Birds may be paedomorphic dinosaurs

In the last few decades we’ve realized that birds descended from theropod dinosaurs, and that the evolution of feathers preceded the evolution of flight. Indeed, many biologists still consider birds to be dinosaurs, since the group “dinosaurs” leaves out some of the descendents of dinosaurs—birds. Regardless of whether one adheres to this convention, we know from genetic and fossil evidence that birds are united with reptiles in one group: the Archosauria, which includes the common ancestor of crocodilians and birds, and all the descendants of that ancestor up to and including modern birds and crocodilians. That means that crocodilians are the living reptiles most closely related to birds (if, that is, you don’t consider birds to be reptiles).

Comparing birds with archosaurs, then, can tell us something about the evolutionary changes that produced our feathered comrades.  And this is what was done in a new paper in Nature by Bhart-Anjan Bhullar et al. What they showed is that the evolution of certain features of birds that distinguish them from their reptilian relatives, namely their large heads, eyes, braincase, and beak—probably evolved through paedomorphosisPaedomorphosis is the evolutionary process in which a descendant retains the juvenile state of an ancestor, retaining certain juvenile features of the ancestor into adulthood and reproductive age.  Paedomorphosis is one example of heterochrony: the process (discussed by Steve Gould in his book Ontogeny and Phylogeny) whereby evolution operates by changing the timing of development.

The classic example of paedomorphosis is the axolotl (Ambystoma mexicanum), a Mexican salamander that becomes sexually mature while retaining the juvenile features that are lost when other salamanders mature, including the gills and the caudal fin along the back:

In the new Nature paper, the authors realized that the juvenile stages of ancient archosaurs resembled the adult stages of birds, their descendants, implying that perhaps the evolution of birds involved a process of heterochrony.  Birds show smaller snouts and bigger eyes and braincases than those of their adult archosaurian relatives, but young archosaurs have the same big heads, big eyes, and short snouts as do adult (and juvenile) birds.

Here’s  figure from the paper showing the difference between juveniles and adults of three living and fossil species: a crocodile (b), a primitive dinosaur (Coelophysis; c), and the famous Archaeopteryx (d; a stem group bird, known in Darwin’s time but now realized to be a transitional form between dinosaurs and birds).  Note the similar change in morphology between adult and juvenile in the alligator and dinosaur, but the general retention of juvenile morphology in Archaeopteryx:

(from the paper): b–d, skulls of selected archosaurs: Alligator 46-day embryo (b, left) and adult (b, right); Coelophysis (primitive dinosaur) juvenile (c, left) and adult (c, right); Archaeopteryx (stem-group bird) juvenile (d, left) and adult (d, right).

To show this pattern statistically, the authors did some fancy manipulations known as Principal-Component Analysis (PCA), parsing out those groups of morphological changes that unite related species. To do this they used both juveniles and adults of ancient archosaurs and theropods, modern birds, and alligators.  I needn’t go into detail, but what they got was confirmation of the impressionistic pattern shown above: archosaurs (including alligators) all undergo elongation of the face, diminution of the eye orbit, and reduction of the neurocranium during development.  Modern birds don’t change nearly so much, and early (fossil) birds change, as expected, in an intermediate way (for example, they undergo a bit of skull elongation). As the authors note:

Evidence for heterochrony is clear. Whereas adults of taxa distantly related to birds (non-eumaniraptorans) cluster together, basally branching bird relatives (eumaniraptorans) cluster with the embryos and youngest juveniles of other non-avian archosaurs (Figs 2 and 3), with the more crownward avialan Confuciusornis [an early bird] nearly identical to embryos and particularly close to the perinate enantiornithine (Figs 3d and 4).

The figure shows the configuration of the embryonic alligator skull, in green, superimposed on the adult alligator skull in red to the left and on the adult skull of the primitive bird Confuciusornis to the right. You can see how much more the the juvenile alligator resembles the adult bird than the juvenile alligator resembles the adult alligator. It’s hard to avoid the conclusion that, during the evolution of birds from their reptilian ancestors, evolution acted to retain the reptilian juvenile skull into adulthood.

Here, by the way, is a wonderful fossil of Confuciusornis, showing its paired tail feathers, and a reconstruction of what the adult looked like:

So maybe birds’ heads evolved simply by stopping the genetic changes that made the heads longer (and the eyes and braincases relatively smaller) during development of their ancestors. Why would evolution act this way? One answer seems obvious: birds need to rely heavily on vision to fly, and they need large brains to harbor the vision centers as well as the neuronal apparatus enabling them to fly.

Note that the postcranial skeleton (the rest of the bird) shows no signs of padeomorphosis. Indeed, some features of the bird body are peramorphic with respect to dinosaurs: that is, they seem to show an extension (rather than a truncation) of development past the adult stage.  The authors say that’s okay because the brain is modular (i.e. genetically independent) of the rest of the skeleton, and can evolve independently, so the paedomorphosis could have involved genetic changes affecting only the skull.  Something similar happened during human evolution: as we can see from our early australopithecine relatives, our postcranial skeletons evolved much faster than our cranium. Lucy, for example (an A. afarensis) has basically an apelike skull sitting atop a skeleton which is much more similar to that of modern humans.

There are other things in the paper that will be of more interest to specialists (e.g., when the various stages of paedomorphic change occurred and what happened with the beak), but I’ve given enough, I think, to show that once more an intriguing evolutionary hypothesis was sitting under our noses all along. Juvenile archosaurs have been around for years, but nobody thought to compare them to early and modern birds.


Bhullar, B.-A. S., J. Marugan-Lobon, F. Racimo, G. S. Bever, T. B. Rowe, M. A. Norell, and A. Abzhanov. 2012. Birds have paedomorphic dinosaur skulls. Nature, advance online publication, doi:10.1038/nature11146.

58 thoughts on “Birds may be paedomorphic dinosaurs

  1. So maybe birds’ heads evolved simply by stopping the genetic changes that made the heads longer (and the eyes and braincases relatively smaller) during development of their ancestors.

    I think I see what you’re driving at, but this seems like a confusing way of putting it, since it’s presumably not the case that bird snout growth stops early. Rather, what’s happening is that orbit and cranial growth continue to keep pace with snout growth, past the point at which they would have stopped in other archosaurs.

    In other words, it’s not that bird snouts are shorter (relative to body size), it’s that the rest of the head is relatively larger.

    Do I have that right?

  2. I guess the real question to ask with regard to changes like this which came from preserving juvenile characteristics in adulthood, is just why the earlier creatures evolved with a youthful stage including those characteristics in the first place?

    1. This was my first question, too. The definition given in the article is: “Paedomorphosis is the evolutionary process in which a descendant retains the juvenile state of an ancestor, retaining certain juvenile features of the ancestor into adulthood and reproductive age.” Gould refers to neoteny in several of his essays; without looking it up and just trying to remember what he said, I think it was that neoteny involved extended development while still in the juvenile stage (that is, of the current animal, not of an ancestral form). And so I thought that maybe they are terms two different phenomena. But, I may be wrong.

      On a much lighter note, while “archaeopteryx” may mean “ancient winged” in Greek; in modern evolution it appears to mean “first crocoduck”!

    2. One convention is that neoteny is one of two kinds of paedomorphism – one of two ways to get it. It involves mainly a slowing down of development rate (rates of cell division and differentiation into tissues) relative to the ancestral condition. The other kind is progenesis or truncation, in which development proceeds at the same rate, but is ended or truncated earlier (again, always relative to the ancestral condition). In practice, it is not always easy to distinguish these, and the both can combine to produce a juvenile-looking descendant.

      In a similar way, a peramorphic character can arise from a speeding up of development (acceleration) OR developing at the same rate but for a longer period (a prolongation of development). Our very high brain-to-body size ratio, compared to other hominoids, is thought to involve a prolongation of the fetal rate of development (even after we’re no longer fetuses) more than an acceleration of brain development.

      So basically, one way to look at it is that heterochrony (paedo- or permorophis) depends on an evolutionary change in the RATE of development, the DURATION of development, or both.

      1. I had only ever seen neoteny used synonymously with paedomorphism. It does make quite a bit of sense to have the distinctions, though. Thanks for the info.

  3. Fascinating! This is an aspect of evolution totally new to me – makes me feel all new in the head! Thanks.

      1. And what gives you the idea computer science is all about electronic computers?

        PCA is an algorithm. Pearson developed a lot of statistical theory and statistical algorithms. By any sensible modern definition of computer science, Pearson’s discovery of PCA would be considered a great achievement in computer science.

        Your comment is somewhat like saying that because Newton was a mathematician, we shouldn’t consider his work as physics.

        1. Your arbitrary definition of anything that makes use of algorithms as “computer science” makes that subject so all-encompassing so as to lose any practical or useful meaning. Given the customary usage of the phrase, your claim that Pearson was performing computer science in 1901 is just silly.

          1. Your arbitrary definition of anything that makes use of algorithms

            No, that’s not the “widely accepted” definition, nor is it my definition. “Making use” of an algorithm is typically a very very different activity from designing general purpose algorithms. The latter though is what (a lot of) computer science, especially theoretical computer science, is all about.

            You might find my definition silly, but the fact that the term “computer science” did not exist in the 1900 does not make algorithmic achievements in that area any less algorithmic and “computer-sciency”. You are welcome, of course, to disagree with what most computer scientists would consider to be the definition of computer science.

            1. I think the key is your assumption that one can arbitrarily decide which earlier (pre-computer) achievements are indeed “in that area”, as you say. I’ll have to take your word for it that most computer scientists think that computer science per se was going on in 1900.

              1. Well, we routinely consider the (unknown) inventors of the decimal system, and the developers and chroniclers of arithmetical algorithms in India and Arabia as the first computer scientists. Al-Khwarizmi is almost considered a patron saint (:D) and lends his name to the centrally important notion of algorithms.

          2. I just realized that this discussion about what really computer science is all about might be severely off-topic, so let’s refrain from prolonging it on this thread. As a professional computer-scientist, I am all too aware of the rather unjust opinions people from other fields typically have about our “new” field, but this is not a fight I want to pick here. If you are one of the commentors here who are offended by my insistence that finding algorithmic procedures like PCA is what a lot of modern computer science is all about, then I offer my unconditional apologies.

            I’d just like to conclude the discussion with this quote from Edsger Djikstra:

            Computer science is as much about computers as astronomy is about telescopes.

            1. OT, but interesting anyway – a lot of us do think of computer science as being concerned with electronics, good to get a more informed perspective.

              1. It’s worth remembering that before WWII, a “computer” was a person who performed calculations.

                For example, from the 1911 Encyclopædia Britannica article, “Observatory”: “There are now two chief assistants, six assistants, and a staff of computers employed.”


  4. Also, I would be interested in knowing how age at tiem of fossilization (juvenile/adult) of fossils is determined or estimated. Wouldn’t the fact that heterochrony is common make it difficult to say whether a given fossil is a different species or just a differently aged specimen of the the same species?

      1. Thanks, I’d imagined disputes like that should occur. This article seems to refer to a few known ways, however, of determining the age of the animal at the time of fossilization:

        ….microscope examinations of thin slices of bone from Triceratops and Torosaurus specimens reveal that individuals attributed to Torosaurus are more mature than any of the ones assigned to Triceratops. Scannella and Horner therefore believe that the fossils that have been categorized as Torosaurus are just Triceratops individuals that reached mature adulthood before they died.

        1. I was fascinated by the suggestion that Torosaurus was just a fully mature Triceratops and have been paying attention to the literature. Another study seems to show that they’re wrong. The paper is in PLoS ONE, so it’s open access.

          Torosaurus Is Not Triceratops: Ontogeny in Chasmosaurine Ceratopsids as a Case Study in Dinosaur Taxonomy
          Longrich, Nicholas R. and Field, Daniel J.
          PLOS ONE Volume: 7 Issue: 2
          DOI: 10.1371/journal.pone.0032623

  5. The dog whisperer on Nat Geo channel has said that his method of applying wolf pack behavior rules to pet dog training has the best success with large breed dogs. But trying to correct/influence the behavior of small breeds is not very easy using his wolf pack techniques. Could this be the start of another example of neoteny? The farther away from an adult wolf a domestic dogs body is then also is their behavior. The small dogs have retained a lot of wolf pup behavior to appeal to their new ‘parents’ (those funny looking nearly hairless primates).

    1. I’m not sure that human manipulation of dog breeding falls into the same category as evolution, so any neoteny isn’t “natural”. The toy breeds are essentially dwarfed to the point that their brains as well as their bodies are abnormal. According to my vet (sorry, I don’t have any other reference), up to 25% of the individuals of some toy breeds suffer from epilepsy or other seizure disorder.

    2. i don’t think this is neoteny …
      firstly because it includes human intervention and secondly because it’s a behaviour … and thirdly its not evolution its manipulation..
      in that way, i mean the behavioural change of a pack of wolfs, if someday a group of men jumps from a tree to tree very efficiently like a monkey and eats raw food… will you be calling it neoteny ????? rubbish…

  6. We can see this (neoteny, I call it) in us humans. I describe humans as apes adapted to a savannah environment while the other apes are adapted to a tropical rainforest environment. I took a picture of a young gorilla skeleton in Prague’s Natural History Museum, a couple of years ago. Apart from the length of it’s arms, it was so similar to a human’s skeleton. The cranium, particularly, was uncannily like a human’s and was large in proportion to the rest of the body frame, unlike the adult which had a much smaller head to body frame ratio. The adult’s skull architecture was also much more robust than the infant gorilla’s.

    So neoteny, in our case, provided the much larger brain-case needed for an increasing brain capacity and faculties needed to survive in the very competitive and hostile environment of East Africa.

    You may say that we are closer to chimpanzees than gorillas but this form of neoteny in cranial development probably holds for chimpanzees too. Their genetics are very similar and in fact, it can be seen that the gorilla is merely a variant of chimpanzee which has adapted to a niche on the upper slopes of volcanoes – where the tree cover is more sparse, the climate is colder and large ground predators are often encountered. For these reasons they had to be able to walk more often on the ground, had to be bigger to withstand the cold, and much more aggressive to defend themselves against lions etc.

    1. All hominoids, and certainly chimps and gorillas as you say, start with a rounded or globular fetal skull that gradually becomes more horizontal, with a sloping back “forehead” and a jaw that projects forward. Our skull and face are neotenic in that we retain the rounded, high brain case characteristic of juvenile apes, and do not develop the forward (prognathous) jaw. That is why juvenile chimps and gorillas look much more human-like than the adults. Less development of the skull in humans accomodates our hyperdeveloped and more globular shaped brain.

      1. Exactly, Frank, and also the higher vaulted and more vertical architecture of infant apes skulls is associated with birthing size limitations.

        Most mammals have a ‘horizontal’ skull architecture and present their snouts first into the birth canal. Apes present the crown of their heads and it can be seen why humans have evolved skulls which are more cylindrical in shape and have flatter faces.

    2. I thought

      a) Human skull neoteny had been long abandoned for the realization that the unique teeth and diet drove the chin reduction that enables a larger brain case, while social communication face/eyes may have independently driven face flattening.

      If the rest of the early hominids were mixing ancestral and derived features, why would the skull with its many functions be different?

      b) Gorillas split earlier than chimps relative to us, and they have evolved dissimilar diets and derived knuckle walk hand mechanisms. (Humans have hands more like the ancestral hominoid AFAIU.)

      So a “variant of chimpanzee” sounds perfectly wrong to me.

      1. Neoteny is the proximate developmental mechanism – the adaptive significance is a separate issue. You have to be careful to distinguish proximate and ultimate (adaptive) causation. And you’re correct that the gorilla is in no way a variant of a chimpanzee.

  7. Hello all,

    I’m the first author of this study. First let me thank you for blogging about our work (although a spelling correction on my name would be great). I just wanted to quickly address the issue of neoteny vs. progenesis. Frank summarizes the definitions that we used, as delineated in Alberch, Gould, Oster, and Wake (1979), Paleobiology. In the main document and in the supplement, we present evidence that the paedomorphic transitions we posit were the result of progenesis, wherein the entire developmental trajectory was truncated. The two pieces of supporting evidence are the accompanying decrease in size and the accompanying decrease in time to maturity, a measurement made possible by the lovely histological studies of fossils performed by talented workers in the field during the last couple of decades.


    1. Whoops! Spelling error corrected. And thanks for stopping by with your comment.

      1. Not a problem. It’s wonderful to have such thoughtful and accurate coverage of our work. I feel flattered and undeserving. The discussion on human neoteny etc. here is very interesting, as well. I’m intimidated by your readers =)

        1. Glad to see that you went with Alberch et al.’s (1979) formalism – heterochronic terminology seems to have become terribly confused since they published. Not that it wasn’t before, but they gave clear alternatives that could be flexibly combined.
          Haven’t had a chance to read the paper yet (candidacy tomorrow), although it is on my to-read list, so maybe you covered this, but is it possible that birds took a paedomorphic route because they were adopting a K-strategy lifestyle? Being able to fly seems to put vertebrates into a protected-lifestyle zone and you’d expect longer lifespan and reduced reproductive output with that.

          1. Hi Lars,

            To a certain extent, of course, r-selection and k-selection are relative terms. The question here is then whether the ancestral bird, as best we can infer from the crown clade, took much more of a k-selection strategy than a pre-paedomorphic dinosaur. Paleognath birds and primitive neognath birds produce relatively large clutches and as a rule have precocial offspring. The clutches seem about the same size as those of similarly sized pre-paedomorphic dinosaurs as far as is known. Moreover, parental care etc. predate birds considerably and go back to before the crown of Archosauria judging from crocodylian behavior. Finally, the paedomorphic events were actually accompanied by a reduction of lifespan and time to reproductive maturity (thus quicker generation time? Allowing radiation? Probably too speculative). There may way have been a number of interesting things going on as far as life history evolution at the origin of birds, but perhaps they were rather more complex, with elements of k-selection among others. Your point is taken, though, that some time between the divergence of crocodilians and extant birds, a more k-selective strategy was adopted. Certainly songbirds show an extreme on that end.


            1. Thanks, Anjan – I was thinking of the generally longer lifespan and reduced clutch (or litter) size in modern bats and birds, compared to similar-sized non-flying endotherms, and how Gould correlated this sort of thing with neoteny. But of course in detail it’s never that simple.

          2. Right, Alberch et al. was a big step forward, but Gould’s book on Ontogeny and Phylogeny is still full of gems. Paedomorphism is not a process itself, but rather the result of processes like neoteny, progenesis, rate hypomorphosis, etc.

  8. Prof. Coyne, thanks very much for posting this. It’s wonderful to read such a clear, concise synopsis of what I’m sure is a very complicated subject.

    1. I’m chiming in to pile on more thanks. The details of the evolution of birds I find to be intriguing as well as confusing. I finally get a chance to try to educate myself on this subject.

    1. A more meat-based diet also supposedly fueled the energy for human brain growth. But many of the anthro ideas I studied are out of fashion these days.

    1. Neoteny & progenesis are two of the Heterochronic processes that can set macroevolutionary changes in a lineage. The outcome of neoteny and progenesis (diff. ways)lead to paedomorphosis or paedomorphic forms- when mature reproductive organs are found along with juvenile features….. in a few words neoteny leads to paedomorphosis.. and same for progenesis… and this is the basic difference between neoteny and paedeomorhosis…

Comments are closed.