Nelson: Developmental Systems Drift

Anyone else interested in developmental systems drift? What it means from a design perspective? Or any other angle?

How prevalent is it compared to developmental system non-drift? I think of Pax-6, for example.

1 Like

Great question, it’s exactly what I would explore in a real conversation about DSD. Pax6 and other remarkable examples of deep homology are in some sense the very opposite of DSD and yet both seem present in phylogeny.

Consider also the conservation of Hox expression in linear sequence both in the genome and the embryo.

1 Like

I still remember the day I read that paper. We were all amazed and excited and thoroughly astonished.

DSD and deep homology should be considered together, and seriously this would make a great topic for a review article or a theme for at least part of a book on design and evolution. Dibs. :wink:

It seems to me that on a general principle borrowed from the game of Jenga(h/t @T_aquaticus), you would predict that more ancient developmental pathways are in general also more resistant to developmental systems drift. Things that get established first have things added on top of them(sort of like the law of superposition), and so over time become more and more resistant to change without destabilizing the whole system added on top.

1 Like

If he’s not a coauthor, I nominate @pnelson for the role of Reviewer 3 :grinning:


As noted above, I think the slide is more than defensible, in context.

But given your reaction, and the fact that online pdfs don’t come with their instructors attached to explain the relevant context, I deleted it. Here is the amended slide deck, with a handful of other corrections and changes:

My reaction to your “I want a real conversation” comments is sorrow (really). See Amos 3:3, which definitely applies here. I am interested in experimental data showing how developmental pathways in animals can be viably and heritably modified. You say that’s unscholarly. OK. I am also interested in how the common descent of the Metazoa can be tested by incongruent patterns. You say common descent is settled science and thus beyond discussion. OK.

That leaves me nothing of interest in the conversation. If I wanted storytelling about deep homology, DSD, and all that, I can get plenty of it at SDB meetings and in the literature.

1 Like

I see the logic. Should be testable.


That is, of course, not what I wrote or what I meant.

Nah, go ahead. Let’s see how common descent can be tested by incongruent patterns. We will, of course, use full descriptions of what common descent predicts, what it doesn’t predict, etc.

That’s what we call science. I think you are politely excusing yourself from science, and I respect that.


My question would be what level of differences is development is deemed “too hard” for evolution to accomplish? Is there a definitive cutoff that you’ve determined a priori, or would it be defined by the level of differences observed between organisms that you’d agree are related by common descent?

In other words, is it in principal possible to say to you “look at these two sister species, they have X large difference in development, but since you agree they’re related then this difference must have been achieved naturally by evolution”, and that would satisfy you?


What are positive and negative controls we can agree upon? If we can’t find any, I suspect this discussion reveals some deeper problems that cannot be bridged.

1 Like

I find your opening slides to be resting on a fundamental problem with how you see theories in science. Why is the theory of evolution supposed to explain how development works? It’s a theory of transgenerational change of populations of organisms, it’s not supposed to explain why some particular developmental pathway works the way it does, it only explains what are the mechanisms (or forces) that cause such developmental pathways to change over generations.

You could go deeper and look at why one protein binds to another, and you’d have to understand that at the level of physical chemistry. Again the theory of evolution explains how proteins change over generations, but other theories (those in biochemistry and protein folding etc.) explain why some set of proteins have the interactions they do, and the emergent properties that result from this. And deeper still there are theories of the physics of atoms and their interactions that explain how come there is such a thing as a molecule. Is evolution supposed to explain that too?

The germ-theory of disease doesn’t explain in what way infectious microorganisms cause disease states. In a way it really just started out by noting a correlation and doing some simple experiments that showed that once some host became infected, it got sick. But it’s not supposed to explain the how. And it would be silly to fault it as being somehow wrong or incomplete for that reason. There are explanations for that, in biochemistry, and in cell biology, and so on. Does that make the germ-theory of disease wrong or useless? Obviously not.

This is why I honestly don’t understand some of these third-wayers and their frankly absurd demand to have the theory of evolution be an explanation for apparently literally everything that happens in biology. It’s ridiculous.


Hi Rum,

You can find discussion of these issues long before the “third-wayers” came along, and by scientists who would never consent to being listed on the Third Way roster:

The central question isn’t explaining “literally everything that happens in biology,” but simply understanding, at least for animals and higher plants, how genetic changes (i.e., the whole panoply of perturbations to DNA, from point mutations through chromosomal rearrangements) can transform phenotypes, via development.

Development must be included in evolutionary dynamics for animals and higher plants, because the process comprises its own set of rules about what DNA changes will be tolerated, versus not, and thus what phenotypes are accessible (versus not) to macroevolution.

Douglas Erwin, for instance, an invertebrate paleontologist at the Smithsonian, who worked closely on macroevolutionary questions with developmental biologist Eric Davidson (d. 2105) during the last two decades of the latter’s life, is not a Third Way advocate. But he is also not a neo-Darwinian, in the sense of being reasonably happy with textbook theory (cf. Brian Charlesworth, Douglas Futuyma, or Russ Lande; the last two attended the Royal Society Extended Synthesis meeting in London in 2016, sat in the front row, and debunked every speaker who took the podium). But Erwin argues strongly that something is very wrong with the theory we inherited from Mayr, Dobzhansky, and crew. In this 2011 open access paper, for example, he posits that evolutionary processes are profoundly non-uniformitarian – Mayr and Dobzhansky would have jumped out of their seats in indignation – such that the genetic and developmental changes which caused the Cambrian Explosion have no modern analogues:

I would also recommend the work of UK developmental geneticist Wallace Arthur. Not a Third Way advocate by a long shot. This book, in particular:

Hi Paul, the link to your Building-Animals-isHard0050720.pdf seems broken. Would you be able to post the link again please? Thanks.

Try this:

Yes, it says “Aminals.” :frowning: For some reason this file has been a non-stop carnival ride.

It all seems like naive falsificationism to me. We expect consrvation of development from common descent. Here are a couple cases of non-conservation. Therefore common descent is falsified. Incidentally, the distribution of bicoid seems a pretty good fit to the tree: one origin of bicoid and losses in two places.


Many thanks. I shall read with interest :slightly_smiling_face:

How do you think this change could occur?

I can think of a number of simple mechanisms. Have you tried?

Here’s one. Suppose that originally gene B (for bicoid) sets the initial body axis. Another gene, call it C (for the next letter) evolves that helps B in some way. Gradually, C becomes a better and better helper, to the extent that it can set the initial axis even if B doesn’t. Then if B gets a disabling mutation, development continues as usual, and B is lost. Now we have a system in which C, not B, is present.