Published online 2 July 1998 | Nature | doi:10.1038/news980702-8

News

Birds and dinosaurs - the debate is over

The fossils of two new species of dinosaur have been discovered in China - dinosaurs with feathers. These creatures effectively close the debate on whether or not birds and dinosaurs share a close evolutionary heritage. The answer is a resounding 'yes'. Yet these fossils make it clear that feathers appeared in evolution long before flight, and long before the appearance of birds in anything like their modern form.

The dinosaurs are described in the 25 June 1998 issue of Nature by the multi-national team of Ji Qiang and Ji Shu-An of the National Geological Museum of China, Beijing, China; Philip Currie of the Royal Tyrrell Museum of Palaeontology, Drumheller, Alberta, Canada, and Mark Norell of the American Museum of Natural History, New York.

Both the dinosaurs come from Liaoning Province, north of Beijing, and are very roughly 145 million years old. Both would have looked, in life, like very stringy turkeys, with long necks and legs, and relatively stubby arms. They would probably have been fast runners, though incapable of flight.

One of the dinosaurs, Protarchaeopteryx, had a switch of feathers at the end of its long tail. The rest of its body may have been covered with downy, feather-like structures similar to those seen in the dinosaur Sinosauropteryx, a smaller and more primitive creature that comes from the same part of China, but in slightly younger rocks. Although doubt has been cast on the relevance of the feather-like body covering of Sinosauropteryx to the origin of 'true' feathers, there is no doubt about the tail-feathers of Protarchaeopteryx. They are indistinguishable from bird feathers: each one has a central stalk, with barbs on either side forming vanes. Yet in most other respects, Protarchaeopteryx is hardly distinguishable from Velociraptor, the fierce hunter that starred in the film Jurassic Park.

The second dinosaur is even more interesting. This creature, Caudipteryx, had vaned feathers not only at the end of its tail, but also on its forearms, which would have looked like miniature wings. Details of its skeleton show that it is more closely related to birds than is Protarchaeopteryx. It also has peculiarities of its own: its jaws lack teeth entirely, except for a fringe of long, pointed teeth at the front of the upper jaw.

Both new dinosaurs, though, are less birdlike even than the earliest-known bird, Archaeopteryx, known from 150-million-year-old rocks in Germany. This creature, dinosaur-like in many ways, but with fully-fledged, birdlike wings, would have been capable of flight of some sort. What the new fossils - of dinosaurs with feathers, but almost certainly flightless - is that feathers appeared before flight, and before birds. Without any further speculation, their very existence rules out the old idea that feathers are features of birds, and birds alone. Conversely, it shows that birds really are the descendants of dinosaurs.

The new finds raise a deeper problem, the same raised by the discovery of any transitional form between what we have hitherto thought of as distinct categories, such as 'bird' and 'dinosaur'. It forces us to think very hard about what we mean by the term 'bird'. Next time you watch the pigeons in the park, ask yourself this question - what, precisely, is a bird? What is it about birds that allows you to see them for what they are, and tell them apart from other animals? What, in short, is the essence of 'birdness'?

This seems like a silly question. Of course everybody knows what a bird is. Birds are animals with beaks and feathers. They eat birdseed and make ugly white streaks down the sides of buildings. We are so convinced that we know what birds are that we hardly give them a second thought.

But this cosy familiarity hides a dangerous fallacy. This fallacy is as follows: we may be confident that we know what birds are like now, but that does not give us licence to guess how birds evolved to reach their present form.

Once we admit this possibility, we must see that the many adaptations of the modern bird skeleton to make it a better airframe could have evolved among the ancestors of birds, for different reasons. Such adaptations include the lack of teeth, the short tail, the rigid, box-like body, the light, hollow bones, the flexible wrist suited to the flapping motion of flight, the collar-bones fused to make a spring-like furcula (the 'wishbone'), and a long list of other features. Velociraptor, for example, had a furcula, flexible wrists and hollowed bones, but it wasn't a bird, and couldn't fly. It is clear that the range of form among modern birds is small subset of that possible among feathered vertebrates. The ancestry of birds contains a much wider range of feathered forms, only a few of which were capable of flight.

A small but vocal minority of researchers has long persisted in the assertion that birds and dinosaurs are not close relatives, but have both acquired birdlike characters independently. The increasing list of close correspondences between birds and dinosaurs goes far deeper than feathers, so it is increasingly hard to claim coincidence unless one has a very good reason for doing so.

Critics have come up with one such reason - discrepancies of stratigraphy. For example, Velociraptor comes from rocks around 70 million years old. It lived much later than supposedly more birdlike creatures such as Protarchaeopteryx and Caudipteryx. This means that dinosaurs such as Velociraptor could not have been ancestral to the more birdlike forms.

But this line of reasoning carries the tacit assumption that the fossil record is always as good as it can be - that is, the age of the rocks in which a fossil is found is a reliable guide to its position in the evolutionary tree of life. Yet the discovery of these new, feathered dinosaurs shows that there are still enormous gaps in our knowledge. Beneath this assumption is another, more serious mistake - that we can take a series of fossils arranged according to their stratigraphic order, and use this order to make statements about ancestry and descent. If this assumption were correct, then of course, 70-million-year-old Velociraptor could not be the lineal ancestor of 145-million-year-old Protarchaeopteryx.

But nobody is claiming such a specific line of ancestry and descent, and here is the reason why: imagine that you are exploring for fossils and you pick one up. Looking at the fossil you have found, you can see that it is a fragment that requires interpretation. It has no labels. It has no identity. It has no birth certificate. It has no pedigree - until you give it one. Looking at this fossil, can you ever prove that it represents the last remains of the lineal ancestor of you, or of any other human or animal? Of course, you cannot.

The best you can do is to suppose that this fossil is your remote cousin. In other words, the most you share is a collateral, rather than a lineal descent. And cousins of different degree can live at different times. Here is a more homely example of what I mean. If you have siblings much older than you are, and they have children, you could have nieces and nephews who are older than you. This may seem rather strange, but discrepancy in age does not, of itself, invalidate the relationship. In the same way, Velociraptor could be the evolutionary 'uncle' or 'aunt' of Protarchaeopteryx, even though it lived very much later.

Another criticism concerns what evolutionary biologists call 'key characters'. Old-fashioned textbooks would assert that the development, in evolution, of certain 'key' features, would mark a qualitative shift in organization. In mammals, the development of a middle-ear containing a chain of three linked bones (rather than the single bone as seen in reptiles) represented this kind of rubicon. An animal without this feature could have had all other mammalian features - hair, milk, live young and so on - but if it had just one middle-ear bone, then it wasn't a mammal, by definition.

With birds, feathers were seen as the key feature. Anything with feathers was a bird, by definition. The discovery of feathered dinosaurs explodes that notion. They have feathers, but they are clearly not 'birds' in the modern sense. Precisely what we do mean by the word 'bird' is thus a vexed question, and the subject of much debate by researchers. The problem can be avoided by ridding ourselves of the very idea of key characters. What is important is not the presence of this feature or that, but the occurrence of a large number of features in combination, and how these features are distributed across the members of the evolutionary family tree.

The new finds show that feathers evolved among dinosaurs, presumably for purposes other than flight. There will be much speculation about why feathers evolved. Could they have been sexual display structures? Could they have acted as aerodynamic vanes to improve stability during fast running, like the spoilers on racing cars? Such speculation is ultimately futile, as we will probably never be able to establish any reason scientifically.

But such speculation runs the risk of making the mistake I outlined above - of presuming to know how animals evolved into their present form. But speculation seem harmless enough. How, then, is it dangerous? Let me explain.

Our almost intuitive knowledge of an 'essence of birdness' largely derives from the long list of extreme specializations that modern birds have to meet the punishing demands of flight. Bird anatomy seems so peculiar, and so specialized, that people have have reasoned that it must have evolved to meet the challenge of flight. This has led to a story of the evolution of birds from small, possibly tree-living reptiles, which evolved flight as a way of getting from tree to tree. Many reptiles today are capable of 'gliding', either using special extensions of skin, or even skin stretched over massively extended ribs. Over millions of generations, these gliding reptiles could have evolved into birds. The 150-million-year-old fossil bird Archaeopteryx, first found in Bavaria in 1861, is the perfect transitional form between reptile and bird, and appears to support this story - that the evolution of birds is indissolubly coupled with the evolution of flight - and of the structures that make flight possible, such as feathers, wishbones and so on.

What is wrong with this story? The error lies in looking at present-day birds, and coming to conclusions, based on the present, about how these things came to be. How can this be wrong? Well, I shall ask the same question in a different way: how can we be certain that the adaptations for flight we see in present-day birds really started out that way? Could assumptions about adaptation be based on human notions of plausibility, rather than on what really happened?

This now seems the case for birds. Over the years, evidence has accumulated for the presence of birdlike features in theropod dinosaurs, even though these dinosaurs did not fly (nor, as far as we know, did they have flying ancestors). Velociraptor, for example, had air-spaces in its bones, and fused wishbones. Such features expose as mistaken all assumptions that certain features evolved 'for' flight, based on what we see in present-day birds. Even though they may confer adaptive 'value' on birds now, these features evolved in non-flying animals, for reasons unconnected with flight - reasons concerning which we can only speculate. Feathers, those quintessentially avian features, can now be added to the growing list. There can no longer be any doubt of the close phylogenetic relationship between birds and theropod dinosaurs. Equally, there can no longer be any doubt that feathers are not key features that automatically mean that their wearer is a bird.

Where, then, does that leave our idea of how to define a bird? Even though the link between birds and dinosaurs is firmly established, the definition of birds has become a difficult problem. The fact is that it is hard to draw a line between birds and dinosaurs that is anything other than arbitrary.

One idea is to restrict membership of the category 'birds' to the latest common ancestor of all present-day birds, and all the descendants of that ancestor, whether living (like the pigeon) or extinct (like the dodo.) At first sight, this seems a good plan - to come back to the start of my argument, everybody knows intuitively what a bird is, based on present-day examples. But this definition runs into problems with the fossil record. There once existed a whole range of flying, feathered, beaked creatures which, were they alive, we'd think of as birds. But detailed consideration of their anatomy shows that they are only remote relatives of present-day birds, so they would not be birds, by our definition. This definition seems perverse, because it is unnecessarily restrictive.

Another idea is to broaden our definition of a 'bird' to include all creatures, living and fossil, that are more closely related to present-day birds than to the group of living creatures most closely related to modern birds - the crocodiles. Unfortunately, the divergence between birds and crocodiles happened a very long time ago, and evolution has come a long way since then. Dinosaurs and many other extinct forms are more closely related to birds than to crocodiles, so this definition would not only include all birds, but all dinosaurs and probably all pterosaurs. We'd have to think of lumbering giants as Diplodocus as birds. This is plainly silly - the definition is too broad to be useful.

The problem is that extinction has taken its toll. Were theropod dinosaurs such as Velociraptor alive today, there would be less of a problem, as we could substitute them for 'crocodiles' in the preceding paragraph, thus restricting the membership of our category 'birds'.

Extinction, then, lies at the heart of our problem of definition. But it also has a wider relevance, and explains the fallacy of charting the evolution of birds according to their present forms alone. That is, the range of forms of creatures that have existed over the 4,500 million years of Earth's history is inestimably greater than the feeble remnant that time has left us. Therefore, it is intuitively wrong of us to take present-day creatures and expect them to be representatives of creatures for all time. As the fossils of Protarchaeopteryx and Caudipteryx show us, any human notions of possibility, based on the limited range or present-day animals, are always subject to rude shocks.

Once we liberate ourselves from the restrictive ideas of key characters, and that features such as feathers evolved with a particular purpose - one limited by what we humans perceive as possible or plausible - the scales fall from our eyes. We see the history of birds as far richer and more interesting than we could ever have imagined possible.

We see running dinosaurs with feathers on tiny wings. Joining Protarchaeopteryx and Caudipteryx is eight-foot-long Unenlagia from Patagonia, which may have been able to 'flap' its arms, bird-fashion.

We see flying birds, such as the crow-sized Rahona, recently described from Madagascar, with an evil, slashing, Velociraptor-like claw - truly a flying dragon.

We see the flightless, stubby-armed Mononykus and Shuvuuia, which, although flightless, seem to be more closely related than Archaeopteryx to modern birds. The skulls and legs of these creatures show features seen in relatively advanced birds, yet they were initially dismissed as aberrant dinosaurs. Yet their relationships imply that flying animals lie in their ancestry.