Russell and Dong, 1994. The affinities of a new theropod from the Alxa Desert, Inner Mongolia, People’s Republic of China. Canadian Journal of Earth Sciences. 30, 2107-2127.

This paper described Alxasaurus and was the main influence placing segnosaurs back into Theropoda after a decade of viewing them as sauropodomorphs or phytodinosaurs. It also began the unecessary replacement of Segnosauria with Therizinosauroidea and includes a heterodox phylogenetic analysis.


Russell and Dong's phylogeny was-

      |  |--Ornithomimidae
      |  `--+--Therizinosauroidea
      |     `--+--Troodontidae
      |        `--Oviraptorosauria

Taxon Issues

Plateosaurus- As this was based on Galton's (1990) concept of the genus, so includes P. engelhardti and Gresslyosaurus (or Plateosaurus) ingens, but not Sellosaurus (or Plateosaurus) gracilis.
Coelophysis- Only based on C. bauri, and not C? rhodesiensis.
Ornithomimidae- The authors include Harpymimus, Garudimimus, Archaeornithomimus, Gallimimus, Struthiomimus, Dromiceiomimus brevitertius, D. samueli, Ornithomimus and Anserimimus. Thus it is more properly Ornithomimosauria.
Therizinosauroidea- The authors include Alxasaurus, "Chilantaisaurus" zheziangensis, Nanshiungosaurus, Enigmosaurus, Erlikosaurus, Segnosaurus and Therizinosaurus.
Troodontidae- The authors include Sinornithoides, IGM 100/44, Borogovia, Saurornithoides, Zanabazar and Troodon.
Oviraptorosauria- Problematically, the authors use this as an OTU and a name for one of their large clades. They use Microvenator, Chirostenotes, Elmisaurus, Oviraptor, IGM 100/42, Rinchenia, Conchoraptor and "Ingenia".
Yangchuanosaurus- Both Y. shangyouensis and Y. magnus are used, including 'Szechuanosaurus' specimen CV 00214 following Carrano et al. (2012).
Allosaurus- Both A. fragilis and the probably conspecific A. atrox and A. "jimmadseni" are used here.
Tyrannosauridae- The authors use Alectrosaurus, Gorgosaurus, Albertosaurus, Daspletosaurus, Alioramus, Tarbosaurus and Tyrannosaurus.
Dromaeosauridae- The authors use Saurornitholestes, Adasaurus, Velociraptor, Deinonychus and Dromaeosaurus.

Coding Issues

1. Plateosaurus (Prieto-Marquez and Norell, 2011), Allosaurus (Chure, 2000), tyrannosaurids (Gorgosaurus- Currie, 2003; Daspletosaurus- Currie et al., 2003) and oviraptorosaurs (Oviraptor- Clark et al., 2002; IGM 100/42- unpublished photos; Rinchenia and "Ingenia"- both Lu, 2004) have premaxilla-nasal contact below the naris.

3. Plateosaurus (Prieto-Marquez and Norell, 2011) lacks palatal premaxillary contact. Baryonyx (Charig and Milner, 1997), troodontids (Saurornithoides- Norell et al., 2006; Zanabazar- Norell et al., 2009) and dromaeosaurids (Velociraptor- Barsbold and Osmolska, 1999; Saurornitholestes, Deinonychus and Dromaeosaurus- all Currie and Varricchio, 2004) have such contact. Tyrannosaurids are polymorphic (present in Tyrannosaurus- Brochu, 2003).

4. Baryonyx (Charig and Milner, 1997; Sereno et al., 1998) and troodontids (Saurornithoides and Zanabazar- both Norell et al., 2009) have posteriorly narrowing nasals.

5. Baryonyx (Charig and Milner, 1997), Yangchuanosaurus (Barsbold et al., 1983) and tyrannosaurids (Gorgosaurus and Daspletosaurus- both Currie, 2003) have prefrontals. Oviraptorosaurs (Rinchenia and "Ingenia"- both Lu, 2004; Conchoraptor- Barsbold et al., 1990; IGM 100/42- unpublished photos) lack them. Dromaeosaurids are polymorphic (Deinonychus- Witmer and Maxwell, 1996; and Dromaeosaurus- Currie, 1995 have them).

6. "Lacrimal, narial or postorbital horns absent (0), present (1)" is a composite character. Only lacrimal horns are coded for here. Oviraptorosaurs (Oviraptor and Rinchenia- both Clark et al., 2002; Conchoraptor- Barsbold et al., 1990) lack lacrimal horns.

7. "Frontals narrow or truncated anteriorly, postorbital ramus projects laterally from orbital margin of frontal (0), very broadly exposed on skull roof, postorbital ramus does not project abruptly laterally from the orbital margin (1)" is a composite character. Being very broadly exposed does not mean frontals are not narrow or truncated, themselves not similar morphologies. The abruptness of the postorbital process' angle from the orbital margin is coded for here. Therizinosauroids (Erlikosaurus- Clark et al., 1994) have an abrupt postorbital process. Plateosaurus (Galton, 1984) and Baryonyx (Sereno et al., 1998) has a confluent postorbital process. Tyrannosaurids (Gorgosaurus and Albertosaurus- both Currie, 2003; Daspletosaurus- Russell, 1970) have extremely reduced orbital margins that should be coded inapplicable.

8. Baryonyx (Sereno et al., 1998) and dromaeosaurids (Saurornitholestes- Sues, 1978; Adasaurus- Turner et al., 2012; Velociraptor- Barsbold and Osmolska, 1999; Deinonychus- Paul, unpublished) have a parietal sagittal crest.

9. Baryonyx (Charig and Milner, 1997), ornithomimosaurs (Gallimimus- Currie and Zhao, 1993) and troodontids (Zanabazar- Norell et al., 2009; Troodon- Currie, 1985) have an ossified orbitosphenoid.

10. "Basicranium not pneumatized, or minimally pneumatized (0), basisphenoid, but not parasphenoid rostrum, strongly expanded and pneumatized (1)" is vague and composite, as there are several pneumatic recesses in the lower braincase including the anterior, otic, subotic, posterior, lateral basipterygoid, internal basipterygoid, basisphenoid, endocranial floor, internal basioccipital and subcondylar. As the subotic recess has a similar distribution to Russell and Dong's coding, its presence is coded for here. Baryonyx (Charig and Milner, 1997) lacks a subotic recess. Allosaurus (Chure and Madsen, 1996) is polymorphic. It cannot be determined in Coelophysis (Colbert, 1989).

11. "Basipterygoid processes moderately long to long, not fused to pterygoids (0), processes abbreviated, fused to pterygoids (1)" is a composite character. As no taxon coded for by the authors has basipterygoid processes fused to their pterygoids (though Avimimus and Nemegtomaia do), basipterygoid process length is coded for here.

12. Troodontids (Saurornithoides- Norell et al., 2009) have anteriorly limited vomers.

13. While the character codes for the closure of the suborbital fenestra, it is open in all taxa, merely reduced in Erlikosaurus and oviraptorids. Thus it is redefined here as "suborbital fenestra less than 35% of orbit height". Troodontids (Saurornithoides- Norell et al., 2009) have large suborbital fenestrae.

14. "Symphyseal region of dentary broad and rather straight (0), slender, medially recurved (1)" is composite, coding for both width and curvature. Only curvature is coded for here. Yangchuanosaurus (coding in Caranno et al., 2012) has a straight dentary. Plateosaurus (Prieto-Marquez and Norell, 2011), Allosaurus (Benson, 2010) and oviraptorosaurs (Microvenator- pers. obs.; Chirostenotes- Currie et al., 1993; Oviraptor- unpublished photo) have a medially recurved dentary. Ornithomimosaurs (Harpymimus lacks curvature- coding in Senter, 2007) and troodontids (Sinornithoides lacks curvature- Currie and Dong, 2001) are polymorphic.

15. Baryonyx (Charig and Milner, 1997) has a coronoid. Ornithomimosaurs lack coronoids (Garudimimus- Kobayashi and Barsbold, 2005; Gallimimus- Hurum, 2001). No coronoid is known for Coelophysis (Colbert, 1989).

16. "Mandibular fenestrum small, rounded (0), large, horizontally oval (1)" is a composite character. Only length is coded for here, with the derived state being an external mandibular fenestra over 12% of mandibular length. Yangchuanosaurus (Dong et al., 1983) and Allosaurus (Chure, 2000) have large fenestrae. Dromaeosaurids are polymorphic (large fenestra in Velociraptor- Norell et al., 2006). Baryonyx (Sereno et al., 1998) is too incomplete to code.

17. Coelophysis (Colbert, 1989) and Yangchuanosaurus (Dong et al., 1983) lack a surangular fenestra. Dromaeosaurids (Velociraptor- Osborn, 1924; Adasaurus- Turner et al., 2012; Deinonychus- Ostrom 1969; Dromaeosaurus- Currie, 1995) have a fenestra. Troodontids are polymorphic (IGM 100/44 lacks a fenestra- Barsbold et al., 1987).

18. "Teeth relatively small and numerous (more than 25 in dentary) (0), teeth large, less than 25 in dentary (1)" is a composite character involving both tooth size and number. Only number is coded for here, with 25 teeth being counted as part of state 0. Baryonyx (Charig and Milner, 1997) has more than 24 dentary teeth. Plateosaurus (Prieto-Marquez and Norell, 2011) and ornithomimosaurs (Harpymimus- Kobayashi and Barsbold, 2005) have less than 25 dentary teeth. Therizinosauroids are polymorphic (Segnosaurus has 24- Barsbold and Maryanska, 1990).

19. This is potentially correlated with character 18, as dentary tooth number would relate to the dentary/maxillary tooth ratio if maxillary number were constant. Dromaeosaurids are polymorphic, as Velociraptor (Barsbold and Osmolska, 1999) has a higher dentary/maxillary tooth ratio than Erlikosaurus. Baryonyx cannot be coded, as the maxilla is incomplete (Charig and Milner, 1997).

20. "Denticles on teeth large, teeth with unconstricted roots (0), denticles small, roots constricted (1)" is a composite character involving serration size and root constriction. Only serration size is coded for here. Yangchuanosaurus (Dong et al., 1983) has small serrations.

21. "Neural spine low and thin on cervicals (0), powerfully constructed (1)" is a composite character involving neural spine height and width. Only height is coded for here. Baryonyx (Charig and Milner, 1997) and Yangchuanosaurus (Dong et al., 1983) have tall neural spines.

22. Baryonyx (Charig and Milner, 1997) has large cervical epipophysis. Troodontids are polymorphic, as IGM 100/44 (Barsbold et al., 1987) and Saurornithoides (Norell and Hwang, 2004) have large epipophyses.

23. Oviraptorosaurs have elongate cervical centra (Microvenator- Makovicky and Sues, 1998; Chirostenotes- Sues, 1997).

24. Oviraptorosaurs lack opisthocoelous cervical centra (Microvenator- Makovicky and Sues, 1998; Chirostenotes- Sues, 1997).

25. "Shaft of cervical ribs slender and usually moderately long (0), broad and short (1)" is a composite character involving both rib length and width. Only length is coded for here. Baryonyx (Charig and Milner, 1997), ornithomimosaurs (Harpymimus- Kobayashi and Barsbold, 2005; Gallimimus- Osmolska et al., 1972; Struthiomimus- Osborn, 1916) and oviraptorosaurs (Chirostenotes- Sues, 1997; IGM 100/42- unpublished photos; Rinchenia and "Ingenia"- both Lu, 2004) have short cervical ribs.

26."Capitular facet lies on anteroventral lamina from transverse process in dorsal vertebrae (0), situated above lamina, on prezygapophyseal base (1)" can vary throughout the dorsal column, with high parapophyses unknown prior to dorsal four in any taxon coded by the authors. It is only coded here for posterior dorsals, excluding the last which can lack a rib or be partially sacralized. Baryonyx (Charig and Milner, 1997) has dorsal parapophyses placed ventral to prezygapophyses. Coelophysis (Colbert, 1989) has parapophyses placed high on posterior dorsals. Oviraptorosaurs are polymorphic, with Microvenator having low parapophyses (Makovicky and Sues, 1998; pers. obs.) and IGM 100/42 having high ones (unpublished photos).

27. Baryonyx (Charig and Milner, 1997), Allosaurus (Gilmore, 1920), ornithomimosaurs (Harpymimus and Garudimimus- Kobayashi and Barsbold, 2005a and b) lack posterior dorsal pleurocoels.

28. Ornithomimosaurs (Harpymimus- coding in Kobayashi, 2004; Garudimimus- Kobayashi, 2004; Archaeornithomimus- Makovicky, 1995) lack sacral pleurocoels. Dromaeosaurids (Saurornitholestes- Makovicky, 1995; Velociraptor- Norell and Makovicky, 1999; Adasaurus- Kubota and Barsbold, 2006) have them.

29. Tyrannosaurids are polymorphic (Gorgosaurus lacks accessory neural spines- Lambe, 1917). Yangchuanosaurus (Dong et al., 1983) does not preserve enough mid caudals to be sure they lacked accessory neural spines.

30. Plateosaurus (Huene, 1926), Allosaurus (Madsen, 1976), tyrannosaurids (Gorgosaurus- Lambe, 1917), ornithomimosaurs (Gallimimus- Osmolska et al., 1972) and oviraptorosaurs (IGM 100/42- unpublished photo) lack neural spines on their distalmost caudals. Therizinosauroids used by Russell and Dong cannot be coded, as none preserve distalmost caudals.

31. Oviraptorosaurs (Rinchenia and "Ingenia"- both Barsbold et al., 2000; IGM 100/42- unpublished photos) have transverse processes on more than 40% of their caudals. Tyrannosaurids (Gorgosaurus- Makovicky, 1995; Lambe, 1917) have transverse processes on less than 40% of their caudals. Ornithomimids are polymorphic, different individuals of Gallimimus straddling the boundary (Osmolska et al., 1972). Because exact caudal count in Alxasaurus (Russell and Dong, 1994) and Sinornithoides (Currie and Dong, 2001) is unknown and estimated counts lead to ~41-43% in both, therizinosauroids and troodontids cannot be coded.

32. Baryonyx (Charig and Milner, 1997) lacks basal chevron processes. Allosaurus (Gilmore, 1920), ornithomimosaurs (Harpymimus- Kobayashi and Barsbold, 2005; Gallimimus- Osmolska et al., 1972), oviraptorosaurs (IGM 100/42, Rinchenia and "Ingenia"- unpublished photos) and troodontids (Sinornithoides- Dong and Currie, 2001; Saurornithoides- Norell et al., 2009; Zanabazar- Barsbold, 1974) have these processes.

33. This character is described as distal chevrons "simple and curving in profile" vs. "much reduced", so is here defined as having distal chevrons with anteroventral processes. Allosaurus (Madsen, 1976) and oviraptorosaurs (IGM 100/42- unpublished photo) have derived distal chevrons. No included therizinosauroid has preserved distal chevrons.

34. This character compares "anteroposteriorly elongate to subcircular" to "subrectangular" coracoids, so is defined here as the proxomodistal length. Therizinosauroids are polymorphic (proximodistally elongate in Segnosaurus- Barsbold and Perle, 1980, and Therizinosaurus- Barsbold, 1979).

35. Another composite character, involving both the triangular vs. rectangular shape and differentiation of the ventral tuber. The latter is used here (differentiated by the capital groove), as Alxasaurus and Segnosaurus have triangular tubers.

36. Baryonyx (Charig and Milner, 1997) and Yangchuanosaurus (Dong et al., 1983) have an ulnar condyle smaller than the radial condyle. Coelophysis (Colbert, 1989), Allosaurus (Gilmore, 1920) and tyrannosaurids (Gorgosaurus- Lambe, 1917; Albertosaurus- Parks, 1928; Tyrannosaurus- Brochu, 2003) have a larger ulnar condyle. Ornithomimosaurs (larger ulnar condyle in Archaeornithomimus- pers. obs. AMNH 6570), therizinosauroids (smaller ulnar condyle in Alxasaurus- Russell and Dong, 1993 and Erlikosaurus- Perle, 1981) and oviraptorosaurs (smaller ulnar condyle in "Ingenia"- unpubloished photo) are polymorphic. The condition in included troodontids (Sinornithoides, Troodon) has not been described.

37. Baryonyx (Charig and Milner, 1997) and Yangchuanosaurus (Dong et al., 1983) lack a posteriorly bowed ulna.

38. Distal carpal I overlaps metacarpal II in dorsal view in Plateosaurus (Huene, 1932), Allosaurus (Chure, 2001), tyrannosaurids (Albertosaurus- Holtz, 2001; Daspletosaurus- Russell, 1970), ornithomimosaurs (Harpymimus- Kobayashi and Barsbold, 2005; Gallimimus- Osmolska et al., 1972) and therizinosauroids (Alxasaurus- Russell and Dong, 1994; Therizinosaurus- Barsbold, 1976). It is innapplicable in oviraptorosaurs, troodontids and dromaeosaurids due to complete fusion between distal carpals I and II.

39. As this character codes for the length of "digit II (metacarpals with phalanges) and (or) digit II (phalanges only)", it needs to be specified to cover only one of those options. The latter is chosen here since no included taxon has digit II shorter when metacarpals are included. Yet this is again problematic as Plateosaurus and Coelophysis are coded 0 due to having digit II supposedly shorter than III, but ornithomimosaurs are coded 0 due to having digit II supposedly shorter than digit I (actually untrue even in Dromiceiomimus and Anserimimus). Thus the character is specified as "manual digit II (phalanges only, including unguals) longer than digit III". Plateosaurus (Huene, 1926; same if metacarpals are added) and ornithomimosaurs (Harpymimus- Kobayashi and Barsbold, 2005; Gallimimus- Osmolska et al., 1972) have digit II longer than III. No included therizinosauroid has known digit II and III length, so it is recoded unknown.

40. This is here specified to be metacarpal III 90% or more length of metacarpal II, to best match Russell and Dong's codings. This is partly correlated with character 39. This is true in Coelophysis (Colbert, 1989) and polymorphic in dromaeosaurids (long mcIII in Deinonychus- Ostrom, 1969).

41. This is here specified to be metacarpal I <40% of metacarpal II length, to best match Russell and Dong's codings. Yangchuanosaurus (Dong et al., 1983), Oviraptorosaurs (Chirostenotes- Currie, 1990; Elmisaurus- Osmolska, 1981; IGM 100/42 and "Ingenia"- Barsbold et al., 1990; Conchoraptor- Barsbold, 1986) have long metacarpals I. Dromaeosaurids are polymorphic (short in Velociraptor- Norell and Makovicky, 1999).

42. Dromaeosaurids are polymorphic (unbowed metacarpal III in Velociraptor- Norell and Makovicky, 1999; Saurornitholestes- coding in Turner, 2008).

43. Tyrannosaurids lack a triangular proximal outline to metacarpal III (Tyrannosaurus- Lipkin and Carpenter, 2008). Troodontids (coding in Rauhut, 2003 from either Sinornithoides or Troodon) have triangular outlines. The condition of a rectangular outline as coded for Coelophysis cannot be confirmed from the literature and is perhaps unlikely given Liliensternus already has a triangular outline (Huene, 1934). It is provisionally recoded as uncertain here.

44. Allosaurus (lip on ungual III- Gilmore, 1920) and troodontids (lip on ungual II in Sinornithoides- Currie and Dong, 2001) are polymorphic.

45. Oviraptorosaurs (Chirostenotes- Currie and Russell, 1988; Rinchenia- unpublished photo; IGM 100/42- unpublished photos) lack dorsoventrally expanded preacetabular processes. Therizinosauroids are polymorphic (Alxasaurus lacks a tall preacetabular process).

46. Baryonyx (Charig and Milner, 1997), therizinosauroids (Nanshiungosaurus- Dong, 1979; Segnosaurus- Zanno, 2010) and troodontids (Troodon- Varricchio et al., 2002) lack a vertical ridge above the acetabulum.

47. Baryonyx (Charig and Milner, 1997), troodontids (Sinornithoides- Russell and Dong, 1994; Troodon- Varricchio et al., 2002) and dromaeosaurids (Saurornitholestes- Hutchinson, 2001; Velociraptor- Norell and Makovicky, 1997; Deinonychus- Ostrom, 1969, 1976) have tapered postacetabular processes. Yangchuanosaurus (Y. shangyouensuis holotype has a tapered process) and therizinosauroids (squared in Segnosaurus- Zanno, 2010) are polymorphic.

48. Tyrannosaurids (Gorgosaurus- Lambe, 1917; Tyrannosaurus- Osborn, 1916) and troodontids (Sinornithoides- Novas, 2004) lack supracetabular crests.

49. Dromaeosaurids (Velociraptor- Norell and Makovicky, 1997; Adasaurus and Utahraptor- coding in Turner, 2008; Deinonychus- Novas, 1996) lack deep brevis fossae.

51. Troodontids (Troodon has a large anterior pubic boot- unpublished illustration) and dromaeosaurids (Deinonychus has a large anterior pubic boot- Ostrom, 1976) are polymorphic.

52. Baryonyx (Charig and Milner, 1997) has a distally expanded ischium.

53. Dromaeosaurids are polymorphic (Utahraptor has a proximally placed obturator process- Britt et al., 2001).

54. Troodontids (Saurornithoides- pers. obs. AMNH 6516; Troodon- Varricchio et al., 2002) and dromaeosaurids (Velociraptor- Norell and Makovicky, 1999; Deinonychus- Ostrom, 1976; Utahraptor- Britt et al., 2001; Adasaurus- coding in Turner, 2008) have cylindrical anterior trochanters. Plateosaurus (Huene, 1926) lacks a projecting anterior trochanter, so is coded inapplicable. Therizinosauroids (Alxasaurus- Russell and Dong, 1994; Segnosaurus- Zanno, 2010) have a completely fused trochanteric crest, so are inapplicable.

55. Coelophysis (Spielmann et al., 2007) has a posteriorly extensive lateral tibial condyle. Dromaeosaurids (Velociraptor- Norell and Makovicky, 1999) have an anteriorly restricted condyle.

56. As the astragalar ascending processes of oviraptorids and troodontids aren't laterally curved, this is restricted to only code for process height. Tyrannosaurids (Gorgosaurus- Lambe, 1917; Albertosaurus- Lambe, 1904; Tarbosaurus- Maleev, 1974), ornithomimosaurs (Harpymimus- Kobayashi and Barsbold, 2005; Garudimimus- Kobayashi and Barsbold, 2005; Archaeornithomimus- pers. obs. AMNH 6570) and dromaeosaurids (Velociraptor- Norell and Makovicky, 1999; Deinonychus- Ostrom, 1969; Utahraptor- Kirkland et al., 1993) have a tall process.

57. Baryonyx (Charig and Milner, 1997) lacks a strongly convex astragalar articular surface of the calcaneum. Allosaurus (Chure, 2000; Gilmore, 1920) and ornithomimosaurs (Gallimimus- Osmolska et al., 1972) have a well developed concavoconvex articulation. Oviraptorosaurs are polymorphic (shallow astragalar concavity for calcaneum in Microvenator- Makovicky and Sues, 1998). Troodon can have unfused astragalocalcanea (Zanno et al., 2011), so troodontids recoded unknown instead of inapplicable. Yangchuanosaurus (Dong et al., 1983) does not expose the astragalocalcanear articular surfaces.

58. Yangchuanosaurus (Dong et al., 1983) lacks an arctometatarsus. Oviraptorosaurs are polymorphic (no arctometatarsus in IGM 100/42, Conchoraptor and "Ingenia"- Barsbold et al., 1990).

59. Baryonyx (Charig and Milner, 1997) and Yangchuanosaurus (Dong et al., 1983) have pedal unguals which are triangular in section. Oviraptorosaurs (uncompressed in Elmisaurus- Osmolska, 1981) and Dromaeosaurids are polymorphic (transversely compressed oval unguals in Adasaurus, Velociraptor and Utahraptor- coding in Senter, 2010).

General analysis conclusions- This analysis is quite small in both character and taxon number. Character-wise it lacks numerous classic examples used to justify traditional clades (e.g. reduced metatarsal I for theropods; sickle claw for deinonychosaurs; enlarged semilunate carpal for maniraptorans) and taxon-wise lacks several standard taxa (Ceratosaurus as a large tetanurine outgroup, Compsognathus or Ornitholestes as basal coelurosaurs, birds to determine which clade is technically Coelurosauria). These may explain the heterodox topology recovered. In part because of the dearth of characters, only 7% (4/59) were correlated, though 36% (21/59) were composite. 25% of the codings (162/649) were wrong, largely due to Baryonyx and oviraptorosaurs being poorly described at the time, many polymorphies (none of which were included by Russell and Dong) and different portions of the composite characters being coded for different taxa. Once corrected, the consensus tree is-

      |  |--Allosaurus
      |  `--+--+--Baryonyx
      |     |  `--Yangchuanosaurus
      |     `--+--Tyrannosauridae
      |        `--Dromaeosauridae

The only differences are no resolution in Maniraptora, and a different topology for basal carnosaurs.

Phylogenetic conclusions- The table shows the number of extra steps needed to accomodate each rearrangement using Russell and Dong's original matrix, and their recoded matrix.

rearrangement original recoded
(Therizinosauroidea,Plateosaurus(Coelophysis,Ornithomimosauria,Troodontidae,Oviraptorosauria,Allosaurus)) (Paul, 1984) 20 14
(Plateosaurus,Yangchuanosaurus,Allosaurus,Troodontidae(Coelophysis,Baryonyx)) (Paul, 1988) 3 3
(Plateosaurus,Baryonyx(Allosaurus,Ornithomimosauria)) (Paul, 1988) 3 3
(Plateosaurus,Baryonyx,Tyrannosauridae(Yangchuanosaurus,Allosaurus)) (Currie and Zhao, 1994) 3 3
(Plateosaurus,Baryonyx,Yangchuanosaurus,Troodontidae,Dromaeosauridae(Allosaurus,Tyrannosauridae)) (Paul, 1988) 5 4
(Plateosaurus,Ornithomimosauria,Tyrannosauridae,Dromaeosauridae(Baryonyx,Yangchuanosaurus,Allosaurus)) (Rauhut, 2000) 3 2
(Plateosaurus,Allosaurus,Yangchuanosaurus,Ornithomimosauria(Tyrannosauridae,Oviraptorosauria,Troodontidae,Dromaeosauridae)) (Sereno, 1999) 9 2
(Plateosaurus,Allosaurus,Yangchuanosaurus(Tyrannosauridae,Ornithomimosauria)) (Holtz, 1994) 2 3
(Plateosaurus,Tyrannosauridae(Ornithomimosauria,Dromaeosauridae)) (Gauthier, 1986) 6 2
(Plateosaurus,Tyrannosauridae,Ornithomimosauria(Therizinosauroidea,Oviraptorosauria,Troodontidae,Dromaeosauridae)) (Sues, 1997) 8 2
(Plateosaurus,Tyrannosauridae,Oviraptorosauria,Troodontidae(Ornithomimosauria,Therizinosauroidea)) (Sereno, 1999) 13 4
(Plateosaurus,Ornithomimosauria,Troodontidae(Oviraptorosauria,Therizinosauroidea)) (Sues, 1997) 4 0
(Plateosaurus,Oviraptorosauria,Therizinosauroidea(Ornithomimosauria,Troodontidae)) (Holtz, 1994) 9 6
(Plateosaurus,Tyrannosauridae,Ornithomimosauria,Therizinosauroidea(Oviraptorosauria,Troodontidae,Dromaeosauridae)) (Sereno, 1999) 12 2
(Plateosaurus,Tyrannosauridae,Ornithomimosauria,Therizinosauroidea,Oviraptorosauria(Troodontidae,Dromaeosauridae)) (Ostrom, 1969) 9 2

Although this was the first modern paper to support theropod 'segnosaurs', it does a rather poor job of testing this cladistically as it does not include several classic theropod (e.g. three manual digits; long preacetabular process) and non-theropod (e.g. little dorsal exposure of lacrimal; metatarsal I contacts tarsus) characters. Regardless, both original and recoded versions find non-avepod therizinosaurs to be unlikely. Besides this, neither version strongly rejects any non-coelurosaur arrangement, showing a lack of applicable characters to this part of the tree. Placing tyrannosaurids in Coelurosauria is similarly easy in both versions, though they stick with dromaeosaurids in the original matrix. While excluding tyrannosaurids from Maniraptoriformes is somewhat possible in the original matrix, enforcing traditional Maniraptora, Chuniaoae and Paraves are all strongly rejected. Yet when recoded, these all only take 2 steps, showing coding errors are responsible for the strongly supported heterodox topology of separating dromaeosaurids from other maniraptoriforms. Somewhat surprisingly, Sereno's ornithomimosaur-therizinosaur clade is strongly rejected by the original matrix, though it is only weakly rejected when recoded. Bullatosauria is strongly rejected by the original matrix, and still moderately rejected when recoded. Finally, Enigmosauria is weakly rejected in the original matrix and in one of the MPTs when recoded. The basic conclusion is that the original matrix strongly rejects most alternative coelurosaur topologies but only due to miscodings, and that properly coded the matrix is too small to reject almost all suggested alternatives except Bullatosauria.

Experiments with controversial taxa- One issue is that since Plateosaurus is the outgroup, it doesn't have a chance to clade with therizinosauroids as e.g. Sereno (1989) proposed. Since Paul (1984) advocated for a Phytodinosauria including these and ornithischians, merely adding the latter isn't a fair test of this alternative hypothesis either. Thus basal ornithischian Heterodontosaurus was added, in addition to Silesaurus and Marasuchus, the latter of which is definitely outside the therizinosauroid-theropod clade. To test whether the 'carnosaur ecotype' was responsible for maintaining a monophyletic Carnosauria, the more basal taxa Herrerasaurus and Ceratosaurus were added. Basal coelurosaurs Compsognathus and Ornitholestes were added to better associate the clade with its relatives. Basal tyrannosauroid Guanlong, basal ornithomimosaur Pelecanimimus, basal oviraptorosaur OTU Incisivosaurus+Protarchaeopteryx, basal troodontid Sinovenator and basal dromaeosaurid Sinornithosaurus were all added to better test the relationships of those groups. Archaeopteryx was added to place birds and define Coelurosauria. Finally, the basal therizinosaurs Falcarius and Jianchangosaurus were added to answer the analysis' main question of therizinosauroid affinities. When ran, 54 trees of 177 steps are found, the consensus of which is-

   |  |--Herrerasaurus
   |  `--Heterodontosaurus
   |  |--Plateosaurus
   |  `--Theropoda
   |     |--Coelophysis
   |     `--+--Baryonyx
   |        `--+--Ceratosaurus
   |           `--+--Tyrannosauridae
   |              |--Ornitholestes
   |              `--Allosauroidea
   |                 |--Yangchuanosaurus
   |                 `--Allosaurus
      |  |--Compsognathus
      |  `--Pelecanimimus
               |  |--Oviraptorosauria
               |  `--Therizinosauria
               |     |--Falcarius
               |     `--+--Jianchangosaurus
               |        `--Therizinosauroidea
                  |  |--Archaeopteryx
                  |  `--Troodontidae

The basal arrangement is either-


|  |--Herrerasaurus
|  `--Heterodontosaurus


      |  |--Herrerasaurus
      |  `--Heterodontosaurus

Protarchaeopteryx/Incisivosaurus is excluded from Therizinosauria and Avialae.

The lack of basal consensus is more proof this matrix lacks characters to distinguish theropods from non-theropods, though therizinosaurs continue to clade with coelurosaurs instead of sauropodomorphs or ornithishians. Interestingly, Ceratosaurus emerges as a carnosaur (as in Currie, 1995), and Ornitholestes groups with allosaurids and tyrannosaurids (as in Paul, 1988). Allosauroidea matches the modern consensus, as does the position of dromaeosaurids. Enigmosauria is favored, continuing the trend in moving therizinosaurs away from troodontids as the matrix is improved. Unfortunately, the low character count was unable to place Guanlong with tyrannosaurids, Pelecanimimus with other ornithomimosaurs, Incisivosaurus/Protarchaeopteryx definitely with caenagnathoids or Sinovenator with other troodontids. Archaeopteryx's placement does confirm Russell and Dong's Oviraptorosauria sensu lato corresponds to Coelurosauria though.