A critique of Sullivan's pachycephalosaur paper,

[Michael Mortimer <mickey_mortimer111@msn.com>]

Sullivan has a new paper out, and the pdf is available on his website at-
http://www.robertmsullivanphd.com/uploads/130_Sullivan__2006__-Pachycephalosauridae.pdf

It's a long paper, and I don't have time to critique it all.  But suffice to 
say, I'm not impressed.  There are a number of problems in the basic 
approach Sullivan takes, which will be illustrated with some examples.

First, Sullivan synonymized Pachycephalosauria with Pachycephalosauridae.  
He calls the former taxon redundant and only uses it in quotes.  No mention 
of their phylogenetic definitions is given at all (one is node-based, the 
other branch-based, thus they could only be used as synonyms if Stegoceras 
were the most basal pachycephalosaur known).  Instead, Sullivan basically 
asserts that the traditionally basal 'homalocephalid' grade taxa are nested 
within high-domed taxa.  He defends this with some spurious reasoning.  For 
instance, because taxa like Dracorex have flat skull roofs, but otherwise 
resemble pachycephalosaurins, the high dome can't be used to help diagnose a 
derived pachycephalosaur clade.  Uh... er... just because a character 
exhibits homoplasy doesn't mean it's impossible to use in diagnoses.  
Sullivan seems to have some typographic concept of taxa, where they are 
defined by 'key characters', and if that key character reverses, the taxon 
is invalid.  Also, flat-domed taxa appear later than some high-domed taxa 
(an unnamed Santonian form), and there can't be a ghost lineage of 
flat-domed taxa because Stygimoloch has a small dome yet lived in the Late 
Maastrichtian.  Uh.... so because some high-domed taxa later reduced their 
dome, other flat-domed taxa can't be more basal, and must have come from 
high-domed taxa too?
Similarly, Sullivan rejects Pachycephalosaurinae, because Sereno (2000) 
seemingly described it as fully-domed (and domes might reduce or be lost!).  
Completely ignoring the fact that Sereno defined the clade phylogenetically, 
which doesn't rely on morphology at all.

Stenopelix is relegated to Ornithischia incertae sedis.  Not because its 
pachycephalosaur characters are invalid, but because the "features are not 
readily preserved in most pachycephalosaurid specimens, and their absence 
and/or presence in non-pachycephalosaurid taxa are not well-established."  
The characters are postcranial (obviously) and easily evaluated in putative 
pachycephalosaur outgroups (ceratopsians, heterodontosaurids, 
hypsilophodonts, etc.).  I don't see how they aren't well established in 
outgroups, and the fact they're not preserved in most pachycephalosaurs (due 
to incompleteness) doesn't matter.  It's like Suzuki et al. (2002) rejecting 
arctometatarsalian characters of alvarezsaurids because they could only be 
observed in Shuvuuia among alvarezsaurids.  Taxa which don't preserve a 
character can be ignored when evaluating that character.

Sullivan's diagnosis for Pachycephalosauridae reads like one of those 
classic 'diagnoses' from the 1950's that lists symplesiomorphic or variable 
characters in addition to synapomorphies.  "Ornithischian dinosaurs with 
thickened, fully-flat
or incipiently to fully-domed frontoparietals."  So... the thickening is all 
that needs to be mentioned then.  "Supratemporal fenestrae absent to 
well-developed."  Uh... this would exclude which states now?

The author questions marginocephalian monophyly because he views the 
low-domed taxa as being derived.  Thus, the parietosquamosal shelf would be 
convergently developed with ceratopsians'.  I don't see the logic here, as 
even some high-domed taxa like Stegoceras have parietosquamosal shelves.  
Sullivan would have to show taxa like Colepiocephale, which lack the shelf, 
are the most basal pachycephalosaurs.  And thus hasn't been done yet.

Finally, anyone following pachycephalosaur taxonomy is familiar with the 
conflicting interpretations of several taxa between Sullivan (2003) and 
Williamson and Carr (2002).  Apparently Ryan and Evans (2005) tried to sort 
out the conflict between the analyses, though I have yet to see the paper.  
I don't have an opinion regarding who's correct, but some of Sullivan's 
reasoning is questionable.  For instance- "Ryan and Evans (2005) resurrected 
Stegoceras breve despite the distinct nature of the strongly down-turned 
parietal and paired nodes that clearly indicate it is not Stegoceras. 
Contrary to their statement, there is no demonstrable growth series for this 
species, and primitive features (i.e., horizontal temporal chamber; frontal 
grooved anteriorly) do not support reference to a more ?primitive? taxon 
(i.e., Stegoceras sensu stricto). The compelling feature is a single row of 
nodes, presumably on the squamosals, that lap onto the strongly downturned 
posterior part of the parietal. This character is not seen in Stegoceras but 
is also present in Prenocephale edmontonensis."
Why one character should be 'compelling' while other characters aren't isn't 
obvious.  We're back to the dark ages of key characters and personal 
evolutionary scenarios.

The paper has its good points however.  There are excellent photos of 
numerous skulls, including Tylocephale and Prenocephale.  Also, there are 
the only detailed evaluations of "Troodon" bexelli and Heishansaurus I've 
seen.

Bakker et al.'s description of Dracorex (available on the same website) is 
just as phylogenetically useless.  After a long spiel about the problems 
with assuming characters have equal weight (with some knee-slappers like 
"PAUP-fiction" thrown in - guh-hyuk), the authors don't provide an objective 
alternative.  Instead, we get a 'scenario' (their term) that's basically 
what cladistic analyses have shown - the long snout and large spike clusters 
of pachycephalosaurins are derived.  Thus Dracorex would have a secondarily 
reduced dome, like its presumed relative Stygimoloch.  Bakker et al. seem to 
misunderstand cladistics the same way Charig and Milner (1990) did when they 
were trying to place Baryonyx in Gauther's (1986) phylogeny.  If a taxon 
exhibits homoplasy, the whole phylogeny is claimed to be doomed.  They say 
"Most recently published phylogenies assume that the large dome evolved once 
and was never lost (Sereno, 1986, 2000). Such an assumption makes Dracorex 
hogwartsia difficult to interpret."  But cladistic analyses assume no such 
thing.  It's just a result of the data entered at the time.  Put Dracorex in 
there and the dome may end up most parsimoniously lost.

In all, the two papers didn't affect my opinion of pachycephalosaur 
phylogenetics at all because they just contain a bunch of unfounded 
assertions.  'Homocephalids' may be derived from high-domed taxa, but a 
stratigraphic difference of two stages and an analogy with Dracorex isn't 
convincing.  Until it is convincing, any affect it has on marginocephalian 
monophyly is hypothetical.  The rationale behind excluding Stenopelix from 
Pachycephalosauria is flawed, and there's no alternative presented in any 
case.  The rejection of Pachycephalosauria and Pachycephalosaurinae 
completely ignores phylogenetic nomenclature, so means nothing.  The papers 
left me with no idea how the authors think pachycephalosaurs are related to 
each other or to other ornithischians, except that pachycephalosaurins are 
monophyletic and Sullivan finds a relationship with ankylosaurs more 
convincing than one with ceratopsians.  Someone needs to enter Dracorex and 
Alaskacephale into Williamson and Carr's matrix and get some science done on 
its phylogenetic affinities.

Mickey Mortimer