Of cats and men: old genes give embryos an hourglass figure

by Matthew Cobb

Cats and humans look very different. Or do they? 32-day old embryos of the two species look pretty identical, as shown here (guess which is the cat! Answer at the end):

Why do they look so similar? This is one of the oddest riddles in biology, which has now been given a new answer following the publication of two articles in Nature today. To understand why this is so important, we need to go back nearly 200 years.

In 1828, Karl von Baer – the first man to observe the human egg – studied a number of embryos of different species and noted that the common features of large groupings of organisms, such as vertebrates, develop before the more specific characters. So, for example, human embryos have tails, and they also have folded structures around their neck that are similar to those seen in fish embryos, and which turn into gills in our water-dwelling relatives.

At the end of the 19th century, the German embryologist Ernst Haeckel turned this observation into an evolutionary “law”, in which he suggested that each organism, as it grows, goes through its evolutionary past. He put this in a fancy phrase: “Ontogeny recapitulates phylogeny”, and provided some neat drawings (these are from Wikipedia):

Ernst Haeckel's Embryological Illustrations from 1874. Taken from Wikipedia. From right to left: Human, dog, deer, pig, chicken, tortoise (note the shell), salamander and fish.

Although this might sound clever, it is not actually true (and it has been suggested that Haeckel’s drawings were not as accurate as they should have been!). Human embryos do not have gills – the embryonic structures become our jaws, not gills. Furthermore, Haeckel’s idea was based on a linear view of evolution that is plain wrong. Species do not evolve in a straight line, but a complex branched pattern. We share a common ancestor with a cat, but our lineages have been going their own way – and evolving differently – ever since.

But even if Haeckel was wrong, it doesn’t get away from the bizarre fact that human embryos have tails. What’s going on there? It turns out that the situation is more complex than Haeckel realised. At the earliest and latest stages of embryonic development, organisms within a given group look very different. But in a middle phase, they all tend to look pretty much the same. This effect is at its strongest in the different animal phyla – vertebrates, arthropods and so on – and is referred to as the “phylotypic” period, because each phylum has a typical common embryonic form which species in that phylum adopt, before turning into their final, species-specific shape.

This “hourglass” shape of embryonic development – specific/common/specific – has long intrigued biologists, but some people felt it might not be true, as it was based primarily on observations of physical similarity, not on the underlying genetic processes. Nature has just published two studies of very different model systems – the fruitfly Drosophila and the zebrafish – that cast a new light on the genetics of the “hourglass”, showing that the oldest genes are involved in the middle, “phylotypic” period.

The study on Drosophila – carried out by a team led by researchers from the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden, Duke University and the University of Manchester – looked at the genes that are expressed during the embryonic development of fruitfly eggs from six Drosophila species, some of which have been separate for 40 million years. Even within these apparently-identical species (one fruitfly looks pretty much like another, unless you are an entomologist – or a fly), there are big differences in the genes that are expressed early and late in development but, just as predicted by the hourglass model, each species tends to express the same genes during the middle, “phylotypic” period.

As you might expect, these middle genes are involved in important processes that are common to all the Drosophila species, such as synthesizing biomolecules, organizing chromosomes and above all controlling anatomical development. Genes that are not involved in development tended not to follow the “hourglass”.

In a partner article, researchers from the Max Planck Institute for Evolutionary Biology and Ruđer Bošković Institute looked at the genes that are activated in 60 different stages of the zebrafish, from the embryo on into the adult. As with Drosophila, they found that the oldest genes were expressed during the middle, “phylotypic” period of embryonic development. Younger, more specific, genes tended to be expressed in the earlier and later phases, with the intriguing exception of the oldest animals, which once again tended to express evolutionarily ancient genes.

Fascinating as these studies are, they only answer half the question. We now know how this “phylotypic” middle phase of the “hourglass” is produced – through the expression of ancient genes. But we do not know why.

The Drosophila group found evidence that natural selection in the form of selective constraint is acting on these patterns of gene expression, perhaps related to the fact that there is a best way of assembling the shared aspects of each body shape, constraining innovation and resulting in the conservation of ancient, highly effective genes involved in development.

The zebrafish group take a slightly different tack and refer to Darwin’s suggestion that environmental influences might be strongest at the earliest and latest phases of development, meaning that in the middle, phylotypic phase, the embryo is less “visible” to adaptation by natural selection, leading to a lack of evolutionary innovation. This would also explain the fact that very old fish express very ancient genes – once an organism is past reproductive age, it generally cannot be “seen” by natural selection.

These two papers shed a new light on one of the most bizarre questions in biology – why do so many embryos look alike? They show that although ontogeny does not recapitulate phylogeny, phylogeny is involved in ontogeny – the way we grow does shed light on our evolutionary past. The “hourglass” is real at the visible and molecular level, and the genes that specify the middle, more general, shape of an organism are some of the oldest in each species’ genome.

Now we need to know why, and above all why did you and I have a tail when we were only a few weeks old?

In commenting on an earlier draft of this post, Jerry responded to this final point this way:

“I think we know the answer to that question–at least sort of, but you imply that it’s a complete mystery.  Clearly our development evolved from an ancestral developmental plan, that that ancestor had a tail, gills, etc.  We retain that plan (yes, we’re not sure why, but probably because changing it would have screwed up everything downstream). Don’t you think you should give the readers at least a BIT of explanation about this stuff? It is, after all, the question with which you begin!”

Well sorry, folks, but Jerry’s pearls of wisdom are as much as you’re going to get on that point. Come up with your own tales about tails in the comments below!

Answer: The cat is on the left. Or maybe it’s the right.

[Edit: There's a nice summary of these papers over at Discovery, together with some useful quotes from the authors.]


Alex T. Kalinka, Karolina M. Varga, Dave T. Gerrard, Stephan Preibisch, David L. Corcoran, Julia Jarrells, Uwe Ohler, Casey M. Bergman & Pavel Tomancak (2010). Gene expression divergence recapitulates the developmental hourglass model. Nature 468, 811–814.

Tomislav Domazet-Lošo & Diethard Tautz (2010). A phylogenetically based transcriptome age index mirrors ontogenetic divergence patterns. Nature 468, 815–818.


  1. Flow
    Posted December 8, 2010 at 1:15 pm | Permalink

    I am skeptical that these papers tell us much. It is well known that early embryos reflect the shape of eggs and the shape of eggs reflects reproductive strategies. A zebrafish embryo, for instance, develops around a huge ball of yolk owing to the fact that fish dump masses of eggs into the water. A Drosophila embryo has to develop real fast as the fruits rot away fast. Thus early embryos are divergent and that this is refected in gene expression and gene ontology does not seem so surprising.

    PS. it is not fair to compare 32 day embryos between two species whose gestation times differ by a factor of 4-5

  2. Vincent
    Posted December 8, 2010 at 1:56 pm | Permalink

    THx Jerry! One of your best posts in a while. This is what I visit your blog for!

    • daveau
      Posted December 8, 2010 at 2:57 pm | Permalink

      Ohhh. That’s gotta smart. For both Jerry and Matthew.

      • whyevolutionistrue
        Posted December 8, 2010 at 3:07 pm | Permalink

        Not so much for MATTHEW . . . .

        • Matthew Cobb
          Posted December 8, 2010 at 3:45 pm | Permalink

          Vincent is really my Mum, Jerry…

      • Launcher
        Posted December 8, 2010 at 6:21 pm | Permalink

        Perhaps Drs Coyne and Cobb (and the other posters) should ‘evolve’ to have different looks to their postings. Maybe ALL CAPS, or cute ~serifs~.

  3. Sam
    Posted December 8, 2010 at 2:19 pm | Permalink

    Maybe someone can explain it to me, but I didn’t see how Haeckel’s idea was based on a linear concept of evolution. Surely you can draw a straight line from any modern species right back to the first self-replicating molecule without thinking that this lineage is the only one in existence, which is all that Haeckel’s idea requires.

    • Ichthyic
      Posted December 8, 2010 at 3:59 pm | Permalink

      by linear is meant “direct line”.

      as if bacteria->jellyfish->fish->lizard->cat->human was a direct line of descent.

      this isn’t the way it has happened, obviously.

      hence we know something Haeckel didn’t:

      evolution took a branching path, not a linear one.

      • Ichthyic
        Posted December 8, 2010 at 4:02 pm | Permalink

        btw, look at Haeckel’s drawings again.

        do you think humans evolved from pigs?

        why not?

      • Ben Breuer
        Posted December 9, 2010 at 6:14 am | Permalink

        I’m not completely sure this is correct: Haeckel did believe in a branching history of life; that’s what his term “phylogeny” is all about. However, he laid the emphasis on the lineage history (the “-geny” part), not so much on the branching (into “phyla”) and how it came about (selection* and all that messy stuff).

        In your example lineage, Haeckel would (probably) not have believed that a jellyfish morphed into a vertebrate, but that the vertebrate shared its lineage with the jellyfish up to the branching time, and that some “signal” of that shared evolutionary history of the two lineages was preserved in a correspondingly early stage of development. Development thus recapitulates lineage history.

        Haeckel was mainly interested in development as copious and credible evidence for evolution. (Fossils were quite rare, with spotty record, and comparative anatomy did not yield much evidence about the early branchings (the early phylogeny) in the animal kingdom.) Effectively he was pushing for (and getting, in his students) a kind of “devo-evo”, evolutionary biology informed by development studies.

        *Haeckel overextended the concept of selection, and considered the influence of natural selection something of a corruption of the precious historical records he thought hidden in development processes. We think somewhat differently today, but Haeckel is not that far from the modern consensus.

        OK. Sermon over. :-)
        NB: Biological (and bio-historical) layperson speaking here.

        • Ben Breuer
          Posted December 9, 2010 at 6:20 am | Permalink

          To add just a bit: Haeckel was a “progressivist.” He did think that vertebrates were more advanced than jellyfish. (So he didn’t get the egalitarian consequences of the tree-of-life concept.) And he extended that idea of “progress” (and qualitative hierarchies) to human groups, and even professions. From p.o.v. of later German history, a distinctly creepy aspect of his thinking.

  4. Insightful Ape
    Posted December 8, 2010 at 3:25 pm | Permalink

    Sorry, I need to ask a question. What is meant by “oldest fish”? Does that mean fish that haven’t changed much during evolutionary time? It can’t be about memebers of one species since we are talking about embryonic phase. But then what does it mean that once you are past a certain age you are invisible to natural selection?That is true for individuals not whole species.
    Also how do we know a gene is ancient? Is it through its existence in the most sister species?Number of mutations? Combination of all?

    • Matthew Cobb
      Posted December 8, 2010 at 3:50 pm | Permalink

      The fish study included *adult* fish, so oldest fish were simply that – the oldest fish in years (or months of whatever). You tell a gene is ‘ancient’ by seeing how many species from how broad a spectrum share it. So for example some of the ‘homeobox” developmental genes are shared by all animals, so we infer they go back to our common ancestor 600 MY ago. Others are found in insects but not centipedes, so must have appeared since those lineages diverged (and long after the human lineage split from the Arthropoda).

    • whrrr
      Posted December 8, 2010 at 6:10 pm | Permalink

      I was confused by this at also, but it is referencing

      “Younger, more specific, genes tended to be expressed in the earlier and later phases, with the intriguing exception of the oldest animals, which once again tended to express evolutionarily ancient genes.”

      This is talking about ancient genes becoming ‘activated’ or expressed in adult fish, not stating that adult fish have highly conserved genomes, which is how i read it at first.

  5. Noam Chomsky
    Posted December 8, 2010 at 4:25 pm | Permalink

    Tim Martin,
    you agreed to email this article if I could not access it. Thank you for trying to help me out. It’s about new genes constructed from already existing genes. Could you please email it to bobsmith99 at catholic.org


    • Tim Martin
      Posted December 8, 2010 at 6:30 pm | Permalink


  6. KP
    Posted December 8, 2010 at 4:33 pm | Permalink

    I happen to know the one on the left is a cat because I use it in my Human Anatomy and Physiology class lecture on embryology.

  7. Nick (Matzke)
    Posted December 8, 2010 at 6:28 pm | Permalink

    Oops, Jerry Coyne –> Matthew Cobb (I missed the author the first time through)

  8. Nick (Matzke)
    Posted December 8, 2010 at 10:21 pm | Permalink

    Comment 7 refers to a post with 2 links which much be stuck in moderation…

    • J.J.E.
      Posted December 8, 2010 at 11:00 pm | Permalink

      Yeah, even regular commenters get stuck in moderation if a high proportion of the text is composed of URLs.

  9. PhiloKGB
    Posted December 9, 2010 at 11:14 am | Permalink

    Obviously the maintenance of developmental pathways has much to do with the continued embryonic presence of ancestral structures. Vertebrate notochords are the main sources of Sonic Hedgehog in the dorsal body; pronephra and metanephra have critical signaling duties in the pathway to amniotic kidneys. Getting from ‘here-to-there’ is now the job these bits have taken on, and getting somites and metanephric kidneys some other way would require massive reengineering projects whose possible selection pressures are not obvious.

  10. Matt G
    Posted December 10, 2010 at 8:54 pm | Permalink

    RE: Flow’s point about a fair way to compare embryos. Is there a convention for normalizing the embryos to gestation time? Are there points at which certain developmental periods end or begin which are marked by a gene being turned on or off, for example?

  11. Dan L.
    Posted December 13, 2010 at 12:58 pm | Permalink

    Fascinating as these studies are, they only answer half the question. We now know how this “phylotypic” middle phase of the “hourglass” is produced – through the expression of ancient genes. But we do not know why.

    Matthew…hadn’t you heard? Science can only tell you how, it can’t tell you why! /biologos

    Sorry, couldn’t resist.

4 Trackbacks/Pingbacks

  1. [...] is another article from Why Evolution Is True You’ve seen the pictures of the various embryos of animals and how similar they look. You’ve also seen how similar they look in the “middle” [...]

  2. [...] the other is a human embryo. Can you tell which is which? For a discussion of this, you can find it here at Jerry Coyne’s blog Why Evolution is True. LikeBe the first to like this [...]

  3. [...] Nature 1, 2 Jerry Coyne‘s blog, PZMeyers, [...]

  4. [...] Uit: Nature 1, 2 Jerry Coyne‘s blog, PZMeyers, Panda’sThumb.    (1)   [...]

Post a Comment

Required fields are marked *


Get every new post delivered to your Inbox.

Join 25,540 other followers

%d bloggers like this: