Sunday, May 27, 2012

The Yurgovuchia analysis isn't all it's cracked up to be

When I first saw Senter et al.'s 2012 analysis for the Yurgovuchia paper, I was excited.  Senter (2007) did a better job coding than the versions of the TWG analysis leading up to it, and a better job at defining characters to be non-composite.  Senter's also gradually increased his taxonomic scope to impressive levels.  But now that I'm integrating the Yurgovuchia matrix data into the Lori matrix, things aren't as great as they seem.

First, that taxonomic scope.  Eotyrannus and microraptorian NGMC 91 are missing without explanation.  Similicaudipteryx isn't there from the Xiaotingia analysis.  Chirostenotes not only lacks Caenagnathus data (Caenagnathus was its own OTU in Senter, 2007 but was deleted from his 2010 paper to decrease computational time, something that might have been useful then since that paper was not primarily a phylogenetic analysis, but which is inexcusable now), but it also lacks data from ROM 43250, the holotype of Epichirostenotes.  Even though I don't think Epichirostenotes can be distinguished taxonomically from Chirostenotes, that would be fine if he made an OTU for Epichirostenotes too.  But no, he didn't.  So we're left with a matrix with no cranial, mandibular, presacral, caudal or pubic data for caenagnathids.  Is it any wonder they're deeply nested within oviraptorids?  For Archaeopteryx, starting in the 2010 paper Senter included "data only from the Thermopolis, Berlin, and Munich specimens. New observations (Mayr et al., 2007) have cast doubt on the conspecificity of the London specimen with those three specimens, so data that came only from the London specimen were deleted from the OTU."  But the holotype is the London specimen!  Thus Senter's OTU does not include the only specimen which is definitely Archaeopteryx.  It's also the only specimen to preserve a lot of braincase data.  Again, he didn't make a separate London OTU, so the data just isn't used.  In Dromaeosauridae, since Senter breaks down Microraptor into the holotype, the CAGS specimens described by Hwang et al. (2002), M. gui and Cryptovolans, we miss out on most cranial data from specimens with complete skulls like IVPP V13475.  He also stopped including Atrociraptor and Saurornitholestes, and deleted the data from the referred Utahraptor specimens.

Perhaps worse are the characters and coding though.  Surprisingly, this is actually coded worse than his 2007 and 2010 versions, with all members of a clade often coded identically regardless of actual variation. As an example, all oviraptorosaurs and paravians are coded as lacking prefrontals. So he recoded Deinonychus incorrectly, and left Sinornithosaurus suspiciously unknown. Or the next character, postorbital bar absent, is incorrectly recoded to give Shuvuuia a complete postorbital bar and leave Avimimus unknown.  Senter also disregards the rule of cladistic analyses to let the tree decide homoplasy.  Since 2007, he's had a pet hypothesis ornithomimosaurs lack promaxillary fenestrae and that the fenestrae below their maxillary fenestrae were not homologous.  This was somewhat defensable in 2007 since other tetanurines have fenestrae arranged horizontally instead of vertically.  But now that Tahara and Larsson (2011) found Dromiceiomimus' lower fenestra communicates with its promaxillary recess, things are unambiguous.  It IS a promaxillary fenestra, even if it was convergently evolved.  The Yurgovuchia paper notes this and addresses it as follows-

"However, the absence of that opening in basal ornithomimosaurs suggests that its appearance in advanced ornithomimosaurs is neomorphic. Therefore, the phylogenetic data matrix used here does not recognize the presence of the promaxillary fenestra in Ornithomimus and other advanced ornithomimosaurs."

But... but... by that reasoning, we might as well just code Avimimus, ornithomimids and derived troodontids as lacking an arctometatarsus, since its absence in basal relatives of those groups show it developed convergently with tyrannosaurids.  Ditto for any character with homoplasy.  By coding this way you completely destroy the point of using PAUP in the first place.  I wonder if this is also the reason for the miscodings alluded to above.  Is Shuvuuia coded as having a complete postorbital bar because it's convergent with the incomplete bar of Aves?  I certainly hope this wasn't Senter's reasoning.

Characters themselves are often "redefined" to be worse.  The 'depth of the jugal beneath the orbit' character has gotten another state for when it tapers anteriorly, but that should be a whole new character.  The character coding for the cross section of the posterior jugal process has been redefined to code for its length and depth, but those aren't the same as cross section AND are independent of each other.  Or take this hilarious example-

"External naris an elongate oblong, with long axis diagonal in lateral view (0) or an elongate oblong with long axis subhorizontal (1) or a short oblong, with length not much greater than height, with long axis subhorizontal (2) [3].  This character replaces a previous one: “Internarial bar rounded (0) or flat (1) [5].” The internarial bar appears flat when the long axis of the external naris is subhorizontal."

Er... the flat internarial bar character refers to its cross section, not its dorsal convexity in lateral view.  A lot of characters from previous TWG analyses are now excluded without comment, so I have to retract my statement on the DML that this analysis is definitely better than prior TWG analyses, because now we have the complication that it's missing some characters.

It goes without saying that no characters are ordered and no codings are polymorphic.

All of this is such a shame, because Senter's 2007 work was such an improvement on other analyses up till that point, and he's been quite a help to me personally over the years.  I'm afraid I can't recommend the Yurgovuchia analysis for those wishing to know coelurosaur phylogeny, and hope that Senter's next version fixes these problems.

References- Senter, 2007. A new look at the phylogeny of Coelurosauria (Dinosauria: Theropoda). Journal of Systematic Palaeontology. 5(4), 429-463.

Senter, 2010. Using creation science to demonstrate evolution: application of a creationist method for visualizing gaps in the fossil record to a phylogenetic study of coelurosaurian dinosaurs. Journal of Evolutionary Biology. 23(8), 1732-1743.

Tahara and Larsson, 2011. Cranial pneumatic anatomy of Ornithomimus edmontonicus (Ornithomimidae: Theropoda). Journal of Vertebrate Paleontology. 31(1), 127143.

Senter, Kirkland, DeBlieux, Madsen and Toth, 2012. New dromaeosaurids (Dinosauria: Theropoda) from the Lower Cretaceous of Utah, and the evolution of the dromaeosaurid tail. PLoS ONE. 7(5), e36790. 

No comments:

Post a Comment