Oliver Rauhut was recently kind enough to send me a copy of his thesis (written in 2000, not 1999 like I cited earlier). It contains lots of new information I believe would be of interest to the list. The topic of the thesis is phylogenetic relationships of basal theropods, although derived taxa are included. An analysis was performed that resulted in 5544 trees of 652 steps each that had a consensus of-
| |__Liliensternus liliensterni
| |__Liliensternus airelensis
| |_Abelisauridae (including Noasaurus)
|___________Magnosaurus (including Eustreptospondylus)
| | |____Chilantaisaurus
| | |___Spinosauridae (Baryonychidae is used)
| |__Carcharodontosauridae (including Acrocanthosaurus)
| |__Santana compsognathid
|_____________Segnosauria (Therizinosauroidea is used)
| |_____Caenagnathoidea (Oviraptorosauria is used)
| |___Dromaeosauridae (including Sinornithosaurus)
|_____Aves (including Rahonavis)
As you can see, Ceratosauria as usually constructed is paraphyletic, which was a main point of the paper. Another major difference from other recent analyses is the inclusion of megalosaurs and spinosaurs in the Carnosauria. I will discuss some new information or different viewpoints below. One great thing about this paper is that Rauhut examined most specimens personally to determine the presence or absence of characters. Unfortunately, I feel Rauhut often makes taxa nomina dubia without sufficient reason and makes many species metataxa, which I feel is inappropriate. I'll defend a few of these cases below, but many others I disagree with require further study. However, even taking these issues into account, the overall quality of the paper is excellent.
Eoraptor has palatal teeth, but the claim of such structures in Massospondylus (Attridge et al., 1985) is incorrect, being the result of fish teeth washing onto the palate before burial (Sereno pers. comm. to Rauhut 1997). Although Eoraptor and herrerasaurids are placed as theropods in the consensus tree, most of the synapomorphies relate to predatory habits and there are many reversals needed, which makes this conclusion tenuous.
The brevis shelf of Caseosaurus is medially expanded as in some crurotarsans, not laterally expanded like Herrerasaurus and most dinosaurs. In addition, the acetabulum is incomplete, so its extent cannot be judged. Thus, Rauhut claims it cannot be shown to be dinosaurian.
The dorsals of Chindesaurus are not tall and craniocaudally short like herrerasaurids. The ilium originally referred to this species is now the holotype of Caseosaurus, so ilial herrerasaurid synapomorphies cannot be determined. Thus, no herrerasaurid synapomorphies are present in the holotype. Also, there is uncertain mixing of the holotype material with a partial Chatterjeea skeleton in the same pocket. Because of these facts, Rauhut refers Chindesaurus to Archosauria indet., but I believe reexamination of the material will show more precise relations.
Rauhut says that the elongate mandibular symphysis of Alwalkeria is unlike theropods, but similar to some crurotarsans.
The paraphyletic traditional Ceratosauria is also seen in Carrano and Sampson 1999. The characters traditionally used to support the Ceratosauria are rather problematic-
- strongly backturned and triangular dorsal transverse processes. Triangular diapophyses are found in many tetanurines and the degree of backturning varies within the dorsal column.
- two pairs of cervical pleurocoels. Because there are no known non-neotheropods with pleurocoels, whether having one or two pairs of pleurocoels is the primitive state cannot be determined. This study suggests having two pairs is primitive. This is also suggested by avian embryological data (Britt, 1993). Liliensternus liliensterni only has a single pair.
- pubic fenestra. Not present in Liliensternus, Ceratosaurus, Elaphrosaurus or abelisaurids. Possibly synapomorphic for a coelophysoid subgroup.
- synsacrum. Varies ontogenetically.
- pelvis fused. Varies ontogenetically and is only partially developed in the old adult Elaphrosaurus holotype.
- trochanteric shelf. Present only in robust individuals and known in Herrerasaurus and basal dinosauriformes.
- sulcus at base of crista tibiofibularis. The characteristic oblique sulcus is only present in Syntarsus. Absent in Segisaurus, Coelophysis, Liliensternus. A longitudinal sulcus like that found in tetanurines is known in Dilophosaurus, Ceratosaurus, Elaphrosaurus and abelisaurids.
- astragalocalcaneum fused to tibia. Ontogenetically variable and absent in the old adult Elaphrosaurus holotype.
- ascending process of astragalus overlapped by fibula. Present in Syntarsus, but not Liliensternus, Dilophosaurus, Ceratosaurus or abelisaurids
- distal tarsals fused to metatarsus. Ontogenetically variable and absent in the old adult Elaphrosaurus holotype.
The neoceratosaur-tetanurine clade is supported by fifteen synapomorphies- presence of a promaxillary foramen in the maxillary; ascending process of the maxillary offset from the anterior margin of the maxillary body; prefrontal displaced medio-posteriorly; ventral margin of the bases of the paroccipital processes situated at, or below the mid-height of the occipital condyle; tooth row ends below the anterior margin of the orbit; retroarticular process of the mandible broad and attachment for the M. depressor mandibulae developed as a posteriorly placed groove; pleurocoels present in axis; cervical pleurocoels developed as foramina that pierce the vertebral body; cervical epipophyses strongly overhang the postzygapophyses; iliac blade expanded ventrally anteriorly; pubic apron has a medial opening distally; lesser trochanter broadened and wing-like; distal end of the femur well rounded; distal articular surface of the tibia broadly triangular in outline; tibial facet present on the calcaneum. It has a bootstrap of 90% and it takes 13 more steps to make the traditional Ceratosauria monophyletic.
Coelophysoidea as traditionally used (Coelophysidae + Dilophosaurus) is also shown to be paraphyletic. Four synapomorphies have been suggested for this clade-
- subnarial gap. True, but the structure is different, that of Dilophosaurus being closer to that of spinosaurids.
- no axial diapophyses. Also in Herrerasaurus and prosauropods, probably symplesiomorphic.
- reduced axial parapophyses. Also in Herrerasaurus and prosauropods, probably symplesiomorphic.
- no axial pleurocoels. This is obviously plesiomorphic, seen in Herrerasaurus, Eoraptor, sauropodomorphs, etc..
Seven additional synapomorphies were found (premaxillary body in front of external nares longer than body below the nares, and angle between anterior margin of the premaxilla and alveolar margin less than 70 degrees; presence of a constriction between the articulated premaxillaries and maxillaries; the lateral rims of the nasals are pronounced and form raised edges; no contact between the squamosal and quadratojugal; presence of enlarged, fang-like teeth in the anterior part of the dentaries; posterior dorsal vertebrae considerably elongated; articular facet of the pubic peduncle of the ilium subdivided into an anterior part that faces anteriorly, and a posterior part that faces ventrally), but were overridden by nine characters grouping Dilophosaurus with ceratosaurs sensu stricto and tetanurines- presence of a promaxillary foramen in the maxillary; prefrontal displaced posteriorly; tooth row ends at the anterior margin of the orbit; cervical pleurocoels developed as foramina that lead into the interior of the vertebral body; axial and cervical epipophyses strongly pronounced and overhanging the postzygapophyses; anterior articular facet of anterior cervicals wider than high; presence of a medial opening in the pubic apron distally; presence of a deep groove in the medial side of the proximal end of the fibula; Mt V flattened transversely and flexed anteriorly in its distal part. Only three steps are needed to make a traditional Coelophysoidea, but if Shuvosaurus is excluded it is just as parsimonious as the tree presented above. This is because the latter taxon lacks a few of the proposed traditional coelophysoid characters- it lacks a subnarial gap and a constriction between the premaxillaries and maxillaries, the lateral margins of the nasals are not raised, and the squamosal contacts the quadratojugal.
Procompsognathus grouped with the coelophysoids in preliminary analyses, but was removed as it contained no unique combination of characters.
Shuvosaurus is a coelophysid based on two characters shared with Syntarsus- forked posterior end of the premaxilla, considerably elongated basisphenoid.
Elaphrosaurus is diagnosed as having a very wide scapular blade (broader than height of vertebral column). The extremely wide scapular blade is yet another character supporting a Chuandongocoelurus + Elaphrosaurus clade, although Chuandongocoelurus' is only 75% as high as the vertebral column.
Noasaurus is included in the Abelisauridae because it shares several characters with the family (short high antorbital part of maxilla, cervical epipophyses hypertrophied and higher than neural spines. Also, no cladistic analysis has been done on abelisauroids, so it is possible Noasaurus is more closely related to some abelisaurids than others. I think the reduced second metatarsal is similar to Masiakasaurus and Velocisaurus, and the sickle claw is similar to Ligabueino. Pedes are poorly known in abelisaurids, so these characters may not mean much. However, Ilokelesia is thought by Headden to be a carnotaurine and lacks a sickle claw. Also, Noasaurus lacks the fused quadrate and quadratojugal seen in Abelisaurus and Carnotaurus, though apparently not Majungatholus. So I think that it is premature to place Noasaurus in the Abelisauridae, but agree that more work needs to be done in this area.
Ligabueino and Velocisaurus scored identically to Elaphrosaurus, so were not included. This is more evidence for a clade containing these genera, probably also including Noasaurus, Masiakasaurus and Laevisuchus.
He suggests Genusaurus is closer to Carnotaurus than Majungatholus, based on the straight dorsal ilial margin and very long vertically oriented ischial peduncle. Also, Rauhut thinks Indosaurus is more similar to allosauroids and that the broad horizontal plate separating the supratemporal fenestrae makes this taxon different from abelisaurids. Novas and Bandyopadhyay (1999) disagree. Interestingly, Indosaurus, Indosuchus and Compsosuchus were first described by Huene and Matley in 1932 (pg. 237), not 1933. So their entries should be changed in George's Dinosaur Genera List and other places.
The Carnosauria clade, containing both traditional carnosaurs, as well as megalosaurs and spinosaurs, is supported by only three synapomorphies- ascending process of the maxillary offset from the anterior rim of the maxillary body and anterior projection of the maxillary body longer than high; cervical vertebrae strongly opisthocoelous; metacarpal I very stout and approximately as broad as long. Arranging the taxa into spinosauroids and allosauroids as in Sereno (1999) requires three more steps. Forcing the topology of Holtz (2000) (with spinosaurids, Eustreptospondylus, Torvosaurus, Piatnitzkysaurus and Afrovenator successively closer to avetheropods, and Monolophosaurus and allosauroids in a monophyletic Carnosauria) requires eleven more steps.
Rauhut uses Baryonychidae in place of Spinosauridae because of uncertainty regarding the association of Spinosaurus' holotype. The cervical vertebra has a low neural spine, which Rauhut considers too different from the high-spined dorsals to belong to the same taxon. Furthermore, he claims the dorsal vertebrae lack the strong pneumatization and laminae of Baryonyx and Suchomimus and are comparable to allosauroids except for the tall spines. As tall neural spines are known in Acrocanthosaurus, he suggests the dorsals may belong to a carcharodontosaurid. The proximal caudal described by Stromer as possibly being too large for the holotype is said to resemble ornithischian caudals more. Quite an interesting problem, as the name Spinosaurus would no doubt stay with whatever taxon the dorsals belong to. I've lost my figures to Stromer 1915 and haven't examined spinosaurids closely in any case, but wouldn't it be odd if Spinosaurus turned out to be an allosauroid? Jurassic Park never had a chance :-) . Additionally, Spinosaurus marocannus was originally distinguished from S. aegyptiacus by its elongate cervical vertebra. However, it is longer because it is a more anterior vertebra than the type of S. aegyptiacus, making the validity of S. marocannus dubious.
Eustreptospondylus is made a junior synonym of Magnosaurus, to form the new combination Magnosaurus oxoniensis. This is because they share the following characters- slight dorsoventral and transverse expansion of anterior dentary; significantly enlarged third dentary tooth; shallow longitudinal groove with rectangular cross section in dentary. Also, the remains of both are almost indistinguishable, differing only in slight differences in the proximal extent of the pubic apron. I think this sounds probable, but cannot comment due to the lack of proper references on either.
Rauhut refers to Becklespinax as Altispinax altispinax, as Huene proposed the name Altispinax for the vertebrae, not the indeterminate tooth of Megalosaurus dunkeri. Thus, Becklespinax is a junior synonym, though altispinax is the first species name available. Everyone change your lists. :-)
"Szechuanosaurus" zigongensis is shown to be more basal than carnosaurs due to the platycoelous cervicals, non-pleurocoelous dorsals and other characters. It is very similar to Xuanhanosaurus in all areas except humeral morphology.
Xuanhanosaurus' description (Dong, 1984) and reexamination (Molnar, 1990) are incorrect in a couple points. The sternum is actually an impression of part of the right coracoid. The carpal and phalangeal joints do not differ from the typical theropod pattern, with highly flexible digits and a somewhat opposable pollex. The arm was actually rather short in comparison to the body, so there is no evidence for quadrupedal motion. Xuanhanosaurus was not included in the final analysis to limit the amount of MPT's, but grouped as either a basal tetanurine or a basal carnosaur.
Although not included in the analysis, the distal tibia outline and proximal extent of the anterior trochantor show Erectopus is a basal tetanurine, not an abelisaurid as has been recently proposed.
Rauhut notes that Yangchuanosaurus shangyouensis, Y. magnus, Sinraptor dongi and S. hepingensis are all nearly identical in morphology. Additionally, all but S. dongi are from the same formation. He suggests they may be synonymous, which would make Yangchuanosaurus shangyouensis the correct name. However, he has not examined them firsthand, so provisionally keeps them separate.
Metriacanthosaurus was not included because it scored identically to sinraptorids.
Rauhut argues against the assignment of Siamotyrannus to the Tyrannosauroidea. He states that the structure in the figure is not a medioventral shelf on the ilium, that other theropods have vertical lateral ilial ridges and proximolateral ischial scars, and that the pubic foot is broken posteriorly, so its length cannot be determined. However, he fails to mention the narrow second and third sacral centra, well-marked insertion of the transverse process of sacral 1 on the ilium and other minor characters cited by Buffetaut et al.. Still, I am inclined to believe Rauhut and Pharris that Siamotyrannus is more likely to be a carnosaur. It grouped with allosauroids in the preliminary analysis, before it was excluded to limit the number of MPT's.
The presence of a (Coelurus + Compsognathidae) clade is unique and supported by only two synapomorphies- cervical epipophyses not overhanging the postzygapophyses; posterior dorsal neural spines lower than long. The same can be said of the (Deltadromeus + Ornithomimosauria) clade- straight humerus in lateral view; presence of a deep and wide groove on the medial side of the proximal end of the fibula. Only one step is necessary to make Deltadromeus a basal coelurosaur. Enforcing the coelurosaurian phylogeny of Holtz (2000) adds 31 steps. Sereno's (1999) phylogeny takes 26 more steps. My most recent phylogeny takes 20 more steps.
Rauhut notes Deltadromeus has a well developed anteromedial ridge on the distal femur. This is more similar to ceratosaurs than coelurosaurs. Additionally, the anteroposteriorly long proximal caudal neural spines, reduced fourth metatarsal and proximally unreduced third metatarsal have me thinking of Elaphrosaurus after all the time I spent looking at the latter genus for "Details on Chuandongocoelurus".
Rauhut assigned Bahariasaurus to the Carcharodontosauridae in 1995 based on characters of referred material. The holotype is said to lack carcharodontosaurid synapomorphies and be very close (if not identical) to Deltadromeus.
Rauhut notes Albertosaurus megagracilis is a nomen nudum, as Paul did not illustrate it, cites the wrong reference and gives no formal diagnosis. Of course, it's just a subadult Tyrannosaurus rex, so it doesn't matter that much.
"Chilantaisaurus" maortuensis came out as a basal coelurosaur before it was removed to limit the number of MPT's.
The absence of transverse processes in Compsognathus is no longer certain. There is a large crack running through the caudal series which makes it impossible to determine their presence. Rauhut also reinterprets the manus so that mcI of Ostrom equals mcII, mcII = mcIII, mcIII = mcIV, I-1 = II-1, II-1 = I-1, ungual I = ungual III. This differs from the suggested reconstruction by Gauthier and Gishlick (2000), where mcI = I-1, mcII = mcII, mcIII = mcIII and a "mystery element" of Ostrom is really mcI. In any case, the pubic foot is actually much longer than reconstructed by Ostrom, being more similar to the C. "corallestris" specimen.
Rauhut still coded AMNH 587 (the referred manus) as Ornitholestes, while it is now known to be Coelurus, which may have had an adverse affect on the analysis. Ornitholestes does not have a nasal horn (contra Paul, 1988), as "the apparent upward flexure of the posterior border of the external nares on the left side of the skull is caused by a break and subsequent ventral displacement of the ascending process of the maxilla and the nasals."
Rauhut notes that there are no comparable elements in Piveteausaurus and Proceratosaurus, so their referral to the same family and genus by Paul (1988) is unfounded. Piveteausaurus differs quite a bit from Ornitholestes, to which it can be compared.
Timimus is said to be indistinguishable from other coelurosaurs, but see http://www.cmnh.org/fun/dinosaur-archive/2001Mar/msg00681.html for why this is untrue. Currie (pers. comm. to Rauhut 1998) thinks Dromiceiomimus may be synonymous with Ornithomimus.
Rauhut uses "Nanshiungosaurus" bohlini, as I suggested in my "Details on Nanshiungosaurus bohlini" post. He bases this on temporal and unspecified morphological differences.
He says that the presence of a posterior ulnar ridge on Avimimus is uncertain, as the proximal end doesn't fit into the proper condyle of the humerus. Nevertheless, Watabe et al. (2000) report new remains show the ulna does indeed have the ridge.
There has been a rumor on the list recently that Koparion is not troodontid. This seems to stem from a statement made by Harris in 1997 that Chure (pers. comm.) no longer believes it belongs in this family. Rauhut says that because constricted roots are known in non-troodontid theropods and Koparion is similar to Compsognathus, it may be a compsognathid. However, he fails to note that other troodontid characters are present in Koparion, including an apex formed by the most distal posterior serration, blood pits between serrations, enlarged serrations and posterior serrations curved apically. At least the last two characters are not found in Compsognathus. Thus, I still believe Koparion is a troodontid.
The distal ends of Sinornithoides' pubes are broken off, so it is impossible to determine if there was a pubic foot. A pubis referred to Troodon (MOR 553S 220.127.116.117) has a pubic foot, but was not associated with Troodon material and has a larger anterior foot and a cranially concave shaft. Sounds like Chirostenotes to me.
Rauhut argues Archaeornithoides is Archosauria incertae sedis, as unserrated teeth with basal constrictions and no interdental plates are known for both crocodilians and avians. However, he does not take the subsidary palatal fenestra into account, which is a coelurosaurian synapomorphy. I recommend it remain in the Maniraptoriformes and possibly in the Troodontidae (as suggested by Currie, 2000).
Rauhut uses a very loose definition of Dromaeosauridae, including the avian Hulsanpes, as well as the more basal deinonychosaur Sinornithosaurus. This may have affected the results. In addition to the definite dromaeosaurids, Adasaurus, Megaraptor and Utahraptor were also included. The cervicodorsal of Variraptor is referred to the Oviraptorosauria because it has two pairs of pleurocoels. As Achillobator also has this feature though, I feel the referral is inappropriate. He refers the taxon to Coelurosauria indet.. The quadrangular posterior articular surface of the last sacral centrum is only known in troodontids, dromaeosaurids, Rahonavis and Archaeopteryx. The pneumatic posterior dorsals and anterior sacrals, and five sacral vertebrae are unlike troodontids. Thus, Variraptor is a eumaniraptoran. Further analysis is difficult, pending a stable eumaniraptoran phylogeny and description of more eumaniraptoran sacra.
If anyone has any questions, feel free to ask them. Now to perform Ken's experiment with segnosaurs in this analysis.....