Two hypotheses have been proposed to explain how flight arose in birds. One hypothesis, the "trees-down" hypothesis, is that bird ancestors were tree dwellers and developed powered flight by gliding from branch to ground. The converse argument, the "ground-up" hypothesis, is that birds developed powered flight directly from a cursorial, or running, ancestor.
Evidence supporting the trees-down hypothesis relates to the inference that early birds appear to have lived in trees. This evidence comes from analysis of the beautiful fossil specimens of Archaeopteryx that preserve not only its foot bones, but also the horny claws (equivalent to human fingernails or the claws of a cat). The claws of Archaeopteryx were long and hooked. In living birds, this type of claw is typically found in birds that live in trees, suggesting that Archaeopteryx may have had similar habits. By examining fossils of the wing and shoulder, functional morphologists have inferred that it was not a skilled flyer. Hence, if it was also a tree dweller, true powered flight probably developed through an intermediate phase that involved gliding down from trees.
The ground-up hypothesis is based on phylogenetic evidence (data concerning the evolution of a genetically related group of organisms). Proponents of this hypothesis point out that the closest relatives of modern birds are dinosaurs like Deinonychus and Troodon. Because they are so closely related to birds, these dinosaurs provide direct evidence about physical characteristics of early birds. Although these dinosaurs were small by typical dinosaur standards, no one has proposed that they lived in trees or flew. Instead, they are viewed as fast-moving runners and denizens of open areas. These observations suggest that primitive birds may have been ground dwellers that developed powered flight from active running and hopping over open areas. Some scientists have even proposed that the extension and flapping of the forelimbs developed as a mechanism for swatting at prey.
Unfortunately, collecting evidence to test hypotheses about the origin of complex functional behaviors such as flight is notoriously difficult, and the problem may remain intractable until new, more powerful methods of study are developed.