Nelson: Developmental Systems Drift

The observation that homologous structures sometimes develop via non-equivalent developmental paths is only an argument against evolution/common descent if you assume that developmental pathways can’t independently evolve in groups following their divergence with other groups. I don’t see why that’s an assumption worth taking seriously.

Indeed. Why worry about it? At Churchill College, Cambridge, last April, Joanna Masel spent a long time during a poster session talking with me about “developmental systems drift” (DSD), which she said was pervasive throughout the animals. Masel said one of her colleagues, just down the hall from her at Arizona, spent a lot of his time collecting examples of DSD, in which homologous adult phenotypes arise via non-homologous developmental pathways.

I listened politely, but wanted to ask her if she had any experimental data in animals showing DSD arising via mutations. I didn’t, because I already knew the answer.

If you open a good comparative anatomy textbook from, say, the late 1970s – Romer & Parsons is an excellent example – you’ll read something like this:

What, then, are the best criteria for the establishment of homology?..Best of all is similarity in developmental history. Embryologic processes in vertebrates tend to be conservative, and organs which are very different in the adult condition may reveal their homology through similarity in embryonic stages. (1977, p. 10)

So morphologies are homologous when they arise via similar developmental pathways, and homologous when they arise via dissimilar pathways. DSD. No reason to worry.

To paraphrase Brian Hall, homology is a statement about pattern, not process. If want to believe that every minute change to developmental pathways can only be achieved by intelligent design, go right ahead, but you shouldn’t suggest (as you did in the video) that evolutionary biologists should expect developmental pathways to be immutable. That makes absolutely no sense.


It’s only a problem if the alternative theory can prove a negative? Is there any evidence that there is developmental variation in animals where common ancestry has been independently verified?

Differing developmental pathways are only a problem if you assume that such pathways can’t evolve. If you assume that, then you already have a problem with evolution, don’t you? Developmental pathways have always been proposed to evolve, so the fact that some homologous structures have different developmental pathways is isn’t some kind of new or distinct “problem”.

What do you mean by “where common descent has been independently verified?” What kind of evidence are you imagining? A fly breeding experiment where an individual manages to development an entirely new developmental pathway in a few months?

1 Like

I would ask a simple question like are the development pathways of all types of mice the same or are their differences?

Why mice? What makes them more “independently verified” to be related than, say, humans and chimps? Flies and beetles? Frogs and sharks?

1 Like

Mice have lots of different species and around 5 generations per year. If they will show that variable developmental paths exist then the case begins to get built that through reproduction and isolation that different developmental paths can be built. I have reasonable confidence that all mice species share a common ancestor and their differences are based on reproductive isolation. @pnelson would you agree with me on this point?

I think I might ask her. I’m not sure, but she might respond they way I would, which is: do you mean convergence?

I’m not aware of anyone investigating the different developmental pathways in different species of mouse. I have no doubt though, that if differences were found, IDers would invoke ID. There is evidence of drift between closely-related species of fruit flies: D. melanogster and D. simulans (Takano, 1998; Takano-Shimizu, 2000), is that good enough?


Hi Steve,

If you don’t mind waiting until Monday (4/27), I’ll indicate where – in the lecture pdf I’ll be posting – the relevant issue with DSD, and evidence for DSD, arises. Convergence is not the issue.

Thanks for your patience. I have a MS revision for Michael Murray to finish today.

Where does this confidence come from? And what, technically, do you mean by “mice”? Mus? Cricetidae? Sigmodontidae? Rodentia?

Works for me. I will be interested to see how you discuss Ralf Sommer’s work in nematodes (we’ve published a couple papers from his group) and the very nice related work of Andrew Fraser on the more familiar Caenorhabditis nematodes. You might not have had the time to see a new review article on DSD; I just found it in the “cited by” section of PubMed for the Fraser group’s nice PLoS Genetics paper but I include it below.

You’re right that convergence is “not the issue,” but IMO it would be a big mistake to ignore it as a huge influence on any consideration of DSD. Because it seems to me that the common theme is that phenotypes, including morphs, can arise through divergent mechanisms. This implies that the mechanism themselves are not coupled in any concrete conceptual way to the phenotype. This is a fact that has fascinated me for more than a decade, and it’s illustrated by thoughts I have had about deep homology: e.g., there is no interesting connection between Notch signaling and neurogenesis, other than the fact that Notch signaling is almost universally involved in neurogenesis. If that’s true, then the systems (Notch signaling) and the outcomes (neurogenesis) are not tightly coupled biologically/mechanistically, and thus the outcome can be constrained while the control system itself drifts around. Hence DSD.

I’ll be interested to see how you discuss those papers and those concepts!


Still hoping to see comments/discussion on DSD. It seems an area of minor activity given the recent review article.

Messy life interrupted here, but I think I’ve beaten it back, at least temporarily. DSD discussion in the monster slide deck coming, as soon as I can find a Discovery staffer in Seattle who has access to their server (for posting the pdf). My apologies for the delay.

1 Like

Nah, it’s okay. And if it’s just a few slides from a talk, it might not be useful for discussion. If you don’t have notes or anything written, then are you sure we’ll be able to have a scientific conversation about this topic?

From a paper discussed by @pnelson on EN:

6.2. Distinct states may be equivalent

Selection can impose external constraints on the form and function of a living system. For example, the environment of the system in the toy model might require ‘cells’ with consistent fonts and case for survival. This would result in the survival of cells with ‘ABC’, ‘abc’, etc., as the values for each of the three entities (electronic supplementary material, figure S3). However, because there are two possible cases (upper versus lower), there would be two distinguishable states that are effectively equivalent for survival in this environment. Such situations could result in unregulated redundancy such that similar functions are performed by different molecules in random sets of cells [33]. Over evolutionary timescales, this type of unregulated redundancy could result in organisms with similar form and function but different underlying molecular mechanisms [34].

So @pnelson, which side of the “its all about DNA” debate do you fall on? If I get your objections to Masel correct, is it safe to think that you are more in the “DNA is everything” camp?

1 Like

I’m intrigued by Jose’s proposals (in that paper), but the challenge for me is his phrase “over evolutionary timescales.” See my monster slide deck, on its way here, Deo volente later today or in the AM tomorrow.

I’m definitely NOT in the “DNA is everything” camp, anyway.

May be of interest:

Article is linked below. It’s excellent so far, I’m about 1/2 through, must get back to day job.

1 Like