Gauthier, 1986. Saurischian monophyly and the origin of birds. Memoirs of the Californian Academy of Sciences 8, 1-55.

This paper was the first numerical cladistic analysis of Mesozoic theropod, and as such was highly influential. For instance, it formed the basis for the phylogeny used in The Dinosauria (Weishampel et al., 1990). Important clades originating here include Tetanurae, a Coelurosauria including taxa closer to birds than to carnosaurs, and Maniraptora.

Phylogeny-

Gauthier's illustrated phylogeny is as follows. Note it excludes Hulsanpes, which can be placed anywhere in Maniraptora according to his codings.

|--outgroup
`--+--Ornithischia 
   `--Saurischia
      |--Sauropodomorpha
      `--Theropoda
         |--Ceratosauria
         |--Procompsognathus
         |--Liliensternus
         `--Tetanurae
            |--Carnosauria
            `--Coelurosauria
               |--Ornithomimidae
               `--Maniraptora
                  |--Compsognathus
                  |--Elmisauridae
                  |--Caenagnathidae
                  |--Microvenator
                  |--Coelurus
                  |--Saurornitholestes
                  |--Ornitholestes
                  `--+--Deinonychosauria
                     `--Avialae

However, the matrix does not generate this cladogram in PAUP. In actuality, Saurornitholestes can fall anywhere in Coelurosauria, including outside Maniraptora and inside Eumaniraptora (the Deinonychosauria + Avialae clade). Thus the actual resulting cladogram is-

|--outgroup
`--+--Ornithischia
   `--Saurischia
      |--Sauropodomorpha
      `--Theropoda
         |--Ceratosauria
         |--Procompsognathus
         |--Liliensternus
         `--Tetanurae
            |--Carnosauria
            `--Coelurosauria
               |*-Saurornitholestes
               |--Ornithomimidae
               `--Maniraptora
                  |*-Hulsanpes
                  |--Compsognathus
                  |--Elmisauridae
                  |--Caenagnathidae
                  |--Microvenator
                  |--Coelurus
                  |--Ornitholestes
                  `--+--Deinonychosauria
                     `--Avialae

Taxon Issues

Outgroup- Gauthier's outgroup includes Pseudosuchia (=Crurotarsi), Ornithosuchidae (now viewed as a crurotarsan clade), Euparkeria (viewed as an avemetatarsalian by Gauthier, but now placed outside Archosauria), Lagosuchus (based largely on Marasuchus), Pterosauria and Herrerasauridae (now viewed as a saurischian and possible theropod clade).
Elmisauridae- Gauthier includes Chirostenotes, Macrophalangia (a junior synonym of Chirostenotes) and Elmisaurus in this family, but excludes Caenagnathus which is now known to be the mandible of Chirostenotes. This clade is now called Caenagnathidae.
Caenagnathidae- Gauthier includes Caenagnathus (the mandible of Chirostenotes) and Oviraptor (including the yet unnamed Conchoraptor and Citipati specimen IGM 100/42) in this family, but excludes "Ingenia" which he places in Ornithomimidae. "Ingenia" is here included, Caenagnathus excluded and the OTU referred to as Oviraptoridae.
Ceratosauria- This includes Ceratosaurus and several coelophysoids- Megapnosaurus, Coelophysis, Segisaurus, Sarcosaurus, Dilophosaurus and the not yet named "Megapnosaurus" kayentakatae, but excludes abelisaurs. Of the latter, Indosaurus and Indosuchus were included in Carnosauria while Elaphrosaurus was included in Ornithomimidae. Liliensternus and Procompsognathus were coded separately.
Carnosauria- This included not only Allosaurus and Acrocanthosaurus but also tyrannosauroids (Alectrosaurus, Dryptosaurus, Albertosaurus, Gorgosaurus, Nanotyrannus [= Tyrannosaurus?], Alioramus, Tarbosaurus, Tyrannosaurus) and two taxa now recognized as abelisaurids (Indosaurus, Indosuchus). The abelisaurids are here ignored, as they are not codable for virtually any characters anyway.
Ornithomimidae- This included all ornithomimosaurs (Archaeornithomimus, Dromiceiomimus, Ornithomimus, O? sedens, Struthiomimus, Gallimimus, Garudimimus, Deinocheirus) as well as the unrelated Elaphrosaurus and "Ingenia" (neither of which was included in my codings of the OTU).
Deinonychosauria- This included its current content of Troodontidae (Saurornithoides, Zanabazar, Troodon) and Dromaeosauridae (Dromaeosaurus, Deinonychus, Velociraptor, Adasaurus), though Saurornitholestes was coded separately.
Avialae- This clade was equivalent to my usage, but only Archaeopteryx, Hesperornithes, Ichthyornis and Aves were recognized at the time.

Coding issues

1. Ornitholestes and oviraptorids (IGM 100/42) have premaxillary subnarial processes with elongate nasal contact. Deinonychosaurs are polymorphic, with the derived state found in Velociraptor, but not Saurornithoides, Zanabazar or Troodon.

2. Gauthier's outgroups (Herrerasaurus, Lesothosaurus, Emausaurus, etc.) all have supratemporal fossae on their frontals, as do other dinosaurian outgroups (Silesaurus, Scleromochlus). Procompsognathus' skull is probably sphenosuchian, so cannot be coded for cranial characters. Ornitholestes and Archaeopteryx both have such fossae as well, unlike Gauthier's codings of unknown and absent.

3. This character uses both neck length vs. presacral length (>36%), but also cervical 6 compared to dorsal 2 length (>20% longer). These are correlated, but may also not always be true in any taxon. Procompsognathus, Liliensternus, Coelurus, Microvenator, Chirostenotes, oviraptorids and deinonychosaurs have elongate necks as well, so should be coded as derived instead of unknown. This is also true of Herrerasaurus, one of Gauthier's outgroups.

4. Gauthier's outgroup Herrerasaurus has laterally placed axial postzygapophyses, while the axis of Coelurus is unpreserved and that of Compsognathus is too poorly preserved to code. Oviraptorids such as IGM 100/42 have the derived state however.

5. Again Herrerasaurus has anterior cervical epipophyses as do basal ornithischians (Heterodontosaurus, Lesothosaurus, Scelidosaurus). Liliensternus also has epipophyses, while the condition in Procompsognathus is unpreserved. Among coelurosaurs, Ornitholestes, oviraptorids (e.g. IGM 100/42) and Saurornitholestes have epipophyses.

6. Yet again the outgroup Herrerasaurus has dorsal hyposphene-hypantrum articulations. Among coelurosaurs, Coelurus, Microvenator, Chirostenotes and Saurornitholestes have them.

7. Herrerasaurus has elongate hands (>45% of humeral+radial length) based on new finds, as do pterosaurs, so the outgroup is coded as apomorphic. With the recent placement of heterodontosaurids as basal ornithischians and Eocursor, it's known ornithischians had elongate hands basally. Based on the length of incomplete manus, Liliensternus, Procompsognathus, Ornitholestes, Coelurus, Compsognathus, Microvenator and Saurornitholestes also have elongate manus.

8. The manus Gauthier used to code Ornitholestes' digit length (AMNH 587) is now referred to Tanycolagreus, leaving Ornitholestes unknown for this character. The basal ornithomimosaur Deinocheirus has a longer third digit than second digit, making that OTU polymorphic. Published remains of Saurornitholestes are still too fragmentary to determine its condition.

9. This character is a composite, as metacarpals IV and V will not necessarily both lay on the ventral sides of III and IV respectively in a taxon. Herrerasaurus has the derived condition, as does Procompsognathus for IV at least, though again the misattributed manus of Ornitholestes means it should be coded as unknown.

10. This character is a composite of several features (manual digit I more robust than digit II; manual ungual I larger than II; metacarpal I 50% or less of metacarpal II length; distal condyles of metacarpal I markedly asymmetrical; phalanx I-1 longer than metacarpal I and is the longest phalanx). Herrerasaurus has features 4, 5 and 6 but lacks the others. Heterodontosaurus has characters 2 and 4. A basal sauropodomorph such as Efraasia has 1, 2, 4, 5 and 6. Dilophosaurus has 2, sometimes 3, 4 and 5. Ceratosaurus certainly lacks 1 and 3, but does have 4. Procompsognathus has 1, 3 and 5 but lacks 6. Allosaurus has 1, 2, 4, 5 and 6. Thus all of these taxa should be polymorphic. Ornitholestes has at least 1, 2, 4 and 5, so is either apomorphic or polymorphic. Compsognathus has 1, 2, 3 and 5, so is coded polymorphic instead of unknown. Coelurus lacks at least 1, but is kept scored unknown. Basal ornithomimosaurs like Deinocheirus have 1, 2, 4, 5 and 6 so are coded polymorphic. Microvenator has at least 4 and 5, but is kept unknown. Elmisaurus and Chirostenotes have 4 and 5. Basal oviraptorids like IGM 100/42 have 1, 2, 3, 4, 5 and 6, so are correctly coded as apomorphic. Deinonychosaurs such as Deinonychus and Velociraptor have 1, 2, 3, 4 and 5, so are coded as polymorphic. Basal avialans like Archaeopteryx have 3, 4, 5 and 6, so are also coded as polymorphic.

11. This character also consists of two parts- reduced mandibular symphysis; reduced overlap of dentary with postdentary bones. The outgroup Herrerasaurus has these features based on recently described material. Liliensternus, Compsognathus, ornithomimids, Microvenator, Chirostenotes and oviraptorids lack mandibular kinesis. Ornitholestes appears to have an intermandibular joint however. Gauthier's ceratosaurs are polymorphic, as Megapnosaurus and Dilophosaurus have a tight fit between the splenial and prearticular suggesting no kinesis, but Ceratosaurus differs. Avialae is also polymorphic, with Archaeopteryx probably lacking a joint while hesperornithids and Ichthyornis have one.

12. The outgroup Herrerasaurus has a dorsally exposed lacrimal, as do basal sauropodomorphs like Efraasia. The skull of Procompsognathus is probably incorrectly referred. The condition in Saurornitholestes can be inferred as derived from its frontal morphology.

13. What Gauthier calls a maxillary fenestra is now termed a promaxillary fenestra. It is now known for both Herrerasaurus and Heterodontosaurus. Tykoski codes it as lacking in Liliensternus, while ceratosaurs are not coded as polymorphic since its absence in the derived Coelophysis and Megapnosaurus is most parsimoniously a reversal. Caenagnathus is now known to lack maxillary fenestrae, while Saurornitholestes has one. Within deinonychosaurs, troodontids known to Gauthier (Saurornitholestes, Zanabazar, Troodon) lack promaxillary fenestrae, while dromaeosaurids (Deinonychus, Velociraptor) have one. Thus deinonychosaurs are polymorphic.

14. Fused vomera are plesiomorphic for ornithischians (Heterodontosaurus, Lesothosaurus). The skull of Procompsognathus is probably incorrectly referred. The vomera are also fused in Ornitholestes, while deinonychosaurs and avialans are polymorphic (fused in Deinonychus and Archaeopteryx, unfused in Velociraptor and Hesperornis).

15. Herrerasaurus, Liliensternus, Ornitholestes and Compsognathus have an expanded ectopterygoid with a ventral fossa. The basal sauropodomorph Pantydraco does as well, making sauropodomorphs polymorphic. Oviraptorids lack the fossa and expansion however.

16. This is a composite character, as odontoid notch width and occipital fossa size are not necessarily correlated. Dilophosaurus and "Megapnosaurus" kayentakatae have a broad odontoid notch, unlike Ceratosaurus, making Ceratosauria polyphyletic.

17. Herrerasaurus has the broadly concave axial intercentrum as in theropods.

18. Procompsognathus, Liliensternus, Ornitholestes, Chirostenotes and oviraptorids have presacral pleurocoels. The sauropodomorph Pantydraco and pterosaurs do as well, making both Sauropodomorpha and the outgroup polymorphic.

19. Scleromochlus, pterosaurs, ornithischians, Liliensternus, Ornitholestes, Chirostenotes, oviraptorids and Saurornitholestes have at least four sacrals.

20. This is another composite character, incorporating distally reduced caudal transverse processes, elongate prezygapophyses and dorsoventrally flattened chevrons. Herrerasaurids have elongate prezygapophyses as well (as noted by Gauthier for Staurikosaurus). Ceratosaurus lacks them though, making Gauthier's Ceratosauria polymorphic. Coelurus and Saurornitholestes have a transition point, while Microvenator, Chirostenotes and oviraptorids lack one (the latter are stated to have one by Gauthier but coded as unknown).

21. Ceratosauria should be polymorphic as Ceratosaurus lacks ossified distal carpals while coelophysoids are correctly coded as having an enlarged carpal block. Compsognathus and Saurornitholestes have an enlarged carpal block, while ornithomimosaurs lack one (which they are stated to by Gauthier, but coded as unknown).

22. Pterosaurs and Herrerasaurus have reduced manual digit V, as does Procompsognathus. Ornitholestes' more complete manus belongs to Tanycolagreus and the holotype is too incomplete to judge this character. Saurornitholestes' described manus is also too incomplete.

23. Herrerasaurus has a reduced manual digit IV, while Ornitholestes and Saurornitholestes are unknown as for the previous character.

24. While this character is stated as "manus with elongate penultimate phalanges", it is best understood as only pertaining to digit II, since digit I was covered in character 10 and digit III is covered in the next character. Pterosaurs, Herrerasaurus, Heterodontosaurus, Eocursor and Efraasia have elongate phalanx II-2, making the outgroup, Ornithischia and Sauropodomorpha polymorphic. Meanwhile, Dilophosaurus has a short phalanx making Ceratosauria polymorphic. Ornitholestes and Saurornitholestes are unknown again. Coelurus has the derived state and Chirostenotes is polymorphic. Avialae should also be polymorphic, as Ichthyornis and numerous other more basal taxa not known by Gauthier have short distal phalanges.

25. Pterosaurs, Heterodontosaurus and Eocursor have manual phalanx III-3 longer than III-1 or III-2, making the outgroup and Ornithischia polymorphic. Ornitholestes is unknown again.

26. This character is another composite- manual unguals enlarged, compressed, pointed, recurved and with large flexor tubercles. Moreover, an enlarged manual ungual I was covered by character 10. Novas (1993) revised this to only apply to digits II and III since sauropodomorphs and Heterodontosaurus have the characters on digit I. Taken this way, Herrerasaurus and pterosaurs still exhibit the character. Heterodontosaurus' and Efraasia's manual ungual IIs would seem to qualify as well, except the latter has a low flexor tubercle. Tyrannosaurids and Compsognathus have rather uncompressed unguals, ornithurine birds have smaller unguals than other theropods, non-Deinocheirus ornithomimosaurs (mentioned by Gauthier as lacking the characters, but coded unknown), Ornitholestes, Compsognathus and many avialans have less curved unguals, and most ornithomimosaurs have low flexor tubercles. Considering this, Ornithomimosauria, Compsognathus and Avialae should be coded as polymorphic. Coelurus does not preserve manual unguals.

27. Scleromochlus and pterosaurs have a long preacetabular process as do ornithischians (noted by Gauthier but coded as absent). Among theropods, it is also present in Microvenator, Chirostenotes and Saurornitholestes.

28. A deeply concave brevis fossa is absent in Microvenator, oviraptorids and most deinonychosaurs (e.g. Adasaurus, Deinonychus, Velociraptor) and is uncertain in Chirostenotes.

29. Gauthier noted pterosaurs have non-sigmoidal femora. Ceratosaurs and carnosaurs sensu Gauthier seem to be polymorphic as some (Dilophosaurus, Allosaurus) have sigmoidal femora while others (Ceratosaurus, Acrocanthosaurus, Tyrannosaurus) don't. Ornitholestes also has a sigmoidal femur.

30. This is a composite character, including both the how closely appressed the tibia and fibula are, and the presence of a fibular crest on the tibia. Gauthier notes pterosaurs (and Scleromochlus) have appressed lower limb elements, as do ornithischians. As both of these groups lack a fibular crest, they are coded as polymorphic. Sauropodomorphs are polymorphic for the presence of a fibular crest, as Saturnalia has one. The tibia of Ornitholestes is unknown, while Coelurus and Chirostenotes have the crest. Sereno and Wild indicate Procompsognathus cannot be coded for this character.

31. Elongate and narrow metatarsi are present in Scleromochlus, pterosaurs, Lagosuchus and most other non-dinosaurian avemetatarsalians. This is also true of basal ornithischians, basal sauropodomorphs and Coelurus. Ceratosaurs and carnosaurs (both sensu Gauthier) are polymorphic as Ceratosaurus, Acrocanthosaurus and Allosaurus have robust metatarsi.

32. Sauropodomorphs are polymorphic, as Efraasia has a symmetrical pes. Ceratosaurs sensu Gauthier are also polymorphic, as Dilophosaurus has an asymmetrical pes while Megapnosaurus' is symmetrical. Liliensternus' and Procompsognathus' pes are asymmetrical. Hulsanpes' pes is also asymmetrical based on the short metatarsal II.

33. In the outgroup, Scleromochlus, pterosaurs and Lagosuchus lack phalanges on pedal digit V, making the outgroup polymorphic. Basal sauropodomorphs Saturnalia and Pantydraco lack phalanges as well, making Sauropodomorpha basally derived for this character.

34. Liliensternus does not preserve the first pedal digit, so whether its metatarsal I contacts the tarsus is unknown.

35. Thin-walled long bones are present in pterosaurs and Marasuchus in the outgroup, making it polymorphic. Anchisaurus and material referred to Thecodontosaurus is also hollowed comparably to theropods, making Sauropodomorpha possibly basally derived.

36. Ornithischia was coded as having an enlarged fang-like tooth in the anterior dentary, but while heterodontosaurids have this, Pisanosaurus and other taxa except for pachycephalosaurs do not. They are thus recoded as polymorphic here. The basal sauropodomorph Panphagia has a fang-like tooth, making sauropodomorphs properly coded polymorphic. Ceratosaurs sensu Gauthier are also polymorphic, as he notes (though he codes them as plesiomorphic), since Ceratosaurus has no fang-like tooth. Elmisauridae is coded as lacking fang-like teeth by Gauthier based on Richardoestesia which was originally incorrectly referred to Chirostenotes. Chirostenotes is now known to have toothless jaws. Saurornitholestes is now known to lack a fang-like tooth.

37. This is another composite character, involving both the maxillary fenestra's size and anteroposterior placement. Procompsognathus' skull is probably incorrectly referred, so it is rescored as unknown. Chirostenotes is now known to lack large maxillary fenestrae, while Saurornitholestes has them. Avialae should be polymorphic, as Hesperornithes and Aves lack maxillary fenestrae. Regarding placement, Tarbosaurus, Tyrannosaurus, Struthiomimus, Dromiceiomimus and Saurornithoides have fenestrae placed at the anterior edge of their antorbital fossae, but are also derived within their respective clades.

38. Scleromochlus has an antorbital tooth row, making the outgroup polymorphic. Procompsognathus is again unknown for this cranial character. Saurornitholestes' maxilla shows its tooth row is antorbital.

39. Ornithomimosaurs do not have axial spine tables (e.g. Gallimimus), contra Gauthier. They are also absent in Microvenator and oviraptorids (e.g. IGM 100/42), but present in Saurornitholestes.

40. This character involves the proximalness of the caudal transition point, and thus correlates with character 20 involving the presence of the transition point. All taxa coded as 0 for 20 should thus be coded unknown for this character (including Ornithischia and Sauropodomorpha), to stop weighting the condition unequally. Herrerasaurus has a proximally placed transition point (22 caudals with transverse processes), so is coded as apomorphic. In actuality, coelophysoids like Megapnosaurus are no less developed than carnosaurs like Allosaurus in the various features used to define this character. Megapnosaurus' transition point is more proximally placed (caudal 21) and its mid caudal vertebrae are more elongate than Allosaurus, with lower neural spines, more reduced transverse processes and equally elongate prezygapophyses. Dilophosaurus is similar (last transverse process on caudal 19, last neural spine on caudal 22). While Megapnosaurus is reported to lack the modified hatchet-like chevrons of Allosaurus and many coelurosaurs, Dilophosaurus is said to have them, and Compsognathus lacks them but is still coded 1 by Gauthier. Thus ceratosaurs are here coded as polymorphic. Ornitholestes and Saurornitholestes both have proximally placed transition points.

41. The outgroup Herrerasaurus has a strap-like scapula. Ceratosaurs sensu Gauthier should be polymorphic, as Ceratosaurus has a strap-like scapula. Carnosauria sensu Gauthier should be polymorphic as well, since tyrannosaurids have highly expanded distal scapulae (as mentioned by Gauthier, though he codes the OTU 0). Ornitholestes is unknown, as it does not preserve a scapula. Compsognathus is also polymorphic, with the two specimens differing. Oviraptorids such as Oviraptor and IGM 100/42 are apomorphic, as is Saurornitholestes.

42. Procompsognathus has a rounded posterior coracoid border, while Microvenator, Chirostenotes, oviraptorids like IGM 100/42, and Saurornitholestes have an elongate posterior process.

43. This character (manus >66% of humerus+radius length) is correlated with 7 (manus >45% of humerus+radius length), so taxa scored 0 for 7 should be scored unknown for this character. Carnosaurs sensu Gauthier should be polymorphic, as some (Allosaurus, Acrocanthosaurus, Albertosaurus) have elongate manus while others (Gorgosaurus, Tyrannosaurus) have short manus. While ornithomimosaurs are coded as having long manus, Gauthier notes is sometimes shorter than 66% of the humerus+radius length, and indeed this seems to be the plesiomorphic state (Deinocheirus, Struthiomimus, Gallimimus) with only Dromiceiomimus having a long manus. Compsognathus has a long manus now that the metacarpal and phalangeal identities are known. Based on their incomplete manus, Liliensternus and Procompsognathus have short manus, while Coelurus and Microvenator have long manus.

44. Carnosaurs sensu Gauthier are polymorphic, since tyrannosaurids (Daspletosaurus, Tyrannosaurus) and Acrocanthosaurus have short articulations for metacarpal II on metacarpal I. Ornitholestes is plesiomorphic for this character, as its articulated manus is referred to Tanycolagreus and its holotypic metacarpal I has a facet for metacarpal II shorter than half its length. Ornithomimosaurs are polymorphic as Deinocheirus has a short surface. Microvenator and Elmisaurus are plesiomorphic.

45. Ornitholestes is unknown due to a lack of manual material. Carnosaurs sensu Gauthier are polymorphic, as Acrocanthosaurus and tyrannosaurids seem to have little overlap between metacarpals II and III. Ornithomimosaurs are also polymorphic, as Deinocheirus lacks much overlap. Compsognathus is plesiomorphic based on the articulated French specimen. Oviraptorids are polymorphic, as Oviraptor lacks the character while "Ingenia" has it. Deinonychosaurs are plesiomorphic based on the articulated manus of Velociraptor while avialans are polymorphic since Archaeopteryx can exhibit both conditions.

46. Procompsognathus has a well developed metacarpal IV, while Ornitholestes, "elmisaurids" and Saurornitholestes have manus which are too fragmentary to exclude the possibility digit IV was present.

47. Procompsognathus does not preserve the ischium, contra Ostrom, so cannot be evaluated for the presence of an obturator process. Saurornitholestes has an obturator process, and Avialae should be coded polymorphic as Archaeopteryx has one.

48. The outgroup is correctly coded as lacking a pubic foot, as herrerasaurids' condition involves a twisted distal pubic flange instead of an actual posteriorly projecting foot. Procompsognathus also lacks a pubic foot, while Ornitholestes does not preserve the distal pubis. Avialae should be polymorphic, as Hesperornithes and Aves lack a pubic foot.

49. Ornithischians (heterodontosaurids, Lesothosaurus, thyreophorans) primitively have a winglike anterior trochanter. Ceratosaurs sensu Gauthier should be polymorphic, as Ceratosaurus and some Megapnosaurus and Dilophosaurus individuals have winglike anterior trochanters. Coelurus also has one, while Microvenator, Chirostenotes, Deinonychosauria (Saurornithoides, Troodon, Deinonychus, Adasaurus, Velociraptor) and Avialae (Archaeopteryx, Hesperornithes, Ichthyornis, Aves) lack the morphology (it is said to be a modified form by Gauthier, who also probably misinterpreted Microvenator's accessory trochanter). Oviraptorids are polymorphic (winglike in "Ingenia", cylindrical in IGM 100/42 and Conchoraptor).

50. Procompsognathus seems to have a short astragalar ascending process, while Coelurus, oviraptorids such as IGM 100/42 and "Ingenia", Chirostenotes and Avialae (Archaeopteryx, juvenile Aves) have tall processes. Ornitholestes and Hulsanpes lack preserves tibiae or tarsals, so are scored unknown.

51. Ornitholestes and oviraptorids (IGM 100/42, "Ingenia") have metatarsal IIIs with limited proximal exposure.

52. This is correlated with character 34, so taxa coded 0 for 34 should be coded unknown for 52. Liliensternus does not preseve pedal digit I, so its metatarsal I length is unknown. Contra Gauthier, the metatarsal I of tetanurines does not seem shorter (compared to metatarsal II) than basal avepods. In actuality, Dilophosaurus (38%), Allosaurus (31%), Velociraptor (31-33%) and Deinonychus (32%) have long metatarsals, while Procompsognathus (18%), Megapnosaurus (25%), "M." kayentakatae (25%), Acrocanthosaurus (27%), Gorgosaurus (23%), Alectrosaurus (14%), Compsognathus (23-24%), Garudimimus (22%), Chirostenotes (22%), IGM 100/42 (24%), Troodon (~15%) and Archaeopteryx have short ones. Thus Ceratosauria, Carnosauria and Deinonychosauria are changed to polymorphic,and Procompsognathus is changed to derived.

53. Carnosaurs sensu Gauthier are polymorphic for the subsidiary palatal fenestra, as tyrannosaurids have one. Oviraptorids (IGM 100/42) lack a fenestra, contra Gauthier. Avialans and deinonychosaurs are polymorphic too, since Dromaeosaurus, Deinonychus, Velociraptor and Archaeopteryx has a fenestra but Saurornithoides, Hesperornis and Aves lack one.

54. This character (depth of ectopterygoid fossa) is correlated with character 15 (presence of fossa), so all taxa coded 0 for 15 (Ornithischia, Oviraptoridae) should be coded ? for 54. Liliensternus is plesiomorphic. Carnosaurs sensu Gauthier should be polymorphic, as Acrocanthosaurus and tyrannosaurids have deep fossae. The depth of ornithomimosaurs' fossae is uncertain.

55. Ceratosaurs should be polymorphic, as Ceratosaurus and coelophysids have cervical ribs fused to their centra. Ornithomimosauria is also polymorphic, as Archaeornithomimus and some Gallimimus and Dromiceiomimus individuals have unfused ribs. Avialae is polymorphic as Archaeopteryx lacks fused ribs. Deinonychosauria and Oviraptoridae are polymorphic, while Gauthier codes both as unknown. Ornitholestes, Coelurus, Compsognathus and Chirostenotes lack fused ribs, while Saurornitholestes has them. Gauhier's suggestion maniraptoriforms with unfused ribs may be ontogenetically young may prove correct, and the juvenile Microvenator is not coded for this reason.

56. Lilensternus and Coelurus lack flexed cervical prezygapophyses, while they are present in Ornitholestes, Compsognathus, Microvenator, oviraptorid IGM 100/42 and Saurornitholestes.

57. Coelurus lacks anterior cervical centra with extremely broad anterior articular surfaces, while Ornitholestes, Microvenator, oviraptorids (IGM 100/42, Conchoraptor, "Ingenia") and Saurornitholestes have broad surfaces.

58. Ceratosaurs sensu Gauthier (Coelophysis, Megapnosaurus, "M." kayentakatae, Segisaurus), Carnosauria sensu Gauthier (Allosaurus, Gorgosaurus, Tyrannosaurus), Compsognathus and deinonychosaurs (Velociraptor) have furculae, whereas ornithomimosaurs are unknown thanks to unossified clavicles (they were coded by Gauthier as having furculae based on "Ingenia").

59. Gauthier notes pterosaurs have fused sterna, so the outgroup is coded as polymorphic. Compsognathus, ornithomimosaurs known by Gauthier and Archaeopteryx lack ossified sternal plates, so are coded 0. Ornithomimosaurs were coded as having fused sterna based on "Ingenia" by Gauthier. Archaeopteryx's condition makes Avialae polymorphic. Oviraptorids are polymorphic as IGM 100/42 has unfused sterna and "Ingenia" varies. Deinonychosaurs have unfused sternal plates based on Velociraptor and Adasaurus. Gauthier codes most coelurosaur taxa unknown because he thinks individuals with unfused plates could have been subadults, but with the continued discovery of unfused sternals in almost all coelurosaurs except alvarezsaurids, some Microraptor specimens and pygostylians, this seems unlikely.

60. Herrerasaurus and pterosaurs both have elongate forelimbs (>50% of presacral and hindlimb length), so the outgroup should be coded as polymorphic. Basal ornithischians like Heterodontosaurus (77% and 56% respectively) and basal sauropodomorphs like Efraasia (54% and 64%) have elongate forelimbs, so these clades should be coded as such. Based on their incomplete forelimbs, Procompsognathus and Liliensternus both have short forelimbs, while Coelurus does as well despite its tibial and pedal lengths being estimates. Microvenator and Chirostenotes have elongate forelimbs when their lengths are extrapolated from other taxa.

61. This character is another composite, consisting of both metacarpal I length (<33% of metacarpal II) and manus elongation (seven times longer than basally wide across metacarpals I and II). Procompsognathus lacks this character, due to its long metacarpal I. In addition its manus is probably broad based on the length of metacarpal II and phalanx II-1. Compsognathus has a short manus based on new reconstructions, Ornitholestes' is unknown thanks to its referred manus belonging to Tanycolagreus, and Coelurus has an elongate manus even though ungual II is unpreserved. Ornithomimosaurs are polymorphic because Deinocheirus and Gallimimus have short manus, while Struthiomimus and Dromiceiomimus have elongate ones. While "Ingenia" lacks elongate hands, this seems to be secondary based on their presence in Oviraptor and IGM 100/42. One problematic aspect is that the only coelurosaurs known to Gauthier that actually have metacarpal I 33% of metacarpal II length or less are Compsognathus, Avialae and maybe Oviraptor. Others (Velociraptor, IGM 100/42, Deinonychus, Chirostenotes, Elmisaurus, "Ingenia", Conchoraptor and ornithomimosaurs) have longer ones. Gauthier notes only ornithomimosaurs in reference to this, and codes them 1 anyway because it is hypothesized to be secondary. This argument is based on their nesting within Coelurosauria though, which is a result of the analysis as opposed to an a priori assumption, so cannot be used for coding purposes. Thus all coelurosaurs except for avialans are coded as polymorphic.

62. This character is correlated with character 25, since phalanx III-3 must be longer than III-1 or III-2 to be longer than both combined. Thus taxa scored as 0 for 25 (Ornithischia and Sauropodomorpha) should be scored unknown for this one. Gauthier states pterosaurs also have manual phalanx III-3 longer than phalanges III-1 and III-2 combined, but the basal Preondactylus and Peteinosaurus do not, so the outgroup is best left as plesiomorphic. Ornithomimosaurs are polymorphic, as Deinocheirus lacks the character. Chirostenotes, Elmisaurus and oviraptorids (Oviraptor, IGM 100/42, Conchoraptor, "Ingenia") lack the character as well, while Ornitholestes is unknown (and even the Tanycolagreus manus lacks it, contra Gauthier).

63. Scleromochlus and pterosaurs have reduced fourth trochanters, so the outgroup is scored as polymorphic. Liliensternus and Coelurus have prominent fourth trochanters, while Ornitholestes, Compsognathus and oviraptorids (e.g. IGM 100/42, "Ingenia") lack one. Deinonychosaurs are polymorphic, as Adasaurus and Velociraptor have well developed fourth trochanters.

64. While Gauthier calls this character "moundlike greater trochanter", he is really referencing the posterior trochanter. Coelurus, Ornitholestes, Compsognathus, ornithomimosaurs, Microvenator and oviraptorids (IGM 100/42, Conchoraptor, "Ingenia") lack posterior trochanters. Among Avialae, Hesperornithes, Ichthyornis and Aves lack the structure, so they should be coded as polymorphic. While Adasaurus lacks a posterior trochanter, this is probably a derived condition within Deinonychosauria (present in Saurornithoides, Troodon, Velociraptor and Deinonychus).

65. This character (ornithomimoid or albertosauroid astragalar ascending process) is correlated with character 50 (allosauroid, ornithomimoid or albertosauroid process), so all taxa coded as 0 for 50 are coded unknown for 65. This includes the outgroup, Ornithischia, Sauropodomorpha, Ceratosauria and Liliensternus. Carnosauria is polymorphic, as tyrannosaurids have tall ascending processes. Coelurus has an allosauroid process, while Compsognathus, Chirostenotes, oviraptorids (IGM 100/42 and "Ingenia") and avialans (Archaeopteryx, juvenile Aves) are derived.

66. Gauthier defines this character as metatarsal I lying on the posterior side of metatarsal II as opposed to the medial side, but Middleton's recent work shows that the metatarsal itself is usually placed closer to medial than posterior in living birds and that the majority of hallux rotation is due to a twisted metatarsal I shaft. In addition, articulated non-avialan theropod feet all show non-rotated halluces including and all non-avialan first metatarsals lack torsion. This includes Compsognathus and Chirostenotes, both coded by Gauthier as having posteriorly placed halluces. The placement of metatarsal I posteriorly in the Compsognathus holotype is thus near certainly due to disarticulation. Liliensternus has an unpreserved digit I, so is unknown. The first metatarsals of ornithomimosaurs (Garudimimus), oviraptorids (IGM 100/42 and "Ingenia") and deinonychosaurs (Troodon, Deinonychus and Velociraptor) lack torsion. Gauthier correctly coded carnosaurs as lacking a posteriorly rotated hallux (Allosaurus, Alectrosaurus, Albertosaurus, Tyrannosaurus), though he stated in his paper the condition was unknown. Avialans are polymorphic as Archaeopteryx and Hesperornithes lack torsion (the second probably a reversal)

67. This character is another composite, involving both the enlargement of the proximal surface of metatarsal IV compard to II (incorrectly compared to III in the text discussion) and the (presumably) anterior narrowing of metatarsal III proximally. Carnosaurs should be polymorphic, as tyrannosaurids have this character (which Gauthier states despite his coding). Ornitholestes, Compsognathus and Hulsanpes have unreduced metatarsal III, while oviraptorids are polymorphic (reduced in IGM 100/42, unreduced in "Ingenia").

68. Carnosaurs are polymorphic, as tyrannosaurids have reduced prefrontals. Ornitholestes and Compsognathus have large prefrontals.

69. Carnosaurs are polymorphic, as Acrocanthosaurus and tyrannosaurids have large axial epipophyses. Microvenator and oviraptorids (IGM 100/42) lack the condition, while basal Avialae does too (Archaeopteryx, Hesperornis, Ichthyornis).

70. Carnosaurs are derived, since both Allosaurus and tyrannosaurids have small hypapophyses. Coelurus, Ornitholestes, ornithomimosaurs (e.g. Dromiceiomimus), Microvenator, Chirostenotes, oviraptorids (e.g. IGM 100/42 and Conchoraptor) and Saurornitholestes have hypapophyses, while Compsognathus seems not to.

71. This character (modifications associated with transition point begin close to base of tail) is not only correlated with characters 40 and 20, but also is a composite involving several features in addition to the low number of caudals with transverse processes, large neural spines and short centra. These are transversely rectangular proximal caudal centra, vertically oriented proximal caudal zygapophyses, and proximal chevrons which are longer than deep. The correlation means taxa scored 0 in characters 20 and/or 40 (like Ornithischia and Sauropodomorpha) should be scored unknown here. Of other taxa, Liliensternus lacks rectangular centra at least, while Procompsognathus, Ornitholestes and Coelurus preserve unmodified proximal caudals. Saurornitholestes has the derived condition.

72. Gauthier notes pterosaurs have a strut-like coracoid, so the outgroup should be polymorphic. The coracoid is subcircular in Compsognathus and ornithomimosaurs (as Gauthier states, but oddly codes both as unknown), as well as Microvenator. It is subrectangular in oviraptorids (IGM 100/42 and "Ingenia") and Saurornitholestes.

73. This character is another composite, with the first part (forelimb >75% of presacral length) correlated with character 60 (forelimb >50% of presacral and hindlimb length) and the second (manus >90% of pes length) correlated with 7 and 43 (manus >45 and >66% respectively of humerus+radius length). Taxa scored as 0 for 60 and 7 and/or 43 (e.g. Sauropodomorpha, Ceratosauria) are scored unknown for this character to reflect the correlation. The outgroup is polymorphic thanks to pterosaurs having both characters. Heterodontosaurus has the first but not the second parts, so Ornithischia is coded as polymorphic. Contra Gauthier, based on its first digit Ornitholestes' manus was probably only ~73% as long as its pes and its forelimb ~60% of its presacral length. Coelurus, oviraptorids (e.g. IGM 100/42 and "Ingenia") and Elmisaurus are also plesiomorphic. As for deinonychosaurs, Velociraptor (~82% of pes length) and Troodon (~73% of pes length) have shorter manus than Deinonychus, so the clade is here coded as polymorphic.

74. Ornitholestes lacks a bowed ulna, while Coelurus has one.

75. This character (semilunate carpal) is correlated with character 21 (enlarged carpal block), so taxa coded as 0 for 21 should be coded unknown here. This includes the outgroup, Sauropodomorpha and Ornithischia. It also includes Ornithomimosauria, which Gauthier coded unknown already, but he did so improperly because he viewed them as potentially deriving from an ancestor with a semilunate carpal. This kind of decision should not be made prior to coding, as the analysis must discover the reversals which exist. Procompsognathus lacks a semilunate based on its preserved distal carpal II (identified as a radiale by Sereno and Wild). Ornitholestes is unknown as it lacks preserved carpal elements (as does the referred Tanycolagreus manus). Compsognathus is plesiomorphic, while Coelurus is unknown as it only preserves distal carpal I. Saurornitholestes has a semilunate.

76. This character is another composite, consisting of lateral bowing of metacarpal III and slenderness of that element compared to metacarpal II. Based on Gauthier's coding, taxa with metacarpal III <70% of metacarpal II's width are coded as derived. Procompsognathus and Liliensternus have the plesiomorphic condition (at least in regard to slenderness for the latter genus). Ceratosaurs should be coded as polymorphic for while all have straight third metacarpals, those of Dilophosaurus and Megapnosaurus are slender. As Gauthier notes, carnosaurs have very slender but unbowed third metacarpals. They are thus here scored as polymorphic, though he scored them as plesiomorphic. Ornitholestes does not preserve metacarpal III, Gauthier's coding being from the referred Tanycolagreus manus. Compsognathus is like carnosaurs in having a thin unbowed metacarpal III (it is technically bowed, but dorsally instead of laterally as shown by the ligament pit on the holotype), so is also scored as polymorphic. Ornithomimosaurs coded by Gauthier all have unbowed metacarpals, but are polymorphic for slenderness (Archaeornithomimus and Gallimimus slender; Deinocheirus, Struthiomimus and Dromiceiomimus robust). Chirostenotes and Elmisaurus must have slender metacarpal IIIs based on their digit III width, but bowing is uncertain (with hindsight is was probably absent as Hagryphus and the Triebold specimen lack it). They are coded as unknown. Oviraptorids (Oviraptor, IGM 100/42, "Ingenia") all have unbowed metacarpal IIIs and are polymorphic for slenderness (Oviraptor and "Ingenia" are slender; IGM 100/42 is robust), so are coded as polymorphic. Deinonychosaurs and avialans scored by Gauthier have thin third metacarpals, but are polymorphic for bowing (bowed in Deinonychus and Archaeopteryx but straight in Velociraptor and Ichthyornis), so are coded as polymorphic.

77. Ornitholestes has a vertical posterior ilial margin, which was drawn incorrectly by Osborn. Chirostenotes and Saurornitholestes both have posteriorly tapered ilia.

78. This is another composite character, involving opisthopuby and ventral extent of the pubic peduncle. These are often found together, but not always. Herrerasaurus has both characters, but since Staurikosaurus does not, the outgroup is left as plesiomorphic. Ornithischians obviously have an opisthopubic pelvis and Heterodontosaurus at least has a pubic peduncle extending far ventral to the ischial peduncle, but are oddly coded as plesiomorphic by Gauthier. Carnosaurs (Allosaurus, Tyrannosaurus) have ventrally extending pubic peduncles but propubic pelves, so are coded as polymorphic. Microvenator and Chirostenotes have ventrally extending pubic peduncles but probably propubic pelves, so are also coded as polymorphic. Oviraptorids include taxa with subequally projecting peduncles and a propubic pelvis ("Ingenia") and those with a ventrally extending pubic peduncle and posteriorly projecting proximal pubis (IGM 100/42), so are also polymorphic. Saurornitholestes has the ventrally extensive pubic peduncle at least, but is coded unknown as the pubic orientation is unknown. Deinonychosaurs are polymorphic as Saurornithoides (the holotype, which Gauthier thought was an undescribed specimen of Barsbold's) has a propubic pelvis. Gauthier recognized this difficulty (though he coded deinonychosaurs as derived), and later finds of Sinovenator and other basal troodontids verify one of his stated possibilities that Saurornithoides' condition is a reversal.

79. This character (pubic boot reduced anteriorly) is correlated with character 48, as taxa without significant pubic expansions cannot have a morphology with a large posterior boot and reduced anterior boot. Thus taxa coded 0 for character 48 are coded as unknown for this character (including the outgroup, Ornithischia, Sauropodomorpha, Ceratosauria and Liliensternus). Microvenator has a significant anterior expansion, while Chirostenotes' anterior boot seems larger than any posterior boot it has. Deinonychosaurs are polymorphic, as Deinonychus and Troodon have significant anterior boots, but Adasaurus and Velociraptor do not.

80. I agree with Gauthier's scoring of ceratosaurs as plesiomorphic, as the condition in derived coelophysids like Megapnosaurus (ischium 64% of pubic length) is seen as a reversal compared to Ceratosaurus, Dilophosaurus, Liliensternus and probably Segisaurus. Contra Gauthier, the condition in Ornitholestes seems to be plesiomorphic (~75%) despite the missing pubic boot. Chirostenotes has the derived condition (50%), but oviraptorids are plesiomorphic (72% in IGM 100/42, 86% in "Ingenia"). As Gauthier notes, Hesperornithes and Aves have elongate ischia. Thus Avialae should be coded as polymorphic, though he codes them as being derived due to hypothesized secondary elongation.

81. This character (obturator process placed distally) is correlated with character 47 (obturator process present), so taxa coded 0 for 47 (such as the outgroup, Ornithischia, Sauropodomorpha, Ceratosauria and Liliensternus) are coded unknown for this character. Contra Gauthier, Compsognathus (22% down the ischium) and Ornitholestes (26%) have proximally placed obturator processes (e.g. compare to 21% in Allosaurus and 27% in Tyrannosaurus). Chirostenotes and Saurornitholestes have distally placed processes. Avialae is coded as derived, as Archaeopteryx has a distally placed obturator process. Gauthier coded the OTU plesiomorphic based on ornithurines' absence of a process and was conservative in interpreting Archaeopteryx.

82. Ornithischia is polymorphic for this character (anterior trochanter separated from greater trochanter by small cleft), as Heterodontosaurus has a trochanteric crest. Compsognathus, Coelurus and Chirostenotes are plesiomorphic. Oviraptorids (IGM 100/42, Conchoraptor, "Ingenia") are derived. Avialae is polymorphic, as Archaeopteryx has a deep cleft between its trochanters.

83. This character (fourth trochanter absent) is correlated with character 63 (fourth trochanter reduced), so taxa coded as 0 for 63 are coded unknown for character 83. This includes Ornithischia, Sauropodomorpha, Ceratosauria, Procompsognathus, Liliensternus and Carnosauria. The outgroup is polymorphic, as pterosaurs and Scleromochlus lack fourth trochanters. Ornitholestes, Compsognathus and oviraptorids (IGM 100/42, Conchoraptor and "Ingenia") lack fourth trochanters. Deinonychosaurs are polymorphic, as Velociraptor and Adasaurus have fourth trochanters.

84. This character (asymmetrical pes) is simply the inverse of character 32 (symmetrical pes), so it should be deleted. All thirteen taxa scored for it are changed to unknown.

General analysis conclusions- Gauthier's analysis is of course the first published cladistic analysis of theropods, so wouldn't be expected to do everything correctly. One issue is the common correlation of characters (14 of the 84 are correlated with others), which inflates the value of certain features (forelimb and manus size, transition point placement, etc.). Another is that many characters actually describe two or more variables that are independent from one another (17 of 84) and should rightfully be split. One flaw common to many early analyses is that the characters are clearly designed to support certain clades Gauthier had in mind, as they are all arranged in order of which clade they help diagnose. This makes the whole thing more a demonstration of his idea than a test of competing ideas. One philosophical issue is that Gauthier often assigns taxa codings based on his subjective opinion of whether a certain state should be present, considering the phylogenetic position he believes them to have. At its worst, this leads to ornithischians being coded as lacking opisthopuby (presumably) because they are not closely related to eumaniraptorans. An explicitly stated example is ornithomimosaurs being coded as uncertain for the presence of a semilunate carpal when they clearly lack one, because Gauthier believed it was possible they evolved from ancestors with it. This is turning the process on its head. Somewhat less egregious are his frequent codings of taxa like Avialae with one state when only one of the two subtaxa exhibits that state (e.g. short ischium in Archaeopteryx but not Ornithurae). While in that particular case and many others, Gauthier is likely correct in his assignment of the plesiomorphic state, this was only (and can only be) verified with future discoveries. For instance, we now know of basal avialans like Shenzhouraptor and Confuciusornis which also have short ischia, showing the condition in Hesperornithes and Aves is derived within Avialae. But in 1986, it was just as possible Archaeopteryx had secondarily shortened its ischium, when we presume to know nothing of its sister group. The all-zero outgroup is never a good idea, especially considering that for 32 of the 84 characters, the derived state was present in some members. Inevitably, Gauthier was limited by the literature available in 1986. In particular, new material of Herrerasaurus, Chirostenotes, Troodon, Saurornitholestes and Velociraptor have been discovered, and most taxa with the exceptions of Liliensternus and Hulsanpes have received recent redescription. Oviraptorids, Coelurus and Ornitholestes were especially poorly described by 1986. One aspect in which Gauthier exceeds most other analyses is his detailed commentary on each character, with illustrations of many. Another positive consideration is that for most fusion characters, he takes the specimens' age into account. While he may be correct in some instances (cervical ribs), in others it seems he was overly cautious (sternal plates). All considering, 486/1512 (32%) of the scorings are incorrect. When corrected, the tree changes to become-

|--outgroup
`--+--Ornithischia 
   `--Saurischia
      |--Sauropodomorpha
      `--Theropoda
         |--Ceratosauria
         |--Procompsognathus
         |--Liliensternus
         `--Tetanurae
            |--Carnosauria
            |--Coelurus
            |--Compsognathus
            `--Coelurosauria
               |--Ornitholestes
               `--Maniraptoriformes
                  |--Ornithomimosauria
                  `--Maniraptora
                     |--Microvenator
                     `--+--Caenagnathidae
                        `--+--Oviraptoridae
                           `--Eumaniraptora
                              |--Saurornitholestes
                              |--Deinonychosauria
                              `--Avialae

Hulsanpes has an uncertain position outside Maniraptoriformes due to only being coded for three characters. It agrees with recent analyses in all respects except the paraphyletic Oviraptorosauria.

Phylogenetic conclusions- For the following calculations, Hulsanpes and Saurornitholestes were deleted, as they agree in all their scored characters with Deinonychosauria in Gauthier's original matrix. The table shows the number of extra steps needed to accomodate each rearrangement using Gauthier's original matrix, and his recoded matrix. A negative number means the arrangement is already most parsimonious, but that many steps are needed to undo it.

rearrangement original recoded
(outgroup,Carnosauria(Sauropodomorpha,Ornithischia)) (Cooper, 1985) 10 2
(outgroup((Ceratosauria,Liliensternus,Coelurus,Compsognathus,Ornitholestes,Ornithomimidae,Caenagnathidae,OviraptoridaeDeinonychosauria)(Carnosauria,Sauropodomorpha))) (Huene, 1914) 40 19
(outgroup(Carnosauria(Ceratosauria,Liliensternus,Procompsognathus,Coelurus,Compsognathus,Ornithomimidae))) (Huene, 1914) 16 12
(outgroup,Ceratosauria,Ornithomimidae,Deinonychosauria(Carnosauria,Coelurus)) (Senter, 2007) 3 0
(outgroup,Ceratosauria,Ornithomimidae,Deinonychosauria(Carnosauria,Compsognathus)) (Olshevsky, 1991) 3 1
(outgroup,Ceratosauria,Deinonychosauria(Carnosauria,Ornitholestes/Ornithomimidae)) (Paul, 1988/Barsbold, 1983) 11 3
(outgroup,Ceratosauria,Coelurus(Ornithomimidae,Deinonychosauria)) 1 -3
(outgroup,Ceratosauria,Compsognathus(Ornithomimidae,Deinonychosauria)) 2 -1
(outgroup,Ceratosauria,Ornitholestes(Ornithomimidae,Deinonychosauria)) 6 -1
(outgroup,Ceratosauria,Deinonychosauria,Avialae(Ornithomimidae,Caenagnathidae,Oviraptoridae)) 6 6
(outgroup,Deinonychosauria,Avialae,Ornitholestes,Compsognathus,Coelurus(Microvenator,Caenagnathidae,Oviraptoridae)) 1 2
(outgroup,Ceratosauria,Deinonychosauria(Oviraptoridae,Avialae)) (Elzanowski, 1999) 1 3
(outgroup,Avialae(Ornithischia,Sauropodomorpha,Ceratosauria,Deinonychosauria)) 64 25
(outgroup,Deinonychosauria,Avialae(Ornithischia,Sauropodomorpha,Ceratosauria,Carnosauria)) 50 16

As the first phylogenetic analysis of theropods, Gauthier's work provided the first objective support for several groups. Saurischia is strongly supported by Gauthier's original data, but recoding leaves it with only weak support (no doubt due in part to including herrerasaurids in the outgroup). An alternative called Phytodinosauria or Ornithischiformes was popular in the late 80's which no numerical analysis has supported. Theropod monophyly was standard in 1986, and explicit alternatives were always rare. One notable exception was Huene's work from the 1910's-1930's which placed carnosaurs sister to sauropodomorphs in the clade Pachypodosauria. This hypothesis is rejected extremely strongly by Gauthier's data, and very strongly by the recoded data as well. Gauthier provides strong support for coelophysoids, Procompsognathus, Liliensternus and Ceratosaurus being basal to other theropods, which he named Tetanurae. The standard alternative was to place these taxa in Coelurosauria instead, but that is strongly rejected by both datasets. Within Gauthier's Tetanurae, he found Carnosauria to be the sister group to a Coelurosauria containing several taxa. However, placing Compsognathus and Coelurus in Coelurosauria was only weakly supported. With recoding the support is even weaker, with Coelurus having an ambiguous position. Placing these taxa within Maniraptora was even more poorly supported, though the evidence for excluding them from this clade remains weak with recoding. On the other hand, both ornithomimosaurs and Ornitholestes were well supported as coelurosaurs originally, but are only poorly supported as such now. Ornitholestes was also moderately supported as a maniraptoran originally, though now it is almost ambiguously placed outside that clade. The monophyly of core maniraptorans (eumaniraptorans and oviraptorosaurs) relative to ornithomimosaurs is moderate in both datasets. Oviraptorosaurian monophyly is weakly rejected in both datasets, as is the idea oviraptorids are closer to birds than deinonychosaurs are. While placing birds or birds and other maniraptorans outside Dinosauria (as Martin, Feduccia, et al. support) was extremely strongly rejected by Gauthier's original data, it remains strongly rejected with the new data.

Experiments with controversial taxa- Holtz (1994) and others criticized Gauthier for assuming the monophyly of Carnosauria and Deinonychosauria, suggesting Allosauridae, Tyrannosauridae, Troodontidae and Dromaeosauridae be coded as separate OTUs. Indeed, Gauthier himself noted that due to the propuby of Saurornithoides and bullatosaurian characters noted by Currie, he should have broken up Deinonychosauria. Similarly, most recent analyses have suggested coelophysoids are not ceratosaurs, and Elaphrosaurus is universally recognized to be a ceratosaur instead of an ornithomimosaur. Charig and Milner (1990) criticized Gauthier's analysis, on both deserved and undeserved grounds, but concluded Baryonyx did not fit into his system of classification. This ignores how cladistic analysis works- a taxon is supposed to be entered into the matrix to find the MOST parsimonious placement. It is not expected that there will be no homoplasy so that one can simply look at each node's characters and plug the new taxon in manually. Gauthier noted the "megalosaur problem", where taxa such as Eustreptospondylus and Megalosaurus were difficult to analyze due to incompleteness and a lack of alpha taxonomy. This has recently been resolved by Benson (2010) and Sadlier et al. (2008). Herrerasaurus is included in the outgroup, which is problematic since it has been more recently recognized as a dinosaur (the ingroup). Finally, Gauthier placed segnosaurs (now therizinosaurs) in Sauropodomorpha, but recent analyses include them in Coelurosauria. Gauthier's matrix will be tested by deleting the outgroup, Ceratosauria, Carnosauria and Deinonychosauria, and adding an outgroup excluding herrerasaurids, Herrerasauridae, Coelophysoidea, Ceratosaurus, Elaphrosaurus, Megalosaurus, Eustreptospondylus, Baryonyx, Allosauroidea (Allosaurus and Acrocanthosaurus), Tyrannosauridae, Segnosauridae, Troodontidae and Dromaeosauridae. Segnosauridae is only coded based on Segnosaurus and Erlikosaurus, as Therizinosaurus and Nanshiungosaurus were not often recognized as being related in 1986. The resulting cladogram is-

|--outgroup
`--+--Ornithischia 
   `--Saurischia
      |--Sauropodomorpha
      `--Theropoda
         |--Herrerasauridae
         `--Avepoda
            |--Coelophysoidea
            `--+*-Procompsognathus
               |--Liliensternus
               `--Neotheropoda
                  |*-Elaphrosaurus
                  |--Ceratosaurus
                  `--Tetanurae
                     |--Eustreptospondylus
                     |--Baryonyx
                     `--Avetheropoda
                        |*-Compsognathus
                        |--Megalosaurus
                        |--Allosauroidea
                        |--Tyrannosauridae
                        |--Coelurus
                        `--Maniraptoriformes
                           |--Ornitholestes
                           |--Ornithomimosauria
                           `--Maniraptora
                              |*-Microvenator
                              |--+--Segnosauridae
                              |  `--Caenagnathidae                         
                              `--+--Oviraptoridae
                                 |*-Troodontidae
                                 |*-Saurornitholestes
                                 `--Eumaniraptora
                                    |--Dromaeosauridae
                                    `--Avialae


Hulsanpes was excluded as it can fall basically anywhere outside Maniraptora. Within that tree, several taxa have unstable positions. Procompsognathus is at least as derived as Liliensternus, but is outside Maniraptoriformes. Elaphrosaurus can go anywhere in Neotheropoda but outside of Avetheropoda. Compsognathus can go anywhere in Avetheropoda outside Maniraptora. Microvenator can go anywhere in Maniraptora outside the Oviraptoridae+Paraves clade. Troodontidae and Saurornitholestes can go anywhere in the Oviraptoridae+Eumaniraptora clade. Several features match recent placements- herrerasaurids are saurischian theropods, Ceratosaurus is closer to tetanurines than coelophysoids, Elaphrosaurus is outside Avetheropoda let alone Ornithomimosauria, Eustreptospondylus and Baryonyx are basal tetanurines, and segnosaurs are maniraptorans.

An interesting experiment is to enforce a traditional Coelurosauria and Carnosauria, since Gauthier's work was one of the first papers to challenge it with an explicit alternative. Herrerasaurids, segnosaurs, dromaeosaurids and avialans were not specified as carnosaurs or coelurosaurs. The resulting tree was 13 steps longer-

|--outgroup
`--+--Ornithischia 
   `--Saurischia
      |--Sauropodomorpha
      `--Theropoda
         |--Herrerasauridae
         `--Avepoda
            |*-Baryonyx
            |--Carnosauria
            |  |--Ceratosaurus
            |  `--+--Megalosaurus
            |     |--Eustreptospondylus
            |     |--Allosauroidea
            |     `--Tyrannosauridae
            `--Coelurosauria
               |--Coelophysoidea
               |--Procompsognathus
               |--Liliensternus
               `--+--Coelurus
                  `--+--Elaphrosaurus
                     `--+--Compsognathus
                        `--+--Ornitholestes
                           `--Maniraptoriformes
                              |--Ornithomimosauria
                              `--Maniraptora
                                 |*-Microvenator
                                 |--+--Segnosauridae
                                 |  `--Caenagnathidae                         
                                 `--+--Oviraptoridae
                                    |*-Troodontidae
                                    |*-Saurornitholestes
                                    `--Eumaniraptora
                                       |--Dromaeosauridae
                                       `--Avialae


Interestingly, when the positions of Ceratosaurus and tyrannosaurids were left to vary as well (since Huene and a few others had considered them coelurosaurs), they both emerged as coelurosaurs. Tyrannosaurids were between Elaphrosaurus and Compsognathus, while Ceratosaurus was just closer to birds than the coelophysoid-grade taxa. This tree was 11 steps longer than the original. Similarly, constraining a diphyletic Theropoda as in Huene (1932) where Megalosaurus, Eustreptospondylus, Allosauroidea and Tyrannosauridae are closer to sauropodomorphs in a Pachypodosauria results in trees 22 steps longer. Baryonyx emerges as a carnosaur (matching Huene's referral of Spinosaurus to that group), herrerasaurids are basal saurischians, while segnosaurids and birds emerge as coelurosaurs.

|--outgroup
`--+--Ornithischia 
   `--Saurischia
      |--Herrerasauridae
      `--+--Pachypodosauria
         |  |--Sauropodomorpha
         |  `--Carnosauria
         |     |--Baryonyx
         |     `--+--Eustreptospondylus
         |        `--+--Megalosaurus
         |           |--Allosauroidea
         |           `--Tyrannosauridae
         `--Coelurosauria
            |--Coelophysoidea
            `--+--Procompsognathus
               |--Liliensternus
               `--+--Ceratosaurus
                  `--+--Coelurus
                     `--+--Elaphrosaurus
                        `--+--Compsognathus
                           `--+--Ornitholestes
                              `--Maniraptoriformes
                                 |--Ornithomimosauria
                                 `--Maniraptora
                                    |*-Microvenator
                                    |--+--Segnosauridae
                                    |  `--Caenagnathidae                         
                                    `--+--Oviraptoridae
                                       |*-Troodontidae
                                       |*-Saurornitholestes
                                       `--Eumaniraptora
                                          |--Dromaeosauridae
                                          `--Avialae


To place herrerasaurids outside Eusaurischia requires 2 more steps, and to place them outside Dinosauria requires 7 more steps. Thus herrerasaurids are moderately supported as saurischians, but only poorly supported as theropods within that clade.

To place Liliensternus and Procompsognathus in Coelophysoidea requires a single additional step, while making coelophysoids ceratosaurs is two steps longer, with Liliensternus and Elaphrosaurus becoming coelophysoids as well. Gauthier's assumption of ceratosaurian monophyly was based on six characters. Of these, the trochanteric shelf is probably plesiomorphic as it is found in the outgroup, Saturnalia and Herrerasaurus. The proximally fused tarsometatarsus is a coelophysid synapomorphy lacking in Dilophosaurus and Ceratosaurus (the one fused metatarsus of this taxon is pathological). The hood-like supracetabular crest would resolve as plesiomorphic, being present in Saturnalia, herrerasaurids, Eustreptospondylus and Megalosaurus. The medial fibular ridge and associated groove would have been resolved as a valid ceratosaurian character though. Same with the tibiofibular sulcus on the distal femur, though Megalosaurus also has this. Finally, Gauthier hypothesizes a narrow pubis may diagnose the group because Allosaurus' is wider, but Megalosaurus, Eustreptospondylus, Tyrannosaurus, Coelurus and others all have slender pubes as well. Thus two of Gauthier's ceratosaur characters are valid, which would exactly balance the two extra steps needed to make coelophysoids ceratosaurs when they are coded separately from Ceratosaurus itself. This ends up making his analysis equivocal in regard to that question, and his assumed monophyly of ceratosaurs not very problematical. Elaphrosaurus was assumed to be an ornithomimosaur (forcing it to be such requires 4 steps), though of the five characters it can be coded for in his diagnosis of the clade, it only has one- gracile humerus with low deltopectoral crest. Of the others, two deserve comment. The metatarsus is said to be narrow and elongate in comparison to tibial length, but the metatarsotibial ratio of Elaphrosaurus (64%) is also similar to some ceratosaurs sensu Gauthier - Segisaurus (62%), Procompsognathus (61%) and Compsognathus (64%). Similarly, the pedal digits are said to be short and stout, but that is only based on comparison to the elongate metatarsus. When compared to a neutral value like proximal metatarsus width, the single preserved digit of Elaphrosaurus (II, excluding the ungual) is longer (179%) than ceratosaurs such as Liliensternus (158%) and Dilophosaurus (138%). The other two ornithomimosaur characters it lacks are a ventrally curved ischium and an arctometatarsus. As it required 4 steps to make Elaphrosaurus an ornithomimosaur, the shared humeral morphology would bring that down to 3 steps, and Elaphrosaurus is only weakly excluded from Ornithomimosauria. As for Procompsognathus and Liliensternus, the latter lacks both ceratosaur synapomorphies and the former is unknown in this regard. However, subadult Dilophosaurus and/or "Megapnosaurus" kayentakatae also lack the features, meaning they may have been present in Liliensternus but merely unexpressed in the subadult syntypes.

Contra Charig and Milner's statements, Gauthier's matrix is capable of handling Baryonyx and resolves it in its correct position as a basal tetanurine. The "megalosaur problem" does not seem to be a major issue, as Eustreptospondylus was resolved in its proper place as a basal tetanurine, while Megalosaurus resolved as slightly more derived, though the data could not exclude an avetheropod placement. Enforcing a Megalosauroidea of Megalosaurus, Eustreptospondylus and Baryonyx is 2 steps longer, with Elaphrosaurus sometimes joining. Forcing Baryonyx and/or Eustreptospondylus to be carnosaurs is 2 steps longer (with or without tyrannosaurids), though Coelurus is sometimes a carnosaur as well and Baryonyx can be a carnosaur or have a position sister to Neotheropoda. However, this does not take into account any carnosaurian characters the three taxa may have. Gauthier provides a long list of carnosaurian synapomorphies, of which Baryonyx has seven (supraorbital crests; narrow and short frontals and parietals; posterolateral surangular ridge; strongly opisthocoelous anterior presacral vertebrae; large transverse processes and neural spines; manual digits II and III reduced; robust postcrania) and lacks one (shortened presacral centra). Similarly, Eustreptospondylus has six (dorsoventrally elongate orbit; supraorbital crests; narrow and short frontals and parietals; strongly opisthocoelous anterior presacral vertebrae; large transverse processes and neural spines; robust postcrania) and lacks two (shortened presacral centra; dorsally expanded preacetabular process). Megalosaurus has five (dorsoventrally elongate orbit; strongly opisthocoelous anterior presacral vertebrae; large transverse processes and neural spines; dorsally expanded preacetabular process; robust postcrania) and lacks one (posterolateral surangular ridge) and has presacral centra of intermediate shortness. It is worth mentioning tyrannosaurids themselves have only weakly or non-opisthocoelous presacrals and reduce manual digit III (which is similarly reduced in other avetheropods) but not II. Also, Ceratosaurus has eight (dorsoventrally elongate orbit; supraorbital crests; narrow and short frontals and parietals; posterolateral surangular ridge; shortened presacral centra; large transverse processes and neural spines; dorsally expanded preacetabular process; robust postcrania). This indicates Gauthier's idea that these characters are size-related (or alternatively basal for Neotheropoda) is correct, and the only character remaining as valid carnosaurian synapomorphy is the reduced external mandibular fenestra, with the reduced manual digit II and strongly opisthocoelous anterior presacrals perhaps uniting allosauroids with megalosauroids. Since Baryonyx or Eustreptospondylus are carnosaurs with 2 more steps and Megalosaurus is already carnosaur-grade, Baryonyx and Megalosaurus are equally parsimoniously allosauroids or basal tetanurines, while Eustreptospondylus is 1 step more likely to be a basal tetanurine than an allosauroid. While Gauthier has been criticized by later authors for assuming carnosaurian monophyly, allosauroids and tyrannosaurids emerge at the same grade even if coded separately, though Megalosaurus, Compsognathus and Coelurus can all be carnosaurs as well. Taking the single valid carnosaurian synapomorphy of Gauthier into account, their monophyly including Compsognathus (as it shares the reduced external mandibular fenestra) would be one step more likely than paraphyly, though Megalosaurus and Coelurus would still have ambiguous placement. Thus Gauthier's assumption of carnosaurian monophyly was non unjustified given his data.

Segnosaurs were viewed by Gauthier as sauropodomorphs, specifically within his clade of "broad-footed" taxa excluding Thecodontosaurus, Efraasia and Anchisaurus (now known to be a more derived sauropodomorph). Segnosaurs known to Gauthier shared 3 characters which he lists as sauropodomorph synapomorphies- lanceolate teeth with coarsely serrated crowns; small skull; tibia shorter than femur. They also share an arched dorsal ilial margin and completely open acetabulum with Anchisaurus and broad-footed taxa. Finally, they are the following with broad-footed sauropodomorphs- large external naris; mandibular condyle set below tooth row; broad pes. However, they lack a transversely compressed dorsal premaxillary process; maxillary teeth increase in height anteriorly; robust forelimbs. These latter three can be ignored though, as it is always possible segnosaurs are the sister taxon of broad-footed sauropodomorphs. The arched dorsal ilial margin and completely open acetabulum can be ignored, as these are also found in caenagnathids, which are the theropod sister taxon of segnosaurids in trees derived from Gauthier's matrix. This leaves 6 characters supporting a relationship to sauropodomorphs, which when subtracted from the 10 extra steps it took to place them there leaves us with low support of 4 steps for theropod relationships. Gauthier's matrix provides low support against Oviraptorosauria, but does not include the mandibular characters which left Gauthier with "no doubt" that Caenagnathus is related to oviraptorids. Gauthier was criticized for assuming deinonychosaurian monophyly in the 1990's, but coding dromaeosaurids and troodontids separately results in troodontids being paravians or deinonychosaurs in some trees and closer to oviraptorids in other trees. Adding Gauthier's listed deinonychosaurian synapomorphies (pedal ungual II enlarged, strongly recurved and strongly compressed; pedal phalanx II-2 shortened; prominent posteroventral heel on phalanx II-2) leads to at least weak support for deinonychosaurian monophyly compared to Avialae. It is certainly better supported than some alternatives, as forcing non-monophyly of core maniraptorans by placing troodontids and oviraptorosaurs closer to ornithomimosaurs as in Holtz (1992) is 6 steps longer, while forcing Bullatosauria (Troodontidae+Ornithomimosauria, which Gauthier mentions as an idea of Currie's) is 7 steps longer.