002) Nature’s Flyers: Birds, Insects, and the Biomechanics of
Flight.
32. X. T. Zheng, Z. H. Zhou, X. L. Wang, et al. (2013) Science.
33. G. Byrnes and A. J. Spence (2011) Integrative and Comparative Biology.
34. M. S. Y. Lee and T. H. Worthy (2012) Biology Letters.
35. G. S. Paul (2002) Dinosaurs of the Air: The Evolution and Loss of Flight in Dinosaurs
and Birds.
36. D. Hu, L. Hou, L. Zhang, et al. (2009) Nature.
37. T. Lingham-Soliar (2010) Journal of Ornithology.
38. J. P. Garner, G. K. Taylor, and A. L. R. Thomas (1999) Proceedings of the Royal
Society of London Series B.
39. K. P. Dial (2003) Science.
40. K. P. Dial, R. J. Randall, and T. R. Dial (2006) BioScience.
41. M. W. Bundle and K. P. Dial (2003) Journal of Experimental Biology.
42. B. W. Tobalske and K. P. Dial (2007) Journal of Experimental Biology.
43. K. P. Dial, B. E. Jackson, and P. Segre (2008) Nature.
4 4. R. Dudley, G. Byrnes, S. P. Yanoviak, et al. (2007) in Annual Review of Ecology,
Evolution, and Systematics.
45. J. A. Oliver (1951) American Naturalist.
46. A. F. Bennett and J. A. Ruben (1979) Science.
47. A. Clarke and H. O. Pörtner (2010) Biological Reviews.
48. R. T. Bakker (1972) Nature.
49. R. A. Thulborn (1973) Nature.

[ 128 ]  On the Wing

 0. F. Seebacher (2003) Paleobiology.
5
51. V. A. Tucker (1968) Journal of Experimental Biology.
52. J. A. Ruben, T. D. Jones, and N. R. Geist (2003) Physiological and Biochemical
Zoology.
53. C. G. Farmer and K. Sanders (2010) Science.
54. P. M. O’Connor (2009) Journal of Experimental Zoology Part a-Ecological Genetics
and Physiology.
55. R. B. J. Benson, R. J. Butler, M. T. Carrano, et al. (2012) Biological Reviews.
56. R. J. Butler, P. M. Barrett, and D. J. Gower (2012) PLoS One.
57. J. Maynard Smith (1952) Evolution.
58. P. Christiansen and N. Bonde (2000) Proceedings of the Royal Society B-Biological
Sciences.
59. L. M. Chiappe and C. A. Walker (2002) in Mesozoic Birds: Above the Heads of
­Dinosaurs.
60. L. D. Martin (1983) in Perspectives in Ornithology.

Bir ds  [ 129 ]

CH A P TER 7

Bats
Wings in the Dark

P eople often find songbirds appealing, but most people seem to be re-
pelled by bats. I suspect this reaction to bats is because they are noc-
turnal, thus creatures of darkness and mystery. Because bats mostly fly in
the dark, few people are aware that they are actually a very widespread
and successful group. Among mammals, only rodents claim more species
than bats, and bats are quite cosmopolitan, inhabiting all continents but
Antarctica, from the tropics to the boreal forests. Flight clearly played a
role in their dispersal. For example, we know bats reached Hawaii from
North America without human help, an inconceivable feat for a tiny non-
flying mammal.
Biologists have long recognized a division between the Old World flying
foxes and fruit bats (“megabats”) and all the other bats (“microbats”).* Mi-
crobats are what those of us from the temperate regions think of when we
think of bats. Microbats tend to be much smaller, they are nocturnal and
use sophisticated echolocation, most eat insects, and they are found all
over the world. Megabats consist of barely 200 species (compared to ap-
proximately 1,000 species of microbats). They are on average larger than
microbats, and the group includes all the largest bats. Most megabats
don’t use echolocation, they are active during the day rather than at night,

*  Megabats were traditionally placed in the suborder Megachiroptera and micro-

bats in Microchiroptera. As we’ll see below, this classification is now obsolete so
I have opted not to use the technical terms. I will instead use the informal terms
“megabat” and “microbat” as a convenient way to distinguish the Old World fruit bat
lineage from the rest of the bats.
and they eat fruit rather than insects. Finally, megabats have a more lim-
ited distribution, being restricted to the Old World tropics and subtropics.
Nevertheless, megabats and microbats are structurally quite similar; al-
though in the past some researchers have questioned their relationship,
they are almost certainly from a single lineage.
Bats unfortunately share parallels with insects when it comes to the
fossil record of flight evolution: the oldest bat fossil is from an animal
that was already a fully powered flyer. Bats’ lightweight skeletons are so
delicate that they don’t leave nearly as many decent fossils as other
mammals, so they have a poor fossil record. Moreover, scientists have
nothing like Archaeopteryx for bats, so fossils tell us neither anything
about how flight evolved, nor about the identity of the direct ancestors
of bats.

CLOSEST REL ATIVES?

Biologists would love to be able to say, “The closest relative of bats are the
[insert name of group here].” After all, bats were the most recent animals
to evolve powered flight, and living mammal groups diversified fairly re-
cently, at least compared to birds and insects. Unfortunately, bats’ anat-
omy is highly specialized. For example, bat wings are highly modified
front limbs. As we saw in Chapter 2, each wing has hyper-elongated fin-
gers supporting a membrane formed by a fold of modified skin. Also, bats’
ears (both external and internal) and their vocal apparatus are modified
for echolocation, an unusual sensing mechanism we will look at later.
These specializations have hindered scientists’ efforts to pin down the lo-
cation of bats on the mammal family tree.
Traditional phylogenies based on anatomical characteristics often
placed bats in a lineage with colugos* and primates called Archonta (for
example, see Fig. 7.1).1 These phylogenies unite bats with primates based on
muscle and reproductive system anatomy and with colugos based on inner
ear and forelimb (wing) anatomy.2 Colugos glide on a wing membrane made
of a sheet of skin stretched between front and hindlimbs. Unlike all other
gliding mammals, colugos have webbed fingers, so the hand acts as part
of the wing (see Chapter 4, Fig. 4.5). Given that much of the bat wing is
supported by modified fingers, colugo wing anatomy thus makes a nice
model for the early evolution of bats from gliding ancestors.3 If bats are

*  Colugos are the large, Southeast Asian gliding mammals sometimes called “flying
lemurs” that we encountered in Chapter 4.

Bats  [ 131 ]
 marsupials

Ancestral anteaters
mammal
Lagomorpha (rabbits)
Rodentia
(mice, squirrels, etc.)
primates
bats Archonta
Dermoptera
(colugos)
shrews

Carnivora
(dogs, cats, bears, etc.)

Artiodactyla (cattle, pigs,

deer, antelope, etc.)
Cetacea (whales, porpoises)
perissodactyla (horses, rhinos)
elephants
Sirenia (manatees)

Figure 7.1:
Traditional mammal phylogeny based on anatomical features (simplified from Novacek;1
some minor groups omitted for clarity).

closely related to colugos, then the evolutionary sequence from a colugo-

like glider to powered flight seems logical and self-evident.
Unfortunately, more recent phylogenies based on genes don’t support
this neat picture. These molecular phylogenies keep primates and colugos
on the same lineage, but bats fall far from this reduced archontan group
(Fig. 7.2). As of this writing, the position of bats on various molecular phy-
logenies has not completely stabilized. These trees generally show bats in
or close to the Laurasiatheria, a lineage including whales, cloven-hoofed
animals (cattle, deer, etc.), horses, and carnivores (dogs, cats, bears, etc.).
Whether bats are nested within Laurasiatheria, or are at the base of the
lineage, or just outside the group remains to be seen, and no consensus yet
exists on which group represents the closest relatives of bats.4–6 Some
phylogenies even show shrews as closely related to bats.7,8 This possible
relationship is intriguing because shrews also use echolocation, although
their system is apparently not nearly as sophisticated as that of bats (see
Box 7.1. Echolocation in Context: The Sensory Suite of Bats).9,10
The fossil record is not much help in teasing out bat evolutionary rela-
tionships. Scientists have described at least two very complete bat fossil
skeletons from the early Eocene (50 to 55 million years ago). Icaronycteris
index was discovered in the 1960s and studied by different researchers

[ 132 ]  On the Wing

 marsupials
elephants
Ancestral Sirenia (manatees)
mammal
anteaters

Lagomorpha (rabbits)
Rodentia
(mice, squirrels, etc.)

primates
Dermoptera (colugos)

shrews

bats
Artiodactyla (cattle, pigs,
deer, antelope, etc.)
Cetacea
(whales, porpoises)
? Perissodactyla
Laurasiatheria (horses, rhinos)
Carnivora
(dogs, cats, bears, etc.)

Figure 7.2:
A molecular (gene-based) phylogeny of the major mammal groups. Various studies differ
in whether shrews and bats should be included in Laurasiatheria. This is a somewhat sim-
plified composite version of trees in Novacek4 and Springer et al.5

over the years.11,12 This species had fully developed wings, spanning 37
centimeters (14½ inches), that were anatomically indistinguishable from
those of living bats; based on the parts of the skull housing the inner ear,
it also used sophisticated, insect-detecting echolocation. Researchers re-
cently discovered the fossil skeleton of Onychonycteris finneyi in the same
fossil beds and from approximately the same age as Icaronycteris.13 Al-
though these two species lived at about the same time, Onychonycteris is
clearly more primitive: it had proportionally shorter wings than all other
bats, and it had claws on all of its fingers.* Moreover, its inner ear does not
show the specializations for echolocation typical of most bats. Neverthe-
less, Onychonycteris was clearly a bat and was very specialized for flight
compared to non-flying mammals. Evidence from the fossils as well as

*  Living bats only have claws on the thumb, or in a few cases, on the thumb and
index finger.

Bats  [ 133 ]
 Box 7.1:  ECHOLOCATION IN CONTEXT: THE SENSORY SUITE
OF BATS

Even though microbats are highly specialized to use sensitive, sophisti-

cated echolocation, bat sonar is a distinctly short-range sense and is
fatiguing to use while bats are perched. Both behavioral experiments
and the physics of calling frequencies show that bats can detect insect-
sized objects only when they are less than 5 or 10 meters (15 to 30 feet)
away. Bats typically fly around 3 to 5 meters per second (7 to 11 mph),
so echolocation detects prey only 1 or 2 seconds before interception.
Although bats always seem to use echolocation to catch flying prey,
they also make extensive use of other senses.
Contrary to the popular phrase “blind as a bat,” bats are not blind. In
fact, when researchers studied vision in microbats, the bats turned out
to have better vision than similar-sized rodents. (Megabats, of course,
are diurnal and have excellent vision.) Researchers have discovered
that when there is enough light to see obstacles, bats stop echolocating
and may even use vision to search for insects on foliage or branches. If
background noise is quiet enough, they may stop calling and listen pas-
sively for insects rustling in leaf litter. Several groups of bats have
evolved to eat pollen and drink nectar, and these bats rely heavily on
their sense of smell to find food.
Bats apparently evolved their advanced sonar to catch insects in
flight, perhaps based on an earlier, simpler system to avoid obstacles
while flying in the dark. Bats can use echolocation for sensing their sur-
roundings, but they are not exclusively dependent on sonar; they have
other well-developed senses as well. This sensory flexibility undoubt-
edly contributed to bats’ success and diversity.

some molecular phylogenies suggest that bats split off from other mam-
mals well before the Eocene, back into the Cretaceous before the dinosaur
extinction. The great antiquity of this split is one of the reasons that sci-
entists have had difficulty discovering bats’ nearest relative. The ancient
split means that their closest relatives among living animals probably di-
verged quite a bit during this long period and may no longer share many
traits with bats.
Most of a bat’s skeleton is heavily modified for flight, and these modifi-
cations make bat fossils instantly recognizable. If we had the fossil skele-
ton of a bat ancestor from before flight evolved, would we recognize it as
being related to bats? I posed this question to a colleague who studies bat

[ 134 ]  On the Wing

biology. He thought such an ancestor would probably be recognizable from
skull anatomy, but without the skull, making a connection to bats might
be quite difficult. Lacking any sort of transition fossil between Onychonyc-
teris and such a possible pre-flight ancestor, and also lacking the ability to
extract genes and construct molecular phylogenies for such fossils, scien-
tists would be hard-pressed to recognize the evolutionary relationship of
a non-flying ancestor with bats.

FLIGHT EVOLUTION

Both the molecular phylogenies and the fossils suggest that bats evolved
flight near the time of the end-Cretaceous extinction event (65 million
years ago), meaning that bats evolved flight long after birds had evolved
powered flight and diversified and quite likely while pterosaurs (Chap-
ter  8) were still around. The living lineages of bats seem to have arisen
near or just before the end of the Cretaceous,14 which implies that the
entire bat lineage and flight must have arisen even earlier. Whether flying
bats had just recently evolved or been around for a long time at the end of
the Cretaceous, the evidence is too sparse to say.

Unquestionably Arboreal

In stark contrast to the situation with bird research, scientists are in near-
universal agreement that bats evolved from arboreal, gliding ances-
tors,15–19 as first proposed by Darwin.20 The reason is simple: bat legs help
support the wing membrane, so how could a running protobat’s legs
become so thoroughly incorporated into the wing? Fast running is a criti-
cal component of the cursorial model, and the basic structure of the bat
wing (not to mention the highly modified legs and backward-facing feet)
seems to preclude running. Most bats are reasonably good climbers; many
roost in trees or tree-holes, and out of all the thousand or so known spe-
cies of bats, only a handful voluntarily spend much time walking on the
ground. Moreover, for decades, scientists considered colugos to be the
closest relatives of bats,1,21,22 which suggested that their common ancestor
was a glider. Nowadays, as genetic studies provide increasingly robust ev-
idence that bats and colugos are not closely related, we think these two
animal groups must have evolved their wings independently. Neverthe-
less, colugos still make a good model for what ancestral protobats must
have looked like. Like bats, colugos have a wing membrane formed from

Bats  [ 135 ]
layers of skin supported by front and hindlimbs, are awkward on the
ground, and are largely nocturnal. Indeed, without ever mentioning colu-
gos explicitly, biologist James D. Smith proposed an evolutionary se-
quence from an arboreal leaper to bats in which the wing he showed for
the intermediate glider (Fig. 7.3) looks identical to a colugo wing.3

Yet Some Raise Questions

A few researchers interested in the evolution of flight in both birds and

bats have raised questions about the gliding-to-flapping transition, as we
saw in Chapter 3 and Chapter 6. Some assert that wings specialized for
gliding would not produce useful force when weakly flapped, and even
that flapping such a wing would cause it to perform more poorly than
during gliding. I am puzzled why this claim persists in the scientific lit-
erature, given that it has repeatedly been refuted both theoretically and
empirically.23–26 Others have argued that flapping a wing designed for
gliding would lead to instability and loss of control. This remains to be
tested and will most likely require something like model tests in a wind

A B C

Figure 7.3:
The evolutionary sequence proposed by James Smith from an arboreal glider to a bat; letters
indicate order of the evolutionary sequence.3 The early stages, particularly stage B, look very
similar to a colugo wing. (Used with kind permission of Springer Science + Business Media.)

[ 136 ]  On the Wing

tunnel to confirm or refute. In any case, no one so far has proposed a rea-
sonable competing theory for flight origins in bats, so scientists generally
accept the arboreal model as the most likely route for achieving powered
flight in bats.

HOW MANY FLIGHT ORIGINS?

Biologists have traditionally separated megabats from microbats within

the bat order, Chiroptera. Megabats are limited to the Old World tropics
and consist of a single family, Pteropodidae, versus about 16 families of
microbats spread around the world. Megabats are mostly diurnal fruit
eaters whereas microbats are largely nocturnal insect eaters. While the
groups are clearly different, the terms “megabats” and “microbats” are
somewhat misleading. In fact, as shown in Figure 7.4, the size ranges of
megabats and microbats overlap substantially, although megabats on av-
erage are significantly larger than microbats. These two groups were
originally considered to be separate suborders—different lineages—of
bats, but we now know that bat relationships are a bit more complex (as
we’ll see).

One Origin or Two for Bats?

Despite their differences, megabats and microbats are so similar in body

form that for over a century, scientists took for granted that they were
both part of one lineage (“monophyletic”). Few scientists took seriously
suggestions that megabats and microbats might have evolved independ-
ently.27 That all changed in the 1980s when John D. Pettigrew and col-
leagues published studies showing that the anatomy of the part of the
brain that processes vision (and the way the eyes are connected to that
brain region) is very similar in megabats, colugos, and primates but dif-
ferent in all other mammals, including microbats (see Box 7.2. Megabat,
Colugo, and Primate Vision: Evolutionary Red Herring).28,29 They inter-
preted their results to mean that megabats are close relatives of pri-
mates but microbats are not. That in turn would mean that powered
flight must have arisen twice, once in megabats and once in microbats.
They went on to develop a phylogeny based on nervous system charac-
teristics that placed megabats close to cologos and primates and posi-
tioned microbats as the lineage that split off earliest among the dozen or
so mammal groups they included. 30 Their argument is bolstered by other

Bats  [ 137 ]
 A

Figure 7.4:
Silhouettes of bats with extended wings, drawn to scale, to show the relative sizes of some
microbats and some megabats. The smallest bat shown here, Pipistrellus pipistrellus (upper
left), has a wingspan of about 23 centimeters (9 inches); although on average, microbats
are smaller than megabats, this diagram shows how much their size ranges overlap.
A. Microbats of the Yangochiroptera lineage. B. Microbats of the Yinpterochiroptera line-
age. C. Megabats (Yinpterochiroptera). (Courtesy of S. T.)

differences: for example, all microbats use echolocation and produce the
sound pulses using their larynx (voice) whereas very few megabats use
echolocation and those that do produce sound pulses by clicking their
tongues.
The anatomical evidence is fairly ambiguous, and other researchers
pointed to strong similarities in forelimb (wing) structure shared by all
bats. 31 Proponents of separate origins countered that this similarity is
to be expected if the wings of megabats and microbats were subjected
to similar, flight-related selection pressures and physical constraints.
Single-­origin proponents countered in turn that even though some of
the fine anatomical details of wing structure differ, the underlying nerve

[ 138 ]  On the Wing

 Box 7.2:  MEGABAT, COLUGO, AND PRIMATE VISION:
EVOLUTIONARY RED HERRING

John Pettigrew did not base his phylogeny uniting megabats (but not
microbats) with colugos and primates on incorrect characteristics:
megabats, colugos, and primates share a set of visual and vision-related
brain features that are very unusual among mammals. These animals
possess color vision, common among birds but rare among mammals.
They also have eyes facing forward with overlapping right and left visual
fields, which gives them binocular vision. Animals use binocular vision
for depth perception—judging distances—but these three groups have
taken it a step further. They have modified the neural wiring that con-
nects the eyes to the brain in such a way that both eyes send part of
their signals to both sides of the brain. This improves the brain’s ability
to compare right and left images and enhances depth perception. If
megabats, colugos, and primates did not inherit these features from a
common ancestor, how did they come to be so similar? Science cannot
give a definitive answer, but we do have some clues.
All three of these groups apparently had a very early history of being
diurnal (day-active), tropical, arboreal fruit eaters. Color vision is easy:
color vision lets animals tell from a distance if many kinds of fruit are
ripe (color being a signal plants probably evolved initially to help fruit-
eating, seed-dispersing birds). Depth perception would also be at a pre-
mium for animals that make their living leaping—or gliding or even
flapping—from one small branch to another high in the tops of very
tall rain forest trees. Reaching out to grab the landing target but miss-
ing it could have terminal consequences. Perhaps depth perception was
also beneficial when reaching out to pluck a ripe fruit (in the tops of
those very tall trees). Our best evidence suggests that megabats and the
colugo-primate lineage independently evolved rewiring of the visual
pathways as a common response to improve depth perception starting
from the standard mammalian pattern. Similar ecological and physical
constraints often lead to convergent anatomies, such as body forms of
sharks and porpoises, although convergence of neural pathways seems
to be rare (but is not well-studied).

innervation pattern for the flight muscles is the same in both groups. 32
This controversy grew rather heated: either the visual systems of mega-
bats and primates evolved convergently or the wings of megabats and
microbats evolved convergently, but anatomical evidence was ­inadequate
to resolve the argument.17,31,33–35

Bats  [ 139 ]
Genes Say “Single Origin”

As researchers began to apply molecular phylogenetic methods to bats and

began building phylogenies based on genes, a consensus began to emerge
from the chaos. Some of the earliest studies firmly supported a single bat
lineage,36 which Pettigrew and supporters of two origins for bats criti-
cized on rather technical chemical and statistical grounds. The tide shifted
decisively when Irish researcher Emma Teeling and colleagues published a
series of studies looking at many genes in many species of bats, as well as
in primates (humans), colugos, dogs, and mice.6,37,38 These studies clearly
showed that megabats and microbats are each other’s closest relatives,
and neither has any close relationship to primates or colugos. These “gene
trees” thus leave no doubt that powered flight evolved only once in bats,
and while the ancestor of bats may have looked somewhat like a colugo,
bats did not arise from the same lineage as colugos (Fig. 7.5).
The molecular studies also showed something quite unexpected: micro-
bats do not form a single lineage because some microbats are actually more
closely related to megabats than to other microbats. Ironically, one of the
first molecular studies to show this split was one co-authored by John Pet-
tigrew, the “two origin” supporter.39 Even before the molecular phylogenies
showed a deep split in the microbats, at least one bat researcher proposed

other mammals
(including primates and colugos)
Laurasiatheria
(including cetaceans, artiodactyles,
carnivores, horses)

Phyllostomidae
Emballonuridae
Yangochiroptera
Vespertilionidae
Molossidae

Megadermatidae
Hipposideridae
Yinpterochiroptera
flapping flight
(ancestral bat) Pteropodidae
(megabats)

Figure 7.5:
This molecular phylogeny of some major bat families shows no close relationship between
bats and colugos and implies a single origin for flight in bats. All bat families other than
Pteropodidae are microbats. (Simplified from the phylogenies in Teeling et al.)38

[ 140 ]  On the Wing

that microbats should be divided into two distinct lineages. Karl Koopman
called these two lineages Yinochiroptera and Yangochiroptera, basing the
names not completely on whimsy: yin is the active principle and yang the
passive principle in Confucian philosophy, and his Yinochiroptera have a
unique mobile bone in their upper jaw.40 At first, some bat researchers
merely saw the molecular phylogenies as confirming Koopman’s microbat
split. The gene-based trees, however, also generally showed that many mi-
crobats (mostly those in Koopman’s Yinochiroptera) belonged on the same
lineage as megabats. Megabats have long been placed in a single family, the
Pteropodidae, so biologists have now combined megabats and their close
microbat relatives into a lineage called Yinpterochiroptera.*
The molecular phylogenies answered some questions but raised others.
Scientists now agree that powered flight arose only once in bats and that
bats are not closely related to colugos (darn, that made such a nice story).
But the microbats in Yinpterochiroptera, such as Old World leaf-nose bats
(Hipposideridae, Fig. 7.5), are all active and sophisticated nocturnal echo-
locators like those in Yangochiroptera, whereas megabats are diurnal and
very few use echolocation. That means that either (1) bats evolved echolo-
cation very early and megabats lost it; or (2) echolocating yinpterochirop-
terans evolved echolocation independently from Yangochiroptera. Given
how central echolocation is to the life of a nocturnal bat and given its close
association with flight in living bats, let’s now take a closer look at echolo-
cation in bats.

HEARING THINGS

Echolocation or sonar † means producing a sound that can be reflected

back from solid objects and listening for the echoes to sense where those
objects are located. Toothed whales (such as dolphins and porpoises) and
microbats have well-developed echolocation systems, and two different
kinds of cave-dwelling birds and at least a few species of shrews seem to
use it as well.9,41 Bats and whales have clearly evolved echolocation for

*  The observant reader might note some redundancy in this new term. “Yinptero-
chiroptera” translates approximately as “active wing hand wing,” and even though
some researchers would prefer a less ad hoc term, this name has become widely
accepted in the scientific literature.
†  “Sonar” usually refers to a human-made device such as used on ships to detect
submerged submarines and echolocation usually refers to an animal’s sensing mech-
anism, but the underlying process is identical; some researchers use them inter-
changeably.

Bats  [ 141 ]
sensing objects when visibility is poor—at night for bats, in murky or very
deep water for whales.
Microbats use their voices to produce ultrasonic sounds, that is, sounds
with such high frequency (pitch) that humans cannot hear them. The
reason bats use such high-frequency sounds is that the wavelength of the
sound determines the smallest object that will return an echo, and wave-
length is inversely related to frequency: you need a very high frequency to
get a short enough wavelength to detect small objects, say, the size of in-
sects. A handy side effect of using such high frequencies is that the sound
can be more narrowly focused and directional; in contrast, lower fre-
quency sounds tend to spread out all over. This directionality has obvious
advantages for sensing the direction of objects. The high frequencies also
have a drawback, however: they attenuate (fade out) very rapidly over dis-
tance so bats must make them very loud, and this requires a lot of power.
These clicks or sound pulses are so intense that if we could hear them, a
bat echolocating at arm’s length would be painfully loud.

Ultrasonic Call Types

Bats use echolocation mainly for two things: detecting and avoiding ob-
jects when they fly in the dark, and insect “hawking” where a flying bat
detects and captures a flying insect. One of the earliest studies of bat sens-
ing showed that intact flying bats could avoid obstacles in total darkness
but not if their ears were plugged.42 This ability pales in comparison to
bats’ ability to detect, track, intercept, and capture insects in flight. Bats
have evolved remarkably sophisticated techniques for zeroing in on flying
insects. Some use frequency-modulated (FM) or broadband calls that give
very precise distance detection whereas others use constant-frequency
(CF) calls that carry farther. Constant-frequency calls also allow the bats
to use the Doppler shift* of the echo to sense their closing speed on the
target insect. Many CF bats can even sense the wingbeat frequency and
orientation of an insect and use that information to decide whether to
bother chasing it. Moreover, as hawking bats close on a target insect, their
calls come faster and faster, so the bat gets more frequent updates of the
target’s position. Bats that hawk in the open tend to use CF calls whereas
bats that hawk in cluttered environments (within a forest) are more likely

*  The Doppler shift is the familiar change in pitch (frequency) due to motion, such
as when a train whistle or siren rises in pitch as it approaches the listener and then
falls in pitch after passing and moving away.

[ 142 ]  On the Wing

to use FM calls. In general, FM bats are more common but some bats use a
combination of both call types depending on the situation.
Separating some microbats into the Yinpterochiroptera and others into
the Yangochiroptera means that CF calls, as well as other specialized fea-
tures of echolocation, evolved independently in the two lineages. This in
turn means that either (1) basic echolocation is ancestral in all bats, and
megabats lost echolocation; or (2) the megabat lineage never had echoloca-
tion, and yangochiropterans and small yinpterochiropterans each evolved
echolocation entirely independently. In the second case, FM and CF styles
are entirely convergent and not inherited from a common ancestor. So far
we don’t have the evidence to answer this conclusively.

Echolocation and Flight

What does all this have to do with flight evolution? Flight and echoloca-
tion in bats are linked by nocturnality and by a literal physical linkage
between the wingbeat and call production.
Bats are probably nocturnal because birds were already well established
and diverse when bats were first evolving flight in the mid- to late Creta-
ceous. The ancestors of bats were most likely already nocturnal and were
able to evolve powered flight at night when birds are largely inactive. Some
scientists have speculated that being nocturnal allowed bats to avoid com-
petition with birds—although they would not have been competing for
food because insect-hawking birds did not evolve until much later. Other
researchers have pointed out that the ancestral owls, hawks, and falcons
were around when the bats first took to the air, so being nocturnal was
probably important to avoid predators.
If you were a small, nocturnal insect eater, having sensitive hearing
might have helped you detect and locate the movement noises of prey. This
passive acoustic detection could have led to some form of echolocation,
which would have been handy for getting around on really dark nights or
detecting jumping or flying insects. We don’t know whether the non-flying
or the gliding ancestor of bats already used some form of echolocation or
whether their descendants evolved it later; either way, it would have been
a huge benefit to a nocturnal flyer. Many dark-adapted mammals rely on
the sense of touch, particularly long whiskers, to find their way in total
darkness. For a flyer moving rapidly through the air, whiskers would not
have been much use for sensing obstacles in time to avoid them.
In order to make their echolocation calls loud enough to detect insects
at a useful distance, bats must put quite a lot of muscle power into

Bats  [ 143 ]
echolocating. In fact, when bats use echolocation while perching, their
metabolic energy use increases tenfold over resting, meaning that echolo-
cation alone requires muscle power akin to vigorous exercise. Bats’ meta-
bolic rates go up even more, by 15 to 20 times the resting rate, when they
are flying, but John Speakman and Paul Racey found in their landmark
study that echolocation while flying does not further increase a bat’s met-
abolic rate.43 In other words, while flying, bats get echolocation “for free.”
When they are searching, they call once per wing beat, so they produce
their calls as a byproduct of flight muscle movements.
This tight linkage between echolocation and flight has led at least some
researchers to suggest that bats evolved echolocation and powered flight si-
multaneously.13 Traditionally, researchers have argued that either echoloca-
tion evolved before flight44 or flight evolved before echolocation.45 To me, a
third possibility seems plausible: the great power requirements of sophisti-
cated, insect-detecting sonar means that as flapping evolved, high-power ech-
olocation became easier and evolved more or less simultaneously. I can even
envision that increasing the strength of echolocation calls in a gliding bat an-
cestor might have stimulated and selected for the earliest, weak flapping.

Echolocation and Size

One curious feature of flight-powered bat echolocation is that it might limit

bat body size. As flapping flyers get bigger, their wingbeat frequency de-
creases. Scientists have suggested that as the ancestors of megabats increased
in size they reached a point where their echolocation calls were not frequent
enough to usefully detect prey. In other words, they traveled so far between
calls that they could not effectively track prey or detect obstacles in time to
avoid them. Large size might also have reduced predation pressure and al-
lowed early megabats to spend more time being active during daylight; in this
case, echolocation would not have been as important and so it was eventually
lost. By the same token, the flight-echolocation linkage may be the reason
the specialized echolocating bats have all remained “micro.” If echolocation
remains beneficial, these bats may be restricted to small body sizes.*

*  Speed and maneuverability may affect size as well. Big flyers are faster, which
seems like an advantage. This speed difference, however, means that smaller flyers
can turn more sharply than big ones; even small bats, as maneuverable as they are,
cannot follow all the twists and turns of an evading insect. They must either try to
intercept, rather than follow, evasive prey, or choose prey that are less evasive. As
they get bigger, large bats may not even be maneuverable enough for interceptions or
following less evasive prey.

[ 144 ]  On the Wing

FLIGHT SPECIALIZATIONS

Bats evolved major anatomical and physiological adaptations to achieve ef-

fective powered flight. The modifications that converted an arm skeleton
into a wing skeleton are pretty obvious (recall Fig. 2.2 from Chapter 2). The
skin of the wing membrane is also heavily modified and quite different
from the skin on the rest of the bat’s body (which is typical mammalian
skin). The wing skin is much more compliant, meaning it stretches and de-
forms more easily.46 Studies by Arnold Song and colleagues on wind-tunnel
models with compliant surfaces show that such flexible wings can have
higher maximum lift coefficients and can be more stall-resistant than rigid
wings.47 Bat wings also contain a complex array of muscle layers, which we
assume are used to adjust the wing’s shape throughout the flight stroke, as
well as a variety of stretchy or tough reinforcing fibers.46 In short, bat
wings are a lot more complex and sophisticated than meets the eye.
Flight imposes high power requirements and, as we saw for birds with
their flow-through lungs, bats have also evolved lung and circulatory
system modifications to improve oxygen delivery to active flight muscles.
Bats have substantially larger lungs and hearts for their size than other
mammals.48 For example, bat hearts can be up to three times bigger than
hearts of other similarly sized mammals. Bats also tend to have more red
blood cells and so more oxygen-carrying hemoglobin in their blood. Fi-
nally, measuring the metabolic rate of animals in flight is quite challeng-
ing, but we have a least a little evidence that bats have higher metabolic
scope than other mammals, meaning that they can sustain an unusually
high metabolic rate for their size.49 Unlike birds, with their highly modi-
fied respiratory system, bats have just evolved incremental improvements
on the typical mammalian system. These changes, however, are clearly
sufficient to give bats the necessary power and stamina to be successful
powered flyers.

REFERENCES
1. M. J. Novacek and A. R. Wyss (1986) Cladistics.
2. J. Shoshani and M. C. McKenna (1998) Molecular Phylogenetics and Evolution.
3. J. D. Smith (1977) in Major Patterns in Vertebrate Evolution.
4. M. J. Novacek (2001) Current Biology.
5. M. S. Springer, M. J. Stanhope, O. Madsen, et al. (2004) Trends in Ecology &
Evolution.
6. E. C. Teeling, O. Madsen, R. A. Van Den Bussche, et al. (2002) Proceedings of the
National Academy of Sciences of the United States of America.
7. R. M. D. Beck, O. R. P. Bininda-Emonds, M. Cardillo, et al. (2006) BMC Evolu-
tionary Biology.

Bats  [ 145 ]
 8. W. J. Murphy, E. Eizirik, W. E. Johnson, et al. (2001) Nature.
9. K. A. Forsman and M. G. Malmquist (1988) Journal of Zoology (London).
10. E. Gould, N. C. Negus, and A. Novick (1964) Journal of Experimental Zoology.
11. G. L. Jepsen (1966) Science.
12. M. J. Novacek (1985) Nature.
13. N. B. Simmons, K. L. Seymour, J. Habersetzer, et al. (2008) Nature.
14. E. C. Teeling (2009) Trends in Ecology & Evolution.
15. U. M. Norberg (1985) American Naturalist.
16. J. M. V. Rayner (1989) in Evolution and the Fossil Record.
17. N. B. Simmons (1995) Symposium of the Zoological Society of London.
18. K. L. Bishop (2008) Quarterly Review of Biology.
19. N. P. Giannini (2012) in Evolutionary History of Bats: Fossils, Molecules and
Morphology.
20. C. Darwin (1860) On the Origin of Species.
21. N. B. Simmons and T. H. Quinn (1994) Journal of Mammalian Evolution.
22. T. A. Vaughan, J. M. Ryan, and N. J. Czaplewski (2000) Mammalogy.
23. U. M. Norberg (1985) in The Beginnings of Birds.
24. R. L. Nudds and G. J. Dyke (2009) Evolution.
25. K. Peterson, P. Birkmeyer, R. Dudley, et al. (2011) Bioinspiration & Biomimetics.
26. N. Vandenberghe, J. Zhang, and S. Childress (2004) Journal of Fluid Mechanics.
27. K. J. Jones and H. H. Genoways (1970) in About Bats: A Chiropteran Symposium.
28. J. D. Pettigrew (1986) Science.
29. J. D. Pettigrew (1991) “Wings or brain . . .” Systematic Zoology.
30. J. D. Pettigrew, B. G. M. Jamieson, S. K. Robson, et al. (1989) Philosophical
Transactions of the Royal Society of London B Biological Sciences.
31. R. J. Baker, M. J. Novacek, and N. B. Simmons (1991) Systematic Zoology.
32. J. G. M. Thewisen and S. K. Babcock (1991) Science.
33. J. D. Pettigrew (1991) “A fruitful, wrong hypothesis . . .” Systematic Zoology.
34. J. D. Pettigrew (1995) in Ecology, Evolution and Behaviour of Bats.
35. N. B. Simmons, M. J. Novacek, and R. J. Baker (1991) Systematic Zoology.
36. W. Bailey, J. Slightom, and M. Goodman (1992) Science.
37. M. S. Springer, E. C. Teeling, O. Madsen, et al. (2001) Proceedings of the National
Academy of Sciences of the United States of America.
38. E. C. Teeling, M. Scally, D. J. Kao, et al. (2000) Nature.
39. J. M. Hutcheon, J. A. W. Kirsch, and J. D. Pettigrew (1998) Philosophical Trans-
actions of the Royal Society of London B Biological Sciences.
40. K. F. Koopman (1984) Bat Research News.
41. H. A. Thomassen, S. Gea, S. Maas, et al. (2007) Hearing Research.
42. J. D. Altringham (2011) Bats: From Evolution to Conservation.
43. J. R. Speakman and P. A. Racey (1991) Nature.
4 4. M. B. Fenton, D. Audet, M. K. Obrist, et al. (1995) Paleobiology.
45. N. B. Simmons and J. H. Geisler (1998) Bulletin of the American Museum of Natu-
ral History.
46. S. M. Swartz, M. S. Groves, H. D. Kim, et al. (1996) Journal of Zoology.
47. A. Song, X. D. Tian, E. Israeli, et al. (2008) AIAA Journal.
48. J. N. Maina (2000) Journal of Experimental Biology.
49. S. P. Thomas (1987) in Recent Advances in the Study of Bats.

[ 146 ]  On the Wing

CH A P TER  8

Pterosaurs
Bygone Dragons

P terosaurs are the last remaining group of powered flyers. I have saved
them for last partly because we know the least about them and partly
because no one has developed any theories of flight evolution for ptero-
saurs aside from those we have already encountered.
Pterosaurs have long been objects of rapt public attention, probably be-
cause some of them were huge. Indeed, they include the largest flying crea-
tures ever known, although many were no bigger than seagulls. Pterosaurs
were reptilian and lived in the Triassic, Jurassic, and Cretaceous periods;
many were huge, and they died out with the non-avian dinosaurs. For this
reason, many people probably assume pterosaurs were a kind of dinosaur.
In fact, while both pterosaurs and dinosaurs (along with birds and croco-
dilians) are archosaurs, they represent two separate and distinct lineages
of archosaurs—in other words, pterosaurs are not dinosaurs. And while
the traditional view is that pterosaurs and dinosaurs are each other’s clos-
est relatives, not all researchers agree.
Pterosaurs, or “pterodactyls” as they are sometimes known (see Box 8.1),
were the first vertebrates to evolve powered flight, following the insects
into the air more than 220 million years ago. They are also the only major
lineage of powered flyers to have gone entirely extinct, disappearing along
with the non-avian dinosaurs at the end of the Cretaceous, about 65 million
years ago. Because we have no living pterosaurs to study, we must make
many inferences and assumptions about them based on living flying ani-
mals that we can study. All modern animals flap to produce thrust and use
variations on the same basic stroke for cruising flight, so we are confident
that pterosaurs did the same. Unfortunately, pterosaur wings were so
 Box 8.1:  WHAT ARE THEY CALLED?

Scientists call the extinct flying reptiles that are the subject of this
chapter “pterosaurs” and they tend to wince at the common use of
“pterodactyl” as a synonym for “pterosaur.” What’s wrong with using
“pterodactyl” as a common name?
The origin of “pterodactyl” goes back to the earliest scientific de-
scription of a pterosaur. The original pterosaur fossil was first described
as a sea creature, but when Georges Cuvier realized it was actually a
flyer, he re-described it and in 1809 gave it the genus name “Ptéro-­
Dactyle.” This name means “wing finger” and refers to the hyper-­
elongated fourth finger that supports most of the wing. The rules for
taxonomic naming were just then starting to gain wide acceptance, and
under those rules, “Ptéro-Dactyle” became the Latinized genus name
“Pterodactylus” for that first pterosaur specimen, which we now know as
Pterodactylus antiquus. Obviously the English common name “pterodac-
tyl” comes directly from Cuvier’s original genus name, so it has a long
history of common usage. The very fact that Cuvier used it as a techni-
cal term, however, causes the problem.
Technically, Pterodactylus is a genus, that is, a group of a few closely
related species out of the hundred-plus known pterosaur species. More-
over, Pterodactylus is the type genus—a sort of exemplar group—for
the suborder Pterodactyloidea, a lineage of later, anatomically distinct
pterosaurs. To a biologist or paleontologist, “pterodactyl” only refers to
the handful of species in the genus Pterodactylus and it can also be con-
fused with “pterodactyloid,” which refers only to members of one out of
several main lineages within the pterosaurs. Thus, to a scientist, “ptero-
dactyl” is ambiguous: is the speaker referring to just the genus, or the
suborder, or to all pterosaurs? I can’t hope to extinguish the use of
“pterodactyl” as a common name, but I do hope readers now under-
stand why scientists much prefer to use “pterosaur” for the name of the
entire group.

different from bird or bat wings (see Chapter 2) that scientists don’t entirely
agree on some of the details, such as how they took off and landed, how ef-
ficient their wings were, or what they did with their wings when not flying.
Hard facts about flight evolution in pterosaurs are even skimpier for
them than for birds or bats. With living animals, we can use molecular
(gene-based) phylogenies to clarify evolutionary relationships, and the
anatomy and physiology of living species can give us pretty strong clues
about how similar structures worked in their extinct ancestors and

[ 148 ]  On the Wing

relatives. In the case of pterosaurs, no human has ever seen an animal fly
by flapping a membranous wing supported by one enormous finger. Simi-
larly, with no live animals to study and almost no information on their
soft tissues, anything researchers say about pterosaur physiology is largely
informed speculation. And finally, we have no early transition fossils for
pterosaurs, and researchers don’t entirely agree on what group may have
been their immediate ancestors. Nevertheless, they were the first verte-
brates to fly under power and they include the largest animals that have
ever flown, so they are worthy of our attention.

ORIGINAL PTEROSAUR FOSSIL DISCOVERIES

Everyone knows what a bird or bee is just from common, everyday experi-
ence. And though most people may never actually see a bat, they will still
know of bats from common cultural references (“blind as a bat”). So when
researchers find a fossil bird or bat, they have a reasonably good idea what
kind of animal they are dealing with. Pterosaurs, in contrast, are not part
of our common, everyday experience. When scientists first encountered
pterosaur fossils over 200 years ago, they came up with a variety of quite
different interpretations for them.*
The first scientifically described pterosaur fossils were from the same
Solnhofen limestone that produced the Archaeopteryx fossils, but the first
pterosaur was found over a half century before Archaeopteryx, in the late
1700s. Cosimo Collini, the first scientist to formally describe a pterosaur
fossil (Fig. 8.1), thought it was some sort of strange sea creature.1 The great
French anatomist Georges Cuvier, however, immediately realized it was a
flying animal;2 he later placed it in the genus Ptéro-Dactyle,3 which became
the basis for the modernized species name, Pterodactylus antiquus (see
Box 8.1). Although Cuvier was well aware that this pterosaur was a reptile,
Samuel Thomas von Soemmerring misidentified a specimen of Pterodacty-
lus as a strange bat that he thought represented a transition between birds
and bats. He published a picture of the animal reconstructed in a very bat-
like posture,4 as shown in Figure  8.2. Although Cuvier and others soon
pointed out Soemmerring’s error and firmly established that pterosaurs
were reptilian and not bats, Soemmerring’s image lingered in the popular
imagination. Indeed, Kevin Padian claims that Soemmerring’s bat-like

*  The history of the earliest studies and diverse interpretations of the first ptero-
saur fossils is recounted in fascinating detail in Peter Wellnhofer’s Illustrated Encyclo-
pedia of Pterosaurs, later reprinted in the United States as The Illustrated Encyclopedia
of Prehistoric Flying Reptiles: Pterosaurs (1996, Barnes & Noble).

P t erosaur s  [ 149 ]
Figure 8.1:
Photo of the original specimen of Pterodactylus antiquus, first described by Cosimo
Collini in 1784 and currently housed in the Bayerische Staatssammlung für Paläontolo-
gie und Geologie, Munich, Germany. (Photo courtesy of S. Christopher Bennett; used by
permission.)

Figure 8.2:
Soemmerring’s bat-like pterosaur reconstruction.4

pterosaur illustration continued to influence the attitudes of both scientists

and the general public into the late 20th century.5,6 Although pterosaur and
bat wings both consist of elongated hand and finger bones supporting a
flexible membrane, the resemblance is superficial. As we will see in this
chapter, pterosaur wings worked rather differently from bat wings.
Over the years since that first Pterodactylus discovery, scientists have
unearthed a number of fossils that preserve some of the wing membrane.
As researchers looked more carefully at those preserved wing membranes,

[ 150 ]  On the Wing

they began to realize that the pterosaur wing membrane was fundamen-
tally different from that of a bat. The bat’s membrane is elastic and flexible,
and its shape (especially camber) is maintained largely by the bones of the
third, fourth, and fifth fingers embedded in the membrane. Pterosaurs, in
contrast, have no finger bones to shape the membrane other than the elon-
gated fourth finger that forms the leading edge. Instead, pterosaur wings
appear to have a dense array of stiffening fibers in most of the wing that
fan out from the wing finger. Researchers are still debating exactly how
these fibers worked, but they do agree that the pterosaur wing membrane
was not simply a flexible sheet flapping in the breeze like a sail or a flag.

PTEROSAUR PHYLOGENY: DINOSAUR COUSINS?

We haven’t discovered any fossils of proto-pterosaurs in the process of

evolving wings, so one way to learn about the early evolution of pterosaurs
is to try to figure out what creatures they are most closely related to; we
can then use that relationship to infer the characteristics of primitive
pterosaurs and their ancestors. On a large scale, pterosaurs are archo-
saurs. As we saw in Chapter 6, archosaurs include dinosaurs—a classifica-
tion now containing birds—crocodiles and relatives, and “thecodonts.”
(“Thecodonts” are no longer considered one lineage but several distinct
lineages of primitive archosaurs.) Archosaurs are defined by a characteris-
tic set of openings in the skull, and pterosaurs are clearly archosaurs, but
who among the archosaurs are their closest relatives?

Traditional View

Scientists have traditionally considered pterosaurs to be closely related to

dinosaurs. As far back as the first year of the 20th century, scientists such
as Harry G. Seeley suggested that pterosaurs shared a close common an-
cestor with dinosaurs.7 Just how close has been much debated. For exam-
ple, German paleontologist Friedrich von Huene published a reconstruction
of a small, primitive archosaur called Scleromochlus taylori in 1914 and
suggested it might be similar to or a relative of the ancestors of ptero-
saurs.8 In 1999, Michael Benton took this a step further and suggested the
Scleromochlus gave rise to both pterosaurs and dinosaurs.9 Others are not
so sure, partly because the Scleromochlus fossil is not very detailed and
partly because it has short front limbs and long hindlimbs, which seems
backward for a proto-pterosaur.10

P t erosaur s  [ 151 ]
 Other researchers have looked for pterosaur ancestors among more
primitive archosaurs like Euparkeria, or even among reptiles more primi-
tive than archosaurs.11 Nevertheless, Kevin Padian and Jacques Gauthier
defined what is closest to the mainstream view by proposing Lagosuchus as
a pterosaur ancestor.12,13 Lagosuchus* was a small (30-cm, 12-inch) bipedal
archosaur that is probably at or very near the base of the dinosaur lineage
(Fig.  8.3). If pterosaurs are also descended from Lagosuchus, that would
make dinosaurs and pterosaurs sister groups—that is, each other’s closest
relatives. A close relationship between Lagosuchus and pterosaurs also has
important implications for the evolution of pterosaur flight, as we will see.

Not Dinosaur Cousins?

Although many paleontologists support a sister-group relationship be-

tween pterosaurs and dinosaurs, not everyone agrees. Paleontologist Mi-
chael Benton initially placed pterosaurs at the very base of the archosaur
tree,14 but later he decided Padian’s analysis was more compelling. Ptero-
saur specialist (and University of Kansas alum) S. Christopher Bennett
has been a prominent critic of the pterosaur-dinosaur connection. In a
1996 paper, he presented a phylogeny that showed pterosaurs down near

non-archosaur reptiles
(lizards, snakes, etc.)
stem
reptiles Proterosuchidae*
Erythrosuchidae*

Proterochampsidae*

crocodile lineage

Euparkeria*

Lagosuchus

pterosaurs
dinosaurs

Figure 8.3:
Phylogeny showing pterosaurs and dinosaurs as sister groups, and both sharing a close
common ancestor with Lagosuchus. Asterisks indicate primitive archosaur groups for-
merly lumped together as “thecodonts.” (Redrawn and simplified from Gauthier.)12

*  The fossil of Lagosuchus is not well preserved. Some paleontologists consider the
better-preserved Marasuchus either a very close relative or the same species, and
Marasuchus gives paleontologists much more anatomical information to work on.

[ 152 ]  On the Wing

the base of the archosaur lineage, among the most primitive archosaurs
and well separated from dinosaurs.15 Although some paleontologists
faulted certain technical aspects of his phylogenetic methods, he was not
alone. Other studies of archosaur relationships, while focused mainly on
more primitive branches, nevertheless place pterosaurs down among
those primitive groups rather than at the top with dinosaurs.16 In a recent,
more refined analysis, Bennett changed his methods to answer his critics
and still got a similar result: pterosaurs splitting off the archosaur family
tree much farther down the trunk than dinosaurs (see Fig. 8.4).17
The mainstream view among paleontologists, however, still places
pterosaurs as very close relatives of dinosaurs. Numerous studies (but all
published before Bennett’s most recent work) contain phylogenies show-
ing that close relationship.18–20 Given the long history of the idea that di-
nosaurs and pterosaurs are sister groups, plus all those relatively recent
studies supporting that view, it remains to be seen whether Bennett’s ar-
guments can shift the majority opinion.
Linking pterosaurs evolutionarily to Euparkeria, Scleromochlus, Lagosu-
chus, or dinosaurs affects how we view the evolution of flight in this group.
All these potential precursors have longer hindlimbs than front limbs. Sci-
entists think Euparkeria may have been able to run short distances on its
hind legs, and Scleromochlus may have been partly or largely bipedal. The
hind legs of Scleromochlus are so long that paleontologists interpret it as a
leaper, somewhat in the mold of a wallaby or a lemur. Researchers have
used the traits of these potential ancestors as a starting point for theories
of flight evolution in pterosaurs.

non-archosaur reptiles
stem (lizards, snakes, etc.)
reptiles Proterosuchidae*
Erythrosuchidae*
pterosaurs
Proterochampsidae*
Euparkeria*
crocodile lineage
Marasuchus
dinosaurs
Scleromochlus

Figure 8.4:
Bennett’s phylogeny moves pterosaurs to a branch among the primitive archosaurs, well
separated from dinosaurs. (Marasuchus on this tree corresponds to Lagosuchus on the tree
in Figure 8.3; asterisks as in Figure 8.3.) Slightly simplified from phylogenies in Bennett.17

P t erosaur s  [ 153 ]
THEORIES OF FLIGHT EVOLUTION

The traditional view of flight evolution in pterosaurs is based on a recon-

struction of the wing membrane attaching along the hindlimbs. If the
hind legs are incorporated into the wings, then running would be diffi-
cult if not impossible. Thus, for most of the 20th century, scientists took
for granted that pterosaurs, like bats, must have initially evolved wings
for gliding. In other words, these scientists assumed that pterosaurs
must have evolved flight via some variation on the arboreal (trees-down)
mechanism.
In the early 1980s, paleontologist Kevin Padian proposed a radically
different scenario that challenged this view.5,21,22 He suggested that scien-
tists had been misled by thinking that pterosaur wing membranes were
necessarily bat-like (harking back to Soemmerring’s misconception). He
proposed that the wing membranes extended only to pterosaurs’ hips, not
down the legs, and made the novel suggestion that pterosaurs were bi-
pedal runners. He reasoned that pterosaurs were close relatives of primi-
tive dinosaurs, and primitive dinosaurs were bipedal runners; therefore,
primitive proto-pterosaurs would most likely have evolved flight via the
cursorial (ground-up) mechanism. Padian called his reconstructions of
narrow-winged, bipedally running pterosaurs “birdlike” reconstructions,
as opposed to the traditional “bat-like” reconstructions (Fig. 8.5). He and
his collaborators presented a new view of pterosaurs as dynamic and agile
on the ground as well as in the air,23 which meshed well with the then-
emerging view of dinosaurs as active, warm-blooded, and birdlike. Given
this interpretation of pterosaurs, Padian’s suggestion of a cursorial origin
of flight in pterosaurs seemed natural and logical, and in many ways it
mirrored his outspoken advocacy of a cursorial evolution of flight in birds.
Unfortunately for this attractive theory, pterosaur specialists began
pointing out that it was not supported by fossil evidence. For one thing, a
number of recently described pterosaur fossils that preserved extensive
wing membrane seem to show that the wing membrane extended all along
the leg to the ankle, so the legs were not free from the wing membranes.24,25
Moreover, contrary to Padian’s reconstruction, both the hip joints and the
foot anatomy of pterosaurs seem to be poorly adapted to running.* 26,27
The final nail in the coffin of bipedally running pterosaurs came when

*  Padian reconstructed his pterosaurs running on their toes with the heel well off
the ground, typical of swift runners like ostriches or antelope. Both foot bones and
trackways show, however, that pterosaurs walked with their heels on the ground,
like us. This arrangement makes for a fine all-purpose walking and climbing foot, but
it is poorly adapted for fast running.

[ 154 ]  On the Wing

 A

Figure 8.5:
Padian’s reconstruction of pterosaurs: (A) with narrow wings (right) compared with the
traditional broad-winged reconstruction (left); (B) as agile runners on the ground. (A
from Padian,22 used by permission of John Wiley & Sons; B from Padian, 5 used by permis-
sion of The Paleontological Society.)

paleontologists described several sets of fossilized tracks that clearly show

pterosaurs to have been quadrupedal—walking on all fours—using the
wings* and hind legs.28,29 The current view is well summed up by the title
of a 1999 review article, “Pterosaurs: Back to the Traditional Model?”:30
the hind leg was incorporated into the wing membrane, pterosaurs (at
least the earlier ones) walked on all fours, and they were not ­bipedal
runners.
If the earliest pterosaurs were quadrupedal walkers rather than bipedal
runners, then a cursorial origin for pterosaur flight does not seem logical.
A cursorial origin requires bipedal running—first, to free up the forelimbs
so they can function aerodynamically, and second, so the animal can run
fast enough for the aerodynamics of its forelimbs to matter. Given the
widespread agreement that early pterosaurs were quadrupeds, most ptero-
saur researchers have returned to the view that pterosaurs must have
evolved through a gliding stage, based on some variation of the arboreal
theory.31 In this view, the ancestors of pterosaurs must have been arboreal

*  The handprints in these trackways are quite unlike handprints of any other ani-
mals, with impressions of the small fingers at an unusual angle plus a large, odd,
trailing impression from the grounded part of the wing finger.

P t erosaur s  [ 155 ]
animals that routinely leaped among branches. They would have initially
evolved flaps of skin on their forelimbs to help steer or extend leaps (or
both). As the flaps enlarged to provide more effective steering, they even-
tually became extensive enough to generate useful lift, and this would
extend leaps and lead to further enlargement. As ancestral pterosaurs
became reasonably adept gliders, steering movements would have evolved
into rudimentary flapping to extend glides and finally into fully developed
flapping for powered flight.

HOW DID PTEROSAUR WINGS WORK?

Padian’s reconstructions, while not ultimately accepted, nevertheless

raised useful questions. He argued (correctly) that the pterosaur wing
membrane functioned very differently from that of the bat wing, and
this argument stimulated much research on the stiffening fibers of the
pterosaur wing membrane. As a result, the general consensus is that
pterosaur wings, while not as narrow as Padian proposed, were probably
narrower than the earlier, traditional, somewhat bat-like depiction. 32
This shape would have made them more aerodynamically efficient and
mechanically sophisticated than wings based on the traditional 20th-
century view.
A simple, stretchy wing membrane running from the wing finger to
the flanks and leg would have flapped like a flag and made an ineffective
wing. The stiffening fibers are the key. Some scientists have proposed that
the fibers were simply battens like those found on sails to prevent flutter-
ing24 or that they gave the wing its camber and transmitted flight loads to
the wing finger.23 Bennett proposed in a more sophisticated analysis that
the fibers mainly kept the membrane spread from front to back, some-
what like the ribs of an umbrella, while at the same time distributing
flight loads away from the tip of the wing finger and more back toward the
hand and arm.32 This arrangement allows the membrane to collapse com-
pactly like a fan when folded, yet retain its fully extended area and camber
when flapping. These fibers were only present in the outer three-quarters
or so of the wing; the shape and movements of the inner part of the mem-
brane without stiffening fibers was probably controlled mainly by the
hind legs.
Imagining pterosaurs walking on the ground may be difficult, but the
fossil trackways show that they walked with their legs fairly close together
under the hips. They walked with the wings slightly splayed to the side,
more or less on the small fingers and the joint at the base of the wing

[ 156 ]  On the Wing

Figure 8.6:
A pterosaur walking quadrupedally, based on Bennett’s reconstruction,28 used by permis-
sion of Taylor & Francis. (Restoration by S. T.)

finger. The joints between the bones of the wing finger itself were basically
immobile and permanently extended so the wing finger would have stuck
up and back alongside the animal’s flanks as it walked (Fig. 8.6). Because
of those immobile wing finger joints, pterosaurs would not have been able
to fold the wing compactly like a bird, and this arrangement might have
been awkward in tight spaces. On the other hand, the non-wing fingers
were strong, well developed, and furnished with strong, curved claws;
unlike birds, pterosaurs probably used their forelimbs as much as their
hindlimbs for climbing.

GROW TH IN PTEROSAURS

Pterosaurs were apparently unique among flapping flyers in that they

could fly at all body sizes—from juvenile to adult. Consider the other
flying animals. Insects, for example, have wings only as adults and adults
never molt, so insects don’t change their body size once they attain flight.
Although birds are not constrained by shedding an exoskeleton, they
don’t become effective flyers until they are very close to adult body size
(mainly because they don’t develop flight feathers on their wings until
they have achieved most of their growth). Even bats, which possess a
more or less complete wing from birth, do not become fully powered
flyers until fairly close to adult body size. Indeed, bats are born at a very
large body size and they grow so rapidly that when they achieve fully
powered flight a few weeks after birth, their wingspan is just slightly

P t erosaur s  [ 157 ]
shorter than an adult’s­—their wings are within the range of adult aspect
ratios and wing loadings. 33
In contrast, pterosaurs seem to have hatched at a very small body size
and to have spent years reaching adult size. The Solnhofen limestone has
produced over 100 fossils of pterosaurs, covering a huge range of body
sizes. For many decades, each new specimen was treated as a new species.
In a set of detailed comparisons in the 1990s, however, Chris Bennett
showed that many of these “species” were actually members of a single
species of different ages.34,35 Based on what appear to be distinct year
classes, Bennett suggested that these moderately large pterosaurs spent at
least two or three years as slow-growing, independent, flying juveniles
before reaching adult size, with adults having wingspans six or eight times
that of the smallest juveniles. (Pterodactylus antiquus had an adult wing-
span of roughly 2 meters or 80 inches so its hatchings might have had
wingspans as small as 28 centimeters or 11 inches.) That these juveniles
were on their own has been confirmed by recent finds of pterosaur fossils
with eggs. The eggs had thin, flexible shells, indicating they were probably
buried rather than brooded like bird eggs.36,37 The small egg size further
supports the idea that pterosaurs hatched at a relatively small body size
and lived independently for a long juvenile period during which they were
presumably capable of flight.38
In a nutshell, any given species of pterosaur must have been able to fly
over a huge range of body sizes. Flying over such a large range of body
sizes is startling not only for its uniqueness among flying animals but be-
cause the aerodynamic properties of wings change substantially over ex-
actly this size range. Small wings have markedly lower lift and higher drag
than geometrically identical large wings at the extremes of this size range
due to the difference in Reynolds number (see Chapter 3, Box 3.1). As a
result, flight would have been more energetically costly and less efficient
for a very young pterosaur, but it would probably have been able to fly
much slower and had more maneuverably than its adult relatives. I am
surprised that I can’t find any analyses of the consequences of this growth
effect. For example, load-carrying and long-distance flight would have
been more difficult, but flight in cluttered habitats like forests would have
been easier for small, juvenile pterosaurs than for their fully grown par-
ents. These differences would have affected everything related to flight­—
from migration ability to foraging style—and they deserve a thoughtful
analysis. Perhaps the changes in wing properties with size were not as
dramatic going from hatchling to adult pterosaurs as going from, say,
hummingbirds to hawks, but we won’t know until some researcher actu-
ally makes the measurements.

[ 158 ]  On the Wing

WHAT FOSSILS CANNOT TELL US

Even the best-preserved pterosaur fossils reveal almost nothing about the
internal organs of the animal. We thus have no direct evidence for what
those organs looked like or how they worked. Were pterosaurs ectother-
mic like crocodiles, or endothermic like birds, or somewhere in between?
Did they have efficient, flow-through lungs like birds or more prosaic tidal
lungs like lizards and mammals? The fossils give tantalizing hints, but so
far, no unequivocal answers. For example, some pterosaurs had exten-
sively pneumatized (air-filled) bones. Some researchers have taken this as
evidence of a birdlike air-sac system and flow-through lungs,39 but per-
haps pterosaurs just evolved skeletal connections with the respiratory
system strictly for weight reduction, independent of the lung arrange-
ment. Without information about the pterosaur’s internal organs, we have
no way of deciding which is more likely.
When fossils are all you have to work with, teasing out anything about
the behavior of the living animal is extremely challenging. Some things
are fairly obvious: teeth, for example, or—if you are really lucky—­stomach
contents can reveal dietary preferences. Vertebrate inner ear structures,
preserved in skull bones, can show how the animal usually held its head;
trackways can tell us how an animal walked or ran. Fossils are mute on
many other forms of behavior: how did they find mates? How social were
they? Did they build nests or migrate? Did they make sounds? For now,
these questions appear to be unanswerable.

Body Mass

Even estimates of body mass in pterosaurs generate controversy. Body

mass is a key element in understanding flight mechanics, so researchers
have put a lot of effort into pterosaur body mass estimates. For example,
body mass estimates for Quetzalcoatlus northropi, the largest known
pterosaur, range from 70 kilograms (around 150 pounds) to 540 kilo-
grams (over half a ton!).40,41 Seventy kilograms seems unrealistically low
for an animal that may have stood over 2 meters (7 feet) tall at the shoul-
der and had the wingspan of a Piper Cub airplane. On the other hand, a
half-ton Quetzalcoatlus would have been quite incapable of flight, so what
use did it have for a wing with a span of over 9 meters (30 feet)? Even the
more reasonable sounding mass estimates—200 to 250 kilograms (440
to 550 pounds)—give one pause: could a 250-kilogram pterosaur really
have landed slowly enough to avoid injury?42

P t erosaur s  [ 159 ]
Heads and Feet

The function of some pterosaur structures remain mysterious. For exam-

ple, many (perhaps most) pterosaurs had large crests on the top or back of
the head (Fig. 8.7). Some of these doubled or tripled the surface area of the
head, and some may have been even further extended by soft tissue.
Structures this large would inevitably have had significant aerodynamic
effects. Paleontologists have suggested that crests were used in flight for
steering,43 and aerodynamic tests have showed that heads with large
beaks and crests make very effective rudders.44 Pterosaurs were unlikely
to have used crests as a primary steering mechanism, however: wings are
much more efficient and effective for producing turns, and living flying
animals steer entirely with wings (a few steer mostly with wings and a
little with tails); pterosaurs without crests surely turned using wings just
like all other flying animals. Moreover, boats have rudders at the back for
a reason: rudders in front are unstable and difficult to keep centered; they
are difficult to straighten once turned and very prone to progressively
tightening turns once a turn starts. While not the primary steering device,
crests could conceivably have been used for fine adjustments to turns or,
more likely, for sudden, rapid turns in urgent situations. An intriguing
advantage of using a crest to turn is that merely turning the head to look

Figure 8.7:
A sample of the variety of pterosaur head crests. (Courtesy of S. T.)

[ 160 ]  On the Wing

at something would cause the pterosaur to swerve in that direction. If not
rudders, what were crests for? Perhaps they were used for some sort of
display, to establish dominance or to attract mates. Whatever the func-
tion, if crests were not for steering, the function must have been very im-
portant because huge crests would have produced significant aerodynamic
costs. For example, these large structures would have increased drag and
made turning the head to look to the side almost impossible in straight
and level flight.
Another pterosaur mystery structure is webbed feet: fossils show that
several pterosaur species had them. “Webbed feet?” you say. “That’s a no-
brainer, they were swimmers.” That is also what paleontologists thought.
But recently, a study by British paleontologist David Hone and his Cana-
dian colleague Donald Henderson used computer models to show that be-
cause they have proportionally huge heads and tiny bodies, pterosaurs
would not have been able to float on the surface of the water without tip-
ping over!45 The problem, as contended by Hone and Henderson, is that
even though some pterosaurs had long necks, they were not flexible
enough to form the sharp S-curve needed to bring the head back over the
center of gravity of the body like a floating pelican. With their heads out in
front of the body, these pterosaurs could only have floated with the head
partly submerged, which would have made breathing a dicey proposition
in any conditions other than flat calm. So some pterosaurs may have had
webbed feet, but if they used them for swimming and Hone and Hender-
son are correct, they did not swim like ducks or geese.

No Molecular Phylogenies

Another limitation of fossils, particularly for building phylogenetic trees,

is that scientists cannot get genes from fossils as old as pterosaurs (Juras-
sic Park notwithstanding); the DNA in the genes breaks down far too rap-
idly. That means researchers trying to reconstruct phylogenies of
pterosaurs and their relatives can’t tap that important source of phyloge-
netic information. Instead, they can only use anatomy—mostly limited
to bone anatomy, to boot—to build their trees. Some might argue that
being limited to anatomical information is not that big a problem; until
about two decades ago, that was what all phylogenies were based on, and
in many cases adding genes to phylogenetic trees did not change them all
that much. Sometimes, however, molecular (genetic) evidence does
matter a lot, as we saw for bats (Chapter  7). The argument about the
placement of pterosaurs within the archosaur lineage is a case in point.

P t erosaur s  [ 161 ]
Lacking molecular data, paleontologists only have anatomical clues to
try to tease out whether the apparent similarities of the hindlimbs of
pterosaurs and dinosaurs are due to inheritance from a common ances-
tor, convergent evolutionary responses to similar selection pressures, or
sheer coincidence. While most recent studies assume common ancestry,
Bennett’s study provides the strongest argument for convergence or
coincidence—­t hat is, against a close dinosaur-pterosaur relationship.17
Ironically, Bennett’s study borrowed techniques for comparing molecular
and anatomical phylogenies and instead used them to compare different
categories of anatomical data.

REASON FOR OPTIMISM?

Rather than ending our look at pterosaurs on a pessimistic note, we have

good reason for a more hopeful outlook. A recent review pointed out that
approximately a third of all known pterosaur species have been discovered
since 2000.38 In other words, thanks to huge numbers of new pterosaur
fossil finds, mostly in China but also in Brazil, scientists have described
half as many new pterosaur species in the last decade or two as for the
preceding two centuries! Many of these new fossils are from sites that
sometimes preserve soft tissue. The bonanza of new fossils has already
helped refine our notions of pterosaurs, and if it continues, at least some
of the currently unanswerable questions may finally get resolved.

REFERENCES
1. C. A. Collini (1784) Acta Academiae Theodoro-Palatinae, Mannheim, Pars Physica.
2. G. Cuvier (1801) Journal de Physique, de Chimie et d’Histoire Naturelle.
3. G. Cuvier (1809) Annales du Muséum national d’Histoire naturelle, Paris.
4. S. T.v. Soemmerring (1817) Denkschriften der koniglichen bayerischen Akademie
der Wissenschaften München, mathematisch-physikalische Classe.
5. K. Padian (1983) Paleobiology.
6. K. Padian (1991) in Biomechanics in Evolution.
7. H. G. Seeley (1901) Dragons of the Air, an Account of Extinct Flying Reptiles.
8. F. v. Huene (1914) Geologische und Paläontologische Abhandlungen NF.
9. M. J. Benton (1999) Philosophical Transactions of the Royal Society of London
Series B-Biological Sciences.
10. P. Wellnhofer (1996) The Illustrated Encyclopedia of Prehistoric Flying Reptiles:
Pterosaurs.
11. R. Wild (1984) Naturwissenschaften.
12. J. A. Gauthier (1986) in The Origin of Birds and the Evolution of Flight.
13. K. Padian (1984) in Third Symposium on Mesozoic Terrestrial Ecosystems.

[ 162 ]  On the Wing

 14. M. J. Benton (1985) Zoological Journal of the Linnean Society.
15. S. C. Bennett (1996) Zoological Journal of the Linnean Society.
16. S. Renesto and G. Binelli (2006) Rivista Italiana Di Paleontologia E Stratigrafia.
17. S. C. Bennett (2013) Historical Biology.
18. S. L. Brusatte, M. J. Benton, J. B. Desojo, et al. (2010) Journal of Systematic
­Palaeontology.
19. D. W. E. Hone and M. J. Benton (2007) Journal of Systematic Palaeontology.
20. S. J. Nesbitt (2011) Bulletin of the American Museum of Natural History.
21. K. Padian (1982) Sciences-New York.
22. K. Padian (1985) Palaeontology.
23. K. Padian and J. M. V. Rayner (1993) American Journal of Science.
24. D. M. Unwin and N. N. Bakhurina (1994) Nature.
25. R. A. Elgin, D. W. E. Hone, and E. Frey (2011) Acta Palaeontologica Polonica.
26. D. M. Unwin (1987) Nature.
27. P. Wellnhofer (1988) Historical Biology.
28. S. C. Bennett (1997) Journal of Vertebrate Paleontology.
29. D. M. Unwin (1997) Lethaia.
30. D. M. Unwin (1999) Trends in Ecology & Evolution.
31. S. C. Bennett (1997) Historical Biology.
32. S. C. Bennett (2000) Historical Biology.
33. L. V. Powers, S. C. Kandarian, and T. H. Kunz (1991) Journal of Comparative
Physiology a-Sensory Neural and Behavioral Physiology.
34. S. C. Bennett (1995) Journal of Paleontology.
35. S. C. Bennett (1996) Journal of Vertebrate Paleontology.
36. J. C. Lü, D. M. Unwin, D. C. Deeming, et al. (2011) Science.
37. D. M. Unwin and D. C. Deeming (2008) Zitteliana Reihe B.
38. D. W. E. Hone (2012) Acta Geologica Sinica-English Edition.
39. R. J. Butler, P. M. Barrett, and D. J. Gower (2009) Biology Letters.
40. S. Chatterjee and R. J. Templin (2004) Geological Society of America Special
Papers.
41. D. M. Henderson (2010) Journal of Vertebrate Paleontology.
42. M. P. Witton and M. B. Habib (2010) PLoS One.
43. W. B. Heptonstall (1971) Scottish Journal of Geology.
4 4. H. R. Jex (2000) Making Pterodactyls Fly (QN Story).
45. D. W. E. Hone and D. M. Henderson (2014) Palaeogeography, Palaeoclimatology,
Palaeoecology.

P t erosaur s  [ 163 ]
CH A P TER  9

Pedestrians Descended from Flyers

Loss of Flight

I have spent much of this book discussing the advantages of flapping

flight, yet lots and lots of flightless animals are descendants of flying
species. We know from their evolutionary relationships that animals
like lice, bedbugs, penguins, and ostriches all had ancestors that were
fully capable of powered flight, yet they themselves are completely flight-
less. These secondarily flightless animals demonstrate that species can
find themselves in situations where the costs of flight outweigh the ben-
efits. In those situations, natural selection will favor the evolutionary
loss of flight.

ANCESTRY CONSTR AINS LOSS OF FLIGHT

The main lineages of flying animals differ dramatically in their propen-

sity to abandon flight. Insects include several major lineages (such as fleas
and lice) and hundreds, maybe thousands, of minor ones that are second-
arily flightless. Many bird lineages also contain flightless members. Aside
from large, flightless runners like ostriches and emus, scientists have
catalogued flightless ducks, geese, ibises, rails, and even a flightless
parrot. The extinct dodos and the threatened kiwis are entirely flightless.
In contrast, all known bats can fly, and all but a tiny handful of bat spe-
cies avoid even landing on the ground. Similarly, to the best of our knowl-
edge, all pterosaurs were flyers. Although the fossil record is incomplete,
we have yet to find a pterosaur with greatly shortened or non-wing-like
front limbs.
 Does the presence of many secondarily flightless insects and birds,
but no secondarily flightless bats or pterosaurs, give us any useful in-
sights? The difference is probably due to the degree of independence
from the wing structure of some walking legs. Insects’ wings are com-
pletely separate from their legs, so alone among powered flyers, insects
did not give up any legs to achieve flight. Although birds did lose the
front limbs in order to evolve flight, the hindlimbs always remained sep-
arate; once they developed powered flight, the hind legs were available to
evolve separately.* Birds could thus evolve hind legs specialized for all
sorts of terrestrial or aquatic locomotion without impairing their ability
to fly. This separation of flying locomotion from non-flying locomotion
means that birds and insects could evolve such competent non-flying
locomotion that under the right conditions they might actually benefit
from giving up flight.
In contrast, bats (and presumably pterosaurs) use a flight mechanism
that incorporates the hind legs into the wing structure. As a result, natu-
ral selection cannot easily act to improve the hindlimbs’ non-flight loco-
motion because such modifications would come at the expense of the
animal’s flight ability: becoming a slow, clumsy runner is unlikely to be an
advantage if it also means becoming a weak, clumsy flyer. This incorpora-
tion of the hindlimbs into the wing seems to have committed bats entirely
to flight as their primary locomotion. As well as we can tell, the same also
applies to pterosaurs.

LESS-THAN-TOTAL FLIGHTLESSNESS

Loss of flight does not have to be permanent or species-wide. Many birds,

especially swimmers, lose the ability to fly when they undergo molt (peri-
odic feather replacement). Most birds molt wing feathers gradually so
their ability to fly is only slightly impaired at any one time, but a number
of birds seek out refuges with adequate food and few predators and molt
all at once. For example, some mergansers (a kind of duck) spend up to
30  days flightless during their molt.1 Similarly, many species of aphids
(tiny plant-feeding insects) are wingless at the beginning of the growing

*  While birds may well have gone through a four-winged gliding stage (Chapter 7),
the hindlimb was never structurally incorporated into the front wing. As the front
limb became more specialized for flapping, the hindlimb appears to have reverted
back to a more terrestrial function. Because the hind “wing” never lost separate toes,
claws, or leg muscles, this would not have been a dramatic transition.

P e d e s t r i a n s D e s c e n d e d f r o m F ly e r s   [ 165 ]
Figure 9.1:
Adult female wingless (left) and winged (right) apple aphids of the same species, Aphis
pomi. (Redrawn by S. T. from Carpenter.)2

season when their host plants are also growing and becoming more abun-
dant. Later in the season, crowding causes the same aphid species to pro-
duce winged adults that can disperse to seek new populations of the host
plant or to head south for the winter (Fig. 9.1).2,3
Many species of insects include both flying and non-flying individuals
simultaneously. Several species of crickets and many moth species include
both long-winged (flying) and short-winged or wingless (non-flying) indi-
viduals.4,5 Sometimes the difference is by gender, with the males flying
and the females being flightless; in other species, either sex can become
flightless. Flightless females often have bigger ovaries and lay more eggs,
so species with both flying and flightless forms seem to exist in a balance
where some situations favor laying more eggs but other situations favor
more dispersal ability.
Social insects like ants and termites also include flying and flightless
individuals in the same species. In these species, wingless, sterile workers
make up the vast majority of the colony’s members. Only reproductive
individuals have wings. These winged individuals provide the main dis-
persal route for the species. They leave the colony, mate, and disperse to
new locations where the survivors found new colonies. In fact, once the
mated queen finds a good location to start a new nest, she promptly sheds
her wings and never flies again. I am tempted to suggest that the subter-
ranean ways of these social insects promoted wing loss in the workers.
Indeed, many other burrowing and wood-boring insects (wood roaches,
numerous beetles) have also lost their wings, yet many additional bur-
rowing and wood-boring species have retained their wings. So burrowing
may or may not contribute to flightlessness, depending on the specific
circumstance.
Often, flightlessness is fixed and permanent. Ostriches are a classic
example. Because of their large size, specializations for running, and

[ 166 ]  On the Wing

relatively small, fluffy wings,* all ostriches are permanently pedestrian.
Similarly, penguins have become so specialized for underwater swim-
ming that their flipper-like wings, albeit powerful, are far too small for
aerial powered flight. Among insects, many such as fleas and lice that live
as external parasites (ectoparasites) on vertebrates are completely flight-
less, evidently dispersing along with their hosts rather than flying.

WHEN AND WHERE NOT TO FLY

Charles Darwin may have been the first to suggest that oceanic islands
might promote flightlessness.6 Flyers on islands may not have anywhere
much to go for dispersal. Moreover, strong oceanic winds and especially
storms might blow a flyer so far from the island that it could not find its
way back. While no problem for a seabird, getting blown far off an island
could be fatal for a wren or a June beetle. Scientists have long argued
about whether islands really promote flightlessness, possibly because the
answer depends on what animal a scientist studies. Although Pacific is-
lands do have many flightless insects (which is what originally caught
Darwin’s interest),7–9 evolutionary biologist Derek Roff has analyzed the
distribution of flightlessness in insects and concluded that flightlessness
is no more common among island insects than among mainland insects.5

Land Birds

Birds, however, are a different story. Flightless island bird species abound,
from the extinct dodos, moas, Hawaiian geese, and Jamaican ibises to
living Galapagos cormorants and New Zealand’s kiwis and flightless par-
rots. Many of these are medium-sized birds that are large enough that
they really can get anywhere on an island by walking, and on many is-
lands, they have no terrestrial predators. Perhaps the best-known exam-
ples are the flightless rails of the Pacific islands. Rails are ground-feeding,
somewhat chicken-like birds that are widely distributed around the world.
On continents, they normally can fly. On many Pacific islands, however,

*  Ostriches can have a wingspan of over 1.5 meters (5 feet), which seems pretty
large until you realize that they stand 1.8 to 2.7 meters (6 to 9) feet tall and weigh
over 90 kilograms (200 pounds). The extinct giant teratorn, which almost certainly
could fly, only weighed three-quarters as much as an ostrich but had wingspans well
over 6.7 meters (22 feet).

P e d e s t r i a n s D e s c e n d e d f r o m F ly e r s   [ 167 ]
they cannot. Phylogenetic studies show that once rails colonize a predator­-
free island, they lose flight fairly quickly in evolutionary terms—120,000
to 500,000 years.10 Environmental physiologist Brian McNab showed that
not only do flightless island birds escape the high energy cost of flying but
their resting energy consumption is lower than that of their flying rela-
tives. That difference allows the flightless birds to survive on less food, or
to produce more eggs on the same food, than their flying relatives.11

Waterfowl

Several different lineages of aquatic birds have become flightless. Pen-

guins are the most obvious example, but they are far from alone. Other
lineages include flightless cormorants, steamer ducks, New Zealand teals,
the extinct flightless ducks and geese of Hawaii, and the recently extinct
Great Auk. Diving birds actually benefit from being denser or heavier for
their size, a condition that obviously conflicts with flight. Shorter, stub-
bier wings are more effective for swimming but less effective for flying.
These birds can use blubber (a fat layer) instead of feathers for insulation;
such fat would be prohibitively heavy for a flyer but it has several advan-
tages for a diver. For example, air trapped in the feathers of a diver com-
presses with increasing depth and changes the bird’s buoyancy whereas
blubber doesn’t compress so it does not affect the bird’s buoyancy as the
bird dives deeper.
The Great Auk, sadly driven to extinction just a couple hundred years
ago, is a particularly revealing example. The auks are a family of birds (Al-
cidae) that both fly in air and swim underwater with their wings. Their
wings are thus a compromise, small for flight but big for swimming. Curi-
ously, the small auks like dovekies and least auklets (18 to 20 centimeters
or 7 to 8 inches long) are the strongest, most agile flyers and the weakest,
slowest swimmers, whereas the large ones like murres (44 centimeters or
17 inches long) are powerful swimmers but marginal flyers. Murres appear
awkward when landing—sometimes requiring several tries and still
sprawling clumsily—and they require such a long takeoff run on the water
surface that they are unable to take off in rough water or without a head-
wind. The Great Auk was the logical extreme: twice as long and five times
as heavy as a murre, it was apparently a very fast, capable, penguin-like
swimmer, but it was totally flightless (see Box 9.1). Evidently, the benefits
of large size for swimming won out over the benefits of marginal flying
ability, so as it got bigger, it evolved much shorter, smaller wings for swim-
ming and lost flight. As well as being flightless, Great Auks were somewhat

[ 168 ]  On the Wing

awkward on land and had little fear of humans due to breeding on rocky
islands and cliffs without predators. They were thus easy pickings for
market hunters who killed them in wholesale batches for their eggs and
feathers. By the mid-1800s, they were gone.13
When diving birds can breed in environments free from terrestrial
predators and can obtain the food they need and get to the places they
need to go by swimming, the costs of flight may come to outweigh the
benefits, as has happened in all these flightless swimmers.

Insects

As for insects, if islands don’t promote flightlessness, what does? Surpris-

ingly, one of the strongest correlations is with very stable, long-lived, un-
disturbed habitats. Quite a few species of deep forest moths have flightless
females, and these are often found in large, undisturbed, virgin forests.5
Probably more familiar are the previously mentioned ectoparasitic insects
like fleas and lice. Flightlessness makes good sense here: if your home and

Box 9.1:  THE ORIGINAL PENGUIN

The term “penguin” originally referred to Great Auks. Linguists dis-

agree on the origin of the term: some derive it from a Welsh phrase for
“white head” but others say it comes from a Spanish or Portuguese word
that means “fat” in some contexts. Regardless of the origin, British sail-
ors who encountered Great Auks in Newfoundland in the 1500s were
calling them “penguins.” Indeed, when Carolus Linneus, the father of
modern taxonomy, described a Great Auk, he gave it the scientific name
Pinguinus impennis.
When Sir Francis Drake’s ship, the Golden Hind, circumnavigated
the globe in the late 1500s, his sailors encountered the birds we now
call penguins in the Straits of Magellan. They called these Southern
Hemisphere birds “penguins,” either because of their close resemblance
to Great Auks or because the sailors did not realize they were different
from Great Auks. (Great Auks and penguins look superficially similar
but they are not closely related.) Possibly due to the popular accounts of
the Golden Hind’s voyage, perhaps aided by the declining Great Auk
populations as well as alternative local names like “garefowl,” by the
1600s “penguin” began to take on its modern meaning, referring solely
to the Southern Hemisphere birds.

P e d e s t r i a n s D e s c e n d e d f r o m F ly e r s   [ 169 ]
your source of food—your host—is mobile, why ever leave? Moreover, if
you can’t find a suitable mate on your current host, just wait till your host
mates. Then you can simply walk onto this new host and look for mates.
Flight for such an insect clearly provides little or no benefit, and the cost
of leaving (and potentially losing) your host could be quite high. In addi-
tion to the major lineages represented by fleas and lice, numerous minor
lineages including bedbugs and sheep keds have gone this route. Sheep
keds look rather tick-like but they are insects in the order Diptera, so they
are actually flightless flies!
Many temperate and high-latitude insects can survive freezing or even
sub-freezing temperatures, but they usually do so hunkered down and
dormant. Very few insects actively feed and reproduce at freezing tem-
peratures; a handful of insects do, however, and can sometimes be found
wandering around on the surface of snow, particularly in alpine regions.
None of these cold-loving insects can fly. The problem is physiological. At
such low temperatures, muscles simply can’t move fast enough to flap the
wings effectively. North American examples of such cold-loving insects
include the snow scorpionflies in the genus Boreas and snow crane flies in
the genus Chionea,12 the latter being another example of flightless flies.

ONCE LOST, GONE FOREVER

If, through natural selection, a species that once could fly evolves loss of
flight, can it reverse the process and regain the power of flight? If, by re-
versal, I mean regaining lost structures such as wing components in their
original form and with their original function, evolutionary biologists
would overwhelmingly say, “No.” Once a structure is lost (assuming the
loss is genetically fixed throughout the species), nature has no way to
rewind or back up to the pre-loss condition. In principle, natural selection
could modify some other structure to function in place of the lost struc-
ture. For example, true flies (Diptera) have lost the egg-laying appendage
or ovipositor of other insects, but a few flies have modified the back end of
the abdomen to function like an ovipositor. When it comes to flight, if a
species doesn’t fly because its wings are too short or its flight muscles are
too small, then under the right conditions, natural selection could con-
ceivably enlarge those structures and allow the species to regain flight (al-
though I have never heard of a case of this happening for a completely
flightless species). If, however, major components of the wings or their
muscles are completely lost throughout the species, as in the total absence
of wings in fleas, then the species will be permanently grounded. (Some

[ 170 ]  On the Wing

researchers think that stick insects may have lost and then reacquired
wings based on a molecular phylogeny,14 but this claim is quite controver-
sial and other scientists have reinterpreted the phylogeny to show no need
for wing reacquisition.)15
You might say, “Well, couldn’t they evolve wings partly or entirely from
new structures, the way the first birds or bats did?” and the answer is,
“Yes, but only under the right conditions.” Considering that such an incipi-
ent flyer would be competing with a whole host of already capable flyers,
and especially considering that the “right conditions” seem to have oc-
curred only four times in the more than 400-million-year history of land
animals, the chances of a secondarily flightless animal “re-evolving” pow-
ered flight are, for all practical purposes, zero.
How widespread is loss of flight? Among birds it is quite rare, having
occurred in fewer than 60 species out of the 10,000 or so scientifically
described species living today. Insects are more challenging to pin down.
Flight loss can vary hugely among lineages (for example, 100% of fleas are
flightless whereas 0% of dragonflies and damselflies are flightless). Based
on the amount of flight loss estimated for most of the major flying lin-
eages,5 an overall estimate for known insect species is in the 5% to 10%
range. Although small on a percentage basis, 5% of nearly a million de-
scribed insect species is still a pretty big number. In fact, that is more or
less the same number as all known vertebrate species (fish, amphibians,
reptiles, mammals, and birds) put together.
Flight obviously has its advantages. As we have seen in this chapter,
however, animals in some situations may not benefit much from being
able to fly and may actually incur evolutionary costs—growing, maintain-
ing, and carrying around wings and flight muscles when flight is counter-
productive. Under these conditions, natural selection will actually favor
loss of flight.

REFERENCES
1. S. R. Craik, J.-P. L. Savard, and R. D. Titman (2009) Condor.
2. P. J. Gullan and P. S. Cranston (2010) The Insects: An Outline of Entomology.
3. G. H. Carpenter (1913) The Life-Story of Insects.
4. D. A. Roff (1984) Oecologia.
5. D. A. Roff (1990) Ecological Monographs.
6. C. Darwin (1860) On the Origin of Species.
7. N. L. Evenhuis (1997) Bishop Museum Occasional Papers.
8. M. J. Medeiros (2008) Zootaxa.
9. C. A. Tauber, M. J. Tauber, and J. G. Giffin (2007) European Journal of Entomology.
10. B. Slikas, S. L. Olson, and R. C. Fleischer (2002) Journal of Avian Biology.

P e d e s t r i a n s D e s c e n d e d f r o m F ly e r s   [ 171 ]
 1. B. K. McNab (1994) American Naturalist.
1
12. G. W. Byers (1969) Evolution.
13. D. N. Nettleship and P. G. H. Evans (1985) in The Atlantic Alcidae: The Evolution,
Distribution and Biology of the Auks Inhabiting the Atlantic Ocean and Adjacent
Water Areas.
14. M. F. Whiting, S. Bradler, and T. Maxwell (2003) Nature.
15. J. W. H. Trueman, B. E. Pfeil, S. A. Kelchner, and D. K. Yeates (2004) Systematic
Entomology.

[ 172 ]  On the Wing

CH A P TER  10

Unifying Themes?

A s we look for shared features of flight evolution among the Big Four­—
 insects, pterosaurs, birds, and bats—we must keep in mind that each
of these groups evolved flight at different times separated by tens or hun-
dreds of million years and under extremely different circumstances. In-
sects probably evolved flight fairly soon after becoming terrestrial, which
would have been before plants had fully colonized land far from water.
Pterosaurs arose more than 100 million years later, after early forms of
trees had evolved and formed widespread forests. Pterosaurs would not
have competed directly with insects because they were so much larger, so
these first flying vertebrates also faced no aerial competition. Birds, on
the other hand, would have been at risk of competition and possibly pre-
dation from pterosaurs when they evolved flight 50 million years or so
after pterosaurs. Perhaps birds evolved flight far inland at a time when
most pterosaurs were largely coastal, or perhaps the protobirds were
enough smaller than pterosaurs that they avoided most competition. Fi-
nally, birds and maybe pterosaurs were present when bats first took to the
skies. Bats apparently avoided both competition and predation from other
large flyers by specializing in nocturnal flight. The ecological and environ-
mental conditions were thus vastly different for each of these lineages as
they evolved flight. Nevertheless, they do have a few features in common.

MAJOR PAR ALLELS

The most obvious common feature of the Big Four flyers is flapping, which
I have used throughout the book interchangeably with powered flight. All
living powered flyers flap their wings for thrust, and everything we know
about pterosaurs suggests that they did as well. Flapping is really a
consequence of using muscles for power (Chapter 2). Muscles don’t allow
continuous rotation, so the next-best option is some sort of oscillating or
see-saw motion. Given the way muscles work and the way wings work,
flapping seems to be the only practical way an animal can produce thrust
for flight, at least in a low density medium like air.* Nonetheless, a transi-
tion from walking to flapping is not trivial. The right and left legs alter-
nate in typical walking, but they have to move simultaneously in flapping.
Interestingly, when my cat climbs a tree in a hurry, he does so with the
right and left front legs moving together, alternating with the back legs,
which also move together. (This is a variation of the “bounding” gait,
which is what locomotion researchers call the hopping gait used by ani-
mals like rabbits.) This kind of climbing might have served as an exapta-
tion for the original flapping motion, which is yet another point in favor
of arboreal climbers being ancestral to flyers.
Some scientists may disagree, but I see directed aerial descent leading
to gliding (“arboreal”) evolution of flight as a possible unifying feature. The
more researchers look, the more they find some form of aerial maneuvera-
bility in climbing animals. Cats always land on their feet, mice and arbo-
real lizards assume a sky-diving posture and make soft landings, and ants,
spiders, and silverfish knocked off a tree branch can steer their fall back to
the trunk. These animals all possess the ability to sense and in some way
influence their movement through the air during a fall. This ability gives
them the sensory and behavioral basis for rudimentary gliding without
any overt aerodynamic structures. Full-fledged gliding would only require
incremental enlargement of aerodynamic surfaces, which in turn explains
why so many kinds of arboreal animals have evolved gliding.
The preponderance of evidence now suggests that all the powered flyers
may have started by gliding from elevated perches. Most researchers agree
that bats and pterosaurs originated flight from the trees down. Some may
not yet accept a trees-down origin for bird flight, although I think the evi-
dence from many recently described feathered dinosaurs is quite compel-
ling. Finally, Yanoviak and colleagues developed the very concept of
directed aerial descent based on insects.1,2 Although the researchers are
hampered by lack of fossil evidence, their theory is so physically and

*  In air, a flyer’s wings need to support the animal’s weight as well as produce
thrust. In water, a swimmer’s weight is mostly supported by the water due to the ani-
mal’s buoyancy; therefore it only needs to produce thrust, not lift. Aquatic animals
can thus make use of intermittent, thrust-only mechanisms like squids’ use of inter-
mittent jetting. Nevertheless, the lift mechanism works just as well under water, and
in addition to sea lion, penguin, and sea turtle flippers that are obviously underwater
flapping wings, a fair number of fish tails—tunas, swordfish, sharks—operate as
wings flapping sideways instead of up and down.

[ 174 ]  On the Wing

biologically reasonable that I would be surprised if directed aerial descent
did not play at least some role in the evolution of flight in insects.
Any animal that flies, whether glider or flapper, must be able to sense
objects at a distance. Flight speeds demand that a flyer be able to detect
obstacles and landing sites long before they are close enough to touch and
with much more accuracy than possible with smell. Flying animals thus
tend to have acute vision. All flyers tend to have very good vision com-
pared to non-flyers, and birds of prey may have the sharpest vision in the
animal kingdom.3 Bats are the exception that proves the rule. Bats actu-
ally have above average vision for their size, but optical physics limits the
ability of small eyes to see detail in very dim light. So bats have evolved
echolocation to replace vision in the dark. Although echolocation can be
amazingly acute, it is a fairly short-range system, so bats revert back to
using vision whenever enough light is available.4
As a rule, modern flying animals are extremely maneuverable. This ma-
neuverability is a consequence of having little built-in or passive stability
(although one study suggests that the flapping motion itself might have
some stabilizing effect).5 The lack of passive stability means that a flying
animal can change directions quickly and easily. The flip side of this is that
if a flapping flyer intends to fly in a straight line, the animal will need to
make continuous, small, rapid course corrections to actively stabilize its
course. Some primitive flyers, such as Archaeopteryx with its long tail,
seem to have had considerable passive stability, but all modern flyers (as
well as the last main pterosaur lineage, the pterodactyloids) have evolved
toward high maneuverability at the expense of stability. This was surely a
response to predation because greater maneuverability makes a flyer
better at evading predators. Flying animals do cover a range of maneuver-
ability, with smaller, slower flyers tending to be more maneuverable and
larger, faster flyers less so. Even large, relatively ungainly flying animals
like swans or cranes are, however, still vastly more agile than even the
most maneuverable fighter jet or competition aerobatic airplane.
Gaining that maneuverability is not just a matter of discarding stabi-
lizing structures. The animal needs to replace stabilizing structures with
active course-correcting responses, and this means modifying the nerv-
ous system to continuously detect and correct for small deviations, typi-
cally using reflexes. So along with acute vision, flying animals all seem to
have evolved modifications to the nervous system to actively stabilize
flight, giving them extreme maneuverability.
Another theme, at least of the vertebrate flyers but less so for insects, is
that the specialized demands of flapping flight have greatly limited the
structural variation and diversity of flyers. As one ornithologist put it,

U nif y ing T hemes?  [ 175 ]

under their feathers, birds all look remarkably similar.6 Of course, within
birds, or within bats, some variation does exist. But if you think of the
variation in body form between a weasel and a domestic cow, or between a
rabbit and an armadillo and an elephant, flying vertebrates just don’t
show that much anatomical diversity. So much of their anatomy is special-
ized for flapping flight that these specializations greatly restrict major
modifications and variations. Indeed, this structural uniformity posed a
significant problem for researchers using anatomical traits to develop the
earliest bird phylogenies. Once gene-based phylogenies became available,
researchers realized that many of the structural characteristics that
seemed to unite various groups were actually convergences driven by the
physical constraints of flapping flight (see Box 10.1. Convergence and Ho-
mology, Trash or Treasure?).
Insects are freed from this constraint partly by their size and partly by
their modular body plan. Streamlining provides very little benefit for very
small animals (Chapter 3), which frees insects to adopt body shapes that

Box 10.1:  CONVERGENCE AND HOMOLOGY, TRASH

OR TREASURE?

As we compare animal flight across the Big Four powered flyers, two key
concepts surface repeatedly: convergence and homology. When scien-
tists look at two animals with similar structures that do the same
thing, they immediately want to know if the two structures evolved
independently due to similar selection pressures and constraints—­
convergence—or whether they are similar because they were inherited
in similar form from a common ancestor—homology. Homologies thus
arise from ancestry whereas convergences arise when different species
face similar constraints.
The constraints that lead to convergence are often physical. For ex-
ample, aerodynamics dictates the properties of effective wings, so
functional wings are normally big, flat, slightly cambered surfaces.
Similarly, flyers like vertebrates that operate at higher Reynolds num-
bers benefit greatly from streamlining. If we ignore the heads and legs
of birds, for instance, their bodies are all strikingly similar, that is,
compact and streamlined.
Physical constraints may be powerful, but animals are also heavily
constrained by their ancestry, so-called historical constraints. Evolution
can only work with the limited selection of building blocks present in a
lineage’s ancestors. For example, animal flyers cannot evolve a propeller
for thrust, because their muscles (and their ancestors’ muscles) do not

[ 176 ]  On the Wing

Box 10.1:  Continued

allow a continuously rotating body part. For this reason, flying animals
are constrained to flapping their wings for thrust. The same applies to an-
atomical properties. Bats are a nice illustration: the ancestors of bats did
not have feathers, so their wings make do with skin and bones instead.
Evolution by natural selection has no mechanism to produce massive
changes in structure or material composition in a single step. A jumping
animal cannot suddenly “decide” to grow full-sized wings the way
Frenchman Gabriel Poulain could bolt a pair of wings to a bicycle in the
1920s. Animals can evolve such large structures only gradually because
even extremely rapid evolution can still take on the order of a million
generations. Similarly, animals have no way to make a jump from the
structural materials that compose their ancestors’ bodies to totally new
and unrelated materials. No matter how much benefit they might gain,
no insect can evolve a graphite-epoxy exoskeleton, no bat can evolve ti-
tanium bones. Animals are limited to incremental changes in structures
built from a limited selection of materials present in their ancestors.
While convergence and homology are both important in evolution, a
biologist’s perspective on these concepts is shaped by the kinds of ques-
tions he or she is exploring. Functional biologists, such as those in my
own field of biomechanics, are often most interested in why different
animals work in similar ways. Why are the wings of both insects and
birds usually cambered? Why are the flapping patterns of all flying an-
imals so similar? Understanding aerodynamics helps us to understand
convergent wing structures and vice versa. Without evolutionary con-
vergences, we don’t have such comparisons, and homologies yield few
functional insights.
In contrast, systematists trying to reconstruct evolutionary history
use homologies as the main basis for building phylogenetic trees. Con-
vergences add confusion. If a systematist inadvertently treats a conver-
gence as a shared, derived characteristic, the resulting tree will be less
accurate, implying shared ancestry where none exists. So to system-
atists, homologies are valuable, whereas convergences need to be recog-
nized and eliminated. Recognizing convergent structures can be
challenging, but building trees with genetic as well as anatomical traits
has greatly improved the process.
Both functional biologists and systematists want to be able to rec-
ognize convergences and to be able to tell them from homologies. The
difference is that evolutionary convergences are valuable and informa-
tive to a functional biologist but detrimental and a source of error to a
systematist.7

U nif y ing T hemes?  [ 177 ]

would be too draggy for larger animals. Also, the flight mechanism is en-
tirely contained in one of the three body regions, the thorax, leaving the
head and abdomen fairly free from flight-related limits. Flight has thus con-
strained the thorax to some extent, but this has been a fairly minor restric-
tion on insect body form; this relaxed constraint allowed insects to evolve
much more anatomical variety than any lineage of flying vertebrates.

DIFFERENCES

Among the Big Four, insects are in a class by themselves. Where flying
insects probably diverge most from the flying vertebrates is in the origin
of the protowings. Whatever structure the earliest insect wing evolved
from, it was not a leg. Flying insects still have the same number of legs as
primitively flightless insects like silverfish. In contrast, natural selection
converted the front legs of birds into wings while leaving the back pair as
fully independent legs. Bats and pterosaurs took this a step further, incor-
porating the hindlimb into the wing structure and thus limiting their ter-
restrial versatility.
Insects also diverge from vertebrate flyers in size. The largest flying
insects—­cicadas, Goliath beetles, big dragonflies, and moths—do overlap
the size range of hummingbirds, but these animals represent the ex-
tremes. The vast majority of flying insects are much smaller than hum-
mingbirds, whereas the vast majority of birds and bats are larger than
hummingbirds. For a flyer, size matters. Tiny wings are inefficient, making
long-distance flight ineffective. Tiny flyers, however, have high surface-
to-volume ratios making vertical takeoff, maneuvering, and hovering
easy, and they have low inertia so collisions cause bounces, not injuries.
Large flyers face the opposite situation. Their more-efficient wings make
long-distance flight economical, but only the smallest can take off verti-
cally or hover, many require long takeoff runs, and collisions with solid
objects can be deadly. Within vertebrates, birds probably cover the widest
size range, bats tend toward the small end, pterosaurs tend toward the
large end, but they all overlap extensively.

STILL EVOLVING?

Throughout this book we have been looking at major evolutionary events,

and some readers may wonder if these animal groups are still evolving.
The answer is an emphatic “yes.”

[ 178 ]  On the Wing

 Natural selection is still acting, but the process is usually far too slow
for humans to observe directly. The loss of flight by rails on islands in
100,000 to 500,000 years (Chapter 9) is extremely rapid in evolutionary
terms and barely an eyeblink to a paleontologist. Yet that is also dozens of
times longer than the entire span of recorded human history, which shows
how unlikely we are to actually witness such significant evolutionary
events. The best we can do is look at patterns of relatedness and past spe-
ciation (in other words, phylogenies) and use those to decipher recent ev-
olutionary events. Such studies tell us that a good bit of speciation and
anatomical modification has occurred among flyers since modern humans
arose a million or so years ago. If we could look back in another 500,000 or
one million years, we would undoubtedly see that some of the groups that
seem to us to be one species of insect or bird are actually undergoing rapid
change and may be in the process of forming new species.

UNANSWERED QUESTIONS

Although tens of thousands (if not millions) of pages in scientific journals

have been devoted to the evolution of animal flight, we still face several
major unanswered questions. Our ignorance is due to those pesky gaps in
the fossil record. Until some of those gaps are filled in, we can place bound-
aries around what is physically or biologically possible, but we can never
conclusively answer certain questions.
Surely the most vexing unanswered questions are when and from what
precursors wings evolved in insects, pterosaurs, and bats. Thanks to those
abundant new Chinese fossils, we are fairly close to answering those ques-
tions for birds. The frustrating gaps, however, mean that we can’t really
pin down when wings evolved in the other three groups. Primitive, flight-
less insects put in a brief appearance very early, followed by a gap of tens
of millions of years, after which fully winged insects appear. As for bats
and pterosaurs, since we have no transition fossils, we cannot even say
with confidence who their direct ancestors were. We cannot tell whether
these groups were flying for an enormously long time or just a brief few
million years before the lifetime of our oldest fossils.
Similarly, without transition fossils or obvious pre-flight ancestors, the
exact function of the protowings—the structures that evolved directly
into wings—remains a mystery for insects, pterosaurs, and bats. Obvi-
ously pterosaur and bat wings started out as front limbs, but what were
the ancestral animals doing with those limbs? Probably not running, but
most likely climbing and eventually gliding. Without fossil evidence,

U nif y ing T hemes?  [ 179 ]

however, we can’t escape the possibility that they were doing something
entirely unexpected. And as for insects, the situation is even worse be-
cause researchers really don’t know what the protowing was, let alone
what it was used for before flight.
We have a pretty good handle on the ancestry of flying insects and
birds, but the direct ancestry of pterosaurs and bats remains an open
question. Bats are so distantly related to other living mammals that these
relationships are not that much help for figuring out their immediate an-
cestors. In contrast, the long association of pterosaurs with dinosaurs ap-
peared to make their ancestry fairly obvious, but that relationship seems
to be unraveling. If pterosaurs are not the sister group of dinosaurs, then
researchers need to do a lot more work to figure out who pterosaurs
evolved from.
Scientists have a long history of debating several “which came first,
chicken or egg?” questions about the order in which some flight-related
specializations evolved. For example, did flight drive the evolution of
feathers, or did feathers evolve first and then become co-opted for flight?
After many decades of discussion, the recent fossil discoveries in China
have pretty much answered this one: feathers evolved well before flight.
In contrast, several other sequence questions remain unanswered. A few
of the most important: Did insect wings evolve from a structure that al-
ready had a joint, or did the protowings first evolve their aerodynamic
function and then evolve a new joint to allow flapping? Did theropod di-
nosaurs evolve one-way lungs before birds arose, or did birds evolve flight
first and then evolve one-way lungs? Ditto for endothermy. Did bats
evolve echolocation before or after evolving powered flight? Either new
fossils or new ways of analyzing old fossils will be needed to answer these
questions.
All of these unanswered questions can be discouraging, but I am opti-
mistic. Scientists using the bounded-ignorance approach, especially when
based on biomechanics and functional morphology, have produced fresh
insights and innovative new theories. Modern phylogenetic methods have
greatly refined our understanding of bird evolution and could possibly do
the same for other groups. Thanks also to the treasure trove of wonderful
fossils described over the last decade from China, some of the major ques-
tions about flight evolution in birds have been definitely or partly an-
swered. While those fossil beds are too young to help us much with flight
origins in pterosaurs or insects, they do show how a new source of fossils
can revolutionize our view of the evolution of flight in one lineage. Per-
haps sources of fossils that can do the same for other animal groups await
discovery by a new generation of researchers.

[ 180 ]  On the Wing

REFERENCES
1. S. P. Yanoviak, R. Dudley, and M. Kaspari (2005) Nature.
2. S. P. Yanoviak, M. Kaspari, and R. Dudley (2009) Biology Letters.
3. R. Fox, S. W. Lehmkuhle, and D. H. Westendorf (1976) Science.
4. J. D. Altringham (2011) Bats: From Evolution to Conservation.
5. T. L. Hedrick (2011) Journal of Experimental Biology.
6. A. S. King and D. Z. King (1979) in Form and Function in Birds.
7. E. O. Wiley and B. S. Lieberman (2011). Phylogenetics: The Theory of Phylogenetic
Systematics.

U nif y ing T hemes?  [ 181 ]

BIBLIO GR A PH Y

Abzug, M. J. and E. E. Larrabee (1997). Airplane Stability and Control: A History of the
Technologies That Made Aviation Possible. Cambridge University Press, Cam-
bridge. 373 pp.
Alexander, D. E. (1986). Wind tunnel studies of turns by flying dragonflies. Journal
of Experimental Biology 122: 81–98.
Alexander, D. E. (2002). Nature’s Flyers: Birds, Insects, and the Biomechanics of Flight.
Johns Hopkins University Press, Baltimore, Maryland. 358 pp.
Alexander, D. E. (2009). Why Don’t Jumbo Jets Flap Their Wings? Flying Animals,
Flying Machines, and How They Are Different. Rutgers University Press, New
Brunswick, New Jersey. 278 pp.
Alexander, R. D. and W. L. Brown (1963). Mating behavior and the origin of insect
wings. Occasional Papers of the Museum of Zoology of the University of Michigan
628: 1–19.
Altringham, J. D. (1996). Bats: Biology and Behaviour. Oxford University Press,
Oxford. 262 pp.
Altringham, J. D. (2011). Bats: From Evolution to Conservation. Oxford University
Press, New York. 324 pp.
Anderson, J. D. (2007). Fundamentals of Aerodynamics. McGraw-Hill Higher Educa-
tion, Boston. 1008 pp.
Bailey, W., J. Slightom and M. Goodman (1992). Rejection of the “flying primate”
hypothesis by phylogenetic evidence from the epsilon-globin gene. Science
256: 86–89.
Baker, R. J., M. J. Novacek and N. B. Simmons (1991). On the monophyly of bats.
Systematic Zoology 40: 216–231.
Bakker, R. T. (1972). Anatomical and ecological evidence of endothermy in dino-
saurs. Nature 238: 81–85.
Beck, R. M. D., O. R. P. Bininda-Emonds, M. Cardillo, F.-G. R. Liu, and A. Purvis
(2006). A higher-level MRP supertree of placental mammals. BMC Evolution-
ary Biology 6: 93.
Beckemeyer, R. J. (2000). The Permian insect fossils of Elmo, Kansas. Kansas School
Naturalist 46: 1–15.
Bennett, S. C. (1995). A statistical study of Rhamphorhynchus from the southern
limestone of Germany: Year-classes of a single large species. Journal of Paleon-
tology 69: 569–580.
Bennett, S. C. (1996). The phylogenetic position of the Pterosauria within the Ar-
chosauromorpha. Zoological Journal of the Linnean Society 118: 261–308.
Bennett, S. C. (1996). Year-classes of pterosaurs from the Solnhofen limestone of
Germany: Taxonomic and systematic implications. Journal of Vertebrate Pale-
ontology 16: 432–444.
Bennett, S. C. (1997). The arboreal leaping theory of the origin of pterosaur flight.
Historical Biology 12: 265–290.
Bennett, S. C. (1997). Terrestrial locomotion of pterosaurs: A reconstruction based
on Pteraichnus trackways. Journal of Vertebrate Paleontology 17: 104–113.
Bennett, S. C. (2000). Pterosaur flight: The role of actinofibrils in wing function.
Historical Biology 14: 255–284.
Bennett, S. C. (2013). The phylogenetic position of the Pterosauria within the Ar-
chosauromorpha re-examined. Historical Biology 25: 545–563.
Bennett, A. F. and J. A. Ruben (1979). Endothermy and activity in vertebrates. Sci-
ence 206: 649–654.
Benson, R. B. J., R. J. Butler, M. T. Carrano and P. M. O’Connor (2012). Air-filled
postcranial bones in theropod dinosaurs: Physiological implications and the
reptile-bird transition. Biological Reviews 87: 168–193.
Benton, M. J. (1985). Classification and phylogeny of the diapsid reptiles. Zoological
Journal of the Linnean Society 84: 97–164.
Benton, M. J. (1999). Scleromochlus taylori and the origin of dinosaurs and ptero-
saurs. Philosophical Transactions of the Royal Society of London Series B-­
Biological Sciences 354: 1423–1446.
Bertin, J. J. and M. L. Smith (1979). Aerodynamics for Engineers. Prentice-Hall, En-
glewood Cliffs, New Jersey. 410 pp.
Béthoux, O. and D. E. G. Briggs (2008). How Gerarus lost its head: Stem-group Or-
thoptera and Paraneoptera revisited. Systematic Entomology 33: 529–547.
Bishop, K. L. (2008). The evolution of flight in bats: Narrowing the field of plausible
hypotheses. Quarterly Review of Biology 83: 153–169.
Boag, D. and M. Alexander (1986). The Atlantic Puffin. Poole (for Blandford Press,
Australia), New York. 128 pp.
Bock, W. J. (1965). The role of adaptive mechanisms in the origin of higher levels of
organization. Systematic Zoology 14: 272–287.
Bock, W. J. (1969). The origin and radiation of birds. Annals of the New York Academy
of Sciences 167: 147–155.
Brusatte, S. L., M. J. Benton, J. B. Desojo and M. C. Langer (2010). The higher-level
phylogeny of Archosauria (Tetrapoda: Diapsida). Journal of Systematic Palae-
ontology 8: 3–47.
Bryant, H. N. and A. P. Russell (1993). The occurrence of clavicles within Dinosau-
ria: Implications for the homology of the avian furcula and the utility of neg-
ative evidence. Journal of Vertebrate Paleontology 13: 171–184.
Bundle, M. W. and K. P. Dial (2003). Mechanics of wing-assisted incline running
(WAIR). Journal of Experimental Biology 206: 4553–4564.
Burgers, P. and L. M. Chiappe (1999). The wing of Archaeopteryx as a primary thrust
generator. Nature 399: 60–62.
Butler, R. J., P. M. Barrett and D. J. Gower (2009). Postcranial skeletal pneumaticity
and air-sacs in the earliest pterosaurs. Biology Letters 5: 557–560.
Butler, R. J., P. M. Barrett and D. J. Gower (2012). Reassessment of the evidence for
postcranial skeletal pneumaticity in Triassic archosaurs, and the early evolu-
tion of the avian respiratory system. PLoS One 7: e34094.
Byers, G. W. (1969). Evolution of wing reduction in crane flies (Diptera—Tipulidae).
Evolution 23: 346–354.

[ 184 ] Bibliography
Byrnes, G. and A. J. Spence (2011). Ecological and biomechanical insights into the evo-
lution of gliding in mammals. Integrative and Comparative Biology 51: 991–1001.
Campbell, K. E. and E. P. Tonni (1983). Size and locomotion in teratorns (Aves, Tera-
tornithidae). Auk 100: 390–403.
Caple, G. R., R. T. Balda and W. R. Willis (1983). The physics of leaping animals and
the evolution of pre-flight. American Naturalist 121: 455–467.
Carpenter, G. H. (1913). The Life-Story of Insects. G. P. Putnam’s Sons, New York.
134 pp.
Carpenter, F. M. (1992). Superclass Hexapoda. In Treatise on Invertebrate Paleontol-
ogy, Part R, Arthopoda 3–4, R. L. Kaesler, Ed. Geological Society of America,
Boulder, Colorado. 655 pp.
Chapman, R. F. (1982). The Insects: Structure and Function. Harvard University Press,
Cambridge, Massachusetts. 919 pp.
Chatterjee, S. and R. J. Templin (2004). Posture, locomotion and paleoecology of
pterosaurs. Geological Society of America Special Papers 376: 1–64.
Chen, P. J., Z. M. Dong and S. N. Zhen (1998). An exceptionally well-preserved the-
ropod dinosaur from the Yixian Formation of China. Nature 391: 147–152.
Chiappe, L. M. and C. A. Walker (2002). Skeletal morphology and systematics of the
Cretaceous Euenantiornithes (Ornithothoraces: Enantiornithese). In Meso-
zoic Birds: Above the Heads of Dinosaurs, L. M. Chiappe and L. M. Witmer, Eds.
University of California Press, Berkeley, California. pp. 240–267.
Christiansen, P. and N. Bonde (2000). Axial and appendicular pneumaticity in Ar-
chaeopteryx. Proceedings of the Royal Society of London Series B-Biological Sci-
ences 267: 2501–2505.
Clarke, A. and H. O. Pörtner (2010). Temperature, metabolic power and the evolu-
tion of endothermy. Biological Reviews 85: 703–727.
Colbert, E. H. (1967). Adaptations for gliding in the lizard Draco. American Museum
Novitates No. 2283: 1–20.
Colbert, E. H. (1970). The Triassic gliding reptile Icarosaurus. Bulletin of the American
Museum of Natural History 143: 89–142.
Collini, C. A. (1784). Sur quelques zoolithes du Cabinet d’Histoire Naturelle de
S.A.S.E. Palatine et de Baviére, à Mannheim. Acta Academiae Theodoro-­
Palatinae, Mannheim, Pars Physica 5: 58–103.
Craik, S. R., J.-P. L. Savard and R. D. Titman (2009). Wing and body molts of male
Red-breasted Mergansers in the Gulf of St. Lawrence, Canada. Condor
111: 71–80.
Currie, P. J. (1991). The Flying Dinosaurs. Red Deer College Press, Red Deer, Alberta,
Canada. 160 pp.
Cuvier, G. (1801). Extrait d’un ouvrage sur les espèces de quadrupèdes dont on a
trouvé les ossmens dans l’intérieur de la terre. Journal de Physique, de Chimie
et d’Histoire Naturelle 52: 253–267.
Cuvier, G. (1809). Mémoire sur le squelette fossile d’un reptile volant des environs
d’Aichstedt, que quelques naturalistes ont pris pour un oiseau, et dont nous
formons un genre de Sauriens, sous le nom de Ptero-Dactyle. Annales du
Muséum national d’Histoire naturelle, Paris 13: 424–437.
Darwin, C. (1860). On the Origin of Species. J. Murray, London. 502 pp.
Davenport, J. (1994). How and why do flying fish fly? Reviews in Fish Biology and
Fisheries 4: 184–214.
Dial, K. P. (2003). Wing-assisted incline running and the evolution of flight. Science
299: 402–404.

Bibliography  [ 185 ]
Dial, K. P., B. E. Jackson and P. Segre (2008). A fundamental avian wing-stroke pro-
vides a new perspective on the evolution of flight. Nature 451: 985–989.
Dial, K. P., R. J. Randall and T. R. Dial (2006). What use is half a wing in the ecology
and evolution of birds? BioScience 56: 437–445.
Douglas, M. M. (1981). Thermoregulatory significance of thoracic lobes in the evolu-
tion of insect wings. Science 211: 84–86.
Dudley, R. (2000). The Biomechanics of Insect Flight: Form, Function, Evolution. Princ-
eton University Press, Princeton, New Jersey. 476 pp.
Dudley, R., G. Byrnes, S. P. Yanoviak, B. Borrell, R. M. Brown and J. A. McGuire
(2007). Gliding and the functional origins of flight: Biomechanical novelty or
necessity? In Annual Review of Ecology, Evolution, and Systematics. Annual Re-
views, Palo Alto. vol. 38, pp. 179–201.
Dunnigan, J. F. (2003). How to Make War: A Comprehensive Guide to Modern Warfare
in the Twenty-first Century. HarperCollins, New York. 672 pp.
Dyke, G. J., R. L. Nudds and J. M. V. Rayner (2006). Flight of Sharovipteryx mirabilis:
The world’s first delta-winged glider. Journal of Evolutionary Biology 19:
1040–1043.
Edwards, J. S. (1986). Arthropods as pioneers: Recolonization of the blast zone on
Mt. St. Helens. Northwest Environmental Journal 2: 63–73.
Elgin, R. A., D. W. E. Hone and E. Frey (2011). The extent of the pterosaur flight
membrane. Acta Palaeontologica Polonica 56: 99–111.
Engel, M. S., S. R. Davis and J. Prokop (2013). Insect wings: The evolutionary devel-
opmental origins of Nature’s first flyers. In Arthropod Biology and Evolution:
Molecules, Development, Morphology. A. Minelli, G. Boxshall, and G. Fusco,
Eds. Springer Verlag, Berlin. pp. 269–298.
Engel, M. S. and D. A. Grimaldi (2004). New light shed on the oldest insect. Nature
427: 627–630.
Evans, J. (1865). One portion of a cranium and of a jaw in the slab containing the
fossil remains of Archaeopteryx. Natural History Review, series 2 5: 415–421.
Evenhuis, N. L. (1997). Review of flightless Dolichopodidae (Diptera) in the Hawai-
ian Islands. Bishop Museum Occasional Papers 53: 1–30.
Farmer, C. G. and K. Sanders (2010). Unidirectional airflow in the lungs of alliga-
tors. Science 327: 338–340.
Feduccia, A. (1996). The Origin and Evolution of Birds. Yale University Press, New
Haven, Connecticut. 420 pp.
Feduccia, A., T. Lingham-Soliar and J. R. Hinchliffe (2005). Do feathered dinosaurs
exist? Testing the hypothesis on neontological and paleontological evidence.
Journal of Morphology 266: 125–166.
Fenton, M. B., D. Audet, M. K. Obrist and J. Rydell (1995). Signal strength, timing, and
self-deafening—The evolution of echolocation in bats. Paleobiology 21: 229–242.
Fisher, H. I. (1957). Bony mechanism of automatic flexion and extension in the pi-
geon’s wing. Science 126: 446.
Flower, J. W. (1964). On the origin of flight in insects. Journal of Insect Physiology
10: 81–88.
Forbes, W. T. M. (1943). The origin of wings and venational types in insects. Ameri-
can Midland Naturalist 29: 381–405.
Forsman, K. A. and M. G. Malmquist (1988). Evidence for echolocation in the
common shrew Sorex araneus. Journal of Zoology (London) 216: 655–662.
Fox, R., S. W. Lehmkuhle and D. H. Westendorf (1976). Falcon visual acuity. Science
192: 263–265.

[ 186 ] Bibliography
Fraser, N. C., P. E. Olsen, A. C. Dooley Jr. and T. R. Ryan (2007). A new gliding tetra-
pod (Diapsida: ?Archosauromorpha) from the Upper Triassic (Carnian) of
Virginia. Journal of Vertebrate Paleontology 27: 261–265.
Frey, E., H.-D. Sues and W. Munk (1997). Gliding mechanism in the Late Permian
reptile Coelurosauravus. Science 275: 1450–1452.
Garner, J. P., G. K. Taylor and A. L. R. Thomas (1999). On the origins of birds: The
sequence of character acquisition in the evolution of avian flight. Proceedings
of the Royal Society of London Series B-Biological Sciences 266: 1259–1266.
Garrouste, R., G. Clement, P. Nel, M. S. Engel, P. Grandcolas, C. D’Haese, L. Lagebro,
J. Denayer, P. Gueriau, P. Lafaite, S. Olive, C. Prestianni and A. Nel (2012). A
complete insect from the Late Devonian period. Nature 488: 82–85.
Gauthier, J. A. (1986). Saurischian monophyly and the origin of birds. In The Origin
of Birds and the Evolution of Flight, K. Padian, Ed. California Academy of Sci-
ence, San Francisco. pp. 1–55.
Gegenbaur, C. (1870). Grundzüge der vergleichenden Anatomie. Wilhelm Engelmann,
Leipzig. 890 pp.
Gegenbaur, C. (1878). Elements of Comparative Anatomy. Macmillan and Co., London.
645 pp.
Giannini, N. P. (2012). Toward an integrative theory on the origin of bat flight. In
Evolutionary History of Bats: Fossils, Molecules and Morphology, G. F. Gunnell
and N. B. Simmons, Eds. Cambridge University Press, Cambridge.
pp. 353–384.
Giebel, C. (1877). Neueste Entdeckung einer zweiten Archaeopteryx lithographica.
Zeitschrift für die Gesammten Naturwissenschaften 49: 326–327.
Godefroit, P., A. Cau, H. Dong-Yu, F. Escuillie, W. Wenhao and G. Dyke (2013). A
Jurassic avialan dinosaur from China resolves the early phylogenetic history
of birds. Nature 498: 359–362.
Gould, E., N. C. Negus and A. Novick (1964). Evidence for echolocation in shrews.
Journal of Experimental Zoology 156: 19–37.
Gould, S. J. (1985). Not necessarily a wing. Natural History 94: 12–25.
Grimaldi, D. A. (2010). 400 million years on six legs: On the origin and early evolu-
tion of Hexapoda. Arthropod Structure & Development 39: 191–203.
Grimaldi, D. A. and M. S. Engel (2005). Evolution of the Insects. Cambridge Univer-
sity Press, New York. 755 pp.
Gullan, P. J. and P. S. Cranston (2010). The Insects: An Outline of Entomology. Wiley-
Blackwell, Hoboken, New Jersey. 565 pp.
Hasenfuss, I. (2002). A possible evolutionary pathway to insect flight starting from
lepismatid organization. Journal of Zoological Systematics and Evolutionary
Research 40: 65–81.
Hedrick, T. L. (2011). Damping in flapping flight and its implications for manoeu-
vring, scaling and evolution. Journal of Experimental Biology 214: 4073–4081.
Heilmann, G. (1927). The Origin of Birds. D. Appleton, New York. 209 pp.
Heinicke, M. P., E. Greenbaum, T. R. Jackman and A. M. Bauer (2012). Evolution of
gliding in Southeast Asian geckos and other vertebrates is temporally con-
gruent with dipterocarp forest development. Biology Letters 8: 994–997.
Henderson, D. M. (2010). Pterosaur body mass estimates from three-dimensional
mathematical slicing. Journal of Vertebrate Paleontology 30: 768–785.
Hennig, W. (1981). Insect Phylogeny. John Wiley, New York. 514 pp.
Heptonstall, W. B. (1971). An analysis of the flight of the Cretaceous pterodactyl
Pteranodon ingens (March). Scottish Journal of Geology 7: 61–78.

Bibliography  [ 187 ]
Hone, D. W. E. (2012). Pterosaur research: Recent advances and a future revolution.
Acta Geologica Sinica-English Edition 86: 1366–1376.
Hone, D. W. E. and M. J. Benton (2007). An evaluation of the phylogenetic relation-
ships of the pterosaurs among archosauromorph reptiles. Journal of System-
atic Palaeontology 5: 465–469.
Hone, D. W. E. and D. M. Henderson (2014). The posture of floating pterosaurs: Eco-
logical implications for inhabiting marine and freshwater habitats. Palaeo-
geography, Palaeoclimatology, Palaeoecology 394: 89–98.
Hornschemeyer, T., J. T. Haug, O. Béthoux, R. G. Beutel, S. Charbonnier, T. A.
Hegna, M. Koch, J. Rust, S. Wedmann, S. Bradler and R. Willmann (2013). Is
Strudiella a Devonian insect? Nature 494: E3–E4.
Hu, D., L. Hou, L. Zhang and X. Xu (2009). A pre-Archaeopteryx troodontid theropod
from China with long feathers on the metatarsus. Nature 461: 640–643.
Huene, F.v. (1914). Beitrage zur Geschichte der Archosaurier. Geologische und Paläon-
tologische Abhandlungen N.F. 13: 1–53.
Hutcheon, J. M., J. A. W. Kirsch and J. D. Pettigrew (1998). Base-compositional
biases and the bat problem. III. The question of microchiropteran monophyly.
Philosophical Transactions of the Royal Society of London Series B-Biological Sci-
ences 353: 607–617.
Jepsen, G. L. (1966). Early Eocene bat from Wyoming. Science 154: 1333–1339.
Jex, H. R. (2000). Making pterodactyls fly (QN Story). Published by TWITT (The
Wing Is the Thing). Retrieved 5/12/2008, from http://www.twitt.org/
QNStory.html.
Ji, Q., P. J. Currie, M. A. Norell and S.-A. Ji (1998). Two feathered dinosaurs from
northeastern China. Nature 393: 753–761.
Jones, K. J. and H. H. Genoways (1970). Chiropteran systematics. In About Bats: A
Chiropteran Symposium, R. H. Slaughter and D. W. Walton, Eds. Southern
Methodist University Press, Dallas. pp. 3–21.
Jusufi, A., Y. Zeng, R. J. Full and R. Dudley (2011). Aerial righting reflexes in flight-
less animals. Integrative and Comparative Biology 51: 937–943.
King, A. S. and D. Z. King (1979). Avian morphology: General principles. In Form
and Function in Birds, A. S. King and J. McLelland, Eds. Academic Press, New
York. vol. 1, pp. 1–38.
Kingsolver, J. G. and M. A. R. Koehl (1985). Aerodynamics, thermoregulation, and
the evolution of insect wings: Differential scaling and evolutionary change.
Evolution 39: 488–504.
Kingsolver, J. G. and M. A. R. Koehl (1994). Selective factors in the evolution of
insect wings. Annual Review of Entomology 39: 425–451.
Koopman, K. F. (1984). A synopsis of the families of bats. Part VII. Bat Research
News 25: 25–27.
Kukalova-Peck, J. (1978). Origin and evolution of insect wings and their relation to
metamorphosis, as documented by fossil record. Journal of Morphology 156:
53–125.
Kukalova-Peck, J. (1983). Origin of the insect wing and wing articulation from the
arthropodan leg. Canadian Journal of Zoology 61: 1618–1669.
Kukalova-Peck, J. (1985). Ephemeroid wing venation based upon new gigantic Car-
boniferous mayflies and basic morphology phylogeny and metamorphosis of
pterygote insects (Insecta, Ephemerida). Canadian Journal of Zoology 63:
933–955.

[ 188 ] Bibliography
Kukalova-Peck, J. (1997). Arthropod phylogeny and “basal” morphological struc-
tures. In Arthropod Relationships, R. A. Fortey and R. H. Thomas, Eds. Chap-
man and Hall, London. pp. 269–279.
Lee, M. S. Y. and T. H. Worthy (2012). Likelihood reinstates Archaeopteryx as a prim-
itive bird. Biology Letters 8: 299–303.
Li, P.-P., K.-Q. Gao, L.-H. Hou and X. Xu (2007). A gliding lizard from the Early Cre-
taceous of China. Proceedings of the National Academy of Sciences of the United
States of America 104: 5507–5509.
Lingham-Soliar, T. (2003). The dinosaurian origin of feathers: Perspectives from
dolphin (Cetacea) collagen fibers. Naturwissenschaften 90: 563–567.
Lingham-Soliar, T. (2010). Dinosaur protofeathers: Pushing back the origin of
feathers into the Middle Triassic? Journal of Ornithology 151: 193–200.
Lü, J. C., D. M. Unwin, D. C. Deeming, X. S. Jin, Y. Q. Liu and Q. A. Ji (2011). An
egg-adult association, gender, and reproduction in pterosaurs. Science 331:
321–324.
Maina, J. N. (2000). What it takes to fly: The structural and functional respiratory
refinements in birds and bats. Journal of Experimental Biology 203:
3045–3064.
Maiorana, V. C. (1979). Why do adult insects not moult? Biological Journal of the Lin-
nean Society 11: 253–258.
Makovicky, P. J. and P. J. Currie (1998). The presence of a furcula in tyrannosaurid
theropods, and its phylogenetic and functional implications. Journal of Verte-
brate Paleontology 18: 143–149.
Marden, J. H. (2003). The surface-skimming hypothesis for the evolution of insect
flight. Acta Zoologica Cracoviensia 46: 73–84.
Marden, J. H. and M. G. Kramer (1994). Surface-skimming stoneflies: A possible
intermediate stage in insect flight evolution. Science 266: 427–430.
Marden, J. H. and M. G. Kramer (1995). Locomotor performance of insects with
rudimentary wings. Nature 377: 332–334.
Marden, J. H., M. R. Wolf and K. E. Weber (1997). Aerial performance of Drosophila
melanogaster from populations selected for upwind flight ability. Journal of
Experimental Biology 200: 2747–2755.
Martin, L. D. (1983). The origin and early radiation of birds. In Perspectives in Orni-
thology, A. H. Brush and G. A. Clark Jr., Eds. Cambridge University Press,
London. pp. 291–353.
Martin, L. D., J. D. Stewart and K. N. Whetstone (1980). The origin of birds: Struc-
ture of the tarsus and teeth. Auk 97: 86–93.
Maynard Smith, J. (1952). The importance of the nervous system in the evolution of
animal flight. Evolution 6: 127–129.
McGuire, J. A. and R. Dudley (2011). The biology of gliding in flying lizards (genus
Draco) and their fossil and extant analogs. Integrative and Comparative Biology
51: 983–990.
McMahon, T. A. (1984). Muscles, Reflexes, and Locomotion. Princeton University
Press, Princeton, New Jersey. 331 pp.
McNab, B. K. (1994). Energy conservation and the evolution of flightlessness in
birds. American Naturalist 144: 628–642.
Medeiros, M. J. (2008). A new species of flightless, jumping, alpine moth of the
genus Thyrocopa from Hawaii (Lepidoptera: Xyloryctidae: Xyloryctinae). Zoo-
taxa 1830: 57–62.

Bibliography  [ 189 ]
Meng, J., Y. Hu, Y. Wang, X. Wang and C. Li (2006). A mesozoic gliding mammal
from northeastern China. Nature 444: 889–893.
Müller, F. (1873). Beiträge zur Kenntniss der Termiten. Jenaische Zeitschrift für
Naturwissenschaft 7: 333–358; 451–463.
Müller, F. (1875). Beiträge zur Kenntniss der Termiten. Jenaische Zeitschrift für
Naturwissenschaft 9: 241–264.
Murphy, W. J., E. Eizirik, W. E. Johnson, Y. P. Zhang, O. A. Ryder and S. J. O’Brien
(2001). Molecular phylogenetics and the origins of placental mammals.
Nature 409: 614–618.
Nesbitt, S. J. (2011). The early evolution of archosaurs: Relationships and the origin
of major clades. Bulletin of the American Museum of Natural History
352: 1–292.
Nettleship, D. N. and P. G. H. Evans (1985). Distribution and status of the Altlantic
Alcidae. In The Atlantic Alcidae: The Evolution, Distribution and Biology of the
Auks Inhabiting the Atlantic Ocean and Adjacent Water Areas, D. N. Nettleship
and T. R. Birkhead, Eds. Academic Press, New York. pp. 53–154.
Neuweiler, G. (2000). The Biology of Bats. Oxford University Press, New York. 310 pp.
Newman, B. G., S. B. Savage and D. Schoulla (1977). Model tests on a wing section
of an Aeschna dragonfly. In Scale Effects in Animal Locomotion, T. J. Pedley, Ed.
Academic Press, New York. pp. 445–477.
Newton, I. (2008). The Migration Ecology of Birds. Academic Press, Boston. 976 pp.
Norberg, U. M. (1985). Evolution of flight in birds: Aerodynamic, mechanical and
ecological aspects. In The Beginnings of Birds, M. K. Hecht, J. H. Ostrom, G.
Viohl, and P. Wellnhofer, Eds. Freunde des Jura-Museums Eichstätt, Eich-
stätt, Germany. pp. 293–303.
Norberg, U. M. (1985). Evolution of vertebrate flight: An aerodynamic model for
the transition from gliding to active flight. American Naturalist 126:
303–327.
Norberg, U. M. (1990). Vertebrate Flight: Mechanics, Physiology, Morphology, Ecology
and Evolution. Springer-Verlag, Berlin. 291 pp.
Novacek, M. J. (1985). Evidence for echolocation in the oldest known bats. Nature
315: 140–141.
Novacek, M. J. (2001). Mammalian phylogeny: Genes and supertrees. Current Biol-
ogy 11: R573–R575.
Novacek, M. J. and A. R. Wyss (1986). Higher-level relationships of the recent eu-
therian orders: Morphological evidence. Cladistics 2: 257–287.
Nudds, R. L. and G. J. Dyke (2009). Forelimb posture in dinosaurs and the evolution
of the avian flapping flight-stroke. Evolution 63: 994–1002.
O’Connor, P. M. (2009). Evolution of archosaurian body plans: Skeletal adaptations
of an air-sac-based breathing apparatus in birds and other archosaurs. Jour-
nal of Experimental Zoology Part a-Ecological Genetics and Physiology 311A:
629–646.
Oliver, J. A. (1951). “Gliding” in amphibians and reptiles, with a remark on an arbo-
real adaptation in the lizard, Anolis carolinensis Voigt. American Naturalist 85:
171–176.
Ostrom, J. H. (1975). The origin of birds. Annual Review of Earth and Planetary Sci-
ences 3: 55–77.
Ostrom, J. H. (1976). Archaeopteryx and the origin of birds. Biological Journal of the
Linnean Society 8: 91–182.

[ 190 ] Bibliography
Padian, K. (1982). Running, leaping, lifting off—flight evolved from the ground up,
not the trees down. Sciences-New York 22: 10–15.
Padian, K. (1983). A functional analysis of flying and walking in pterosaurs. Paleobi-
ology 9: 218–239.
Padian, K. (1984). The origin of pterosaurs. In Third Symposium on Mesozoic Terres-
trial Ecosystems, W. E. Reif and F. Westfall, Eds. Attempto Verlag, Tübingen.
pp. 163–168.
Padian, K. (1985). The origins and aerodynamics of flight in extinct vertebrates.
Palaeontology 28: 413–434.
Padian, K. (1991). Pterosaurs: Were they functional birds or functional bats. In Bio-
mechanics in Evolution, J. M. V. Rayner and R. J. Wootton, Eds. Cambridge
University Press, Cambridge. pp. 146–160.
Padian, K. and J. M. V. Rayner (1993). The wings of pterosaurs. American Journal of
Science 293A: 91–166.
Paul, G. S. (2002). Dinosaurs of the Air: The Evolution and Loss of Flight in Dinosaurs
and Birds. Johns Hopkins University Press, Baltimore. 460 pp.
Pennycuick, C. J. (1972). Animal Flight. Edward Arnold, London. 68 pp.
Pennycuick, C. J., T. Alerstam and B. Larsson (1979). Soaring migration of the
common crane Grus grus observed by radar and from an aircraft. Ornis Scan-
dinavica 10: 241–251.
Peterson, K., P. Birkmeyer, R. Dudley and R. S. Fearing (2011). A wing-assisted run-
ning robot and implications for avian flight evolution. Bioinspiration & Biomi-
metics 6: 046008.
Pettigrew, J. D. (1986). Flying primates? Megabats have the advanced pathway from
eye to midbrain. Science 231: 1304–1306.
Pettigrew, J. D. (1991). A fruitful, wrong hypothesis—Response. Systematic Zoology
40: 231–239.
Pettigrew, J. D. (1991). Wings or brain—convergent evolution in the origins of bats.
Systematic Zoology 40: 199–216.
Pettigrew, J. D. (1995). Flying primates: Crashed, or crashed through? In Ecology,
Evolution and Behaviour of Bats, P. A. Racey and S. M. Swift, Eds. pp. 3–26.
Pettigrew, J. D., B. G. M. Jamieson, S. K. Robson, L. S. Hall, K. I. McAnally and
H. M. Cooper (1989). Phylogenetic relations between microbats, megabats
and primates (Mammalia: Chiroptera and Primates). Philosophical Transac-
tions of the Royal Society of London Series B-Biological Sciences 325: 489–559.
Powers, L. V., S. C. Kandarian and T. H. Kunz (1991). Ontogeny of flight in the little
brown bat, Myotis lucifugus— behavior, morphology, and muscle histochem-
istry. Journal of Comparative Physiology a-Sensory Neural and Behavioral Physiol-
ogy 168: 675–685.
Rayner, J. M. V. (1985a). Mechanical and ecological constraints on flight evolution.
In The Beginnings of Birds, M. K. Hecht, J. H. Ostrom, G. Viohl and P. Well­
nhofer, Eds. Freunde des Jura-Museums Eichstätt, Eichstätt, Germany.
pp. 279–288.
Rayner, J. M. V. (1985b). Cursorial gliding in protobirds: An expanded version of a
discussion contribution. In The Beginnings of Birds, M. K. Hecht, J. H. Ostrom,
G. Viohl, and P. Wellnhofer, Eds. Freunde des Jura-Museums Eichstätt, Eich-
stätt, Germany. pp. 289–292.
Rayner, J. M. V. (1988). The evolution of vertebrate flight. Biological Journal of the
Linnean Society 34: 269–287.

Bibliography  [ 191 ]
Rayner, J. M. V. (1988). Form and function of avian flight. Current Ornithology
5: 1–66.
Rayner, J. M. V. (1989). Vertebrate flight and the origins of flying vertebrates. In
Evolution and the Fossil Record, K. Allen and D. Briggs, Eds. Belhaven Press,
London. pp. 188–217.
Rayner, J. M. V. (1999). Estimating power curves of flying vertebrates. Journal of
Experimental Biology 202: 3449–3461.
Rees, C. J. C. (1975). Aerodynamic properties of an insect wing section and a
smooth aerofoil compared. Nature 258: 141–142.
Renesto, S. and G. Binelli (2006). Vallesaurus cenensis Wild, 1991, a drepanosaurid
(Reptilia, Diapsida) from the late Triassic of northern Italy. Rivista Italiana Di
Paleontologia E Stratigrafia 112: 77–94.
Robinson, P. L. (1967). Triassic vertebrates from lowland and upland. Science & Cul-
ture 33: 169–173.
Roff, D. A. (1984). The cost of being able to fly—a study of wing polymorphism in 2
species of crickets. Oecologia 63: 30–37.
Roff, D. A. (1990). The evolution of flightlessness in insects. Ecological Monographs
60: 389–421.
Ross, A. J. (2010). A review of the Carboniferous fossil insects from Scotland. Scot-
tish Journal of Geology 46: 157–168.
Ruben, J. A., T. D. Jones and N. R. Geist (2003). Respiratory and reproductive paleo-
physiology of dinosaurs and early birds. Physiological and Biochemical Zoology
76: 141–164.
Russell, A. P., L. D. Dijkstra and G. L. Powell (2001). Structural characteristics of
the patagium of Ptychozoon kuhli (Reptilia: Gekkonidae) in relation to para-
chuting locomotion. Journal of Morphology 247: 252–263.
Schmidt-Nielsen, K. (1972). Locomotion: Energy cost of swimming, flying and run-
ning. Science 177: 222–227.
Schmidt-Nielsen, K. (1990). Animal Physiology: Adaptation and Environment. Cam-
bridge University Press, Cambridge. 602 pp.
Seebacher, F. (2003). Dinosaur body temperatures: The occurrence of endothermy
and ectothermy. Paleobiology 29: 105–122.
Seeley, H. G. (1901). Dragons of the Air, an Account of Extinct Flying Reptiles. Methuen
& Co., London. 239 pp.
Shear, W. A., P. M. Bonamo, J. D. Grierson, W. D. I. Rolfe, E. L. Smith and R. A.
Norton (1984). Early land animals in North America: Evidence from Devo-
nian age arthropods from Gilboa, New York. Science 224: 492–494.
Shipman, P. (1998). Taking Wing: Archaeopteryx and the Evolution of Bird Flight. Simon
& Schuster, New York. 336 pp.
Shoshani, J. and M. C. McKenna (1998). Higher taxonomic relationships among
extant mammals based on morphology, with selected comparisons of re-
sults from molecular data. Molecular Phylogenetics and Evolution 9:
572–584.
Simmons, N. B. (1995). Bat relationships and the origin of flight. Symposium of the
Zoological Society of London 67: 27–43.
Simmons, N. B. and J. H. Geisler (1998). Phylogenetic relationships of Icaronycteris,
Archaeonycteris, Hassianycteris, and Palaeochiropteryx to extant bat lineages,
with comments on the evolution of echolocation and foraging strategies in
Microchiroptera. Bulletin of the American Museum of Natural History
235: 1–182.

[ 192 ] Bibliography
Simmons, N. B., M. J. Novacek and R. J. Baker (1991). Approaches, methods, and
the future of the chiropteran monophyly controversy—Reply. Systematic Zo-
ology 40: 239–243.
Simmons, N. B. and T. H. Quinn (1994). Evolution of the digital tendon locking
mechanism in bats and dermopterans: A phylogenetic perspective. Journal of
Mammalian Evolution 2: 231–254.
Simmons, N. B., K. L. Seymour, J. Habersetzer and G. F. Gunnell (2008). Primitive
Early Eocene bat from Wyoming and the evolution of flight and echolocation.
Nature 451: 818–821.
Slikas, B., S. L. Olson and R. C. Fleischer (2002). Rapid, independent evolution of
flightlessness in four species of Pacific Island rails (Rallidae): An analysis
based on mitochondrial sequence data. Journal of Avian Biology 33: 5–14.
Smith, J. D. (1977). Comments on flight and the evolution of bats. In Major Patterns
in Vertebrate Evolution, M. K. Hecht, P. C. Goody, and B. M. Hecht, Eds.
Plenum Press, New York. pp. 427–437.
Socha, J. J. and M. LaBarbera (2005). Effects of size and behavior on aerial perfor-
mance of two species of flying snakes (Chrysopelea). Journal of Experimental
Biology 208: 1835–1847.
Soemmerring, S. T. v. (1817). Über einen Ornithocephalus brevirostris der Vorwelt.
Denkschriften der koniglichen bayerischen Akademie der Wissenschaften
München, mathematisch-physikalische Classe 6: 89–104.
Song, A., X. D. Tian, E. Israeli, R. Galvao, K. Bishop, S. Swartz and K. Breuer (2008).
Aeromechanics of membrane wings with implications for animal flight. AIAA
Journal 46: 2096–2106.
Speakman, J. R. (1993). Flight capabilities in Archaeopteryx. Evolution 47: 336–340.
Speakman, J. R. and P. A. Racey (1991). No cost of echolocation for bats in flight.
Nature 350: 421–423.
Springer, M. S., M. J. Stanhope, O. Madsen and W. W. de Jong (2004). Molecules
consolidate the placental mammal tree. Trends in Ecology & Evolution 19:
430–438.
Springer, M. S., E. C. Teeling, O. Madsen, M. J. Stanhope and W. W. de Jong (2001).
Integrated fossil and molecular data reconstruct bat echolocation. Proceedings
of the National Academy of Sciences of the United States of America 98:
6241–6246.
Stafford, B. J. (1999). Taxonomy and Ecological Morphology of the Flying Lemurs (Der-
moptera, Cynocephalidae). Ph.D. Dissertation, City University of New York.
464 pp.
Staniczek, A. H., P. Sroka and G. Bechly (2014). Neither silverfish nor fowl: The enig-
matic Carboniferous Carbotriplura kukalovae Kluge, 1996 (Insecta: Carbo-
triplurida) is the putative fossil sister group of winged insects (Insecta: Pter-
ygota). Systematic Entomology: DOI: 10.1111/syen.12076.
Swan, L. W. (1970). Goose of the Himalayas. Natural History 70: 68–75.
Swartz, S. M., M. S. Groves, H. D. Kim and W. R. Walsh (1996). Mechanical proper-
ties of bat wing membrane skin. Journal of Zoology 239: 357–378.
Tauber, C. A., M. J. Tauber and J. G. Giffin (2007). Flightless Hawaiian Hemerobi-
idae (Neuroptera): Comparative morphology and biology of a brachypterous
species, its macropterous relative and intermediate forms. European Journal
of Entomology 104: 787–800.
Teeling, E. C. (2009). Hear, hear: The convergent evolution of echolocation in bats?
Trends in Ecology & Evolution 24: 351–354.

Bibliography  [ 193 ]
Teeling, E. C., O. Madsen, R. A. Van Den Bussche, W. W. de Jong, M. J. Stanhope
and M. S. Springer (2002). Microbat paraphyly and the convergent evolution
of a key innovation in Old World rhinolophoid microbats. Proceedings of the
National Academy of Sciences of the United States of America 99: 1431–1436.
Teeling, E. C., M. Scally, D. J. Kao, M. L. Romagnoli, M. S. Springer and M. J. Stan-
hope (2000). Molecular evidence regarding the origin of echolocation and
flight in bats. Nature 403: 188–192.
Tennekes, H. (1996). The Simple Science of Flight: From Insects to Jumbo Jets. MIT
Press, Cambridge, Massachusetts. 137 pp.
Thewisen, J. G. M. and S. K. Babcock (1991). Distinctive cranial and cervical inner-
vation of wing muscles: New evidence for bat monophyly. Science 251:
934–936.
Thomas, S. P. (1987). The physiology of bat flight. In Recent Advances in the Study of
Bats, M. B. Fenton, P. A. Racey, and J. M. V. Rayner, Eds. Cambridge Univer-
sity Press, New York. pp. 75–99.
Thomassen, H. A., S. Gea, S. Maas, R. G. Bout, J. J. J. Dirckx, W. F. Decraemer and
G. D. E. Povel (2007). Do Swiftlets have an ear for echolocation? The func-
tional morphology of Swiftlets’ middle ears. Hearing Research 225: 25–37.
Thorington, R. W. Jr. and L. R. Heaney (1981). Body proportions and gliding adapta-
tions of flying squirrels (Petauristinae). Journal of Mammalogy 62: 101–114.
Thulborn, R. A. (1973). Thermoregulation in dinosaurs. Nature 245: 51–52.
Tobalske, B. W. and K. P. Dial (2007). Aerodynamics of wing-assisted incline run-
ning in birds. Journal of Experimental Biology 210: 1742–1751.
Trueman, J. W. H., B. E. Pfeil, S. A. Kelchner and D. K. Yeates (2004). Did stick in-
sects really regain their wings? Systematic Entomology 29: 138–139.
Tucker, V. A. (1968). Respiratory physiology of house sparrows in relation to high-
altitude flight. Journal of Experimental Biology 48: 55–66.
Tucker, V. A., T. J. Cade and A. E. Tucker (1998). Diving speeds and angles of a gyr-
falcon (Falco rusticolus). Journal of Experimental Biology 201: 2061–2070.
Unwin, D. M. (1987). Pterosaur locomotion: Joggers or waddlers? Nature
327: 13–14.
Unwin, D. M. (1997). Pterosaur tracks and the terrestrial ability of pterosaurs. Le-
thaia 29: 373–386.
Unwin, D. M. (1999). Pterosaurs: Back to the traditional model? Trends in Ecology &
Evolution 14: 263–268.
Unwin, D. M. and N. N. Bakhurina (1994). Sordes pilosus and the nature of the
pterosaur flight apparatus. Nature 371: 62–64.
Unwin, D. M. and D. C. Deeming (2008). Pterosaur eggshell structure and its impli-
cations for pterosaur reproductive biology. Zitteliana Reihe B 28: 199–207.
Vandenberghe, N., J. Zhang and S. Childress (2004). Symmetry breaking leads to
forward flapping flight. Journal of Fluid Mechanics 506: 147–155.
Vaughan, T. A., J. M. Ryan and N. J. Czaplewski (2000). Mammalogy. Saunders Col-
lege Publishing, Philadelphia. 565 pp.
Videler, J. J. (2005). Avian Flight. Oxford University Press, New York. 269 pp.
Vogel, S. (1994). Life in Moving Fluids: The Physical Biology of Flow. Princeton Univer-
sity Press, Princeton, New Jersey. 467 pp.
Vogel, S. (2001). Prime Mover: A Natural History of Muscle. W.W. Norton, New York.
370 pp.
Walker, A. D. (1972). New light on origin of birds and crocodiles. Nature 237:
257–263.

[ 194 ] Bibliography
Warrick, D. R. and K. P. Dial (1998). Kinematic, aerodynamic and anatomical mech-
anisms in the slow, maneuvering flight of pigeons. Journal of Experimental
Biology 201: 655–672.
Wellnhofer, P. (1988). Terrestrial locomotion in pterosaurs. Historical Biology
1: 3–16.
Wellnhofer, P. (1996). The Illustrated Encyclopedia of Prehistoric Flying Reptiles: Ptero-
saurs. Barnes & Noble Books, New York. 192 pp.
Whalley, P. and E. A. Jarzembowski (1981). A new assessment of Rhyniella, the earli-
est known insect, from the Devonian of Rhynie, Scotland. Nature 291: 317.
Whiting, M. F., S. Bradler and T. Maxwell (2003). Loss and recovery of wings in
stick insects. Nature 421: 264–267.
Wiest, F. C. (1995). The specialized locomotory apparatus of the freshwater hatchet-
fish family Gasteropelecidae. Journal of Zoology 236: 571–592.
Wigglesworth, V. B. (1963). Origin of wings in insects. Nature 197: 97–98.
Wigglesworth, V. B. (1973). Evolution of insect wings and flight. Nature 246: 127–129.
Wigglesworth, V. B. (1976). The evolution of insect flight. In Insect Flight, R. C.
Rainey, Ed. Blackwell Scientific Publications, Oxford. pp. 255–269.
Wild, R. (1984). Pterosaurs from the upper Triassic of Italy. Naturwissenschaften
71: 1–11.
Wiley, E. O. and B. S. Lieberman (2011). Phylogenetics: The Theory of Phylogenetic Sys-
tematics. Wiley-Blackwell, Hoboken, New Jersey. 406 pp.
Witton, M. P. and M. B. Habib (2010). On the size and flight diversity of giant ptero-
saurs, the use of birds as pterosaur analogues and comments on pterosaur
flightlessness. PLoS One 5: e13982.
Wootton, R. J. (1972). Nymphs of Palaeodictyoptera (Insecta) from the Westpha-
lian of England. Palaeontology 15: 662–675.
Wootton, R. J. (1986). An approach to the mechanics of pleating in dragonfly wings.
Journal of Experimental Biology 125: 361–372.
Wootton, R. J. and C. P. Ellington (1991). Biomechanics and the origin of insect
flight. In Biomechanics in Evolution, J. M. V. Rayner and R. J. Wootton, Eds.
Cambridge University Press, Cambridge. pp. 99–112.
Yanoviak, S. P. (2010). Gliding ants of the tropical forest canopy. In Ant Ecology, L.
Lach, C. Parr, and K. Abbott, Eds. Oxford University Press, Oxford.
pp. 223–224.
Yanoviak, S. P. and R. Dudley (2006). The role of visual cues in directed aerial de-
scent of Cephalotes atratus workers (Hymenoptera: Formicidae). Journal of
Experimental Biology 209: 1777–1783.
Yanoviak, S. P., R. Dudley and M. Kaspari (2005). Directed aerial descent in canopy
ants. Nature 433: 624–626.
Yanoviak, S. P., M. Kaspari and R. Dudley (2009). Gliding hexapods and the origins
of insect aerial behaviour. Biology Letters 5: 510–512.
Yanoviak, S. P., Y. Munk and R. Dudley (2011). Evolution and ecology of directed
aerial descent in arboreal ants. Integrative and Comparative Biology 51:
944–956.
Yanoviak, S. P., Y. Munk, M. Kaspari and R. Dudley (2010). Aerial manoeuverability
in wingless gliding ants (Cephalotes atratus). Proceedings of the Royal Society of
London Series B-Biological Sciences 277: 2199–2204.
Zheng, X. T., Z. H. Zhou, X. L. Wang, F. C. Zhang, X. M. Zhang, Y. Wang, G. J. Wei,
S. Wang and X. Xu (2013). Hind wings in basal birds and the evolution of leg
feathers. Science 339: 1309–1312.

Bibliography  [ 195 ]
INDE X

Note: Page numbers followed by “n” indicate footnotes. Page numbers in italics
indicate illustrations, and may also include text discussion.

aerobic muscle, 24 flight of, 27, 105, 109, 175

aerobic scope, 121, 145 on phylogenies, 106, 112
air combat, 13 on phylogeny, by Godefroit team,
air sacs, respiratory, 123 115
airfoil, 42, 43 skeletal specializations of, 126
airspeed. See speed teeth of, 106
albatross, 5, 8, 9, 58 thecodonts and, 105
altitudes, bird migration, 6 Archonta, 131, 132
anaerobic muscle, 24 Archosauria, 106, 147, 151, 152, 153
angle of attack, 39, 40 Arctic tern, migration distance, 7, 8
antagonistic muscles. See muscles area, wing, 44. See also surface area
ants, directed aerial descent in, 48–49, Argentavis, 9
64, 87 Arthropleura, 94
aphids, 99, 165, 166 Arthurdactylus conandoylensis, 12
arboreal animals, directed aerial articulation, wing, 30. See also shoulder
descent by, 87, 88 joint; wing articulation
arboreal theory, 108, 110 aspect ratio, 45, 48, 62, 63
Archaeopteryx in, 109 auks, 13, 168
bats and, 135 Aurornis, 115, 116
biomechanics supports, 111 automatic wing extension, 31, 33
climbing as exaptation for, 174 Aves, 106. See also birds
dinosaurs and, 120 Avialae, 106, 107, 115, 116
directed aerial descent supports, 174
Microraptor supports, 115 bat wings, 32–33, 178
pouncing proavis, compared with, bats
118 ancestry of, 180
pterosaurs and, 154, 155 arboreal flight evolution in, 174
Archaeognatha, fossil, 75 in Archonta, 131
Archaeopteryx, 103–105, 104, 109 automatic wing extension in, 33
and Avialae, 115, 116 colugos and, 131, 135
Dienonychus and, 110 convergent evolution in, 139, 143
evolutionary relationships of, 107 echolocation by, 133, 134, 141, 142,
feathers of, 114 144
bats (continued) beak as adaptation, 125
families, molecular phylogeny of, flight origin, 173
140 hand and finger bones, 30
flight evolution in, 135–136, 143 largest, 9
flight muscles power echolocation, migration altitudes, 6
144 pneumatic bones, 124
flight origin in, 137–140, 173 shoulder joint, 123
fossil skeletons of, 132 wing structure, 30–31, 31
front limb modifications of, birds
136, 145 arboreal flight evolution of, 174
growth and flight in, 157 Archaeopteryx, 104, 106, 109, 110,
heart and lungs of, 145 126. See also Archaeopteryx
Icaronycteris, 132 as archosaurs, 151
limited structural variation in, 176 Avialae, 106
megabats and microbats, 130, 137, biomechanics and flight evolution
140 theories for, 111
megabats and primates, 137 Confuciusornis, 112, 126
microbats, not a single lineage in, cormorants, Galapagos flightless,
140 167
migration distance of, 7 and crocodilians, 106, 123
Onychonycteris, 133 dinosaur clavicles and, 106
on phylogeny, 132, 133, 137 as dinosaur subgroup, 110
physiological modification to directed aerial descent in, 120
increase power of, 145 endothermy in, 121
poor fossil record of, 131, 135 feathered dinosaurs related to, 113
predation by, 14, 15, 142 feathers of, 19, 30, 114, 117, 180.
sensing (other than echolocation) See also feathers
in, 134 flightless rails, 167
shrews, relationship to, 132 flightlessness in, 167, 171
vision in, 175 growth and flight, 157
wing structure of, 32–33, 178 largest, 9
Yangochiroptera, 141 limited structural variation in, 176
Yangochiroptera and in Maniraptora, 106
Yinpterochiroptera, 138, 140 migration distances of, 7–8
Yinochiroptera, 141 more primitive than Archaeopteryx,
Yinpterochiroptera, 141 116
beak, birds’, 125 Neornithes, 127
bedbugs, secondarily flightless, 100 one-way lungs of, 122, 123, 180
bee flies, 14, 99 origin of, 5
bees, 15, 99 Ornithurae, 127
beetles, 1, 8, 19, 99 pouncing proavis theory and,
Bennett, S. Christopher, 152, 158, 162 118–119
“Big Four” powered flyers, 4–5, 37, 173 protowing of, 178
pterosaurs, only extinct members, skeletal specializations of, 125
147 tails of, 125
biomechanics, 111, 118–119 theropod-bird relationship
biramous limbs (crustaceans), 82 (skepticism of), 110
bird wing skeleton, 125
aspect ratios, 62 bones
automatic wing extension, 31 bat wing, 32

[ 198 ] Index
 bird wing, 31 damselflies, 92, 93, 95
pneumatic, 124 Darwin, Charles
pterosaur wing, 33, 34 arboreal theory proposed
“bounded ignorance”, 82–85, 86, 89 by, 108
brain. See nervous system bat flight evolution described
bristletails, 75, 79 by, 135
evolution of complex structures,
camber, 30, 42, 151, 156, 177 explained, 23
Carbotriplurida, 76, 89 loss of flight discussed by, 167
Caudipteryx, 112, 113, 118, 119 use of tree diagram, 16
China. See Liaoning fossil beds Deinonychus, 106, 107, 108, 109
Chiroptera, 137. See also bats design in evolution, 28
chord, wing, 45 Dial, Kenneth, 119
Chrysopelea, 68 dinosaurs
clavicles (dinosaur), 105, 106 Archaeopteryx as transition from,
Coelurosauravus, 64, 65 105
cold-blooded animals. See ectotherms as archosaurs, 151
collarbones. See clavicles Aurornis, 115, 116
Collembola, 74, 75 Caudipteryx and Protarchaeopteryx,
colugos, 68 112, 113
and bats, 131, 135, 136 clavicles in, 105, 106
gliding, 69 Deinonychus, 106, 107, 108
on the ground, 71 endothermy in, 122
on phylogeny, 132, 133 feathered, 113, 117
vision, similarities with megabats, Liaoning fossil beds, 111, 112
139 lungs of, one-way, 123, 180
wing loading and speed, 69 Maniraptora, 106, 107
wing structure, 68 Microraptor, 113, 114, 115, 120
complex structures, evolution of, 23 phylogeny of, 107, 152, 153
condors, 6, 9, 58 pneumatic bones of, 124
Confuciusornis, 112, 115, 126 Scleromochlus and pterosaurs and,
control-configured flyers, 26 151, 153
convergent evolution, 17, 139 secondarily flightless, 116
due to flight constraints, 176 thermoregulation in, 117, 121,
versus homology, 176 122
cormorants, Galpagos flightless, 167 theropods, 106, 107. See also
cost of locomotion, flight versus theropods
walking, 3, 55, 71 Velociraptor, 106
courting in flight, 14 very small, 120, 121
crocodilians, 106, 123, 151, 152, 153 Diptera (true flies). See flies
Crustacea, limb terminology of, 82 directed aerial descent, 71, 87, 174
cursorial theory, 109, 110 discovery, 48–50
Archaeopteryx and, 109 as unifying feature, 174
biomechanics and, 111 in vertebrates, 69, 120
compared with pouncing proavis distance
theory, 118 long migrations, 7–8
flapping for thrust on ground, 111 trades off against speed, 28
Microraptor and, 115 dodo, loss of flight, 167
pterosaurs and, 154 downstroke, 51, 52, 53
Cuvier, Georges, 148, 149 Draco, 60, 61, 63, 67

Index  [ 199 ]
drag, 29, 40 flight origin in insects, 76
induced, 44 gliding and, 47, 50, 55
lift-to-drag ratio, 44 historical constraints on, 22, 176
shape and, 29 of insect tracheal system, 79
and size, 45 insect wing origin, theories, 77
stall and, 41 of living animals, 178, 179
streamlining and, 29, 42, 46 living animals as analogs for, 86, 119
teminal velocity and, 71 of loss of flight ability, 100
thrust overcoming, 50 of lungs, one-way, 123, 180
viscous versus pressure, 46 maneuverability and stability in, 27
dragonfly, 92, 93, 95, 171 megabat and microbat relationship,
aerial predation by, 14 140
courtship flight of, 14 of Neornithes, 127
flapping pattern of, 95 oribatid mites evolving new joint, 80
gliding of, 48 paranotal lobe theory in, 80–81
laying eggs in flight, 14 of pneumatic bones, 124
lift to drag ratio in, 45, 58 of pterosaur, no molecular
Odonatoptera, 92–94 phylogenies for, 161
paleopterous wing hinge of, 90 rails, loss of flight, 167, 168, 179
wing structure of, 21 relationships, 19
ducks, seasonal loss of flight, 165 sexual selection, 81
Dudley, Robert, 49, 66, 87 evolutionary relationships, 16, 19.
See also phylogenies (phylogenetic
eagle, 5, 6, 9, 58 trees)
echolocation, 141, 142 evolutionary trees, 16. See also
evolution of, 143 phylogenies (phylogenetic trees)
in fossil bats, 133 exaptation, 22, 69, 88, 122, 174
multiple origins (microbats), 141 exite or exopodite (crustacean limb),
and vision, 175 81, 82
ectotherms, 121, 159 exoskeleton, insect, 35
eggs, laying, in flight, 14
Ellington, Charles, 84 falcon, 6, 11, 14
Enantiornithes, 107, 115, 127 falling, 71, 174
endotherms, 117, 121–123, 159 feathered dinosaurs, 117
energetics, flight, 3, 158 Aurornis, 116
energy conservation, 55, 71 Caudipteryx and Protarchaeopteryx,
Engel, Michael, 75, 76 112
Ephemeroptera, 92. See also mayflies Microraptor, 113
evolution pouncing proavis theory, 118–119
arthropod limb structure and, 82 feathers, 31
of bat front limb, 136 Archaeopteryx, 104, 114
bat molecular phylogeny, 140 Aurornis, 116
benefits of gliding, 70–71 Caudipteryx, 112
of complex structures, 23 Confuciusornis, 126
convergent, 139, 143, 162, 176 contour, 31
design in, 28 on dinosaurs, 113, 122
of echolocation, 143 evolution of, 117
of endothermy in birds, 121 as insulation, 117, 122
experiments, 84, 85, 87 Microraptor, asymmetrical, 113
of feathers, 113, 117 origin of, 180

[ 200 ] Index
 pennaceous, 112 laying eggs in, 14
on phylogenies, 19 maximum altitudes, 6
primary, 30, 31, 113, 114 maximum speed, 5, 6
Protarchaeopteryx, 112 power requirements for, 24, 121
symmetrical, Caudipteryx, 113 powered, 2, 50
featherwing beetle, 8 sensing requirements of, 26, 61
fish, 70 speed, 3, 5, 6, 54
flapping underwater, 13
and aerobic scope, 121 unpowered, 2
in arboreal and cursorial theories, vision, acute, requirement for, 26,
108, 109 61, 72, 175
Archaeopteryx, 109 flight evolution
Avialae, defined by, 106 arboreal theory in, 108, 111, 120,
bats, 137 135, 154, 174
bats, on phylogeny, 140 Archaeopteryx and competing
Big Four, in common, 173 theories, 109
bird lungs and power for, 123 of bats, 135–143, 136, 140
dragonfly pattern, 95 biomechanics and flight origin
Enantiornithes, 127 theories, 111
in evolution of insect flight, 89 bird wing skeleton modifications,
forces, 24, 52, 53 125
forewings and hindwings held bird-theropod relationship
together, 99 (skepticism of), 110
grasshopper pattern, 95 confused with bird ancestry, 109
helicopter (analogy), 50 cursorial theory, 108–110, 155
and hovering power, 54 directed aerial descent, 88, 120, 174
limits structural variation, 176 feathers and, 117, 180
muscles, 24, 144, 174 flapping for thrust on ground, 111
similar for all flying animals, 24 gill theory, 77–79
for thrust, 50–52 gliding and, 47, 55, 72, 84, 88
for thrust on ground, 111, 120 of gliding animals, 69, 72
transition from walking, 174 historical constraints and, 176
upstroke and downstroke insects, 76, 89
movements, 51 models of ancestral, 82–85
versus gliding, 55 wing origin, traditonal theories,
weak, 55, 72, 80, 89, 136, 156 77
wing movements, 51 wings not from legs, 97
wing-assisted incline running and, Kukalova-Peck’s theory, 81
119 living animals as analogs for, 86, 119
fleas, 100, 167, 169–171 loss of flight ability, 100, 164
flies, 15, 99 modifications to nervous system for,
flightless, 170 61, 175
halteres, 100 paranotal lobe theory, 80–81
flight, 2 pouncing proavis theory, 118–119
benefits of, 3–4 of pterosaurs, and arboreal theory,
courting in, 14 154, 155
energy cost, 3 of pterosaurs, and cursorial theory,
evolutionary success and, 4 154, 155
flapping, Big Four, 4, 173 of pterosaurs, birds and bats,
gliding and evolution of, 72 potential interactions, 173

Index  [ 201 ]
flight evolution (continued) Liaoning fossil beds, 111
requirement for acute vision in, 26, mayflies, 93
61, 72, 175 mayfly nymphs, 98
Rhyniognatha (ancient insect fossil), Microraptor, 113, 114, 115
75 nymphs with large wing pads, 98
silverfish falling behavior and, 87 oldest springtails, 74
skeletal specializations in birds for, Ornithurae, 127
125 Onychonycteris, 133
solar collecting and insect Palaeodictyopterida, 91
protowings, 84 pterosaur, 150, 158, 162
surface-skimming theory, 86–87 pterosaur trackways, 155, 156
transition from walking to flapping, Rhyniognatha, 75
174 springtail, 75
weak flapping and, 55, 72, 80, 89, frogs, gliding, 69
136, 156 fruit bats, Old World. See megabats
wing-assisted incline running fruit fly, 6, 58
theory, 119–120 furculas, 105, 106
flight muscles, soaring birds, 58
flightlessness, secondary. See loss of Garner, Joseph, 118
flight gecko, gliding, 67
flying animals, 8, 9, 10, 178 geese, 6, 7, 9
amphibian, 13 Gegenbaur, Carl, 77
limited structural variation in, 175, genus (plural genera), 11
176 giant insects, Carboniferous, 94
maneuverability of, 27 gigantothermy, 122n
require good vision, 26 gill theory of insect wing origin, 77–79,
flying fish, 70 85, 98
flying foxes, Old World. See megabats exites in, 81
flying lemurs, 68. See also colugos Kukalova-Peck’s modification of, 81
flying lizards. See Draco problems, 79
flying squirrels, 63, 68 Wigglesworth, Vincent,
force, flapping, 24, 52, 53. See also drag; supporting, 81
lift; thrust; weight gills, insect, 77, 78
fossils glide angle
Archaeopteryx, 103–105, 104 ants, 49, 64
bats, 131, 132 model, ancestral insect, 85
bristletail, 75 glide tests, insect models, 85, 87
Carbotriplurida, 76 gliders
Caudipteryx and Protarchaeopteryx, animals, 60
112 in equilibrium glide, 62, 63
Confuciusornis, 112, 126 hypothetical ancestral insects, as
earliest winged insects, 76 models of, 84
Elmo fossil beds, 94 lift-to-drag ratio and glide angle, 47
Enantiornithes, 127 gliding, 2, 47–50, 63
evolutionary relationships and, 19 ants, glide angle of, 49
feathered dinosaurs, 112 arboreal and cursorial theories
giant insects, 94 and, 108
Icaronycteris, 132, 133 ballistic and aerodynamic phases, 62
incomplete record, 179 biomechanics of bird flight evolution
insects and insect wings, 89 and, 111

[ 202 ] Index
 directed aerial descent and, 87, 88 gliding reptiles. See gliding animals
Draco, 60 goose. See geese
evolution of, 50 grasshoppers, flapping pattern, 95
evolutionary benefits of, 70–71 Great Auks, 13, 168, 169
and falls, 70, 71 great blue heron, 6
favors large animals, 58 griffenflies (“giant dragonflies”), 93, 94
flight, as a form of, 72 Grimaldi, David, 75
in flight evolution, 174 ground up theory. See cursorial theory
bats, 135, 136 growth, pterosaur, effect on flight,
insects, 84, 88, 89 157–158
pterosaurs, 154, 155
high lift-to-drag ratio improves, 48 halteres (fly stabilizers), 100
Microraptor, 115 Hasenfuss, I., 87
model tests, extinct gliding hatchetfish, 70
lizards, 66 hawks, 14, 15, 58
in paranotal lobe theory, 80 Heilmann, Gerhard, 105
powered by gravity, 47 historical constraints, 22, 176
small size and, 48 homology, 176
soaring, 48 homonomous wings, 95
versus flapping, 55 honeybees, 6, 7, 11, 14
weight and speed in, 47 house flies. See flies
wing-assisted incline running hovering, 12–13, 51
omits, 119 power requirements for, 54
gliding animals, 48, 61 size and, 12, 55, 56
angles of attack in, high, 63 structural requirements for, 55
ants, 64 hummingbird, 52
aspect ratios of, 63
bat ancestor as, 135 Icaronycteris, 132, 133
colugos, 68, 69 Icarosaurus, 64, 65
don’t soar, 62 induced drag, 44
Draco, 60 insect exoskeleton, 35
energy conservation and, 71 insect protowings, 83
escape from predators by, 70 as courtship displays, 80
as evolutionary endpoint, 72 as exites of leg, 81
extinct lizards, 65 as gills, 77
extinct mammals, 66 as paranotal lobes, 80
extinct reptiles, 64, 65, 67 as solar collectors, 82
flying fish, 70 insect wings
flying squirrels, 68 anatomy of, 36
frogs, 69 articulation of, 36
geckos, 67 inability to molt, 93
glide paths of, 62, 63 membrane of, 35
living today, 67–70 not modified legs, 35
Microraptor, 115 pleating for stiffness, 35, 36
nervous systems of, 61 structure of, 35–37
non-equilibrium glides in, 62, 63 veins in, 35
snakes, 68 insects
wing lengths constrained in, 63 age of flight origin of, 76, 173
wings of, 61 arboreal flight evolution in, 174
gliding lizards, extinct, 65 bedbugs, 100

Index  [ 203 ]
insects (continued) tracheal system of, 79
body form variation of, 176, 178 uniramous limbs of, 82
Carboniferous giants, 94 upstroke of, 52
Carbotriplurida fossil, 76 wing hinge
cold-loving, and loss of flight, 170 neopterous, 90
earliest winged, 76 origin, 180
Ephemeroptera, 92 paleopterous, 90
fleas, 100 wing origin, questions about, 179
flight evolution in, 77, 79–81 insulation, 117, 122
flight origin theories, traditional, 77 interference, mechanical, wing
fly halteres, 100 hinges, 96
forewings, dominant, 99 islands, loss of flight on, 167
fossils, 75, 76, 78, 89, 91, 93, 94
four-winged plan, 94, 99 June beetle, green, 1, 19
gill theory, 77–79, 81
gills of, aquatic, 77, 78 Kaspari, Michael, 87
gliding by, 48 Kingsolver, Joel, 82, 84
homonomous wings, 95 Koehl, Mimi, 82, 84
immature, terminology, 78 Koopman, Karl, 141
integrated flight evolution Kramer, Melissa, 86
hypothesis, 89 Kuehneosaurus, 64, 65
Kukalova-Peck’s theory, 81 Kukalova-Peck, Jarmila, 81
largest known, 93, 94
legs did not become wings in, 97, 178 L/D. See lift-to-drag ratio
lice, 100 lactic acid in muscles, 25
locked forwings and hindwings, 99 Lagosuchus, 152, 153. See also
loss of flight in, 100, 167, 169, 171 Marasuchus
mayflies, 92 laminar flow, 46
fossils, 93 largest flyers, 9
immature, with gills, 78 larva (immature insect), 78
subimago, 92 Laurasiatheria, on phylogeny, 133, 140
model tests of ancestral, 82–85 Liaoning fossil beds, 111, 112, 126
nymphs and naiads, 78 lice, loss of flight in, 100, 167, 169, 170
Odonatoptera, 92–94 lift, 39, 40
oldest fossils of, 74 angle of attack and, 40
only fly as adults, 92, 157 Bernoulli’s equation and, 40
origin, 4, 74 hovering and vertical takeoff, 12
Palaeodictyopterida, 91, 92 on insect protowings, 84, 89
paranotal lobe theory, 80–81 relative wind and, 40
phylogeny of major groups, 90 speed and, 43
protowings as solar collectors, 84 stall and, 41, 42
Rhyniognatha (ancient fossil), 75 trailing edge orientation and, 41
silverfish, 76. See also silverfish lift-to-drag ratio, 44
single functional wing pair in gliding, 47, 48
flies, 99 maximum, 44n
size, versus vertebrates, 178 and size, 58
species with flying and flightless loss of flight
individuals, 166 bats and pterosaurs, absent in, 165
stoneflies and surface skimming, in birds, 165, 168, 171
86–87 energy conservation and, 168

[ 204 ] Index
 in Great Auk, 168 form a lineage with some microbats,
in insects, 165, 166, 169, 170 140
on islands, 167 linked to primates, 137, 139
legs and, 165 on molecular phylogeny, 140
ostriches and, 166 vision in, 134, 139
penguins and, 167, 168 Meganeuropsis, 93
reversal, improbability membrane, wing, 32, 34, 35, 36
of, 170 microbats, 130, 138
seasonal, 165 difference from megabats, 137
social insect worker caste, 166 echolocation in, 142
winged and wingless aphids, 166 on molecular phylogeny, 140
lungs (flight modifications of), 122, 145, not a single lineage, 140
159, 180 sensing (other than echolocation),
134
mammals, 66, 132, 133 Microraptor, 113, 114, 115, 120, 121
mandibles, dicondylic, 75 migration distances, 7–8
maneuverability, 27 minimum power speed, 54
in animal flight evolution, 27 mites, oribatid (de novo joint
bird tails and, 125 evolution), 80
of birds, 44 models, testing of
of colugos, 69 extinct gliding lizards, 66
of dragonflies, 96 hypothetical ancestral insect, 82–85
predators, avoiding with, 175 silverfish falling, 87
pterosaurs, size and, 158 molting
and stability, lack of built-in, 175 feather replacement and
maneuvering, pterosaur head crests flightlesness in birds, 165
and, 160 insect, 92
Maniraptora, 106 insect wings incapable of, 97
Deinonychus and Velociraptor, 106 mayfly subimago, 92
evolutionary relationships of, Monarch butterfly migration
107 distance, 7
Liaoning fossil beds, 112 Müller, Fritz, 77
Microraptor, 113 murres, 13, 168
on phylogeny, 107, 115 muscle tissue, 23
secondarily flightless, 116 aerobic versus anaerobic, 24–25
Marasuchus, 152n, 153 fuels, 24–25
Marden, James, 86 myoglobin in, 25
mating in flight, 14 operating principles, 23
mayflies, 92 oxygen use, 24–25
feeding, 15 muscles
fossils, 93, 98 antagonistic pairs, 23
gills on, immature, 78 constrain flapping, 24, 176
immature, 78, 98 endotherms versus ectotherms,
mating in flight, 14 121
subimago, 92, 97 evolution of insect wing and, 95
Maynard Smith, John, 27 flapping, 111, 174
McGuire, Jimmy, 66 and hovering, 56
Mecistotrachelos, 66, 67 lactic acid and, 25
megabats, 130, 138 power for echolocation, 143
difference from microbats, 137 myoglobin, 25

Index  [ 205 ]
naiad (immature aquatic insect), 78 paranotal lobe theory (insects), 80–81,
natural selection, 21 85, 88
complex structures and, 23 partridge chicks, and WAIR theory, 119
design in, 28 pelicans, 5, 15
for gliding, 88 penguins, 167–169
in gill theory, 77–79 Pettigrew, John D., 137, 140
on living animals, 179 phylogenetic systematics, 16–19
on maneuverability and stability phylogenetic trees. See phylogenies
(flight), 27 (phylogenetic trees)
in paranotal lobe theory, 80 phylogenies (phylogenetic trees), 16
sexual selection and, 81 Archaeopteryx and birds, 106, 107
on solar collecting protowings, 84 Archaeopteryx-Dienonychus
Neoptera, 90, 95, 96, 97 relationship, 110
neopterous wing articulation, 90, 96 arthropod limb structure and, 82
Neornithes, 127 assembling, 17
nervous systems of Avialae and theropods, by
active stabilizing by, 175 Godefroit team, 115
controlling wing beat pattern, 25, 26 of bat families, molecular, 140
directed aerial descent and, 49 bird-dinosaur relationships, 106,
gliding and, 88 107, 115
of gliding animals, 61 computers and, 17
vertical take off reflex, 13 convergence versus homology, 176
wing-assisted incline running, examples, 18
evolution of, 120 genetic (DNA), 17
nighthawks, 6, 15 insect-crustacean relationship
Norberg, Ulla, 111 shown by, 82
nymph (immature insect), 78, 98 insects on, 90, 92
Liaoning fossil beds, evidence from,
Odonatoptera (including Odonata), 92, 112
93, 94 of mammals based on anantomy,
Onychonycteris, most primitive fossil 132
bat, 133 of mammals, molecular, 133
“opposite birds”. See Enantiornithes Maniraptora and birds, 113
oribatid mites (de novo joint evolution), megabats and microbats closely
80 related, 140
Ornithurae, 107, 115, 127 molecular, 17
ostriches, loss of flight in, 166 molecular, show microbats form two
Ostrom, John, 106, 109, 121 lineages, 140
oxygen pouncing proavis theory and,
atmospheric, and giant insects, 94 118–119
and bat respiratory systems, 145 pterosaurs and dinosaurs as sister
bird lungs obtaining, 123 groups, 152
use by muscles, 24 pterosaurs, no molecular phylogenies
for, 161
Padian, Kevin, 149, 152, 154, 155 reconstruction, 16–19
Palaeodictyopterida, 91, 92 separate pterosaurs from dinosaurs,
Palaeoptera, 90, 92 152, 153
paleopterous wing articulation, 90, 92. shared derived characteristics
See also wing articulation and, 16
parachuting, 49, 80, 88 splits megabats and microbats, 137

[ 206 ] Index
pigeon wing structure, 21 bat-like reconstruction of, 149, 150
pilots, in air combat, 13 as bipedal runners, 154, 155
pinions (primary feathers). See feathers bird-like reconstruction of, 155
pleural appendage theory, 77–79 body mass estimates for, 159
plovers, 6, 15 crests of, 160
pneumatic bones, 124, 159 cursorial evolution disputed, 155
pouncing proavis theory, 118–119 dinosaurs, relationship to, 151
power endothermy versus ectothermy, 159
for echolocation in bats, 143, 144 fingers of, 33
endothermy and, 121 fossil, 150
for flapping, 24 no transition, 149
flight versus walking, 24 showing growth, 158
and fuel economy in flight, 55 trackways, 155, 156
and speed, 28, 53, 54 growth, size, and flight, 157–158
U-shaped power curve of, 54 Lagosuchus as ancestor, 152
powered flight, 2, 50 largest, 10
Archaeopteryx, 109 lungs of, 159
bat evolution and, 143 origin of, 5
Enantiornithes, 127 phylogeny of, 152, 153, 161
flapping, Big Four, 173 pneumatic bones of, 159
gliding in the evolution of, 72 “pterodactyl” as common name, 148
thrust from flapping, 50–52 Ptéro-Dactyle, first described, 149
two origins in bats, 137 quadrupedal, 155
powered flyers, “Big Four”, 4–5 Scleromochlus, link to dinosaurs, 151
predation, maneuverability and, 175 size, 57
predators, aerial, 14, 96 stability (flight), 27
primary feathers, 30, 31, 114, 126 stiffening fibers, in wing membrane,
primates, 131, 139 156
on phylogeny, 132, 133 swimming in, 161
vision, similarities with megabats, time of flight origin, 173
139 walking in, 156, 157
Protarchaeopteryx, 112 webbed feet of, 161
prothoracic lobes of wing membrane of, 150, 154
Palaeodictyopterida, 91 wing structure of, 33–35, 178
protowings, 22, 111, 178. See also insect puffins, 13, 15
protowings pygostyle, 125, 126
Pteranodon ingens, 9, 58
Ptéro-Dactyle, 148, 149 Quetzalcoatlus, 10, 11, 159
pterodactyls. See pterosaurs
Pterodactylus, 148, 149, 150, 158 Racey, Paul, 144
Pteropodidae (megabats), 137, 139, 140, rails, loss of flight in, 167, 179
141 Rayner, Jeremy, 111
pterosaur wing, 34 relative wind, 40
bones, 33 respiratory systems, 122, 145
stiffening fibers, 34 Reynolds number, 45, 46, 158
pterosaurs, 10 Reynolds, Osborne, 46
ancestry, 180 Rhyniognatha, ancient insect fossil, 75
arboreal flight evolution in, 154,
155, 174 sandpipers, 6, 7, 15
as archosaurs, 151 Scleromochlus, 151, 153

Index  [ 207 ]
sexual selection, 81 stability-configured airplanes, 26
shared derived characteristics, 16, 17 stall, 41, 42, 63
Sharovipteryx, 66, 67 steering in dragonflies, 96
shoulder joint, birds’ and ancestors’, stiffness of insect wings, 35
123, 124, 126 stoneflies, 86–87
shrews, 132, 133 streamlining, 29, 42, 46
silverfish, 75n, 76, 79, 81, 87, 89 structure
falling behavior, 87 bat wings, 32–33, 178
primitively flightless, 100 bird wings, 30–31
size, 57 insect wings, 35–37, 36
and drag, 45 pterosaur wings, 33–35, 34
growth, and pterosaur flight, subimago (mayfly), 92
157–158 surface area, aerodynamic, 44, 56, 70
hovering and, 12, 56 surface-skimming theory of insect
insects versus vertebrates, 178 flight origin, 86, 109
largest flyers, 9 surface-to-volume ratio, 56, 71, 178
limited by echolocation in bats, 144 swans, 5
microbats versus megabats, 138 swifts, 14, 15
and protowing functions, insect, 84 swimming in pterosaurs, 161
pterosaur body mass estimates, 159
smallest flyers, 8 takeoff, vertical, 12–13
and weight, 57 Taylor, Graham, 118
wing structure and, 27 Teeling, Emma, 140
and wings, effectiveness of, 85 temperature-regulation. See
smallest flyers, 8 thermoregulation
snake, gliding, 68 teratorn, 9
soaring, 48, 58 terminal velocity, 71
Socha, John J., 68 Teyler Museum and unrecognized
social insects, winged and wingless Archeopteryx fossil, 110
individuals, 166 thecodonts, 105, 151, 152
Soemmerring, Samuel Thomas von, 149 thermoregulation, 117, 121
Solnhofen limestone, 103, 120, 149, 158 theropods, 106
sonar. See echolocation Caudipteryx and Protarchaeopteryx,
span, wing, 45 112
Speakman, John, 144 and cursorial theory, 110
species in evolution, 16, 18 endothermy in, 122
species naming, 11 evolutionary relationships of, 107
speed feathered, 113, 117
for best fuel economy, 54 phylogeny, including Avialae, 115
flight advantage of, 3, 5 very small, 120, 121
flight, of large birds, 6 Thomas, Adrian, 118
glider, weight and, 47 thrips, 8, 57
lift and, 43 thrust, 50–52, 174
low aspect ratio wings and, 63 tracheal system (insects), 79, 94
maximum (flight), 6 tracks, fossil pterosaur, 155, 156
minimum power, 54 trees down theory. See arboreal theory
and power, 28, 53 Troodontidae, 107, 115, 117
trades off against distance, 28 Tucker, Vance, 123
springtails, 74, 75 turbulent flow, 46
stability (flight), 26, 27, 175 Tyrannosaurus, 117

[ 208 ] Index
unanswered questions bird, 31, 30–31
bat and pterosaur ancestry, 180 insect, 35–37, 36
functions of protowings, 179 pterosaur, 33–35, 34
timing of wing origins, 179 speed and, 28
wing precursors in insects, wing-assisted incline running (WAIR),
pterosaurs and bats, 179 119–120
uniramous limbs, 82 winglets
upstroke, 51, 52, 53 on model of hypothetical ancestral
uropatagium (bat tailwing), 32 insect, 83, 85
U-shaped power curve, 54 of nymphs, 98
Paleaeodictyopterida, 91
veins, insect wings, 35, 36 on silverfish model, 87
Velociraptor, 106, 108 wings
vertebrates, size, versus insects, 178 area and lift on, 44
vertical takeoff, 12–13 articulation of, insect, 80
viscosity, 46 aspect ratio of, 45
vision basic attributes, 29, 30
in bats, 134 bat, 32–33
requirement for flight, 175 bat, lift and thrust on, 33
similarities in, megabats, colugos bat, membrane (patagium) of, 32
and primates, 139 bat, most primitive, 133
Volaticotherium, 66 bird, structure of, 30–31
vultures, 45, 58 changing shape, 31
cylinder acting as, 84, 85
walking versus flight, 5, 174 dragonfly, motions of, 96
wandering albatross, 8 dragonfly, structure of, 21
warm-blooded animals. See endotherms effectiveness and size, 85
wasps, 15, 99 flapping forces on, 50, 52, 53
webbed feet in pterosaurs, 161 flapping movements of, 51
weight flapping, for thrust, 50, 174
glider speed and, 47 forewings and hindwings flapped
and hovering ability, 56 together, 99
and size, 57 gliding with, 47
terminal velocity and, 71 grasshopper, flapping pattern of, 95
wing articulation, 30 hinge interference, dragonfly, 96
Confuciusornis, 126 and historical constraints, 176
dragonfly (interference), 96 homonomous, 95
insect, 36, 89 inability to molt, in insects, 93
neopterous, 90 induced drag of, and tip effects, 44
origin, 180 insect
paleopterous, 90, 91, 92 anatomy of, 36
wing extension, automatic, 31, 33 four-winged plan, 94, 99
wing membrane legs did not become wings, 97
bat, 32 neopterous hinge, 90
insect, 36 paleopterous hinge, 90
pterosaur, 34, 150, 154, 156 pleating, 35
wing muscles and hovering, 54 lift-to-drag ratio of, 44
wing pads (nymphs), 98 locked forwings and hindwings, 99
wing structure modifications of bat front limb
bat, 32–33, 178 structure, 145

Index  [ 209 ]
wings (continued) wingspan, largest, 8, 9
modified into halteres (flies), 100 Wootton, Robin, 84
physics dictate form of, 176 work of flapping, 24
pigeon, structure of, 21
pterosaur, 34, 155, 156 Xianglong, 64
scaling structure and weight of, 57
size affects hovering, 56, 57 Yangochiroptera, 138, 140, 141
of smallest flyers, 8 Yanoviak, Stephen, 48, 87, 88,
stall resistance, 145 174
stiffening fibers of, pterosaur, 34 Yinochiroptera, 141
structural materials of, 21 Yinpterochiroptera, 138, 140, 141
structure and scaling of, 27
swimming with, 13 Zygentoma (silverfish), 76, 87

[ 210 ] Index