Some quick backstory: lots of sauropods have long, overlapping cervical ribs, like the ones shown here in Sauroposeidon (diagram from this old post):

These long cervical ribs are ossified tendons of ventral neck muscles, presumably longus colli ventralis. We know they’re ossified tendons because of their bone histology (Klein et al. 2012), and we suspect that they’re longus colli ventralis because those tendons look the same in birds, just less ossified, as in this rhea (same specimens as these even older posts: 1, 2):

Diplodocoids have apomorphically short cervical ribs, which never extend very far past the end of their respective centra and sometimes don’t overlap at all. Still, we assume the long ventral neck muscles were there, just without long ossified tendons. Which brings me to Apatosaurus, which has cervical ribs that are anteroposteriorly short but famously massive, extending very below and/or to the sides of the cervical centra — for a truly breathtaking example see this post. Here are C3 through C7 in CM 3018, the holotype of Apatosaurus lousiae (Gilmore 1936: plate 24):

At least for me, it’s hard to resist the temptation to mentally scoot those vertebrae together into articulation, and imagine that the very swoopy-looking and maybe even down-turned cervical ribs allowed the ventral tendon bundles to wrap around the bottom of each cervical rib protuberance, something like this:

But it’s just not so, because like all 2D images, Gilmore’s plate distorts 3D reality. If you get to see the mounted skeleton in person, it’s clear that the cervical ribs are all more or less in line, and none of them are pointed at the big protuberances, which stick way out ventrolaterally.

Here I’ve drawn in the likely trajectories of the longus colli ventralis tendons. My little red pathways don’t precisely match the cervical ribs as mounted, but there’s a lot of distortion and restoration going on. For example, comparing with Gilmore’s plate we can see that the cervical ribs of C5, which point downward compared to all the others, only do that because someone forced them to — the whole anterior portion of the rib, where the shaft would actually join to the capitulum and tuberculum, is reconstructed. Even if I’m a little off, it’s clear that the cervical ribs shafts point backward, they’re all more or less in two parallel lines, and none of them point down and out toward the ventrolateral processes. The photo contains a mountain of useful morphological information that you’d never get from the lateral views.

My takeaways from all this:

  1. If a person has only seen 2D images of a specimen, and especially if those 2D images have only been orthogonal views with no obliques, their little island of knowledge is surrounded by at least a sizeable lake of ignorance, if not a small ocean.
  2. That doesn’t mean that seeing specimens in person is the only antidote — 3D models and 3D prints are extremely useful, and for specimens that are difficult to manipulate because of their size or fragility, they may be more useful than seeing or handling the specimen, at least for some questions.
  3. For Apatosaurus specifically, those ventrolateral processes cry out for explanation. They’re fairly solid knobs of bone that stick way out past the ossified tendons of the ventral-most neck muscles. That’s a super-weird — and super-expensive — place to invest a bunch of bone if you’re not using it for something fairly important, especially in a lineage that had just spent the last 80-100 million years making their necks as light as possible.
  4. Pursuant to that last point, we’re now in — ugh-ouch-shame — our 8th year of BrontoSMASH!!, with still just the one conference presentation to show for it (Taylor et al. 2015). Prolly time we got moving on that again.

References

Over on Mastodon (sign up, it’s great!), Jim Kirkland posted a baby Utahraptor caudal vertebrae for #FossilFriday. Here it is:

And after a bit of virtual prep work:

My first reaction was just “That’s pretty!“. My second, which I admit should have been my first, was “Wait a sec — how the heck do those things articulate?

The issue is that both the prezygs and the postzygs overhang the centrum by so much. If we imagine three of these babies consecutively, there are basically two options.

First, the centra articulate closely, with what we might feel intuitively is a reasonable cartilage gap; and the zygs cross over:

Does something like this ever happen? Not in sauropods, for sure, but it could be correct — if the zyg facets are some way short of the tips of their processes, so that the most distal parts of each process are pre-epipophyses and epipophyses rather than prezygs and postzygs per se.

The other interpretation is this, with the zygs overlapping near the end as in sensible dinosaurs, and much more spaced out centra:

If this is right, then (in this respect) baby Utahraptor tails resembled camel necks in having big intervertebral spaces, which in life were filled with big cartilage plugs.

 

.

.

.

SPOILER SPACE

.

.

.

Have a think about this before reading on.

.

.

.

SPOILER SPACE

.

.

.

 

 

OK, here is the horrible truth.

Dromaeosaur tails do overlap their zyg processes as in the first mock-up above: but they do much, much worse than this!

Here is the truly perverted figure 37 of Ostrom’s classic 1969 monograph on Deinonychus — the publication that catalysed the whole Dinosaur Renaissance:

As you can see, the zygapophyseal processes are grotesquely elongated, and overlap in long stiffening bundles with those of successive vertebrae (part C of the figure). The actual zyg facets are small, and close to the origins of these processes (see parts A and B of the figure). And the chevrons are also hideously protracted beyond their natural length to form stiffening bundles beneath the tail that complement those above the tail.

To add insult to injury, the chevrons even face in the wrong darned direction, extending anteriorly along the tail rather than posteriorly as in all decent animals. Yes: in Ostrom’s illustrations, we’re seeing the vertebrae in right lateral view, i.e. anterior is to the right.

All of this confirms that I was so, so right two decades ago to focus so completely on proper dinosaurs instead of these nasty mutant ones. Ugh.

Vertebrae of Haplocanthosaurus (A-C) and a giraffe (D-F) illustrating three ways of orienting a vertebra: articular surfaces vertical — or at least the caudal articular surface vertical (A and D), floor of the neural canal horizontal (B and E), and similarity in articulation (C and F). See the paper for details! Taylor and Wedel (2002: fig. 6).

This is a lovely cosmic alignment: right after the 15th anniversary of this blog, Mike and I have our 11th coauthored publication (not counting abstracts and preprints) out today.

Taylor, Michael P., and Wedel, Mathew J. 2022. What do we mean by the directions “cranial” and “caudal” on a vertebra? Journal of Paleontological Techniques 25:1-24.

This one started back in 2018, with Mike’s post, What does it mean for a vertebra to be “horizontal”? That post and subsequent posts on the same topic (one, two, three) provoked interesting discussions in the comment threads, and convinced us that there was something here worth grappling with. We gave a presentation on the topic at the 1st Palaeontological Virtual Congress that December, which we made available as a preprint, which led to us writing the paper in the open, which led to another preprint (of the paper this time, not the talk).

Orienting vertebrae with the long axis of the centrum held horizontally seems simple enough, but choosing landmarks can be surprisingly complex. Taylor and Wedel (2022 fig. 5).

This project represented some interesting watersheds for us. It was not our first time turning a series of blog posts into a paper — see our 2013 paper on neural spine bifurcation for that — but it was our first time writing a joint paper in the open (Mike had started writing the Archbishop description in the open a few months earlier). It was also the last, or at least the most recent, manuscript that we released as a preprint, although we’ve released some conference presentations as preprints since then. I’m much less interested in preprints than I used to be, for reasons explained in this post, and I think Mike sees them as rather pointless if you’re writing the paper in the open anyway, which is his standard approach these days (Mike, feel free to correct me here or in the comments if I’m mischaracterizing your position).

So, we got it submitted, we got reviews, and then…we sat on them for a while. We have both struggled in the last few years with Getting Things Done, or at least Getting Things Finished (Mike’s account, my own), and this paper suffered from that. Part of the problem is that Mike and have far too many projects going at any one time. At last count, we have about 20 joint projects in various stages of gestation, and about 11 more that we’ve admitted we’re never going to get to (our To Don’t list), and that doesn’t count our collaborations with others (like the dozen or so papers I have planned with Jessie Atterholt). We simply can’t keep so many plates spinning, and we’re both working hard at pruning our project list and saying ‘no’ to new things — or, if we do think of new projects, we try to hand them off to others as quickly and cleanly as possible.

Two different ways of looking at a Haplocanthosaurus tail vertebra. Read on for a couple of recent real-life examples. Taylor and Wedel (2022: fig. 2).

Anyway, Mike got rolling on the revisions a few months ago, and it was accepted for publication sometime in late spring or early summer, I think. Normally it would have been published in days, but the Journal of Paleontological Techniques was moving between websites and servers, and that took a while. But Mike and I were in no tearing rush, and the paper is out today, so all is well.

One of the bits of the paper that I’m most proud of is the description of cheap and easy methods for determining the orientation of the neural canal. For neural canals that are open, either because they were fully prepped or never full of matrix to begin with, there’s the rolled-up-piece-of-paper method, which I believe first appeared on the blog back when I was posting photos of the tail vertebrae of the Brachiosaurus altithorax holotype. For neural canals that aren’t open, Mike came up with the Blu-tack-and-toothpick method, as shown in Figure 12 in the new paper:

A 3d print of NHMUK PV R2095, the holotype of Xenoposeidon, illustrating the toothpick method of determining neural canal orientation. Taylor and Wedel (2022: fig. 12).

I know both methods work because I recently had occasion to use them, studying the Haplocanthosaurus holotypes (see this post). For CM 572, the neural canal of the first caudal vertebra is full of matrix, so I used a variant of the toothpick method. I didn’t actually have Blu-tack or toothpicks, so I cut thin pieces of plastic from the edge of an SVP scale bar and stuck them in bits of kneadable eraser. It worked just fine:

The neural canal of caudal 2 was prepped, so I could use the rolled-up-piece-of-paper method:

(Incidentally, Mike and I refer to our low-tech orientation-visualizers as “neural-canal-inators”, in honor of Dr. Heinz Doofenshmirtz from Phineas and Ferb.)

In the above photos, notice how terribly thin the base of the neural arch is, antero-posteriorly. Both of these vertebrae are in pretty good shape, without much breakage or missing material, and their morphology is broadly consistent with that of other proximal caudals of Haplocanthosaurus, so we can’t write this off as distortion. As weird as it looks, this is just what Haplo proximal caudals were like. And with the neural canals held horizontally, the first two caudals end up oriented like so:

Now, as we pointed out in the paper, the titular question is not about determining the posture of the vertebrae in life, it’s about defining the directions ‘cranial’ and ‘caudal’ for isolated vertebrae — Mike asked the question back when for the holotype (single) dorsal vertebra of Xenoposeidon. But an interesting spin-off for me has been getting confronted with the weirdness of vertebrae whose articular surfaces are nowhere near orthogonal with their neural canals. I tilted those CM 572 Haplo caudals so that their neural canals were horizontal partly because that’s the preferred orientation that Mike and I landed on in the course of this work, but also partly because to me, that’s a more arresting image than the preceding ones with the articular faces held vertically. I’m both freaked out and fascinated, and that seems like a promising combination — there are mysteries here that cry out to be solved.

As usual, we have loads of people to thank. In addition to all those listed in the Acknowledgments of the new paper, I’m grateful to Matt Lamanna and Amy Henrici of the Carnegie Museum of Natural History for letting me play with study the Haplo specimens in their care. Mike and I also owe a huge thanks to the editorial team at the Journal of Paleontological Techniques. We reached out to them a few days ago to ask if it might be possible to get our in-press paper done and out in time for SV-POW!’s anniversary weekend, and they pitched in to make it happen.

What’s next? We weighed the evidence and formulated what the best solution we could think of. Now it’s up to the world to decide if that was a useful contribution. The comment thread is open — let’s find out.

One of the benefits of being me is that my friends often make me cool dino-themed stuff for my birthday (f’rinstance). This year, it was this dinosaur dig cake from my friend Jenny Adams. Yes, it’s a vulgar, overstudied theropod,* but I take the requisite amount of joy from how thoroughly blown apart its skeleton is. Plus, the skull and cervicals are pneumatic (in vivo, if not in choco), so it’s a least plausibly interesting (i.e., not an ornithopod), and it looks cool (i.e., not Camarasaurus).

*I’m morally obligated to thank Paul Barrett for this wonderful phrase, which I use pretty much every chance I get.

Should you want to replicate this glycemic index Chicxulub, here’s the stratigraphic breakdown, starting from the bedrock (bedchoc?):

  • base layer is a regular chocolate cake,
  • but with added chocolate chips,
  • topped with vanilla frosting, to hold down:
  • a whole package of Oreos crumbled into faux dirt
  • surrounding the vanilla-flavored white chocolate dinosaur bones

Jenny made the dino bones using a set of (new, clean) plastic sand molds, like these:

You can find a zillion like ’em online by searching for ‘dinosaur sand toys’ or ‘dinosaur sand mold’.

Anyway, I can report that the excavation has been most enjoyable, but with about half the ‘quarry’ left to explore, the number of fossils recovered intact continues to hover near zero — we’ve been grinding them up to use as dietary supplements. Good thing it’s just a theropod!

I’ve been in contact recently with Matt Lamanna, Associate Curator in the Section of Vertebrate Paleontology at the Carnegie Museum of Natural History — which is obviously the best job in the world. Among a batch of photos that he sent me recently, I seized on this gem:

Tyrannosaurus rex, Diplodocus carnegii, Apatosaurus louisae and multiple mostly juvenile individuals of Homo sapiens. Photograph taken between 1941 and 1965. Courtesy of Carnegie Museum of Natural History.

There’s so much to appreciate in this picture: the hunchbacked, tail-dragging Tyrannosaurus; the camarasaur-style skull on the Apatosaurus; the hard-to-pin-down archaic air of Diplodocus.

But the thing I love about it is the 1950s kids. (Or, to be fair, maybe the 1940s kids or early 1960s kids, but you get the point.) They way they’ve all been asked to look up at the tyrannosaur skull, and are obediently doing it. How earnest they all appear. How they’re all dressed as tiny adults. How self-consciously some of them have posed themselves — the thoughtful kid one in from the left, his foot up on the plinth and his chin resting on his hand; the cool kid to his right, arms crossed, interested but careful not to seem too impressed.

Where are these kids now? Assuming it was taken in 1953, the midpoint of the possible range, and assuming they’re about 12 years old in this photo, they were born around 1941, which would make them 81 now. Statistically, somewhere around half of them are still alive. I wonder how many of them remember this day, and the strange blend of awe, fascination, and self-consciousness.

This is a time-capsule, friends. Enjoy it.

Last Thursday I gave a public lecture for the No Man’s Land Historical Society in the Oklahoma Panhandle, titled “Oklahoma’s Jurassic Giants: the Dinosaurs of Black Mesa”. It’s now on YouTube, on the No Man’s Land Museum’s channel.

There’s a point I want to make here, that I also made in the talk: we can’t predict the value of natural history collections. The first sauropod vertebrae that Rich Cifelli and Kent Sanders and I CT scanned back in the spring of 1998 belonged to what would become Sauroposeidon, but most of the ones we scanned after that were Morrison specimens collected by J. Willis Stovall’s crews from the Oklahoma Panhandle between 1934 and 1941. Those scans formed the core of the pneumaticity research that fleshed out the Sauroposeidon papers (Wedel et al. 2000a, b), and was more fully developed in my Master’s thesis and the papers that came out of that (Wedel 2003a, b).

OMNH 1094, a mid-cervical vertebra of Brontosaurus in right lateral view. If you’ve seen one of my talks or my first few papers, you’ve seen this vert. I just realized that I have almost all the photos I need to do a proper multi-view; stand by for a future post on that.

So the foundation of my career was built in large part from collections that had been made 60 years earlier, decades before CT was invented. I’ll also note here that Xenoposeidon — Mike’s fourth paper (Taylor and Naish 2007), but the one which really launched his career as a morphologist — is based on a specimen collected in the 1890s. Natural history collections are not only resources for making comparisons, but also the engines of future discovery, and building and maintaining them is one of the most significant contributions to science that we can make.

I thank a bunch of folks at the end of the talk, but I especially want to thank Brian Engh for the use of his art, and Anne Weil for inviting me to collaborate on the sauropod material from the Homestead Quarry. Looking forward to more adventures!

References

Morphological variation in paramedullary airways; yellow = spinal cord, green = diverticula. The spectrum of variation is discretized into four groups: i, branches of intertransverse diverticula contact spinal cord at intervertebral joints; ii, branches of intertransverse diverticula extend partially into the vertebral canal, but remain discontinuous; iii, paramedullary diverticula form sets of tubes that are continuous through vertebral canals of at least two consecutive vertebrae; iv, continuous paramedullary diverticula anastomose with supravertebral diverticula. Each variant is depicted diagrammatically (A, dorsal view; B, E, H, & K, transverse view) and shown in two CT scans; images in each column correspond to the same morphology. Morphology i: C, cormorant; D, scrub jay. Morphology ii: F, bushtit; G, common murre. Morphology iii: I, red-tailed hawk; J, black-crowned night heron. Morphology iv: L, M, pelican. (Atterholt and Wedel 2022: figure 5)

New paper out:

Atterholt, Jessie, and Wedel, Mathew J. 2022. A computed tomography-based survey of paramedullary diverticula in extant Aves. The Anatomical Record, 1– 22. https://doi.org/10.1002/ar.24923

Quick aside, which will soon be of historical interest only: so far, only the accepted-but-unformatted manuscript is available, with the final, fully-formatted ‘version of record’ due along at some point in the future. We’re not sure when that will be — could be next week, could be months from now — which is why I’m following my standard procedure and yapping about the new paper now. This has paid off in the past, when papers that were only available in accepted ms form were read and cited before the final version was published. UPDATE on April 9: the formatted version of record is out now, as an open-access article with a CC-BY license, and I swapped it for the ‘accepted ms’ version in the links above and at the end of this post.

This paper has had a weirdly drawn-out gestation. Jessie and I hatched the idea of it way back in 2017, when we were teaching in the summer anatomy course together. I learned that Jessie had a big war chest of CTs of dead birds, and I’d been obsessed with supramedullary diverticula in birds and sauropods for some time already (e.g., an SVPCA talk: Wedel et al. 2014). There were detailed published descriptions of the supramedullary diverticula in a handful of species — namely chickens, turkeys, and pigeons — but no broad survey of those diverticula across living birds. Jessie had the CT scans to do that big survey, which we got rolling on right away. She presented our preliminary results at SVPCA in 2018 (Atterholt and Wedel 2018), and by rights the paper should have been along shortly thereafter. More on that in a sec.

One thing that may seem odd: we use the term ‘paramedullary diverticula’ instead of the more familiar and established ‘supramedullary diverticula’. That’s because these diverticula are not always dorsal to the spinal cord; sometimes they’re lateral, sometimes they’re ventral, and sometimes they completely surround the spinal cord, like an inflated cuff. So we decided that the term ‘paramedullary’, or ‘next to the spinal cord’, was more accurate than ‘supramedullary’, or ‘above the spinal cord’, for describing this class of diverticula.

Observed variation in the shape, arrangement, and orientation of paramedullary diverticula relative to the spinal cord; yellow = spinal cord, green = diverticula. A, paired diverticula dorsal to spinal cord in an ostrich. B, paired diverticula lateral to spinal cord in a bushtit. C, paired diverticula ventral to spinal cord in a violet turaco. D, three diverticula dorsal to spinal cord in an ostrich. E, four diverticula dorsal to spinal cord in an eclectus parrot. F, single, c-shaped diverticulum dorsal to spinal cord in an ostrich. G, diverticula completely surrounding spinal cord and pneumatizing vertebra in a violet turaco. H, no paramedullary diverticula present in a Pacific loon. I, diverticula completely surrounding spinal cord in a pelican. (Atterholt and Wedel 2022: figure 6)

I will have more to say about the science in other posts, and you can get the scientific backstory in this post and this one and the abstracts cited above and linked below. The rest of this post is mostly about me, so if you stick around, buckle up for some advanced navel-gazing.

There’s no one reason why this paper didn’t come out sooner. In short, I hit a wall. We went through a curriculum change at work, and suddenly the annual schedule that I’d relied on for a decade was completely upended. I had some unexpected challenges in my personal life. But the biggest problem was that my attitude toward research and writing had changed, for the worse.

When I was fresh out of grad school I had this kinda snotty attitude that my research was MINE, and wherever I was teaching was just, like, a paycheck, man, but they don’t own me, or my research. And as my teaching and committee responsibilities ramped up I still felt like research and writing was something I did for myself, and that my mission was to steal however many hours I could away from the “day-job work” to get done the things that I really wanted to do. Like a guerilla insurgency. In retrospect, it was a pretty good attitude for getting stuff done.

But somewhere along the way, I stopped thinking about research as something that belonged to me, something that I did for myself, and started thinking about it as part of my job. (This also maybe is not so flattering in what it reveals about how I think, or at least thought, about my job.) Instead of using my research time as a source of energy and a wellspring of satisfaction and positivity, I starting thinking of it only as a sink. And it happened so insidiously that I didn’t even realize it. My productivity plummeted, and I didn’t understand why. I was restless and depressed, and I didn’t understand that either. At the level of my superficial thoughts I still wanted to get research done, but my subconscious was turned off to it, so I just spun my wheels.

Then the pandemic hit. I’d always been a pretty optimistic, upbeat person, but I found myself just staring off into space franticizing about all the horrible things going on in the world, or staying up too late doom-scrolling the news. I slept too little, and poorly, and by the end of 2020 I felt worn down to a nub.

Osteological evidence of paramedullary diverticula. A, pocked texturing inside the vertebral canal of a pelican (LACM 86262). B, pneumatic foramen on the roof of the vertebral canal of an albatross (Phoebastria nigripes, LACM 115139). C, pneumatic foramina in the floor of the vertebral canal of an ostrich (Struthio camelus, LACM 116205). D, deep pneumatic fossae in the roof of the vertebral canal of an Eastern moa (Emeus sp., LACM unnumbered). (Atterholt and Wedel 2022: figure 7)

Then a series of positive things happened:

  • I received a long, heartfelt email from Jessie (fittingly!), asking after me and laying out a plan for getting the paper done and out. It was the kick I needed to look inside and start picking myself apart, to figure out what the heck was going on. Much of this post is cribbed from my reply to her.
  • I got a little break from lecturing in the spring of 2021, and that gave me the space to get a couple of things finished and submitted — the pneumatic variation paper with Mike in January (Taylor and Wedel 2021), and the Haplocanthosaurus neural canal paper, which was submitted even earlier in January, although it came out much later (Wedel et al. 2021; more on that publication delay in a future post).
  • Finally, I had young, energetic coauthors who were moving projects forward that required modest levels of effort from me, but which paid off with highly visible publications that I’m proud to be an author on, including the saltasaur pneumaticity paper (Aureliano et al. 2021) and the ‘Sauro-Throat’ paper (Woodruff et al. 2022).

It’s impossible to overstate how energizing it was to get new things done and out, and how much it helped to have collaborators who were putting wins on the board even when I was otherwise occupied. One of those collaborators was Jessie, who just kept pushing this thing forward — and, sometimes, pushing me forward — until it was done. So the paper you can read today is a testament not only to her acumen as a morphologist, but also to her tenacity as a scholar, and as a friend.

The part of the paper I’m happiest about is the “Conclusions and Directions for Future Research”, which points the way toward a LOT of further studies that need to be done, both on extant birds and on fossil archosaurs, ranging from bone histology to functional morphology to macroevolution. As we wrote in the concluding sentence of the paper, “We hope that this study serves as a foundation and an enticement for further studies of this most unusual anatomical system, in both extinct and extant archosaurs.”

I can’t wait to see what comes next.

References

For reasons that would be otiose, at this moment, to rehearse, I recently found myself in need of a hemisected turkey cervical. Happily, I own five skeletonised turkey necks, so it was with me the work of a moment to select a candidate. But now what? How to hemisect it? We have  discussed plenty of hemisected things here at SV-POW!, but they tend to have been produced using heavy machinery such as a band saw: something that I singularly lack.

SPOILER: I found a way. Here is a domestic turkey Meleagris gallopavo domesticus, 9th cervical vertebra, hemisected, in right medial view. Read on to discover the extremely high-tech approach that yielded this prize. It’s propped up on some kind of turkey bone to help me get a good medial perspective, I am thinking maybe the pygostyle?

One idea was to use an angle-grinder: not to cut down the midline of the vertebra — it would be much too blunt and powerful for a small, delicate vertebra — but to use as a sanding surface, locking the grinder in place and holding the vertebra up against the spinning plate. That might work well, assuming I could find a way to secure the angle grinder safely, but as it happened my need for a hemisected vertebra came up during a power cut. (Thanks, Storm Eunice!)

So I did it the way the Amish do their vertebral hemisections: by hand, simply by rubbing the vertebra against a sheet of sandpaper:

CT scanning: the Amish method.

This is not as laborious as you might think. I used a single sheet of medium-grade sandpaper, and it took maybe 15–20 minutes. And I just rubbed back and forth while exerting downward pressure. Initially I worked my way only through the prezygapophyseal ramus, which is the part of the turkey cervical that extends the furthest laterally. Once I was satisfied that the plane between eroded prezyg and the intact postzyg was parasagittal, I just kept the vertebra parallel to the sandpaper and kept rubbing. (Sorry I didn’t think to get a photo at this stage.)

One thing that took me by surprise is that there was so very much bone dust. I mean, I am an idiot that this surprised me, since the whole purpose of this exercise was to reduce one half of this vertebra to bone dust. But the lesson to be learned here is to do it on the easily-cleaned bathroom floor — not on the desk next to the computer keyboard and above a carpet. Learn from my mistakes, folks!

Anyway, after some work on the prezyg/postzyg pair, here’s how the vertebra was looking:

You can see straight away that the prezyg ramus, postzyg ramus and parapophyseal ramus are extensively pneumatized, honeycombed with small, irregular air-spaces. In this image it looks like the region of bone between the pre- and postzygs is much more solid, but this is an illusion: what we’re seeing here is a section through the cortical  bone of the neural arch, cut parallel to the surface. Let this be a warning not to over-interpret individual slices of CT-scans!

Once we get a little deeper, we see that the whole wall of the neural arch — and indeed the centrum and the neural spine — is honeycombed, just like the zyg rami:

Now we have another area of what I’m going to call Phantom Apneumaticity: the posterior part of the centrum looks like solid bone, apart from a few pneumatic spaces in the posteroventral extremity. Again, this is an illusion.

Here’s the next place I stopped:

Here, the Phantom Apneumaticity is even more striking: seeing just this as a CT slice could easily mislead someone into thinking that almost the whole of the posteroventral part of the centrum is solid bone. But again, it’s just that we’re very close to the surface of the bone, and seeing a slice parallel to that surface.

This last image also shows an important point of technique: there is a low convex ridge running across the phantom apneumatic area from the top of the cotyle to the base of the centrum. This is where I had changed the angle I was holding the vertebra at, so I accidentally sanded the posteroventral part of the vertebra more than the rest. I found that it was important during this process to keep checking the angles, and to adjust: making sure I wasn’t sanding more from the front than the back, or from the top than the bottom, or leaving a ridge like this.

Also in this last photo you can see that I was just beginning to break through into the neural canal: the anterior part of it is now exposed, between the anterior part of the neural spine and the anterior articular surface. At this stage I sighted along from in front to get a sense of how much further I had to go:

Domestic turkey Meleagris gallopavo domesticus, 9th cervical vertebra, most of right side removed, in anterior view with dorsal to the right. Propped up on the coracoid of a different, larger, turkey.

Quite a way, I guess. Here it is rotated and cropped, so you can more easily recognise it:

Domestic turkey Meleagris gallopavo domesticus, 9th cervical vertebra, most of right side removed, in anterior view. You can see that the neural canal is still mostly intact.

More sanding was required. I sanded some more.

You’ve already seen the final result up at the top of the page, but here is a cleaned-up version of that image, oriented according to Definition 3 of Taylor and Wedel (2019):

Domestic turkey Meleagris gallopavo domesticus, 9th cervical vertebra, hemisected, in right medial view.

And if that isn’t beautiful, what is?

The exciting thing is, anyone can make one of these. Matt’s already explained how to extract and clean up bird vertebrae and given you some ideas of what to do with them. Prepare out some turkey vertebrae and get going with the sandpaper!

I leave you with one more image: the hemisected vertebra in anterior view, oriented with dorsal to the top, and mirrored so it make up a complete vertebra once more. Enjoy!

References

 

I was looking more closely at the turkey skeleton from my recent post, and zeroed in on the last two dorsal (= thoracic) vertebrae. They articulate very well with each other and with the first vertebra of the sacrum, with the centra and zygapophyses both locking in so that there can only have been very little if any movement between them in life. Here they are, in right lateral view:

Last two dorsal (= thoracic) vertebrae of a mature domestic turkey Meleagris gallopavo domesticus, in right lateral view.

Before we move on, it’s worth clicking through to the full-size version of this image and wondering at both the quality of modern phone cameras (a Pixel 3a in this case) and the variety of textures on these little bones. There is smooth, finished bone on the sides of the neural spines; very fine pits and bumps on the zygapophyseal facets where the thin layer of hyaline cartilage attached; rougher texture in the parapophyseal facets where thicker cartilage attached; and very rough texture on the ends of the transverse processes, where there was relatively thick cartilage.

And there is, unsurprisingly in a bird, pneumaticity everywhere. In the more anterior vertebra alone (to the right) the photo shows pneumatic openings (from bottom to top) low on the centrum (below the parapophysis), high on the centrum (below the lateral process),  in the hollow between the lateral process, the posyzyg and the centrum, on the lateral surface of the prezygapophyseal ramus, and on the rear surface of the lateral process. There are others that are obscured in this photo, including on top of the lateral process where it meets the neural spine. Here they are, pointed out for you (with the hidden one shown translucently):

Last two dorsal (= thoracic) vertebrae of a mature domestic turkey Meleagris gallopavo domesticus, in right lateral view. Pneumatic openings on penultimate vertebra highlighted with red lines; obscured opening above lateral process shown as translucent.

OK, that was the B-movie. Now to the main feature. The next photo shows the same two vertebrae, folded away from each other so that we see the anterior face of the posterior vertebra (on the left) and the posterior face of the anterior vertebra (on the right).

Last two dorsal (= thoracic) vertebrae of a mature domestic turkey Meleagris gallopavo domesticus. Left: last dorsal vertebra in anterior view; right: penultimate dorsal vertebra in posterior view.

Again, do click through to see the exquisite detail, especially the complex of pneumatic features on the anterior face of the neural spine of the last dorsal (on the left) and on the posterior face of the left lateral process of the penultimate dorsal (on the right).

And … in the articular facets of the centra?

Seriously, what the heck is going on here? It doesn’t make sense  that there would be pneumatic openings in articular surfaces, because by definition something else (in this case the adjacent vertebra) is abutted hard up against then, so there is no way for a diverticulum to get in. For the same reason, you don’t get vascular foramina in articular surfaces because there is no way to get an artery in there. And there is no hint in these vertebrae of channels along either articular surface that diverticula or arteries could  possibly have laid in.

And yet, there those big openings are. What are they?

I discussed this with Matt, in case it’s Well Known Phenomenon that I’d somehow not heard about but it seems it is not. What we know for sure is that these openings are present, and that they are not mechanical damage inflicted during preparation. So what are they?

What else even is there for them to be? What penetrates bone apart from diverticula and blood vessels? Nerves follow the blood vessels, so it can’t be nerves in the absence of blood vessels.

By the way, there are similar but smaller openings in the posterior face of the last dorsal (the one on the right in the photo), but none anywhere else along the postcervical column: not on the anterior surface of the penultimate dorsal, not on the front or back of the sacrum, and not in any of the other dorsals.

One possibility we considered is that the vertebrae were locked together in life and that a pneumatic space inside the centrum of the last dorsal worked right through into the penultimate one. But that doesn’t work: the openings are not aligned. Also, those in the penultimate dorsal are definitely blind (i.e. they do not connect to deeper internal air-spaces) and those in the last dorsal probably are, too.

We do not know what is going on here.

Help us! Is this kind of thing common in turkeys? Have people seen it in other taxa? Do we know what it is?

Naturally I was grateful when Cary invited me to be part of the team working on Dolly, the diplodocid with lesions in its neck vertebrae (Woodruff et al. 2022; see previous posts on Dolly here and here). I was also intellectually excited, not only to see air-filled bones with obvious pathologies, but also for what those pathologies could tell us about Dolly and other sauropods. That’s the part of our new paper I want to unpack in this post.

We have a lot of evidence that air-filled bones in birds are a good model for air-filled bones in extinct dinosaurs. And we have several lines of evidence (not just air-filled bones; see Schachner et al. 2009, 2011, 2020) that the respiratory systems of many dinosaurs functioned broadly like those of living birds. But we have less direct evidence than we’d like, so every additional bit of information is welcome.

Diving into Diverticula

In birds, the air-filled bones in the neck and body are connected to the respiratory system by air-filled tubes. These tubes sometimes get called air sacs, in the sense that they are sacs filled with air, but we also refer to them pneumatic diverticula, to distinguish them from the respiratory air sacs in the torsos of birds that ventilate the lungs. Imagine blowing up some rubber gloves and sticking them inside a bird* and you’ll have a pretty good mental model of the system — the inflated ‘palm’ area of each glove is like one of the respiratory air sacs, the air-filled glove fingers are the diverticula, and the rubber material of the glove is the pneumatic epithelium that lines the air sacs and the diverticula alike. 

* Please don’t actually try that.

The cartoon above presents an unrealistically simplified picture of the respiratory system in dinosaurs and birds. For one thing, I omitted the windpipe or trachea — that blue tube going up the neck represents the diverticula that run alongside, and often inside, the neck vertebrae, parallel to the trachea but separated from it by whole sheets of muscle. Also, the cartoon only shows diverticula of the air sacs, but diverticula can also originate from the lungs themselves (see O’Connor 2006: 1211 and Schachner et al. 2020: 16-19). Here’s the actual respiratory system of a pigeon, with the trachea and lungs in pink and the air sacs and their diverticula in blue (Muller 1908 fig. 11):

I think it’s pretty natural to look at that illustration and wonder where the heck the guts go, since it certainly looks like the air sacs are occupying the entire volume of the torso. The answer is that the air sacs enclose the viscera “as do the shells of a nut”, in the memorable formulation of Wetherbee (1951: p. 243), describing the air sacs of the English sparrow.


Fig. 4. Reconstruction of the distribution of pneumatic diverticula in diplodocids and dicraeosaurids. A. Schematic drawing of midcervical vertebra of Diplodocus in left lateral aspect (A1), in dorsal aspect with single neural spine (A2) and in dorsal aspect with bifurcate neural spine (A3). The partitioning of pneumatic diverticula at the lateral surface of the vertebral corpus is hypothetical, based on the strongly divided pneumatic fossae. Schwarz et al. (2007: fig. 4A).

Furthermore, the pneumatic diverticula around the vertebrae in birds are complex, and we are fairly certain that they were also complex in sauropods, because they left so many distinct traces. The most detailed reconstructions of the cervical diverticula in sauropods that I know of are those of Daniela Schwarz and colleagues (2007), as shown above. For what it’s worth, I think those reconstructions are not just reasonable but perhaps conservative; I think there’s a good chance that the diverticular network around the vertebrae was even more complex and extensive. 

The rubber-glove model also lets us see that the diverticula are cul-de-sacs. We know that diverticula can anastomose, or merge, to form networks, and there is a possibility that if diverticula from different air sacs anastomosed, different pressures in those air sacs might allow some air to circulate through the diverticular network. Maybe — the anterior and posterior air sacs fill and empty at the same time, so there might not be a pressure differential to exploit. If air circulates in the diverticula at all in birds, it probably happens in the dorsal vertebrae, where diverticula from different parts of the respiratory system have the best opportunity to anastomose. But the far ends of the diverticular network are always dead ends, and we assume that air diffuses in and out of those terminal diverticula fairly slowly. We’re stuck with assumptions because no-one’s ever checked, experimentally, to determine the rate of diffusion or circulation of air in the diverticula. But it’s hard to imagine much circulation in the terminal diverticula, with no air reservoir or pump at the far end.

Reconstruction of soft−tissues in the neck of Diplodocus. A. Transverse cross−sections through cervical vertebra with bifurcate neural spine in the diapophysis region (A1) and in caudal third of vertebra (A2). B. Transverse cross−sections through cervical vertebra with single neural spine in diapophysis region (B1) and in caudal third of vertebra (B2), dashed outlines representing possible craniocervical extensor muscle analogous to m. biventer cervicis of extant birds or m. transversospinalis capitis of extant crocodylians. Schwarz et al. (2007: fig. 7A-B).

Here’s another great illustration from Schwarz et al. (2007), showing cross-sections of the neck of Diplodocus with hypothetical soft tissues restored. Bone is black, muscle is pink, and the pneumatic diverticula are blue. As this diagram makes clear, the air spaces in the bones are themselves extensions of the diverticula (that much is true regardless of how extensive we make the reconstructed diverticula outside the vertebrae). Instead of smooshing an inflated rubber glove into a duck, imagine smooshing one into a vertebra of a duck — or a Diplodocus — so that all of the empty spaces are occupied by some blobby bit of inflated-glove finger. All of air spaces in the bone would be lined by rubber-glove material, which in this metaphor is the same pneumatic epithelium that lines both the respiratory air sacs and their pneumatic diverticula, outside the bones or inside them.

I get to see this firsthand in the gross anatomy lab in our unit on head and neck anatomy. As we open up the skulls of the cadavers, the air-filled epithelial balloons that fill the sinuses sometimes pull away from, or completely out of, their bony recesses. (I’ll bet I could demonstrate the same thing with the sinuses of a pig or sheep head — I should give that a shot and post the resulting photos or videos here.) The point is, the pneumatic epithelium is in intimate contact with the bone, lining every pneumatic fossa, foramen, and internal chamber; this will be really important later on.

Incidentally, one question I get a lot is whether the diverticula, inside or outside the bones, contributed to gas exchange in sauropods. The answer is, probably not. We know from dissections and histological work on birds that the respiratory air sacs, their diverticula, and the diverticular spaces inside the skeleton are all relatively avascular, meaning that the tissues get enough blood to stay alive, but aren’t specialized for gas exchange. Furthermore, physiological experiments on living birds have shown that about 95% of the gas exchange happens in the lungs, and almost all of the remaining 5% happens in the paired abdominal air sacs (Magnussen et al. 1976), probably because they are so large and so intimately in contact with the guts (Wetherbee’s nutshell metaphor), which are well-supplied with blood. We also know from bone histology that the air-filled bones of extinct dinosaurs are essentially identical to those of modern birds (Lambertz et al. 2018), so there’s no evidence that they functioned any differently.

A simplified diagram of the sauropod respiratory system. What I’ve labeled “air tubes” here are the pneumatic diverticula. Air holes in the vertebrae are also known as pneumatic foramina. The shapes of the lungs and air sacs are speculative, but the minimum extent of the pneumatic diverticula is not–although it could be an underestimate (e.g., diverticula might have gone even further down the tail, and just not left any diagnostic traces on those vertebrae).

A final piece before we get back to Dolly: we know from lots of anecdotal observations, and some actual experiments, that air-filled bones have to stay connected to the outside to form in the first place, and to stay healthy afterward. This is true of both human sinuses and postcranial pneumatic bones in birds, so it’s reasonable to assume that it’s a general feature of all air-filled bones (see Witmer 1997 for lots of relevant citations and discussion). This is a pretty handy thing to know, because if we find an air-filled vertebra way out in the tail, we know pneumatic diverticula of the respiratory system got at least that far. ‘At least’ because pneumatic diverticula can make diagnostic traces on bones, but they don’t always do so. That means that the diverticular network can easily be more extensive than its skeletal traces, but not less so — see Wedel and Taylor (2013) for more on that.

To sum up, we suspect the following things about pneumatic diverticula around the vertebrae of sauropods, including Dolly:

  1. The diverticula were complex, based on the traces they left on the bones, and similarly complex diverticula in birds.
  2. The diverticula were patent, or open, maintaining an open connection to the outside by way of the respiratory air sacs, lungs, and trachea, because that’s how air-filled bones work in living birds and mammals.
  3. Despite being complex and ultimately open to the outside in one direction, the diverticula were cul-de-sacs, with little or no active circulation of air — especially in the neck.

With that in mind, what does the distribution of infected bone in Dolly tell us about sauropods?

Infections and inferences

Here’s another illustration of the respiratory system of the pigeon, this time a dorsal or top-down view, from Muller (1908: fig. 12):

Okay, that was a bit of a bait-and-switch: I promised you Dolly and gave you another pigeon. But that’s only to help you understand this similar cartoon I drew, which represents Dolly’s respiratory system and neck vertebrae, also seen from the top down:

Like the earlier cartoon, this is pretty simplified. For instance, I got lazy and didn’t draw all of the neck vertebrae. In life, Dolly probably had 15 or 16 neck vertebrae, like other diplodocids, and we know that the three with lesions are C5-C7 because they were found articulated. Here I drew just enough vertebrae to make my points, and I left off the head and all the other extraneous bits. Also, I’ve drawn the diverticula that run up the neck originating from cervical air sacs, as in pigeons (Muller 1908), but there is evidence that in ostriches those diverticula may originate from the lungs themselves (Schachner et al. 2020). Whether the diverticula come from the lungs or the air sacs is probably not an answerable question for sauropods, and for my purposes here, it doesn’t matter, only that the diverticula are connected back to the core respiratory system.

Three things struck us about the distribution of the infected bone in Dolly’s neck:

  1. The lesions are all in vertebrae that are a long way up the neck, far from the lungs and respiratory air sacs in the torso.
  2. The lesions are clustered in serially-adjacent vertebrae, instead of being scattered up and down the neck randomly.
  3. The lesions are present bilaterally, on both left and right sides of the affected vertebrae.

Well, as opposed to what? We can imagine a scenario in which the lesions were scattered randomly, not just up and down the neck, but also on left and right sides, like so:

If the infection had been carried in the blood, we might expect such a random pattern. In that case, it would be an extreme coincidence if a blood-borne infection, which could go anywhere in the body, only manifested in the air spaces on the sides of three consecutive vertebrae. The clustering of the Dolly’s lesions, in the air spaces on both sides in three adjacent vertebrae, far up the neck, points to a different cause.

Recall that diverticula are lined by epithelium, and that in air-filled bones, the epithelium is right up against the bone tissue. The infection in Dolly almost certainly started out as an infection in the diverticulum, which was so severe that it spread to the underlying bone. In exactly the same way that the air spaces in the bones are the skeletal footprints of the diverticula, the lesions in Dolly’s vertebrae are the skeletal footprints of infected epithelium lining the diverticula, like so:

The infection may have gotten so severe, far up Dolly’s neck, precisely because there was little airflow so far from the lungs and air sacs. Airborne bacteria or fungal spores could have floated into the diverticula by diffusion, come to rest against the epithelium in warm, dark, humid conditions, and gone wild. It’s also possible that a huge swath of Dolly’s respiratory system was infected, but the infection only got severe enough to spread to the underlying bone in cervical vertebrae 5-7, in which case the actual infection might have looked something like this:

Just like a diverticulum can contact a bone without producing a distinct trace, the pneumatic epithelium could be infected without producing a bony lesion. Thought experiment: how many times have you had a sinus infection, and how many people do you know who have had sinus infections? And how many of those sinus infections were severe enough, and lasted long enough, to produce bony lesions like we see in Dolly? Probably very few — such things do happen in humans, and the medical literature has plenty of cases (and if this has happened to your or a loved one, you have my full sympathy) — but on a population level, the fraction of respiratory infections that produce bony lesions is miniscule. Similarly, it’s very likely that much more of Dolly’s respiratory system was infected than we can tell from the skeleton.

A cervical vertebra of an ostrich with some of the pneumatic diverticula traced on.

The presence of infected bone on both left and right sides of C5-C7 in Dolly is also telling. If the diverticula on the left and right sides of the neck were separate, the symmetrical pattern of infection would be another extreme coincidence. But in birds there are opportunities for diverticula from the left and right sides of the neck to meet and anastomose, especially the supravertebral diverticula on the neural arch (shown above), and the supramedullary diverticula inside the neural canal. Based on pneumatic traces on the vertebrae we infer that the same diverticula were present in sauropods, as shown up above in the Diplodocus figures from Schwarz et al. (2007), and those left-and-right communications probably allowed the infection to develop more or less symmetrically. Or to put it another way, the symmetrical infections are additional evidence that the diverticula on the left and right sides of the neck were connected across the midline, and birds show that there are several ways that could have happened.

CT scans of cervical 7 of MOR 7029. Photograph and scan model of the vertebra ((A,B) respectively). The colored lines in (B) correspond to the scan slices (and scan interpretative drawings below). White arrows point to the external feature, while black arrows denote the hyperintense bone and irregular voids. (C) Comparison of the abnormal tissue composition of MOR 7029 (left), compared to that of a ‘normal’ diplodocine (right). Text and white arrows indicate the various features different shared/differentiated between the two. For the interpretative drawings, white = ‘normal’ bone, grey = hyperintense bone, black = irregular voids. Woodruff et al. (2022: fig. 2).

Another possibility is that a good chunk of the internal structure Dolly’s vertebrae was infected, and the lesions that we see on the surface are just the groady tips of big, disgusting icebergs of infected bone. In fact, that’s pretty much what the CT scans show. So possibly the infection started on one side of each vertebra and basically burrowed through to reach the other side. That would probably take weeks or months, whereas the infection could have spread across the midline through diverticula in hours or days, so I think the latter scenario is still the most plausible explanation for the presence of the lesions on both sides of the affected vertebrae.

In summary, I don’t think Dolly tells us anything surprising that we didn’t suspect before. Rather, the pattern of infection in Dolly makes perfect sense if the diverticula of sauropods were essentially bird-like, and that pattern is difficult to explain any other way.

Finding skeletal traces of a respiratory infection in Dolly was still a crazy lucky break, and that’s something I’ll discuss more in the next post in this series.

References