[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Dracorex's phylogenetic position examined with science
After the big deal made by Sullivan and Bakker et al. about Dracorex making
"all previous phylogenetic analyses ... inadequate", what actually happens
when Dracorex is included in Williamson and Carr's (2002) matrix? Does the
universe implode? Well, let's see.
First, I made a NEXUS file from Williamson and Carr's matrix. I left
characters ordered, but altered some of the codings so as to avoid weighting
characters. For instance, character 35 is "Parietosquamosal shelf nodes:
absent (0); present (1).", but character 36 is "Parietosquamosal shelf
nodes: absent (0); in more than two rows (1); in a single row, sometimes
accompanied by ventrolateral corner nodes (2)." This weights the presence
of parietosquamosal shelf nodes, since their absence gets a state in two
characters. So I changed state 0 in character 36 to inapplicable. I also
did this for 26 vs. 21, 34 vs. 51, 41 vs. 40, 43 vs. 8, 47 vs. 46, and 53
vs. 52. Finally, I switched the order of states in character 47 (which
describes which end of the parietosquamosal bar is taller) to be a logical
progression when the character is ordered (medial end taller > ends subequal
> lateral end taller).
I added Dracorex, the undescribed Dracorex-like skull with the low dome
(referred to here as the Galiano taxon), Alaskacephale and
Micropachycephalosaurus. I would add Triebold's specimen ('Sandy'), except
that I don't know exactly which parts are real and which are reconstructed.
I also added Yinlong as another outgroup, because it is similar to
pachycephalosaurs in several ways.
The first 100000 most parsimonious trees had the following strict consensus-
| `--+--UCMP 130051
| |--Stegoceras validum
| `--Sphaerotholus buchholtzae
The position of Micropachycephalosaurus is extremely tenuous, as it is based
on the original brief description and line drawings. The position of
Yinlong should be considered as an indication we need to examine Stenopelix
and Wannanosaurus in light of the fact we now know basal ceratopsians are
very pachycephalosaur-like. Perhaps these other fragmentary basal
pachycephalosaurs are basal ceratopsians instead. Using Yinlong as the
outgroup instead of Psittacosaurus leaves a polytomy between Wannanosaurus,
Stenopelix and Micropachycephalosaurus.
Wannanosaurus and Ornatotholus are based on immature specimens, and both
Sullivan and Williamson and Carr believe the latter to be a juvenile
Stegoceras. It's thus possible Wannanosaurus would have a more derived
position if adult specimens were known. Excluding these juvenile specimens
does not influence tree topology.
The main differences between Williamson and Carr's (2002) and Sullivan's
(2003) trees are-
Hansuessia (including Gravitholus and UCMP 130051 in Sullivan's view; both
Stegoceras sp. indet. in Williamson and Carr's view) and Colepiocephale are
pachycephalosaurines sister to Pachycephalosaurini in Sullivan's tree. Here
they are 'stegocerines', as in Williamson and Carr's published tree.
Sullivan considers buchholtzae a synonym of edmontonensis, and places it,
goodwini and breve/brevis in Prenocephale. In Williamson and Carr's
published tree, buchholtzae and goodwini are sister to pachycephalosaurins,
brevis is a basal 'stegocerine', and edmontonensis is a pachycephalosaurid
that can be excluded from pachycephalosaurins, Prenocephale+Tylocephale and
the present clade of 'stegocerines'. Here, the results are more equivocal.
buchholtzae and goodwini could clade with Prenocephale+Tylocephale or
pachycephalosaurins. edmontonensis is a pachycephalosaurine in this tree,
and can fall into Sphaerotholus (but not Pachycephalosaurini or
Prenocephale+Tylocephale). This makes a synonymization with buchholtzae
possible, as Sullivan suggests. It should be noted the only reason
Tylocephale clades with Prenocephale is because both are Asian.
Note I didn't include Sullivan's characters in the analysis, nor did I check
the accuracy of the codings. So this post isn't a test of his ideas on
pachycephalosaurid phylogeny. I would recommend against referring to
American taxa as Prenocephale though, especially if you keep Tylocephale
separate. Similarly, I would recommend referring to Colepiocephale and
Hansuessia instead of Stegoceras lambei and S. sternbergi respectively,
until their positions are more definitively determined. edmontonensis might
be best referred to Sphaerotholus, as this could work in both phylogenies.
breve/brevis is a more difficult case.
Most importantly for this post, however, is that Dracorex still clades with
pachycephalosaurins, despite its flat skull and large supratemporal
fenestrae. Apparently a little homoplasy did not destroy the cladogram.
The Galiano taxon and Alaskacephale are also pachycephalosaurins, as
predicted. Despite "the conceptual challenges presented by the incongruous
combination of skull adaptations" in Dracorex, or the "unwarranted"
assumption of "mixing apples and oranges, bowling balls and cannon balls,
BB?s and seeds from currants" that Bakker et al. likened to unweighted
numerical cladistic analyses, the cladogram places Dracorex where Sullivan's
and Bakker's subjective scenario does. If people would just try using
cladistics instead of complaining about it, they may be surprised at its
effectiveness. But note that the fact Dracorex is secondarily domeless does
not mean all domeless pachycephalosaurs (Homalocephale, Goyocephale..) are.
Nor does it invalidate Pachycephalosauria or Pachycephalosaurinae, as there
are still several taxa outside of (Stegoceras + Pachycephalosaurus), and
they're placed there for more reasons than lacking a dome. In fact, besides
Pachycephalosauria itself, Pachycephalosaurinae, Pachycephalosauridae and
the basal placement of Goyocephale are the best-supported nodes in the
cladogram. Indeed, excluding the 'key character' "dome present" leaves the
tree intact. 'Key character' indeed.