Recent paleontological and neontological studies on feathers and feather-like integumentary structures have improved greatly our understanding of the origin and early evolution of feathers. New observations on some non-avian dinosaur specimens preserving integumentary structures, in combination with recent paleontological and neontological data, provide additional insights into this important evolutionary issue.

Five major morphogenesis events are inferred to have occurred sequentially early in feather evolution before the origin of the Aves, and they are: 1) appearance of filamentous and tubular morphology, 2) formation of follicle and barb ridges, 3) appearance of rachis, 4) appearance of planar form, and 5) formation of pennaceous barbules.

These events produce several morphotypes of feathers that are common among non-avian archosaurs but are probably lost later in avian evolution, and they also produced several morphotypes of feathers that are nearly identical or identical to those of modern birds. While feathers of non-avian dinosaurs exhibit many unique features of modern feathers, some of them also possess striking features unknown in modern feathers. Several models of evolutionary origin of feathers based on developmental data suggest that the origin of feathers is a completely innovative event and the first feathers have nothing to do with reptilian scales. We believe, however, that the defining features of modern feathers might have evolved in an incremental manner rather than in a sudden way. Consequently, an evolutionary model characteristic of both transformation and innovation is more acceptable for feather evolution. The function of the first feather is inferred to be neither related to flight nor to insulation. Display or heat dissipation, among others, remains viable hypotheses for initial function of feathers. An integrative study is promising to provide much new insights into the origin of feathers.

Feathers are the most complicated integumentary derivative of birds, and also the most characteristic feature of them. A typical feather is composed of hierarchical branches of rachis, barbs and barbules on a tubular structure called calamus, and it is either radially symmetrical or essentially in a planar form. Some feathers lack a rachis, but most feathers have a shaft (calamus of basal tubular portion and rachis of the rest solid portion) with barbs forming vanes on either side which are secondarily branched to form barbules. This basic plan shows a great diversity, and results in a wide variety of forms, which differ in: implantation (in the skin or on the skeleton), thickness and stiffness of the rachis, relative sizes of rachis and barbs, type, spacing and placement of barbs, symmetry and curvature of vanes, presence and structure of an afterfeather, and melanin pigmentation (Stettenheim, 2000). Largely corresponding to the diverse morphology, feathers also serve a diverse of functions, including physical protection, thermal regulation, locomotion, display, tactile sensation, and water repellency, among others (Stettenheim, 2000).

Mainly due to the diverse morphologies and functions of modern feathers, there is little consensus on the possible earliest morphology and initial function of feathers. Early fossil feathers also fail to provide relevant information. Feather-like structure has been claimed to be present in the Early Jurassic theropod (Kundrat, 2004), but this interpretation receives little attention due to their structure is preserved only as imprint. The Late Jurassic Archaeopteryx specimens preserve fine feather impressions, but they are identical in morphology to modern feathers. Similarly, other known basal birds all have feathers of fully modern form (Zhang and Zhou, 2006; Zhou and Zhang, 2006). In general, the known feathers of basal avians provide little significant information concerning the origin of feathers. Though the highly specialized tail feathers with an undifferentiated vane region on either side of the central rachis in many basal birds (Zhang and Zhou, 2000) and even in a non-avian theropod (Zhang et al., 2008b) have been suggested to represent a type of primitive feathers, their implications for understanding the origin of feathers have been questioned mainly due to their relatively late appearance in feather evolution (Prum and Brush, 2002; Xu, 2002).

Most non-avian diapsids have scaled skin. However, Longisquama insignis, a small diapsid reptile with uncertain systematic position, has recently been suggested to have non-avian feathers (Jones et al., 2000). Its highly specialized elongated dorsal scales have recently described as feather-like in many details and have been regarded as non-avian feathers (Jones et al., 2000). However, this interpretation has been criticized from the perspective of both preservation and morphology (Reisz and Sues, 2000; Prum, 2001; Unwin and Benton, 2001).

Understanding the evolution of structures must lie in a phylogenetic framework. Over the last few decades, a wealth of fossil evidence has been found to support the theropod hypothesis of bird origins, and numerous systematic analyses also strongly corroborate this hypothesis, suggesting that birds are nested deeply within a group of theropod dinosaurs called Coelurosauria (Gauthier, 1986; Sereno, 1999). It is therefore expected that simpler feather-like integumentary structures should be present in the closest evolutionary relatives of birds among the coelurosaurian theropods.

Over the last decade, numerous dinosaur specimens preserving soft tissue have been recovered from the Early Cretaceous Jehol Group of northern China, the lacustrine beds of uncertain Jurassic-Cretaceous age in western Liaoning, and the Jurassic Daohugou Formation of eastern Nei Mongol (Xu and Zhang, 2005; Xu and Norell, 2006; Zhang et al., 2008b; Xu et al., 2009a). In general, these findings suggest that: 1) early feathers are simple filamentous structures and have appeared at least in basal coelurosaurian theropods; 2) more complex pennaceous feathers evolved early in maniraptoran theropods and some non-avian theropods even had flight feathers with asymmetrical vanes; and 3) the original function of feathers has nothing to do with flight. The discoveries of these specimens have significantly advanced our understanding of the origin and early evolution of feathers, though admittedly still much information is needed to understand some critical stages for feather evolution. In the present paper, we will review recent paleontological and neontological data relevant to the origin and early evolution of feathers, comment on their implications, and propose an evolutionary scenario to describe the origin and early evolution of feathers.

This original paper is available at http://www.ivpp.cas.cn/cbw/gjzdwxb/xbwzxz/200911/P020091104362654347399.pdf