杏吧原创

DAWN, and something stirs among the cycads鈥攑rimitive plants of the
Jurassic. A basking dragonfly points its huge compound eyes skyward, and a
hundred hexagons fill with the last image they will ever record. The dragonfly鈥檚
wings power up, but it is too late. Its attacker is already changing course to
make the kill. The insect鈥檚 exoskeleton crunches beneath the plummeting mass of
scales and feathers. The early bird catches its worm.

A flight of fancy? Perhaps. But it is becoming increasingly clear that we
need to rethink our ideas about why birds took to the air. A dazzling array of
new fossil finds is bringing about a renaissance in palaeo-ornithology, and with
it the dawning realisation that established theories fall far short of
explaining certain key features of the fossil record. Why, for example, did
birds evolve functional wings long before they began to develop lighter
skeletons for efficient flight? And why are the wings of the first known bird,
Archaeopteryx, so poorly designed for flight when its individual
feathers are so beautifully adapted to generate lift?

It was questions like these that led me, Joseph Garner and Adrian Thomas of
Oxford University to propose a radical new theory for the origin of feathered
flight. Until now, researchers believed the driving force behind the evolution
of wings was the benefits that flight confers. In other words, birds developed
wings to fly so that they could get one up on ground-based competitors and
predators. But what if feathered wings did not evolve for flight, but for
aerodynamic control in a leaping attack? Then the pieces of the puzzle start to
fit together and flight itself emerges almost by accident.

Imagine a small theropod鈥攕omething like Velociraptor, but half
the size鈥攖hat ambushes its prey by jumping down from rocks, ledges or
small trees. After leaving its perch, our animal has only limited inertial
control over its orientation and trajectory, using limb and body movements in a
similar, though less acrobatic way to an Olympic high-diver performing a pike
and double somersault. Natural selection favours improved aerial control. So the
creature鈥檚 descendants evolve structures that create drag鈥攄owny fringes,
perhaps鈥攐n their arms and tail, leaving the legs clear for a kicking
attack. Simple limb or tail movements now let these animals vary the amount of
drag generated on either side of their body to control their orientation and
trajectory, a bit like a boat鈥檚 rudder.

As time passes, the drag-inducing structures on the animal鈥檚 limbs become
broader and flatter. This allows the animal to generate a greater turning force,
improving its agility and manoeuvrability in the air. This is because the
flattened structures now create large amounts of drag when held across the
airflow, but are highly streamlined when they lie parallel to it鈥攖he
principle known in rowing parlance as feathering. The flattening also means that
the animal鈥檚 limbs create lift when the feathers lie at a suitably small angle
to the airflow, so lift-based control can supplant the primitive drag-based
mechanism. Lift-based control is enhanced through natural selection, and
eventually there is enough lift to sustain true flight, as the primitive pounce
becomes a swoop. It is a small step for the muscles controlling the wings to
switch to generating a useful flapping stroke.

The idea that wings evolved as control surfaces differs fundamentally from
the established theories for the origin of feathered flight. Though some degree
of control is essential for successful flight, these theories all suppose that
wings evolved to extend a leap, rather than control it. But here any agreement
ends鈥攖he established theories diverge into two opposing camps over how
this happened. In the arboreal models animals living in trees evolved wings to
enhance their mobility. After progressive improvements in lift production a
steep parachuting descent becomes a shallow glide. Flapping flight follows as a
late innovation. Cursorial models, on the other hand, suppose that a
fast-running biped evolves lift-producing surfaces so it can run or leap more
efficiently. Again, the animal鈥檚 wings become better and better at producing
lift and flapping flight eventually evolves, with or without an intermediate
gliding stage.

The debate over the origins of flight has always been heavily polarised
between the two extremes. It is hard to see what middle ground there might be.
Arboreal theories work with gravity, while cursorial theories must overcome it.
You might just as well try to invent a sport somewhere between skydiving and the
triple jump. Ironically, the only point of agreement between the established
theories鈥攖hat wings evolved to keep an animal airborne鈥攁lso
conflicts strongly with evidence from the fossil record. If wings evolved for
flight, then we would expect a host of other flight adaptations to appear early
in the evolution of birds. But the fossil record tells a different story.

Modern birds, for example, go to all sorts of lengths to lighten the load for
flight, laying their eggs sequentially to avoid carrying too many around at
once. Bar-tailed Godwits will even reabsorb their gut before they migrate. But,
while the skeleton of the Magnificent Frigatebird weighs less than its feathers,
the skeleton of Archaeopteryx is unremarkably reptilian in structure.
In fact, apart from being small, Archaeopteryx has no novel
weight-reducing adaptations at all. Only in more advanced fossil birds, such as
Confuciusornis from China, are the teeth and bony jaws replaced by a
horny beak, and the bony tail reduced to a short stub, or pygostyle. Why would
an early flier like Archaeopteryx go to all the trouble of evolving
complex wings for flight, but then fail to evolve simple weight-reducing
adaptations like these?

Also, why do the preserved flight feathers of Archaeopteryx extend
only along the hand and lower arm, leaving a substantial gap between the wing
and body wall? 鈥淭he trouble with having this gap is that the wing becomes all
tip,鈥� says evolutionary aerodynamicist Adrian Thomas, 鈥渁nd that greatly reduces
the efficiency of the wings.鈥� If you fill in the gap, flight efficiency is
massively enhanced. Modern birds have a relatively short upper arm and the gap
is filled by an additional set of feathers called tertials. More advanced fossil
birds, such as Eoalulavisfrom Spain, have a shorter upper arm than
Archaeopteryx, so it is much more likely that their flight feathers could
have filled the gap.

But neither the gap in the wing nor the weight of the body is at odds with
the new theory. The best place to put a control surface is at the ends of the
limbs鈥攎aximising the lever arm upon which the control forces act. So we
would expect feathers to appear first at the ends of the arms, only bridging the
gap between wing and body when efficient flight became important. And a lighter
skeleton would reduce acceleration in a controlled fall, giving the prey of our
pouncing proto-bird more time to escape, so the late appearance of
weight-reducing adaptations in the avian fossil record is exactly what we would
expect.

Then there is that great paradox of avian evolution: the origin of feathers.
The established theories have great difficulty explaining how feathers came to
be so perfectly designed for lift production. Evolution is a gradual process,
but it is hard to see how a simpler stage in feather evolution could have
produced useful lift. Without the microscopic hook-like barbules that hold
together adjacent barbs on the feather, for example, you lose the integrity of
the entire vane. So feathers must have played another role before lift
production, but their exquisitely complex structure seems like gross overkill
for, say, regulating body temperature.

This isn鈥檛 a problem with the new theory which predicts that feathers first
appeared as downy structures, generating drag like the parachute on a dandelion
seed. As natural selection enhanced drag-based control, the feather stiffened
into a flat vane, and adjacent barbs linked together on a stiff central shaft.
After flattening and stiffening, feathers could finally generate useful lift
when held at a suitable angle to the airflow, but only after they had first
become adapted for drag production.

Nesting ducks

The pouncing theory may do a good job of explaining the key features of the
avian fossil record, but just how plausible is it? One attraction is that there
are good modern analogues for each of its stages. The downy flightless young of
tree-nesting ducks like the Goldeneye, for example, use drag-based control to
stabilise themselves as they jump from their nest hole to the ground. Unfledged
owlets use a mixture of lift and drag generated by their developing wings, to
control their descent as they pounce upon potential prey from a raised perch.
And perching predatory birds such as kestrels, kookaburras and adult owls often
generate asymmetric lift forces with their wings, dropping down on their prey in
a kind of controlled fall.

Of course, there are good modern analogues for the established theories of
bird flight too. It is hard to find a group of tree-dwelling animals that has
not evolved gliding or parachuting flight at some point or another. There are
flying squirrels, flying frogs and even flying snakes. Supporters of the
cursorial theory, on the other hand, might look to the roadrunner, which prefers
to run rather than fly, rarely landing on a perch that it cannot reach within a
hop and a flap from the ground. But the evidence from such comparisons is not
always as strong as proponents of the established theories suggest. Supporters
of the arboreal theory, for example, point out that the pronounced curvature of
the claws of Archaeopteryx makes them look remarkably like the highly
curved claws of modern tree-dwellers. 鈥淭rue, but the raptorial claws of closely
related theropods like Deinonychus are even more tightly curved,鈥� says
Joseph Garner, now at the University of California in Davis. 鈥淭he same logic
would force us to conclude that this three-metre long, eighty-kilogram biped
also lived in trees!鈥�

Modern analogues aside, what we really needed to back up our theory were some
intermediary fossils to fill the gap between the featherless theropods and fully
fledged birds like Archaeopteryx. So, imagine our delight when two
feathered theropods, named Caudipteryxand Protarchaeopteryx,
were discovered in China last year. Nobody suggests that these dinosaurs could
fly鈥攖heir forelimbs are too short, and the preserved feathers cover too
small an area to form a useful flight surface. The only preserved feathers on
Protarchaeopteryx are on the chest and tail, but the two specimens of
Caudipteryx have feathered wings as well as a tail. What is most
interesting, though, is that the wing feathers of Caudipteryx are
attached only to the hands鈥攁n ideal arrangement for drag-based control. If
these feathers were there only to slow a fall, not to control it, you would
expect to find them along the whole arm. Could this be the intermediate fossil
that we were looking for?

Both feathered dinosaurs have a heavy toothed skull and long bony tail, and
Caudipteryx even carried stomach stones, or gastroliths. Clearly
carrying less weight was not an issue for these goose-sized animals, so it is
unlikely that they represent early stages of either the arboreal or cursorial
models. On the other hand, Caudipteryx and Protarchaeopteryx sit
sufficiently far down the avian family tree to be unlikely candidates for the
flightless descendants of an earlier flying ancestor. Besides, their skeletons
show no hangovers from flight鈥攕uch as a laterally directed shoulder joint
to hold the wing out flat. Perhaps Caudipteryx and
Protarchaeopteryx really do represent an early stage of our new model, using
drag-based control to steer themselves in a pouncing attack.

The only difficulty with all this is that in modern animals, stomach stones
are usually a means of weighing down the bodies of diving animals or grinding up
tough plant material鈥攏either of which fits well with Caudipteryx
as a leaping predator. In fact, gastroliths would be potentially lethal for a
meat-eating animal, since their grinding action can splinter a bone into a
thousand stomach-lacerating pieces.

The clue to what Caudipteryx was eating lies in its piercing teeth.
They are certainly not the teeth of a herbivore, and look very much like the
teeth of insectivorous lizards today. Caudipteryx probably used its
gastroliths to squeeze insects鈥� thick exoskeletons and extract the liquid
contents, just as the Aardwolf does today. This sort of hyaena is one of only a
handful of terrestrial mammals with stomach stones. Most theropods were medium
sized carnivores like Deinonychus, or extremely large carnivores like
Tyrannosaurus, so the switch to a diet of insects may well have been
what initiated the decrease in size that was a necessary prerequisite for the
evolution of flight. The picture that we are building up of Caudipteryx,
as an insectivorous biped with feathers at the tips of its arms, certainly
seems to support our ideas

Living fossils

Unfortunately, even if Caudipteryxdoes represent an intermediate
stage in our model, it is too young a fossil to have been a direct ancestor of
modern birds. The Chinese limestone deposits in which Caudipteryx was
found have just been reliably dated to the early Cretaceous, making them at
least 20 million years younger than the late Jurassic limestone in which
Archaeopteryx became entombed. Feathered flight itself must have evolved in
the early Jurassic or before: some time prior to the appearance of
Archaeopteryx. This suggests that Caudipteryx may have been a
living fossil in its own time. Indeed, the Chinese fossil beds in which it was
found seem to be a kind of refuge for outmoded flora and fauna. A cornucopia of
primitive plants and animals have turned up there, many of them already remnants
from a bygone era while they were alive.

That leaves us with one final question to answer. How was a proto-bird like
Caudipteryx able to survive alongside fully fledged birds like
Confuciusornis? The truth is, we don鈥檛 know for sure. If wings did evolve
for flight then the inept Caudipteryx would have been at a distinct
disadvantage. But open your mind to the possibility that there were other
benefits to feathery forearms, and you start to see how this creature could have
survived 20 million years beyond its time.