|
Ken Kinman suggested that I analyze segnosaurs as
basal theropods. He believes that segnosaurs may be polyphyletic if they
are compared to a range of tetanurines without using synapomorphies for
Coelurosauria and more exclusive clades. I will do this utilizing Rauhut's
(2000) analysis of theropods, which focuses on non-coelurosaurs. The
details of this analysis can be seen in my previous post, and a copy of the
thesis is available to those of you who contact me offlist.
Therizinosauroidea is an OTU in the analysis and is the sister group of
Oviraptorosauria, as in other recent analyses (Makovicky and Sues 1998, Holtz
2000, mine unpublished). First, I coded Beipiaosaurus, Alxasaurus,
Therizinosaurus and "segnosaurids" as separate OTU's and entered them into the
matrix instead of Therizinosauroidea. Remember that no segnosaur
synapomorphies were used, as they were pointless in the original analysis.
This results in 16632 trees of 676 steps (24 more than the original run).
The tree is identical to the original, except within Maniraptora, where the
topology is now-
|_____________Beipiaosaurus
|
|____Alxasaurus
|
|___Therizinosaurus
| |___"Segnosauridae"
|______________Caudipteryx
|
|_____Microvenator
|
|___Caenagnathoidea (Oviraptorosauria is used)
|
|___Avimimus
|______________Troodontidae
|
| |___Dromaeosauridae (including
Sinornithosaurus)
|
|_________Unenlagia
|
|_____Aves (including Rahonavis)
|______________Bagaraatan
Note the two changes- the segnosaur-oviraptorosaur
clade is not present in some trees, and Bagaraatan is resolved as a basal
paravian (as in my analysis). Of course, a segnosaur-oviraptorosaur clade
is present in some trees, so such a result requires 0 more steps. This
result is important, as it shows segnosaurs group together even without
synapomorphies of the group included. Also, Beipiaosaurus is the most
basal member, as would be expected.
Next, I removed all coelurosaurian,
tyrannoraptoran, maniraptoriform, maniraptoran and segnosaur-oviraptorosaur
characters from the analysis (as well as those characters supporting
intermediate, but unnamed clades). This is easily done, as Rauhut includes
a list of all characters that diagnose each taxon. I excluded those
characters found with DELTRAN, which supports parallels over reversals.
This excluded 66 characters, leaving only 158. A potential problem with
doing this is that certain coelurosaurian clades may be diagnosed by
ordered multistate characters. For example, Maniraptora may
be characterized by state 2 of character 50, and Tetanurae may be
characterized by state 1 (which includes state 2). By excluding character
50, both support for Maniraptora and Tetanurae is lost. This could be
solved by making multistate ordered characters multiple characters, one for each
state, then excluding the ones with coelurosaurian states. Such a process
is time consuming however, so I did not try it. Another problem is that
some characters characterizing coelurosaurian clades may also characterize more
basal clades in parallel. Thus, excluding these characters may lower
support for more basal clades as well. Keeping these problems in mind, a
huge number of MPT's were generated (>55000). These are 434 steps long
and very unresolved, as might be expected if every character supporting
Coelurosauria and higher clades was removed.
_____Euparkeria
|_____Marasuchus
|_____Ornithischia
|_____Sauropodomorpha
|_____Eoraptor
|_____Staurkiosaurus
| |_Herrerasaurus
|______Shuvosaurus
|________Liliensternus
liliensterni
| | |_Liliensternus
airelensis
|
|_Gojirasaurus |
|__Segisaurus
|
|_Syntarsus
|
|_Coelophysis
|________Dilophosaurus
|___________Ceratosaurus
|
|__Elaphrosaurus
|
|_Abelisauridae (including Noasaurus)
|_________Piatnitzkysaurus
|_________"Szechuanosaurus" zigongensis
|__________Magnosaurus (including Eustreptospondylus)
| | |_______Monolophosaurus
| | |_______Poekilopleuron
| | |_______Torvosaurus
| | | |____Chilantaisaurus
| | | |____Spinosauridae
(Baryonychidae is used)
| | |_______Afrovenator
| | |_______Allosaurus
| | |_______Sinraptoridae
| | |_______Neovenator
| | |_______Carcharodontosauridae (including
Acrocanthosaurus)
|
|_________Deltadromeus
| |________Stokesosaurus
|
|____Tyrannosauridae
|____________Proceratosaurus
|____________Coelurus
|____________Sinosauropteryx
|
|___Compsognathus
|
|__Santana compsognathid
|_____________Ornitholestes
|_____________Ornithomimosauria
|_____________Bagaraatan
|_____________Beipiaosaurus
|_____________Alxasaurus
|
|_____Therizinosaurus
|
|_____"Segnosauridae"
|_____________Caudipteryx
|_____________Microvenator
|_____________Caenagnathoidea
(Oviraptorosauria is used)
|_____________Avimimus
|_____________Troodontidae
|_____________Dromaeosauridae
(including Sinornithosaurus)
|_____________Unenlagia
|_____________Aves
(including Rahonavis)
Note that although Beipiaosaurus is in the
tetanurine polytomy, Therizinosauroidea is one of the few clades
remaining. Not even oviraptorosaur or eumaniraptoran clades remain, which
is unexpected, as I did not exclude their characters. This is very
good evidence for therizinosauroid monophyly. I examined several random
trees (numbers 1, 10000, 20000, 30000, 40000 and 50000) and noticed the
phylogeny around segnosaurs was the same (sister group of oviraptorosaurs) and
Beipiaosaurus grouped with therizinosauroids, but Proceratosaurus was jumping
around to various positions in the tree. As the taxon is only known from a
partial skull, it no doubt had even fewer codable characters now and could not
be placed precisely. This also explains why Oviraptorosauria and
Eumaniraptora were not found- Proceratosaurus was a member of those clades in
some trees. Thus, I deleted it and ran the analysis again. Yet
again, many trees were generated (>32000; 434 steps long) and there was a
large polytomy. Note that because the number of steps in these trees is
the same as the previous analysis, Proceratosaurus was adding no information and
did not affect tree topology. This time however, the consensus tree was as
follows-
_____Euparkeria
|_____Marasuchus
|_____Ornithischia
|_____Sauropodomorpha
|_____Eoraptor
|_____Staurkiosaurus
| |_Herrerasaurus
|______Shuvosaurus
|________Liliensternus
liliensterni
| |
| |_Liliensternus airelensis
| |
|_Gojirasaurus | |
|__Segisaurus
| |
|_Syntarsus
| |
|_Coelophysis
|
|______Dilophosaurus
|___________Ceratosaurus
|
|__Elaphrosaurus
|
|_Abelisauridae (including Noasaurus)
|__________Piatnitzkysaurus
|__________"Szechuanosaurus" zigongensis
|__________Magnosaurus (including Eustreptospondylus)
|___________Monolophosaurus
|___________Poekilopleuron
|___________Torvosaurus
|___________Chilantaisaurus
|___________Spinosauridae (Baryonychidae is used)
|___________Afrovenator
|___________Allosaurus
|___________Sinraptoridae
|___________Neovenator
|___________Carcharodontosauridae (including Acrocanthosaurus)
|___________Deltadromeus
| |________Stokesosaurus
|
|____Tyrannosauridae
|____________Coelurus
|____________Sinosauropteryx
|_____________Compsognathus
|_____________Santana
compsognathid
|_____________Ornitholestes
|_____________Ornithomimosauria
|_____________Bagaraatan
|_____________Alxasaurus
| | |
|_____Therizinosaurus
| | |
|_____"Segnosauridae"
| |
|_______Beipiaosaurus
|
|_________Caudipteryx
|
|_____Microvenator
|
|_____Caenagnathoidea (Oviraptorosauria is used)
| |_____Avimimus
|_____________Troodontidae
|_________Dromaeosauridae (including Sinornithosaurus)
|_________Unenlagia
|_____Aves (including Rahonavis)
So even without coelurosaur, maniraptoriform,
maniraptoran, segnosaur-oviraptorosaur, segnosaur or therizinosauroid
characters, segnosaurs are monophyletic and the sister group of
oviraptorosaurs. Additionally, Beipiaosaurus is the most basal taxon, as
is expected. There is no tendancy for segnosaurs to be para- or
polyphyletic and no indication they clade with any non-maniraptoran taxa.
This is in opposition to Ken's hypothesis and suggests that the current
consensus of paleontologists, that segnosaurs are monophyletic and sister group
to oviraptorosaurs, is correct. If Ken would like me to try another
experiment, he is welcome to suggest one.
Mickey
Mortimer |