[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Re: Mickey's experiment (segnosaurs)
Ken Kinman wrote-
> The idea of a Deltadromeus-Ornithomime clade does not "feel" right
> me at all (ooooppps, there's that "unscientific" intuition rearing its
> head again). This is the one part of Rauhut's topology that jumped out at
> me, and I agree with Mickey that Deltadromeus is too primitive to be a
> sister group to ornithomimes. If Deltadromeus is a coelurosaur at all, it
> is a very basal form, more basal than even Tyrannosaurids, much less
Okay. First, as I said, Rauhut's study was mainly on non-coelurosaurs. The
coelurosaurian relations were interesting, but not looked at as closely as
ceratosaur and carnosaur relations. This probably explains some odd things
like Deltadromeus grouping with ornithomimosaurs and troodontids being
deinonychosaurs. The statements I made about how only two characters
support the Deltadromeus-ornithomimosaur clade and how only one step is
needed to make Deltadromeus a basal coelurosaur were directly from Rauhut's
thesis. Oliver knows this is an unlikely grouping and commented on it
"However, straight humeri are found in a variety of theropods (ceratosaurs,
advanced spinosauroids, tyrannosaurids), and the morphology of the humerus
in Deltadromeus is otherwise unlike that of ornithomimosaurs. Likewise, the
distribution of a deep groove in the proximal fibula is patchy in theropods,
and more information on the detailed morphology and probable biological
significance of this character is needed for a better evaluation of its
value for phylogenetic reconstructions. Several plesiomorphic characters
indicate that Deltadromeus might be a much more basal coelurosaur: the
mid-caudal chevrons show an intermediate morphology between the rod-like to
L-shaped chevrons of basal tetanurans and the skid-like chevrons of almost
all other coelurosaurs. Furthermore, the pubic boot is broadly triangular in
ventral view, as in all basal neotheropodans and tetanurans, but unlike the
narrow boot with subparallel margins as found in all other coelurosaurs.
Finally, a horizontal groove is present anteriorly on the astragalar
condyles. Such a groove is known in ceratosaurs and carnosaurs, but it is
not found in other coelurosaurs. Since placement of Deltadromeus at the base
of Coelurosauria only requires one additional step, it seems likely that
more material of this taxon might show it to be a basal coelurosaur rather
than a sister-group to ornithomimosaurs."
I do however think that Deltadromeus is basal to both ornithomimosaurs and
tyrannosauroids. It comes out this way in my most recent analysis and has
done so nearly every time, with the odd exception when it is a basal
> I have a feeling Mickey's experiment will show this. What is more
> exciting is the unexpected results that it might generate. Mickey seems
> have a real knack for teasing interesting possibilities out of all those
> characters (so many characters, that frankly it sometimes makes my head
> swim). Maybe Sereno should be a little worried. Especially since Mickey
> is just getting started and I for one already trust his intuition a lot
> than Sereno's (particularly when it comes to cladistic analysis).
> certainly isn't the deciding factor, but my experience is that those who
> have a more intuitive feel for characters have an easier time detecting
> convergences (and other homoplasies) than those whose intuition is not as
> well developed.
Although I suppose the experiment did place Deltadromeus away from
ornithomimosaurs, I'm uncertain why it would be expected. Removing useful
characters from an analysis makes it less accurate, not more so. Any
unexpected results would presumedly be due to the absence of evidence to the
contrary, such as tyrannosauroids grouping with carnosaurs before I removed
Proceratosaurus. All the coelurosaur and tyrannoraptoran characters used by
Holtz and others to place tyrannosaurs in the Coelurosauria were removed, so
the small amount of carnosaur characters present took over.
Thanks a lot for the compliments and comparisons to Sereno. I really
appreciate it. What you have to realize about Sereno's analyses is that
they're not done the same way as mine, Tom's or Oliver's. Sereno for the
most part only includes characters that support his tree. Thus he is not
really testing to see if his phylogeny is more parsimonious, but rather
representing his ideas on phylogeny with a cladogram. The fact that his
analyses have less characters included that would contradict his phylogeny
can be seen in bootstrap and consistancy index values. If you ever ran one
of his matrices through PAUP, you would see almost every bootstrap value
(support for nodes) is over 90%. Sereno's 1999 tetanurine analysis took
2.08 seconds to bootstrap and all but one clade is over supported over 90%.
Mine would take an eternity to bootstrap, but Oliver's took 484 hours(!) and
only five clades (out of 37) were supported over 90%. His consistancy
indices (CI) are also extremely high. The CI for his 1999 Tetanurae
analysis is .67, my coelurosaur analysis' is .35, Oliver's theropod
analysis' is .42 and Holtz's (2000) theropod analysis' is .44. Thus,
Sereno's analyses are not really comparable to others', as they are
constructed differently. Hmm, I'm off on a tangent :-) . In any case.....
While I do like your compliments about my knack for getting interesting
results (is that a good thing :-) ), they had no part in the current
experiment, as I was simply using Oliver's matrix and removing specified
> I am, of course, disappointed that Therizinosauria are not
> in your experiment, as I really expected them to be so. I'll accept that.
> That they are coelurosaurs seems to be strongly supported by your
> However, I still think they may be slightly paraphyletic, with the
> of Therizinosauroids (Alxasaurus, Therizinosaurus, and segnosaurids)
> splitting off as a separate clade just before Beipiaosaurus, which would
> a plesion by itself or maybe even a basal oviraptorosaur sensu lato).
Well, accepting segnosaurs as coelurosaurs and Therizinosauroidea as
monophyletic is a start.....
> However, what REALLY struck me was the Deltadromeus-tyrannosaur
> that emerged. Holy Smokes, I wasn't expecting that at all!!!! My
> reaction to this is that tyrannosaurs may have split off earlier in the
> coelurosaur radiation than generally believed.
Well, as I said above, it sometimes works that way in my analysis. And both
Holtz (unpublished) and I now have tyrannosaurs basal to maniraptoriformes.
Still, the results from the experimental analysis are more likely to be
wrong than the complete analysis, merely because we're losing good
> But finally back to therizinosaurs again, I think it might be
> interesting to see what your tree topology would look like if
> were completely omitted from the analysis. In that case, would the
> Alxasaurus-Therizinosaurus-segnosaur clade still emerge and fall out
> to oviraptorosaurs? At least that is the one further test that I think
> might show something interesting.
Let's see what happens....
I ran Rauhut's matrix, excluded the 66 coelurosaurian etc. characters, and
deleted both Proceratosaurus and Beipiaosaurus. I ran it and over 75000
MPT's resulted. There are 431 steps, three less than before, showing
Beipiaosaurus' absence affected the result. The consensus is-
| | |_Liliensternus airelensis
| |_Abelisauridae (including Noasaurus)
|__________Magnosaurus (including Eustreptospondylus)
|___________Spinosauridae (Baryonychidae is used)
|___________Carcharodontosauridae (including Acrocanthosaurus)
| | |______Therizinosaurus
| | |______"Segnosauridae"
| |_________Caenagnathoidea (Oviraptorosauria is used)
|_________Dromaeosauridae (including Sinornithosaurus)
|_____Aves (including Rahonavis)
As you can see, even without Beipiaosaurus, therizinosauroids are
monophyletic and still closest to oviraptorosaurs. The presence of Coelurus
around this clade is rather odd, but probably the result of the fragmentary
nature of the material.