Giraffatitan's dorsals are just WEIRD.

Posted by Nima On Sunday, March 16, 2014 8 comments

Some really interesting stuff here. First off, the last "name that dinosaur" contest was a good one, Zach Armstrong won. It was indeed the La Invernada titanosaur, a relatively small species that doesn't have a name but is pasted all over South American paleontology websites. The reasonably complete foot allows us to place this animal at the hub of lithostrotia, close to Epachthosaurus.

Second, the bizarre derived titanosaur Yongjinglong datangi has been described in PLoS One. A crowning moment for both Chinese paleontology and open-access research. So long Cretaceous Research, Acta Geologica Sinica and other paywalled journals.This odd creature is from the Hekou Group, so it was probably in the same ecosystem as Huanghetitan and Daxiatitan. Details here.

But the craziest thing to come to my attention is from an old favorite. 

I just realized while looking at revising my Giraffatitan skeletal that most previous restorations seem to have either botched the shape of some of the dorsals to look too generic or followed Dr. Werner Janensch's rather hasty full body skeletal instead of his far more detailed engravings of the actual fossil material of the primary specimen, HMN SII.

Here's my original reconstruction, which you can see on DeviantArt:

This uses SII as well as a number of other specimens to fill in the hindlimbs, shoulders, hips, head and tail. Lets take a look at the dorsals.

There's a bit of uncertainty as to which dorsal was truly the last. However, the last two shown in this reconstruction right before the sacrum (the ones whose neural spines seem to neatly interlock with each other) are fused at the centrum joint. HMN SII was a subadult individual (judging by the lack of fusion in the coracoid, and unfused scapula found in similar-sized individuals) that either was getting near puberty or had some unique pathologies such as DISH or ankylosing spondylitis (this pair of bones shows some ossified ligaments on the neural spines which may also have fused together given enough time). This pair of fused rear dorsals is labeled as D11 and D12 (the final two dorsals) by Taylor (2009), but if you follow Janensch (1950) they should actually be D10 and D11. D12 on the other hand, looks as illustrated on the left in the image below.

Last 3 dorsals in Giraffatitan, from HMN SII and the even larger HMN fund no. (which also includes the caudal series that Janensch frankensteined onto the rear end of SII).
It's from a larger individual (the centrum, is thicker while the top of the neural spine is eroded off), but clearly not the same bone as either of the two fused ones. And it looks far more typical in shape for a terminal dorsal that hooks into the front end of the sacrum. So my original skeletal as well as all the others by Greg Paul, Scott Hartman, etc. are probably wrong and will have to be revised.

However the main point to take note of isn't even this discrepancy, but rather the bone that sits in front of the fused pair in the SII specimen. Note the red box around this bone.

The dorsal vertebra in front of the fused pair has a long centrum. In fact, it looks freakishly long because of vertical crushing. I have "uncrushed" it a bit. The original is so bizarre it looks like it came from a totally different species, but it was found together with the rest of the same specimen.

Giraffatitan HMN SII Dorsal 9 (per Janensch, 1950) or 10 (if you follow Taylor, 2009), reversed. Arrows show direction of geological crushing. The centrum used to be deeper in life, the lower neural arch was taller and not smushed into the centrum, and the neural spine tilted rearward instead of forward. The prezygapophyses also seem to be worn off, as is much of the diapophysis which has been crushed forward.

Now even if you correct for crushing, that's still going to be a very long centrum compared to the vertebrae both before and behind this one. And its rear rim has a totally different angle from the other centra, meaning that between this bone and the next one down (the first on the fused pair) there is an odd dip in the spine, a sort of "lordosis" or "anti-hunchback" posture. And however you restore the end of the centrum (its upper portion is missing and represented by a dotted line here), there is still going to be a BIG gap between the neural spine of this vertebra and the next (even with the spine tilted the correct way, uncrushed). But the gap is often ignored in the schematic literature.

Four different reconstructions of the Giraffatitan torso, primarily based on HMN SII. (A) Greg Paul, 1988; (B) Scott Hartman, 2012; (C) Asier Larramendi, 2013; (D) Nima Sassani, 2011.

None of these have the order correct with the D12 based on Janensch (HMN fund no 8). One of the two speculative middle dorsals has to be removed to make room for D12 at the back end and still keep the count at 12 dorsal vertebrae, which is typical of basal titanosauriformes. But notice how some of these skeletals (notable mine and Asier's) do show the big gap and also the odd "return up" of the subsequent fused pair's neural spines. Greg Paul ignores this feature but does at least half-bake the gap, while Scott Hartman totally omits both of these very distinctive features. But they are natural and can't solely be attributed to crushing.

The point is that the dorsal column as a whole needs to be reworked. In fact the dorsals of HMN SII are a lot less complete than often believed.

Hey, at least when you bother to include HMN fund no 8, you only are missing one dorsal in the sequence to make it a full count of 12. Note hoe the neural spines shorten so radically between D4 and D7. This is quite a bit different that you see in currently existing skeletal restorations, which ignore D12 from HMN fund no 8, and bump D7 back to the D8 position to make things look a bit more gradual (adding an imaginary middle dorsal in the process to keep the count at 12 vertebrae). We can see that if you follow Janensch's explicit instructions (which even he failed to incorporate into his full skeletal recon) then such a position is no longer tenable.
So even though almost the whole dorsal series is present in some form, many of the neural spines and arches are broken and missing,and even the centra show a lot more variation than Paul or Hartman restored. Even the new revised mount in Berlin omits a lot of these details in its sculpted replica bones. So the spine will have to be radically revised. Just how radically? Take a look at this:

HMN SII + HMN fund no 8 (D12, scaled down by 10% to SII)

This is even more bizarre than previously thought. With just the baseline amount of de-crushing necessary to make the vertebrae articulate, so that we avoid unnecessary artificial distortions, the spine is kinked at both ends of the by-now-notorious Dorsal 9. Even it you ignore the pathology argument (and you probably should, since D9, D10 and D11 are all very symmetrical, with no anomalies in lateral curvature), the odd shape and angle of D9 is even stranger than even myself and Asier Larramendi had restored it. While the gap between it and the fused pair is now smaller in the neural spines (which makes sense since the tips of the spines in D9 and D10 almost interlock at this angle), the gap below the zygapophyses (which the spinal cord would have run through) is still gigantic. Woe betide any young Giraffatitan that got bitten there.

Another interesting feature is that there seems to be another dip between dorsals 3 and 4, (or rather an upcurve of the anterior dorsals at D3) which may mean that the tall neural spines in this region came out looking less hump-like than traditionally depicted, and the spine profile of the live animal may have actually been more of a straight incline. And this would clearly make the angle of the anterior dorsals steeper and the neck even higher and more vertical... without having to add an insane amount of upward kink at the base of the neck the way Research Casting International did for the updated Berlin mount of Giraffatitan. I suppose it was easier to alter one joint than redo four of them, but then again closer attention should have been paid to how they reconstructed those other anterior dorsals in the first place. Janensch wasn't making up the shape of the cotyles, and D3 and D4 show very little vertical crushing. There should actually be an upcurve at D3, not a downcurve or a hump.

Even Greg Paul's new 2010 version doesn't come close.

There is still a slight hump in the soft tissue there (which looks excessive anyway) and the tips of the neural spines definitely form a hump. But the centra form a straight line. If they were restored as per Janensch's engravings (and dorsals D3 and D4 are not crushed, so there's no need to "straighten" them out) then D3 and everything in front of it would form a steeper angle and less hump without needing such deep nuchal muscles.

Note that D3 and D4 show almost no crushing in the centrum so the articulation angle even at Osteological Neutral Pose still results in an upward tilt of D3, which makes it easier for D2 and D1 to arch up by fewer degrees and still support a vertical neck, with a minimum of strain or flexion on any one joint, far less than in either of Greg Paul's versions or the updated Berlin mount. Take that, Kent Stevens.

So yes, I will be revising my interpretation pretty heavily. Giraffatitan may turn out to be a bit of a sail-back... in the same sense as Acrocanthosaurus.

Stay tuned for more updates, Giraffatitan's dorsals aren't the only weird thing about this beast.

Bwwaaaaroooooo! Atchooo!


Janensch, W. 1950a. Die Skelettrekonstruktion von Brachiosaurus brancai. Palaeontographica, Supplement 7 (I, 3):97-103.

Janensch, W. 1950c. Die Wirbelsäule von Brachiosaurus brancai. Palaeontographica, Supplement 7 (I, 3):27-93.

Paul, G.S. (1988). "The brachiosaur giants of the Morrison and Tendaguru with a description of a new subgenus, Giraffatitan, and a comparison of the world's largest dinosaurs". Hunteria, 2(3): 1–14.

Paul, G.S. (2010). The Princeton Field Guide to Dinosaurs. Princeton, NJ: Princeton University Press

Taylor, M.P. (2009). "A Re-evaluation of Brachiosaurus altithorax Riggs 1903 (Dinosauria, Sauropod) and its generic separation from Giraffatitan brancai (Janensh 1914)." Journal of Vertebrate Paleontology, 29(3): 787-806.