The Origin Of Feathers
Transcript of a paper presented
April 18, 1998 at Dinofest
In 1860, less than one year after the publication of The Origin of Species, Charles Darwin wrote the following to Asa Gray, "...I remember well the time when the thought of the eye made me cold all over, but I have got over this stage of the complaint, and now small trifling particulars of structure often make me very uncomfortable. The sight of a feather in a peacock's tail, whenever I gaze at it, makes me sick!" I hope that my talk this afternoon does not make someone ill. I think that anyone who studies birds or their ancestors would agree with Darwin's strong characterization of the problem: feathers, especially those on the wings, which are involved in flight, are a morphological wonder. Their complexity suggests a very rich evolutionary history. Although the topic of feather evolution is full of debate, one fact that is plainly obvious: feathers are multifunctional. Feathers are most important in thermoregulation, locomotion, and communication. However, it seems unlikely that feathers spontaneously began as such a dynamic structure, instead, over time, various selection forces allowed new features to be incorporated into the whole.
Today, there exist two "main" feather origin hypotheses. The first suggests that feathers evolved directly for flight through a series of stages. The other proposes that feathers were originally used like hair for thermoregulation. I would hazard a guess that most everyone in this room subscribes to one of these two hypotheses. I am here to talk about a third possibility: the informational origin and function of the initial protofeathers. Some of you might know of it as the display hypothesis.
This idea is not new; yet, researchers have basically ignored it. My intent today is not to convince you that "yes, this did happen" but instead "yes, it is plausible."
E. Mayr of Harvard was one of the first to suggest seriously that feathers could have evolved this way. In a paper entitled Evolutionary Novelties which was published in 1960, he writes that "there is every reason to believe that the group of reptiles ancestral to the birds already had feathers, even though they might have been acquired either for temperature control, as an epigamic character, or in some other way not connected with flight."
The idea was extended in 1982 in paper presented by Richard Cowen and Jere Lipps in what they called the display hypothesis. Since then, Cowen has re-iterated and modified their model in the several editions of his textbook, History of Life. In the 1994 edition, Cowen renamed the model the "display and fight" hypothesis, suggesting that wing flapping behaviors, bluffing, and fighting all might have been necessary for the evolution of powered flight.
I proposed the informational origin hypothesis in 1997. Although it is similar to the hypothesis of Cowen and Lipps, there are some significant differences. The informational origin hypothesis is much more generic than the current "display and fight" hypothesis.
Differences In The Two Models
Cowen has suggest a very active role for behaviors such as wing flapping, bluffing, and fighting as part of the evolution of the feather and instrumental in the evolution of flight. The informational origin hypothesis is not so bold and is more like the embryonic version of the original 1982 display hypothesis. The information origin model is more concerned with the very initial appearance and early evolution of protofeathers than the transition from protofeathers to flight.
The informational origin hypothesis recognizes that the kinds of information conveyed by the incipient protofeathers can not be known. These might have been extremely species-specific including sexual cues (male to female), social messages (parent to offspring), and territorial threats (male to male). Protofeathers could have also augmented other more general information such as warning marking (like venomous snakes) or cryptic colors (for camouflage). The informational hypothesis suggests that any initial displays would have been much more passive, simple, and probably grafted into existing behaviors. The word "display" is a misnomer. It assumes behaviors where information can be transmitted very passively from sender to recipient. A male cardinal is always red, sending information to recipients all the time, but it is not actively displaying each moment. Also, if the early protofeathers were involved in cryptic coloration, this can hardly be thought as of display, but a survival strategy via distorting the real data by providing misinformation in order to blend into the environment.
What Is information?
What do I mean by information? There is a sender and a recipient, although the sender does not need to be purposefully sending a signal. The information from the sender to the recipient causes a change in the recipient. At some level, the recipient is now "aware" of new data about the environment. It usually elicits a behavioral response. This can be as simple as fight or flee reaction. The fact that a certain bird has a blue head and a specific call sends several cues to female birds of that same species. The male bird did not have control over the color of his feathers or the kind of mating call he uses, as both are genetic programs. Likewise, the female responses are often instinctual, tied to stereotyped, species-specific cues. If cue A, then response B. A large set of simple, interacting rules like this one are used by the majority of animal species, resulting in complex behaviors.
Using Occam's razor, what kinds of information could be revealed by such new structures as protofeathers? I think the hypothetical "transred" model best illustrates this. At some point in time, T0, redness started being utilized as an informational cue in the avian ancestor. For this thought experiment, let us pretend that redness says to other individuals "I'm the male of the species." Females pick mates by several criteria; one of the important ones is how red the males are. Selection would produce males in each generation that were more red than the prior one. This strategy could be best summarized as "the redder, the better." By time T1, most males are covered by extremely red scales. Suddenly, at T2, spontaneous mutations occur, producing the novel, prefeather structure. This new phenotype would have either a negative or positive effect on fitness. If negative, it would be expected that natural selection would eliminate if from the population. If positive, selection should increase the prevalence of the trait. Those males that are "transred" might even attract more mates, even though they are the same size. They just appear to be more red than the other males. In this example, it is possible to see how this phenotype could have a selective advantage without a species also having to develop an entire set of new behaviors. The information in the "transred" model could have been any message. Redness could have been any informational cue, such as a warning signal to other animals. Red could have been instructing "yes, I am poisonous."
This "transred" model fits well with the "behavior evolves first" hypothesis which reasons that behaviors of some kind will occur before major morphological adaptations will take place. The webbed feet of ducks can provide an example. It is assumed that the webbed feet of duck ancestors occurred after they were living in aquatic habitats, not before.
The informational origin hypothesis assumes that the avian ancestors had good vision, perhaps even color vision, as visual cues must be read accurately by the recipient in order to work. Second, this hypothesis assumes that avian ancestors could have used all the kinds of informational cues used by animals today.
Objections To The Hypothesis
The first main objection to the informational origin hypothesis is that it can not be tested. It is hard to test; however, it is not impossible. By looking at extant birds, it is observed that feathers are being used for a number of behaviors, but most prominently, species-specific sexual cues. We can never know exactly what behaviors pre-avian creatures engaged; however, we can make an educated guess that similar sexual behaviors might have been some of the earliest. With this is mind, we would expect any initial fossils with protofeathers to be completely covered where the scales and later fossils to have certain areas, like tails, to have larger, wider feather, being used for such behaviors. There is good evidence already for this from fossil like Sinosauropteryx and Protarchaeopteryx. More data is needed; yet, assumptions and models can be tested against the current evidence.
Another objection is that there can be no selective force for the initial protofeathers. Natural Selection can not act on a structure that is not present. I agree. If feathers arose from scales, this argument is easily pushed aside. But what if feathers are a novel structure? How could there be any informational selection on the first protofeathers? No pre-avian ancestors is wishing feathers into place so it can attract more mates. This is not an argument for orthogenesis! I think the "transred" model I talked about earlier can provide an answer. If protofeathers spontaneous appeared via mutations, then the phenotype of those individuals would have been altered, effecting their overall fitness one way or another. Protofeathers could have been advantageous almost immediately, if they were grafted onto already existing cues and behaviors.
What Are feathers?
For a long time, it has been assumed that feathers were modified reptilian scales. Recent molecular studies by several researchers do question this basic assumption. The new data does give good reasons to think that feathers are not modified scales at all, but a completely novel structure. Another possibility is that pre-avian ancestors had evolved a new type of scale, consisting of different proteins from the reptile scales found today, and it is this that became the protofeather. The debate has not been fully settled, but it does illustrate how limited our data on feather evolution really is. However, the informational origin hypothesis is workable either way.
The first important fossil is Longisquama, a Triassic thecodont. Earlier researchers working on bird evolution had proposed that this animal might have been capable of limited flight. Cladistic analysis shows it is unlikely that this species is directly related to birds; however, it does have some very long dermal structures which are reminiscent of feathers. Some have gone so far to suggest these were part of a gliding apparatus, but this seems to be stretching reality to the limit. It is very likely that these were used for another purpose. It does not take too much imagination to think these elongated scales might have been used to convey information. They might have had another function, like defense; however, asserting they should have had some informational capability is not unreasonable. What can be gleaned from this fossil is that a very strong selective force had shaped and enhanced this very unique, feather-like phenotype.
For the past year, scientists, laypeople, and the media have been excited over the "feathered" Chinese dinosaur, Sinosauropteryx. I will not spend much time talking about this fossil as others have more eloquently analyzed it today. If you would like to learn more about this fossil, I will point you to the January 8, 1998 issue of Nature. It seems both statistically and cladistically unlikely that the ancestor of all birds was unearthed here, so it really should not be viewed as pre-avian. Second, it is not clear if the feather-like structures on the surface of this theropod are related to feathers at all. Nevertheless, feather-like structures cover the surface of the fossil, implying this morphological change does indeed have an advantage. This fossil species fits into the display model very well. The initial protofeathers would be expected to cover the body more or less equally, like scales, hair, and skin. Only after they had appeared could natural selection occur. The informational origin and function hypothesis would predict an animal much like Sinosauropteryx to exist at one time, early in the evolution of feathers. Whatever this creature was, it certainly looked very different from other theropods. The males and females certainly would have recognized others of their species quickly. Likewise, the other species would have been able to instantly notice the plumed Sinosauropteryx.
It is unfortunately that the fossil that could provide the some wonderful data is not well studied. Found in China, Protarchaeopteryx, is indeed a very interesting case study. It has been suggested that this species is a flightless theropod and not a bird, but this still needs more analysis. The feathers are symmetrical, with large, long feathers on the tail. Bird or theropod, the symmetrical shape and size of the feathers instructs us that this creature was no longer flying, if it ever flew at all. Natural selection kept these non-flight feathers intact and large, strongly suggesting that these were extremely important for display functions. If this fossil really is a non-avian theropod, then it will become very clear that feathers not only evolved before birds, but also were used for functions besides flight.
It is hard to analyze the feathers of Archaeopteryx for any single trait as a multitude of selective forces were simultaneously interacting. Clearly, this "first bird" could fly and the feathers have been strong shaped by selection for this. The huge feathers on the wings and tail are also what one might expect from display function. The fact that the long, bone and muscle tail is retained, instead of a feather one like in modern birds, does suggest that the tail was being used for displays. The long feathers on the tail are very apparent, just as one would expect if Archaeopteryx was using it for display as well as flight.
The informational origin of feathers would not be plausible at all if modern birds did not use their feathers in so many behaviors and displays. Besides obvious examples like the Peacock's tail and the various Birds of Paradise, feathers are extremely important in both sexual selection and simple species recognition in almost every, if not all, genus of birds. The fact the songbirds augment their feather behaviors with another layer of song-related behaviors is just an example of how important such behavior is too the lifestyle of birds.
Another interesting fact about modern birds is that they are highly influenced by what psychologists and ethnologists called supernormal stimuli. For example, studies have shown that various Geese species will abandon their own nest to sit on a "superegg." These supereggs were huge, white soccer balls. The normal stimulus would have been their own, white, round, egg. Supernormal stimulus follows one simple rule: more is better. Experiments with Java Sparrows and Doves have shown how birds of one species can act as supernormal stimuli. To the Java Sparrow, the Doves were just more "birdy" than others of their own species were. When given a choice, they would prefer to pair with Doves and would display their full range of social behaviors with the Doves. In fact, they would threaten other Sparrows that approached their Dove. Other experimenters have worked directly with feathers. By artificially manipulating the tails of several species of birds, an immediate effect on their fitness resulted. Birds, especially males, with shortened tails had a lowered fitness (i.e. less progeny) when compared to normal males. Males that had their tails lengthened had a much better time attracting mates and had a higher fitness. Of course, as R.A. Fisher pointed out in 1930, a male's tail can only grow so long until it is more costly to have than advantageous for mate attraction. A tail 50 feet long would certainly make it difficult to fly, walk, and avoid predators, regardless how attractive he would appear to the opposite sex. A final example come from Hawaii, where earlier this century male and female ducks both looked very similar, much like the female Mallard Duck of North America. When male Mallard Ducks with their characteristic green heads were introduced to the islands, female ducks preferred the green headed ducks and choose their mates accordingly. Greenness, an informational cue for maleness, was automatically selected. Today, all the male ducks in Hawaii have the traditional Mallard duck markings and green heads.
Since many bird species exhibit these stereotypic, genetic behaviors, it does suggest that their ancestors also had such behaviors. Responses to supernormal stimuli, if also part of the behavior schema of avian ancestors, could be the missing clue to how such protofeathers spread among the original population. If the informational origin of feathers is accepted as a real possibility, we should expect that other fossils might eventually be unearthed that have evidence of specialized feathers, such as long feathers on the tail. The current evidence, both extinct and extant, makes it hard to completely ignore the idea that feathers were informational devices very early and used for displays before birds had true, powered flight.
A Final Note
Modern researchers use a reductionist approach. This allows us to take the complex world and break it down into a simple models: If X, then Y. Although this makes for great papers, it is not how the real world-or natural selection-works. It is really more like this: If AB, Then XYZ; yet, if C, then only XZ, etc. This is something we all know intuitively, but often forget when discussing adaptionist models. In modern birds, many selective forces-thermoregulation, locomotion, communication, predation, environment, lifestyle, all work together to maintain and shape the feathers in each species. Different niches create distinct morphologies; the life histories of shorebirds and penguins produce specialized, yet disparate feathers. At some point in time, a synergy of natural forces, all acting and interacting shaped the kinds of feathers found on Archaeopteryx. We should not be surprised if the same thing has happened to protofeathers even earlier. Certainly protofeathers could have had more than one function. We should recognize that from day one, many selective forces were interacting on the initial protofeathers, resulting in one of the most complex and beautiful structures in the living world, the feathers of modern birds.
Thank you very much.
© 1998 Thom Quinn