Brachiosaurusbrancai is not Brachiosaurus

September 9, 2009

Today sees the publication of the new Journal of Vertebrate Paleontology, and with it my paper on the two best-known brachiosaurs and why they’re not congeneric (Taylor 2009).  This of course is why I have been coyly referring to “Brachiosaurus” brancai in the last few months … I couldn’t bear to make the leap straight to saying Giraffatitan, a name that is going to take me a while to get used to.

But before we go lunging into the details, here is my skeletal reconstruction of Brachiosaurus proper, taken from the paper:

Taylor-SVP-Brachiosaurus-fig7-reconstruction-R3-480px

Skeletal reconstruction of Brachiosaurus altithorax, with Homo sapiens and Canis familiaris for scale, from Taylor (2009:fig. 7). White bones represent the elements of the holotype FMNH P 25107. Light grey bones represent material referred to B. altithorax: the Felch Quarry skull USNM 5730, the cervical vertebrae BYU 12866 (C?5) and BYU 12867 (C?10), the “Ultrasauros” scapulocoracoid BYU 9462, the Potter Creek left humerus USNM 21903, left radius and right metacarpal III BYU 4744, and the left metacarpal II OMNH 01138. Dark grey bones modified from Paul’s (1988) reconstruction of Giraffatitan brancai. Scale bar equals 2 m.

Those of you familiar with Greg Paul’s classic reconstruction of Giraffatitan brancai will immediately recognise that Real Brachiosaurus is rather differently proportioned, especially in having a longer torso and tail.

This paper has been in the works for some time, and while it was in review and then in press at JVP, it led a double life as Chapter 2 of my dissertation.  (For most of its gestation period, the paper’s title was just “Brachiosaurus brancai is not Brachiosaurus“, and the folder where I keep all the project files is still called “bb-is-not-b”).  In the end, I chickened out and went for a longer, more formal, title.

So why are the two species not congeneric?  Well, it’s a long story, and you can read about the detail in the paper, but the bottom line is that virtually every bone that is known from both species differs in significant respects between them.

Of course, I am not the first to suggest that the African brachiosaurid that we know and love isn’t exactly Brachiosaurus.  Credit for that goes to Greg Paul, who more than twenty years ago executed a then-new skeletal reconstruction of that species (the very same reconstruction that is now considered the classic), and in doing so noticed some differences between the American type species Brachiosaurus altithorax and the African referred species “Brachiosaurus” brancai (Paul 1988).  Paul hedged his bets, though: rather than erect a new genus for the African animal, he proposed a subgenus Brachiosaurus (Giraffatitan), so that the full name of the species would become Brachiosaurus (Giraffatitan) brancai; and that of the type species would become Brachiosaurus (Brachiosaurus) altithorax.  Unsurprisingly, this cumbersome nomenclatural scheme did not catch on, and I have not been able to locate a single subsequent reference to these subgenera in the literature.

xxx

Second caudal vertebrae of Brachiosaurus altithorax and Brachiosaurus brancai, equally scaled, from Taylor (2009:fig. 3). A, B, B. altithorax holotype FMNH P 25107; C-G, B. brancai referred specimen HMN Aa. A, C, posterior; B, D, F, right lateral; E, G, anterior. A-B modified from Riggs (1904:pl. LXXV); C-E modified from Janensch (1950a:pl. 2), F-G modified from Janensch (1929:fig. 15). Scale bar equals 50 cm.

That didn’t mean the idea was dead, though: three years later, George Olshevsky’s self-published mega-revision of dinosaur taxonomy proposed raising the name Giraffatitan to genus level (Olshevsky 1991).  Although this genus became popular on the Internet (it cropped up, for example, in Mike Keesey’s much-lamented Dinosauricon web-site), it was almost completely ignored in the technical literature, and even Greg Paul himself subsequently seems to have reverted to using the name Brachiosaurus brancai (e.g. Paul 1994:246).

Why was the new name overlooked?  Partly, I suspect, just because it’s so butt ugly — everyone knows and loves Brachiosaurus brancai, and the name itself has a definite poetry to it that Giraffatitan sorely lacks.  But mostly it’s because Paul didn’t really make a case for the separation that he proposed — wrongly stating, for example, that “the caudals, scapula, coracoid, humerus, ilium, and femur of B. altithorax and B. brancai are very similar” (Paul 1988:7).

That’s how things stood a few years back when I started to take a serious interest in Migeod’s Tendaguru brachiosaurid, which lives in the basement of the Natural History Museum in London.  It quickly started to seem to me that it wasn’t the same thing as what everyone means by Brachiosaurus, but to make sense of it all, I needed first to figure out what the Brachiosaurus actually does mean.  That meant visiting the type material of both species, in Chicago and Berlin, and really looking closely.

Well, I don’t want to go on all day — apart from anything, England play Croatia in a World Cup qualifier in just over an hour — so I’ll just show you some of the the differences between the dorsal vertebrae of the two species.  (You’ll have seen the caudals up above — I just threw them in to break up all that text).

xxx

Dorsal vertebrae of Brachiosaurus altithorax and Brachiosaurus brancai in posterior and lateral views, equally scaled, from Taylor (2009:fig. 1). A, B, E, F, I, J, M, N, B. altithorax holotype FMNH P 25107, modified from Riggs (1904:pl. LXXII); C, D, G, H, K, L, O, P, B. brancai lectotype HMN SII, modified from Janensch (1950a:figs. 53, 54, 56, 60-62, 64) except H, photograph by author. Neural arch and spine of K sheared to correct for distortion. A, D, E, H, I, L, M, P, posterior; B, F, G, J, N, right lateral; C, K, O, left lateral reflected. A, B, dorsal 6; C, D, dorsal 4; E-H, dorsal 8; I-L, dorsal 10; M, N, P, dorsal 12; O, dorsals 11 and 12. Corresponding vertebrae from each specimen are shown together except that dorsal 4 is not known from B. altithorax so dorsal 6, the most anterior known vertebra, is instead shown next to dorsal 4 of B. brancai. Scale bar equals 50 cm.

Lots and lots of differences here — I will quote from the Systematic Paleontology section on the type species: “Postspinal lamina absent from dorsal vertebrae (character 130); distal ends of transverse processes of dorsal vertebrae transition smoothly onto dorsal surfaces of transverse processes (character 142); spinodiapophyseal and spinopostzygapophyseal laminae on middle and posterior dorsal vertebrae contact each other (character 146); posterior dorsal centra subcircular in cross-section (character 151); posterior dorsal neural spines progressively expand mediolaterally through most of their length (“petal” or “paddle” shaped) (character 155); mid-dorsals about one third longer than posterior dorsals (see Paul, 1988:7); middorsals only about 20% taller than posterior dorsals (see Paul, 1988:8); dorsal centra long (Janensch, 1950a:72) so that dorsal column is over twice humerus length (Paul, 1988:8); transverse processes of dorsal vertebrae oriented horizontally (Paul, 1988:8); dorsal neural spines oriented close to vertical in lateral view; dorsal neural spines triangular in lateral view, diminishing smoothly in anteroposterior width from wide base upwards; deep inverted triangular ligament rugosities on anterior and posterior faces of neural spines” …. *gasp*

So anyway: the upshot of all this is that “Brachiosaurus” brancai differs from Brachiosaurus altithorax more than, say, Barosaurus does from Diplodocus; and so it must be placed in its own genus … and that genus has to be Giraffatitan, because of the ICZN’s principle of priority.  And THAT is why the very end of the paper — the last sentence of the Acknowledgements — reads:

Finally, I beg forgiveness from all brachiosaur lovers, that so beautiful an animal as “Brachiosaurus” brancai now has to be known by so inelegant a name as Giraffatitan.

Anyway, go and read the paper; full-resolution figures are freely available if you want to look more closely than the JVP’s PDF allows.

References

(And, yes, Randy, I know what your comment is going to say; go ahead and say it anyway, it’ll give me a chance to explain why your approach is wrong :-))

Advertisements

53 Responses to ““Brachiosaurusbrancai is not Brachiosaurus

  1. andy Says:

    More “everything I knew about dinosaurs when I was 10 is wrong” then. Anyone who messes with Brachiosaurus deserves to get eaten by a rampaging Manospondylus gigas. :-P

  2. Mike Keesey Says:

    I never have understood the hatred for the name — I see no problem with it. Perhaps it would be better formed as “Giraffotitan“? Other than that, how is it worse than Anatotitan or Gondwanatitan?

  3. Mike Taylor Says:

    Well, Mike, I guess if you don’t feel the Name Pain, that’s all to the good — I won’t try to persuade you that it’s horrible, because if I succeeded you wouldn’t thank me. It just feel fantastically lame to me that, given the most awesome of all dinosaurs — in fact, the most awesome object of any kind anywhere in the universe — the best you can do in naming it is “really big giraffe”.

  4. 220mya Says:

    Mike – I didn’t see your snarky comment until the second read through. Before I make a comment, can you answer this question: if you added the BMNH (sorry NHM) Tendaguru brachiosaurid to the matrix, do you think it would fall out closer to brancai or to altithorax, or would you just get a polytomy? Same question for Sauroposeidon and other supposed brachiosaurs. If this question impinges on any sort of manuscript that might be in prep, I understand if you don’t want to answer.

  5. Mike Taylor Says:

    Hi, Randy — sorry, was that comment uncalled-for? :-)

    Your question on adding the Archbishop (i.e. the NHM brachioaaur) and/or Sauroposeidon to the matrix from the Brachiosaurus paper … is a good one. To my shame, I’ve not done that, but I will OF COURSE be doing it for the Archbishop description — which is, now that this one is out, pretty much at the top of my TODO list. While doing that, it would seem churlish not to throw my old buddy Sauroposeidon in, too, but that’s as far as I’ll go: I’m not going to make an effort to score all the other putative basal titanosauriforms out there, as I know of at least two other projects that are under way to do this properly. So for my purposes, the phylogenetic analysis is merely part of the furniture that need in the background of my descriptive work.

    But since you asked what I THINK … I have a hunch that the Archbishop is going to fall out closer to Brachiosaurus than to Giraffatitan; Sauroposeidon of course can’t be found as sister to Brachiosaurus as it’s known only from cervicals, which are not known from Brachiosaurus (at least, not convincingly referred cervicals). So I would not be totally shocked if I found ((Brachiosaurus, Archbishop), (Giraffatitan, Sauroposeidon)).

    Then again, the hypothesis that Sauroposeidon is a brachiosaur has never really been tested … it’s possible that when I throw it in the matrix, it will come out as a titanosaur or something. Who can tell?

  6. Mike Keesey Says:

    I think of it more as, “the primordial gargantuan from the hoary days before gods who happens to resemble a giraffe from certain angles”.

    Oh, and on this: “Mike Keesey’s much-lamented Dinosauricon web-site….” I hope it’s the demise that’s lamented, not the site itself! :)

    Should have a rad new site up this year sometime….

  7. LeeB Says:

    Is there any particular reason why it would not be useful to include Lusotitan in your matrix as well.
    Or is this not as closely related to Brachiosaurus as has been previously thought?

    LeeB.

  8. Mike Taylor Says:

    Lee, it would be great to include Lusotitan. But I’ve not seen the material (nor that of Cedarosaurus, Sonorasaurus, etc.) and I don’t want to wait many years until I’ve seen all that stuff before I can describe the Archbishop. Like I said, others are working on basal titanosauriform phylogeny for its own sake; but that’s not the focus of my work.

  9. Matt Wedel Says:

    Then again, the hypothesis that Sauroposeidon is a brachiosaur has never really been tested … it’s possible that when I throw it in the matrix, it will come out as a titanosaur or something. Who can tell?

    Someone who’s read Wedel et al. (2000b) and Naish et al. (2004). Unless the verts contain a fair number of titanosaur-o-centric characters that me, Kent, Rich, Darren, both Daves, the other Kent, our reviewers, and our editors missed, and there’s been a global conspiracy of silence among our readers all these years.

    Okay, seriously, yes it’s true that Sauroposeidon has never been included in a proper phylogenetic analysis. But I would like to think that we’re not so narrow-minded that computerized analyses are the only form of hypothesis testing that we recognize. The hypothesis that Sauroposeidon is a brachiosaurid–and you know that I correctly see this as a proposition and not a certainty–has been tested to the extent that in all but one of the phylogenetically informative characters that anyone has pointed out to date, it is more similar to Giraffatitan and the American Brachiosaurus sp. material than to anything else. The odd man out is fully camellate internal structure, which as far as I know is only present among sauropods in Sauroposeidon and most (but probably not all) somphospondylans. That one character (so far) does not outweigh the rest (and “Angloposeidon” provides a plausible intermediate between Giraffatitan/Brachiosaurus sp. and Sauroposeidon for that and other characters–something that Darren and I have been planning to write up for ages now).

    We could test this by coding these up and running the analysis, and it would be great if someone did that just for the sake of form, but I think we’d learn more about the process than about the affinities of Sauroposeidon. For one thing, there is so much missing data that including Sauroposeidon (and other relevant taxa like “Angloposeidon”, the Croatian brachiosaur, et al.) in a good-sized analysis would probably just produce a monster polytomy. Let’s say that it destroyed all resolution within Macronaria. That wouldn’t mean that Sauroposeidon ought to be properly considered Macronaria incertae sedis, it would mean that current phylogenetic techniques aren’t good at dealing with lots of missing data. The majority of characters would still point to a brachiosaurid identity.

    Or maybe we’d find out that Sauroposeidon is closer to Giraffatitan than to titanosaurs, because we coded up the handful of characters that we’ve got and, as expected, that’s how they pointed. Unless someone comes up with a good number of new characters based on cervical vertebrae AND Sauroposeidon shares more of those characters with titanosaurs than with brachiosaurs, I don’t see any other outcome.

    Don’t get me wrong, I’m not anti-phylogenetics (far from it), and I’m not trying to be defeatist. I can’t wait until someone puts Sauroposeidon into a proper analysis. But I’ve had no interest in doing so myself, because until those new characters show up I’ve got better things to do than let a program give me a result I can arrive at on my own. Also, as you state, there are a couple of folks working on sorting out basal titanosauriform phylogeny, and I can always hope that they’ll include S. (and I’ll also completely understand if they don’t).

    All of this reminds me that I’ve not yet gotten around to explaining why, as a reviewer, I was satisfied that Qiaowanlong is a brachiosaurid. Maybe I will do a whole post on brachiosaurid cervicals soonish. On that note, I gotta get back to the abdominal viscera (of stinkin’ humans, sigh).

  10. 220mya Says:

    Mike;

    For the sake of discussion, if the topology is as you say, then I do support the generic separation of altithorax and brancai. Now, of course, as you might surmise, if the two sub-clades are well-supported, I would also advocate putting altithorax and the NHM Tendaguru taxon in the same genus, and brancai and Sauroposeidon in the same genus. Of course, this is all for the sake of discussion because we don’t actually know what the results would be, and I suspect the sub-clades you outline probably wouldn’t be that well-supported given some of the missing data problems you mention.


  11. Very cool looking paper. Will take me a while to read through it all.

    Regarding cladistic analyses, the traditional worry is that the characters that are similar in Brachiosaurus, Giraffatitan and Sauroposeidon are symplesiomorphies within Titanosauriformes. This wouldn’t necessarily be obvious because our idea of titanosaurs probably consists of the taxa we’re familiar with- Euhelopus and genera at least as derived as Andesaurus. But if some of the numerous recently described basal titanosaurs from Asia and North America share some of the supposed brachiosaurid characters, they turn these characters into titanosauriform symplesiomorphies that were lost in derived titanosaurs. This is a way we could find a surprising outcome without having to find many new characters- we would be finding new distributions instead. Of course current sauropod analyses are probably deficient on brachiosaurid characters because normally only one brachiosaurid is used, which is no doubt why Brachiosaurus becomes paraphyletic in one step in Taylor’s analysis. Or alternatively, maybe some of the basal titanosaurs share characters with only some traditional brachiosaurids, again making these characters which are gained on the titanosaur stem then lost again before derived titanosaurs appear. Or they might support subclades of basal titanosaurs. In both of these cases, the characters are almost certainly lacking from current matrices and would not even be noticed unless someone sits down and really compares each taxon to each other taxon. Unfortunately most matrices of any group consist largely of characters we notice in the most derived members of each subgroup, which is why so many trees are Hennigian combs. Maybe these scenarios wouldn’t happen in the present case, but they’re issues in tree construction that are often overlooked.

    As for your hypothetical comprehensive macronarian analysis, I completely agree we’d get a polytomy if we included everything. But there are ways to still see the underlying structure. You just prune taxa from the trees (as opposed to excluding them from the analysis) until you get as much resolution as possible. It does require some degree of repetition for each taxon, and is MUCH easier in TNT than in PAUP, but in the end you can say “taxon x can only go in these positions” for every taxon you included. It’s important to do this instead of just deleting or excluding taxa from the analysis though, since even fragmentary taxa with unique character distributions can have large effects on tree topology.

    The kinds of issues I noted above are those which make me extremely cautious about accepting Qiaowanlong or any of these new taxa as having any relatively certain position within Macronaria (or even being macronarians, honestly). It’s like deciding Rahonavis’ position based on comparison to only Deinonychus, Troodon and pygostylians. Without Unenlagia, Microraptor, Buitreraptor, Mahakala, Jinfengopteryx, Archaeopteryx, Shenzhouraptor, etc. to flesh out the character distributions, both humans and PAUP would be misled.

  12. Matt Wedel Says:

    But if some of the numerous recently described basal titanosaurs from Asia and North America share some of the supposed brachiosaurid characters, they turn these characters into titanosauriform symplesiomorphies that were lost in derived titanosaurs. This is a way we could find a surprising outcome without having to find many new characters- we would be finding new distributions instead.

    That is a pretty darn good point. Your point about throwing in everyone and pruning after is also well taken.

    Remind me again why you’re not publishing? :-)/2

  13. Mike Taylor Says:

    Mickey Mortimer wrote: “Of course current sauropod analyses are probably deficient on brachiosaurid characters because normally only one brachiosaurid is used, which is no doubt why Brachiosaurus becomes paraphyletic in one step in Taylor’s analysis.”

    This is a very good and important point that I didn’t bring out strongly enough in the paper. There are indeed more characters that could be added to the matrix to support brachiosaur monophyly.

    But wait — we don’t add characters IN ORDER TO support a particular conclusion, do we?

    Well, actually yes, sort of. No-one would deliberately set out to mine characters that support a pre-selected phylogenetic hypothesis. But it is a fact that the way you notice characters in the first place is by looking at whatever taxa you’re working on; and when you see a character that they share, well, it would be perverse NOT to add it to the matrix. For example, looking a the Archbishop dorsals, I ask myself, is this similar to Brachiosaurus altithorax? And part of the answer is that, yes, it resembles it in the possession of a five-way junction of lamina at the base of the transverse process. But when I come to code that character, I will quite likely find that it’s unique to brachiosaurs. So will I have added a character IN ORDER TO support brachiosaur monophyly? Not on purpose.

    (On the other hand, I might find when I come to look at it that this character is widespread, and actually pulls Brachiosaurus down towards Camarasaurus, or something.)

    So: I do have another more phylogenetically focussed project rumbling along in the background (which I don’t want to blab about at this early stage), and that will hopefully give us more characters to work with when evaluation dorsal vertebrae. It’ll be interesting to see what those characters do to the current candidate brachiosaurs.

  14. Mike Taylor Says:

    Inspired by this very post, Lockwood DeWitt of Outside the Interzone takes his pet Brachiosaurus altithorax for a walk: http://outsidetheinterzone.blogspot.com/2009/09/in-my-dreams.html

  15. David Marjanović Says:

    Those of you familiar with Greg Paul’s classic reconstruction of Giraffatitan brancai will immediately recognise that Real Brachiosaurus is rather differently proportioned, especially in having a longer torso and tail.

    Then why is the entire tail in dark grey, implying it’s unknown in Brachiosaurus and copied from Giraffatitan?

    I never have understood the hatred for the name — I see no problem with it.

    It should definitely be -titanus.

    It just feel fantastically lame to me that, given the most awesome of all dinosaurs — in fact, the most awesome object of any kind anywhere in the universe — the best you can do in naming it is “really big giraffe”.

    So calling it “arm lizard” is better?

    That magnificent animal, a lizard? You can’t be serious.

    But wait — we don’t add characters IN ORDER TO support a particular conclusion, do we?

    No – we throw in all parsimony-informative characters that we can find and that don’t seem to be correlated to others. The more, the better.

    OK, according to some SVP meeting abstract or other, there does seem to be a limit where (for any given number of taxa) the addition of yet more characters doesn’t lead to a noticeable improvement; but I’m not sure if that limit has ever been reached in practice. Maybe by Livezey & Zusi, but I doubt it.

    As a rule of thumb, below 3 times as many characters as taxa I wouldn’t publish.

    You can never have enough taxa according to the same simulation studies (and also my limited experience, and Mickey’s less limited one).

    For parsimony (but not likelihood or Bayesian analysis!!!), missing data are a much smaller problem than most people apparently still think. As was shown by a series of simulations by Wiens and others, mostly in the June 2003 issue of JVP, what matters is not the proportion of missing data, but the absolute amount of known data.

    Closely related taxa known from nonoverlapping body parts can decrease resolution because they cannot be found as sister-groups even if they are (Nemegtosaurus and Quaesitosaurus on the one hand and Opisthocoelicaudia come to mind as a possible sad example); and taxa that lie far away from the ingroup increase the possibility of long-branch attraction. But apart from such cases, and outright nomina dubia, throw in as many taxa as you can. Every taxon has an influence on the position of every other taxon. Phylogenetic analyses don’t respond in a linear fashion to changes in the size of the dataset or (as my last two papers confirm, and the next one will again) to mistakes in the matrix.

    (On the other hand, I mind find when I come to look at it that this character is widespread, and actually pulls Brachiosaurus down towards Camarasaurus, or something.)

    This is why it’s important to check the presence of every character state in every taxon. I’m right now working on the revision of a matrix that had two identical characters, worded differently, apparently taken from autapomorphy lists of different clades and just coded “present” for the clade in question and “absent” for the entire rest of the matrix…

  16. David Marjanović Says:

    …and Opisthocoelicaudia on the other hand, obviously.

  17. Mike Taylor Says:

    Those of you familiar with Greg Paul’s classic reconstruction of Giraffatitan brancai will immediately recognise that Real Brachiosaurus is rather differently proportioned, especially in having a longer torso and tail.

    Then why is the entire tail in dark grey, implying it’s unknown in Brachiosaurus and copied from Giraffatitan?

    Not quite — the second caudal is preserved in the BOBA holotype FMNH P25107, and it’s very informative (and illustrated above). To quote the paper:

    More significant are the differences between the single known caudal vertebra of Brachiosaurus and those of Giraffatitan. Many caudals of the latter are known, and are remarkably consistent in morphology, while the single known caudal of the former is unambiguously associated with the remainder of the specimen and differs from those of Giraffatitan in two mechanically significant ways: first, although it is from a similarly sized animal as the G. brancai type specimen, and is comparable in anteroposterior length, it is taller in both the centrum and the neural arch (Fig. 3B, D); and second, the transverse broadening of the neural spine towards its extremity allows a much greater area for ligament attachment ­ about 2.25 times as great. The former character certainly indicates that the tail was taller in the American taxon, and the latter suggests that it was longer, perhaps by about 20%-25%.

  18. Mike Taylor Says:

    It just feels fantastically lame to me that, given the most awesome of all dinosaurs — in fact, the most awesome object of any kind anywhere in the universe — the best you can do in naming it is “really big giraffe”.

    So calling it “arm lizard” is better?

    Touche!

  19. Andreas Johansson Says:

    It should definitely be -titanus.

    Any particular reason, beyond the pun potential? The shorter form is commoner in Latin and, AFAIK, the only one in Greek.

  20. Andreas Johansson Says:

    Also, why am I suddenly inflicted with some sort of floating purple cartoon head for an avvie?

  21. Matthew Says:

    Let’s talk about Giraffatitan’s horizon. Mike Taylor states that it is present in:
    “Middle and Upper Saurian Members, Tendaguru Formation”.

    In recent review of Tendaguru (ref. below) Bussert et al. proposed to call this two Menbers as Middle and Upper Dinosaur. First is problaby late Kimmeridgian and second Tithonian (and maybe in part even Cretaceous). This is of course just nomelclature and case of uncertain age, but they stated that Giraffatitan (there called Brachiosaurus) is present in Lower Dinosaur Member too, which could be as early as Callovian, but albo could be quite young – probably late Kimmeridgian. So which remains come from Lower Dinosaur Member? I was wondering that could be “The Archbishop”?

    Ref:
    Bussert, R., Heinrich W-D., Aberhan, M. (2009) “The Tendaguru Formation (Late Jurassic to Early Cretaceous, southern Tanzania): definition, palaeoenvironments, and sequence stratigraphy” Fossil Record 12(2): 141-174

  22. Nathan Myers Says:

    Andreas: You reminded somebody of Barney. Go to gravatar.com to make your own.

  23. Mike Taylor Says:

    Hi, Matthew. It’s all very well for Bussert et al. to use the Anglicised names Upper Dinosaur Member and so on, but there is a substantial literature using the older name, and I’m sticking with it. In biological taxonomy we have a principle of priority — I don’t know whether the same is true in stratigraphic taxonomy, but it seems like a good way to go about things anyway.

    As for the age of the top of the Tendaguru Formation: yes, I know it’s somewhat contentious, and that it has been claimed to extend into the Lower Cretaceous. That may be so, but the most recent work at the time I submitted this paper argued for a wholly Jurassic age, so that’s what I went with (being no stratigrapher myself, and frankly not being very interested). Sorry, no refs. to hand for this.

    Finally, and more interesting to me, brachiosaurs in the Lower Saurian Member? Russell et al. (1980) tabulated the occurrence of all Tendaguru Formation dinosaurs across the members, and listed 1 “Brachiosaurus” individual in the Lower, 26 in the Middle and none in the Upper Saurian Bed. They also listed 3 in the Lower and 4 in the Upper Transitional Sands. So brachiosaurs certainly seemed to peak in the Middle Saurian Member; by contrast, dipldocines (catalogued as “Barosaurus”) increased in frequency from a low start.

    And, no, the Archbishop does not seem to be from the Lower Saurian Member. I’ll need to chase some tails before I write the strat section of that paper, though.

  24. Zach Miller Says:

    Congrats on the paper! I just downloaded it, but I’ll have it in print soon enough–I got three copies of JVP in a box since becoming a member (all at once, that was wierd) so I expect the new issue to crop up in my mailbox any day now.

    Can’t wait for SVP, either. I can congratulate you in person (hopefully)!


  25. Mike,

    In stratigraphic nomenclature–like zoological–there is a principle of priority. But there is also a “type specimen” (or rather, “type section”) issue as well. What Bussert et al. did was finally formally define the subunits of the Tendaguru Formation (including the Tendaguru Formation itself) with specific boundaries and such. All previous Tendaguru stratigraphy was essentially “nomen nudum” (to use a taxonomic analogy).


  26. the upshot of all this is that “Brachiosaurus” brancai differs from Brachiosaurus altithorax more than, say, Barosaurus does from Diplodocus; and so it must be placed in its own genus

    Well, actually no, it’s not necessary, Mike:o) Of course, that’s what people in these cases usually do (and why not, inventing new names is always a good fun; although your situation was different, of course, because there already was a name), but brancai can still be Brachiosaurus (it’s still nomenclaturally separated from B. altithorax, which still can be its closest known relative, because we don’t know [you didn’t test] the positions of, say, Sonorasaurus thompsoni, “Brachiosaurus” nougaredi, Lusotitan atalaiensis, Sauroposeidon proteles, Cedarosaurus weiskopfae or Paluxysaurus jonesi…).

    Anyway, although I think that the phylogenetic analysis was unnecessary in this case, I have to say that I like your paper very much! It’s not difficult to read (actually it’s evident that you’ve enjoyed writing:o)), it’s informative and accurate. Congratulations!

  27. Mike Taylor Says:

    Daniel wins the prize for making the comment I was expecting Randy to make.

    Stand by for a rebuttaldiscussion … in a forthcoming SV-POW! post with a title suggested by Jim Farlow! (That should get the punters in.)

  28. Mike Keesey Says:

    For consistency, you could also sink Barosaurus into Diplodocus (and it’d still be about equivalent to an insect superspecies, if that). :)


  29. […] y puede descargarse en pdf aquí. Hay mucha más información (en inglés) aquí. […]


  30. “But wait — we don’t add characters IN ORDER TO support a particular conclusion, do we?

    Well, actually yes, sort of. No-one would deliberately set out to mine characters that support a pre-selected phylogenetic hypothesis.”

    Or would they? This is actually a very easy thing to check for in an analysis. It’s provided by a little thing called the Consistancy Index (CI). Basically, the CI is a ratio of characters in the analysis versus times each character changes states in the cladogram. So every convergence or reversal will cause the CI to lower. Thus if you have a CI of 1.00, either you examined the one clade magically lacking homoplasy, you got extremely lucky in finding only the characters that would support your phylogeny and not seeing the ones that don’t, or you planned your cladogram to have that result. In general then, the lower CI the better, since it indicates less bias. For some examples, here are the CIs of big sauropod analyses.

    Wilson and Sereno, 1998 – 0.81
    Sereno, 1999 – 0.80
    Wilson, 2002 – 0.66
    Calvo and Salgado 1995 – 0.655
    Upchurch, 1998 – 0.553
    Harris, 2006 – 0.526
    Upchurch et al., 2004 – 0.514

    So most look decent, and the two newest are best, but the analyses involving Sereno are problematic. 0.80 isn’t bad for Sereno though. It’s actually the lowest of the CIs in his 1999 dinosaur supertree publication, which range from that to 0.97(!) (basal Aves and Prosauropoda).

  31. Mike Taylor Says:

    I’d not previously come across the idea that low CI is a good thing because it shows you didn’t cheat; I am used to the idea that high CI is a good thing because it shows your clade is well resolved. But actually I am inclined to think that high CI merely indicates a small number of characters and taxa, as CI pretty uniformly declines as the size of the matrix increases.


  32. I’d agree with your idea about matrix size being correlated somewhat to CI, but only in that a huge matrix will generally have a low CI, assuming correct coding. This can still fit with my idea, since it basically says there are only so many characters being used in analyses so far (to ~300 in sauropod matrices), so if you were designing a matrix to get a certain result you would only include a subset of characters and taxa. This makes your analysis smaller than competing analyses which include all the additional homoplasious characters and taxa. But it’s certainly possible to have a small unbiased analysis. Let’s see what the data say, with amount of data (characters x taxa) listed after the CI.

    Wilson and Sereno, 1998 – 0.81, 1090
    Sereno, 1999 – 0.80, 1508
    Wilson, 2002 – 0.66, 6318
    Calvo and Salgado 1995 – 0.655, 637
    Upchurch, 1998 – 0.553, 5330
    Harris, 2006 – 0.526, 9930
    Upchurch et al., 2004 – 0.514, 12669

    So yes, Sereno’s analyses are certainly smaller than Upchurch’s and more recent ones, and there’s a trend if you graph it out. But also note Calvo and Salgado’s and Wilson’s analyses have about the same CI, but differ by an order of magnitude in size.

    (hope you all don’t mind me turning SVPOW into Mickey’s Cladistics Corner this week) ;)

    Oh, and now that I’ve read a bit more of the Giraffatitan paper, I can comment. I think you did a good job showing the species are different, and that most of Janensch’s Brachiosaurus synapomorphies are more widely distributed. However, to really clinch your case I think you would need to show that the four characters Janensch listed which are limited to brachiosaurids are also present in species besides altithorax and brancai, and that this is also true of Brachiosaurus synapomorphies subsequent authors have (hypothetically) proposed. So I wonder, which brachiosaur-grade taxa also have the following?

    – Anterior dorsal vertebrae with long diapophyses.
    – Neural spines low in posterior dorsals, taller anteriorly.
    – Ilium with strongly developed anterior wing.
    – Ilium with compressed pubic peduncle.

  33. Vasha Says:

    Well, I may be in a very small minority, but I love the name Giraffatitan and was hoping it would become widely used.

  34. Philip Kahn Says:

    Very nice, Mike! And interesting analysis.

    On a random aside, for high-res figures that are mostly text and solid lines, consider saving them as a “PNG” file. It’s lossless so the text stays sharp (and no generational losses from editing), and for simple images like text or screenshots without pictures in them, they are actually smaller than JPGs.

  35. Mike Taylor Says:

    Hi, Philip. I do of course submit all my figures as lossless formats — typically TIFF or, when the journal allows it, PNG. The images in this post are very much JPEG-appropriate — I can only assume that you’re talking here about the high-resolution version of Fig. 6 (the cladogram) at http://www.miketaylor.org.uk/dino/brachio/images/figures/Taylor-SVP-Brachiosaurus-fig6-phylogeny-R3.jpeg

    That is a JPEG rather than a PNG just because it’s simpler for me to manage the site if all the figures of the same type — but as it happens, it’s rock solid and, to my eye at least, artifact-free.

    Or were you thinking of a different image?

  36. Philip Kahn Says:

    Yep, I was talking about the phylogeny — I opened that one up for a quick overview first, and noticed it was a PNG. Nevermind then!

  37. M. O. Erickson Says:

    Why is it that in the sekeletal reconstruction of Brachiosaurus altithorax, you didn’t use this skull?

    Originally thought to belong to “Morosaurus” (Camarasaurus), Carpenter and Tidwell (1998)redescribed it and came to the conclusion that it actually belongs to Brachiosaurus altithorax.

    I havn’t had a chance to read the paper, is the issue touched upon in there?

    Ref.

    Carpenter, K. and Tidwell, V. (1998). “Preliminary description of a Brachiosaurus skull from Felch Quarry 1, Garden Park, Colorado.” Pp. 69–84 in: Carpenter, K., Chure, D. and Kirkland, J. (eds.), The Upper Jurassic Morrison Formation: An Interdisciplinary Study. Modern Geology, 23(1-4).

  38. M. O. Erickson Says:

    I just saw that the skull on the skeletal is indeed supposed to be the Felch Quarry skull USNM 5730, but it looks nothing like USNM 5730. This is strange.

  39. Mike Taylor Says:

    The skull that I used in my Brachiosaurus altithorax reconstruction (Taylor 2009:fig. 7) is indeed the Felch Quarry skull USNM 5730, the image modified from Carpenter and Tidwell (1998:fig. 2A). What makes you think that the image you linked to is USNM 5730? It looks like a perfectly well-behaved caramasaur skull to me, and the ascending process of the maxilla is much more slender than that of the Felch Quarry skull — compare with Carpenter and Tidwell (1998:fig. 5A).

  40. M. O. Erickson Says:

    I’ve seen that thing I linked to labelled as “Brachiosaurus altithorax in a few places on the web, and since my version of Carpenter and Tidwell (1998) unfortunately lacks the figures, I’ve been walking around assuming that that that skull is USNM 5730. Never mind me, I’m just a retard.

    By the way, I just read the paper, and I agree with your assesment of Brachiosaurus and Giraffatitan.

  41. M. O. Erickson Says:

    “That magnificent animal, a lizard? You can’t be serious.”

    It sounds to me like your saying lizards aren’t magnificent. If that is the case, I don’t share your veiws in the remotest sense.

    I’ve never quite understood all of the hatred toward –saurus. I mean, sure, dinosaurs aren’t literally lacertillians, but they ARE reptiles, and the lizard is typically thought of (at least by layfolk) as the very epitome of a reptile. Besides, you can always just translate it as “reptile” instead.

  42. Mike Keesey Says:

    “-ornis” would be much more phylogenetically accurate.

  43. Mike Taylor Says:

    It sounds to me like your saying lizards aren’t magnificent. If that is the case, I don’t share your veiws in the remotest sense.

    Yes, lizard are magnificent. But the thing is, they are nowhere near as magnificant as sauropods :-)

    I’ve never quite understood all of the hatred toward -saurus. I mean, sure, dinosaurs aren’t literally lacertillians, but they ARE reptiles, and the lizard is typically thought of (at least by layfolk) as the very epitome of a reptile. Besides, you can always just translate it as “reptile” instead.

    Speaking only for myself, I agree that the usual “lizard” translation of “-saurus” is not a good reason to dislike it. In fact, if we want to translate in a way that preserves actual intent rather than historical meaning, I think you could even translate it as “dinosaur” — so that Argentinosaurus is “Argentina dinosaur”. (Yes, I know that -saurus is used from plenty of other animals apart from dinosaurs; nevertheless, that’s how the suffix is understood by the general public).

    But there is another and much more compelling reason to hate “-saurus”, and that is because it so fantastically boring. Really, it indicates the most apalling nomenclatural apathy. If you find the world’s biggest land animal in Argentina and the best name you can come up with is “Argentina dinosaur”, then something is not as it should be.

    So … although I am on record of not being a fan of Giraffatitan, Giraffasaurus would been much worse.

  44. David Marjanović Says:

    I mean, sure, dinosaurs aren’t literally lacertillians, but they ARE reptiles

    “Reptile” is a term that should be completely retired. Like… like “thecodontian”.

    Besides, you can always just translate it as “reptile” instead.

    That would be cheating, because no such category existed in Classical Greek.

  45. Asier Says:

    Congratulations for your new paper!!!

    It´s awosme! although I have some divergences about your mass calulations, because you have simplifed too much the mass calculations. I think that is imposible that two individuals (Giraffatitan lectotype and your reconstruction) had the same mass in the legs and neck… I think that Brachiosaurus was heavier.

    My criticism is enterily constructive ;-)

    Have you drawn the skeletal reconstruction?

    Best regards,


  46. […] (although Janensch’s species “Brachiosaurus” brancai is now referred to the separate genus Giraffatitan).  And it certainly is a classic: "Brachiosaurus" brancai (i.e. […]


  47. […] model), thus I bet it is a typical Giraffatitan-based “Brachiosaurus” (see SVPOW posts here, […]


  48. […] (I threw in the other well-known Morrisson-Formation sauropods Camarasaurus and Brachiosaurus, even though Woodruff and Fowler don’t mention them, just because it was easy to do and I was interested to see what would happen.  And when I say Brachiosaurus, I mean B. altithorax, not Giraffatitan.) […]


  49. […] I could have sent my 2009 paper on Brachiosaurus to Nature, and the reviewers would (presumably) have said “this is good science, but not […]


  50. […] my 2009 brachiosaur paper, I gave rather short shrift to the sacrum of Brachiosaurus — in part because there is no […]


  51. […] MB.R.2181. (previously known as “Brachiosaurus” brancai HM S II — yes, it’s changed genus and specimen number, both recently, but […]


  52. […] (1950) paper on the vertebral column of Giraffatitan (which he called Brachiosaurus brancai, wrongly as it turns out) is in many ways a superb piece of work. Together with a separate paper on the skull of […]


  53. […] so Giraffatitan is a legitimate holder of the crown. (Confusing matters further, it used to be thought to be a species of Brachiosaurus). But there were definitely bigger sauropods than that — just not known from such complete […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: