Development is epigenetic

May 14, 2011 • 10:21 am

by Greg Mayer

One of the points I stress to students in my evolution class is that development is epigenetic: organisms develop from a less differentiated state to a more differentiated state. In modern terms, genes, the intraembryonic environment, and the extraembryonic environment interact to produce the organism through a sequence of stages going from an undeveloped to a mature state. The general point (though not the part about genes) has been known for a couple of centuries, so it might seem it wouldn’t be necessary to emphasize it, but the alternative view of development– preformationism— has a surprising hold on people’s minds. Preformationism maintains that development is essentially growth: there is in the germ cells a differentiated organism, which grows or unfolds during the course of development.

A homunculus inside a sperm, by Nicolaas Hartsoeker, 1695. This is not how development works.

Preformationism, though wrong, is frequently reinforced by the common (though badly mistaken) practice of referring to DNA or the genome as a “blueprint” for the organism. It is of course no such thing. A blueprint is a two dimensional representation of a three dimensional object. There is, in a blueprint, a scaled representation of all the parts of the object. We can tell, for example, that the window on the second floor is 4 m above and 2m to the left of the door. There is nothing like that in your DNA: there isn’t a gene for your left eye, which is a scaled distance away from the gene for your right eye. Your DNA (and your development) is much more akin to a recipe. In a raisin cake recipe, there isn’t a line in the recipe that says place a raisin 2 cm in from the upper left hand corner (there would be, if we had a blueprint for the cake). Rather, if you combine the right ingredients, in the right sequence, in the right environment, the result is a cake with raisins distributed through it at a certain density.

Richard Dawkins expounds the recipe analogy in The Blind Watchmaker (1986, pp. 295-296):

A recipe in a cookery book is not, in any sense, a blueprint for the cake that will finally emerge from the oven…. a recipe is not a scale model, not a description of a finished cake, not in any sense a point-for-point representation. It is a set of instructions which, if obeyed in the right order, will result in a cake.

Now, we don’t yet understand everything, or even most things, about how animals develop from fertilized eggs. Nevertheless, the indications are very strong that the genes are much more like a recipe than like a blueprint. Indeed, the recipe analogy is really rather a good one, while the blueprint analogy, although it is often unthinkingly used in elementary textbooks, especially recent ones, is wrong in almost every particular. Embryonic development is a process. It is an orderly sequence of events, like the procedure for making a cake…

The genes, taken together, can be seen as a set of instructions for carrying out a process, just as the words of a recipe, taken together, are a set of instructions for carrying out a process.

The reason this is important for students of evolution is that most of evolution is the modification of pre-existing structures, and these structures arise in the organism via a process of epigenetic development. Thus, most of evolution is the modification of pre-existing developmental programs.  Evolution doesn’t swap out one adult structure for another, but rather alters developmental programs, which results in differences in adults.  To understand phenotypic evolution, we must understand the variations which alterations of the developmental program can give rise to, their natures, and frequency. These studies are the domain of what has come to be called “evo-devo“. (From the evidence of at least vertebrate paleontology [this, this, and this], we can expand the generalization and say that most of evolution is the gradual, adaptive, modification of pre-existing developmental programs.)

I’ll finish this post with three brief observations. First, doesn’t it seem strange that the man who has provided the most compelling way of seeing the complex and interactive nature of development has been so frequently assailed for being reductionist and atomistic? Second, the fact that certain genes that have a major effect on development (Hox genes) are arranged along chromosomes in antero-posterior order of their influence on the developing body, while not preformationist, is nonetheless an intriguing and unexpected correspondence of the spatial arrangement of the body and the genes. And third, I’m using epigenetic in the original embryological/morphological sense, from which C.H. Waddington derived the term “epigenetics” in 1942, and not the recent odd usage, in which epigenetic means ‘heritable variation not associated with nucleic acid variation’ or, even more oddly, “all the weird and wonderful things that can’t be explained by genetics“; the hijacking of the word, and the conflation of the pseudo-neologism with Waddington’s ideas, have been nicely explicated in a paper by David Haig. Epigenesis is too useful a concept to lose the word for it.


Dawkins, R. 1986. The Blind Watchmaker. W.W. Norton, N.Y.

Haig, D. 2004. The (dual) origin of epigenetics. Cold Spring Harbor Symposia on Quantitative Biology 69:1-4.

48 thoughts on “Development is epigenetic

  1. Evolution doesn’t swap out one adult structure for another, but rather alters developmental programs, which results in differences in adults.

    You know…that’s rather obvious now that you point it out, yet it really hadn’t occurred to me before to think of it like that.

    It’s not that the proto-giraffes with longer necks survived better, but rather that the proto-giraffes had genes which resulted in embryonic development patterns that resulted in childhood development patterns that resulted in adults with longer necks that survived better.

    And, of course, even that’s an oversimplification….

    Thanks for this essay!


  2. Your final point is a great one, but I fear that the cat is already out of the bag. There have been whole reviews on epigenetics (and its role in evolution) that focus strictly on the “recent odd usage,” and I suspect that there are lots of young biologists for whom the word epigenetics conjures up images of specific cases of DNA methylation and its effects on offspring, and NOT the way that cells and tissues talk to each other during the complex unfolding of development in an organism with highly differentiated tissues. I DO agree that we need a word to encapsulate what is meant by epigenetics in the traditional usage. When I lecture about the intersection between evolution and development, I find that I have to use the word ‘epigenetics’ but I have to stop and explain the disparate uses in current biology.

    1. The head of my department when I was in grad school was Pere Alberch, one of the founders of modern evo-devo, and so the Waddingtonian concept was clear and well-known to me and my cohorts. I think the Waddingtonian concept still predominates among embryologists, systematists, and morphologists. For geneticists and developmental biologists it varies. More molecular types lean to the recent, odd usage, while evolutionary and morphogenetic types lean toward the Waddingtonian. (And to be clear, ‘epigenesis’ and ‘epigenetic’ long predate Waddington’s usage. He coined the noun form, ‘epigenetics’, but consistent with the earlier word’s meaning.)


      1. Just been reading about him in Mark Blumberg’s “Freaks of Nature”. He says (note on p.265 about JC’s rejoinder to Olivia Hudson on ‘Hopeful Monsters) “Coyne’s unwavering commitment to gradualism (in the traditional Darwinian sense) makes me uneasy. Regardless, by the end of such disputes , including that between Bateson and the Darwinians, it becomes clear that continuity and discontinuity are relative terms that are best set aside so we can focus on empirical findings.”

      2. I’m still not sure why you call usage that is more delimiting and practical “odd” but this molecular biologist favored usage (methylation and histone modification) will clearly reign supreme as it is routinely applied in biomedical research, the field(s) with all the money.

    2. There’s a somewhat parallel confusion with the word “gene”, which nowadays most people treat as synonymous with “protein-coding region of DNA”, whereas historically it has been a much broader word.

      In Dawkin’s The Selfish Gene, for example, he takes great care to explain precisely what he means by “gene”, and it’s not the predominant modern usage.

      I sometimes wonder if people who reject gene-centered selection are confused by this – that perhaps they learn that, in development, it’s largely not which proteins you make, but when and where you make them that’s important. If they assume that “gene” means the bits that encode proteins, then the idea that gene-centered selection is correct seems intuitively wrong. But the historical definition, and the one used by Dawkins, includes all regulatory non-coding DNA as well – indeed, any DNA which affects phenotype at all, along with DNA that affects only its own replication (such as meiotic drivers, which make sense only with gene-centered selection).

      Confusion over just what people are saying is part and parcel with an evolving language. I agree with Mayer that the purloining of well-established words for tangential or even incompatible concepts should be resisted. Not having biological training, I was not even aware of the older meaning of “epigenetics”, and it took a lot of contextual inference for me to figure out that he was definitely not talking about new and unimproved “epigenetics”.

      1. I agree with Mayer that the purloining of well-established words for tangential or even incompatible concepts should be resisted.

        I agree as well, from the other end of the progression; being of the ‘old school’ (or simply, old!) I’d been totally flummoxed to find that ‘epigenetics’ no longer seemed to mean what I thought it did.

        (Which brings up the issue of changing definitions causing older literature to become more and more incomprehensible with time…)

  3. This is not how development works.

    I love that caption’s dry understatement.

    CJBATM (chuckling just barely audibly to myself)

  4. Did they really know what sperm cells looked like back then, or is it just a really good coincidence?

    1. Hartsoeker was a Dutch astronomer and lens maker who is supposed to be the first person to view sperm under a microscope. The shape is pretty accurate; the little guy inside, not so much. More on Hartsoeker here and here; the latter link has a discussion of his embryological views (he was, of course, a preformationist).


      1. Even more on the whole 17th century egg and sperm thing and what they thought they were seeing in my book “Generation” (in the US) or “The Egg & Sperm Race” in the UK.

      2. Homunculus is such a good word – use it as an insult to ‘belittle’ someone! I am so horrid…

  5. I agree with the “recipe” description…but are you saying that the Hox genes are one exception to this and at least this gene cluster looks like a blue print?

    1. No, the Hox genes are not a blueprint. Like most other genes, they produce a gene product that interacts with all sorts of other gene products to produce morphological features. The first gene in the Hox cluster does not ‘resemble’ the anterior part of an organism, and later ones do not resemble more posterior parts. But that there is parallelism in spatial arrangement along the body and chromosomes is striking– and I don’t know why. It will be really interesting to find out.


      1. Thanks. Do you think there is some kind of base level of preformationism, in that the polarity of the fertilized egg which plays a role in development in certain organisms, appears to be pre-established in the reproductive cycle?

        1. I thought egg (embryo?) polarity was established in a variety of ways, including maternal hormone gradients in some embryos!? (And I believe your comment indicate so.)

          If so, wouldn’t preformationism immediately fall as explanation in any and all cases?

          1. Torbjorn,

            I think I agree with you but my question is …is there such a thing as an unpolarized egg….or pre-polarized egg?


            1. Think of it a symmetry breaking. The egg is pretty symmetric before fertilisation, but fertilisation itself breaks spherical symmetry- there’s a point where the sperm enters. Attachment so a surface also breaks symmetry.

      2. Studies by Iimura suggest how the sequential clustering of the Hox genes ends up corresponding to their anterior-posterior expressions, at least in mesodermal tissues. Cells in the primitive streak undergo a progressive derepression of the Hox genes starting at the Hox1 end. Some cells undergo an epithelial to mesodermal transition after only the Hox1 gene is expressed. These cells move anteriorly. This movement somehow prevents further derepression so that only Hox 1 is expressed. Other cells begin their anterior movement after both Hox1 and Hox 2 are expressed. They similarly cannot derepress Hox 3. Since they started movement later, they are posterior to the Hox1 only cells. Etc.

  6. Thanks Greg – I was never aware of such a silly idea that things were somehow preformed and just get bigger, though I always suspected that the religious anti-abortion folks must believe such nonsense to make claims that even a blastocyst is somehow a fully formed human.

  7. I can still remember the shock of insight from seeing a diagram of the epigenetic landscape in one of C.H. Waddington’s books when reading it over 40 years ago now.

    I am impressed to this day by the usefulness of the “landscape” concept in diverse fields of science. In chemistry the potential energy surfaces of molecules and reactions are a landscape representation. In evolutionary theory there are of course fitness landscapes and finally on the grandest scale the cosmological landscape of string theory.

    They are all of course higher dimension ed potential energy surfaces simplified to a 3-D representation in a 2-D projection. I wish the current evolutionary biologists would draw them the right way though, like physicists, chemists and the late great C. H. Waddingtion did. The vertical axis is potential energy and stable states are the lowest levels on this axis. So there are fitness valleys not fitness hills. A test “marble”, “ping-pong ball” atom or organism will tend to roll down to the lowest level in the metaphor of terrestrial gravity operating in a landscape.

    Great post a timely reminder of the importance of C.H. Waddington’s work

    1. I’m not sure energy goes through other than as analogy, even if an idealized population “ball” may strive after increased fitness. (For example, there is the problem of attaining an extrema in the first place.)

      But seeing the definition of absolute fitness resulting in possible values between 0 and infinity, it makes “infinitely” more sense to switch the ratio around. It would instead make the after/before sense odd. So YMMV.

  8. I’m loving the recipe analogy.
    In a recipe you have variation, when you want a table spoon of butter your most likely never going to use the exact same amount.
    Sometimes you make a mistake and use a teaspoon, or 2 table spoons.
    When you use a different ingredient as a replacement it’s hard to tell until the product is finished what changes all the different strangely correlated properties will incur.
    Recipes passed down generation to generation can incur heritable differences, slight changes that accumulate, or they could remain as identical as possible given that each baker or cook has slight quirks (like always using heaping table spoons and so forth).
    So many neat little comparisons to make. hmmm

  9. Im not as learned on all this stuff as most people here, so maybe Im just not seeing it yet, but I dont think recipe is a good analogy. A recipe is a set of instructions for building something, and someone must interpret and follow those directions. So who, in the case of genes, is reading the recipe and following the directions? (I can almost hear the religious nuts answer, “God!”). My understanding is that genes make copies of themselves, cells split into two, etc *by themselves*. So there are no instructions because there is no one to instruct. Thus genes and the cells that contain them are neither a blueprint nor a recipe, but merely a “situation” that results in replication and growth.

    1. Any analogy will break down eventually – the point of it is only to make things a bit easier to understand and to eliminate more common and more serious misunderstandings than it risks creating.

      Genes plus their cellular environment are indeed that situation. The recipe analogy is for the genes. To extend it to the genes plus their cellular environment, you’d have to make the RNA out as, say, idiot cooks that mindlessly follow recipes – recipes that are so simple in their basic instructions that a complicated molecule can be their chef.

  10. You know, I’m willing to bet that most of the authors who use the blueprint analogy have not even thought about what a blueprint is and is not. Indeed, I’m willing to bet a jelly-filled doughnut that a lot of them have never even seen a blueprint.

    What we have here may not be so much a bad analogy (“DNA is like a blueprint”) as it is that the meaning of “blueprint” has been extended to encompass all sorts of things that aren’t much like a real blueprint (if they are at all).

    Or, to put this another way, if you took an actual blueprint and put it in front of one of the offending authors and said “now explain to me just how DNA is like this diagram”, you would probably get a lot of confusion, hand waving, and backpedaling. They’ve simply been lazy and borrowed a bad analogy without ever thinking through what they’re actually saying.

    It’s good to have the falsity of the blueprint-DNA analogy pointed out so clearly. Maybe this will lead to better text books sometime in, say, the next million years.

    1. Ever since I read RD’s recipe analogy I have been surprised how rarely it gets used compared with ‘blueprint’, particularly by experts who should know better.

  11. The word ‘evolution’ was formerly used as a synonym for preformation and stayed connected to embryology for a long time.

    James T. Costa’s Annotated Origin has a note on the last sentence of The Origin of Species that makes a nice connection with Prof. Mayer’s post:

    “[T]he very last word is the only use of the word ‘evolve’ or its cognates in the book — ironic, given that ‘evolution’ is now synonymous with Darwin’s model of common descent by natural selection. In his day the word was more closely associated with embryological development, and indeed Darwin’s usage in this last sentence may be invoking an image of the embryo’s unfolding developmental complexity, as natural selection endlessly spins out those forms most beautiful and wonderful.”

    1. A great edition, that annotated version. Darwin devotes only a part of Chapter 13 to embryology however though it was clearly important in his view.

  12. It is probably worth mentioning that according to some muslims, “preformationism” is refuted in koran, and they consider this a “scientific miracle”. This is based on the following Koranic statement: “we created bones, and covered
    them with muscles”. Even if a refutation of preformationism, it is still wrong-bones and muscles develop simultaneously, not in tandem-but that’s the claim anyway.

  13. Great post. I’ve come across the blueprint analogy quite a few times but I never questioned its appropriateness or gave much thought to what a blueprint really is. Thanks for the enlightenment.

  14. I took to the recipe analogy as “so bloody obvious” when I stumbled on it, and subsequently realized that the “blueprint” terminology that I had unthinkingly used was terribly confusing.

    That Dawkins originated this I didn’t know, and so lately, so good for him and thanks for the information!

    1. Though now I remember that this was one of those few cases where I actually figured it out before hearing it elsewhere. Because I was very proud at the time when I realized it was “out there”.

      Dawkins, huh? Wow.

  15. Sadly, I think the battle over the word epigenetics may already have been lost.

    I quite literally was befuddled beyond understanding when I first started researching epigenetics.

    It may be time to coin an entirely different word to describe BOTH usages. Because as it currently stands, the available one serves neither cause.

  16. I was thinking about the recipe metaphor in relation to bread making, which I do a couple of times a week. With bread, you have to do things in the right order, and minor variations still produce bread: use olive oil instead of butter and you get a different texture (‘Italian bread’), leave out the fat altogether and you get again a different texture (‘French bread’) that will go stale quickly. Totally mechanical and predictable: add these ingredients in the right order at the right time, perform some mechanical effects (mixing kneading dividing heating) again at the right times, and the result is a loaf of bread.
    Then surprise surprise I found there is a huge industry comparing breadmaking to the ‘spiritual life’ e g;
    quote: “The measuring, the kneading and rising, baking and eating, are reflections of the holy process in which God forms us, Peters said.” Is nothing sacred?

  17. Isn’t the recipe metaphor just another simplification of a highly dynamic and complex process? Development isn’t a purely genetic process. The suggestion that you can build an organism simply from genetics is laughable. Or, put another way: How does the recipe analogy explain polyphenism?

    1. Perhaps we’d see polyphenism in a batch of cookies if the oven temperature were uneven? Some cookies might be dark and crispy, others light and chewy.

      1. Yes! But Dawkins never says anything about how the temperature variations, etc. of this magical oven might play a role in development. His emphasis is on “instructions (emphasis) which, if obeyed in the right order, will result in a cake,” and that genes are these instructions. Follow instructions and-presto-cake. Or: genes encode development. This analogy, while certainly an improvement over the blueprint model, falls flat when we consider that any phenotype is the product of both genetics and the environment.

        If I grow a frog embryo in a solution containing mercury or retinoic acid or ethanol, it’s not going to exhibit anything close to the range of phenotypes that we might call ‘normal.’ The ‘instructions’ are still intact, but we end up with something different. If we’re talking in terms of cake, Dawkins’ analogy leads us to believe that a cake baked at the North Pole, in a jungle, or underwater will all be the same, just as long as you follow the directions.

        1. I don’t have Dawkins’ texts to hand, so I’ll quote myself: “if you combine the right ingredients, in the right sequence, in the right environment…” (emphasis added). The right environment for a cake is an oven; the right environment for an embryo will vary, but will not be a bath of retinoic acid. There’s nothing in the recipe metaphor to imply the environment is irrelevant; in fact, precisely the opposite, recipes (and development) depend crucially on the environmental situation of the ingredients.


Leave a Reply