Fish are not inside-out insects

It’s a shame I have to say that right from the beginning.

I’m beginning to develop a distaste for computer models of biological processes, which is a shame. From Andrulis to Fleury to Pivar, the field is tainted with people who don’t know a lick of biology but are good at inventing algorithms that go spinning off into never-never land, spawning odd and suggestive shapes that, in their happy ignorance, they assign to real organisms.

The latest is a guy named Eric Werner who, in a surprising change of pace, has a model that does not involve whorls, spirals, vortices, or toroids, the usual objects of crackpot obsessions, and is instead about inversions. He has a model for the production of skeletons, which is his, and which explains both exoskeletons and endoskeletons. “Exo”, “endo”, get it? All you have to do to turn one into the other is to turn the animal inside out.

That’s the other thing about these kinds of modelers — it’s got to be a really simple transformation that does the job all in one step.

An intriguing unanswered question about the evolution of bilateral animals with internal skeletons is how an internal skeleton evolved in the first place. Computational modeling of the development of bilateral symmetric organisms suggests an answer to this question. Our hypothesis is that an internal skeleton may have evolved from a bilaterally symmetric ancestor with an external skeleton. By growing the organism inside-out an external skeleton becomes an internal skeleton. Our hypothesis is supported by a computational theory of bilateral symmetry that allows us to model and simulate this process. Inside-out development is achieved by an orientation switch. Given the development of two bilateral founder cells that generate a bilateral organism, a mutation that reverses the internal mirror orientation of those bilateral founder cells leads to inside-out development. The new orientation is epigenetically inherited by all progeny. A key insight is that each cell contained in the newly evolved organism with the internal skeleton develops according to the very same downstream developmental control network that directs the development of its exoskeletal ancestor. The networks and their genomes are are identical, but the interpretation is different because of the cell’s inverted orientation. The result is inside-out bilateral symmetric development generating an inside-out organism with an internal skeleton.

My first thought was…an interesting suggestion. We know from the homology of the patterning molecules involved that vertebrates and invertebrates are upside-down relative to each other, so at some point an ancestor flipped (or more likely, the ancestor was morphologically ambiguous in the dorsal-ventral axis), so let’s think about whether that’s feasible. And then my second thought was…wait, no way. That makes no sense at all.

So I read the paper. I was right, it makes no sense at all.

First thing I noticed was that the acknowledgements thank Francis Hitching, a notorious crank, Martin Brasier (no problem there, he’s a paleontologist specializing in Cambrian evolution…but also not a developmental biologist), and Cellnomica, a company that makes the modeling software. I looked. Eric Werner is the president and CEO of Cellnomica, which sort of means he was happily thanking himself for allowing him to use his software, which is nice, I suppose.

But except for Brasier, I’m already unimpressed. That doesn’t matter, though; he will sway me by the data and the evidence, right?

Next problem: there isn’t any. This is one of those totally evidence-free papers; the author didn’t bother to look up anything about the induction of skeletal elements in arthropods and vertebrates, cites nothing but three (!) papers all written by Eric Werner (!!), all published unreviewed in arXiv (!!!), and builds everything from a simplistic premise about how axis information is inherited epigenetically in dividing cells. There actually is a substantial literature on the inheritance of the orientation of cytoskeletal elements in dividing cells in flies and nematodes, for instance — but it’s not as trivial as what Werner proposes, and he doesn’t cite any of it, anyway.

The whole thing consists of the graphical output of simulation runs on his software, like this:

Orientation transform results in inside-out growth. The bilateral multicellular organism on the left Fig.a, where the red cells are on the outside, develops from the founder cells next to its right in Fig.b that have a Back-to-Back orientation. The reversal from a Back-to-Back orientation to a Face-to-Face orientation in the founder cells Fig.b is epigenetically inherited in the progeny. The result of this orientation switch is inside-out development seen in the organism on the right Fig.c where the red cells are on the inside. The transformation is reversible going in either direction depending only on the initial orientation of the bilateral founder cells.

He hasn’t even questioned his premises. Is there evidence of cells producing mirror-image progeny (actually, I recall that there is…but it’s not quite as uniform as he proposes)? Is there reason to think from, say evidence in the fossil record, that ancient chordates are inside-out arthropods? Nope, and Brasier should have been able to tell him so. Is there developmental evidence that this is how skeletons form? For instance, are the progenitors of internal skeletons homologous to the cells of the arthropod cuticle that produce there exoskeleton?

And that’s where I stopped and told myself, “obviously not”. The arthropod cuticle is produced by ectodermal cells that produce chitin. If you turned one literally inside out, the ectoderm would become the endoderm, the lining of the gut…a phylogenetically ancient tissue with homology between arthropods and chordates. It does not produce the chordate skeleton. That job is done by mesodermal derivatives…a tissue that forms in roughly similar ways by ingression/involution of cells during gastrulation. Mesoderm forms muscle and connective tissue in both arthropods and chordates, and produces bone and cartilage in addition in chordates.

The story is complete abiological and ahistorical bollocks. It’s really nothing but a pointless exercise in making a computer program run its paces, and is about as relevant to evolution as Spore, another game that attempted to model a science and failed abysmally.

It also makes me curious about something else: it was published in arXiv, which is mainly a repository of physics papers, with some abstract biological/mathematical stuff trickling in. Is the physics collection as plagued with drivel as the few samples of biology papers I’ve seen shoveled in there?

Werner E (2012) How to Grow an Organism Inside-Out: Evolution of an internal skeleton from an external skeleton in bilateral organisms. arXiv:1207.3624v1

(via Tommy Leung)


  1. #1 abadidea
    August 29, 2012

    arXiv has no barriers by design, so yes you can find crackpottery in every field if you look. I hear it is a nice graveyard for flawed P = NP proofs.

    On arXiv one should check the citations and use common sense, just as you have.

  2. #2 Michael Kelsey
    SLAC National Accelerator Laboratory
    August 29, 2012

    As abadidea noted, arXiv has few barriers. There is some minor moderation, including a requirement that uploading to a specialty area requires co-sponsorship by an “endorser” (an author registered by the arXiv moderators as legitimate).

    Uploading to the arXiv is *NOT* publication, and really should not be labelled as such. It is a repository, equivalent to a public FTP server (remember those, you old timers?), where an author can put his final draft for colleagues to read before it comes out in a journal. That was the original intent.

    Obviously, arXiv has broadened it’s effect substantially beyond just a way to distribute preprints through a one-stop shop. But it should never be considered “publication” (except by those crackpots who also consider their personal Web sites as “publication”).

  3. #3 chris
    August 30, 2012

    “What happens to the gut when you turn the animal inside out” was pretty much my first thought, and he doesn’t even bother trying to answer it? That’s not very impressive.

  4. #4 Eric Lund
    August 30, 2012

    One thing to add regarding the arXiv is that physics (unlike some other fields, so I hear) has a longstanding tradition of freely sharing preprints. The arXiv lets you do that without paying postage or photocopying costs, and conversely you can read anybody’s preprint without being on a preferred distribution list. Papers in the arXiv are indexed by the ADS abstract database and are considered citable.

    However, it is very much a caveat lector environment. In my grad school days, cranks still occasionally sent manuscripts to real physics professors. Today they have the arXiv to do it. (There are also actual journals, most notoriously Physics Essays, which sometimes publish crank papers.) Certainly, when the only citations are to other arXiv manuscripts by the same author, that raises a red flag.

  5. #5 David Marjanović
    Museum für Naturkunde, Berlin
    August 30, 2012

    We know from the homology of the patterning molecules involved that vertebrates and invertebrates are upside-down relative to each other, so at some point an ancestor flipped (or more likely, the ancestor was morphologically ambiguous in the dorsal-ventral axis)


    PZ, didn’t you read the comments of the last thread about this?

    First of all, it’s nothing short of sloppy to talk about “invertebrates” in this context. You mean arthropods. (If not specifically Sophophora melanogaster.)

    In mollusks, the genes that do dorsal/ventral patterning in arthropods and vertebrates don’t do that; they’re not expressed only dorsally or ventrally.

    Combined with a few other lines of evidence*, the most parsimonious conclusion by far is that nothing has flipped; arthropods and vertebrates have undergone some evolution in their patterning of the dorsoventral axis, and they happen to have done this in opposite ways.

    * Like… vertebrates have 1 dorsal nerve cord, and arthropods have 2 ventral ones; but enteropneusts have 1 dorsal and 1 ventral one, mollusks have 2 ventral and 2 ventrolateral ones, and flatworms have 8 in total, arranged in an octogon: 2 ventral, 2 ventrolateral, 2 dorsolateral, 2 dorsal.

  6. #6 Richard M. Meyerson
    August 30, 2012

    (produce there exoskeleton?)?? should read “produce their exoskeleton?”

  7. #7 owlbolix
    September 5, 2012

    Computer models?
    If they work for climate science they should also work for biological processes.
    Oh, wait…

  8. #8 David Marjanović
    Museum für Naturkunde, Berlin
    September 6, 2012

    They do, if they’re not as ridiculously simplistic as in this case.

New comments have been temporarily disabled. Please check back soon.