INSIGHTS on March 7, 2017 http://science.sciencemag.org/ PERSPECTIVES Downloaded from PALEONTOLOGY story that is beginning to be corroborated by biomechanical studies. According to this story, the development of flight was chaotic, A Mesozoic aviary with different dinosaurs experimenting with different airborne behaviors using dif- Biomechanical models are key to understanding how ferent airfoil and feather arrangements (see the figure), until ultimately only modern dinosaurs experimented with different ways of flying birds survived. The growing understanding of the di- By Stephen L. Brusatte of powered flight—that is, active flapping nosaur–bird transition stems from a trove that generates thrust—has been widely of fossils of early birds and their closest he evolution of birds from a group regarded, sometimes explicitly but often dinosaur cousins collected over the past of small dinosaurs between 170 mil- implicitly, as a long evolutionary march two decades. Most notable are thousands lion and 150 million years ago has in which natural selection progressively emerged as a textbook example of a refined one subgroup of dinosaurs into School of GeoSciences, University of Edinburgh, Grant major evolutionary transformation in ever-better aerialists. However, recent fossil Institute, James Hutton Road, Edinburgh EH9 3FE, UK. T the fossil record (1). The attainment discoveries reveal a much more interesting Email: [email protected] MARSHALL TODD ILLUSTRATION: 792 24 FEBRUARY 2017 • VOL 355 ISSUE 6327 sciencemag.org SCIENCE Published by AAAS DA_0224Perspectives.indd 792 2/22/17 10:35 AM The paravian dinosaur Microraptor glides right direction. The authors used several through the canopy on its four wings. mathematical models to predict whether various winged dinosaurs and early birds of these “protofeathers,” which were soft, could employ various styles of flapping pliable, and arranged in an orderly se- flight seen in modern birds. The verdict: quence (6). In some later meat-eating thero- Except for a handful of paravians, the dino- pods, these simple filaments evolved into saurs did not have the wing sizes or configu- larger, flattened, branching structures—the rations necessary for powered flight. pennaceous (quill-like) feathers that form Furthermore, looking at trends across the wings of today’s birds. the family tree, Dececchi et al. found no Wings are essential for flight; they are the pattern of progressive refinement in aerial airfoils that provide lift. Thus, it is no sur- ability beginning with the origin of wings. prise that they have been a particular em- This supports the hypothesis that wings phasis of recent research. When the Chinese developed for nonflight reasons (such as fossils began to emerge in the late 1990s, display, egg brooding, or something else they showed something unexpected: Many entirely) and that their early evolution theropod dinosaurs had wings of varying was not shaped by selection for aerial flap- shapes and sizes, and many advanced par- ping. Only much later, it seems, did some avian species, like Microraptor, which are paravians evolve the right combination of closely related to birds, had wings on both small body size, large wings, and other ana- the arms and legs (2). Furthermore, wings tomical features to begin powered flight. It first appeared even earlier in much more was at that point that selection was able to primitive, sheep-to-horse–sized theropod mold these animals into more effective fly- dinosaurs that are widely held to be too ing machines. large to fly (7). In 2014, Foth et al. proposed There is another twist to Dececchi et al.’s on March 7, 2017 a provocative hypothesis: Wings originally study. Although some paravians like Mi- evolved as display structures and were later croraptor may have been able to power them- repurposed as airfoils (8). selves through the air, the authors found that Testing this hypothesis, however, is diffi- not all paravians had this ability. Nor was cult. How can you demonstrate that some- the common ancestor of paravians and birds thing in a fossil was definitively a display clearly a lift-producing flapper. Coupled with ornament? One clue comes from pigment- the many morphological differences among bearing structures called melanosomes, winged paravians and early birds, this sug- which have been identified in fossil feathers gests that powered flight may not have been by high-resolution microscopy. The results a singular innovation of the lineage that led reveal that theropod wings were endowed to modern birds, but a behavior that many with various colors, as might be expected different groups of small, feathered, winged http://science.sciencemag.org/ if they were used for decoration (9). But paravians achieved independently. this does not necessarily confirm that they This idea is bolstered by the discovery, evolved for this purpose. reported by Xu et al. in 2015, of a small Much more fundamental questions re- theropod, called Yi qi, with a bat-like mem- main, however. For all the buzz about feath- brane of skin stretched between its fingers ered dinosaur discoveries, there is little and body (12). All other known winged clarity on which dinosaurs could actually dinosaurs have pennaceous feathers. It Downloaded from fly. Answering this requires quantitative seems likely that Yi qi’s skin membrane is biomechanical studies that test whether an airfoil, probably used for gliding, as the different wing and feather configurations, animal probably didn’t have the wing flex- of Chinese fossils with exquisite details of across a range of dinosaurs, enabled vo- ibility to flap. The unusual construction of feathers and other soft tissues (2). When lant activity—be it powered flight, incipient its wings is striking evidence that different placed together on a family tree, these fos- flapping flight, gliding, or other airborne forms of dinosaur flight evolved in differ- sils show that many anatomical compo- behaviors. This, in turn, would help untan- ent groups. nents of the modern-bird flight apparatus gle which dinosaurs could fly, using which The Mesozoic Era (from 252 million to evolved piecemeal over tens of millions of styles, and how flight performance evolved 66 million years ago) must have been aflut- years of dinosaur evolution, for reasons un- across the dinosaur family tree. Most work ter with an aviary of dinosaurs, experiment- related to flight (1, 3). in this area has focused on the four-winged ing with different ways of navigating the air Even the most quintessential avian fea- Microraptor, shown by mathematical and (see the figure). But much work remains to ture of all—feathers—did not evolve as a physical models to have been a capable be done to better understand this heady pe- flight adaptation. They began as hairlike glider (10). Microraptor is known from riod of evolution. Mathematical models are filaments, which the earliest dinosaurs ac- many specimens amenable to study, but it a good start, but the next big breakthrough quired, most likely, for insulation (4) or to alone is not a panacea for understanding will come with more advanced anatomical camouflage their bodies (5). Xing et al.’s the origin of flight. We must know more models of early birds and their close dino- stunning discovery of a dinosaur tail in am- about other winged dinosaurs, too. saur relatives—physical ones (10, 13) that ber, announced in late 2016, gives the first A biomechanical study published last can be subjected to wind-tunnel experi- glimpse at the three-dimensional structure year by Dececchi et al. (11) is a step in the ments, and digital ones that can be analyzed SCIENCE sciencemag.org 24 FEBRUARY 2017 • VOL 355 ISSUE 6327 793 Published by AAAS DA_0224Perspectives.indd 793 2/22/17 10:35 AM INSIGHTS | PERSPECTIVES The route from wings to flight PROTEOME Wings evolved in some early dinosaurs, but flight capabilities arose later. It is likely that only birds were capable of full-powered flight. Dinosaurs are not to scale. Quantifying Psittacosaurus Could not ;y protein Filaments Dilong (dis)order Protofeathers Could not ;y Pennaceous feathers A study of bacteria, yeast, Bat-like skin membrane and mammalian cells maps Ornithomimus Probably could not ;y the melting behavior of cellular proteins By Christine Vogel Caudipteryx Could not ;y wenty-five years ago, Chothia pre- dicted that the structural domains of Yi all proteins can be classified into about Plausible glider 1000 folds (1). Later studies refined this number; however, scientists also found that some proteins or parts of proteins Protofeathers; on March 7, 2017 could not ;y Tnever assume a specific fold. These regions are called intrinsically unstructured or disor- Dinosaurs Microraptor dered (2). Oncogenes such as p53 or breast Plausible glider/incipient cancer 1 (BRCA1) contain long disordered ;apping ;ight stretches, and aggregation of the disordered Theropods a-synuclein is thought to underlie Parkin- Maniraptoriforms Zhenyuanlong son’s and Alzheimer’s diseases (3, 4). On page Had wings, but Probably could not ;y 812 of this issue, Leuenberger et al. (5) map probably could not ;y the thermodynamic stabilities of more than Pennaraptorans 8000 proteins across four organisms, provid- Had wings; possible ing insights not only into the evolution of incipient ;ight protein structure and expression in cells but http://science.sciencemag.org/ Anchiornis also into possible molecular causes and con- Plausible glider/incipient sequences of human disease. Paravians ;apping ;ight Had wings; possible Many early studies of protein structural incipient ;ight stability relied on careful examination of one protein at a time, studying either its crys- Archaeopteryx tallographic structure or the temperature- Incipient or more advanced dependent change in its optical properties, Downloaded from ;apping ;ight such as circular dichroism (CD) (6). Later, Avians (birds) other methods were developed, but they were Had wings; capable either slow, not quantitative, or relied purely of powered ;ight Modern bird on computational predictions (7).