Our neural spine bifurcation paper is out

March 15, 2013

Wedel and Taylor 2013 bifurcation Figure 4 - classes of bifurcation

Figure 4. Cervical vertebrae of Camarasaurus supremus AMNH 5761 cervical series 1 in anterior view, showing different degrees of bifurcation of the neural spine. Modified from Osborn & Mook (1921: plate 67).

Today sees the publication of my big paper with Mike on neural spine bifurcation, which has been in the works since last April. It’s a free download here, and as usual we put the hi-res figures and other supporting info on a sidebar page.

Navel-gazing about the publication process

This paper is a departure for us, for several reasons.

For one thing, it’s a beast: a little over 13,000 words, not counting tables, figure captions, and the bibliography. I was all geared up to talk about how it’s my longest paper after the second Sauroposeidon paper (Wedel et al. 2000), but that’s not true. It’s my longest paper, period (13192 vs 12526 words), and the one with the most figures (25 vs 22).

It’s the first time we’ve written the paper in the open, on the blog, and then repackaged it for submission to a journal. I have several things to say about that. First, it was more work than I expected. It turns out that I definitely do have at least two “voices” as a writer, and the informal voice I used for the initial run of blog posts (linked here) was not going to cut it for formal publication. So although there is very little new material in the paper that was not in the blog posts, a lot of the prose is new because I had to rewrite almost the whole thing.

I have mixed feelings about this. On one hand, last May kinda sucked, because just about every minute that wasn’t spent eclipse chasing was spent rewriting the paper. On the other hand, as Mike has repeatedly pointed out to me, it was a pretty fast way to generate a big paper quickly, even with the rewriting. It was just over two months from the first post in the destined-to-become-a-paper series on April 5, to submission on June 14 (not June 24 as it says on the last page of the PDF), and if you leave out the 10 days in late May that I was galavanting around Arizona, the actual time spent working on the paper was a bit under two months. It would be nice to be that productive all the time (it helped that we were basically mining everything from previously published work; truly novel work usually needs more time to get up and going).

Wedel and Taylor 2013 bifurcation Figure 18 - Barosaurus and Supersaurus cervicals

Figure 18. Middle cervical vertebrae of Barosaurus AMNH 6341 (top) and Supersaurus BYU 9024 (bottom) in left lateral view, scaled to the same centrum length. The actual centrum lengths are 850 mm and 1380 mm, respectively. BYU 9024 is the longest single vertebra of any known animal.

You may fairly wonder why, if almost all the content was already available on the blog, we went to the trouble of publishing it in a journal. Especially in light of sentiments like this. For my part, it’s down to two things. First, to paraphrase C.S. Lewis, what I wrote in that post was a yell, not a thought. I never intended to stop publishing in journals, I was just frustrated that traditional journals do so many stupid things that actually hurt science, like rejecting papers because of anticipated sexiness or for other BS reasons, not publishing peer reviews, etc. Happily, now there are better options.

Second, although in a sane world the quality of an argument or hypothesis would matter more than its mode of distribution, that’s not the world we live in. We’re happy enough to cite blog posts, etc. (they’re better than pers. comms., at least), but not everyone is, and the minimum bound of What Counts is controlled by people at the other end of the Overton window. So, bottom line, people are at least theoretically free to ignore stuff that is only published on blogs or other informal venues (DML, forums, etc.). If you want to force someone to engage with your ideas, you have to publish them in journals (for now). So we did.

Finally, ever since Darren’s azhdarchids-were-storks post got turned into a paper, it has bothered me that there is an icon for “Blogging on Peer-Reviewed Research” (from ResearchBlogging.org), but not one (that I know of) for “Blogging Into Peer-Reviewed Research”. If you have some graphic design chops and 10 minutes to kill, you could do the world a favor by creating one.

Hey, you! Want a project?

One of the few things in the paper that is not in any of the blog posts is the table summarizing the skeletal fusions in a bunch of famous sauropod specimens, to show how little consistency there is:

Wedel and Taylor 2013 NSB Table 1 - sauropod skeletal fusions

(Yes, we know that table legends typically go above, not below; this is just how they roll at PJVP.)

I want this to not get overlooked just because it’s in a long paper on neural spine bifurcation; as far as I’m concerned, it’s the most important part of the paper. I didn’t know that these potential ontogenetic indicators were all mutually contradictory across taxa before I started this project. Not only is the order of skeletal fusions inconsistent among taxa, but it might also be inconsistent among individuals or populations, or at least that’s what the variation among the different specimens of Apatosaurus suggests.

This problem cries out for more attention. As we say at the end of the paper:

To some extent the field of sauropod paleobiology suffers from ‘monograph tunnel vision’, in which our knowledge of most taxa is derived from a handful of specimens described decades ago (e.g. Diplodocus carnegii CM 84/94). Recent work by McIntosh (2005), Upchurch et al. (2005), and Harris (2006a, b, c, 2007) is a welcome antidote to this malady, but several of the taxa discussed herein are represented by many more specimens that have not been adequately described or assessed. A comprehensive program to document skeletal fusions and body size in all known specimens of, say, Camarasaurus, or Diplodocus, could be undertaken for relatively little cost (other than travel expenses, and even these could be offset through collaboration) and would add immeasurably to our knowledge of sauropod ontogeny.

So if you’re looking for a project on sauropod paleobiology and you can get around to a bunch of museums*, here’s work that needs doing. Also, you’ll probably make lots of other publishable observations along the way.

* The more the better, but for Morrison taxa I would say minimally: Yale, AMNH, Carnegie, Cleveland Museum of Natural History, Field Museum, Dinosaur National Monument, BYU, University of Utah, and University of Wyoming, plus Smithsonian, University of Kansas, OMNH, Denver Museum, Wyoming Dinosaur Center, and a few others if you can swing it. Oh, and Diplodocus hayi down in Houston. Check John Foster’s and Jack McIntosh’s publications for lists of specimens–there are a LOT more out there than most people are familiar with.

References

About these ads

35 Responses to “Our neural spine bifurcation paper is out”

  1. Casey Says:

    Cool stuff (well, not the bifid neural spines part–a bit niche ;) ). But the blog-to-paper phenomenon is interesting. I just recently ripped off my own blog post from a few years ago to use as an intro to paper (which I swear I’ll have the revisions turned back this weekend!) and was also slightly dinged on the conversational manner of the prose (its a paper cut from a different cloth as well, but we’ll talk about that some other time).

    But I was thinking…why can’t a paper read like a talk, or sound like it was me actually presenting the paper to you (without the cursing). What I have always liked most about Lovecraft is that you can read the words out loud, and the material is even more eerie and scary—campfire material at its best. So, if you could read my words, reading the way I might present something (assuming its presented well)…isn’t that also an effective means of communicating science (particularly anatomy)? This isn’t anything new, just something I’ve been contending with today. Congrats on the paper!

    Finally might I suggest reposting the Figures here or somewhere?, because the pdf has awfully-super fuzzy figures of supposedly gorgeous sauropod vertebrae.


  2. So far, favorite figure is this one, fig. 9. I prefer diagrammatic figures to help express ideas than figures of fossils with interpretive text. I would love to see a size-based correlation for this figure, along with all taxa known with bifidy (there’s those titanosauriforms in the mix; would be interesting to explore that angle).

  3. Matt Wedel Says:

    Quick hits, before I get on to the meat:

    Casey, the full-size figures are already posted here–as noted in the second sentence of the post. :-)

    Jaime, we like that figure, too. Stand by for a post by Mike explaining how it came it be. I had not thought of trying to depict size in the diagram, but it’s a worthwhile idea, especially for the camarasaurs. And, yeah, a global look at neural spine bifurcation in sauropods is needed–maybe someone will use this a springboard to go do that.

    Thanks to you both for your kind words.

  4. Matt Wedel Says:

    Now, Casey wrote:

    I just recently ripped off my own blog post from a few years ago to use as an intro to paper and was also slightly dinged on the conversational manner of the prose

    First: yay! More blogging-into-peer-reviewed-research. This really needs to be a Thing.

    Five years ago I wrote this:

    Write as informally as you think you can get away with (i.e., without sacrificing the science). Your papers will make better reading and more people will read them. There’s a reason people still read The Origin for pleasure. Here’s a simple heuristic: if reviewers aren’t criticizing you for being informal, your prose is too stilted. Reviewers frequently knock me for being too informal, and that helps me find the balance. YMMV.

    I still mostly believe all that, although there is a difference between writing accessibly and writing with no net at all. That’s what I was trying to overcome in reworking the prose for this paper. It’s an interesting problem and one I’d like to see discussed more.

    But I was thinking…why can’t a paper read like a talk, or sound like it was me actually presenting the paper to you (without the cursing). What I have always liked most about Lovecraft is that you can read the words out loud, and the material is even more eerie and scary—campfire material at its best. So, if you could read my words, reading the way I might present something (assuming its presented well)…isn’t that also an effective means of communicating science (particularly anatomy)?

    That is an awesome question, and it immediately makes me think this: why, in 2013, does a paper have to be a paper? Why couldn’t it be a video? What if there was an outlet that would let you post a video in which you talk through all your research, with linked hi-res figures and a written version that could be translated to Braille, ASL, etc., for people with impaired vision or hearing? Why does something like that have to be any less real and citation-worthy than a traditional paper?

    What if all of our conference talks were archived on the web and had DOIs or equivalent so we could cite them?

    That’s a disturbing thought. Not because I don’t want it to happen–I think it would be awesome–but because it’s the first time I’ve had to think of PeerJ as being old-fashioned.


  5. I like Chris Bennett’s tone: he’s walking the reader through his process, conversational with technical expertise. I’m not allowed to get away with it, and I’m not sure I could effectively pull it off. Definitely not even a casual reference to Madonna when discussing anoles, O no.

  6. Andy Farke Says:

    Congrats on getting this out! The paper looks great, and I can’t wait to read it in more depth. Two quick questions:

    1) Any particular reason for not putting scale bars on most of the images? I found it a little odd to read through and see references to centrum length, but not a corresponding scale bar. I guess there may be some issues of parallax for structures so big and wide. . .was this in your thoughts?

    2) Have you considered uploading data and images to figshare, in addition to your svpow.com data page? This would give each individual file a DOI, permanent archiving at a recognized repository, etc.

  7. Matt Wedel Says:

    1) Any particular reason for not putting scale bars on most of the images? I found it a little odd to read through and see references to centrum length, but not a corresponding scale bar. I guess there may be some issues of parallax for structures so big and wide. . .was this in your thoughts?

    Yep, that’s it exactly. Especially for the multi-view figures, like fig. 1 and fig. 7, it is just impossible to get one scale bar that works, and I hate hate HATE untrustworthy scale bars (which IMHO is almost all of them). For comparisons, we found that scaling things to the same centrum length (or height or whatever) was more useful, and giving people actual measurements in the captions means they have real data instead of I-measured-it-with-the-included-scale-bar-so-I-guess-it-might-be “data”.

    2) Have you considered uploading data and images to figshare, in addition to your svpow.com data page? This would give each individual file a DOI, permanent archiving at a recognized repository, etc.

    We’ve already done that for the Nexus files. I can’t think of any reasons not to do it for the figures, too. Gimme a few days–getting 25 files posted to FigShare only takes about 12 minutes, maybe, but getting them all posted to FigShare with good descriptions and no embarrassing errors takes a little longer, and I’m slammed right now. But I will do it–thanks for the prod.

  8. Mike Taylor Says:

    Thanks, Andy.

    I guess the truthful answer to #1 is that they just didn’t seem relevant to the points we were making, which were all to do with morphology and relative size. I don’t think the lack of scalebars impedes our argument at any point. That said, I do agree that it’s good practice to always include them, and I guess if I could do it all again, I would add them before submission.

    What would be the actual benefit of going through the rather cumbersome FigShare process for each of our 24 figures?

  9. Andy Farke Says:

    Cool! Thanks for the info…I agree that it can take a little to make sure all of the metadata are in line. (I forgot to mention that I am a figshare Advisor–pretty much exactly as it sounds, but I should have noted that potential conflict-of-interest in my first post)

  10. Andy Farke Says:

    Matt’s explanation was pretty satisfactory for me. I guess the main reason I like to see scale bars (when applicable) is that they allow reuse of images for research in ways the original author may not have anticipated–e.g., geometric morphometrics, etc. But, parallax is a problem as Matt noted–maybe it’s time for us to start stating where we place our scale bar relative to the specimen? Hmm. . .something I should have done for my ceratopsian skull papers.

    Permanence (figshare is backed up via CLOCKSS) and citable doi are the two reasons for me. . .yes it is cumbersome, but then again good metadata and documentation take time. However, I do see how figshare could improve usability here. It’d be nice to be able to duplicate metadata from another entry more easily; that would certainly reduce some of the drudgery.

  11. Matt Wedel Says:

    Regarding scale bars, Mike wrote:

    I do agree that it’s good practice to always include them

    Just to be clear, I don’t, for the reasons I gave above.

    and I guess if I could do it all again, I would add them before submission.

    Full disclosure: Mike wanted to include scale bars, and as first author I vetoed them, so if people are unhappy, their unhappiness should be directed solely at me.

  12. Matt Wedel Says:

    I guess the main reason I like to see scale bars (when applicable) is that they allow reuse of images for research in ways the original author may not have anticipated–e.g., geometric morphometrics, etc.

    Is the figure-with-scalebar more useful than the figure-with-actual-measurement-noted-in-caption? I’m genuinely curious, not trying to be snarky. I personally would a thousand times rather have a measurement than a scalebar. Surely anyone who wants to reuse this can just read the caption, and if they absolutely need scalebars, add them themselves.

    But, parallax is a problem as Matt noted–maybe it’s time for us to start stating where we place our scale bar relative to the specimen?

    If we’re going to use scale bars for 3D macro specimens, I think that’s imperative. And even then, with big sauropod elements it is just about impossible to get far enough away to remove the effects of parallax. I noted that in this post, and I have another post on the subject in the works.


  13. To Matt and Mike: Scale bars in humongous specimens obviously become almost unusable due to parallax, unless you were getting really intensive in serial photos at varying angles (for photogrammetry and whatnot), or you’re really close up where parallax is not an issue. But one wonders whether a scale of some sort with a disclaimer in the Methods section to describe the problem, BUT that the scale bar is fixed to observed centrum length. One could also include a 2D bar, for height and length, set to cent.l. and post.cent.h. or whatever to allow the viewer to have somewhat of examining structures at scale for the figures. Otherwise, one wonders if diagramatic figures of the vertebrae, corrected for parallax, would be BETTER, leaving the photos (scalebar-less) for Figshare.

  14. Mike Taylor Says:

    Full disclosure: Mike wanted to include scale bars, and as first author I vetoed them

    Is that true? Once again, I have no memory whatsoever of this. My memory is becoming a worry.

  15. Mike Taylor Says:

    … it is just about impossible to get far enough away to remove the effects of parallax.

    s/parallax/perspective/?

  16. Matt Wedel Says:

    Full disclosure: Mike wanted to include scale bars, and as first author I vetoed them

    Is that true? Once again, I have no memory whatsoever of this. My memory is becoming a worry.

    Yours may not be the only one. :-)

    I am certain–certain–that at some point in the last few months you brought up the fact that the photos in the paper have no scale bars, and I said that that was deliberate. I think it happened in Gchat. But to my immense irritation, I can’t find the conversation I’m thinking of, and the search facility isn’t turning it up, either.

    But going back farther I turned up something interesting. A pair of messages on April 23, 2012, with the subject line “Re: I have the ball, for a while at least”, included this exchange:

    Matt: We can list the lengths of verts in the caption. I am not opposed to also putting in scale bars, but obviously measurements count for more.

    Mike: I almost feel like we should make a point of omitting scale bars. Because, MYDD.

    So we’ve both been anti-scale-bar, at least at some point on this project.

    s/parallax/perspective/?

    Both! Perspective with respect to the bones not being in strict orthogonal views, and parallax with respect to any included scale bars. If a measurement tool is not in the same plane as the thing being measured, that counts as parallax. And I think that once objects get sufficient three-dimensionality, it’s just about impossible to accurately capture their dimensions with a scale bar–some part will be closer than the scale bar and therefore look bigger, some part will be farther away, etc. Admittedly this is tangled up in the perspective problem. IF we could get photos of big bones free of perspective distortion–like from infinite distance with infinite zoom–we could make scale bars free of parallax errors.

    But they’d still be dumb without accompanying measurements, and kinda redundant with them. The lameness of scale bars runs deep.

  17. Matt Wedel Says:

    Otherwise, one wonders if diagramatic figures of the vertebrae, corrected for parallax, would be BETTER, leaving the photos (scalebar-less) for Figshare.

    Oddly enough, Mike and I have just been talking about how to do this, for one specimen at least. It may be a couple of months before we’re in a position to actually do it, but I’ll try to remember to link back to this comment thread when we do. Watch this space.

  18. Casey Says:

    Ah yes, carefully reading the post may have actually answered my question. I say keep the scale bars out. People need to stop doing 2D morphometrics [full stop], based on photographs [2nd full stop], anyway, don’t enable them. For the same points about parallax above…how do you know that you’re getting the right landmark? Otherwise, I like creative uses of silhouettes. If you guys used silhouettes of mandibles or pterygoids as scale bars next to your vertebrae, I might pay more attention.

  19. Andy Farke Says:

    I’ll concede to Matt on measurements in caption *potentially* being just as good better than scale bars for image reuse. I’m reflexively in favor of scale bars because I’ve never known any better, and will confess that this discussion has gotten me thinking in a very good way about how useful they actually are. Upon this deeper consideration, I have a few semi-random points to note:

    1) Using lengths instead of scale bars requires careful justification and documentation of how the length was measured–i.e., centrum length or orbit width or metatarsal length or whatever is not a simple measurement, and unless it’s defined precisely it’s hazardous. In other words, it’s not necessarily solving the problem.

    2) I still strongly argue in favor of scale bars for composite figures that include two elements of vastly different size that are not depicted to scale. This is not so much for purposes of morphometrics, but for ease of reader comparison. As a reader, it’s a lot more intuitive to see two elements placed side by side with scale bars than to have to hunt in the caption to find their size, do the mental math, and then try to scale them in my head. The alternative is to depict elements to scale, but of course this doesn’t work well when they’re of very disparate size.

    3) Along the same lines, I personally find scale bars more intuitive than a measurement statement in the caption for grokking the size of specimens. This will probably just take practice to get around, but there you have it.

    4) It’s completely hazardous to try and do detailed morphometrics or take detailed measurements off of complex 3D objects depicted in 2D, but realistically people are still going to do it, and in some cases they have no choice (Spinosaurus holotype? Anything behind glass at the AMNH?) As mentioned above, I concede Matt’s point that measurements are more desirable for scaling, so this is more in response to Casey’s comment.

    This whole discussion has been a pretty awesome one, and has really challenged my assumptions about how I’m going to deal with scale bars in the future. At the end of the day, I’m going to leave them in, but for big and complex objects I’m going to be much more careful about stating where the scale bar was in the original photograph, whenever possible.

  20. Casey Says:

    So Andy, since this came up above, you also suggested Figshare in a recent revisions package I received, but you didn’t state that you were on the advisory board of it when you did so. As an author, why should I have to read advertisements from the editor for their pet project?<<<cynical interpretation. I don't think you are getting $ or kickbacks or perks from it (well, you should I suppose) and I assume you mean the best by it. But is there some line there that we shouldn't cross when it comes to either innocently championing a product (as an "objective" editor) versus something more sinister when dealing with reviews?

  21. Mike Taylor Says:

    Hmm, that really is a cynical interpretation, Casey! Surely the reason Andy is on the FigShare advisory board is because he already thinks it’s A Good Thing? I agree that he should have thought to mention it in his review, but it hardly seems necessary to entertain any narrative but the obvious one.

  22. Casey Says:

    Think of it as an All Yesterday’s style look at science dissemination. No harm intended :D.


  23. Since scale bars (or the lack thereof) are all the rage in this discussion, let me come down on the side of not including them with photos of specimens. Having attempted to scale specimens from the scale bars in many a paper, I can say they are so frequently in error that it’s painful. Worse, it seems that the inclusion of scale bars is often used in lieu of publishing measurements.

    Now of course people can mismeasure bones (or misrecord those measurements), but in my experience I’ve found published numbers to be far more reliable. I know the SVPOW team have devoted some past entries to pleading for people to include measurements of specimens, and I can’t second that loudly enough.

    I concede to Andy that scale bars can be useful as something of a heuristic tool for readers to informally gauge size, but for that sort of usage to be ok in my mind it would need to be an explicit choice that in no way reduces the need for publishing dimensions.

  24. Mike Taylor Says:

    It’s interesting. I’ve always included scale bars, at least before this paper. But the more I read here, and the more I think about them, the more I think that they are actively misleading more often than not. I find myself increasingly inclined to omit them by design in future papers, if necessary including measurements in the figure caption.

  25. John Scanlon, FCD Says:

    Scale bars should be represented as fuzzy, and with a note in the caption saying ‘Scale approximately one cubit (indicative only, not for morphometric use)’. Then they’ll be useful for their only legitimate purpose.

  26. Mike Taylor Says:

    That’s not a bad idea, John. Although come to think of it, the use of a human figure for scale is a sort of a fuzzy scalebar.


  27. Actually, the use of human silhouettes is not a bad idea at all, although we’d need other proxies for smaller bones (not everyone works on sauropods Mike).

  28. Mike Taylor Says:

    And that is where they make their big mistake.

  29. Sarah W Says:

    Nice work, guys! I like the sidebar page part but point out that the high-res images could be posted to MorphoBank or MorphBank and become part of the actual published data. :)

  30. Mike Taylor Says:

    Are there really both MorphoBank and MorphBank? Yes, I see that there are — here and here. What a terrible failure of naming. I might start a site called MorphaBank.

    I am not totally convinced that images posted to one or other those sites (or FigShare or DataDryad) are much more “published” than those on this site.

  31. Sarah W Says:

    Another benefit of MorphoBank is that you can post the nexus files as part of the project. We did this for the Nyasasaurus project and it meant that all the supplements and bonus images were in one place.

  32. Sarah W Says:

    Well, MorphoBank at least has NSF support and assigns accession numbers (as you would get for genetic sequences at GenBank). I think they are part of the published record if you publish those accession numbers as part of the paper. I like the idea of pointing readers of the paper somewhere (your site, or MorphoBank). I think the difference is that you are pointing to a site in the paper, and can cite accession numbers. Using DataDryad and MorphoBank also means you can’t manipulate data after the fact, and that your data is somewhere with other data. So people might search for one thing and also find yours, and make connections that way.


  33. […] Also, as noted in this post, it is odd that in this specimen of Apatosaurus the cervical ribs had not fused to the first two vertebrae, even though they normally do, and despite the fact that the vertebrae had fused to each other, even though they normally don’t. Further demonstration, if any were needed, that sauropod skeletal fusions were wacky. […]


  34. […] from this post. For the pros and cons of scale bars in figures, see the comment thread after this post. MYDD is, of course, a thing […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 345 other followers

%d bloggers like this: