The Significance of the Feathered Chinese Dinosaurs

The 1990s will go down in history as a time when one of the most significant fossil deposits ever discovered was brought to light. The Lower Yixian Formation (Chaomidianzi Formation of some) and allied rock units in western Liaoning Province, People's Republic of China, have yielded a wealth of fossil vertebrates, preserving—in often astounding abundance—an entire fauna in all its diversity (Luo, 1999;

Swisher et al., 1999). The basal birds from these deposits are discussed in this volume by Zhou and Hou (Chapter 7). With regard to the origin of birds, several additional taxa are relevant, particularly because of the preservation of integumentary remains interpreted to be feathers or featherlike filaments. At this writing, six theropod taxa (other than the indisputable birds) have been reported to have "feathery" skin (with, no doubt, more taxa on the way): Sinosaurop-teryx prima (Ji and Ji, 1996; Chen et al., 1998), Protar-chaeopteryx robusta (Ji and Ji, 1997; Ji et al., 1998), Caudipteryx spp. (Ji et al., 1998; Zhou and Wang, 2000; Zhou et al., 2000), Beipiaosaurus inexpectus (Xu et al., 1999a), Sinornithosaurus millenii (Xu et al., 1999b; Ji et al., 2001, if the juvenile dromaeosaurid pertains to this species), and Microraptor zhaoianus (Xu et al., 2000). The obvious significance for the debate on avian origins is that if feathers are truly present in nonavian theropod dinosaurs, then this should effectively close the door to any opposition to the theropod hypothesis. The debate, for all intents and purposes, will be over.

Thus, it is necessary to assess these claims carefully. See also the chapters in this volume by Clark, Norell, and Makovicky (Chapter 2) and Zhou and Hou (Chapter 7) for their assessments. The controversy began when Ji and Ji (1996: translation courtesy of Chen P.-J. and P. J. Currie) identified feathers in Sinosauropteryx and argued that (1) they were similar to modern down in lacking rachis and barbs, and (2) they were restricted to a median frill running from the head to the tip of the tail dorsally and onto the ven-tromedian surface of the tail. For Ji and Ji (1996), the presence of feathers required the referral of Sinosauropteryx to Aves. In the subsequent furor, there was a retreat from their interpretation as true feathers, being instead "proto-feathers" (e.g., Brush et al., 1997). Moreover, the status of Sinosauropteryx as a bird was questioned, as it clearly had the skeletal anatomy of a small theropod dinosaur. Indeed, Chen et al. (1998), based on additional specimens, formally referred Sinosauropteryx to Compsognathidae, a clade of relatively basal coelurosaurs. These authors also presented the first in-depth morphological analysis of the integumentary structures, describing them as coarse, probably hollow, filaments up to 40 mm in length; a chemical or elemental analysis has not been published. The ultimate question, of course, is, What makes these feathers or even "proto-feathers" (Unwin, 1998)?

Indeed, Geist et al. (1997) and Feduccia (1999b) suggested that the structures in Sinosauropteryx were in fact not feathers at all and, moreover, that they were not external, epidermal appendages of any kind. Rather, they argued that, based on comparative anatomy, the fossil structures more closely resembled collagenous fibers supporting a midsagit-tal dermal frill, that is, internal structures that became frayed in the process of decomposition. Although the ap parent midline distribution of the filaments is indeed fully consistent with dermal frills, which are widely present in modern squamates and even well known in some dinosaur groups (e.g., sauropods: Czerkas, 1994; hadrosaurids: Lull and Wright, 1942), the filaments are not actually in the median plane in all regions but rather are in some places offset, such as the head region, which is not preserved in a straight lateral view (Padian et al., 2001). In fact, a routine finding with the Liaoning birds and dinosaurs is that the feathers or filaments are preserved as a halo around the skeletal remains. Thus, the "midline frill" is perhaps more safely interpreted as an artifact resulting from the animals being preserved lying more or less on their sides, such that the halo would roughly correspond to the median plane. Geist et al. (1997) suggested that another problem with the feather interpretation in Sinosauropteryx is that although some specimens may show a ruffle of fibers extending along the tail, another specimen shows a smooth outline along the tail.

It is valid to question whether these shortcomings falsify the feather hypothesis or may simply be ascribed to vagaries of preservation. Nevertheless, the inference of feathers in Sinosauropteryx has such profound implications— not only for the origin of birds but also for the origin of feathers and endothermy—that we should be compelled by the weight of evidence before accepting such momentous claims. In the acknowledged absence of calamus, rachis, and barbs, the identification of these structures as "true" feathers in Sinosauropteryx is clearly unjustified. Also problematic is the inference of "protofeathers." Although true feathers certainly had epidermal precursors that lacked such definitive attributes as rachis and barbs, how would we recognize them? Chemical analysis showing unique feather proteins might provide valid evidence, but again such studies have not been performed. Significantly, according to Prum's (1999, 2000) developmental model of feather evolution, the filaments of Sinosauropteryx are entirely consistent with an early stage of feather evolution. Moreover, Padian et al. (2001) argued that these filaments have enough morphological attributes in common with feathers that it is fair to accept that the filaments pass the similarity test of homology with avian feathers. Finally, the notion of feather precursors in Sinosauropteryx is significantly enhanced by the feathered theropods from the Yix-ian discussed later, leading Padian (1998:729) to state that "doubts [raised by Geist et al. (1997)] can now be put to rest." Thus, in effect, the filaments of Sinosauropteryx might be regarded as passing the congruence test of homology, as well (Padian et al., 2001). Nevertheless, the evidence in Sinosauropteryx obviously should be judged on its own merits, and stemward inferences based on crownward observations require considerable justification (i.e., level II inference; Witmer 1995a).

Shortly after the announcement of Sinosauropteryx, Ji and Ji (1997) announced the discovery of another feathered creature, Protarchaeopteryx robusta. They regarded it as the sister group of Archaeopteryx, even placing it within Ar-chaeopterygidae, but Ji et al. (1998) removed it from a position within Aves. Unlike the case of Sinosauropteryx, the feathers attributed to Protarchaeopteryx are absolutely indisputable, with clear rachis and barbs. Thus, if a phyloge-netic placement outside birds is justified, then a feathered nonavian theropod would be at hand. Unfortunately, the unique specimen of Protarchaeopteryx (NGMC 2125) is quite poorly preserved, and many attributes either are open to interpretation or beyond reliable observation (e.g., about 50% missing data according to Ji et al., 1998). I was unable to confirm the two plesiomorphies identified by Ji et al. (1998) that would deny Protarchaeopteryx a higher position—a short frontal process of the premaxilla and serrated teeth. The premaxilla is badly damaged, and the teeth seemed to lack clear serrations; Ji et al. (1998) regarded the serrations as so small (7-10/mm) that they were not visible even with my hand lens, but one then wonders if something so small can be regarded as truly a "serration." It may lack a reversed hallux, which would be an important plesiomor-phy, but neither foot of the holotype is well preserved. Given the current state of our knowledge of Protarchaeopteryx, it is difficult to predict whether better specimens will show it to be outside or within Aves. It is even conceivable that a sister group relationship with Archaeopteryx, as originally suggested by Ji and Ji (1997), will be borne out (see Elzanowski, Chapter 6 in this volume). Certainly, Protarchaeopteryx is very close to the transition to birds, which makes its state of preservation all the more frustrating.

Much better preserved, however, is the material of Caudipteryx (Ji et al., 1998; Zhou and Wang, 2000; Zhou et al., 2000). Caudipteryx in many ways seems to be the perfect "feathered dinosaur." It possesses clearly "avian" feathers (i.e., with calamus, rachis, and barbs), yet, unlike those of Archaeopteryx, these feathers are not part of a flight apparatus, and hence Caudipteryx obviously did not fly. Moreover, Caudipteryx lacks many of the derived bony features unique to "proper" birds and hence has justifiably been hailed as the first animal to be discovered that is both indisputably feathered and indisputably not a bird. Indeed, Caudipteryx truly begs the question of just what may be called a "bird" in the colloquial sense of the word.

As mentioned, feathers are known for the two widely studied specimens described by Ji et al. (1998); the several new specimens reported by Zhou and Wang (2000) and Zhou et al. (2000) confirm a consistent pattern. In their preserved state, well-developed feathers are largely restricted to the manus and distal portion of the tail. As far as can be discerned, the inner and outer vanes are symmetrical about the rachis. Ji et al. (1998) and Zhou and Wang (2000) reported preservation of filamentous structures in the body regions, but the real question is whether the distribution of true feathers was more extensive in life or actually restricted to the tips of the hands and tail.

The association of true feathers with the skeletons of Caudipteryx is beyond any doubt, which is important because the skeleton is decidedly nonavian—that is, this is no chimeric association. The following discussion is not intended to be a description of the bony anatomy of Caudipteryx but rather a tabulation of its primitive, non-avian attributes (see also Zhou and Wang, 2000). Although it is more customary in this cladistic age to enumerate derived characters, documentation of the primitive characters of this feathered creature is necessary to counter claims that Caudipteryx is in fact "a secondarily flightless bird, a Meso-zoic kiwi" (Feduccia, 1999a:4742; 1999b; see also Jones et al., 2000b). In addition to feathers, another significant avian apomorphy would be the shortened tail. In Caudipteryx there are only 22 caudal vertebrae, the same number as in Archaeopteryx and fewer than in any other known nonavian theropod (Ji et al., 1998). Moreover, the distal portion is clearly very stiff, although, as correctly noted by Ji et al. (1998; see also Zhou et al., 2000), definitely not fused into a pygostyle (or a "protopygostyle," as Feduccia [1999a] called it). Other than its short length and distal stiffening, nothing about the tail is particularly birdlike. Its distal caudal vertebrae have very short centra (Ji et al., 1998), rather than the elongate distal centra observed in Archaeopteryx and Ra-honavis (Forster et al., 1998). Moreover, the proximal caudal haemal arches (chevrons) are very long and spatulate, again unlike those of basal birds and unlike those of even most derived nonavian coelurosaurs. The closest match to the tail of Caudipteryx may well be among oviraptorosaurs. As particularly well demonstrated by Nomingia (Barsbold et al., 2000a,b), oviraptorosaurs display the following derived characters: a reduced number of caudal vertebrae (24 in Nomingia—only 2 more than in Caudipteryx), rigid distal tail with short centra, relatively long transverse processes on the proximal caudals (Sereno, 1999a), and elongate and spatulate haemal arches. Although Barsbold et al. (2000a,b) regarded the tail of Nomingia as bearing a "pygostyle," it is certainly not homologous (or even that similar) to the avian structure, and I would tend to reserve that name for pygostylian birds (see Chiappe, Chapter 20 in this volume). In any event, the shortened tail of Caudipteryx is not particularly birdlike and is basically matched by the tails of oviraptorosaurs.

Zhou et al. (2000) advanced a few additional birdlike characters that, although they still regarded Caudipteryx as a nonavian dinosaur, "indicate that its phylogenetic position remains a debatable issue." Not having examined their new specimens firsthand, I cannot comment in detail on the birdlike attributes, but a few points are pertinent. Of the birdlike characters that they advance, some are clearly homoplasies (e.g., manual phalangeal formula of 2-3-2), some are more widely distributed in maniraptorans (e.g., tooth form, uncinate processes), and some are open to interpretation (e.g., the "partially reversed" hallux). They also pointed to the remarkably short trunk (only nine thoracic vertebrae) and elongate hindlimbs, birdlike features that had earlier attracted the attention of Jones et al. (2000b).

Jones et al. (2000b) argued that Caudipteryx possessed a strikingly birdlike attribute relating to the location of the center of mass and the proportions of the trunk and hindlimb; these parameters were entirely unlike those of any known nonavian theropods but indistinguishable from those of cursorial birds. They provided three alternatives to explain these data. First, perhaps simply Caudipteryx apo-morphically and convergently developed a locomotor style similar to that of cursorial birds. Second, perhaps Caudipteryx was a nonavian theropod that had flight in its ancestry. And third, perhaps Caudipteryx was in fact "a secondarily, flightless, post-Archaeopteryx, cursorial bird" (Jones et al., 2000b). The authors clearly favor this third hypothesis. Testing all three hypotheses is firmly within the realm of phylogenetic analysis, and the paper of Jones et al. (2000b) was a functional analysis, not a comprehensive phylogenetic study.

Nevertheless, despite the presence of true feathers, birdlike hindlimb proportions, and perhaps other, less certain features, Caudipteryx displays a variety of plesiomorphic characters throughout the skeleton that, when taken together, clearly place it outside Aves. Ji et al. (1998) listed three such characters. Their first two characters are very similar and relate to the quadratojugal and its contact with the quadrate and squamosal. I concur that the quadratojugal of NGMC 97-9-A bears the primitive character of a relatively long dorsal (squamosal) process that probably is sutured to the quadrate, and the new specimens reported by Zhou et al. (2000) confirm this arrangement. Caudipteryx clearly lacks the small quadratojugal of birds, including such basal birds as Archaeopteryx, Confuciusornis, and enantiornithines.

The other plesiomorphic trait cited by Ji et al. (1998) involves the retention of a prominent, triangular obturator process of the ischium. Again, I fully agree, and I regard the shape of the ischium as one of the clearest manifestations of the position of Caudipteryx outside Aves. Basal birds have complex ischia (Forster et al., 1998) that generally are characterized by a small (or even absent) obturator process and instead a large, tablike (i.e., rectangular) proximodorsal process extending up toward the ilium. This is the condition in, for example, Archaeopteryx, Confuciusornis, and enantior-nithines. The very birdlike theropod Unenlagia comahuensis (Novas and Puerta, 1997) presents the intermediate condition of possessing both a large obturator process and a proxi-modorsal process. The shape and orientation of the ischium of Caudipteryx are clearly visible in the new material described by Zhou and Wang (2000; see also Zhou et al., 2000). Both ischia are well preserved and show only a single process that is large and triangular. This shape is exactly like that of the obturator process of, say, dromaeosaurids and ovirap-torosaurs. Another primitive character thus would be the absence of the proximodorsal process.

A number of other primitive characters of Caudipteryx can be added to those discussed by Ji et al. (1998; see also Zhou and Wang, 2000, and Zhou et al., 2000). For example, the jugal is a typically nonavian theropodan jugal with a very large postorbital process. Birds, on the other hand, have lost the postorbital process of the jugal or, at most, have reduced it to a small process. Even taxa that retain a postorbital bone and a dorsotemporal arch (e.g., Archaeopteryx, the Catalan enantiornithine nestling, alvarezsaurids) lack a large postorbital process of the jugal and basically have a jugal bar. The postorbital bone of the enantiornithine Proto-pteryx has a long jugal process that might reach the jugal, but such a contact is not clear on the specimens (Zhang and Zhou, 2000). The only certain exception is Confuciusornis, which curiously possesses a complete postorbital bar formed by contact of the postorbital and jugal bones. But even in Confuciusornis most specimens have a relatively small postorbital process of the jugal (in some cases, little more than a bump), and the postorbital bone makes up almost all of the bar (Martin et al., 1998; Peters and Ji, 1998; Chiappe et al., 1999; Hou et al., 1999; Zhou and Hou, Chapter 7 in this volume). It may be noted here that the Eichstätt specimen of Archaeopteryx displays a somewhat bifid caudal extremity to the jugal. The dorsal prong of this bone could be interpreted as a postorbital process (e.g., Paul, 1988), but it seems to be situated too far caudally to reach the ventral ramus of the postorbital as preserved in the Berlin specimen (see also Chiappe et al., 1999); hence, I tend to agree more with the restoration of Archaeopteryx produced by Chatterjee (1991).

Another primitive character of Caudipteryx is the relatively very deep mandibular fenestra, as evidenced by the deep caudal embayment of the dentary of the paratype skull. Absence of a mandibular fenestra had been thought to characterize Aves because such an opening is absent in Archaeopteryx, hesperornithids, Ichthyornis, and neornithines, but the discovery of mandibular fenestrae in Confuciusornis (Martin et al., 1998; Chiappe et al., 1999; Zhou and Hou, Chapter 7 in this volume) makes this assessment a bit problematic. Nevertheless, the fenestra in Confuciusornis is not nearly as deep as in Caudipteryx and has an unusual form and thus may well be a reversal. The mandibular fenestra of Caudipteryx, on the other hand, is very comparable to that of dromaeosaurids, oviraptorosaurs, and other nonavian coelurosaurs and thus represents the primitive condition.

The thoracic girdle of Caudipteryx is also quite primitive and has none of the avian apomorphies seen in members of Aves. For example, the scapula has a relatively broad blade with a pronounced distal expansion, indicating the retention of a broad suprascapular cartilage. The blade clearly is not the slender and elongate structure seen in all basal birds. The shape of the coracoid is more or less that of a conventional nonavian coelurosaur coracoid, with a quadrilateral shape, proximal supracoracoidal nerve foramen, and moderate biceps tubercle. The coracoid certainly is not the elongate "straplike" bone seen in ornithothoracine birds. Another primitive trait here relates to the orientation of the girdle in that it is located on the lateral aspect of the thorax with the scapula at an angle to the axial column rather than on the dorsal aspect of the thorax with the scapula parallel to the column. The former condition is the primitive condition, whereas the latter condition is observed in all birds (Jenkins, 1993), including Archaeopteryx, Confuciusornis, and other basal birds. One hesitates to make too much of the orientation of elements in two-dimensional specimens, but, taken at face value (and all specimens agree on this point), Caudipteryx again displays the primitive condition.

The pelvic girdle of Caudipteryx presents primitive, non-avian characters beyond the ischiadic shape noted earlier. For example, the ilium is relatively very tall directly above the acetabulum, and its preacetabular portion is not expanded cranially; this is the typical condition for most non-avian coelurosaurs. In birds, on the other hand, the ilium is relatively low, with a greatly elongate preacetabular portion (see Elzanowski, Chapter 6 in this volume; Zhou and Hou, Chapter 7 in this volume). The pubic apron is extensive in Caudipteryx, measuring about 56% of total pubic length in the holotype. This is considerably more than the 45% measured in the London Archaeopteryx, the bird with the longest known pubic apron, and may even exceed that of some dromaeosaurids (Norell and Makovicky, 1997, 1999). Finally, Zhou and Wang (2000) and Zhou et al. (2000) noted that the pubis is not retroverted (as argued by Feduc-cia, 1999b) but rather is directed cranially, as in most non-avian theropods.

The picture that emerges from this brief survey of Caudipteryx is of a feathered theropod dinosaur that is probably well outside the avian lineage. I have not performed a more extensive formal analysis, but it seems readily apparent that it would be much less parsimonious to include Caudipteryx within Aves. And this point leads to the question of the phylogenetic position of Caudipteryx. The analysis of Ji et al. (1998) was not very inclusive, using only Velociraptor as an outgroup to the Chinese taxa and birds. My initial study of the specimens suggested a number of derived features pointing to oviraptorosaur relationships for Caudipteryx, including the following. The jaws are almost completely edentulous in Caudipteryx, which is indeed a re semblance to oviraptorosaurs, but even more striking is the conformation of the jaws. In both groups, the dentary is very deep between the mandibular fenestra and the sym-physeal portion, which is deflected ventrally; moreover, the symphyseal portion is medially inflected, and the caudal processes of the dentary diverge widely around the mandibular fenestra, both of which are attributes of ovi-raptorosaurs (Makovicky and Sues, 1998). The premaxilla has an extensive prenarial portion (which is also an avian apomorphy), and the naris itself is retracted (extensively in oviraptorosaurs). Finally, the maxilla is very characteristic, being a relatively small, rostrally displaced triangular element. Many of the postcranial elements compare well with oviraptorosaurs but also with other clades of coelurosaurs. However, the tail of Caudipteryx, as detailed previously, is quite similar to that of oviraptorosaurs in that both are short and proximally very thick. Given that my observations were not part of a comprehensive phylogenetic analysis, it was gratifying to see these impressions of an ovirap-torosaurian Caudipteryx borne out by numerous cladistic analyses presented at the Ostrom Symposium at Yale University in 1999 (e.g.,by P. C. Sereno,T. R. Holtz, M. A. Norell, and P. J. Currie), suggesting broad independent discovery of these relationships (and a heartening affirmation of phylogenetic systematics). Barsbold et al. (2000a,b) also regarded Caudipteryx as a basal oviraptorosaur. More significant, Caudipteryx was included in the very extensive phylogenetic analysis of Sereno (1999a). Sereno scored Caudipteryx for 204 characters (only 16% missing data) and found not only that Caudipteryx is well outside Aves but also that it is indeed a basal oviraptorosaur, sharing a dozen characters with oviraptoroids.

Thus, feathers of essentially modern structure do indeed predate the group conventionally known as "birds." This finding may seem shocking, but it is to be expected. This surprise again may relate to the pervasive sense of Archaeopteryx as truly the Urvogel, or "first bird." Common sense, of course, dictates that the elaborate feathers of Archaeopteryx, arranged as they are in their "modern" array of primaries and secondaries, must have had predecessors. However, Caudipteryx will likely remain difficult for some to accept, perhaps because it is such a dramatic repudiation of opposition to the theropod origin of birds.

In fact, more "feathered dinosaurs" are likely to come to light. For example, Xu et al. (1999a) described Beipiao-saurus, a new therizinosauroid theropod from the Lower Yixian Formation that bears filamentous dermal structures that are perhaps similar to those of Sinosauropteryx. These structures lack the unambiguous feather structure seen in Caudipteryx (i.e., they lack calamus, rachis, and barbs), but they are clearly present on areas of the body that cannot be explained away as remnants of a median frill. In Bei-piaosaurus, filamentous structures are associated with ele ments of both fore- and hindlimbs. The best-preserved filaments are attached to the ulna, where some approach 70 mm in length. Xu et al. (1999a) describe some filaments as distally branched and with hollow cores. These filaments have the same "protofeather" problems as did those of Sinosauropteryx (i.e., Are filaments truly the evolutionary precursors of feathers?), but their association with the limbs and their considerable length clearly indicate that they are some kind of epidermal appendage rather than an artifact of desiccating dermal collagen (see also Prum, 1999).

Another nonavian theropod with preserved integumentary filaments is the Yixian dromaeosaurid Sinornithosaurus millenii (Xu et al., 1999b). Unfortunately, the filaments are not in their natural positions, and thus, for example, the cluster of filaments adjacent to the skull cannot be reliably attributed to the head region. A very significant finding of Sinornithosaurus is a negative one, and that is the absence of hand and tail feathers. No true feathers (i.e., with rachis and barbs) of the sort seen in birds and Caudipteryx have been recovered with Sinornithosaurus. This is a bit troubling because the phylogenetic hypothesis of Sereno (1999a) predicts that, minimally, hand and tail feathers should be found in dromaeosaurids. However, given that the integumentary structures are not in life position and that the Sinor-nithosaurus specimen is generally jumbled somewhat on the slab, it is probably best not to make too much of this absence and assume that it is preservational. Close examination reveals some details suggesting that the filaments of Sinor-nithosaurus are more structured than those of Sinosaurop-teryx and Beipiaosaurus and hence more similar to avian feathers. Xu et al. (2001b) documented branching of some filaments, the compound construction of filamentous bundles, and even the occurrence of basal tufts of filaments, all features indicative of a more structurally complex integumentary covering. This report was followed shortly by the announcement by Ji et al. (2001; see also Norell, 2001) of a new specimen of a juvenile dromaeosaurid that is very similar to Sinornithosaurus and may even be the same species. The specimen preserves the integument in place and affirms the complex nature of the integument in dromaeosaurids. Not only does the juvenile specimen show branching and tufted filaments, but it also shows fibers branching off of a central axial filament—that is, it shows structure that could be interpreted as being the rachis and barbs of a "true" feather. Moreover, Ji et al. (2001) argued that the structures were so well ordered that birdlike barbules almost certainly had to have been present. As in Caudipteryx, the tail and fore-limbs have the best-organized integumentary structures.

Microraptor, the tiny dromaeosaurid reported by Xu et al. (2000), lacks hand and tail feathers but has the now typical filamentous coat. As in the juvenile dromaeosaurid, some integumentary impressions bear a rachislike structure, suggesting that true feathers might have been present in this animal, although this finding awaits confirmation with better-preserved material.

It is also relevant at this point to mention the findings of Schweitzer et al. (1999) on the biochemistry and morphology of fibrous integumentary structures recovered from the head region of the alvarezsaurid Shuvuuia from Mongolia (see also Chiappe, Norell, and Clark, Chapter 4 in this volume). Schweitzer et al. (1999) reported two important observations about these structures. First, biochemical studies are consistent with their being composed of beta keratin, a protein found in the feathers and scales of sauropsids. Second, the structures were apparently hollow. At present, the only structures known to be both hollow and composed of beta keratin are avian feathers. These findings are more provocative than conclusive, and, given the controversial phylogenetic position of alvarezsaurids (see Novas and Pol, Chapter 5 in this volume), one should be hesitant to make too much of these findings. Nevertheless, they may be legitimate evidence for feather or featherlike structures outside Aves.

In sum, the significance of these Chinese (and Mongolian) fossils for the debate on avian origins, in one sense, should be minimal. That is, we should not be surprised at the identification of feathers in a group of animals that a broad consensus had always thought was close to avian ancestry. The discovery of feathers in, say, Caudipteryx simply adds one more apomorphy to the long list of derived characters linking birds with theropod dinosaurs. The disproof of feathers in any of these Chinese forms would simply remove one character; all the others would remain. Likewise, forcing Caudipteryx to be within Aves because of its possession of true feathers (Cai and Zhao, 1999) would not automatically strip it of its clear theropod heritage. Thus, in this context, feathered dinosaurs are not that important. But, of course, in this high-profile, high-energy debate, rhetoric, regrettably, sometimes seems paramount to evidence. Feathers—that quintessentially avian trait—have always been the great definer of birds. 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.

0 0

Post a comment