FORGOTTEN GIANTS, #2: Argentinosaurus

Posted by Nima On Saturday, February 5, 2011 21 comments

It has been a LONG time, dino-fans. But after SVP 2010, its aftermath, writing and working on tons of other projects, I thought I'd show you all something that's already been making the rounds on DeviantArt.

My second installment of the Forgotten Giants series, Argentinosaurus!

Argentinosaurus huinculensis is widely considered to be the biggest dinosaur. Though Amphicoelias fragillimus would have been far larger, and even today, Puertasaurus is the biggest dinosaur known from currently existing remains, that still doesn't reduce the fact that Argentinosaurus still holds many records.

* The largest femur currently known
* The tallest dorsal vertebrae currently known
* The longest tibia currently known (no, Bruhathkayosaurus doesn't count. Here's why.)
* Possibly the largest sacrum ever discovered
* The largest replica/speculatively sculpted fiberglass skeleton mount anywhere (there's one copy in the Fernbank Museum, another in Germany's Senckenberg Museum, and another at Museo Carmen Funes in Plaza Huincul, near the site of the dinosaur's discovery).
* The longest dorsal column of any titanosaur
* The largest (and possibly only) dinosaur to have a museum built on-site specifically to house its remains
* The biggest dinosaur named for a country

So it's almost a forgone conclusion that this giant would be #2 in my series. It's not really "forgotten" at all, but it's still a very important member of the titanosauria, and still not very well-understood. The remains uncovered and described by Dr. Jose Bonaparte and Dr. Rodolfo Coria are pretty limited, but the size of them is astounding. Vertebrae over a meter tall. A femur 8 feet long. And hips as wide a car.

Within a few years it had become an international sensation. Finally, Ultrasauros was out and Argentinosaurus was the new biggest dinosaur. By the late 90s, its name was well known, and it became the subject of a number of long-forgotten TV specials with badass music but dismally high narrator turnover rates.

Ahh yes, the 90s. Good times...

Argentinosaurus is today still famous as the "biggest dinosaur" despite Supersaurus being longer, Sauroposeidon being taller, and the legendary Amphicoelias fragillimus being far longer and far heavier. Even since 2005, when the even more massive Puertasaurus was discovered, Argentinosaurus is still the iconic "record holder" in the eyes of the public. So even though it's pretty popular and far from forgotten, it's still a very important titanosaur, so you knew I was going to give it the Paleo-King treatment sooner or later!

However, there was a bit of a challenge - though huge, the known bones of Argentinosaurus comprise only a small portion of the skeleton. Several dorsal vertebrae, a tibia, a referred femur shaft, a partial sacrum, parts of the ilia and a fragment of pubis. There's also a second referred femur which may be from Argentinosaurus but I've never seen it figured.

This scarcity of fossil remains hasn't stopped people from trying to fill in the huge gaps. The Fernbank Museum's Argentinosaurus mount in Atlanta, Georgia, which is made of mostly speculative fiberglass models along with a few casts of the original bones, is one such attempt. But it's got a host of mistakes in the speculative parts.

Fernbank Museum mount

Senckenberg Museum mount (Frankfurt, Germany)
Museo Carmen Funes mount (Plaza Huincul, Argentina)

These three mounts are all identical, and even have similar poses, but are photographed from different angles. See if you can find the following errors in them (there may be even more that I didn't catch):

* Abnormally angular nasal arch
* Lower jaw looks strangely too much like that of Giganotosaurus - huge side hole and all!
* Teeth are far too few and thin, appear to be made of wire
* Neck is too horizontal (probably a result of the designers and welders reading too much Kent Stevens...)
* Most neck vertebrae are just clones/recasts of one another; three in a row identical, then the next three...
* Cervical ribs are drooping down at the ends instead of overlapping snugly in double rows along the base of the neck.
* Ribs are excessively deep in the chest region to shorten far too rapidly as you go back to the hips - a 'triangular pan-pipe' rib cage profile not seen in any sauropods.
* Ilia are curved inward toward the backbone at their front tips, big mistake! In titanosaurs (and all macronarians for that matter) they are curved OUT. Titanosaurs had a wide belly, flared-out hips make sense. For that matter, the rib cage is also a bit too narrow!
* Sacrum is missing large portions of the front two sacral ribs (the original fossil was missing them too -why didn't the preparators account for this and reconstruct the missing portions?)
* Sacrum is not fused to the ilia - in fact there are huge gaps separating it from them. The reason for these garish gaps can only be guessed at.
* The radius and ulna are improperly aligned in both arms. They should be one in front of the other, not side by side like the tibia and fibula. A little knowledge of vertebrate anatomy could go a long way here...
* The hands lack phalanges and thumb claws. The reasons for NOT getting rid of them are actually quite strong for titanosaurs, especially basal ones like Argentinosaurus. Suffice it to say that even derived titanosaurs like Diamantinasaurus had thumb claws - they just don't get preserved in most titanosaur hands because of the loose cartilage connections in the phalanges, which dessicated easily, allowing the thumb claws to be washed away.
* The feet are far too large and have an excessive fourth claw. This is not found on most sauropods.
* The metatarsals and toes are arranged in an almost linear pattern, rather than the semi-columnar, circular pattern that they would normally articulate in. And yet we criticize Chinese museum workers for making the exact same mistake with their sauropods... sheesh.
* The ankles have an extra "squashed mound of dough" bone beneath the real ankle bones, the Astralagus and the Calcaneum (both of which are too large anyway). This squashed mass is not a real bone at all, nor is it based on anything found in real fossils. One can only assume it represents ossified cartilage, but the ankle cartilage of sauropods NEVER ossified!

Those awful feet ! Linear arrangement of the digits, extra 4th claw, no vestigial 5th phalanx, 4th metacarpal is mysteriously thicker than the 3rd one, ankle bones WAY too big to match the tibia and fibula, and the feet are practically the size of Texas for a dinosaur that was only 110 ft. long!
You could probably sleep under those feet at night!

As for illustrations of Argentinosaurus.... some are great, some are okay, and most are downright terrible. The terrible ones are too many to list. But here is a sample of some of the better ones out there:

Aldo Chiappe. While his version isn't necessarily the most accurate, I do like how he really gives it the feeling of taking up a LOT of space. Great shading and the texture just draws you in...

Unknown artist. Neck needs a bit of work, but the head is unmistakably macronarian.
Illustration from the "Gigasaurier" German exhibition at the Senckenberg Museum (artist unknown). Neck is far too horizontal, but the head is definitely macronarian, which is a plus. Furthermore, I like the fact that the legs are not so "overbulked" with this one.

Fabio Pastori. Although I'm not a huge fan of his textures for either Argentinosaurus or Mapusaurus, he did get the basic shapes generally right.
Dinoraul. This is a 3D model, that is apparently available for sale. Probably the best 3D render of Argentinosaurus, though the neck seems a bit short and stocky. The head is excellent, nice big nose, but I can't help noticing the head's remarkably familiar "squashed" appearance - I suspect this guy must have ripped off my Puertasaurus' head in some capacity... (I'm NOT making any money off of this model, I guarantee you that).

Steve Kirk. This painting was done for one of the later editions of Dougal Dixon's Dinosaurs, which was probably one of the best dinosaur books a kid could ask for back in the 90's (the early editions didn't include Argentinosaurus). Very good size effect, if not the best leg/neck proportions. Thumbs up for the vertical neck! Unfortunately I can't post a larger image, but here it is if you want to see it close up.

Unknown artist. Nice long neck, though too horizontal, and the torso seems too short for Argentinosaurus. The pose, the juvenile, and the incorrect diplodocid-like head, give away that this artist probably copied Mark Hallett's famous Mamenchisaurus painting to some extent. A wonderful painting by the way, though outdated.
Greg Wenzel. He's is definitely working in the style of Greg Paul. Very nice and all around realistic painting of Argentinosaurus. Good skin texture, nice plant life, long semi-vertical neck, and macronarian head typical of basal titanosaurs. Skin color is a bit bland, but at least the shape and perspective have no problems. I just wish there was a bigger version online...

Greg Wenzel, different portion of the same painting. Though this scan is of very bad quality and distorted color, the painting itself looks pretty good anatomically. Nice big belly, long torso, legs not overbulked, and a long, near-vertical neck. The flock of birds really gives an idea of this animal's huge size and scale.

Wikidino image by Timothy Bradley. These digital paintings of many dinosaurs are all over internet forums. Most are actually pretty good. Though the neck may be a bit too short, the posture is excellent. Large torso, boxy head, tail high in the air. Massive, but not overbulked. It LOOKS like an Argentinosaurus, errors or no errors.

But what of the scientific illustrations? I mean actual skeletal diagrams...

Well, there are only really two. Greg Paul's, and Ken Carpenter's....

Modified from Paul (1994) and Carpenter (2006), respectively.

It is not a mistake that Greg Paul's skeletal shows fewer bones than Ken Carpenter's - it was done earlier, just one year after after Bonaparte and Coria's description paper came out (it didn't have diagrams of a couple of the vertebrae, or the femur shaft, which was dug up and referred to Argentinosaurus later). Though why he left out the sacrum is a mystery, as it was in the original paper.

The main difference aside from the bones, is obvious - the shape. Carpenter's version has much shorter arms and a very short thing neck. It's almost embarrassingly small... it's basically a scaled up Saltasaurus! In 2006 there seems to have been a trend towards "re-sizing" Argentinosaurus. Paul's version was considered by many in the field to be too brachiosaur-like, with its long vertical neck and tall arms. Carpenter proposed a radical re-interpretation of Argentinosaurus based on very late, derived titanosaurs like Saltasaurus - which reduced the length of Argentinosaurus to only about 80-90 feet or so. But this revision is not as reliable as it seems. Saltasaurus (which lived in the Maastrichtian epoch) was one of the last titanosaurs to evolve, millions of years after Argentinosaurus (Cenomanian epoch) was long extinct. It's not very likely that the early titanosaurs looked exactly like the last ones. Additionally, Saltasaurus was a small species, far smaller and shorter than Argentinosaurus, and specialized for low-grazing like a diplodocid or dicraeosaurid. The skulls of other small, short-necked titanosaurs like Bonitasaura bear this out - the teeth have become crowded in the front of the mouth and very thin, much like Diplodocus, in a wild case of convergent evolution. But something as huge and massive as Argentinosaurus must have been a treetop feeder - and it would have needed a very long neck. This is true of most if not all of the big basal titanosaurs. But that doesn't mean all of the late stage saltasaur-type titanosaurs were little short-necked grazers. There were some, like Alamosaurus, that had far longer necks, longer arms, and grew much larger than Saltasaurus. Very long necks and long arms didn't die with Brachiosaurus - they survived well into the end of the Cretaceous.

Alamosaurus sanjuanensis. Modified (by Matt Wedel) from Lehman and Coulson (2002) It punches a hole in the notion that all late-evolving titanosaurs were small-armed and short-necked. Of course, that tells us practically nothing about more basal titanosaurs.

Rapetosaurus as well, had an insanely long neck, and was a rather small saltasauroid/nemegtosaurid. And of course there were already several non-saltasauroid titanosaurs known by the time Carpenter wrote his paper that likely had huge necks. Puertasaurus, Phuwiangosaurus, possibly Huabeisaurus.... Yet from all the available reference taxa, Carpenter chose to use little puny low-grazing Saltasaurus as the silhouette shape for Argentinosaurus. Now I know preparing a paper has a lot of time strain and difficult decisions to make, and Ken Carpenter is a very knowledgeable and busy paleontologist with a lot of research going on (I actually met him at SVP, the guy's immensely accomplished). But I just don't understand why he picked Saltasaurus as the ideal body morph for a far larger and more primitive animal. I would have used something a bit closer to the basal end of titanosauria or titanosauriformes, something whose dorsal vertebrae actually resembled those of Argentinosaurus a bit... like Euhelopus, Phuwiangosaurus, Paluxysaurus, Sonidosaurus, heck, even Giraffatitan would be useful here. And they all had very long, probably vertical necks. In subsequent years, Futalognkosaurus turned up, a transitional titanosaur more advanced and younger than Argentinosaurus but more primitive and older than Saltasaurus - and it's a huge beast, with a completely preserved neck that even most brachiosaurs would have envied.

Small neck? I'll show YOU! Futalognkosaurus dukei, from Calvo, et. al. (2008)

So maybe using a "brachiosaur-like" model for Argentinosaurus isn't so bad after all. Monster necks seem to have been the rule rather than the exception in titanosauriforms both before and after its time. And Greg Paul's version of Argentinosaurus doesn't look all that much like a brachiosaur. It's something different. It's actually not copied from any other animal the way Carpenter's version is copied straight from Saltasaurus.

I had a hunch that Paul's version was generally correct. But that long whiplash tail bugged me. It's too diplodocid-like, and even the whip-tails wound on later titanosaurs weren't so thin and limp at the end. Also, Ken Carpenter's version does have two major advantages that Greg Paul's doesn't - it's got ALL the bones currently known (though it heavily distorts their shapes) and it gives an idea of the shape of the pelvis. It's also got a bit of pubis, which wasn't mentioned in the original description paper, but which may be present in some grainy internet photos. This was the clue for dissecting the actual number and shape of the missing bones.

The original bones on display in Plaza Huincul, Argentina.

Most problematic of all was the pelvis - how do you restore it when it's only known from a broken central piece of sacrum and slivers of ilium and pubis that aren't even documented? You do it very carefully.

Argentinosaurus partial sacrum (ventral view) and tibia (medial view) ...taken (and mysteriously flipped) by someone on the net, from Bonaparte and Coria (1993)

That stripe-shaded region is the first sacral vertebra's centrum, which is badly eroded but still visibly much larger than the other sacral centra. The sacrum, even though missing most of the 1st sacral ribs, much of one side, and the entire 6th vertebra, is still as long and almost as wide as the tibia. A rear as wide as your shin? That's a really big butt. Probably the biggest butt on record when complete, though there are some serious challengers to that title.The rough patch at right is a sort of articular surface to both the ilium and the hip socket. Replicating the other half of the sacrum digitally was pretty easy.

 Next, adding the ilia and the 6th sacral vertebra, and fixing up the 1st sacral ribs.

Then, some modding, slimming the ilia, and cleanup of the remaining bones...

On second though, the wider ilia of the earlier version make more sense. They should be more flared out than the ilia of a brachiosaur, but a bit less than in a saltasaur.

Okay, that's the basic hip structure for now. Turns out that it was flipped backwards (the original paper has it the other way around and I didn't realize this when I had first found the diagram online) - no problem, I fixed it later as you will soon see. Next came the actual walking profile of the dinosaur itself. Despite the apparent similarity, I did not look at Greg Paul's version for direct reference here - I simply started my own from scratch (and facing the other direction).
 This early sketch needed a lot of work, but it's a good start. The 12 dorsal vertebrae may be too many. Most titanosaurs had 11, a few even had just 10. There's some droop to the tail tip, but its far less thin and whip-like than the Greg Paul version. Remember, early titanosaurs like Argentinosaurus were basically evolutionary nephews of Euhelopus and Brachiosaurus. They weren't even close to mimicking diplodocids like the much later saltasaurs did. And most titanosaur whip-tail endings that have been discovered are of the late-stage saltasaur variety.

 The drawing, shaded, wrinkled, and patterned. The top of the neck patterned similar to the way David Peters painted it in Don Lessem's book, Supergiants. Great book by the way (and beautifully illustrated), if you're interested in the biggest dinosaurs (or at least the ones that were considered the biggest back in the 90s).

From the original skeleton sketch and using Greg Paul and Ken Carpenter's skeletals as references, a rough skeletal diagram of the known parts took shape, with the left leg separated from the hips.

Then I refined it and shaded the missing portions....
 And fixed the sacral count to six.... sort of.

 First stage cleanup  - captions, bone diagrams, scale bar, and human figure. The dorsal count is also reduced to 11. Probably a more accurate number.

 Next, the tail was extended, and a front view of the dinosaur and the sacrum in ventral view (now correctly flipped) were added. I also added the sliver of ilium, which based on the sacrum's state of preservation, actually does appear to be the right ilium, not the left ilium as in Carpenter's skeletal. To get a good approximation of the sacrum's correct shape (and that of the ilia) from various angles, I referred to these diagrams:

Titanosauriform/titanosaur ilia and pubes, from Lehman and Coulson (2002). (Back then Isisaurus was known as "Titanosaurus" colberti.)

Sacra and ilia of Neuquensaurus, Saltasaurus, Opisthocoelicaudia, and strangely enough, Camarasaurus. From Salgado (1997). All of these hips show the ilia flared out at the front, and very strongly flared out in titanosaurs - unlike the warped speculative ilia of the Fernbank mount, which oddly curve inwards...

 Then the pubis was shortened and thickened a bit, using Ken Carpenter's rather fuzzy interpretation of the fragment, and the pubis of Andesaurus for some reference. Though the classification of Argentinosaurus as an "andesaurid" is increasingly under attack these days, it does have several basic features in common with Andesaurus, and is about the same age. So I did not make the pubis identical, just similar.

 Some lateral details of the vertebrae were altered using the description paper diagrams for anterior reference, and a posterior dorsal pic from Andesaurus for posterior reference.

 A wider torso in front view, and modifying a Wikipedia image of the partial femur for a front-view diagram. The picture was from a bad angle, but another one from a different angle in Mazzetta, et. al. (2004) helped give a fuller idea of the femur's 3D form. Also front and rear views of some vertebrae are added, based on the description paper's drawings.

 Next I made the limbs a bit thicker in both views and re-scaled the femur ends to conform more with complete titanosaur femurs. This diagram from Lehman and Coulson (2002) was very useful:

 Next, it became clear that many of the dorsal vertebrae still had mistakes. So I looked back to the description again:

 Anterior and three posterior dorsals. Cut and pasted from Bonaparte and Coria (1993)

 Eventually it turned out that the diapophyses got much taller and more buttressed at the rear of the dorsal column than I had though, and the neural spines also seemed to go retrograde and tilt forward in the posterior dorsals. So pretty much every detail of the vertebrae had to be revised.

Some cheap anonymous photos of the dorsals floating around on the net were also useful.

Also the sacral neural spines were revised, to look more natural and closer to the fossil (the only halfway decent view of the top of the original sacrum is this):

(Notice the bit of right ilium on the sacrum. Another fragment of it lies in front of the near end of the femur.)
The sacral spines are very squashed, so in life they would have been a bit taller. Though maybe not as heightened as the ones in the Fernbank replica: 

Yeah, those sacral spines look a bit too tall... like almost diplodocid-tall... I still can't get over the fact that the 1st sacral ribs are 90% missing in this cast (hence the huge unnatural gaps in front of the sacrum), the 6th sacral is missing entirely, that the sacrum isn't flush with the ilia anywhere along its length, and that the speculative (and somewhat too long) ilia are curving in at the front instead of out. It's like they just casted the sacrum raw, without restoring any of the missing parts and worn edges first, and wrung the ilia through a giant vise! Who gave the orders to do this horrible mutilation of such a grand dinosaur?

So in the end, after all the revision and rants, you get massively revised buttressed dorsals, and a scalloped sacrum ridge with some cool wavy lines. And one more rear view of a vertebra. With every possible bit of visual data extracted from the 1993 description paper and every photo available on the web, I'd say it's done!

 This is it, the final product.

Feel free to comment on any part of this process. Or, feel free to challenge my arguments if you're one of the sculptors for the Fernbank mount. I'd really appreciate their side of the story. Enjoy :)

P.S. I've never heard of the skull of Argentinosaurus being found, but there is a mysterious skull that has popped up on the internet from time to time, that has no name but is rumored to be Argentinosaurus.

If anyone has information on what this strange skull really is, by all means let me know.


Zach Armstrong said...

First off, the majority of titanosaurs that we know of had 10 dorsal vertebrae, not 11. Secondly, Argentinosaurus is by far the largest dinosaur excluding Amphicoelias fragillimus and the questionable Bruhathkayosaurus. Puertasaurus is definitely smaller than Argentinosaurus...

The Argentinosaurus painting that you said is in the style of Greg Paul is by Gregory C. Wenzel, as is the painting right below it, which are actually cropped portions of the same painting. The painting appears in its entirety in the first edition of "The Complete Dinosaur" edited by James Farlow and M.K. Brett-Surman.

You have done a great job on your life reconstruction, as usual, and you are definitely a superb artist. That being said, I disagree in a number of aspects with your partial skeletal restoration, like dorsal vertebrae count, order of the dorsals, restoration of the sacrum and ilium, etc.

The Editor said...

The artist of the wikidino picture is Timothy Bradley,he made over 400 picture like these. You can see his page here:
I have been searching the web for his art for a long time. It was originally on the Jurassic Park Institute page, but when that site was closed down it was moved to a yahoo kids page. Later, that too was shut down. The picture have been scattered all over the internet.

Nima said...

Thanks for the compliments Zach. I'll update the Greg Wenzel references. That's a really good painting, pretty much how I always expected Argentinosaurus would look. Just a few things I'd like to point out:

Is that "majority" of titanosaurs made up of mostly saltasaurs/derived lithostrotians? I don't know of anything as basal as Argentinosaurus (or slightly more basal) that has 10 dorsals. Indeed most basal titanosaurs are not known from complete dorsal columns so you can't make that call. Phuwiangosaurus is, and it had 12 dorsals. And 12 is the basal condition in titanosauriforms, being the standard in Brachiosaurs and "Euhelopodids". Between their time and the Maastrichtian there's a huge gap, and I doubt that titanosaurs went from 12 dorsals to just 10 overnight. There were in all likelihood some mid-Cretacoeus forms with 11 dorsals. If someone finds a complete Andesaurus or Argentinosaurus torso, we'll know for sure. Even among derived forms, the pattern is not constant, Bonitasaura is currently restored with 12 dorsals. So the case for Argentinosaurus having just 10 is not as strong as it sounds, considering most complete titanosaur dorsal columns are saltasaurs from the Maastrichtian, far removed from Argentinosaurus. Argentinosaurus is not a Saltasaurus clone, Dr. Carpenter....

As for Puertasaurus's size - it's bigger than Argentinosaurus whether I like it or not. My schematic estimated it at 135 ft. long. Even if you scale it down by 10% (which some have proposed based on the bones in their present state) it's still longer than Argentinosaurus. Scale it down by 20% (far too small IMO, but whatever) and it's nearly the same length as Argentinosaurus (108 ft.) and still considerably wider (therefore more voluminous and massive). And that's still assuming pretty conservative tail proportions Even scaling down by 20 or 10%, you could still achieve lengths of 130 ft. by giving it a longer, more typically titanosaurian tail.

If you mangle things even more, and if, as you claim for lognkosaurs in general, the anterior dorsals were not necessarily shorter than the posterior ones,reducing the dorsal column length of my Puertasaurus by 20% or more would still yield a comparable volume to Argentinosaurus, as the torso is far wider. If it's the width you have a problem with... I can't help you there - read the paper and take another good look at those huge wing-like diapophyses.

To be fair, your Puertasaurus a few months back had similar dimensions to mine, and you also estimated it was bigger than Argentinosaurus with some pretty complex maths, so I don't see the point in shrinking it down to look like a scrunched up mini-accordion with a barely any stride room between its arms and legs :P

Leo said...

Hi Nima and compliments for this brilliant post! I truly have nothing to add but this note: the Dougal Dixon Argentinosaurus is not by D. Dixon himself, but Steve Kirk!
By the way, it seems that a pterosaur by S. Kirk was featured (officially? I doubt) on the poster for Steven Spielberg's forthcoming project called Terra Nova. Check it out on my blog:


Zach Armstrong said...

The “majority” I refer to is indeed mostly composed of saltasaurs and derived lithostrotians. However, that is not the point. My problem, Nima, is when you say something like "The 12 dorsal vertebrae may be too many. Most titanosaurs had 11, a few even had just 10." The fact is, the majority of titanosaurs for which we have complete dorsal columns had 10 vertebrae. Your original statement (and my reply) had nothing to do with the perceived derived-ness or basal-ness. You simply said that “most titanosaurs had 11”, which is, at best completely unsubstantiated by evidentiary support, but at worst is in complete contradiction with the facts. And, if you notice, the initial part of my comment which you refer to was not directed at your reconstruction of Argentinosaurus but was directed at your statement that said, "The 12 dorsal vertebrae may be too many. Most titanosaurs had 11, a few even had just 10." My differing opinion in how Argentinosaurus should be restored was not directly related to the first part of my original comment and so the basal-ness of derived-ness of Argentinosaurus was not directly relevant to that.

The fact is, only one titanosaur, Opisthocoelicaudia (a derived lithostrotian to boot!) was originally described with 11 dorsals and one of those (the first one) might actually be a posterior cervical instead. Except for Phuwiangosaurus, which is known from 12 dorsals, the rest of known titanosaurs had 10 (taxa that are incomplete in the dorsal column, such as Bonitasaura , should not factor into the discussion).

I also think you misunderstand, or at least misinterpret, how dorsal vertebrae develop in an evolutionary context. Just because Phuwiangosaurus had 12 dorsals and derived titanosaurs had 10 dorsals does not mean there had to be an 11-dorsal "intermediate" in between. Standard evolutionary theory does not require such a linear pattern and in fact often contradicts such expected patterns. It could be that titanosaurs went from 12 to 10 dorsals right away. The thing is, we don't know. We need more fossils. Even if there was an intermediate step with 11 dorsals, this does not mean the majority had 11 dorsals.

You said that “There were in all likelihood some mid-Cretacoeus forms with 11 dorsals.” Even if true, that does not support your statement that ‘the majority had 11’. In fact, Malawisaurus had 10 but was from the EARLY Cretaceous. On the other end, Futalognkosaurus, from the LATE Cretaceous also had 10. So there is no consistent temporal distribution in the number of dorsal vertebrae, which further undermines your statement. If you wanted to insist on your statement that the majority had 11 dorsals, you should have prefaced that with “In my unsubstantiated opinion, the majority of titanosaurs had 11 dorsals because…” That would have been better, but still misleading.

In fact, Argentinosaurus is probably not as basal as commonly assumed, which further bolsters support that it probably had around 10 dorsal vertebrae.

Zach Armstrong said...

As for the size of Puertasaurus compared to Argentinosaurus, I have a few things to say.
In your reconstruction, what I get using GIMP’s measuring tool is that the 4 meter scale-bar is 139 pixels, and the distance between the pre- and post-zygapophyses of the 9th cervical is 52.5 pixels. This means the cervical is about 151 cm between the pre- and post-zygapophyses. The actual length as measured and given in the description of Puertasaurus is actually 118 cm. So the length is 28% longer than it should be. The total length of Puertasaurus , as portrayed in your drawing, is about 43.8 meters long from tail tip to the tip of the skull, measured along its vertebral axis using the provided scale bars. However, the guy for scale appears to be 6 ft and 1/2 inch tall based off of the scale-bar. The guy for scale is Eugene Sandow, who was actually 5 ft 9 and 1/4 inches tall. So the scale bar is about 4.5% too large from what it is supposed to be, meaning, as reconstructed, the Puertasaurus is about 41.88 meters long (this is close to your estimated length of 135 ft based on your reconstruction, which is about 41.15 meters). Even taking that into account, the cervical vertebrae is still ~22.5% too long. This means, accounting for that fact, the actual length is 1/1.225=81.63% of what one should get based on the drawing. So the actual length would be ~34 meters. This would reduce my original mass estimated based off of your drawing of 136.6 tonnes to (0.8163)^3=54.39% of my original estimate to about 74 tonnes, which is about 10-15 tonnes smaller than an estimated mass of 85-88 tonnes for Argentinosaurus (that is, assuming Argentinosaurus was roughly comparable to Malawisaurus in dimensions. If the latter was 2.52-2.59 tonnes, and the comparable vertebral elements in the former are about 3.24 time as large in linear dimensions as the latter, then it was about 34 times more massive for an extrapolated mass of about 85-88 tonnes).
Once, I re-do my Futalognkosaurus skeletal (again), I will do a hypothetical reconstruction of Puertasaurus and we’ll see how big each of ‘em turns out. The overall dimensions of the vertebrae in length and height are smaller in Puertasaurus than in Argentinosaurus, meaning that Puertasaurus was only larger in one of the three dimensions, while it was actually smaller in the other two (the height of the dorsal in Puertasaurus was 106 cm tall; even the incomplete dorsal in Argentinosaurus were 115 cm tall, with the projected height of a complete one at about 159 cm tall; the functional length in of the dorsal in Puertasaurus was about 30.3 cm compared to a functional length of the shortest dorsal vertebrae in Argentinosaurus of about 42.3 cm ) which means we should predict an overall smaller animal.

Nima said...

Yikes, looks like another dead discussion...

I appreciate your efforts with the math Zach, but the fact is, whenever you are scaling dinosaurs based on a couple of bones, the results are bound to be problematic at best. The real message to take home here isn't that Puertasaurus was 34m long or 40m long, or 74 tonnes as opposed to 90 or 100 tons. It's that estimating size of a dinosaur with such a paucity of material is by its very nature complicated and a few simple scaling formulas could still produce results that are way off.

I favor an upper-limit length of up to 41m based on my restoration. It may have only been 34m or as much as 43m. That's not the point here, because so little of the skeleton is known. What is known as that this creature definitely rivaled Argentinosaurus in size, and in my view was larger, and I made an original attempt to draw the thing. If Sandow and the scale bar need a few adjustments so be it, but I think Puertasaurus will still come out larger. I especially doubt your using Malawisaurus as a scale reference for Argentinosaurus's proprtions - Argentinosaurus, to me anyway, seems to be proportionally less wide relative to its torso length than Malawisaurus or any lognkosaur for that matter, and also proportionally less wide in the sacrum (which is easily evident from the shape of the preserved sacrum)... if it had lognkosaur proportions given the length of its dorsal column it would indeed be the biggest dinosaur - but that would mean a super-wide rib cage and huge wide diapophyses like Malawisaurus, Mendozasaurus, or Puertasaurus - and that simply flies in the face of the actual fossil evidence for Argentinosaurus. It was creature with a thinner (but possibly more elongated) body.

Nima said...

Long story short, Argentinosaurus probably had radically different proportions from Puertasaurus (and other lognkosaurs), and that's pretty much ALL you can really prove from the few known bones without extrapolating or inferring anything additional. My inferences are more visual and yours are more theoretical. Personally I'm skeptical of any purely numbers-based approach, there are so many ways to do a double integration volumetric model and different options for the masses, so if someone says Argentinosaurus weighs 80 tons today and 50 tons tomorrow, it really makes no difference in the long run because they are just re-crunching numbers without making a better restoration or finding more data. And it's not going to sway my thinking either way because I don't know every detail of how they were interpreting the fossil evidence OR my restoration or any other reference. Your mathematical winds have blown many ways over the past few months, and I'm not that fanatically dedicated to casting aside everything I know and sailing to the tune of any one of them.

All the same, 34m still beats Argentinosaurus's 33m. And with the masses, it's also pretty tough going. My Puertasaurus seems to have an oversized cervical mainly because I accounted for the fact that the cervical is heavily eroded and broken in several places, and I reconstructed it a bit differently than Dr. Novas did. Based on other titanosaur cervicals it seemed likely that the entire bone was a bit larger than it appears, especially the rear, which probably had more to it than the portions that Dr. Novas's team restored with plaster. Sure you can disagree with almost any aspect of my restoration - it is by necessity largely guesswork after all - but you can't "disprove" any of it by simply making calculations about the whole body of an animal based on a few bones and undoing a slightly oversized scale bar.

BTW, if you want to mock me and call things "unsubstantiated" and "misleading" on my blog, there are easily a hundred ways I could call many aspects of your sauropod restorations "unsubstantiated" and "misleading". So far I haven't, because insulting others is generally not something I'm fond of, but if you like, that could easily be arranged. Your stuff has a lot of dodgy anatomical guesswork and purely theoretical numbers that you try to pass off as absolute fact (a tactic which isn't acceptable even if some of your numbers happen to be close to correct). The choice is yours :)

Matt Martyniuk said...

Great post, but why criticize any restoration with a more horizontal neck? Animals move, and surely such a pose would not only have been possible but necessary, as these things did have to drink. While it may not have been "habitual" or "osteologically neutral" (whatever those are supposed to be) doesn't mean it was never exhibited in life (and I suspect in the case of some of these mounts, it's necessary to fit the thing indoors!).

I don't have much to say in your discussion with Zach but calling data "unsubstantiated" or "misleading" is in no way a personal insult, nor should it be construed as mocking, IMO. It may simply mean you haven't provided support for your claims in a clear enough manner.

Nima said...

Thanks for the input Matt.

I'm not bashing horizontal necks like it's any matter of principle. But most horizontal-necked restorations have nothing to do with drinking at the water hole. I have no problem with the inevitability that macronarians would sometimes have a horizontal neck for a few seconds, but with most paintings it's hard to tell if the neck is in motion. If it's a painting of a herd of macronarians with some in vertical pose and some lowering their heads at a river, that's more realistic.

I was mainly disagreeing with habitually horizontal-necked paleo-art that's inspired by either SNAFUism or outdated misconceptions of sauropod neck structures, not images that show an occasional lowering of the neck to drink. BTW, if macronarians gathered around waterfalls, they would not have to lower their heads to drink.

To be honest I usually don't take things personally, and I'd like to agree with you - but this behavior has regretfully been a pattern. I've had to tell people commenting on here to cool it several times before. I try to make the posts here real jam-packed with facts instead of fluff, so I'm hoping to get a nice large spectrum of input and ideas on here with most posts, but more often than not what happens instead is one or two people starting World War III over a couple of poorly understood technical issues and probably scaring off anyone else who felt like commenting. Furthermore it isn't some impersonal data that got labeled "misleading", if you follow the flow of the comments it reads like it was contextualized in far more personal terms. I admit my statements about Argentinosaurus dorsal vertebra counts have a dose of guesswork, but that's because no complete dorsal column is known for any basal species very close to Argentinosaurus (as opposed to several known for derived Lithostrotia). So restoring it with just 10 dorsals would be an equally big assumption, no matter how many Maastrichtian saltasaurs you dig up. Now if people found a complete ARTICULATED Argentinosaurus postcranial series and it had only 10 dorsals and someone said there must be 11, THAT would truly be misleading.

Anonymous said...

Thanks for the excellent post Nima, really interesting demonstration of your process.

Ezequiel said...

Hi Nima! Great post! this is far from what I can appreciate, since I work with fossil ferns (well, maybe Argentinosaurus eated some of them...).

Regarding the Illustrations, the first unknown (the one with the man at the right) is from Jorge Blanco (here is a page in ... german? of him, with a sketch almost identical to the illustration in the blog), the next unknown (Gigasaurier) I'm almost sure it was made by Jorge Gonzalez (, he has another Argentinosaurus on that page), and the one after Dougal Dixon is from Lucas Fiorelli (, the three Argentinian illustrators.

Best wishes!


Zach Armstrong said...

My point on the size, Nima, was not to dogmatically say that Puertasaurus was 74 tonnes vs. 136 tonnes. My point was that, when we compare the fossil material, it is not clear that Puertasaurus was larger than Argentinosaurus, as I demonstrated by comparing measurements of the vertebrae. If you want to argue that Puertasaurus was bigger, fine. But WHY do you think it was bigger? To me, the actual fossil evidence does not indicate that it was bigger. True, we only have one dorsal vertebrae and one cervical vertebrae to go off of so anything we can say is going to be at best tentative. My problem is when you appear to suggest that it was definitely bigger--the truth is, the actual evidence we have is ambiguous to say the least. The math and resulting numbers I demonstrated were not meant to be taken as gospel truth, but to challenge your assumption. You can criticize my methods all you want, but do you honestly have any better way to compare size? Your reconstruction, as well as mine, is a hypothesis, not a demonstrated fact. IMO, it would've been better to say that Puertasaurus would be bigger than Argentinosaurus assuming the reconstructions turn out to be accurate as feasibly possible with the limited info we have. As I showed above, I think your Puertasaurus reconstruction is overly optimistic when it comes to the size of the preserved elements as portrayed.

As for the comparing Malawisaurus to Argentinosaurus, you are right--to a degree. Malawisaurus may not be a good model to scale for Argentinosaurus and I have repeatedly qualified my extrapolated mass estimate for Argentinosaurus off of Malawisaurus with something like "assuming they are comparable in dimensions". I try my best to be explicit when I make certain assumptions, and I feel that you should be more explicit too, however that will obviously be a matter of personal choice and opinion to do so. Argentinosaurus probably was not built like Puertasaurus, but the fact the the vertebrae of the former are taller and longer than the latter would suggest that, overall, it was bigger in volume even if more slender in build.

You say that your inferences are more visual and mine are more theoretical. That is not really true--since I have created my own multi-view reconstructions (so that is visual) AND I have done my best to work with the numbers. The two are not, and should not be, mutually exclusive. Using either by itself is bound to get inaccurate results which is why I prefer to use BOTH methods together to complement each other and hopefully make a more true to life reconstruction.

I find it intriguing that you say that to accept my mathematical ideas would cause you to throw away everything you know. Graphic double integration is a proved and tried method of getting mass estimates better than any other mass estimation method (but it still has an error of up to +/- 20%) for extinct animals. I'm not making this stuff up. I would like to know how YOU get your size estimates, because as far as I can tell, I cannot repeat them—something which is a prerequisite for scientific thinking and logic.

I highly doubt that the vertebrae were so eroded as to be restored as 22.5% longer than they actually are preserved. You justify it by saying it is significantly eroded. That is true for the neural spine and posterior part of the centrum. But the pre- and post-zygapophyses (for which my measurements in your drawing were based on) appear in tact and erosion of those elements is certainly not described in the paper. Thus, restoring those larger is highly conjectural which means it is unfalsifiable. This means using it as a basis for a reconstruction to show it is bigger than Argentinosaurus is not really based on any solid evidence--which was my point originally.

And actually, if you scale it off of Malawisaurus proportions, you get an estimated length of about 37 meters for Argentinosaurus, not 33 m.

Zach Armstrong said...

BTW, I was not mocking you or degrading you when I called your assumption of 11 dorsals unsubstantiated and misleading. That's just a fact, as I demonstrated earlier--unless you can provide evidence for the majority of titanosaur taxa having 11 dorsals. Otherwise, your position is unsubstantiated. It would've been much more accurate and truthful if you had said basal titanosaurs had 12 and intermediate and derived ones had 10, so you hypothesized it had 11 because it was more derived than Phuwiangosaurus (which had 12) and more basal than Malawisaurus (which had 10). I would've been comfortable with that, but not with "the majority of titanosaus had 11 dorsals". What I said wasn’t to mock or insult you, but to disprove your reasoning.

You’re right I can’t “disprove” your reconstruction, because being able to disprove or falsify something is a characteristic of a scientific hypothesis or theory based on testable assumptions. Your assumptions do not appear to me to be, at least currently, testable because they involve too much speculation. What I can say is that your reconstructions are highly speculative and do not conform well to what we do know, and therefore are unlikely to be accurate.

If you want to call me out on my reconstructions and reasoning or insult me, go ahead. My recons are there and you can do your own GDI estimate, relate the results and potentially disprove them if they are inaccurate. I make clear my assumptions to as best I can, highlight uncertainties and room for disagreement. If you honestly disagree with what I have said above, then SHOW ME where I am wrong—not by appealing to personal preferences or distrust of mathematics—but by showing me scientific results that disprove what I have said (regarding, for instance the number of dorsals, their size, etc.).

I have never intentionally insulted you in any way. I HAVE strongly challenged your reasoning and your referenced data and have not minced words when doing so—but I didn’t do it to insult you or defame your character. If you honestly feel that is what I was doing, then I can’t change your mind. I disagree that my comments were “contextualized in far more personal terms”. I have great respect for you and your abilities and talents, which is why I feel comfortable challenging what I feel is flawed reasoning on your part. If you don’t like that, then I won’t comment on your stuff anymore. I’m not trying to make you look stupid or careless. What I AM trying to get you to do is get you to make strong arguments based on good reasoning, and challenging you when you don’t. That’s what happens in science. If you don’t like it, then I’ll stop visiting. The choice is yours.

Nima said...

@ Optimisticpainter: Thank you! I will try to post more things like this in the future. The process is just as important as the result.

@Ezequiel: Thanks, I will correct the references soon. BTW, extinct ferns are a very interesting area of study, I actually have a hard time deciding what plants to put in each scene I draw. So if you know what the most common ferns and other plants looked like in the major famous faunas (Morrison, Tendaguru, Dashanpu Quarry, Antlers, Gansu, Mangchuan, Huincul, Candeleros, Rio Limay, Pari Aike, Allen, Nemegt, Hell Creek, etc.) I'd be very interested to know.

As for Argentinosaurus eating ferns.... I don't think it was a fern-eater since most basal titanosaurs seem to be treetop feeders. But you never know, even bona fide treetop feeders like Giraffatitan could have bent down and gobbled up a few ferns here and there.

Well Zach, maybe you got me. Not in the best way, but I just might take up the challenge and re-scale my Puertasaurus and see how it stacks up against my other dinosaurs when I have the time. As for scaling Argentinosaurus based on Malawisaurus, I'm not surprised that gave a bigger estimate than my version, because Malawisaurus has a shorter torso relative to its body length than my Argentinosaurus. I don't think scaling up from Malawisaurus is even a reliable method for Argentinosaurus, but if it was a longkosaur then it would indeed be the longest titanosaur. It's just that it wasn't a lognkosaur and its vertebrae are totally different. The 20% margin of error on graphic double integration is a good point, especially considering what you said about my measurements being off.

There's a good possibility we'll never get the sizes right unless a lot more bones are found, but for now the best anyone can do is come up with their own models and illustrate them. To some extent it's curiosity. People just are burning to know what was the biggest in any category. Which is fine as long as it's not taken too literally and eternally when the fossil material is so sparse. GDI is obviously far from useless, it's just that it can wildly varying results depending on what you put in. This is true of any mathematical formula, as Greg Paul pointed out in 1988, because even a small difference in body shape can produce a very big difference in volume and mass. The best trick, I think, is to extrapolate what is easily indicated by the few bones present. For example, a very wide ribcage when you have huge wide diapophyses is not unreasonable. A very long neck when a mid-cervical vertebra is huge. A long torso when the dorsal centra are long. 11 dorsal vertebrae when you have 9 that are known (according to Ken Carpenter) and enough morphological variance in two anterior spots to reasonably accommodate two additional vertebrae with transitional features. I'm not saying it's absolutely right, but you can't say it's absolutely wrong or unscientific as opposed to using a very distant relative as verbatim reference. I try to read the bones and the trends they indicate, not just count them. I did this for many species for about 10 years before I started this blog, and I devoured a good number of papers and SV-POW material before starting it as well.

The strange thing is that the Fernbank mount looks to be around 37m, and almost everything aside from the dorsals is oversized, and even the dorsals are too many - 13 of them!

Zach Armstrong said...

I think scaling from Malawisaurus gives us a good general idea of its size, although maybe not as close a range as one would like. For instance, using the +/- 20% error (that is the maximum experimental error found for GDI; this was from oblique view diagrams of living/recently dead specimens; otherwise the error is around +/- 10%) , we could get a range of 68-102 tonnes assuming my lower bound GDI estimate for Argentinosaurus of 85 tonnes. This encompasses the mass of 74 tonnes I get for Puertasaurus based on a scaled down version of your drawing. So it is possible that Puertasaurus was bigger--certainly the uncertainty allows for it. I do not view it as likely, but that is why specific techniques like GDI can be useful to compare.

You're right that GDI will give wildly varying estimates depending on certain assumptions (e.g., ribcage width, tissue density, scaling errors, etc.). However, we can explicitly state these assumptions which means we can standardize comparisons and then can make educated guesses as to which assumptions are most likely. So, what is important is to minimize the amount of junk data we put in by making restorations as accurate as possible. One way we can do this is by making the least amount of speculation as possible. For fragmentary individuals, limiting the latter is a significant problem (as mentioned).

As an example taken from what you just stated: you think that diapophyseal width gives a good indication of how wide the ribcage is. That is a testable hypothesis. But instead of just saying that and restoring the width arbitrarily, try to quantify it. You have large hard copies of Greg Paul's skeletals, can you derive any pattern from diapophyseal width to ribcage width from them? My guess is no, but to date, no one has done the necessary work to verify or falsify that which puts any additional assumptions on very.

Same thing with the other things: you say a huge cervical means a long neck. Ok, good. But how long? Try to quantify it, show predictable patterns.

You said a long torso when dorsal centra are long. Ok, good. But how long? Again, try to quantify the trend and show predictable patterns that emerge.

Once you do that, then your research and reconstructions can be even more informative, and more importantly, it becomes falsifiable and repeatable which characterizes good research.

Leo said...

@Ezequiel: Please note that what you labelled "Dougal Dixon Argentinosaurus" actually isn't drawn by D. Dixon, but was an illustration made by Steve Kirk, featured in a book by D.Dixon, as I already wrote in an almost ignored previous comment here!


Ezequiel said...

@Leo: That`s true! my mistake! (it is now changed in the blog, too)

@Nima: one thing I shoud do is try to make a "compendium" of known floras for such typical ecosystems... If I do any progress, I'll let you know. And about sauropods eating ferns... well, you have Tree Ferns, mostly in the Order Cyatheales (e.g. genera Cyathea, Alsophila, Lophosoria) which reach considerable (even more than 15 meters/45 feet tall) sizes... and that may be of interest of an Argentinosaurus. At least one species of such plants has been recorded in Early Cretaceous floras of Patagonia, and similar taxa persist in the Oligocene of Neuquén province, thus probably Argentinosaurus met some of these big ferns (but who knows if he liked them, or preferred the classic Araucaria?).


PS: here´s a photograph of one tree fern

Anonymous said...

Fascinating and awesome life restoration, Nima. Zach has a lot of good points to make, but, as you pointed out, until better material is found, all this hubbub is over educated guesswork.

What is the chance that the largest individual of any of these species has been found? Do we know enough about the giant titanosaurs to venture a guess as to average size? With dinosaurs like pachyrhinosaurus, yes. But how many putative argentinasaurs and puertasaurs have been found yet? Are there any un-described bones that you might have knowledge of that might add to our fund of knowledge, as happened with supersaurus?

That would be interesting!

Anonymous said...

!!!!!!!!!!!!!!!!!! its huge like yo

Anonymous said...

Can I post your Argentinosaurs reconstruction at Wikipedia (either under CC-BY license or under fair use)?

MathKnight, a dinosaur editor and admin in the Hebrew Wikipedia

Post a Comment