Me vs. Makovicky et al.- comparison and consensus

[Michael Mortimer <mickey_mortimer111@msn.com>]

Jaime Headden wrote-

> Given the number of "improbable" portions of this topology, the only part 
> of
> the tree I might even not question so far is the (carnosaur (tyrannosaur
> (ornithomime + maniraptor))) generality and perhaps the bird topologies, 
> but
> that is due to a [I profess] unfamiliarity with the distinct morphologies 
> of
> these birds. Thus this entire tree is rife with improbabilities that 
> outweighs
> any of the "improbabilities" in the TWG analyses, Rauhut's topology in both
> thesis and published, and Holtz's analyses.

Is Jaime right?  Is my tree more improbable than either the TWG's numerous 
permutations, Rauhut (2003) or Holtz (2000, 2004)?  Well, it's only fair to 
compare the taxa included in both analyses when comparing their results.

When you only compare the taxa shared between my analysis and Makovicky et 
al.'s, the only differences are in my analysis-
1. Compsognathus is closer to birds than to Sinosauropteryx + Huaxiagnathus.
2. Ornithomimosaurs are closer to birds, moreso than compsognathids and 
Ornitholestes.
3. Alvarezsaurids and enigmosaurs switch places relative to birds.
4. Incisivosaurus is closer to caenagnathoids than Caudipteryx is.
5. Avimimus is outside Caenagnathoidea instead of being a caenagnathid.
6. Microvenator is an oviraptorid instead of a caenagnathid.
7. Citipati is a caenagnathid instead of an oviraptorid.
8. Patagonykus and Alvarezsaurus switch places.
9. Dromaeosaurs are sister to birds instead of to troodontids.
10. Archaeopteryx is a troodontid instead of a bird.
11. Byronosaurus is basal to IGM 100/44.
12. Mei is a basal dromaeosaur instead of a basal troodontid.
13. Sinornithosaurus is closer to dromaeosaurids than to Microraptor.
14. Unenlagia is a dromaeosaurid instead of outside the Microraptor + 
Dromaeosaurus clade.
15. Buitreraptor + Rahonavis are sister to Archaeopteryx instead of the 
Microraptor + Dromaeosaurus clade.

1. Compsognathidae was supported by only three characters in the TWG matrix, 
two of which are in my analysis (mesial serrations absent on teeth, poorly 
developed semilunate carpal; fan-shaped dorsal neural spines not included).  
Compsognathus was nested within the family based on an additional character 
(large olecranon process), also not in my analysis.  Needless to say, I 
don't find their evidence for placing Yixian taxa in Compsognathidae 
particularily strong.

2. Rauhut (2003) and Holtz (2004) both agree with my analysis on this point.

3. This varies within runs of the TWG matrix, and Holtz (2004) agrees with 
my analysis.

4. This was supported by three characters in the TWG matrix, all of which 
are in mine (coronoid prominance on surangular; toothless maxilla; 
crenulated dentigerous margin of premaxilla).  Similarly, Senter et al. 
(2004) support it with three characters, two of which are in mine (toothless 
dentary; manual phalanx II-1/II-2 ratio), and the other which is flawed 
(frontal vs. parietal length, since either bone can change length 
independantly).  I've suggested before the lack of a dorsal splenial 
process, extremely convex mandibular articulation, and other characters may 
place Incisivosaurus closer to caenagnathoids than Caudipteryx is.  
Especially since toothlessness is converged on so much in different 
maniraptoriformes.  I wouldn't commit to either alternative yet.

5. I know you agree with my analysis here, as do Maryanska et al. (2002) and 
Osmolska et al. (2004).

6. This is generally a trichotomy, though Maryanska et al. (2002) agrees 
with my position.  You don't think either the TWG nor I are correct, and 
place it outside Caenagnathoidea.

7. When Lu (2004) included more oviraptorosaur taxa in the TWG matrix, he 
got this same result in 93% of his mpt's.  Probably a result of neither the 
TWG matrix nor mine (yet) being made to determine caenagnathoid topology.

8. Though universally found when tested in published analyses, this was 
based on one character in the TWG matrix (tibiotarsus fused).  Notably, 
Shuvuuia must reverse its tibio-tarsal fusion (though it does have 
astragalo-calcaneal fusion).  Even matrices made to examine alvarezsaurid 
phylogeny don't support this very strongly, as Chiappe's analyses (1996, et 
al., 1998; 2002) always support it based on only two characters- better 
developed procoely in sacrum, posterior sacrum with better developed keels 
on ventral midline.  Though I have 'posterior sacral articulation convex' 
and 'posterior sacrum with ventral midline keel' in my analysis, neither is 
quantified yet to incorporate the less developed morphologies of 
Alvarezsaurus.  Novas (1996) added supracetabular crest and fourth 
trochanter present to support this node.  However, the fourth trochanter is 
absent in Parvicursor and only incipient in Shuvuuia.  Also, the 
supracetabular crest is described and illustrated as present in 
Alvarezsaurus by the TWG and Bonaparte (1991), so this is controversial and 
I've coded it as unknown.  Being a hindlimb character, fourth trochanter 
morphology has yet to be included in my analysis.  These latter two 
characters are somewhat dependant on topology though, as ornithomimosaurs 
have supracetabular crests (as do Microvenator and basal therizinosaurs) and 
fourth trochanters (as do Avimimus and basal therizinosaurs), so their 
absence could be apomorphies of Alvarezsaurus.  Plus, there are characters 
agreeing with my topology (e.g. coracoid bent at level of tubercle in 
Patagonykus, but not Alvarezsaurus or Mononykus), and Patagonykus lived 
earlier than Alvarezsaurus and is the largest alvarezsaurid, so some data 
make more sense that way.  While I still favor the TWG arrangement, I don't 
think mine is out of the question.

9. Although recent analyses tend to support deinonychosaurian troodontids, I 
could see any of the three possible topologies here being true.  Possible 
synapomorphies for dromaeosaurs + birds excluding archaeopterygids and 
troodontids include the medially displaced coracoid foramen (reversed in 
dromaeosaurids), enlarged anterior cervical epipophyses, and ossified 
uncinates (also in oviraptorosaurs).

10. This is pretty unique, but it's only a difference of three nodes to get 
from basal troodontid to basal avialan in my tree, or two nodes in Makovicky 
et al.'s tree (because it doesn't have Jinfengopteryx).  None of the 
characters supporting this are particularily convincing (e.g. squamosal 
recess; posteriorly placed quadrate foramen; lateral dentary foramina within 
groove; nasolacrimal duct passes lateral to lacrimal; basisphenoid recess 
absent; ophthalmic branch of trigeminal nerve completely separated from main 
branch before exiting braincase; metotic foramen reduced to a slit or 
closed), partially because the latter four also occur in birds (at least in 
ornithurines sensu Chiappe).

11. This is only supported by one character in the TWG matrix - the enlarged 
otosphenoidal crest.  Norell et al. (2000) added two more (basisphenoid 
recess absent; large axial epipophyses).  All are in my analysis.  Norell et 
al. code Byronosaurus incorrectly for both, hence their absence in diagnosis 
for the TWG analyses.  I don't see it being very hard to overturn one 
character.  One obvious character supporting my topology is the serrated 
teeth of IGM 100/44 (since Senter et al. confirmed Sinovenator lacks them).

12. This has only been tested by the TWG matrix besides mine.  They support 
Mei as a troodontid based on eight characters, seven of which I include (the 
other is pedal).  Normally my analysis agrees with theirs, but several of 
the characters are also found in basal birds and/or basal dromaeosaurs, so 
may actually diagnose Paraves and/or Deinonychosauria and be reversed in 
other clades.  For instance- birds besides Archaeopteryx have dorsoventrally 
flattened internarial bars; birds and Sinornithosaurus lack squamosal - 
quadratojugal contact; Sinornithosaurus and Buitreraptor have lateral 
dentary grooves; Graciliraptor has dorsally grooved distal caudals; 
Archaeopteryx, Rahonavis and Buitreraptor have short dorsal transverse 
processes.  The remaining characters are a robust fourth metatarsal and 
small tightly packed anterior dentary teeth.  While I'm not throwing Mei out 
of Troodontidae, it isn't as definitively placed there as one might think.

13. This is sometimes found in the TWG results (e.g. Hwang et al., 2002; 
Makovicky et al., 2003), based on the lack of basal tooth constriction in 
Sinornithosaurus and dromaeosaurids.  Additionally, Sinornithosaurus has 
mesial tooth serrations, as in dromaeosaurids.  The Buitreraptor analysis 
found different results because three pelvic characters were added, none of 
which are in my analysis.  Senter et al. (2004) supported Sinornithosaurus 
as a microraptorian based on twelve characters (only four of which are in my 
analysis so far), though several are seen in ornithurines (sensu Gauthier) 
too, so are synapomorphies of the dromaeosaur + bird clade in my latest 
topology.  I'm inclined to agree Sinornithosaurus is a microraptorian (as I 
haven't included most of its suggested synapomorphies), though I don't find 
the alternative unlikely.

14. This has been in every TWG result until Makovicky et al's latest 
modification.  Dromaeosaurid characters of Unenlagia include the anteriorly 
concave preacetabular process (also seen in Shenzhouraptor, which is a 
dromaeosaur in my tree), and dorsal neural spine tables (also seen in 
ornithurines, but not basal troodontids, Buitreraptor, Archaeopteryx or 
Rahonavis; so this fits my topology).  It's placed by Rahonavis and 
Buitreraptor in the TWG matrix due to four characters, only one of which is 
in mine (supracetabular crest).  Achillobator has three of these (dorsally 
concave postacetabular process; elongate obturator process; mesopuby), yet 
is a dromaeosaurid in Makovicky et al.'s tree and mine (it's miscoded for 
the first two in Makovicky et al.'s matrix).  Unenlagia is sister to 
Rahonavis in Makovicky et al.'s tree based on five characters, three of 
which are in mine.  Three of these (all dorsals pleurocoelous; six sacrals; 
lateral cuppedicus ridge extending posteriorly) are found in some 
dromaeosaurids too.  The other two (elongate preacetabular process; large 
proximodorsal ischial process) and the supracetabular crest support 
Makovicky et al.'s hypothesis, but the support for a dromaeosaurid Unenlagia 
isn't much worse.  I'm willing to go either way.  One might argue placing 
Achillobator in Unenlagiinae would decrease the reasons for placing 
Unenlagia in Dromaeosauridae, but then you'd have to explain the 
dromaeosaurid characters of Achillobator (large serrated teeth; short 
cervicals; highly elongate distal caudal prezygapophyses and chevrons; 
etc.).

15. You may have noticed that Archaeopteryx has six of the unenlagiine and 
Rahonavis + Unenlagia characters suggested by Makovicky et al. (all dorsals 
pleurocoelous; short dorsal transverse processes; lateral cuppedicus ridge 
extending posteriorly; supracetabular crest; mesopuby; elongate 
preacetabular process), and is polymorphic for three (dorsally concave 
postacetabular process; large proximodorsal ischial process; elongate 
obturator process).  The only one it lacks is six sacral vertebrae.  So it's 
not surprising Rahonavis and Buitreraptor come out next to Archaeopteryx in 
my tree.  Of the seven characters placing them by dromaeosaurs in Makovicky 
et al.'s tree, my analysis has five.  The sinusoidal supratemporal fossa 
edge is only known in Buitreraptor among unenlagiines, and is also present 
in the basal Ukhaa Tolgod troodontid and Troodon, though absent in Mei and 
Saurornithoides.  So it has a somewhat ambiguous distribution among 
deinonychosaurs, though Archaeopteryx does lack it.  I don't feel 
comfortable coding any non-dromaeosaur paravian for paraquadrate foramen 
size, including Mei and Archaeopteryx (the only two coded for it by TWG). In 
Mei, the dorsal quadratojugal process is missing, while Archaeopteryx never 
has the two bones visible posteriorly.  Buitreraptor does have unconstricted 
tooth roots, but Microraptor doesn't.  If Microraptor is outside 
Sinornithosaurus + dromaeosaurids as in my tree, this would be an ambiguous 
synapomorphy of Buitreraptor + Microraptoria + Dromaeosauridae at best.  
Distally placed cervical epipophyses are actually present in pygostylians, 
Archaeopteryx and most troodontids (Mei, Sinornithoides, Byronosaurus).  Of 
these, Makovicky et al. only coded Mei for the character.  So this doesn't 
support placing Buitreraptor with dromaeosaurids.  Stalked dorsal 
parapophyses are found in Buitreraptor and dromaeosaurs, but also in 
Confuciusornis and Mei (and further down, alvarezsaurids).  They are thus 
possibly diagnostic for Paraves or Eumaniraptora instead, and lost in 
Archaeopteryx and derived troodontids.  The posteriorly bifurcated chevrons 
that supposedly join Buitreraptor, Rahonavis and dromaeosaurids 
(Velociraptor, Graciliraptor, Microraptor) are also found in the troodontids 
Jinfengopteryx and Sinornithoides, as well as Shenzhouraptor.  They are 
absent in Achillobator, Saurornitholestes, Mei, Sinovenator, derived 
troodontids and Archaeopteryx.  So they could support a basal troodontid or 
basal ornithurine (sensu Gauthier) position as well.  Finally, the 
ginglymoid metatarsal II of Rahonavis, Neuquenraptor and Buitreraptor is 
seen not only in dromaeosaurids, but also in scansoriopterygids, the basal 
troodontid IGM 100/44 and Confuciusornis (the latter two miscoded by 
Makovicky et al.).  So it works in my topology where Rahonavis and 
Buitreraptor are in a clade with scansoriopterygids.

While writing this post, I've noticed several characters miscoded in 
Makovicky et al.'s (2004) matrix, most of which I've commented on above 
(under 12, 14 and 15).  After correcting those codings and rerunning it, the 
resulting tree is different from the published one.  The Paraves portion is-

|--+--Adasaurus
|  |--Achillobator
|  |--Utahraptor
|  |--Dromaeosaurus
|  |--Saurornitholestes
|  |--IGM 100/1015
|  `--+--Velociraptor
|     `--Deinonychus
`--+--Sinornithosaurus
  |--Microraptor
  `--+--+--Buitreraptor
     |  |--Unenlagia+Neuquenraptor
     |  `--+--Rahonavis
     |     `--+--Archaeopteryx
     |        `--Confuciusornis
     `--+--+--Mei
        |  `--Sinovenator
        `--+--Sinornithoides
           `--+--Byronosaurus
              `--+--Troodon
                 `--+--Saurornithoides mongoliensis
                    `--Saurornithoides junior

IGM 100/44 has an unstable position within the troodont+bird clade, but is 
never inside Archaeopteryx+Confuciusornis, Mei+Sinovenator or 
Sinornithoides+derived troodontids.

- You'll notice troodontids are now sister to birds instead of to 
dromaeosaurs.  The third possibility, not seen in the usual TWG matrix or my 
latest run.  I noted under #9 that I consider this plausible, and this is 
proof it's not to hard to recover.
- Byronosaurus and Sinornithoides switched places, which is unsurprising 
since this varies between TWG runs.
- Microraptorians aren't necessarily monophyletic.  I noted under #13 that 
this wasn't unlikely.
- Microraptorians are sister to troodontids + birds, which is quite an odd 
result.  This is based on- posteriorly placed maxillary fenestra; 
squamosal-quadratojugal contact absent; carotid processes; lobate 
preacetabular process; proximodorsal ischial process; obturator process 
placed at distal end of ischium; subarctometatarsus; posterior or lateral 
metatarsal IV flange; elongated obturator process.  The only relevent coding 
I changed was Sinornithosaurus' lack of squamosal-quadratojugal contact (Xu 
and Wu, 2001).
- There's a clade consisting of Buitreraptor, Unenlagia, Rahonavis, 
Archaeopteryx and Confuciusornis.  This is sort of a blend between Makovicky 
et al.'s results and mine.  It's very similar to mine, except Unenlagia gets 
moved from Dromaeosauridae (agreeing with Makovicky et al.) and 
Confuciusornis (and presumably other birds) gets moved from being sister to 
dromaeosaurs.  It recovers the Unenlagiinae of Makovicky, except it includes 
his two birds, Archaeopteryx (agreeing with my analysis) and Confuciusornis. 
 It's not surprising considering what I wrote above under #14 and #15.  
Note there is no Unenlagiinae exclusive of birds though, as Unenlagia and 
Buitreraptor are in a polytomy with a Rahonavis + birds clade.  The latter 
are united by - reduced pubic apron; metatarsal III not reduced proximally; 
posterior or lateral metatarsal IV flange absent; mid-dorsal ischial 
process; elongate ulna.  None of these are in my analysis yet.  In my 
analysis, Rahonavis is sister to Buitreraptor exclusive of Archaeopteryx 
based on enlarged dorsal hypapophyses, and three distal chevron characters.  
So I'd favor combining Rahonavis and Archaeopteryx to the exclusion of 
Buitreraptor, as in the modified Makovicky matrix.  Unenlagia shares an 
enlarged proximodorsal ischial process with Rahonavis + birds; Buitreraptor 
shares an elongate humerus with them; Unenlagia and Buitreraptor share a 
lobate postacetabular process to the exclusion of Rahonavis + birds.  So 
assuming Unenlagia really is part of this clade, how exactly it relates to 
Buitreraptor and Rahonavis + birds seems unresolved.
Are birds a member of this clade though, or are they somewhere else (like 
sister to dromaeosaurs, or sister to dromaeosaurs + troodontids)?  The TWG 
matrix can't test this very well, as it only includes Confuciusornis among 
ornithurines sensu Gauthier.  Of the 'unenlagiine' synapomorphies, 
pygostylians lack dorsal pleurocoels, short dorsal transverse processes 
(ornithuromorphs at least), a supracetabular crest, mesopuby, dorsally 
concave postacetabular processes, and elongate obturator processes.  They 
are polymorphic for elongate cuppedicus fossae, and they have elongate 
preacetabular processes, large proximodorsal ischial processes, and more 
than five sacrals.  They do have four of the five Rahonavis + Archaeopteryx 
found by the TWG though.  Although twelve characters are found to place 
Confuciusornis sister to Archaeopteryx in Makovicky et al.'s matrix, 
Archaeopteryx lacks two (reversed hallux; proximodorsal lips on manual 
unguals absent), all birds except archaeopterygids and confuciusornithids 
lack a fused scapulocoracoid, other 'unenlagiines' may have two (unfused 
parietals; splenial not exposed broadly laterally), and other basal 
paravians have three (asymmetrical remiges; reduced number of caudals; 
reduced number of caudals with transverse processes).  Four may be good 
(biceps scar on deltopectoral crest; distal tarsals fused to metatarsus; 
metatarsus fused; sickle claw absent), though I've examined none of them, 
and only the Wellnhoferia specimen has fused metatarsals.  Opposing these, 
there are some characters suggesting Rahonavis is closer to birds than 
Archaeopteryx is - trochanteric crest; fibula doesn't reach tarsus; m. 
iliofibularis tubercle faces posteriorly; proximal tarsals fused together.  
At the moment, I accept birds as part of the Archaeopteryx + Rahonavis 
clade, but the evidence for placing them closer to either taxon seems about 
even.  This assessment could use a lot more work though.

At the moment, I find the following most likely, but I'm just as iffy on 
basal paravian topology as ever.

|--+--Microraptoria
|  `--Dromaeosauridae
`--+--+--Mei
  |  `--other Troodontidae
  `--+--Buitreraptor
     |--Scansoriopterygidae
     |--Unenlagia+Neuquenraptor
     `--+--Rahonavis
        |--Archaeopterygidae
        `--Pygostylia

Additional studies would be needed to judge the position of Shenzhouraptor 
et al. and Jinfengopteryx.

Mickey Mortimer