Argentinosaurus: smaller than you think?
April 15, 2010
Dorsal vertebrae from Argentinosaurus (center) and Supersaurus (either side). The vert on the left is the holotype of Ultrasauros, and the one on the right is the holotype of Dystylosaurus, but both of those taxa have been sunk into Supersaurus. Found on teh intert00bz.
As often happens here, a comment thread got to be more interesting than the original post and ended up deserving a post of its own. In this case, I’m talking about the thread following the recent Mamenchisaurus tail club post, which got into some interesting territory regarding mass estimates for the largest sauropods. This post was inspired by a couple of comments in particular.
Zach Armstrong wrote:
I don’t trust Mazzetta et al.’s (2004) estimate, because it is based off of logarithmic-based regression analyses of certain bone lengths, which a recent paper by Packard et al. (2009) have shown to overestimate the mass by as much as 100 percent! This would mean the estimate of 73 tonnes given my Mazzetta would be reduced to 36 tonnes.
To which Mike replied:
Zach, Mazzetta et al. used a variety of different techniques in arriving at their Argentinosaurus mass estimate, cross-checked them against each other and tested their lines for quality of fit. I am not saying their work is perfect (whose is?) but I would certainly not write it off as readily as you seem to have.
Weeeeell…Mazzetta et al. did use a variety of measurements to make their mass estimates, but they did it in a way that hardly puts them above criticism. First, their estimates are based on limb-bone allometry, which is known to have fairly low accuracy and precision (like, often off by a factor of 2, as Zach noted in his comment). Second, the “raw data” for their allometric equation consists of volumetric mass estimates. So their primary estimation method was calibrated against…more estimates. Maybe I’m just lazy, but I would have skipped the second step and just used volumetric methods throughout. Still, I can see the logic in it for critters like Argentinosaurus where we have limb bones but no real idea of the overall form or proportions of the entire animal.
Anyway, the accuracy of their allometric estimates is intertwingled with their volumetric results, so if their volumetric estimates are off…. The volumetric estimates used a specific gravity of 0.95, which to me is unrealistically high. Taking into account the skeletal pneumaticity alone would lower that to 0.85 or 0.8, and if the critter had air sacs comparable to those of birds, 0.75 or even 0.7 is not beyond the bounds of possibility (as discussed here and also covered by Zach in his comment).
Now, Mazzetta et al. (2004) were not ignorant of the potential effects of pneumaticity. Here’s what they wrote about density (p. 5):
The values from Christiansen (1997) were recalculated using a slightly higher overall density (950 kg/m^3), as the 900 kg/m^3 used in that paper may be slightly too low. Most neosauropods have extensively pneumatised vertebrae, particularly the cervicals, which would tend to lower overall density. However, these animals are also very large, implying a proportionally greater amount of skeletal tissue (Christiansen, 2002), particularly appendicular skeletal tissue, and consequently, they should have had a higher overall density.
This is pretty interesting: they are arguing that the positive allometry of skeletal mass as a fraction of body mass–which is well documented in extant critters–would offset the mass reduction from pneumaticity in animals as big as sauropods. I haven’t given that enough thought, and I definitely need to. My guess–and it is a guess–is that the effects of skeletal allometry were not enough to undo the lightening imposed by both PSP (~10%) and pulmonary air sacs (another ~10%, separate from the lungs), but I haven’t done any math on this yet. Fodder for another post, I reckon.
Getting back to Mazzetta et al., some of the volumes themselves strike me as too high, like ~41,500 liters for HM SII. That’s a LOT more voluminous than Greg Paul, Don Henderson, or Mike found for the same critter. The 16 metric ton Diplodocus and 20.6 metric ton Apatosaurus used by Mazzetta et al. are also outside the bounds of other recent and careful estimates. Not necessarily wrong, but definitely at the upper end of the current spectrum.
Mazzetta et al. got a mass estimate of 73,000 kg for Argentinosaurus, but (1) they used a density that I think is probably too high even if skeletal allometry is considered, (2) at least some of the volumetric mass estimates that form the “data” for the limb-bone regressions are probably too high, and (3) even if those problems were dealt with, there is still the general untrustworthiness of limb-bone regression as a mass estimation technique. 1 and 2, if fixed to my satisfaction, would tend to push the estimated mass of Argentinosaurus down, perhaps significantly (the effect of 3 is, if not unknowable, at least unknown to me). Given that, Zach’s ~52 metric ton estimate for Argentinosaurus is very defensible. (Probably worth remembering that I am a sparse-wing fanatic, though.)
None of this means that Mazzetta et al. (2004) were sloppy or that their estimate is wrong. Indeed, one of the reasons that we can have such a deep discussion of these points is that every link in their chain is so well documented. And there is room for honest disagreement in areas where the fossils don’t constrain things as much as we’d like. You cannot simply take a skeleton, even a complete one, and get a single whole-body volume. The body masses of wild animals often fluctuate by a third over the course of a single year, which pretty well buries any hope of getting precise estimates based on skeletons alone. And no one knows how dense–or sparse–sauropods were. I haven’t actually done any math to gauge the competing effects of skeletal allometry on one hand and PSP and air sacs on the other–and, AFAIK, no one else has either (Mazzetta et al. were guessing about pneumaticity as much as I’m guessing about skeletal allometry). Finally, Argentinosaurus is known from a handful of vertebrae and a handful of limb bones and that’s all, at least for now. If we can’t get a single body volume even when we have a complete skeleton, we have to get real about how precise we can be in cases where we have far less material.
The upshot is not that Argentinosaurus massed 73 metric tons or 52 or any other specific number. As usual, the two-part take home message is that (1) mass estimates of sauropods are inherently imprecise, so all we can do is make our assumptions as clear as possible, and (2) even the biggest sauropods might have been smaller than you think. ;-)
Reference
Mazzetta, G.V., Christiansen, P., and Farina, R.A. 2004. Giants and bizarres: body size of some southern South American Cretaceous dinosaurs. Historical Biology 2004:1-13.
Sauropod Vertebra Picture of the Week 
April 16, 2010 at 9:09 am
On the effect on mass of limb-bone allometry: Matt and I discussed this recently in a different context and he made the point that there is good an consistent evidence that sauropod limb bones scaled isometrically through ontogeny (i.e. the retained the same proportions, changing only in absolute size, as the animals grew up). Evidence is briefly summarised by Taylor (2009:796) as it happens —
“It is notable that the [Giraffatitan] juvenile left humerus HMN XX19 has a GI of 8.63, and so is as gracile as the humeri of adult specimens, corroborating in B. brancai the findings of Carpenter and McIntosh (1994:277) for Apatosaurus, Ikejiri et al. (2005:176) for Camarasaurus, and Tidwell and Wilhite (2005) for Venenosaurus Tidwell, Carpenter and Meyer, 2001 that sauropod limb bones, unlike their vertebrae, scale isometrically during ontogeny.”
But I pointed out that this in itself tells us nothing about how sauropod limb bones scale through evolution: it’s quite possible that, for a species B descended from species A, all humeri of specimens of A have the same proportions whatever the growth stage of the specimens, and all humeri of specimens of B have the same proportions whatever the growth stage of the specimens, and the proportions of A and B are different.
At least, it’s possible given what we know right now. To the best of my knowledge, no-one has ever studied this, so we don’t have data. BUT this would be a neat little self-contained stats-across-phylogeny project for someone with appropriate skills and available time. So if anyone out there wants to take it on, please do!
April 16, 2010 at 10:28 am
it reads “ultrasaurOs” at the bottom of the first picture
April 16, 2010 at 10:30 am
by the way, sure that was the smallest kid they found around to make the verts look even bigger
April 16, 2010 at 7:59 pm
Matt, it is interesting that you now mention the increasing allometry of the skeletal mass, as I had just assumed that wouldn’t affect the overall density. As the skeletal mass increases proportionally, I would assume the pneumatic areas would also increase along with the bones (i.e., the pneumatic fossae and foramina in large sauropod vertebrae do not appear to be proportionately smaller than in “normal” sized sauropods), canceling out the affect of the increased relative skeletal mass.
Quick question: how many dorsals are have been discovered for Argentinosaurus? In the original description there are 6 described dorsals, but in Ken Carpenter’s paper on Amphicoelias fragillimus, 9 are shown preserved, and in pictures I’ve seen of the preserved dorsal series at museums show 7 dorsals are known. So, does anyone know how many are known, since there are conflicting numbers going on here(and if so, if and where the other dorsals were described?)?
April 16, 2010 at 11:25 pm
It is pretty amazing to think that a 55 metric tons estimate is considered surprisingly small. Sauropods were crazy huge, especially these titanosaurs…
>>I would assume the pneumatic areas would also increase along with the bones
I’m sure it does. But there is pneumaticity in flesh as well as bone (to quote an earlier SV-POW post, “the volume of air inside the vertebra was dinky compared to the probable volume of air outside.” And bone-with-air-spaces will be heavier than flesh-with-air-spaces. This post says 0.63 g/cu cm for Alamosaurus pneumatic vertebrae; considering that we’re talking neck density estimates as low as 0.3 g/cu cm, adding more pneumatic bone is definitely going to raise the density.
April 17, 2010 at 8:26 am
it reads “ultrasaurOs” at the bottom of the first picture
No kidding.
As the skeletal mass increases proportionally, I would assume the pneumatic areas would also increase along with the bones (i.e., the pneumatic fossae and foramina in large sauropod vertebrae do not appear to be proportionately smaller than in “normal” sized sauropods), canceling out the affect of the increased relative skeletal mass.
I am tempted to assume that myself, but until someone does the math we’re just whistling in the dark. It definitely needs doing.
April 17, 2010 at 4:05 pm
Hmmmm…. this whole situation looks tricky. Limb-bone allometry does have some major problems, especially if it’s used “in the dark” without some kind of decent reconstruction of the animal’s non-skeletal form. It all depends if the sauropod in question is more proportionally “big-boned” like Apatosaurus, or “big-fleshed” like Brachiosaurus.
With Argentinosaurus, any such reconstruction is pretty difficult and speculative. If you use Greg Paul’s old skeletal (which contains only a fraction of the Argentinosaurus material now known, but has still held up pretty well over the years), the creature could easily mass 80 tons or more, especially if you restore the torso with a round or barrel-shaped cross-section (I’m aware this violates the “Sauropods were tacos, not corn dogs” rule, but then again titanosaurs seem to be breaking the conventional rules of sauropod physiology all over the place, and torso width is one possible example of this, as hip width definitely is). Ken Carpenter’s recent skeletal includes all the material but is far smaller than Paul’s, since Carpenter basically just copied and pasted the bones onto a blown-up Saltasaurus silhouette – a method that I have major disagreements with. Andesaurids and saltasaurids are opposite ends of the titanosaur spectrum, and in between these two extremes there are already creatures known to have far grander proportions (and necks) than Saltasaurus.
What’s really telling though is how much LESS “excavated” the Argentinosaurus vert is than the two Supersaurus verts. It’s like a block of concrete among wood carvings. Now of course we have to take into consideration the fact that titanosaur spinal pneunaticity largely took the form of internal camellae rather than the classic external camerae of Jurassic sauropods, but I actually think this made for denser bones overall and probably a denser sauropod. Also when I look at pneumatic bones of many animals today, it’s not all air in the internal bubbles of spinal bones, mostly it’s either fluid or marrow, both of which added some weight.
The main problem with the math so far is that there’s not enough of Argentinosaurus known, but one thing’s for sure, it certainly dwarfed Giraffatitan and had a crazy-long torso to boot. So with all that plus the (apparently) denser bones, I’m still leaning towards a bigger Argentinosaurus until more fossils turn up.
April 17, 2010 at 5:27 pm
Now of course we have to take into consideration the fact that titanosaur spinal pneunaticity largely took the form of internal camellae rather than the classic external camerae of Jurassic sauropods, but I actually think this made for denser bones overall and probably a denser sauropod.
Why do you think that? Is this a hunch, or do you have some data? From my own work, it seems that some camellate vertebrae were no more pneumatic than camerate vertebrae, but some–such as those of Early Cretaceous brachiosaurids–were a LOT less dense. And certainly I have not found any evidence that any camellate vertebrae were more dense than camerate vertebrae, as you suggest.
Also when I look at pneumatic bones of many animals today, it’s not all air in the internal bubbles of spinal bones, mostly it’s either fluid or marrow, both of which added some weight.
Really? This is not something I’m familiar with. What bones have you looked at, and in what animals? I’m not trying to be a jerk, I’m honestly curious. Some of the things you assert in this comment are strongly at odds with what I’ve found, and if I’ve been getting it wrong, I’d like to be pointed to the contradictory evidence.
April 17, 2010 at 8:48 pm
Matt, that’s extremely diplomatic.
For some reason, I am in an undiplomatic mood right now, so let me translate: Nima, you seem to be talking crap. Give us evidence or cut it out. We don’t want random uninformed speculation sprinkled over this blog as though it were fact.
(Matt, I assume that’s a fair paraphrase.)
April 17, 2010 at 10:31 pm
Oops, I did not mean my comment to offend or sound adversarial in the slightest. So sorry if anyone saw it that way.
The only “fact” I stated was that titanosaur verts are camellate rather than camerate. The rest was my opinion, where I said “I think” (I hope that came across as an opinion and nothing more). I did not discount the possibility that I may be wrong, so I think it’s a bit unfair to say I treated speculation as fact (And I don’t exactly consider said speculation uninformed)…
So I shall explain:
Basically I was under the impression that the density of camellate bones stacks up as you add the mass of the little walls of the camellae – as opposed to camerate bones, where there are huge empty regions with NO honeycombed bone, surrounded by extremely thin walls that themselves have some camellae within. Camellate verts like those of titanosaurs at least have a “puffed out” surface and all those internal camellae walls in areas where camerate sauropod verts have nothing at all. This Brachiosaurus cervical is just thin sheets of bone and huge gaping camerae and fossae, there is very little bone for the camellae to even take up (yet there are some camellae, and these lighten the bone further still).
https://svpow.wordpress.com/?s=brachiosaurus+sp.
Whereas in this Saltasaurus caudal, there are thicker walls of bone between the camellae, and this being a slice of honeycomb-like air-pockets, it seems to me that when all the walls of the camellae are added up as you go through the slices/scans, there is more bone density in there than in brachiosaurs.
Now if there was a scan of a titanosaur cervical available rather than a caudal, I admit this would be a much more valid comparison. One thing that is certain is that titanosaur camellae are not very large, this section of a Alamosaurus cotyle is full of them:
My hunch was basically that the bone walls of all those camellae added up will result in a denser bone than the paper-thin excavated bones of animals like Mamenchisaurus and Brachiosaurus. This was my honest inference, no intention to talk crap. And if I ever did feel an urge to waste people’s time or talk crap about something, paleontology (or any type of science) would be the last thing on my mind. Sorry if that’s what it seemed like.
One thing that really burns for an answer is that no matter how pneumatic the skeleton of any titanosaur was, that still leaves huge amounts of soft tissue that were not pmeumatic, not to mention several tons of fluids (which is a subject I’m not sure has ever been posted here or anywhere else – if there’s a post, please point me to it).
April 17, 2010 at 11:33 pm
Well I think I know what caused the confusion earlier. And the main clue was in Matt’s statement that “it seems that some camellate vertebrae were no more pneumatic than camerate vertebrae, but some–such as those of Early Cretaceous brachiosaurids–were a LOT less dense.” If you count the air sacs in the camerae of camerate sauropod verts, then it’s true that the camellate species were no more pneumatic…. but they were technically less dense if you exclude the big external air sacs from the density equation’s volume figure for the camerate types – which by necessity must be done since the actual volume of the complete external air sacs is impossible to measure.
When I envisioned the density of a brachiosaur vertebra, I considered the air sacs in the camerae as “part” of the vertebra’s structure, thus past of its volume, much as the internal camellae of titanosaur vertebrae are by default part of their volume – my bad for not mentioning that big detail. I just realized that since the full extent of the air sacs in a brachiosaur can not be known with certainty, the raw vertebra’s volume (without the external air sac volume) is what is normally used for the density equation. I’m guessing that if an approximate volume for the external air sacs were included in the vert’s volume figure (let’s say, one for the most conservative configuration shown in Daniela Schwarz-Wings’ paper), then the bone density would be comparable to or lower than a camellate sauropod.
April 18, 2010 at 8:40 am
“I just realized that since the full extent of the air sacs in a brachiosaur [i.e. its soft tissue] can not be known with certainty, the raw vertebra’s volume (without the external air sac volume) is what is normally used for the density equation.”
That is the only way in which sauropod vertebra density has been measured — it’s the ASP (air-space proportion) of Wedel (2005). And, once more, the actual measured numbers for ASP are generally higher for camellate/somphospondylous taxa than for camerate. Matt’s papers are freely available at http://sauroposeidon.wordpress.com/publications/ — read them.
April 19, 2010 at 5:43 am
Also I’m still curious about this:
Also when I look at pneumatic bones of many animals today, it’s not all air in the internal bubbles of spinal bones, mostly it’s either fluid or marrow, both of which added some weight.
I have a hard time making sense of this because the only extant animals that are known to have pneumatic vertebrae are birds and some osteoglossomorph fish (true story, their vertebrae are pneumatized by the swim bladder). If you’re looking at the ‘spinal bones’ of other animals, like, say, alligators or cows, then they’re not pneumatic, so they don’t bear on the density of pneumatic bones one way or another. And if you’re pointing out that some vertebrae of birds are filled with marrow or fluid, then clearly they’re not the pneumatic ones, either. It’s been known for ages that many birds don’t pneumatize the entire vertebral column. So, sorry to be dense, but what are you talking about?
April 19, 2010 at 10:55 am
To answer Zach’s question, thought I’d add that when I looked at Argentinosaurus last year, there were 7 dorsal vertebrae preserved (one more than described). I think the additional dorsal vertebra was subsequently excavated from the site/prepared after the publication (along with a few additional elements also not described in the paper). No idea what the basis is for the 9 dorsal vertebrae in the Carpenter (2006) paper.
April 20, 2010 at 12:01 am
Just a quick thought on the discussion about skeletal pneumaticity lightening the skeleton: I’m not sure we’ve been thinking about pneumaticity in the right way. The relationship between skeletal mass and body mass appears pretty constant regardless of the degree of pneumaticity in the body, suggesting pneumatic structures have no effect on overall mass (Prange et al. 1979). As I understand it, pneumatisation of the skeleton does not remove skeletal mass but redistributes it, making for bones with expanded linear dimensions but without adding additional mass. Thus, animals with extensive PSP are not strictly lighter for being pneumatic, but can achieve greater overall size with no effect on mass. There are probably structural benefits to this, too: larger beam diameters are stronger for a given mass and all that, so this probably also helps in attaining large size. As such, I agree with Mazzetta et al. (2004) in that bigger sauropods would have heavier skeletons than smaller ones – if they were anything like modern animals, they almost certainly would – but disagree that this would counter the effect of pneumaticity: the animal’s mass would reflect it’s skeletal mass, and skeletal pneumaticity would have nothing to say about it.
April 20, 2010 at 2:25 am
The relationship between skeletal mass and body mass appears pretty constant regardless of the degree of pneumaticity in the body, suggesting pneumatic structures have no effect on overall mass (Prange et al. 1979). As I understand it, pneumatisation of the skeleton does not remove skeletal mass but redistributes it, making for bones with expanded linear dimensions but without adding additional mass. Thus, animals with extensive PSP are not strictly lighter for being pneumatic, but can achieve greater overall size with no effect on mass.
That’s one interpretation of the results of Prange et al. (1979), and the one that Prange et al. favored. Prange et al. (1979) floated the hypothesis that when the skeleton of a bird becomes pneumatized, the total amount of bone in the skeleton doesn’t change because an amount of bone equal to that removed by pneumatization is reinvested elsewhere in the skeleton. That’s a nifty idea, and it would solve the apparent paradox, but AFAIK it hasn’t been tested. And there are a couple of other possibilities.
First, almost no birds pneumatize the entire skeleton. In many taxa, only the presacral vertebrae and humeri are pneumatic. And there are shedloads of little passerines that hardly pneumatize the skeleton at all. So I’ve often wondered: maybe pneumaticity just isn’t widespread enough (in the skeleton, and among taxa) to shift the curve for birds. That is, pneumaticity could lighten those individuals and taxa that have it, but still have a negligible impact on the curve for the clade as a whole. I don’t know enough about stats to know how a thousand essentially apneumatic passerines would affect the curve relative to the comparative handful of “hyperpneumatic” taxa like pelicans and vultures.
Second, when the whole-body density of birds has been measured–for example, by Hazlehurst and Rayner (1992)–it has been found to be a lot lower than that of mammals. And the volumes I’ve seen given for the lungs and pulmonary air sacs–10-20% of the whole-body volume–are generally not big enough to account for that on their own.
Admittedly this is pretty insubstantial in either direction. Someone needs to go through all of the taxa used by Prange et al. (1979) and figure out how much PSP was present in each one, or maybe it would be faster to just redo the study and keep track of the PSP this time. Also, the hypothesis of Prange et al. that bone removed during pneumatization is reinvested somewhere else is open to investigation. One could follow baby birds through the weeks during which their skeletons are being pneumatized and see if the skeletal mass as a fraction of body mass decreases (PSP lightens the skeleton) or stays the same (it doesn’t). And we could use a lot more work on the in vivo densities of birds–whole-body densities, skeletal densities, and individual element densities.
And I know that there was just a paper out in the last few weeks purporting to have solved the problem, by finding that the bone tissue of birds is denser than that of mammals. I haven’t read that one yet, but I know of a couple of published values that don’t agree. And if it is true that the bone tissue of birds is denser, then it’s even stranger that their whole-body densities are so low, which might point to an even bigger role for PSP in lightening the body.
Lots of good questions, and not nearly enough data. If any of you readers are wondering what to do your PhD work on, look around. This field is wide open.
April 20, 2010 at 9:01 pm
It’s obvious what pneumatization does for bones: it gives them more bending strength per unit weight. But what good is soft-tissue pneumatization supposed to do? Sure, some of it in the right places can improve breathing. Sure, some of it in other places may provide attractive convexity, for display. But, really, how much of this soft-tissue pneumatization can you use? Is there any benefit to inflating balloons at random places in the body cavity? All that comes to mind is improving the lever arm of muscles that cross it.
April 23, 2010 at 3:26 pm
“(true story, their vertebrae are pneumatized by the swim bladder)”
Holy crap that’s cool.
July 27, 2020 at 8:28 pm
Bump!
I’d just like to remind everyone that, following Greg Paul (2019), the best estimates for Argentinosaurus are 65-75 tonnes, no more and no less.
Depending on whether you lowballed or highballed before this is either good or bad news.
July 28, 2020 at 11:07 am
Well: more precisely, following Greg Paul (2019), Greg Paul’s best estimates for Argentinosaurus are 65-75 tonnes, no more and no less. Others may legitimately reach different conclusions. Paul may be correct, but there mere fact that his paper is the most recently published doesn’t in itself mike it particularly more likely.