EvoWiki is now a project of the RationalMedia Foundation.
We are moving all content to RationalWiki.
See the EvoWiki project page for details!

Evolution of feathers

From EvoWiki

Jump to: navigation, search

The Quintessence of Birdness--The Origin, Evolution, and Debate, on Feathers



As long as scientists have studied birds, they have puzzled over and marvelled at that most intricate and perplexing of avian traits--the feather. It is an amazing feat of biological engineering, functional and aesthetic (though birds care little of this, one might argue). Long has the feather delighted creationists and the proponents of so-called "intelligent design" with its fantastically complex structure. AIG has labelled the feather as quite simply something which cannot have evolved. Ornithologists, too, have been guilty of imputing to the feather, by virtue of its amazing properties, almost divine origin (e.g., Feduccia 1996). And yet the question has been and still is, "How did feathers originate and how have they evolved since then?"

We know a great deal about feathers, their function, the molecular and developmental pathways by which they are derived, their gross morphology, and so on. The great mystery is how they came about. It is a debate which, arguably, is at the very heart of avian phylogenetics, whether this is warranted, or not. Ostrom (1976) argued that if you solved the ancestry of Archaeopteryx you had solved the ancestry of birds. Many, whether of ornithological or paleontological inclination, would argue that once you have solved the origin of feathers (which includes isolating those taxa that display their most basal precursors), the debate on avian origins is effectively over.

While such a precise degree of resolution is perhaps not possible, the past quarter century has seen vast advances in this field of ornithological research, and today we stand on the brink of putting to rest these age-old questions.

Seeking Resolution Within the Context of the Intuitively Facile

Historically, two major schools of thought have dominated research into feather evolution. One has attempted to use ancestral feather function to generate hypotheses about early feather morphology. According to the other, ancestral morphologies of more derived feathers are hypothesized according to integument found in modern bird lineages; that is, the integument found in more basal lineages were assumed to be themselves more basal (Dyck 1985). In both, there has been a great deal of "common sense" involved in such explanations (Regal 1985), perhaps to their detriment.

The latter is seriously deficient, and is easily dispelled. The obvious problem is that the earliest known (chronologically speaking) feathers are already modern (Prum & Brush 2002), and the "basal" feathers of derived birds display derived characters that rule them out as ancestral; for example, the differentiated barbs of down feathers are clearly derived from pennacous feathers (Prum 1999, Dyck 1985).

If these feathers truly represented a basal morphology, then given that modern birds share common ancestry with Archaeopteryx, we would have to assume that the asymmetrical remiges in both Neornithes and Archaeopteryx are convergent, a possibility so remote as to be effectively rejected. Rather, the feathers of the most basal extant neornithine lineages should be seen as "secondarily simplified" (Prum & Brush 2002). Complicating the matter are potentially confused neornithine phylogenies. Proponents of “ratite” (e.g. Cracraft & Clarke 2002) holophyly and the concomitant assumption that ratites are primitive neornithes, have argued that their feathers are the most primitive (Lowe 1935). But the holophyly of "Ratitae" has been called into question, and perhaps is not a basal stem. The same is much clearer for various other Neornithes that were under consideration, including penguins.

While a full review of all the various early theories of feather origin that subsume the former methodology is beyond the scope of a primer such as this, which seeks to discuss the principal theories for feather origin as the debate now stands, a brief catalogue of prior models is nonetheless in order. Historically, functional hypotheses have essentially been polarized into two prevailing schools of thought. Feduccia (1985, 1996) has incorrectly expressed this dichotomy as one between adherents of the arboreal origin of flight and terrestrial origin of flight, but it is more accruately considered in terms of adaptation and exaptation. Those who hold to the adaptive school (e.g., Savile 1962, Bock 1965, Parkes 1966, Feduccia 1980, 1996) have argued that feathers arose in an aerodynamic context, either before or during the advent of homeothermy. In contrast to this view is that first elucidated by Ostrom in a series of publications in the 1970s in which it was postulated that feathers originally appeared to insulate endothermic proavians and were only later exaptated for flight. This idea found popular support in the work of Robert Bakker (Bakker 1975, 1986), and has since become the leading explanation for the original context in which feathers or feather precursors appeared, whether this is merited or not.

The principal argument of the so called adaptive school is the known relationship between even the smallest increase in integumentary surface area and the ability to sustain a glide or parachuting effect in an arboreal form. While this much is undeniably correct, perhaps the most glaring flaw in this hypothesis is the reliance it places upon a classical view of feather evolution in which feathers can be derived via simple (relatively speaking) distal ramification and segmentation of elongate scales. Yet the combined molecular and embroygenic data suggest that feathers are much more than "scales gone frayed", which has been assumed implicitly or explicitly by most researchers until recently. Feathers are composed of a sub-class of beta-keratin called feather (phi) keratins. Beta-keratins as a whole are remarkably distinct from alpha-keratins, which are present in the epidermis of every amniote, being much smaller and occuring as tandem repeats throughout the genome. Within beta-keratins, different classes of phi-keratin are themselves rather distinct. The sub-class that forms avian scutate scales, claws and beaks are larger than those present in feathers. The only keratins that are similar to those of other reptiles are the nonfeather type beta-keratins that make up the avian reticulate scales.

Morphogenesis is also wildly different. Feathers and scutate scales begin development as a visibly distinct placode. However, this is not present in reticulate scales and all reptilian scales, whose placode remains indistinguishable from the epidermis (Maderson & Alibardi 2000). Subsequent stages of development are unknown outside of feathers, from the branching of barbs, the tubular invagination that forms of the follicle, to the barb ridges. Even the planar surface, so loved by functionalist advocates of an aerodynamic origin, is not homologous. Scales are inherently planar; their surfaces are derived directly from the placode. Planar feathers (pennaceous) , by contrast, obtain their shape only after its growth from the cylindrical sheath, making them primarily cylindrical, and only secondarily flat (Prum & Brush 2002, Prum 2003). Any theory of feather evolution, then, must account for this.

For these reasons, various hypotheses, arising from speculation about function or the phylogenies of exant birds, and riding on the mistaken notion of "scales gone frayed", have fallen by the wayside. For the most part. Although ostensibly suitable "sturcturally homologous scales" are still sometimes advanced (Jones et al 2000, Zhang & Zhou 2002) there is a considerable amount of equivocation in their presentation (compare, for example, Feduccia in Jones et al 2000 with Feduccia 1999). Old habits, it seems, die hard.

The alternative hypothesis, in which basal feather precursors or even the general remigial architecture observed in flying birds arose in a strictly thermoregulatory context, has generally observed that feathers serve a demonstrable role in biothermal energetics and serve as insulatory mechanisms. The principal data in support of this hypothesis is phylogenetic, as opposed to the aerodynamic scenario; i.e., the thermoregulatory model most closely matches the best supported phylogeny of Aves. The greatest strength of the thermoregulatory model is that is closely congruent with the most current data on the embryogenesis of feathers. Nevertheless there are numerous objections, the simplest and most emphasized (e.g., Parkes 1966, Feduccia 1996) being: "why feathers?" Further objections raised and best summarized by Feduccia (1996) include:

a) Down feathers are secondarily derived in birds and thus cannot represent an ancestral feather morphology

b) Hairlike integumentary derivatives would do just as well as feathers for thermoregulation, as is seen in mammals.

c) Flight feathers degenerate in cursorial flightless birds

d) Pterylae contradict an initially thermoregulatory role

e) Feathers are fundamentally designed to be aerodynamic structures

The first of these objections is entirely irrelevant, as no one postulates that down feathers seen in extant birds represent the original feather morphology in any context, whether thermoregulatory or aerodynamic. The second objection is valid but the preponderance of evidence argues that the earliest feathers were simple, filamentous derivatives. The third objection, heavily stressed by Feduccia (1985, 1996, 1999a, b) is also irrelevant in that it merely demonstrates that aerodynamic pressures constrain the morphology of remiges in extant birds. The same can be said for the presence of pterylae, whose existence and conformation in extinct archaic birds cannot be known. Indeed, in some ratites, (e.g., Struthio and Rhea) pterylosis is not known, at least in the conformation of the secondary feathers. And lastly, the aerodynamic properties of feathers merely indicate that aerodynamic pressures govern the morphological features we observe in feathers amongst today's birds.

A logical objection to the hypothesis that feathers arose in an endothermic form is to argue that endothermy and feathers as insulatory mechanisms arose simultaneously in a reciprocal feedback loop, perhaps correlated with the initial ecological shift towards an arboreal lifestyle (e.g., Bock 1985).

Although the thermoregulatory and aerodynamic functional hypotheses have been at the center of the debate, they are by no means the only ones to have been advanced. Ostrom (1974, 1976) proposed that feathers arose primarily to aid in the capture of prey in a highly cursorial insectivorous (loosely defined) form, a form he argued was typified by Archaeopteryx. Distal ramification to increase the effectiveness of the "fly swatter" would have led to the derivation of advanced feathers later coopted for aerodynamic function. Ostrom later abandoned this model entirely after it received extensive criticism (e.g., Martin 1983) and revision by Caple et al. (1983).

Hypotheses in which the origin of feathers has been tied to sexual selection, have been proposed in one form or another (e.g., Stephen 1974, Cowen & Lipps 1982). While they do logically invoke the known influence of sexual selection upon evolutionary trajectories, they lack the ability to explain why such advanced integumentary derivatives as feathers would appear in lieu of some simpler and embryogenically and morphogenically more conservative structure, or any ability to predict the morphology of basal precursors. While these hypotheses cannot be entirely ruled out, it does not seem probable that they alone can account for the origin of feathers and their subsequent evolution.

More generally, there are problems with these explanations above and beyond what has already been mentioned. Obviously, generating hypotheses about morphology based on hypotheses of adaptive (or exaptive) function are probably not the strongest inferences. There is, in addition, a tendency for advocates of these views to reject homologous structures because of their functional preconceptions; that is, to attempt to establish homology according to function rather than form. There is no better example than Feduccia and his cohorts strenuously arguing that coelurosaurian integument is little more than frayed collagen fibers. Consequently, they also have a tendency to accept analogous structures that seem to fit; for example, the appendages of Longisquama (if they are appendages at all). Further still, this is part of a deeper tendency to conflate ones favorite avian phylogeny with feather function. For example, Feduccia insists birds are clearly derived from a "thecodont" stock and were arboreal. Therefore feathers must have functioned in an aerodynamic context.

The Feather: From Placode to Pterylae

In the past ten years or so, the field of evolutionary developmental biology has largely revolutionized our understanding of feather embryogenesis and ontogeny, which has in turn clarified our view of how feathers first appeared, regardless of the reason for which they appeared.
The feather placode above a condensation of dermal cells
Previous attempts, as mentioned above, did not take the complex heirarchical nature of feather development into account when hypothesizing basal feather morphs, and were therefore led down untenable paths. To understand feather evolution, one must first understand how they come into being.

Feather development begins with an epidermal placode situated above a condensation of dermal cells which specifies the particular feather's location.

From below, dermal cells work themselves upwards, forcing the epidermis into a finger-like projection called the papilla, or feather bud. Signaled by the dermis, the epidermal cells around the base of the papilla then sink down, creating an invagination called the lumen, or follicle cavity. Subsequent morphogenesis proceeds from the epidermal collar. Along its length, keratinoctyes proliferate and form barb ridges.
The papilla, or feather bud
These barb ridges are helically displaced as they grow, eventually making their way to the anterior midline and fusing to form the rachis ridge, which later becomes the feather rachis. Opposite the rachis ridge, new barb ridges spring out of the collar, these fusing with the rachis ridge anteriorly. On the barb ridges themselves, peripheral cells organize themselves into horizontal layers. Following the death of cells in the middle, those on either side become the paired barbules, with those more central fusing to become the ramus.

Finally, the whole structure, which until this point has remained essentially tubular, opens up. The outer surface becomes the dorsal surface of the fully developed feather, and the interior becomes the ventral. It should now be clear precisely why the planar surfaces of scales and feathers are not homologous; scales develop from the anterior and posterior surfaces of the placode directly, feathers round-aboutly develop their surface from the inner and outer surfaces of the cylindrical collar.

Also of great importance is the hierarchically contingent nature of the developmental processes. Barbs can only form on a collar, and a rachis can only be formed after the growth and displacement of the barbs. Distal and proximal barbules cannot close a vane unless they have barbules of some sort to grow from, which themselves originate from barbs. One step necessarily precedes the other.

The Evo-Devo of Feathers

Guided by the hierarchically contingent developmental process described above, Prum (1999) proposed a theory that seamlessly harmonizes the morphogenetic, biochemical and paleontological data in a way previous theories have failed to do. It involves essentially five stages, one built on top of another, broadly mimicking feather development while explicitly not being based on the discredited Haeckelian “law” of recapitulation.

The first stage is hypothesized to have originated with the first feather follicle. As above, the dermis would have pushed the epidermis into a collar, with the epidermis sinking around its base. This would have yielded a hollow, tubular structure much like the calamus of modern feathers. Stage II involves the origin of barbs. Derived from the collar, these would have opened up into a simple “tuft” extending from a calamus. Stage III has two stages which the theory cannot distinguish between in terms of temporal origination; either could have occurred first. What Prum labels IIIa involves the helical displacement of the stage II barbs and their fusion to form the rachis on the midline. The fully developed feather would have been pinnate, and superficially quite similar to modern feathers. With the evolution of stage IIIb, stage II barbs would have evolved barbules and ramus. Together, both stages would yield an open pennaceous feather complete with a rachis, ramus, barbs, and barbules. The following stage, stage IV, sees the evolution of distal and proximal barbules, built off IIIb, which would have hooked together and closed the vane. Fully developed, these are essentially modern, but symmetrical, feathers. All subsequent morphologic variety is subsumed under stage V, including asymmetrical flight feathers, and down.

How these changes are accomplished is a rather complicated matter, and all the intricacies have not yet come to light. What is known, however, is that plesiomorphic developmental pathways were co-opted and changed in such a way that novel feather structures were developed. An illustrative examples comes by way of Harris et al (2002), who looked at patterns of Shh and Bmp2 expression in a chicken (Gallus), duck (Anas), and alligator (Alligator). What they found was that at the placode stage, when both feathers and scales are just condensations of dermal cells, there is a conserved expression of Shh in the posterior domain, and of Bmp2 along the anterior border -- in both timing and polarity -- in each. Subsequent to this stage, we see derived coexpression of Shh and Bmp2 in the distal epithelim at which point the papilla is growing. Subsequent patterns of expression are also unique to feathers.

What all of this shows it that the placode and placode development are plesiomorphic in archosaurs, and that the Shh-Bmp2 module was probably co-opted during development multiple times, leading to a novel feather morphology after each. That is, in the primitive scaled precursor to feathered theropods in which Shh was expressed posteriorly and Bmp2 anteriorly, mutation altered this such that Shh and Bmp2 now additionally expressed themselves in the distal epithelium. Although this is my no means the entire story, the role of Shh and Bmp2 are illustrative of the molecular pathways Prum and his colleagues envision.

The Evolution of Feather Keratins

As noted above, phi-keratins are remarkably different from both alpha-keratins and other beta-keratins. Those of various feathers and their parts are a heterogeneous bunch, all with a mass of roughly 10.6 Kd. Those of scutate scales, claws and beaks yeild a similar electrophoretic array, but are larger, with a mass of 14.5 Kd (Brush 1996). This size difference, according to Walker & Bridgen (1976), is due to a repeating tripeptide sequence (Gly-Gly-X, where X is either Phr, Leu or Tyr) of 3 Kd. Other differences exist in the precise specifications for the beta-pleated sheath, and shorter/longer globular portions.

Because feather specific phi-keratins are clearly similar enough to establish homology, and non-feather classes broadly so, Brush proposed that an ancestral non-feather type phi-keratin gene (recently discovered in alligator claws by Sawyer et al. 2000, making it plesimorphic in archosaurs) underwent duplication and subsequent deletion of the Gly-Gly-X region, resulting in the two distinct sizes. Subsequent duplication and modification explain the similarity of all the smaller feather phi-keratins (Brush 1993, 1996, Prum & Brush 2002).

It is unknown if feather specific phi-keratins were present in the most basal feathers. Brush (1996, 2001) suggested they were, but Prum (1999) has argued that the morphological novelty of the feather itself probably preceded it.


Brush, A.H. 1980. Chemical heterogeneity in keratin proteins of avian epidermal structures: Possible relationships to structure and function. In The Skin of Vertebrates, Eds R. Spearman & P. Riley, pg 87-109. Linnean Society of London, London.

Brush, A.H. 1985. Convergent evolution of reticulate scales. Journal of Experimental Zoology 36: 303-308.

Brush, A.H. 1996. On the origin of feathers. Journal of Evolutionary Biology 9: 131-142.

Brush, A.H. 2001. The beginnings of feathers. In New Perspectives on the Origin and Early Evolution of Birds, Eds J. Gauthier & L.F. Gall, pg. 171-179. Peabody Museum of Natural History, New Haven.

Dyck, J. 1985. The evolution of feathers. Zoologica Scripta 14: 137�154.

Feduccia, A. 1980. The Age of Birds. Harvard University Press, Cambridge.

Feduccia, A. 1985. On why the dinosaurs lacked feathers. In The Beginnings of Birds, Eds M.K. Hecht et al, pg. 75�79. Freunde des Jura-Museums, Eichstatt.

Feduccia, A. 1996. The Origin and Evolution of Birds, First Edition. Yale University Press, New Haven.

Feduccia, A. 1999. The Origin and Evolution of Birds, Second Edition. Yale University Press, New Haven.

Jones, T.D. et al. 2000. Nonavian feathers in a late Triassic archosaur. Science 288: 2202�2205.

Maderson, P.F. & Alibardi, L. 2000. The development of the sauropsid integument: a contribution to the problem of the origin and evolution of feathers. American Zoologist 40: 513�529.

Martin, L. 1983. The origin of birds and of avian flight. Current Ornithology, 1: 105-129.

Ostrom, J.H. 1976. Archaeopteryx and the origin of birds. Biological Journal of the Linnean Society 8: 91-182.

Prum, R. 1999. Development and Evolutionary Origin of Feathers. Journal of Experimental Zoology 285: 291-306.

Prum, R. 2003. Are current critiques of the theropod origins of birds science? rebuttal to Feduccia (2002). The Auk 120: 550-561.

Prum, R. & Brush A.H. 2002. The evolutionary origin and diversification of feathers. The Quarterly Review of Biology 77: 261-295.

Zhang, F. & Zhou, Z. 2000. A primitive enantiornithine bird and the origin of feathers. Science 290: 1955�1959.

Personal tools