Giraffatitan's weird head gets even weirder

Posted by Nima On Saturday, July 9, 2016 19 comments

In the last post on Giraffatitan, we focused on just how strange the head is, and explored some hints about the ontogeny of the animal's face.

However after a deeper exploration of the actual fit of the skull bones, dumping all stylizations and previous conventions of illustrating this iconic brachiosaur, a few things started dawning after being hidden and dissociated for mission of years. Giraffatitan is even weirder than I thought last time.

Not that it's easy to tell from three fragmentary skulls and a fourth that, while largely complete, has undergone massive distortion from crushing. That skull, HMN t1, which was reconstructed in the 1930s, was cast in fiberglass recently by Research Casting International (RCI) in their 2007 revamp of the Humboldt Museum's dinosaur hall - one that was long overdue. The cast was scaled up by around 15% or so on a 3D printer to match the body of the larger HMN SII, whose associated (and far less complete) skull SMN S116 was significantly larger than HMN t1. Apparently an earlier cast of HMN t1 existed as far back as the 30s and stood in a glass case in front of the old mount.

The problem, of course, with HMN t1 is the crushing and distortion, which results in an asymmetric skull today.

From the sides the distortion is more apparent.

 Left: moderate vertical crushing in the upper jaw. Right: more severe crushing in upper jaw, including lateral splaying of the lip region and artificial progmathism and splaying of the premaxilla and snout tip. This actually results in a different observable lip line on one side than on the other.  Of course the teeth are seriously falling out of their sockets here. They did not extend out that far in life.

Another problem is that the warping and crushing is in more than one direction, so that you are literally getting a different face looking at it from different angles. Judging the ideal "shape it should be" from a few photos at odd diagonal angles is asking for trouble. So how do you reliably uncrush this thing evenly, without photographic distortion on top of physical distortion, and get an idea of what the skull originally looked like?

Well you can go based on photos by amateur photographers from slightly off angles in a small cramped basement room, or go by professional drawings from the past, or use published photos. I prefer published photos from the paper, but for Janensch (1935) these are rather old and grainy, and I assumed a better result could be had from bigger, newer, sharper full-color photos, or from supposedly well-measured professional drawings of the skull in its hypothetical pristine form.

BIG mistake.

Initially the design for the Giraffatitan skulls in the skeletal redux went like this:

So I literally redesigned the skull four different ways before getting something that I could feel reasonably satisfied wasn't based on distortions and actually looked natural, like the interior of a living face, not some emaciated plastic toy. Something that didn't look out of place on an appropriately fleshed out Giraffatitan.

The first version was on the old Giraffatitan skeletal I posted. The drawing I used for inspiration (artist unknown) was rather grainy, and I ended up exaggerating the proportions and the shape of the teeth somewhat. On a 1950s brachiosaur drawing this head may have looked okay, but the shape of the nose and the jaws just seemed contrived based on what I had seen of the skulls - and the snout was a bit too beak-like in profile.

The second version came to me after hunting down a photo from a not-quite-profile angle on the web. Upping the contrast and then editing out the further premaxilla yielded a good snout profile, and this time with the nasal arch editing looking much better. The nasals of HMN t1 do appear a bit flattened so will need to be edited each time. Here the angle itself helped counteract the appearance of the crushed snout that plagues ride-view verbatim restorations.We end up with more robust jaws and a more believable gumline for a brachiosaur. But still, this image was based on a photo from an angle and so necessitated some distortion due to perspective as well.

Out of frustration some may resort to simply taking Janensch's drawing of a "de-crushed" composite skull as the true path. The problem here is that Janensch made a glaring error - the snout in his engraving is far too long. I shortened it a bit, but even then this version seems to shrink the nose and the rear skull and overgrow the snout and jaws. None of the Giraffatitan skulls have these proportions, they all reflect proportionally shorter jaws than that.

Finally a real edge-on profile photo of the right side of the skull surfaced on the internet. It was poorly lit and grainy, but it was the best profile available at the time - the picture was taken from some distance, so no "fish-eye" shape distortion, and also no angle distortion. Of course the crushing was still there, but now there was no extra visual illusion on top of it to undo. Rapidly this became a line-drawing, but then the flattened upper jaw and prognathic snout tip had to be corrected. With the jaws deepened to make up for crushing and possible erosion, the teeth back in their sockets, and the back of the skull at its proper proportions, this fourth attempt looked like the answer.

Based on it, I crafted the previous incarnation of the ontogenic sequence of Giraffatitan skulls, with some more modification.

Unfortunately, this assumed the other skulls were more or less identical to HMN t1. And it also utilized an excessive amount of morph change from the original despite compensation for crushing being necessary. A better photo was needed. Actually several better ones were needed for these skulls turned out to be unique individuals with different faces.

Looking closer at photos of the skulls, it became clear that this little happy family just looked wrong.

The largest skull, HMN S116, does not look like an exact copy of HMN t1. It is clear we are seeing one species here, but there is individual variation, ontogenic variation, and maybe even some sexual variation. Just looking at the individual skulls and fragments, this is not readily observable. But just wait til you try to reconstruct the gaps.

Pretty messy, pretty horrible. But let's clean up the process a bit...

How to draw accurate Giraffatitan skulls without going insane

It is often helpful to invert colors in MS Paint and work "in negative" - it allows you to avoid distracting and potentially artificial structures and visual illusions caused by too many changes between black and white regions. Now the process of following the skull photos much more closely than in the last set of reconstructions becomes very simple. The published photos from Janensch (1935) are rather grainy compared to more recent ones, but at least they were taken professionally, from proper lateral angles at a good distance, and thus can be used to make a skull recon while both removing crushing and avoiding the pitfalls of having to worry about camera angle distortion from amateur photos of the skulls (or of t1 anyway, since the other skulls have never been reconstructed or cast, and are off limits to the public). Reversing one side of the skull and overlapping it in Paint and Pixia allows you to get an idea of the relative crushing and distortion in different directions on both sides of the skull, and average their outlines to compensate for it. Some additional decrushing was also done with the snouts, which were all a bit more flattened than normal.

So in the end we have a rather different set of skulls than the speculative versions in the last post. Interestingly enough, the large HMN S116 has an absolutely huge nose, even by the standards of the more famous HMN t1. While the nasal arches are not preserved in S116, the enormous and massively buttressed shape of the upper maxillary process means that the nasals begin higher up on the skull than in HMN t1. In addition, the higher slope of the maxilla's upper surface indicates the nasal arch was also more elongated from front to rear (relative to the snout) than in t1. This overall indicates a nose that was oversized in all dimensions relative to t1. The lower jaw by contrast seems a bit undersized.

This can be easily explained as the result of ontogeny, as the large S116 - probably the same animal as the huge mounted postcrania labeled HMN SII - is actually still growing, its coracoids being unfused to scapulae, though it is still more mature than the smaller t1. However, there is probably more to this bulbous difference in nasal size than just ontogeny.

Note that the immature HMN S66, which is smaller than t1, also shares the large S116's trait of very large and tall upper maxillary processes and thus nasal bones that are rooted very high on the head. The nasal of S66 is flattened, which is to be expected as it has disconnected from the premaxillary (whose upper portion, making up the lower half of the nasal arch, has long broken off and disappeared) . However judging by the high-sloping upper surfaces of the maxillae in this specimen, the full nasal arch was likely also proportionally taller and longer than in t1. The fact that both the more mature S116 and the slightly smaller and (likely) less mature S66 have significantly more massive and taller upper maxillary processes and larger noses overall than t1, as well as a different shape to the maxillary processes altogether, indicates we may actually be looking at sexual dimorphism - perhaps with the large S116 and the much smaller S66 both being males, and the intermediately sized t1 being a female.

This possibility indicates that dimorphism in Giraffatitan could have progressed, at least in the skull, from a relatively young age. HMN SII/S116 was roughly 74ft. long, even with the substitution of the smaller correct tail HMN Aa for the oversized tail "HMN Fund no" used in the mounted exhibit. Judging by the unfused coracoids (and overlapping unfused scapula from the similar-sized HMN Sa9 - which may also be part of the same individual), the animal was likely a subadult, perhaps in its tens or early 20s assuming these animals took around 30 years to reach adulthood, which seems to be the indication in osteological sauropod studies. HMN S66, by a very rough estimate, was probably around 50ft. long, and may have been in its early teens. Unfortunately there has not been much histological work done on Giraffatitan to determine the ages of various specimens so these are speculations for now, but it is likely that if we are seeing sexual dimorphism in skulls, it probably began well before Giraffatitan reached physical maturity.

Of course, adult Giraffatitans (of which HMN XV2 and "HMN Fund no" may be examples) would have had even bigger heads. As these larger specimens, likely ranging between 85 and 90ft. long when alive, are not known from shoulder material, whether they are full-grown or not is impossible to determine. So the typical adult size of Giraffatitan - let alone its upper limit - is not determinable with any certainty, and neither is its maximum likely adult skull size. But we can at least scale up S116 to get a rough model of how big XVs's skull may have been.

Eventually thus we end up with an ontogenic sequence, which can be compared to other brachiosaurs known thus far:

Yes, those are some pretty huge skulls. And it makes sense, as they needed a big head, and especially a big mouth, to pack down all the food needed to grow to such huge sizes and beyond. HMN XV2 could have taken in 30gk in a single bite (though given how Jurassic conifer tendrils were built, much of each bite would have been air). And things get even stranger when you realize that even in the smaller HMN t1, the braincase was about 500 cc's, far larger than in many dinosaurs, and comparable to a chimpanzee brain, which is considered pretty large in terms of raw size. Nobody will ever see sauropods as "pin-headed" ever again.

So to recap, not all Giraffatitan skulls were copies of HMN t1. There is significant variation, enough to suggest a possible dimorphism in addition to ontegenic changes.

Giraffatitan's head is just WEIRD.

Posted by Nima On Monday, January 4, 2016 60 comments

The Jurassic weirdness continues!

The last post on Giraffatitan focused on the torso, and how it had often been inaccurately restored. But I didn't appreciate just how strange this animal's spine was until I got down to business, and started articulating images of the bones to see exactly how the centra and zygapohyses actually fit together. In the process I discovered that the actual 12th dorsal, though published by Werner Janensch in his 1950 monograph, was never scaled or reproduced in the mounted skeleton, nor was it used by any of the previous artists who had done skeletals of Giraffatitan. Greg Paul, Scott Hartman, Stephen Czerkas and or course (ironically) Janensch himself had left it out of their full-body skeletals. In addition it appears that they all changed the bizarre proportions of dorsal 9 - which has a relatively compact neural arch but a hugely elongated centrum - in order to make it fit in sequence such that the spine was more or less straight. But D9 (as heavily restored in plaster by Janensch anyway) has to be tilted upwards by around 40 degrees in order to have the short hyposphene reach far back enough to properly lock into place with D10, which indicates that their angle of articulation is anything but straight, and that D9 probably fits into the dorsal column like an upward-pointing wedge of sorts.

The result is a bizarre double-kink in the lower dorsals which both reinforces the lower back and makes the torso shorter and more compact. The fact that D10's centrum (again, going off of Janensch's restoration) has a condyle that is tilted up and back further reinforces this tilted angle of articulation demanded by the hyposphene of D9, as does the resulting snug fit of the neural spines of D9 and D10, without an excessive gap between them. Oddly Janensch doesn't carry over the weird shapes of both bones to his own full-body skeletal, though he does illustrate them individually in his paper, odd shapes and all, just it as his team restored them.

But this is far from the only strange thing about Giraffatitan that has been overlooked for decades.
Nearly every part of its body turned out to have unexpected features not included in ANY previous restorations. And one of the most commonly oversimplified, blurred, or just flat-out distorted parts in many restorations is... the head.

That's right, Giraffatitan's head is truly weird. A marvel of natural engineering and stress distribution through struts that in some places appear thinner than a human finger. The skull was light and hollow, yet could get up to a meter long (estimated size for adult individuals such as HMN XV2). And yet it was packed with big teeth resembling a cross between spoons and railroad spikes, built to crunch through hard branches high in the ancient conifers.

As you can see in the above picture, the skull is partially reconstructed with plaster, including one of the eye struts and the region just below the base of the nasal crest.

But there are in fact four skulls in existence. At least that is how many Janensch mentioned.

Three of them are missing a great deal of material, but the most well-known one, HMN t1, is nearly complete. We know this skull very well. Anyone who has seen photos of the Berlin mounted specimen (mostly based on HMN SII) has also probably seen this t1 skull, which is actually from a smaller individual. A scaled up cast of this skull was mounted on the skeleton itself in 2007, replacing an older crude sculpted skull.

The skull you see at the feet (or rather hands) of the Giraffatitan in these photos is actually only a cast of HMN t1. The real skull is stored in a museum vault and is (supposedly) off limits to the public.

Now at one point this face was cute.

Then it got fossilized and crushed. A few pieces such as the upper part of the eye socket are missing or broken. The upper jaw is partially collapsed in the middle, causing the sides of the maxillae to turn up and flare out. The sides of the jaws are thus artificially bowed out sideways. This led Dr. Matt Wedel to comment that it looks like a toilet seat today. Honestly I feel sorry for this poor creature. But sauropod skulls being delicate and easily smushed is a fact of life. Some of them had such loose connections between the skull bones that they actually dislocated during fossilization!

The crushing is easier to see from the side:


The snout has been flattened in the center, and to some extent the top of the nasal arch has also been squashed. Also notably, the teeth appear artificially long because they have slipped out of their sockets (or been pushed out by inclined crushing during fossilization) and the roots are visible. The skull itself had to be glued together from many fragments, and when first excavated was a bit of a jumble, like this:

Overall I would say given the shape the bones were in, Janensch and his staff had done a pretty good job of rebuilding the skull. Some of the skull elements were actually warped in the fossilization process, which makes sense as the bone layers are extremely thin.

But the trickiest aspect of this whole story is that there are a number of different ways the skull could have looked in real life. The crushing was uneven, which means the left and right sides of the skull appear rather different, with the right upper jaw considerably flatter than the left. Also we may be dealing with the possibility of ontogeny, that the skull of Giraffatitan would have changed shape with age and maturity. This is usually not a big concern in sauropods, as they do not develop any horns or massive butting surfaces on their heads, but that doesn't preclude the possibility that the shape of the head itself changed with age.

In trying to reconstruct a profile of Giraffatitan's head, I had to get around a few things.

First, the specimen I'm using for the skeletal is HMN SII, so the skull has to closely match the S116 skull, which is from the same or similar-sized individual. This skull has somewhat different proportions to some of the bones than HMN t1, although part of this may be due to either ontogeny or sex of the individual. That said, I wanted to create a reconstruction that adequately combines the most consistent aspects of all the skulls and eliminates crushing so that we can see the "ideal" morph of how SII's head would have looked on the living animal.

This was going to be a literal headache. It didn't help that Janensch and other early authors had themselves illustrated the "generic" Giraffatitan skull a number of different ways, with varying proportions.

So in brief, below, is the progress of morphs, trying to get the uncrushed proportions just exactly right (with a similar but shorter process for the referred Felch Quarry skull of Brachiosaurus - also an immature specimen - shown below it.)

By comparison with many photos from different angles and all the known Janensch engravings, gradually a more complete picture emerged. And so with a few remixes for different specimens, ultimately the conclusion was that the typical Giraffatitan head - hypothetically a mix of t1 and S116 - would look as follows.

So after about 30 variations and tweaks, this is what we've got. Overall a LOT better than the ugly derpy overbite version you see in most books and websites (basically a caricature of the crushed t1 skull), or for that matter the oversimplified blurry Greg Paul version which is sorely lacking in detail and deviates substantially from the fossils in several ways.

Pauly DERP that you can get sued for imitating. 

So yes, Giraffatitan - when uncrushed - has a rather different head than we've long been lead to believe. Feel free to comment below.