Tuesday, January 17, 2012

Agnolin and Novas's (2011) cheating for unenlagiine monophyly?

Here's a quick example of the fun I find when checking matrices for my analysis.  Agnolin and Novas (2011) took Hu et al.'s (2009) matrix and recoded some paravians, as well as adding three new characters to support unenlagiine monophyly- maxilla elongate, postantral wall extensive, teeth grooved.  They got a result that's fairly unique in placing unenlagiines as avialans instead of dromaeosaurids.  I'm adding the data into my matrix and notice the matrix has five new characters added instead of three, all of which are only present in Buitreraptor and Austroraptor.  Well, except the last one, which is seemingly the third they list (grooved teeth) since its codings would suggest it is tooth-based (though wrongly coded for Garudimimus, ornithomimids and Confuciusornis) and present in Sinornithosaurus (and incorrectly in Compsognathus).  As if two phantom characters weren't bad enough, the authors clearly didn't actually check if taxa had them, instead merely going through the matrix and marking anything with cranial material '00000'.  This includes taxa which can't be coded for maxilla elongation or maxillary postantral wall elongation like Tanycolagreus, Coelurus, Mononykus, Falcarius, Alxasaurus, Protarchaeopteryx and Caenagnathus, and even some taxa with no cranial material like Archaeornithomimus and Anserimimus.  Sure none of those taxa are close to unenlagiines so their coding really doesn't matter here, but it sets such a bad precident.  Someone who didn't notice this could go through the matrix and see Protarchaeopteryx is coded as lacking a posteriorly extensive postantral wall and think the authors must have examined the material since it's certainly not illustrated or stated in the literature, and thus code it that way in their own matrix.  But no, I've seen the specimen and the whole maxillary area is crushed so much that you can't even tell if the hole in it is the antorbital fenestra or just damage, let alone see any internal details of the thing.  I'm not even going to go through their new character codings because I can't trust them one bit.

And by the way, contra Agnolin and Novas, their matrix generates 16210 most parsimonious trees of 1409 steps, not 145 trees of 1440 steps.  I was suspicious about the added unenlagiine characters, both real and phantom, but it turns out that the trees have the same topology when either the two phantom characters or all five new characters are excluded.  So I guess they weren't cheating for unenlagiine monophyly after all.

It's amazing how young most theropod specimens are

Theropod analyses contain so many characters that are ontogenetically variable.  Not just the numerous fusion characters (e.g. premaxillae, parietals, dentaries, cervical ribs, pygostyle, scapulocoracoids, pelvis, ischia, tibiotarsi, tarsometatarsi), but also less obvious things.  Skull shape, orbit shape, maxillary fenestra position, postorbital orbital process presence, nasal rugosity, premaxillary tooth serration, pubic boot size, and more.  We know that if you try to code juveniles for these characters, you get the wrong topology.  Juvenile tyrannosaurines end up basally placed in that clade (Carr, 2005; near certainly Brusatte et al., 2009 for Alioramus); juvenile confuciusornithids end up basal to Pygostylia (Gao et al., 2008 for Zhongornis).  And yet, basically no one accounts for this when coding.  You might think that it shouldn't matter that much since comparatively few juvenile theropods are known, but I'm checking basically phylogeny-neutral fusions like braincase, neurocentral sutures and between sacral centra for my upcoming analysis and I'm surprised by just how few adult specimens we have.  Among basal coelurosaurs, for instance-

Tanycolagreus' holotype has visible and open dorsal sutures, free sacral centra and is 63% the size of another specimen.
Coelurus' holotype has visible cervical and some dorsal sutures, open sutures on other dorsals and on caudals, and free sacral centra.
Ornitholestes' holotype has an unfused braincase, visible presacral and some proximal caudal sutures.
Nedcolbertia has open proximal caudal sutures on even the largest specimen.
Nqwebasaurus has open dorsal and caudal sututes.
Santanaraptor has visible caudal sutures.
Mirischia has open and visible dorsal sutures, open sacral sutures and sutured sacral centra.
Compsognathus' largest specimen has open and visible cervical sutures, visible dorsal, sacral and caudal  sutures, and free and sutured sacral centra.  The holotype is 62% its size.
Huaxiagnathus has open caudal sutures.
Sinocalliopteryx has a completely unfused braincase, at least visible cervical sutures and visible dorsal sutures.
Juravenator has open sacral and caudal sutures and free sacral centra.
Scipionyx has everything free of course.
In fact, among sampled basal coelurosaurs (excluding tyrannosauroids and maniraptoriforms), only Aniksosaurus seems to be perhaps an adult based on obliterated sutures on a cervical and a caudal.

Similicaudipteryx's holotype is probably an adult based on obliterated presacral and sacral sutures and has a pygostyle.  Referred specimen STM22-6 is 64% its size (length proxied by femoral length) and also has one, but STM4-1 is 17% its size and doesn't.  Zhongornis is 26% the size of adult Confuciusornis and didn't either, but specimens are known only 64% the size of the largest adults and have pygostyles.  Did Sinosauropteryx have a pygostyle?  Of course not you'd say, the holotype has a series of sixty-four unfused caudal vertebrae (the two larger described specimens don't preserve the distal tail).  The largest specimen lacks fused neurocentral sutures on its proximal caudals, has sutured but unfused sacral centra, and did not fuse at least one sacral rib to its vertebra.  So it's immature and the holotype 64% its size.  Thus the holotype must be quite a bit less than 64% adult size which falls within the range of pygostyle uncertainty in Similicaudipteryx, and so would be better coded unknown.  Of course I consider it unlikely given its phylogenetic position, but codings shouldn't assume that.  Now remember that every ontogeny-related character has its own timing (that probably varied between taxa, but our sample sizes and description quality doesn't even let us go there), and you can see the problem. 

References- Carr, 2005. Phylogeny of Tyrannosauroidea (Dinosauria: Coelurosauria) with special reference to North American forms. Unpublished PhD dissertation. University of Toronto. 1170 pp.

Gao, Chiappe, Meng, O'Conner, Wang, Cheng and Liu, 2008. A new basal lineage of Early Cretaceous birds from China and its implications on the evolution of the avian tail. Palaeontology. 51(4), 775-791. 

Brusatte, Carr, Erickson, Bever and Norell, 2009. A long-snouted, multihorned tyrannosaurid from the Late Cretaceous of Mongolia. Proceedings of the National Academy of Sciences. 106(41), 17261-17266.

Friday, January 13, 2012

Please do a better job at alvarezsauroid taxonomy

Among my Database updates were two new alvarezsaurid papers- Agnolin et al. (2012) describing Bonapartenykus and Xu et al. (2012) monographing Linhenykus.  They both raise some interesting issues for the clade, showing alvarezsaur authors need to get their taxonomic act together.

Bonaparte (1991) initially put Alvarezsaurus in Alvarezsauria, but this was largely ignored since his Alvarezsauridae covered known taxa for almost two decades.  Similarly, Livezey and Zusi (2007) used Alvarezsauroidea, but that was basically an afterthought in their huge inaccurate Aves analysis and like Alvarezsauria was redundant with Alvarezsauridae as then known.  Hu et al. (2009) were the first to propose a phylogenetic definition for either and went with Alvarezsauroidea, defining it as a stem-based clade.  Choiniere et al. (2010) then first published a proposed non-alvarezsaurid alvarezsaur (Haplocheirus), and followed Hu et al.'s choice.  Later authors have similarly used Alvarezsauroidea, so we have consensus.  Hoorah.  Except... now Agnolin et al. want to start using Alvarezsauria again "in order to emphasize this morphologically distinctive theropod group", and propose basically the same definition, considering it a senior synonym of Alvarezsauroidea.  Yet due to the ICZN, Bonaparte named Alvarezsauroidea the same time he named Alvarezsauridae, so there's no date or author priority, and plenty of big theropod clades end in -oidea (Coelophysoidea, Megalosauroidea, Tyrannosauroidea).  As for the subjective level of morphological distinction, might I remind the authors they're writing in the twenty-first century.  None of their reasons justify disrupting a taxonomic consensus.

Similarly, once people decided to follow the ICZN and use Parvicursorinae over Mononykinae, the former was given a node-based definition equivalent to the one Mononykinae had (Choiniere et al., 2010).  But now the same group of authors (Xu et al., 2012) want to change it to a stem-based definition that ends up including more basal taxa like Linhenykus and Xixianykus too.  Why?  Because the latter "are morphologically very similar to members of this node-based Parvicursorinae" and "the most informative option is to treat Parvicursorinae as a stem-based taxon."  But under their broader Parvicursorinae, the most basal forms will undoubtedly be very similar to Patagonykus, so by their same logic we would then expand the definition again.  That's why we should be past these 50's-style typological arguments- evolution is a continuum.  Again, even ignoring the faulty logic, phylogenetic taxonomy is only really useful if we leave the names and definitions alone so that when anyone says "Parvicursorinae" everyone will know where it goes in their phylogeny.  It's as if people are using phylogenetic taxonomy, but don't really get the point of why it was introduced in the first place.

On a related note, Agnolin et al. propose Patagonykinae for Patagonykus and Bonapartenykus, but never properly define it with a node- or stem-based definition (and external reference taxon for the latter).  Not that it's particularly useful, as the clade is united by all of two synapomorphies and has only been found in one analysis of course.

So far, the detailed alvarezsauroid analyses have been based on adding taxa to Longrich and Currie's (2009) little alvarezsaurid analysis, which results in Xixianykus and Linhenykus being basal as noted above.  Instead, Agnolin et al. add Ceratonykus, Albinykus, Xixianykus and Linhenykus to Choiniere et al.'s big theropod analysis.  Seems smart, until you realize that Choiniere et al. just left tons of known states uncoded.  For instance, of all theropods, only Stokesosaurus langhami(!) is coded as lacking the traditional alvarezsaurid character of procoelous caudals.  So the analysis is terribly untrustworthy.  In any case, Agnolin et al. find a novel topology where Ceratonykus and Albinykus are basal (proposed by their original describers to be sister to the derived Mononykus and Shuvuuia respectively), Albertonykus is sister to Mononykus instead of being basal, and Linhenykus is sister to Shuvuuia instead of being basal.  Compare below.

Previously published consensus-

|--outgroups
`--+--Linhenykus
   `--+--Albertonykus
      `--+--Xixianykus
         `--+--Parvicursor
            `--+--+--Shuvuuia
               |  `--Albinykus
               `--+--Mononykus
                  `--Ceratonykus

Agnolin et al. (2012)-

|--outgroups
`--+--+--Ceratonykus
   |  |--Xixianykus
   |  `--Albinykus
   `--+--Parvicursor
      `--+--+--Shuvuuia
         |  `--Linhenykus
         `--+--Mononykus
            `--Albertonykus

While I don't trust the latter matrix as noted above, I am open to alternate alvarezsaurid topologies since Longrich and Currie's matrix is small, and hasn't included Albinykus or Bonapartenykus (I added all other known alvarezsauroids and reinforced Alifanov and Barsbold's suggested placement for Ceratonykus at least).  What's irritating though is that Agnolin et al. go on to use their topology to propose and define two new clades.  One is Mononykini (Mononykus olecranus <- Parvicursor remotus, Patagonykus puertai, Alvarezsaurus calvoi), based on one character which Parvicursor actually also has.  Whoops!  The other is Ceratonykini, defined as (Ceratonykus oculatus + Xixianykus zhangi).  This is based on two characters, which only Xixianykus and Albinykus have.  Whoops again!*  Do I even have to say that Agnolin et al. don't test to see how many extra steps the old consensus topology takes to constrain in their analysis?  Note that when applied to the standard topology, their definition of Ceratonykini ends up covering a much wider set of taxa, including Mononykini.  So not only do we have the awkward nested tribes, but the concept is different than intended by Agnolin et al..  And the whole thing is based on meager character evidence that's incorrect anyway.  Surely if you find a new topology from the consensus, you should wait until someone else finds it in a different matrix before applying a name to your novel clade.  Unless your new analysis has strong support compared to the old topology, and you should pay attention to see if that support is based on accurate anatomy.  And even if you feel the urge to name a new clade with strong and accurate support, please choose a definition that works with the old topology if possible.  Agnolin et al. could have just called it Xixianykini and defined it as (Xixianykus zhangi <- Mononykus olecranus).  That retains the same position in both trees (sister to Parvicursorinae), being unaffected by the controversial taxa, and Xixianykus actually has the proposed diagnostic characters.  It's so easy.

So alvarezsaur authors, please don't redefine clades (especially for typological reasons) and please don't name new clades based on poor, inaccurate support, with definitions that don't work well in other topologies.

* Okay, I was writing this post at my parents' from memory of morphology, and now that I'm back home with the Ceratonykus pdf, it may also have the supposed ceratonykine characters (distal tarsals fused to metatarsals; metatarsals fused).  At least the right metatarsus seems to, though the left is more ambiguous.  Fusion is never explicitly mentioned, though they are called 'tarsometatarsals' [sic].  I was right about Parvicursor's second cnemial crest though, ha!  In any case, I stand by my point that such low character support is problematic to name new clades from.

References- Bonaparte, 1991. Los vertebrados fosiles de la Formacion Ryo Colorado, de la ciudad de Neuquen y cercanyas, Cretacico superior, Argentina. Museo Argentino de Ciencias Naturales "Bernardino Rivadavia", Revista (Seccion Paleontologya). 4, 15-123.

Livezey and Zusi, 2007. Higher-order phylogeny of modern birds (Theropoda, Aves: Neornithes) based on comparative anatomy. II. Analysis and discussion. Zoological Journal of the Linnean Society. 149 (1), 1-95.

Hu, Hou, Zhang and Xu, 2009. A pre-Archaeopteryx troodontid theropod from China with long feathers on the metatarsus. Nature. 461, 640-643.

Longrich and Currie, 2009. Albertonykus borealis, a new alvarezsaur (Dinosauria: Theropoda) from the Early Maastrichtian of Alberta, Canada: Implications for the systematics and ecology of the Alvarezsauridae. Cretaceous Research. 30(1), 239-252.

Choiniere, Xu, Clark, Forster, Guo and Han, 2010. A basal alvarezsauroid theropod from the Early Late Jurassic of Xinjiang, China. Science. 327, 571-574. 

Agnolin, Powell, Novas and Kundrat, 2012. New alvarezsaurid (Dinosauria, Theropoda) from uppermost Cretaceous of north-western Patagonia with associated eggs. Cretaceous Research. DOI: 10.1016/j.cretres.2011.11.014

Xu, Upchurch, Ma, Pittman, Choiniere, Sullivan, Hone, Tan, Tan, Xiao and Han, 2012. Osteology of the alvarezsauroid Linhenykus monodactylus from the Upper Cretaceous Wulansuhai Formation of Inner Mongolia, China, and comments on alvarezsauroid biogeography. Acta Palaeontologica Polonica. http://dx.doi.org/10.4202/app.2011.0083

Tuesday, January 3, 2012

(Almost) Famous: I'm (mis)quoted in Feduccia's new book!

Today Brad McFeeters informed me that Feduccia (2011) quoted me from a 2003 DML post in his new book!  This is so great...

"Among the more amusing websites available for this group is the message board "Dinosaur Mailing List" (sponsored by the University of Southern California), whose mission is to "foster dialogue on science pertaining to dinosaurs."  Although there is much of interest on the site, sifting the grain from the chaff can be a daunting task, and numerous amateurs use the site to vent their anger at dissenters, standing like Swiss Guards at the Vatican over all matters pertaining to dinosaurology, including entries in Wikipedia.  After I was interviewed as Discover's "scientist of the month" in 2003 (the article was titled "Plucking apart the Dino-Birds"), a new thread appeared on the Dinosaur Mailing List site.  "'Tis Time to Get Medieval on Alan Feduccia" stated: "Let Feduccia make his bold claims...Let the General go into battle with his army...But!!!! On the other side of the table will be sitting our knights."  A follow-up entry even called for "a scientific crucifixion."  It indeed seems that the weaker the science, the more contentious and vitriolic its adherents.[6]"

Going to the references, we find...

"6. Dinosaur Mailing List [dinosaur@usc.edu; sponsored by the University of Southern California]; M. Mortimer, "'Tis Time to Get Medieval on Alan Feduccia," Dinosaur Mailing List, 3 January 2003, 22:36:31-0880."

But wait, I never wrote that.  As Brad noted, it was Kris Kripchak!  He misread my reply!  Feduccia can't even get quotation right.  What's especially amusing is my response to Kris embodies respect for evidence and science, and the ideal of letting dissenters be heard-

"The problem of course, is that they are ignoring the evidence. It would be interesting if some scientific organization were capable of forcing scientists to address evidence against their theories, the alternative being to lose their degrees or something. But this would only lead to stagnation, as there are also many honest scientists out there who are trying to support potentially correct alternative theories, but might be prevented if they couldn't yet deal with the problems in their ideas. Critics force us to closely examine our theories and are useful for catalyzing paradigm shifts and theory modifications. I suppose a committee could be formed to examine cases and determine when someone was just stubbornly holding on to an idea with no reason, but it would all be too subjective to work. Methodologies are as controversial as morphologies are.

Besides, ABSRDists, despite their innumerable flaws, perform useful work for the paleontological community. They describe taxa (Cryptovolans, Scansoriopteryx, Eoenantiornis, etc.) and perform research (Ruben et al. on nasal turbinates, Harwell et al. on croc respiration) that may not have been performed as quickly otherwise. And mortality will see to it that defunct theories are banished from scientific journals eventually, as younger workers are less likely to believe falsehoods as evidence against them piles up.

Science works, it just takes time."

But hey Feduccia, if you really want a 2003 DML quote by me, here's one that's properly attributed and which I still stand by-

Nov. 26 "Indeed, Feduccia and his colleagues are either incredibly ignorant, or are horrendous liars."