Feathered dinosaurs

Feathered dinosaurs is a term used to describe dinosaurs, particularly maniraptoriform dromaeosaurs, that were covered in plumage, either filament-like intergumentary structures with few branches, to fully developed pennaceous feathers complete with shafts and vanes. Feathered dinosaurs first came to realization after it was discovered that dinosaurs are closely related to birds. Since then, the term "feathered dinosaurs" has widened to encompass the entire concept of the dinosaur–bird relationship, including the various avian characteristics some dinosaurs possess, including a pygostyle, a posteriorly oriented pelvis, elongated arms and forelimbs and clawed hand, and clavicles fused to form a furcula. A substantial amount of evidence demonstrates that birds are the descendants of theropod dinosaurs, and that birds evolved during the Jurassic from small, feathered maniraptoran theropods closely related to dromaeosaurids and troodontids (known collectively as deinonychosaurs). Less than two dozen species of dinosaurs have been discovered with direct fossil evidence of plumage since the 1990s, with most coming from Cretaceous deposits in China, most notably Liaoning Province. Together, these fossils represent an important transition between dinosaurs and birds, which allows paleontologists to piece together the origin and evolution of birds.

Despite integumentary structures being limited to non-avian dinosaurs, particularly well-documented in maniraptoriformes, fossils do suggest that a large number of theropods were feathered, and it has even been suggested that based on phylogenetic analyses, Tyrannosaurus at one stage of its life may have been covered in down-like feathers, although there is no direct fossil evidence of this. Based on what is known of the dinosaur fossil record, paleontologists generally think that most of dinosaur evolution happened at relatively large body size (a mass greater than a few kilograms), and in animals that were entirely terrestrial. Small size (&lt;1 kg) and arboreal habits seem to have arisen fairly late during dinosaurian evolution, and only within maniraptora.

Birds were originally linked with other dinosaurs back in the late 1800s, most famously by Thomas Huxley. This view remained fairly popular until the 1920s when Gerhard Heilmann's book The Origin of Birds was published in English. Heilmann argued that birds could not have descended from dinosaurs (predominantly because dinosaurs lacked clavicles, or so he thought), and he therefore favored the idea that birds originated from the so-called 'pseudosuchians': primitive archosaurs that were also thought ancestral to dinosaurs and crocodilians. This became the mainstream view until the 1970s, when a new look at the anatomical evidence (combined with new data from maniraptoran theropods) led John Ostrom to successfully resurrect the dinosaur hypothesis. The best and most complete review of bird origin theories is that provided by Witmer (1991).

Fossils of Archaeopteryx include well-preserved feathers, but it was not until the early 1990s that clearly nonavian dinosaur fossils were discovered with preserved feathers. Today there are more than twenty genera of dinosaurs with fossil feathers, nearly all of which are theropods. Most are from the Yixian Formation in China. The fossil feathers of one specimen, Shuvuuia deserti, have even tested positive for beta-keratin, the main protein in bird feathers, in immunological tests.

Early hypotheses
Shortly after the 1859 publication of Charles Darwin's The Origin of Species, the ground-breaking book which described his theory of evolution by natural selection, British biologist and evolution-defender Thomas Henry Huxley proposed that birds were descendants of dinosaurs. He compared skeletal structure of Compsognathus, a small theropod dinosaur, and the 'first bird' Archaeopteryx lithographica (both of which were found in the Upper Jurassic Bavarian limestone of Solnhofen). He showed that, apart from its hands and feathers, Archaeopteryx was quite similar to Compsognathus. In 1868 he published On the Animals which are most nearly intermediate between Birds and Reptiles, making the case. The leading dinosaur expert of the time, Richard Owen, disagreed, claiming Archaeopteryx as the first bird outside dinosaur lineage. For the next century, claims that birds were dinosaur descendants faded, with more popular bird-ancestry hypotheses including 'crocodylomorph' and 'thecodont' ancestors, rather than dinosaurs or other archosaurs.

Aboreal dinosaurs
Thanks to the discovery of such theropods as Microraptor and Epidendrosaurus, we do now have small forms exhibiting some features suggestive of a tree-climbing (or scansorial) way of life. However, the idea that dinosaurs might have climbed trees goes back a long way, and well pre-dates the dinosaur renaissance of the 1960s and 70s.

The idea of scansoriality in non-avian dinosaurs has been considered a 'fringe' idea, and it's partly for this reason that, prior to 2000, nobody had attempted any sort of review on the thoughts that had been published about the subject. The oldest reference to scansoriality in a dinosaur comes from William Fox, the Isle of Wight curator and amateur fossil collector, who in 1866 proposed that Calamospondylus oweni from the Lower Cretaceous Wessex Formation of the Isle of Wight might have been in the habit of 'leaping from tree to tree'. The Calamospondylus oweni specimen that Fox referred to was lost, and the actual nature of the fossil remains speculative, but there are various reasons for thinking that it was a theropod. However, it's not entirely accurate to regard Fox's ideas about Calamospondylus as directly relevant to modern speculations about tree-climbing dinosaurs given that, if Fox imagined Calamospondylus oweni as resembling anything familiar, it was probably as a lizard-like reptile, and not as a dinosaur as they are currently understood.

During the early decades of the 20th century the idea of tree-climbing dinosaurs became reasonably popular as Othenio Abel, Gerhard Heilmann and others used comparisons with birds, tree kangaroos and monkeys to argue that the small ornithopod Hypsilophodon (also from the Wessex Formation of the Isle of Wight) was scansorial. Heilmann had come to disagree with this idea and now regarded Hypsilophodon as terrestrial. William Swinton favored the idea of a scansorial Hypsilophodon, concluding that 'it would be able to run up the stouter branches and with hands and tail keep itself balanced until the need for arboreal excursions had passed', and in a 1936 review of Isle of Wight dinosaurs mentioned the idea that small theropods might also have used their clawed hands to hold branches when climbing.

During the 1970s, Peter Galton was able to show that all of the claims made about the forelimb and hindlimb anatomy of Hypsilophodon supposedly favoring a scansorial lifestyle were erroneous, and that this animal was in fact well suited for an entirely terrestrial, cursorial lifestyle. Nevertheless, for several decades Hypsilophodon was consistently depicted as a tree-climber.

In recent decades, Gregory Paul has been influential in arguing that small theropods were capable climbers, and he not only argued for and illustrated scansorial abilities in coelurosaurs, he also proposed that as-yet-undiscovered maniraptorans were highly proficient climbers and included the ancestors of birds. The hypothesised existence of small arboreal theropods that are as yet unknown from the fossil record later proved integral to George Olshevsky's 'birds come first' (BCF) hypothesis. Olshevsky argued that all dinosaurs, and in fact all archosaurs, descend from small, scansorial ancestors, and that it is these little climbing reptiles which are the direct ancestors of birds.

Ostrom, Deinonychus and the Dinosaur Renaissance
In 1964, the first specimen of Deinonychus antirrhopus was discovered in Montana, and in 1969, John Ostrom of Yale University described Deinonychus as a theropod whose skeletal resemblance to birds seemed unmistakable. Since that time, Ostrom had become a leading proponent of the theory that birds are direct descendants of dinosaurs. During the late 1960s, Ostrom and others demonstrated that maniraptoran dinosaurs could fold their arms in a manner similar to that of birds. Further comparisons of bird and dinosaur skeletons, as well as cladistic analysis strengthened the case for the link, particularly for a branch of theropods called maniraptors. Skeletal similarities include the neck, the pubis, the wrists (semi-lunate carpal), the 'arms' and pectoral girdle, the shoulder blade, the clavicle and the breast bone. In all, over a hundred distinct anatomical features are shared by birds and theropod dinosaurs.

Other researchers drew on these shared features and other aspects of dinosaur biology and began to suggest that at least some theropod dinosaurs were feathered. The first restoration of a feathered dinosaur was Sarah Landry's depiction of a feathered "Syntarsus" (now renamed Megapnosaurus or considered a synonym of Coelophysis), in Robert T. Bakker's 1975 publication Dinosaur Renaissance. Gregory S. Paul was probably the first paleoartist to depict maniraptoran dinosaurs with feathers and protofeathers, starting in the late 1980s.

By the 1990s, most paleontologists considered birds to be surviving dinosaurs and referred to 'non-avian dinosaurs' (all extinct), to distinguish them from birds (aves). Before the discovery of feathered dinosaurs, the evidence was limited to Huxley and Ostrom's comparative anatomy. Some mainstream ornithologists, including Smithsonian Institution curator Storrs L. Olson, disputed the links, specifically citing the lack of fossil evidence for feathered dinosaurs.

Modern research and feathered dinosaurs in China
The early 1990s saw the discovery of spectacularly preserved bird fossils in several Early Cretaceous geological formations in the northeastern Chinese province of Liaoning. South American paleontologists, including Fernando Novas and others, discovered evidence showing that maniraptorans could also move their arms in a bird-like manner. Gatesy and others suggested that anatomical changes to the vertebral column and hindlimbs occured before birds first evolved, and Xu Xing and colleagues proved that true functional wings and flight feathers evolved in some maniraptorans, all strongly suggesting that these anatomical features were already well-developed before the first birds evolved.

In 1996, Chinese paleontologists described Sinosauropteryx as a new genus of bird from the Yixian Formation, but this animal was quickly recognized as a theropod dinosaur closely related to Compsognathus. Surprisingly, its body was covered by long filamentous structures. These were dubbed 'protofeathers' and considered to be homologous with the more advanced feathers of birds, although some scientists disagree with this assessment. Chinese and North American scientists described Caudipteryx and Protarchaeopteryx soon after. Based on skeletal features, these animals were non-avian dinosaurs, but their remains bore fully-formed feathers closely resembling those of birds. "Archaeoraptor," described without peer review in a 1999 issue of National Geographic, turned out to be a smuggled forgery, but legitimate remains continue to pour out of the Yixian, both legally and illegally. Many newly described feathered dinosaurs preserve horny claw sheaths, integumentary structures (filaments to fully pennaceous feathers), and internal organs. Feathers or "protofeathers" have been found on a wide variety of theropods in the Yixian, and the discoveries of extremely bird-like dinosaurs, as well as dinosaur-like primitive birds, have almost entirely closed the morphological gap between theropods and birds.

Features linking birds and dinosaurs
Over two dozen distinct anatomical features are shared by birds and theropod dinosaurs.
 * 1) Pubis (one of the three bones making up the vertebrate pelvis) shifted from an anterior to a more posterior orientation (see Saurischia), and bearing a small distal "boot".
 * 2) Elongated arms and forelimbs and clawed manus (hands).
 * 3) Large orbits (eye openings in the skull).
 * 4) Flexible wrist with a semi-lunate carpal (wrist bone).
 * 5) Double-condyled dorsal joint on the quadrate bone.
 * 6) Ucinate process of the ribs.
 * 7) Lateral exposition of the glenoid in the humeral joint.
 * 8) Posterior
 * 9) Hollow, thin-walled bones.
 * 10) 3-fingered opposable grasping manus (hand), 4-toed pes (foot); but supported by 3 main toes.
 * 11) Reduced, posteriorly stiffened tail.
 * 12) Elongated metatarsals (bones of the feet between the ankle and toes).
 * 13) S-shaped curved neck.
 * 14) Erect, digitgrade (ankle held well off the ground) stance with feet postitioned directly below the body.
 * 15) Similar eggshell microstructure.
 * 16) Teeth with a constriction between the root and the crown.
 * 17) Functional basis for wing power stroke present in arms and pectoral girdle (during motion, the arms were swung down and forward, then up and backwards, describing a "figure-eight" when viewed laterally).
 * 18) Expanded pneumatic sinuses in the skull.
 * 19) Five or more vertebrae incorporated into the sacrum (hip).
 * 20) Posterior caudal vertebrate fused to form the pygostyle.
 * 21) Straplike scapula (shoulder blade).
 * 22) Clavicles (collarbone) fused to form a furcula (wishbone).
 * 23) Hingelike ankle joint, with movement mostly restricted to the fore-aft plane.
 * 24) Secondary bony palate (nostrils open posteriorly in throat).
 * 25) Pennaceous feathers in some taxa. Proto-feathers, filaments, and integumenatary structures in others.

To date, these features have been discovered in dromaeosaurids, oviraptorosaurids, troodontids, alvarezsaurids, ornithomimids, therizinosaurids, and tyrannosaurids.


 * Source 1: Are Birds Really Dinosaurs? Dinobuzz, Current Topics Conserning Dinosaurs. Created 9/27/05. Accessed 7/20/09. Copyright 1994-2009 by the Regents of the University of California, all rights reserved.
 * Source 2: Kurochkin, E., N. 2006. Parallel Evolution of Theropod Dinoaurs and Birds. Entomological Review 86 (1), pp. S45-S58. Some of the more interesting similarities are discussed here:

Feathers
Archaeopteryx, the first good example of a "feathered dinosaur", was discovered in 1861. The initial specimen was found in the solnhofen limestone in southern Germany, which is a lagerstätte, a rare and remarkable geological formation known for its superbly detailed fossils. Archaeopteryx is a transitional fossil, with features clearly intermediate between those of modern reptiles and birds. Discovered just two years after Darwin's seminal Origin of Species, its discovery spurred the nascent debate between proponents of evolutionary biology and creationism. This early bird is so dinosaur-like that, without a clear impression of feathers in the surrounding rock, at least one specimen was mistaken for Compsognathus.

Since the 1990s, a number of additional feathered dinosaurs have been found, providing even stronger evidence of the close relationship between dinosaurs and modern birds. Most of these specimens were unearthed in Liaoning province, northeastern China, which was part of an island continent during the Cretaceous period. Though feathers have been found only in the lagerstätte of the Yixian Formation and a few other places, it is possible that non-avian dinosaurs elsewhere in the world were also feathered. The lack of widespread fossil evidence for feathered non-avian dinosaurs may be due to the fact that delicate features like skin and feathers are not often preserved by fossilization and thus are absent from the fossil record. A recent development in the debate centers around the discovery of impressions of "protofeathers" surrounding many dinosaur fossils. These protofeathers suggest that the tyrannosauroids may have been feathered. However, others claim that these protofeathers are simply the result of the decomposition of collagenous fiber that underlaid the dinosaurs' integument. The Dromaeosauridae family, in particular, seems to have been heavily feathered and at least one dromaeosaurid, Cryptovolans, may have been capable of flight.

Skeleton
Because feathers are often associated with birds, feathered dinosaurs are often touted as the missing link between birds and dinosaurs. However, the multiple skeletal features also shared by the two groups represent the more important link for paleontologists. Furthermore, it is increasingly clear that the relationship between birds and dinosaurs, and the evolution of flight, are more complex topics than previously realized. For example, while it was once believed that birds evolved from dinosaurs in one linear progression, some scientists, most notably Gregory S. Paul, conclude that dinosaurs such as the dromaeosaurs may have evolved from birds, losing the power of flight while keeping their feathers in a manner similar to the modern ostrich and other ratites.

Comparisons of bird and dinosaur skeletons, as well as cladistic analysis, strengthens the case for the link, particularly for a branch of theropods called maniraptors. Skeletal similarities include the neck, pubis, wrist (semi-lunate carpal), arm and pectoral girdle, shoulder blade, clavicle, and breast bone.

Wishbones
The wishbone (or furcula) was long believed to be a structure unique to birds, formed by the fusion of the two collarbones (clavicles) into a single V-shaped structure that helps brace the skeleton against the stresses incurred while flapping. The reptiles ancestral to birds, therefore, should, at the very least, show well-developed clavicles. But by 1926, when Danish artist and naturalist Gerhard Heilmann published his monumentally influential The Origin of Birds, no clavicles had been reported in any theropod dinosaur. Noting this fact, Heilmann suggested that birds evolved from a more generalized archosaurian ancestor, such as the aptly-named Ornithosuchus (literally, “bird-crocodile”), which is now believed to be closer to the crocodile end of the archosaur lineage. At the time, however, Ornithosuchus seemed to be a likely ancestor of more birdlike creatures.

In the 1980s, paleontologists began to realize that theropods not only possessed clavicles, but also possessed clavicles that were fused into a wishbone. In 1983, Rinchen Barsbold reported the first dinosaurian furcula from a specimen of the Cretaceous theropod Oviraptor. A furcula-bearing Oviraptor specimen had previously been known since the 1920s, but because at the time the theropod origin of birds was largely dismissed, it was misidentified for sixty years. p. 9

Following this discovery, paleontologists began to find furculae in other theropod dinosaurs. Wishbones are now known from the dromaeosaur Velociraptor, the allosauroid Allosaurus, and the tyrannosaurid Tyrannosaurus rex, to name a few.

In 2000, Alex Downs reported an isolated furcula found within a block of Coelophysis bauri skeletons from the Late Triassic Rock Point Formation at Ghost Ranch, New Mexico. While it seemed likely that it originally belonged to Coelophysis, the block contained fossils from other Triassic animals as well, and Alex declined to make a positive identification. Currently, a total of five C. bauri furculae have been found in the New Mexico Museum of Natural History's (NMMNH) Ghost Ranch, New Mexico Whitaker Quarry block C-8-82. Three of the furculae are articulated in juvenile skeletons; two of these are missing fragments but are nearly complete, and one is apparently complete. Two years later, Tykoski et al. described several furculae from two species of the coelophysoid genus Syntarsus (now Megapnosaurus), S. rhodesiensis and S. kayentakatae, from the Early Jurassic of Zimbabwe and Arizona, respectively. Syntarsus was long considered to be the genus most closely related to Coelophysis, differing only in a few anatomical details and slightly younger age, so the identification of furculae in Syntarsus made it very likely that the furcula Alex Downs noted in 2000 came from Coelophysis after all. By 2006, wishbones were definitively known from the Early Jurassic Coelophysis rhodesiensis and Coelophysis kayentakatae, and a single isolated furcula was known that might have come from the Late Triassic type species, Coelophysis bauri.

Avian air sacs
Large meat-eating dinosaurs had a complex system of air sacs similar to those found in modern birds, according to an investigation which was led by Patrick O'Connor of Ohio University. The lungs of theropod dinosaurs (carnivores that walked on two legs and had birdlike feet) likely pumped air into hollow sacs in their skeletons, as is the case in birds. "What was once formally considered unique to birds was present in some form in the ancestors of birds", O'Connor said. In a paper published in the online journal Public Library of Science ONE (September 29, 2008), scientists described Aerosteon riocoloradensis, the skeleton of which supplies the strongest evidence to date of a dinosaur with a bird-like breathing system. CT-scanning revealed the evidence of air sacs within the body cavity of the Aerosteon skeleton.

Gizard stones
Another piece of evidence that birds and dinosaurs are closely related is the use of gizzard stones. These stones are swallowed by animals to aid digestion and break down food and hard fibres once they enter the stomach. When found in association with fossils, gizzard stones are called gastroliths.

Heart and sleeping posture
Modern computed tomography (CT) scans of a dinosaur chest cavity conducted in 2000 found the apparent remnants of complex four-chambered hearts, much like those found in today's mammals and birds. The idea is controversial within the scientific community, coming under fire for bad anatomical science or simply wishful thinking. The type fossil of the troodont, Mei, is complete and exceptionally well preserved in three-dimensional detail, with the snout nestled beneath one of the forelimbs, similar to the roosting position of modern birds. This demonstrates that the dinosaurs slept like certain modern birds, with their heads tucked under their arms. This behavior, which may have helped to keep the head warm, is also characteristic of modern birds.

Reproductive biology
A discovery of features in a Tyrannosaurus rex skeleton recently provided more evidence that dinosaurs and birds evolved from a common ancestor and, for the first time, allowed paleontologists to establish the sex of a dinosaur. When laying eggs, female birds grow a special type of bone in their limbs between the hard outer bone and the marrow. This medullary bone, which is rich in calcium, is used to make eggshells. The presence of endosteally derived bone tissues lining the interior marrow cavities of portions of the Tyrannosaurus rex specimen's hind limb suggested that T. rex used similar reproductive strategies, and revealed the specimen to be female. Further research has found medullary bone in the theropod Allosaurus and ornithopod Tenontosaurus. Because the line of dinosaurs that includes Allosaurus and Tyrannosaurus diverged from the line that led to Tenontosaurus very early in the evolution of dinosaurs, this suggests that dinosaurs in general produced medullary tissue. Medullary bone has been found in specimens of sub-adult size, which suggests that dinosaurs reached sexual maturity rather quickly for such large animals.

Brooding and care of young
Several Citipati specimens have been found resting over the eggs in its nest in a position most reminiscent of brooding. A dinosaur embryo was found without teeth, which suggests some parental care was required to feed the young dinosaur, possibly the adult dinosaur regurgitated food into the young dinosaur's mouth (see altricial). This behaviour is seen in numerous bird species; parent birds regurgitate food into the hatchling's mouth.

Molecular evidence and soft tissue
One of the best examples of soft tissue impressions in a fossil dinosaur was discovered in Petraroia, Italy. The discovery was reported in 1998, and described the specimen of a small, very young coelurosaur, Scipionyx samniticus. The fossil includes portions of the intestines, colon, liver, muscles, and windpipe of this immature dinosaur.

In the March 2005 issue of Science, Dr. Mary Higby Schweitzer and her team announced the discovery of flexible material resembling actual soft tissue inside a 68-million-year-old Tyrannosaurus rex leg bone from the Hell Creek Formation in Montana. After recovery, the tissue was rehydrated by the science team. The seven collagen types obtained from the bone fragments, compared to collagen data from living birds (specifically, a chicken), reveal that older theropods and birds are closely related.

When the fossilized bone was treated over several weeks to remove mineral content from the fossilized bone marrow cavity (a process called demineralization), Schweitzer found evidence of intact structures such as blood vessels, bone matrix, and connective tissue (bone fibers). Scrutiny under the microscope further revealed that the putative dinosaur soft tissue had retained fine structures (microstructures) even at the cellular level. The exact nature and composition of this material, and the implications of Dr. Schweitzer's discovery, are not yet clear; study and interpretation of the specimens is ongoing.

The successful extraction of ancient DNA from dinosaur fossils has been reported on two separate occasions, but upon further inspection and peer review, neither of these reports could be confirmed. However, a functional visual peptide of a theoretical dinosaur has been inferred using analytical phylogenetic reconstruction methods on gene sequences of related modern species such as reptiles and birds. In addition, several proteins have putatively been detected in dinosaur fossils, including hemoglobin.

Feather structures and anatomy
Feathers vary in length according to their position on the body, with the filaments of the compsognathid Sinosauropteryx being 13 mm and 21 mm long on the neck and shoulders respectively. In contrast, the structures on the skull are about 5 mm long, those on the arm about 2 mm long, and those on the distal part of the tail about 4 mm long. Because the structures tend to be clumped together it is difficult to be sure of an individual filament's morphology. The structures might have been simple and unbranched, but Currie &amp; Chen (2001) thought that the structures on Sinosauropteryx might be branched and rather like the feathers of birds that have short quills but long barbs. The similar structures of Dilong also appear to exhibit a simple branching structure.

Exactly how feathers were arranged on the arms and hands of both basal birds and non-avian maniraptorans has long been unclear, and both non-avian maniraptorans and archaeopterygids have conventionally been depicted as possessing unfeathered fingers. However, the second finger is needed to support the remiges, and therefore must have been feathered. Derek Yalden's 1985 study was important in showing exactly how the remiges would have grown off of the first and second phalanges of the archaeopterygid second finger, and this configuration has been widely recognized. p. 129-159

However, there has been some minor historical disagreement over exactly how many remiges were present in archaeopterygids (there were most likely 11 primaries and a tiny distal 12th one, and at least 12 secondaries), and also about how the hand claws were arranged. The claws were directed perpendicularly to the palmar surface in life, and rotated anteriorly in most (but not all) specimens during burial. p. 129-159 It has also been suggested on occasion that the fingers of archaeopterygids and other feathered maniraptorans were united in a single fleshy 'mitten' as they are in modern birds, and hence unable to be employed in grasping. However, given that the interphalangeal finger joints of archaeopterygids appear suited for flexion and extension, and that the third finger apparently remained free and flexible in birds more derived than archaeopterygids, this is unlikely to be correct; it's based on a depression in the sediment that was identified around the bones.

Like those of archaeopterygids and modern birds, the remiges of non-avian theropods would also have been attached to the phalanges of the second manual digit as well as to the metacarpus and ulna, and indeed we can see this in the fossils. It's the case in the sinornithosaur NGMC 91-A and Microraptor. Surprisingly, in Caudipteryx, the remiges are restricted to the hands alone, and don't extend from the arm. They seem to have formed little 'hand flags' that are unlikely to have served any function other than display. Caudipteryx is an oviraptorosaur and possesses a suite of characters unique to this group. It is not a member of Aves, despite the efforts of some workers to make it into one. The hands of Caudipteryx supported symmetrical, pennaceous feathers that had vanes and barbs, and that measured between 15–20 centimeters long (6–8 inches). These primary feathers were arranged in a wing-like fan along the second finger, just like primary feathers of birds and other maniraptorans. No fossil of Caudipteryx preserves any secondary feathers attached to the forearms, as found in dromaeosaurids, Archaeopteryx and modern birds. Either these arm feathers are not preserved, or they were not present on Caudipteryx in life. An additional fan of feathers existed on its short tail. The shortness and symmetry of the feathers, and the shortness of the arms relative to the body size, indicate that Caudipteryx could not fly. The body was also covered in a coat of short, simple, down-like feathers.

Evolution
Feathers are extremely complex integumentary structures that characterize a handful of vertebrate animals. Although it is generally acknowledged that feathers are derived and evolved from simpler integumentary structures, the early diversification and origin of feathers was relatively unknown until recently, and current research is ongoing. Since the theropod ancestry of birds is widely supported with osteological and other physical lines of evidence, the precursors of feathers in dinosaurs are also present, as predicted by those who originally proposed a theropod origin for birds.

Original function
The functional view on the evolution of feathers has traditionally focussed on insulation, flight and display. Discoveries of non-flying Late Cretaceous feathered dinosaurs in China however suggest that flight could not have been the original primary function. Feathers in dinosaurs indicate that their original function was not flight, but of a different nature. Theories include insulation brought around after they had metabolically changed from their cold-blooded reptilian ancestors, to increasing running speed. It has been suggested that vaned feathers evolved in the context of thrust, with running, non-avian theropods flapping their arms to increase their running speed.

Evolutionary models
The following is the generally acknowledged version of the origin and early evolution of feathers:
 * 1) The first feathers evolved; they are single filaments.
 * 2) Branching structures developed.
 * 3) The rachis evolved.
 * 4) Pennaceous feathers evolved.
 * 5) Aerodynamic morphologies appeared. (curved shaft and asymmetrical vanes)

This scenario appears to indicate that downy, contour, and flight feathers, are more derived forms of the first "feather". However, it is also possible that protofeathers and basal feathers disappeared early on in the evolution of feathers and that more primitive feathers in modern birds are secondary. This would imply that the feathers in modern birds have nothing to do with protofeathers.

Protofeathers, filaments, and stage 1 structures
Models of feather evolution are often proposed that the earliest prototype feathers were hair-like integumentary filaments similar to the structures of Sinosauropteryx and Dilong.

In Prum's model of feather evolution, hollow quill-like integumentary structures of this sort were termed Stage 1 feathers. The idea that feathers started out as hollow quills also supports Alan Brush's idea that feathers are evolutionary novelties, and not derived from scales. However, in order to determine the homology of Stage 1 feathers, it is necessary to determine their proteinaceous content: unlike the epidermal appendages of all other vertebrates, feathers are almost entirely composed of beta-keratins (as opposed to alpha-keratins) and, more specifically, they are formed from a group of beta-keratins called phi-keratins. No studies have yet been performed on the Stage 1 structures of Sinosauropteryx or Dilong in order to test their proteinaceous composition, however, tiny filamentous structures discovered adjacent to the bones of the alvarezsaurid Shuvuuia have been tested for beta-keratin, and the structures were discovered to be composed of beta-keratin. Alvarezsaurids have been of controversial phylogenetic position, but are generally agreed to be basal members of the Maniraptora clade. Due to this discovery, paleontologists are now convinced that beta-keratin-based protofeathers had evolved at the base of this clade at least.

Vaned, pennaceous feathers
While basal coelurosaurs possessed these apparently hollow quill-like 'Stage 1' filaments, they lacked the more complex structures seen in maniraptorans. Maniraptorans possessed vaned feathers with barbs, barbules and hooklets just like those of modern birds. However, their bodies were not covered in vaned feathers as are those of the majority of living birds: instead, it seems that they were at least partly covered in the more simple structures that they had inherited from basal coelurosaurs like Sinosauropteryx. This condition may have been retained all the way up into basal birds: despite all those life restorations clothing archaeopterygids in vaned breast, belly, throat and neck feathers, it seems that their bodies also were at least partly covered in the more simple filamentous structures. The Berlin Archaeopteryx specimen appears to preserve such structures on the back of the neck though pennaceous vaned feathers were present on its back, at least.

Though it has been suggested at times that vaned feathers simply must have evolved for flight, the phylogenetic distribution of these structures currently indicates that they first evolved in flightless maniraptorans and were only later exapted by long-armed maniraptorans for use in locomotion. Of course a well-known minority opinion, best known from the writings of Gregory Paul, is that feathered maniraptorans are secondarily flightless and descend from volant bird-like ancestors. While this hypothesis remains possible, it lacks support from the fossil record, though that may or may not mean much, as the fossil record is incomplete and prone to selection bias.

Display feathers
The discovery of Epidexipteryx represented the earliest known examples of ornamental feathers in the fossil record. Epidexipteryx is known from a well preserved partial skeleton that includes four long feathers on the tail, composed of a central rachis and vanes. However, unlike in modern-style rectrices (tail feathers), the vanes were not branched into individual filaments but made up of a single ribbon-like sheet. Epidexipteryx also preserved a covering of simpler body feathers, composed of parallel barbs as in more primitive feathered dinosaurs. However, the body feathers of Epidexipteryx are unique in that some appear to arise from a "membranous structure." The skull of Epidexipteryx is also unique in a number of features, and bears an overall similarity to the skull of Sapeornis, oviraptorosaurs and, to a lesser extent, therizinosauroids. The tail of Epidexipteryx bore unusual vertebrae towards the tip which resembled the feather-anchoring pygostyle of modern birds and some oviraptorosaurs. Despite its close relationship to avialan birds, Epidexipteryx appears to have lacked remiges (wing feathers), and it likely could not fly. Zhang et al. suggest that unless Epidexipteryx evolved from flying ancestors and subsequently lost its wings, this may indicate that advanced display feathers on the tail may have predated flying or gliding flight.

According to the model of feather evolution developed by Prum & Brush (2002, 2003) feathers started out ('stage 1') as hollow cylinders, then ('stage 2') became unbranched barbs attached to a calamus. By stage 3, feathers were planar structures with the barbs diverging from a central rachis, and from there pennaceous feathers. The feathers of Epidexipteryx may represent stage 2 structures, but also suggests that a more complicated sequence of steps in the evolution of feathers took place.

Use in predation
Several maniraptoran lineages were clearly predatory and, given the morphology of their manual claws, fingers and wrists, presumably in the habit of grabbing at prey with their hands. Contrary to popular belief, feathers on the hands would not have greatly impeded the use of the hands in predation. Because the feathers are attached at an angle roughly perpendicular to the claws, they are oriented tangentially to the prey's body, regardless of prey size. p. 315 It is important to note here that theropod hands appear to have been oriented such that the palms faced medially (facing inwards), and were not parallel to the ground as used to be imagined.

However, feathering would have interfered with the ability of the hands to bring a grasped object up toward the mouth given that extension of the maniraptoran wrist would have caused the hand to rotate slightly upwards on its palmar side. If both feathered hands are rotated upwards and inwards at the same time, the remiges from one hand would collide with those of the other. For this reason, maniraptorans with feathered hands could grasp objects, but would probably not be able to carry them with both hands. However, dromaeosaurids and other maniraptorans may have solved this problem by clutching objects single-handedly to the chest. Feathered hands would also have restricted the ability of the hands to pick objects off of the ground, given that the feathers extend well beyond the ends of the digits. It remains possible that some maniraptorans lacked remiges on their fingers, but the only evidence available indicates the contrary. It has recently been argued that the particularly long second digit of the oviraptorosaur Chirostenotes was used as a probing tool, locating and extracting invertebrates and small mammals and so on from crevices and burrows. It seems highly unlikely that a digit that is regularly thrust into small cavities would have had feathers extending along its length, so either Chirostenotes didn't probe as proposed, or its second finger was unfeathered, unlike that of Caudipteryx and the other feathered maniraptorans. Given the problems that the feathers might have posed for clutching and grabbing prey from the ground, we might also speculate that some of these dinosaurs deliberately removed their own remiges by biting them off. Some modern birds (notably motmots) manipulate their own feathers by biting off some of the barbs, so this is at least conceivable, but no remains in the fossil record have been recovered that support this conclusion.

Homology and comparison to modern feathers
Some feather morphologies in non-avian theropods are comparable to those on modern birds. Single filament like structures are not present in modern feathers, although some birds possess highly specialized feathers that are superficial in appearance to protofeathers in non-avian theropods. Tuft-like structures seen in some maniraptorans are similar to that of the natal down in modern birds. Similarly, structures in the fossil record composed of a series of filaments joined at their bases along a central filament bear an uncanny resemblance to the down feathers in modern birds, with the exception of a lack of barbules. Furthermore, structures on fossils have been recovered from Chinese Cretaceous deposits that are a series of filaments joined at their bases at the distal portion of the central filament bear a superficial resemblance to filoplumes. More derived, pennaceous, feathers on the tails and limbs of feathered dinosaurs are nearly identical to the remiges and retrices of modern birds.

Rahonavis
The ability to fly or glide has been suggested for at least two dromaeosaurid species. The first, Rahonavis ostromi (originally classified as avian bird, but found to be a dromaeosaurid in later studies) may have been capable of powered flight, as indicated by its long forelimbs with evidence of quill knob attachments for long sturdy flight feathers. The forelimbs of Rahonavis were more powerfully built than Archaeopteryx, and show evidence that they bore strong ligament attachments necessary for flapping flight. Luis Chiappe concluded that, given these adaptations, Rahonavis could probably fly but would have been more clumsy in the air than modern birds.

Microraptor
Another species of dromaeosaurid, Microraptor gui, may have been capable of gliding using its well-developed wings on both the fore and hind limbs. Microraptor was among the first non-avian dinosaurs discovered with the impressions of feathers and wings. On Microraptor, the long feathers on the forelimbs possess asymmetrical vanes. The external vanes are narrow, while the internal ones are broad. In addition, Microraptor possessed elongated remiges with asymmetrical vanes that demonstrate aerodynamic function on the hind limbs. A 2005 study by Sankar Chatterjee suggested that the wings of Microraptor functioned like a split-level "biplane", and that it likely employed a phugoid style of gliding, in which it would launch from a perch and swoop downward in a 'U' shaped curve, then lift again to land on another tree, with the tail and hind wings helping to control its position and speed. Chatterjee also found that Microraptor had the basic requirements to sustain level powered flight in addition to gliding.

Microraptor had two sets of wings, on both its forelegs and hind legs. The long feathers on the legs of Microraptor were true flight feathers as seen in modern birds, with asymmetrical vanes on the arm, leg, and tail feathers. As in bird wings, Microraptor had both primary (anchored to the hand) and secondary (anchored to the arm) flight feathers. This standard wing pattern was mirrored on the hind legs, with flight feathers anchored to the upper foot bones as well as the upper and lower leg. It had been proposed by Chinese scientists that the animal glided and probably lived in trees, pointing to the fact that wings anchored to the feet of Microraptor would have hindered their ability to run on the ground, and suggest that all primitive dromaeosaurids may have been arboreal. Sankar Chatterjee determined in 2005 that, in order for the creature to glide or fly, the wings must have been on different levels (as on a biplane) and not overlaid (as on a dragonfly), and that the latter posture would have been anatomically impossible. Using this biplane model, Chatterjee was able to calculate possible methods of gliding, and determined that Microraptor most likely employed a phugoid style of gliding—launching itself from a perch, the animal would have swooped downward in a deep 'U' shaped curve and then lifted again to land on another tree. The feathers not directly employed in the biplane wing structure, like those on the tibia and the tail, could have been used to control drag and alter the flight path, trajectory, etc. The orientation of the hind wings would also have helped the animal control its gliding flight. Chatterjee also used computer algorithms that test animal flight capacity to test whether or not Microraptor was capable of true, powered flight, in addition to passive gliding. The resulting data showed that Microraptor did have the requirements to sustain level powered flight, so it is theoretically possible that the animal flew on occasion in addition to gliding.

Therizinosaurs
Skin impressions from the type specimen of Beipiaosaurus inexpectus indicated that the body was covered predominately by downy feather-like fibers, similar to those of Sinosauropteryx, but longer, and are oriented perpendicular to the arm. Xu et al., who described the specimen, suggested that these downy feathers represent an intermediate stage between Sinosauropteryx and more advanced birds (Avialae).

Unique among known theropods, Beipiaosaurus also possessed a secondary coat of much longer, simpler feathers that rose out of the down layer. These unique feathers (known as EBFFs, or elongated broad filamentous feathers) were first described by Xu et al. in 2009, based on a specimen consisting of the torso, head and neck. Xu and his team also found EBFFs in the original type specimen of B. inexpectus, revealed by further preparation. The holotype also preserved a pygostyle-like structure. The holotype was discovered in two phases. Limb fragments and dorsal and cervical vertebrae were discovered initially. The discovery site was re-excavated later on, and this time an articulated tail and partial pelvis were discovered. All come from the same individual.

The holotype has the largest proto-feathers known of any feathered dinosaur, with the author and paleontologist Xing Xu stating: "Most integumentary filaments are about 50 mm in length, although the longest is up to 70 mm. Some have indications of branching distal ends.". The holotype also had preserved dense patches of parallel integumentary structures in association with its lower arm and leg.

Thick, stiff, spine-like structures were recovered sprouting from the new specimen's throat region, the back of its head, its neck and its back. New preparation of the holotype reveals that the same structures are also present on the tail (though not associated with the pygostyle-like structure). The EBFFs differ from other feather types in that they consist of a single, unbranched filament. Most other primitive feathered dinosaurs have down-like feathers made up of two or more filaments branching out from a common base or along a central shaft. The EBFFs of Beipiaosaurus are also much longer than other primitive feather types, measuring about 100-150 millimeters (4-6 inches) long, roughly half the length of the neck. In Sinosauropteryx, the longest feathers are only about 15% of the neck length. The EBFFs of Beipiaosaurus are also unusually broad, up to 3 mm wide in the type specimen. The broadest feathers of Sinosauropteryx are only 0.2 mm wide, and only slightly wider in larger forms such as Dilong. Additionally, where most primitive feather types are circular in cross section, EBFFs appear to be oval-shaped. None of the preserved EBFFs were curved or bent beyond a broad arc in either specimen, indicating that they were fairly stiff. They were probably hollow, at least at the base.

In a 2009 interview, Xu stated: "Both [feather types] are definitely not for flight, inferring the function of some structures of extinct animals would be very difficult, and in this case, we are not quite sure whether these feathers are for display or some other functions." He speculated that the finer feathers served as an insulatory coat and that the larger feathers were ornamental, perhaps for social interactions such as mating or communication.

Tyrannosauroids


Long filamentous structures have been preserved along with skeletal remains of numerous coelurosaurs from the Early Cretaceous Yixian Formation and other nearby geological formations from Liaoning, China. These filaments have usually been interpreted as "protofeathers," homologous with the branched feathers found in birds and some non-avian theropods, although other hypotheses have been proposed. A skeleton of Dilong was described in 2004 that included the first example of "protofeathers" in a tyrannosauroid. Similarly to down feathers of modern birds, the "protofeathers" found in Dilong were branched but not pennaceous, and may have been used for insulation.

The presence of "protofeathers" in basal tyrannosauroids is not surprising, since they are now known to be characteristic of coelurosaurs, found in other basal genera like Sinosauropteryx, as well as all more derived groups. Rare fossilized skin impressions of large tyrannosaurids lack feathers, however, instead showing skin covered in scales. It is possible that "protofeathers" were present on areas of the body not preserved with skin impressions. Alternatively, secondary loss of "protofeathers" in large tyrannosaurids may be analogous with the similar loss of hair in the largest modern mammals like elephants, where a low surface area-to-volume ratio slows down heat transfer, making insulation by a coat of hair unnecessary. Therefore, as large animals evolve in or disperse into warm climates, a coat of fur or feathers loses its selective advantage for thermal insulation and can instead become a disadvantage, as the insulation traps excess heat inside the body, possibly overheating the animal. Protofeathers may also have been secondarily lost during the evolution of large tyrannosaurids, especially in warm Cretaceous climates. Tyrannosaurus at one stage of its life may have been covered in down-like feathers, although there is no direct fossil evidence of this.

Dromaeosaurids
There is a large body of evidence showing that dromaeosaurids were covered in feathers. Some dromaeosaurid fossils preserve long, pennaceous feathers on the hands and arms (remiges) and tail (rectrices), as well as shorter, down-like feathers covering the body. Other fossils, which do not preserve actual impressions of feathers, still preserve the associated bumps on the forearm bones where long wing feathers would have attached in life. Overall, this feather pattern looks very much like Archaeopteryx.

The first known dromaeosaur with definitive evidence of feathers was Sinornithosaurus, reported from China by Xu et al. in 1999. Many other dromaeosaurid fossils have been found with feathers covering their bodies, some with fully-developed feathered wings. Several even show evidence of a second pair of wings on the hind legs, including Microraptor and Cryptovolans. While direct feather impressions are only possible in fine-grained sediments, some fossils found in coarser rocks show evidence of feathers by the presence of quill knobs, the attachment points for wing feathers possessed by some birds. The dromaeosaurids Rahonavis and Velociraptor have both been found with quill knobs, showing that these forms had feathers despite no impressions having been found. In light of this, it is most likely that even the larger ground-dwelling dromaeosaurids bore feathers, since even flightless birds today retain most of their plumage, and relatively large dromaeosaurids, like Velociraptor, are known to have retained pennaceous feathers. Though some scientists had suggested that the larger dromaeosaurids lost some or all of their insulatory covering, the discovery of feathers in Velociraptor specimens has been cited as evidence that all members of the family retained feathers.

Maniraptora
Maniraptorans are so far the only group of dinosaurs where true, pennaceous feathers were present. More simple integumentary structures were present elsewhere in coelurosaurian theropods, including in compsognathids and tyrannosauroids.

NGMC 91-A, the Sinornithosaurus-like theropod informally dubbed "Dave", possessed unbranched fibers in additional to more complex branched and tufted structures. Protarchaeopteryx is well known for its fan-like array of 12 rectricial feathers, but it also seems to have sported simple filament-like structures elsewhere on the tail. Soft and downy feathers are preserved in the chest region and tail base, and are also preserved adjacent to the femora.

In 2007, quill knobs were found on the ulna (one of 2 "forearm" bones) of a Velociraptor, and the researchers concluded that this theropod had modern-looking feathers, composed of a rachis (stem) and vanes formed by barbs. They thought this might indicate that Velociraptor had flying ancestors - but pointed out that the feathers might have had functions other than flight, e.g. for display or for shielding a nest.

Caudipteryx
Caudipteryx has clear and unambiguously pennaceous feathers, like modern birds, and several cladistic analyses have consistently recovered it as a nonavian, oviraptorid, dinosaur, it provided, at the time of its description, the clearest and most succinct evidence that birds evolved from dinosaurs. Lawrence Witmer stated:

"The presence of unambiguous feathers in an unambiguously nonavian theropod has the rhetorical impact of an atomic bomb, rendering any doubt about the theropod relationships of birds ludicrous.”"

- Lawrence Witmer

However, not all scientists agreed that Caudipteryx was unambiguously non-avian, and some of them continued to doubt that general consensus. Paleornithologist Alan Feduccia sees Caudipteryx as a flightless bird evolving from earlier archosaurian dinosaurs rather than from late theropods. Jones et al. (2000) found that Caudipteryx was a bird based on a mathematical comparison of the body proportions of flightless birds and non-avian theropods. Dyke and Norell (2005) criticized this result for flaws in their mathematical methods, and produced results of their own which supported the opposite conclusion. Other researchers not normally involved in the debate over bird origins, such as Zhou, acknowledged that the true affinities of Caudipteryx were debatable.

Anchiornis
Anchiornis is notable for its proportionally long forelimbs, which measured 80% of the total length of the hind limbs. This is similar to the condition in early avians such as Archaeopteryx, and the authors pointed out that long forelimbs are necessary for flight. It is possible that Anchiornis was able to fly or glide, and may have had a functional airfoil. Anchiornis also had a more avian wrist than other non-avian theropods. Anchiornis has hind leg proportions more like those of lower theropod dinosaurs than avialans. Faint, carbonized feather impressions were preserved in patches in the type specimen. Feathers on the torso measured an average of 20 mm in length, but the feathers were too poorly preserved to ascertain details of their structure. A cladistic analysis indicated that Anchiornis was a basal member of the Avialae and the sister taxon of Aves.

Velociraptor
Fossils of dromaeosaurids more primitive than Velociraptor are known to have had feathers covering their bodies, and fully developed, feathered wings. The fact that the ancestors of Velociraptor were feathered and possibly capable of flight long suggested to paleontologists that Velociraptor bore feathers as well, since even flightless birds today retain most of their feathers.

In September 2007, Alan Turner, Peter Makovicky, and Mark Norell reported the presence of quill knobs on the ulna of a Velociraptor specimen from Mongolia. Fourteen bumps approximately 4mm apart were found in a straight line along the bone, directly corresponding to the same structures in living birds, the bumps serving as an anchor for the secondary feathers. These bumps on bird wing bones show where feathers anchor, and their presence on Velociraptor indicate it too had feathers. According to paleontologist Alan Turner,

"A lack of quill knobs does not necessarily mean that a dinosaur did not have feathers. Finding quill knobs on Velociraptor, though, means that it definitely had feathers. This is something we'd long suspected, but no one had been able to prove."

Co-author Mark Norell, Curator-in-Charge of fossil reptiles, amphibians and birds at the American Museum of Natural History, also weighed in on the discovery, saying:

"The more that we learn about these animals the more we find that there is basically no difference between birds and their closely related dinosaur ancestors like velociraptor. Both have wishbones, brooded their nests, possess hollow bones, and were covered in feathers. If animals like velociraptor were alive today our first impression would be that they were just very unusual looking birds." According to Turner and co-authors Norell and Peter Makovicky, quill knobs are not found in all prehistoric birds, and their absence does not mean that an animal was not feathered – flamingos, for example, have no quill knobs. However, their presence confirms that Velociraptor bore modern-style wing feathers, with a rachis and vane formed by barbs. The forearm specimen on which the quill knobs were found (specimen number IGM 100/981) represents an animal 1.5 meters in length (5 ft) and 15 kilograms (33 lbs) in weight. Based on the spacing of the six preserved knobs in this specimen, the authors suggested that Velociraptor bore 14 secondaries (wing feathers stemming from the forearm), compared with 12 or more in Archaeopteryx, 18 in Microraptor, and 10 in Rahonavis. This type of variation in the number of wing feathers between closely related species, the authors asserted, is to be expected, given similar variation among modern birds.

Turner and colleagues interpreted the presence of feathers on Velociraptor as evidence against the idea that the larger, flightless maniraptorans lost their feathers secondarily due to larger body size. Furthermore, they noted that quill knobs are almost never found in flightless bird species today, and that their presence in Velociraptor (presumed to have been flightless due to its relatively large size and short forelimbs) is evidence that the ancestors of dromaeosaurids could fly, making Velociraptor and other large members of this family secondarily flightless, though it is possible the large wing feathers inferred in the ancestors of Velociraptor had a purpose other than flight. The feathers of the flightless Velociraptor may have been used for display, for covering their nests while brooding, or for added speed and thrust when running up inclined slopes.

Oviraptorosaurs
Evidence for feathered oviraptorosaurs exists in several forms. Most directly, two species of primitive oviraptorosaurs (Caudipteryx) have been found with impressions of well developed feathers, most notably on the wings and tail, suggesting that they functioned at least partially for display. Secondly, at least one oviraptorosaur (Nomingia) was preserved with a tail ending in something like a pygostyle, a bony structure at the end of the tail that, in modern birds, is used to support a fan of feathers. Similarly, quill knobs (anchor points for wing feathers on the ulna) have been reported in the oviraptorosaurian species, Avimimus portentosus. Additionally, a number of oviraptorid specimens have famously been discovered in a nesting position similar to that of modern birds. The arms of these specimens are positioned in such a way that they could perfectly cover their eggs if they had small wings and a substantial covering of feathers.

The bodies and limbs of oviraptorosaurs are arranged in a bird-like manner, suggesting the presence of feathers on the arms which may have been used for insulating eggs or brooding young. Members of Oviraptoridae possess a quadrate bone that shows particularly avian characteristics, including a pneumatizatized, double-headed structure, the presence of the pterygoid process, and articular fossa for the quadrratojugal.

Oviraptorids were probably feathered, since some close relatives were found with feathers preserved (Caudipteryx and possibly Protarchaeopteryx). Another finding pointing to this is the discovery in Nomingia of a pygostyle, a bone that results from the fusion of the last tail vertebrae and is responsible in birds to hold a fan of feathers in the tail. Finally, the arm position of the brooding Citipati would have been far more effective if feathers were present to cover the eggs.

Ornithomimosaurs
In 1997, filament-like integumentary structures were reported to be present in the Spanish ornithomimosaur Pelecanimimus polyodon. Furthermore, one published life restoration depicts Pelecanimimus as having been covered in the same sort of quill-like structures as are present on Sinosauropteryx and Dilong. However, a brief 1997 report that described soft-tissue mineralization in the Pelecanimimus holotype has been taken by most workers as the definitive last word 'demonstrating' that integumentary fibers were absent from this taxon.

However, the report describing soft-tissue mineralization described soft-tissue preservation seen in one small patch of tissue, and the absence of integument here does not provide much information about the distribution of integument on the live animal. This might explain why a few theropod workers (notably Paul Sereno and Kevin Padian) have continued to indicate the presence of filamentous integumentary structures in Pelecanimimus. Feduccia et al. (2005) argued that Pelecanimimus possessed scaly arms and figured some unusual rhomboidal structures in an effort to demonstrate this. The objects that they illustrate do not resemble scales and it remains to be seen whether they are anything to do with the integument of this dinosaur. A full description/monograph on this dinosaur has yet to be published, which might have more information on this subject.

Psittacosaurus
The integument, or body covering, of Psittacosaurus is known from a Chinese specimen, which most likely comes from the Yixian Formation of Liaoning. The specimen, which is not yet assigned to any particular species, was illegally exported from China, in violation of Chinese law, but was purchased by a German museum and arrangements are being made to return the specimen to China.

Most of the body was covered in scales. Larger scales were arranged in irregular patterns, with numerous smaller scales occupying the spaces between them, similarly to skin impressions known from other ceratopsians, such as Chasmosaurus. However, a series of what appear to be hollow, tubular bristles, approximately 16 centimeters (6.4 in) long, were also preserved, arranged in a row down the dorsal (upper) surface of the tail. However, according to some scientists "[a]t present, there is no convincing evidence which shows these structures to be homologous to the structurally different [ feathers and protofeathers] of theropod dinosaurs." As the structures are only found in a single row on the tail, it is unlikely that they were used for thermoregulation, but they may have been useful for communication through some sort of display.

''Tianyulong
Tianyulong is notable for the row of long, filamentous integumentary structures apparent on the back, tail and neck of the fossil. The similarity of these structures with those found on some derived theropods suggests their homology with feathers and raises the possibility that the earliest dinosaurs and their ancestors were covered with analogous dermal filamentous structures that can be considered as primitive feathers (proto-feathers). The filamentous integumentary structures are preserved on three areas of the fossil: in one patch just below the neck, another one on the back, and the largest one above the tail. The hollow filaments are parallel to each other and are singular with no evidence of branching. They also appear to be relatively rigid, making them more analogous to the integumentary structures found on the tail of Psittacosaurus than to the proto-feather structures found in avian and non-avian theropods. Among the theropods, the structures in Tianyulong are most similar to the singular unbranched proto-feathers of Sinosauropteryx and Beipiaosaurus. The estimated length of the integumentary structures on the tail is about 60 mm which is seven times the height of a caudal vertebra. Their length and hollow nature argue against of them being subdermal structures such as collagen fibers.

Such dermal structures have previously been reported only in derived theropods and ornithischians, and their discovery in Tianyulong extends the existence of such structures further down in the phylogenetic tree. However, the homology between the ornithischian filaments and the theropods proto-feathers is not obvious. If the homology is supported, the consequence is that the common ancestor of both saurischians and ornithischians were covered by feather-like structures and that groups for which skin impression are known such as the sauropods were only secondarily featherless. If the homology is not supported, it would indicate that these filamentous dermal structures evolved independently in saurischians and ornithischians, as well as in other archosaurs such as the pterosaurs. The authors (in supplementary information to their primary article) noted that discovery of similar filamentous structures in the theropod Beipiaosaurus bolstered the idea that the structures on Tianyulong are homologous with feathers. Both the filaments of Tianyulong and the filaments of Beipiaosaurus were laong, singular, and unbranched. In Beipiaosaurus, however, the filaments were flattened. In Tianyulong, the filaments were round in cross section, and therefore closer in structure to the earliest forms of feathers predicted by developmental models.

Some scientists have argued that other dinosaur proto-feathers are actually fibers of collagen that have come loose from the animals' skins. However, collagen fibers are solid structures; based on the long, hollow nature of the filaments on Tianyulong the authors rejected this explanation.

Detractors from the mainstream
A small minority, including ornithologists Alan Feduccia and Larry Martin, continues to assert that birds are instead the descendants of earlier archosaurs, such as Longisquama or Euparkeria. Embryological studies of bird developmental biology have raised questions about digit homology in bird and dinosaur forelimbs. However, due to the cogent evidence provided by comparative anatomy and phylogenetics, as well as the dramatic feathered dinosaur fossils from China, the idea that birds are derived dinosaurs, first championed by Huxley and later by Nopcsa and Ostrom, enjoys near-unanimous support among today's paleontologists.

BADD, BAND, and the Birds Came First hypothesis
The non-standard, non-mainstream Birds Came First (or BCF) hypothesis proposed by George Olshevsky suggests that while there is a close relationship between dinosaurs and birds, but argues that, merely given this relationship, it is just as likely that dinosaurs descended from birds as the other way around. The hypothesis does not propose that birds in the proper sense evolved earlier than did other dinosaurs or other archosaurs: rather, it posits that small, bird-like, arboreal archosaurs were the direct ancestors of all the archosaurs that came later on (proper birds included). George was aware of this fact, and apparently considered the rather tongue-in-cheek alternative acronym GOODD, meaning George Olshevsky On Dinosaur Descendants. This was, of course, meant as opposite to the also tongue-in-cheek BADD (Birds Are Dinosaur Descendants): the term George uses for the 'conventional' or 'mainstream' view of avian origins outlined in the first two paragraphs above. 'BADD' is bad, according to BCF, as it imagines that small size, feathers and arboreal habits all evolved very late in archosaur evolution, and exclusively within maniraptoran theropod dinosaurs.

Claimed temporal paradox
The temporal paradox, or time problem is a controversial issue in the evolutionary relationships of feathered dinosaurs and birds. It was originally conceived of by paleornithologist Alan Feduccia. The concept is based on the apparent following facts. The consensus view is that birds evolved from dinosaurs, but the most bird-like dinosaurs, and those most closely related to birds (the maniraptorans), are known mostly from the Cretaceous, by which time birds had already evolved and diversified. If bird-like dinosaurs are the ancestors of birds they should be older than birds, but Archaeopteryx is 155 million years old, while the very bird-like Deinonychus is 35 million years younger. This idea is sometimes summarized as "you can't be your own grandmother". The development of avian characteristics in dinosaurs supposedly should have led to the first modern bird appearing about 60 million years ago. However, Archaeopteryx lived 150 million years ago, long before any of the bird changes took place in dinosaurs. Each of the feathered dinosaur families developed avian-like features in its own way. Thus there were many several different lines of evolution. Archaeopteryx was merely the result of one such line.

Numerous researchers have discredited the idea of the temporal paradox. Witmer (2002) summarized this critical literature by pointing out that there are at least three lines of evidence that contradict it. First, no one has proposed that maniraptoran dinosaurs of the Cretaceous are the ancestors of birds. They have merely found that dinosaurs like dromaeosaurs, troodontids and oviraptorosaurs are close relatives of birds. The true ancestors are thought to be older than Archaeopteryx, perhaps Early Jurassic or even older. The scarcity of maniraptoran fossils from this time is not surprising, since fossilization is a rare event requiring special circumstances and fossils may never be found of animals in sediments from ages that they actually inhabited. Secondly, fragmentary remains of maniraptoran dinosaurs actually have been known from Jurassic deposits in China, North America, and Europe for many years. The femur of a tiny maniraptoran from the Late Jurassic of Colorado was reported by Padian and Jensen in 1989. Third, if the temporal paradox would indicate that birds should not have evolved from dinosaurs, then what animals are more likely ancestors considering their age? Brochu and Norell (2001) analyzed this question using several of the other archosaurs that have been proposed as bird ancestors, and found that all of them create temporal paradoxes&mdash;long stretches between the ancestor and Archaeopteryx where there are no intermediate fossils&mdash;that are actually worse. Thus, even if one used the logic of the temporal paradox, one should still prefer dinosaurs as the ancestors to birds.

Quick & Ruben (2009)
In Quick & Ruben's 2009 paper, they argue that modern birds are fundamentally different from non-avian dinosaurs in terms of abdominal soft-tissue morphology, and therefore birds cannot be modified dinosaurs. The paper asserts that a specialized 'femoral-thigh complex', combined with a synsacrum and ventrally separated pubic bones, provides crucial mechanical support for the abdominal wall in modern birds, and has thereby allowed the evolution of large abdominal air-sacs that function in respiration. In contrast, say the authors, theropod dinosaurs lack these features and had a highly mobile femur that cannot have been incorporated into abdominal support. Therefore, non-avian theropods cannot have had abdominal air-sacs that functioned like those of modern birds, and non-avian theropods were fundamentally different from modern birds. However, this was not mentioned in the paper, but was of course played-up in the press interviews. The papers never really demonstrate anything, but merely try to shoot holes in a given line of supporting evidence. It has been argued that respiratory turbinates supposedly falsify dinosaur endothermy, even though it has never been demonstrated that respiratory turbinates really are a requirement for any given physiological regime, and even though there are endotherms that lack respiratory turbinates. The innards of Sinosauropteryx and Scipionyx also supposedly falsify avian-like air-sac systems in non-avian coelurosaurs and demonstrate a crocodilian-like hepatic piston diaphragm, even though personal interpretation is required to accept that this claim might be correct. Even though crocodilians and dinosaurs are fundamentally different in pelvic anatomy, some living birds have the key soft-tissue traits reported by Ruben et al. in Sinosauropteryx and Scipionyx, and yet still have an avian respiratory system. For a more detailed rebuttal of Quick &Ruben's paper, see this post by Darren Naish at Tetrapod Zoology.

Fossil evidence
After a century of hypotheses without conclusive evidence, especially well-preserved (and legitimate) fossils of feathered dinosaurs were discovered during the 1990s, and more continue to be found. The fossils were preserved in a Lagerstätte — a sedimentary deposit exhibiting remarkable richness and completeness in its fossils — in Liaoning, China. The area had repeatedly been smothered in volcanic ash produced by eruptions in Inner Mongolia 124 million years ago, during the Early Cretaceous Period. The fine-grained ash preserved the living organisms that it buried in fine detail. The area was teeming with life, with millions of leaves, angiosperms (the oldest known), insects, fish, frogs, salamanders, mammals, turtles, lizards and crocodilians discovered to date.

The most important discoveries at Liaoning have been a host of feathered dinosaur fossils, with a steady stream of new finds filling in the picture of the dinosaur-bird connection and adding more to theories of the evolutionary development of feathers and flight. Norell et al (2007) reported quill knobs from an ulna of Velociraptor mongoliensis, and these are strongly correlated with large and well-developed secondary feathers. Behavioural evidence, in the form of an oviraptorosaur on its nest, showed another link with birds. Its forearms were folded, like those of a bird. Although no feathers were preserved, it is likely that these would have been present to insulate eggs and juveniles.

Genuine feathers?
There have been claims that the supposed feathers of the Chinese fossils were a preservation artifact. Despite doubts, the fossil feathers have roughly the same appearance as those of birds fossilized in the same locality, so there is no serious reason to think they are of different nature; moreover, no non-theropod fossil from the same site shows such an artifact, but sometimes show unambiguous hair (some mammals) or scales (some reptiles).

The Archaeoraptor fake
In 1999, a supposed 'missing link' fossil of an apparently feathered dinosaur named "Archaeoraptor liaoningensis", found in Liaoning Province, northeastern China, turned out to be a forgery. Comparing the photograph of the specimen with another find, Chinese paleontologist Xu Xing came to the conclusion that it was composed of two portions of different fossil animals. His claim made National Geographic review their research and they too came to the same conclusion. The bottom portion of the "Archaeoraptor" composite came from a legitimate feathered dromaeosaurid now known as Microraptor, and the upper portion from a previously-known primitive bird called Yanornis.

List of dinosaur genera preserved with evidence of feathers
A number of non-avian dinosaurs are now known to have been feathered. Direct evidence of feathers exists for the following genera, listed in the order currently accepted evidence was first published. In all examples, the evidence described consists of feather impressions, except those marked with an asterisk (*), which denotes genera known to have had feathers based on skeletal or chemical evidence, such as the presence of quill knobs.


 * 1) Avimimus* (1987)  p. 536
 * 2) Sinosauropteryx (1996)
 * 3) Protarchaeopteryx (1997)
 * 4) Caudipteryx (1998)
 * 5) Rahonavis* (1998)
 * 6) Shuvuuia (1999)
 * 7) Sinornithosaurus (1999)
 * 8) Beipiaosaurus (1999)
 * 9) Microraptor (2000)
 * 10) Nomingia* (2000)
 * 11) Cryptovolans (2002)
 * 12) Scansoriopteryx (2002)
 * 13) Epidendrosaurus (2002)
 * 14) Psittacosaurus? (2002)
 * 15) Yixianosaurus (2003)
 * 16) Dilong (2004)
 * 17) Pedopenna (2005)
 * 18) Jinfengopteryx (2005)
 * 19) Sinocalliopteryx (2007)
 * 20) Velociraptor* (2007)
 * 21) Epidexipteryx (2008)
 * 22) Anchiornis (2009)
 * 23) Tianyulong? (2009)


 * Note, filamentous structures in some ornithischian dinosaurs (Psittacosaurus, Tianyulong) and pterosaurs may or may not be homologous with the feathers and protofeathers of theropods.

Primitive feather types
The evolution of feather structures is thought to have proceeded from simple hollow filaments through several stages of increasing complexity, ending with the large, deeply rooted, feathers with strong pens (rachis), barbs and barbules that birds display today.

It is logical that the simplest structures were probably most useful as insulation, and that this implies homeothermy. Only the more complex feather structures would be likely candidates for aerodynamic uses.

It is not known with certainty at what point in archosaur phylogeny the earliest simple “protofeathers” arose, or if they arose once or, independently, multiple times. Filamentous structures are clearly present in Pterosaurs, and long, hollow quills have been reported in a specimen of Psittacosaurus from Liaoning. It is thus possible that the genes for building simple integumentary structures from beta keratin arose before the origin of dinosaurs, possibly in the last common ancestor with pterosaurs – the basal Ornithodire.

Most of the theropods known to have feathers are maniraptorans. Only a few non-maniraptoran theropods are known to have them as well. At present, the most primitive (known) theropod dinosaur with integumentary filaments is Sinosauropteryx, a compsognathid (Jurassic/Cretaceous, 150-120 mya), whose body was covered with feather-like structures that look like hollow tubes, or hairs. They may or may not have had barbs, like downy (plumaceous) feathers. Another very primitive theropod, Dilong paradoxus (Early Cretaceous), an ancestor of Tyrannosaurus rex, had similar simple feather structures. The alvarezsaurid Shuvuuia is sometimes found to be outside the maniraptora, but consensus right now places it as a maniraptoran. The first dinosaur fossils from the Yixian formation found to have true flight-structured feathers (pennaceous feathers) were Protarchaeopteryx and Caudipteryx (135-121 mya). Due to the size and proportions of these animals it is more likely that their feathers were used for display rather than for flight. Subsequent dinosaurs found with pennaceous feathers include Pedopenna and Jinfengopteryx. Several specimens of Microraptor, described by Xu et al. in 2003, show not only pennaceous feathers but also true asymmetrical flight feathers, present on the fore and hind limbs and tail. Asymmetrical feathers are considered important for flight in birds. Before the discovery of Microraptor gui, Archaeopteryx was the most primitive known animal with asymmetrical flight feathers.

Phylogeny and the inference of feathers in other dinosaurs
Feathered dinosaur fossil finds to date, together with cladistic analysis, suggest that many types of theropod may have had feathers, not just those that are especially similar to birds. In particular, the smaller theropod species may all have had feathers and possibly even the larger theropods (for instance T. rex) may have had feathers, in their early stages of development after hatching. Whereas these smaller animals may have benefited from the insulation of feathers, large adult theropods are unlikely to have had feathers, since inertial heat retention would likely be sufficient to manage heat. Excess internal heat may even have become a problem, had these very large creatures been feathered.

Fossil feather impressions are extremely rare; therefore only a few feathered dinosaurs have been identified so far. However, through a process called phylogenetic bracketing, scientists can infer the presence of feathers on poorly-preserved specimens. All fossil feather specimens have been found to show certain similarities. Due to these similarities and through developmental research almost all scientists agree that feathers could only have evolved once in dinosaurs. Feathers would then have been passed down to all later, more derived species (although it is possible that some lineages lost feathers secondarily). If a dinosaur falls at a point on an evolutionary tree within the known feather-bearing lineages, scientists assume it too had feathers, unless conflicting evidence is found. This technique can also be used to infer the type of feathers a species may have had, since the developmental history of feathers is now reasonably well-known. Nearly all paleontologists regard birds as coelurosaurian theropod dinosaurs. Within Coelurosauria, multiple cladistic analyses have found support for a clade named Maniraptora, consisting of therizinosauroids, oviraptorosaurs, troodontids, dromaeosaurids, and birds. Of these, dromaeosaurids and troodontids are usually united in the clade Deinonychosauria, which is a sister group to birds (together forming the node-clade Eumaniraptora) within the stem-clade Paraves.

Other studies have proposed alternative phylogenies in which certain groups of dinosaurs that are usually considered non-avian are suggested to have evolved from avian ancestors. For example, a 2002 analysis found oviraptorosaurs to be basal avians. Alvarezsaurids, known from Asia and the Americas, have been variously classified as basal maniraptorans,  paravians, the sister taxon of ornithomimosaurs, as well as specialized early birds. The genus Rahonavis, originally described as an early bird, has been identified as a non-avian dromaeosaurid in several studies. Dromaeosaurids and troodontids themselves have also been suggested to lie within Aves rather than just outside it. p. 472

The scientists who described the (apparently unfeathered) Juravenator performed a genealogical study of coelurosaurs, including distribution of various feather types. Based on the placement of feathered species in relation to those that have not been found with any type of skin impressions, they were able to infer the presence of feathers in certain dinosaur groups. The following simplified cladogram follows these results, and shows the likely distribution of plumaceous (downy) and pennaceous (vaned) feathers among theropods. Note that the authors inferred pennaceous feathers for Velociraptor based on phylogenetic bracketing, a prediction later confirmed by fossil evidence.