Friday, April 5, 2013

Excellent new paper on titanosauriforms

We interrupt our regularly scheduled dismantling of Peters' phylogeny to bring you a non-theropod paper.

Basal titanosauriforms and macronarians have one of the most contentious phylogenies among dinosaurs.  Mannion et al. redescribe Lusotitan, which used to be Brachiosaurus atalaiensis, finding it to be just basal to Titanosauriformes.  They also redescribe "Brachiosaurus" nougaredi, which I learned is known from unassociated elements, only one of which can be located (an indeterminate titanosauriform metacarpal III).  Just with this the paper would be great, as I love redescriptions of old taxa, but the authors went further and performed a large phylogenetic analysis (with continuous characters) centering on this part of the tree AND reviewed all previous records of basal titanosauriforms.  It's like the titanosauriform equivalent of Carrano et al.'s tetanurine paper.  I'm really not caught up on the lit for these things, as there were plenty of facts I didn't know.  For instance, Agustinia's "osteoderms" are actually rib fragments and ischia.

One thing I did disagree with is their take on LapparentosaurusBothriospondylus madagascariensis was named by Lydekker (1895) based on a dorsal and other material which Mannion believes to be indeterminate.  Bonaparte (1986) erected Lapparentosaurus madagascariensis for that material and what's subsequently been referred to the species, but made two dorsal neural arches described by Ogier (1975; part of a partial skeleton) the holotype.  You might think he was just naive and tried to reset the holotype, but he explicitly erects madagascariensis as a new species as opposed to crediting Lydekker, so it seems he was naively trying to ignore priority instead.  Of course by giving it the same species name, this has engendered confusion.
Mannion et al.'s take is that Bonaparte both referred Lydekker's madagascariensis to the new genus Lapparentosaurus AND erected Lapparentosaurus madagascariensis as a homonym, and thus Ogier's material should be renamed, except that it's what people today think of as being Lapparentosaurus, so the ICZN should be petitioned to conserve the genus for Ogier's material.
I think this is more work than needs to be done.  Bonaparte only says the remains studied by Lydekker "are considered here to represent a new genus, Lapparentosaurus, with the species name L. madagas-madagascariensis [sic]."  Then he goes on to formally erect Lapparentosaurus madagascariensis as a gen. et sp. nov. with his new holotype.  So he never formally refers Lydekker's species to his new genus, just the material it was based on.  Thus we have no homonym, only seemingly indeterminate Bothriospondylus madagascariensis and valid Lapparentosaurus madagascariensis.  If they were synonymous, we'd use Lydekker's madagascariensis for the species name, but since they aren't we don't have a problem.  Certainly not one requiring the ICZN.

If I have one general complaint about the paper, it's that some taxa are declared indeterminate without valid justification, like Pukongosaurus ("based on a lack of diagnostic features"), Agustinia ("The absence of diagnostic features"), Dinodocus, Macrurosaurus, Jiutaisaurus ("All of the features listed as diagnostic by Wu et al. (2006) are more widespread amongst sauropods"), Arkharavia ("The holotype specimen lacks any diagnostic features (all are more widespread – see the current analysis)"), Amargatitanis ("the proposed autapomorphies in Apesteguía (2007) are all more widespread or based on incompletely preserved elements") and Rugocaudia ("all of the characters used to diagnose Rugocaudia are more widespread amongst titanosauriforms and do not form an unusual character combination (see this analysis), or are of dubious diagnostic value (see also D’Emic & Foreman, 2012)").  Just saying something lacks diagnostic characters or character combinations, or that the proposed diagnostic characters/combinations are more widespread are both assertions.  Explicitly describing which two or more taxa share each proposed character or combinarion is better, but Mannion et al. usually don't even go that far.  What's really needed is to go over all the known characters of taxon, and show which two or more other taxa share them all, or indicate they aren't useful for distinguishing species.

Let's use Arkharavia as an example, since it's so simple (... I said before I wrote this paragraph).  It was based on a partial caudal and some referred caudals, the latter of which Mannion et al. identify as hadrosaurian.  I agree with this.  They go on to state all the proposed diagnostic features of the holotype are more widespread, and cite their current analysis as the reference.  Alifanov and Bolotsky (2010) really only propose two characters to diagnose Arkharavia that can be observed in the holotype.  The first is heterocoelous proximal caudal centra, and if we check Mannion et al.'s analysis, that's not a character used.  They do say the centrum is mildly procoelous in Arkharavia, which again I agree with, but never say this differs from the original description or that it nullifies a suggested autapomorphy.  So one of the two characters is not more widespread nor analyzed in their matrix, but is a misinterpretation by the original describers.  The second suggested autapomorphy is the shortness of the centrum compared to its height.  And here, Mannion et al. do use that as a character (C26), which could separate Arkharavia in the continuous character analysis.  Arkharavia has a value of 0.4 (kudos to Mannion et al. for describing their characters so well, btw), which is also scored that way in Camarasaurus and Tangvayosaurus (both the lowest values in the matrix).  So is Arkharavia indeterminate compared to these two genera?  Well no, because neither of these has procoelous centra which Mannion et al. just commented on.  Both are coded 0 for that character (C27), while Arkharavia would be a 1 (or a 0.1 in the continuous version).  Ironically, though they say it's an "indeterminate somphospondylan based on its mildly procoelous, anteroposteriorly short centrum", this is actually a unique character combination in their own matrix!  And based on these two characters alone, it would end up equally parsimonious as a camarasaurid or sister to Tangvayosaurus/Fusuisaurus, so would be Macronaria instead of definitely in Somphospondyli.  Now do I think this character combination means Arkharavia is valid?  No.  The degree of procoely can change in sauropod tails, and doesn't even always follow a simple progression.  Similarly, the elongation of centra changes between vertebrae, and several taxa are only 10% away in Mannion et al.'s matrix.  You would have to examine the degree of variation in each within taxa and see if the position of Arkharavia's holotype in its tail could be constrained.  But note you'd have to do this to support either option- validity or indeterminacy.  This is the start of the kind of work that would need to be done, and Mannion et al. don't come close or even get the summary of why they think it's invalid correct.  Then you'd need to look at the other morphologies preserved in the holotype, which probably also vary throughout the caudal series, and may be scorable in their matrix so that Macronaria incertae sedis isn't the most parsimonious placement, etc etc..  To their credit, the authors do assume indeterminacy less than most authors, and call several other taxa only provisionally indeterminate until they are studied in more depth.

Okay, I have another general complaint as well, and that's that no alternate positions for taxa were tested.  The authors do run a bootstrap-esque test, but those aren't useful for trees with incomplete taxa since they easily destabilize clades that are otherwise well supported.  They do have a handy Table 7 that lists alternative placements for taxa by different authors, which will make it easy for someone to check their homoplasy in the future though.  Furthermore, the analysis in general is well done, with ordered characters, various tests of how to work in continuous states, and well described characters.  Their phylogenetic definitions are also top notch, using eponymous type species.  Maybe with Giraffatitan falling closer to Cedarosaurus, Abydosaurus and Sonorasaurus than to Brachiosaurus, those refusing to use the new genus name will finally change their mind.

Oh, and I hate the continuation of the myth that "Titanosaurus is indeterminate and thus its co-ordinated rank-taxa must be abandoned (Wilson & Upchurch, 2003)."  Why oh why did Wilson and Upchurch start the legend that indeterminate taxa can't have eponymous ranks in the ICZN?

But besides these minor problems, the paper's excellent.  And it's open access, at the address below!  I highly recommend anyone interested in sauropods check it out, as while it doesn't solve all the problems with basal macronarians, it starts us on the right track for many of them.

Reference- Mannion, Upchurch, Barnes and Mateus, 2013. Osteology of the Late Jurassic Portuguese sauropod dinosaur Lusotitan atalaiensis (Macronaria) and the evolutionary history of basal titanosauriforms. Zoological Journal of the Linnean Society. advance online publication.
DOI: 10.1111/zoj.12029
http://onlinelibrary.wiley.com/doi/10.1111/zoj.12029/pdf

No comments:

Post a Comment