From a previous post of mine:
The cladogram for the evolution of flight looks like this:
(Note -- each name along the top is a known transitional fossil; and those aren't all that have been discovered.) Here's a more detailed look at the middle section:
Fossils discovered in the past ten years in China have answered most of the "which came first" questions about the evolution of birds from dinosaurs.
We now know that downy feathers came first, as seen in this fossil of Sinosauropteryx:
That's a close-up of downy plumage along the backbone. Here's a shot of an entire fossil
Sinosauropteryx was reptilian in every way, not counting the feathers. It had short forelimbs, and the feathers were all the same size. Presumably, the downy feathers evolved from scales driven by a need for bodily insulation.
Next came Protarchaeopteryx:
It had long arms, broad "hands", and long claws:
Apparently this species was driven by selection to develop more efficient limbs for grasping prey. One of the interesting things about this species is that the structure of the forelimb has been refined to be quite efficient at sweeping out quickly to grab prey, snap the hands together, then draw them back towards the body (mouth?). The specific structures in question are the semilunate carpal (a wrist bone), that moves with the hand in a broad, flat, 190 degree arc, heavy chest muscles, bones of the arm which link together with the wrist so as to force the grasping hands to spread out toward the prey during the forestroke and fold in on the prey during the upstroke. Not only is this a marvelously efficient prey-grabbing mechanism, but the same mechanism is at the root of the wing flight-stroke of modern birds. Evolution often ends up developing a structure to serve one need, then finds it suitable for adaptation to another. Here, a prey-grasping motion similar in concept to the strike of a praying mantis in a reptile becomes suitable for modifying into a flapping flight motion.
Additionally, the feathers on the hands and tail have elongated, becoming better suited for helping to sweep prey into the hands.
Next is Caudipteryx:
This species had hand and tail feathers even more developed than the previous species, and longer feathers, more like that of modern birds:
However, it is clear that this was still not a free-flying animal yet, because the forelimbs were too short and the feathers not long enough to support its weight, and the feathers were symmetrical (equal sized "fins" on each side of the central quill). It also had very reduced teeth compared to earlier specimens and a stubby beak:
But the elongation of the feathers indicates some aerodynamic purpose, presumably gliding after leaping (or falling) from trees which it had climbed with its clawed limbs, in the manner of a flying squirrel. Feathers which were developed "for" heat retention and then pressed into service to help scoop prey were now "found" to be useful for breaking falls or gliding to cover distance (or swooping down on prey?).
Next is Sinornithosaurus:
Similar to the preceding species, except that the pubis bone has now shifted to point to the back instead of the front, a key feature in modern birds (when compared to the forward-facing publis bone in reptiles). Here are some of the forearm feathers in detail:
Long feathers in detail:
Artists' reconstruction:
Next is Archaeopteryx:
The transition to flight is now well underway. Archaeopteryx has the reversed hallux (thumb) characteristic of modern birds, and fully developed feathers of the type used for flight (long, aligned with each other, and assymetrical indicating that the feathers have been refined to function aerodynamically). The feathers and limbs are easily long enough to support the weight of this species in flight. However, it lacks some structures which would make endurance flying more practical (such as a keeled sternum for efficient anchoring of the pectoral muscles which power the downstroke) and fused chest vertebrae. Archaeopteryx also retains a number of clearly reptilian features still, including a clawed "hand" emerging from the wings, small reptilian teeth, and a long bony tail. After the previous species' gliding abilities gave it an advantage, evolution would have strongly selected for more improvements in "flying" ability, pushing the species towards something more resembling sustained powered flight.
Next is Confuciusornis:
This species had a nearly modern flight apparatus. It also displays transitional traits between a reptilian grasping "hand" and a fully formed wing as in modern birds -- the outer two digits (the earlier species had three-fingered "hands") in Confuciusornis are still free, but the center digit has now formed flat, broad bones as seen in the wings of modern birds.
Additionally, the foot is now well on its way towards being a perching foot as in modern birds:
It also has a keeled sternum better suited for long flight, and a reduced number of vertebrae in the tail, on its way towards becoming the truncated tail of modern birds (which while prominent, is a small flap of muscle made to look large only because of the long feathers attached).
From this species it's only a small number of minor changes to finish the transition into the modern bird family.
(Hey, who said there are no transitional fossils? Oh, right, a lot of dishonest creationists. And there are a lot more than this, I've just posted some of the more significant milestones.)
There's been a very recent fossil find along this same lineage, too new for me to have found any online images to include in this article. And analysis is still underway to determine exactly where it fits into the above lineage. But it has well-formed feathers, which extend out from both the "arms" and the legs. Although it wasn't advanced enough to fully fly, the balanced feathering on the front and back would have made it ideally suited for gliding like a flying squirrel, and it may be another link between the stage where feathers had not yet been pressed into service as aerodynamic aids, and the time when they began to be used more and more to catch the air and developing towards a "forelimbs as wings" specialization.
So in short, to answer your question about how flight could have developed in birds, the progression is most likely some minor refinement on the following:
1. Scales modified into downy feathers for heat retention.
2. Downy feathers modified into "straight" feathers for better heat retention (modern birds still use their body "contour feathers" in this fashion).
3. Straight feathers modified into a "grasping basket" on the hands (with an accompanying increase in reach for the same purpose).
4. Long limbs with long feathers refined to better survive falls to the ground.
5. "Parachute" feathers refined for better control, leading to gliding.
6. Gliding refined into better controlled, longer gliding.
7. Long gliding refined into short powered "hops".
8. Short powered flight refined into longer powered flight.
9. Longer powered flight refined into long-distance flying.Note that in each stage, the current configuration has already set the stage for natural selection to "prefer" individuals which better meet the requirements of the next stage. Evolution most often works like this; by taking some pre-existing ability or structure, and finding a better use for it or a better way to make it perform its current use.
