Wednesday, May 16, 2018

Testing alternative stem bird topologies in Cau, 2018

In the comments to my last post describing Cau's (2018) new paper detailing the acquisition of characters on the line to Aves, reader AOF requested a post similar to what I did six years ago with the Carrano et al. (2012) tetanurine analysis.  Namely, testing alternative topologies using constraint analyses to see how many more steps they would require.  I think these kinds of things can be illuminating.  I've often said that we shouldn't think of a new cladogram as just 'the best new hypothesis', but rather check individual components of the tree to see how likely or unlikely they are to be correct.  Cau's 2018 matrix has a reduced taxon sample, which could easily change the number of steps compared to a complete sample.  On the other hand, I think Andrea tries to include all proposed characters in his MegaMatrix, which could make this a more honest measure of comparative topology length than most studies.  I'm not sure which variable overrides the other.  Then we have score correctness, which I've never checked in a Cau matrix, so I'm taking that at face value.

Andrea sent me his NEXUS file, but whereas the paper reports 3072 MPTs of 6790 steps, I found 10872 MPTs of that length.   Eoraptor and Buriolestes aren't always sauropodomorphs, Pisanosaurus is sometimes an ornithischian, Asilisaurus, Silesaurus and Sacisaurus are an unresolved trichotomy, Enantiornithes can be paraphyletic with Zhongjianornis among them, and Fake Ornithuromorpha is less resolved, as Patagopteryx and Apsaravis can be outside Hongshanornis+Aves, Archaeornithura can be a songlingornithid, etc..  I think Andrea's philosophy would be that these things vary with taxon inclusion, so aren't a definite part of his data.  The absence of any included spinosaurids, carcharodontosaurines, parvicursorines or Avimimus unfortunately makes some weird 80s and 90s hypotheses untestable.  After six more years of experience, I've added a new category "less likely" because I think that factors like scoring accuracy and taxon inclusion can have a larger influence.  As I said in the 2012 post, the corrected TWG matrix needed 15 more steps to get a monophyletic Deinonychosauria which is the most common outcome for TWG matrices today.  I'll say the Lori matrix recovers at least one hypothesis found to be "unlikely" here, so even that's not the kiss of death.

Basically ambiguous

0 steps- Maniraptoromorph Compsognathus.

0 steps- Megaraptoran Gualicho.

1 step- Ceratosaurus closer to abelisauroids than Elaphrosaurus.

1 step- Megalosauroid piatnitzkysaurids.

1 step- Coelurosaurian Zuolong.  Ends up as a basal maniraptoromorph between Coelurus and Ornitholestes.

1 step- Coelurid Aorun, as in its original description.  It moves to Maniraptoromorpha with Coelurus.

1 step- Coelurid Tanycolagreus, as in its original description.  Coelurus moves into Tyrannosauroidea with Tanycolagreus.

1 step- Compsognathid Sinosauropteryx and/or Sinocalliopteryx.

1 step- Tyrannosauroid Sinocalliopteryx or Coelurus.

1 step- Anchiornithid or archaeopterygid Xiaotingia.

1 step- Scansoriopterygids closer to Aves than Archaeopteryx.

1 step- Sapeornis closer to Aves than Confuciusornis.

1 step- Zhongjianornis sister to Pygostylia, as in its original description.

1 step- Ichthyornis closer to Aves than Hesperornis, the consensus until the recent cranial redescription of Ichthyornis.

2 steps- Theropodan Eodromaeus.

2 steps- Chilesaurus just outside Avepoda, which was my best guess back in 2015 considering the results of its original misscored matrices and my subjective feelings of what would plausibly reverse.

2 steps- Ceratosaurian Gualicho.  It has an uncertain placement within the clade, though is excluded from Abelisauria.

2 steps- Megaraptora as coelurosaurs just outside of Tyrannoraptora.  This was my result back in 2010 after adding Benson's Neovenatoridae data to my theropod supermatrix (since superseded by the Lori analysis).  Bicentenaria is at this level too, while Guanlong and Tanycolagreus become maniraptoromorphs.

2 steps- Maniraptoran Ornitholestes.

2 steps- Hongshanornithid Parahongshanornis, as in its original description.

Cast YPM 56693 of the Mononykus olecranus holotype pes in plantar view, courtesy of Senter.

Quite likely to be true

3 steps- Saurischia.  Herrerasaurs and Eodromaeus are theropods.  Note that while some of these constraints were tested in the Ornithoscelida paper, the studies differ in both taxonomic content and characters used, so that MegaMatrix results don't necessarily correlate with Ornithoscelida paper results and should not be seen as scooping anything we find. 

3 steps- Ornithischian Daemonosaurus. Chilesaurus is sister to Averostra.

3 steps- Theropodan Herrerasaurus/Tawa/Daemonosaurus.  These each take three steps more, and other herrerasaurs follow when one is constrained.

3 steps- Dilophosaurid Liliensternus, as in Paul (1988).

3 steps- Ceratosauria sensu lato, combining Neoceratosauria and Coelophysoidea.  Chilesaurus is the most basal theropod, Elaphrosaurus plus Limusaurus are the basal neoceratosaurs, and Gualicho is the most basal tetanurine.

3 steps- Eustreptospondylus and/or Megalosaurus outside Avetheropoda.  If one is constrained, the other follows.

3 steps- Megaraptoran Eotyrannus, as in Novas et al. (2013).

3 steps- Maniraptoran Coelurus.

3 steps- Troodontid Aurornis, Anchiornis and/or Xiaotingia.

3 steps- Chongmingia sister to Ornithothoraces, as in p2 of its original description. 

4 steps- Ornithischian silesaurids.  Lewisuchus is outside Dinosauria, Saurischia exists, and Asilisaurus and Pisanosaurus form a silesaur grade to either side of Silesauridae.

4 steps- Phytodinosauria.  Eoraptor and Buriolestes are sister to Dinosauria, while herrerasaurs and Eodromaeus are theropods.

4 steps- Ornithischian Chilesaurus.  Ornithoscelida occurs.

4 steps- Theropodan EoraptorBuriolestes, herrerasaurs and Eodromaeus also theropods.

4 steps- Abelisaurid Eoabelisaurus, as in its original description.

4 steps- Metriacanthosaurids outside Allosauria (Allosaurus plus Carcharodontosaurus).  Acrocanthosaurus joins Neovenator, so this also covers carcharodontosaurid Neovenator.

4 steps- Tyrannosauroid Compsognathus, as in Olshevsky (1991).  Surprised this one is so parsimonious.  A Compsognathidae with Aorun, Bicentenaria and Sinosauropteryx are the basalmost tyrannosauroids, with Sinocalliopteryx, Coeluridae including Tanycolagreus and Guanlong successively closer to core tyrannosauroids.

4 steps- Maniraptoromorph Tanycolagreus and/or Guanlong.

4 steps- Ornithomimosaurian Gualicho.  Tested due to Rauhut (2003) finding the very similar Deltadromeus in this position.  Note that ornithomimosaurian Elaphrosaurus is 36 steps longer, so the cases aren't that similar.

4 steps- Therizinosaurian Jianchangosaurus.  Still outside Falcarius plus Beipiaosaurus, and Cau said in a comment to the last post Jianchangosaurus was still an alvarezsauroid even after adding Erlikosaurus and Shuvuuia.  I'm not revealing much by saying the Lori analysis finds Jianchangosaurus to be a therizinosaur between Falcarius and Beipiaosaurus as in its original description.  Seems fishy...

4 steps- Archaeopterygid Jinfengopteryx, as in its original description.

4 steps- Archaeopterygid Anchiornis.

4 steps- Oviraptorosaurian scansoriopterygids.  They have an uncertain position within the clade, and therizinosaurs are still sister to oviraptorosaurs.

Allosaurus fragilis holotype tooth (YPM 1930) in ?lingual view.  Courtesy of the YPM.

5 steps- Dilophosaurus/Cryolophosaurus closer to Averostra than Coelophysis.

5 steps- Non-avetheropod Compsognathus, as in Novas (1992).  Very surprised this is so easy to get.  It's even more extreme than Novas' version, where Compsognathus was at least closer to avetheropods (his Tetanurae) than Piatnitzkysaurus and Eustreptospondylus, because in the constrained trees Carnosauria still has the same content as Cau's MPTs.  Sinosauropteryx joins it.

5 steps- Maniraptoromorph Eotyrannus.  I'm surprised by this, since I figured the result in other matrices was due to a lack of tyrannosauroid characters, which I think are all in the MegaMatrix.

5 steps- Non-tyrannoraptoran Coelurus, as in Paul (1988).

5 steps- Avemetatarsalian alvarezsauroids, as in Sereno (1999).  Jianchangosaurus is still a basal alvarezsauroid.

5 steps- Haplocheirus compsognathid/coelurid grade, as in Alifanov and Saveliev (2011).  Wasn't there some analysis that recovered it here too?  I just constrained it to be outside Maniraptoriformes (including alvarezsaurids).

5 steps- Non-pennaraptoran therizinosaurs, which fall out sister to Pennaraptora like the current consensus.

5 steps- Chongmingia a basal ornithurine (sensu Gauthier) outside Shenzhouraptor and Pygostylia, as in p1 of its original description.  The Lori analysis recovers it in a different position than  p1, p2 or Cau's analysis.

Less likely

6 steps- Non-eusaurischian saurischian EoraptorBuriolestes follows, but herrerasaurs and Eodromaeus are theropods.

6 steps- Theropodan Guaibasaurus.  Non-dinosaurian Eoraptor and Buriolestes, and this recovers Phytodinosauria.

6 steps- Eustreptospondylus closer to Neotetanurae than Megalosaurus, as in Holtz (2000).  I'm actually surprised this is so unlikely.

6 steps- Carnosaurian Sinosauropteryx, as in Longrich (2002).  Though Longrich's phylogeny was a bit different in having megalosaurids and metriacanthosaurids outside Avetheropoda.

6 steps- Ornithomimosaurian Haplocheirus, as in the Bayesian analyses of Cau and Lee and Worthy (2011).

6 steps- Arctometatarsalian therizinosaurs, as in Sereno (1999).

6 steps- Shenzhouraptor closer to Aves than Sapeornis.  

7 steps- Sauropodomorph Staurikosaurus but not Herrerasaurus, as in pachypodosaur Staurikosaurus of Kischlat (2000).

7 steps- Megalosauroid Monolophosaurus.  Megalosauroidea remains in Carnosauria.

7 steps- Coelurosaurian Neovenator.  Not sure if this has been suggested in print before, but I noticed quite a few coelurosaur-like characters when scoring Neovenator for the Lori matrix.  It forms the most basal coelurosaur clade with Aorun and Gualicho.

7 steps- Maniraptoran Compsognathus.

7 steps- Paravian alvarezsaurids, though note the lack of parvicursorines probably affects these numbers.  They (including Jianchangosaurus and Haplocheirus) emerge as the most basal paravians.

7 steps- Basal paravian Anchiornis, Aurornis, scansoriopterygids, Serikornis and/or Xiaotingia.

7 steps- Archaeopterygid Rahonavis, as in Forster et al. (1998).

7 steps- Fake-Ornithuromorphan Confuciusornis, as in Kurochkin (2006).  I really thought this would be more difficult to achieve than enantiornithine Confuciusornis (below).

Box of Archaeornithimimus asiaticus elements from AMNH 6576, with my identifications (dc- distal caudal, dt- distal tarsal, pedal except m 1-1 and m 3-2).  Is that a proximal metatarsal I in the upper left?  Courtesy of the AMNH.

8 steps- Non-eusaurischian saurischian HerrerasaurusEodromaeus and sometimes Eoraptor become herrerasaurs and Buriolestes is one node more stemward.

8 steps- Tetanurine Cryolophosaurus, as in Carrano et al. (2002).  Dilophosaurus stays in Coelophysoidea.

8 steps- Ceratosaurian megalosaurids, as in Britt (1991).  Chilesaurus falls out in a polytomy with megalosaurids and other ceratosaurs.

8 steps- Megalosauroid piatnitzkysaurids, with Megalosauroidea outside Avetheropoda.  Since this is the Carrano et al. consensus, I thought it would take less steps.

8 steps- Monolophosaurus sister to Avetheropoda, as in Smith et al. (2007).  Megalosaurids and piatnitzkysaurids fall out as more basal tetanurines.

8 steps- Arctometatarsalian tyrannosauroids, AKA Tyrannosaurus closer to Ornithomimus than to birds as in Holtz (1994).  I'm very surprised this is so parsimonious.  Coelurus and Bicentenaria join Tyrannosauroidea, but Gualicho leaves to be a ceratosaur.

8 steps- Maniraptoran tyrannosauroids, as in Sereno (1999).  This is accomplished more by moving ornithomimosaurs (including Gualicho) stemward to be the most basal coelurosaurs except for Zuolong

Somewhat possible

9 steps- Classic late 80s to early 90s topology where Staurikosaurus is sister to Herrerasaurus plus Dinosauria.  Tawa plus Daemonosaurus are closer to dinosaurs than both, while Sanjuansaurus follows Herrerasaurus.

9 steps- Piatnitzkysaurus outside Orionides, as in Rauhut (2003).  Condorraptor follows Piatnitzkysaurus, and megalosauroids fall outside Avetheropoda.  Surprised this is so high.

9 steps- Carnosaurian Tyrannosaurus, which brings megaraptorans, Gualicho and Bicentenaria to form the sister group of Allosauroidea (including Monolophosaurus).  I bet this is more parsimonious than most readers would assume given published topologies over the past two decades.

9 steps- Avialan Caudipteryx, as in its original description.  The rest of Oviraptorosauria follows it, though troodontids are still closer to Aves.

9 steps- Avialan Microraptor, as in Agnolin and Novas (2013).  Weirdly becomes the most basal troodontid, with that family closer to Aves than scansoriopterygids and anchiornithids.

9 steps- Avialan Unenlagia, as in its original description and Agnolin and Novas (2013).  Halszkaraptorines are unenlagiids, which are outside the Troodontidae plus Ornithes clade.

9 steps- Deinonychosauria.  Scansoriopterygids are oviraptorosaurs, while Jinfengopteryx and anchiornithids are avialans.

9 steps- Archaeopterygidae sister to Troodontidae.  Anchiornithines fall out as archaeopterygids. 

10 steps- Monolophosaurus outside Orionides, as in Carrano et al. (2002).  Megalosaurids and piatnitzkysaurids form successively closer outgroups to Avetheropoda.

10 steps- Fukuivenator excluded from Alvarezsauridae plus Therizinosauria plus Pennaraptora as in its original description.  It emerges as the sister to other maniraptorans.

10 steps- Eumaniraptora excluding troodontids as in Agnolin and Novas (2013).  Scansoriopterygids are oviraptorosaurs.

10 steps- Dromaeosaurid Xiaotingia, as in Senter et al. (2012).  Falls out in Microraptoria.

10 steps- Dromaeosaurid Balaur, as in its original description.  Falls out sister to Unenalagiinae plus Halszkaraptorinae.

11 steps- Tyrannosauroid Acrocanthosaurus, as in Bakker et al. (1988).  Tyrannosauroids become carnosaurs, with Sinraptor, Acrocanthosaurus and Bicentenaria successively closer to the 'core tyrannosauroid' clade of Eotyrannus, Gualicho, megaraptorans and TyrannosaurusTanycolagreus and Guanlong are now maniraptoromorphs.

11 steps- Alvarezsauroid Nqwebasaurus.  Alvarezsauroids emerge sister to ornithomimosaurs, with Haplocheirus and Jianchangosaurus forming a basal [edit] arctometatarsalian clade.

11 steps- Mahakala outside Unenlagiinae plus Eudromaeosauria (Halszkaraptor follows), as in most TWG matrices (though Senter et al. 2012 recovered it sister to unenlagiines like Cau).


12 steps- Alvarezsauroid Chilesaurus, where it emerged in the Lori matrix back in 2015.

12 steps- Ornithuran (sensu Gauthier) oviraptorosaurs, as in Maryanska et al. (2002).  Constraining Khaan to be closer to Meleagris than Archaeopteryx results in oviraptorosaurs (including scansoriopterygids) being the first clade to diverge from the avian stem after Archaeopteryx.

12 steps- Basal paravian Jinfengopteryx, as in Foth et al. (2014). 

13 steps- Coelophysoid Elaphrosaurus, as in Paul (1988).  Ceratosauria sensu lato forms, Elaphrosaurus is outside core coelophysoids and Limusaurus and sometimes Gualicho follow.

13 steps- Compsognathid Nqwebasaurus, as in Novas et al. (2013).  Compsognathids (including Aorun) become ornithomimosaurs.

13 steps- Dromaeosaurid Rahonavis.  Emerges in the unenalgiine plus halszkaraptorine clade.

14 steps- Neovenatorid megaraptorans, though this actually moves Neovenator out of Carnosauria into Tyrannosauroidea, so isn't that similar to Benson's topology.

14 steps- Enantiornithine Confuciusornis.

15 steps- Sauropodomorphan Chilesaurus, where it emerges as the most basal member.

15 steps- Carnosaurian Ceratosaurus, as in Currie (1995).  Other ceratosaurs follow.

15 steps- Alvarezsaurids closer to Aves than dromaeosaurids or troodontids.  Haplocheirus and Jianchangosaurus remain behind as ornithomimosaurs.  Note the lack of parvicursorines probably affects this number.

15 steps- Archaeopterygid Unenlagia, as in Forster et al. (1998).  Buitreraptor remains in Dromaeosauridae.

Cladogram of archosauromorphs after Kischlat (2000).  Note saurischian Marasuchus and sauropodomorph Staurikosaurus.

16 steps- Saurischian Marasuchus, as in Kischlat (2000).  Ornithischian silesaurs result, and Lewisuchus sister to Eodromaeus plus avepods.

16 steps- Non-avetheropod Sinraptor as in Longrich (2002).  Carnosauria becomes a grade, Acrocanthosaurus joins Neovenator, and Guanlong and Tanycolagreus become maniraptoromorphs. 

18 steps- Basal paravian Archaeopteryx.  Deinonychosauria forms, and anchiornithids and scansoriopterygids are further from Eumaniraptora.

19 steps- Coelophysoid ornithischians or ornithischians sister to Neotheropoda, as in Baron (2017).  Chilesaurus is the most basal theropod.

19 steps- Basal deinonychosaur Archaeopteryx, as in Xu et al. (2011).   Anchiornis, Aurornis and Serikornis are one node closer to Dromaeosauridae plus Troodontidae.

20 steps- Monolophosaurus sister to Guanlong, as in Carr (2006) who proposed they were an adult and juvenile of the same species.  The pairing resolves as sister to Tyrannoraptora.

21 steps- Carnosaurian megaraptorans.  Er, wow.  The Carrano et al. consensus is blown out of the water.  They don't even group with Neovenator, instead (including Gualicho and Bicentenaria) being in a trichotomy with megalosaurids and an Allosauroidea including piatnitzkysaurids.

I'm drawing the line here for plausibility

22 steps- Theropodan Marasuchus, as in Olshevsky (1991). Silesaurids, Eoraptor+Buriolestes and herrerasaurians are also theropods.

23 steps- Avialan therizinosaurs, as in Maryanska et al. (2002).  One of the odder parts of the classic 'oviraptorosaurs are birds' analysis is that they recovered therizinosaurs as closer to birds than dromaeosaurids or troodontids, which was only briefly mentioned in the text, while they removed Troodontidae and Therizinosauria from their figured cladogram.  Constraining this result leads to oviraptorosaurs being dragged along, and the whole of Caenagnathiformes is sister to taxa closer to Aves like scansoriopterygids, anchiornithids, Archaeopteryx, etc..

24 steps- 'Allosaur' Ornitholestes, as in Paul (1988).   Although Paul includes Ornitholestes in his Allosauridae, he views that family as paraphyletic to tyrannosaurids and his figure 10-1 shows Allosaurus closer to tyrannosaurids than Ornitholestes.  I thus only specified Ornitholestes to be closer to Allosaurus than megalosaurids, piatnitzkysaurids, Compsognathus and birds.  The resulting tree has ornitholestiids (including Zuolong) sister to core allosauroids including Monolophosaurus (which was considered closer to Allosaurus by Paul too- pg. 307), but tyrannosauroids and compsognathids are coelurosaurs.

24 steps- Ornithuran (sensu Gauthier) alvarezsaurids.  The most crownward alvarezsaurids were ever proposed to be, closer to Aves than Archaeopteryx.  They end up just crownward of anchiornithids, and weirdly form a clade there with scansoriopterygids and oviraptorosaurs.  As usual, the absence of parvicursorines probably affects the numbers.

24 steps- Archaeopterygid Protarchaeopteryx, as in Paul (2002).  Xiaotingia and scansoriopterygids are also closer to Archaeopteryx than Aves in these trees.

25 steps- Tyrannosauroids sister to Pennaraptora, as in Sereno (1999).  Like Sereno's trees, alvarezsauroids and therizinosaurs (Beipiaosaurus) form an expanded Arctometatarsalia, though now joined by Ornitholestes, Aorun and CompsognathusCoelurus becomes a tyrannosauroid.

26 steps- Ornithischian alvarezsaurids, as in Alifanov and Barsbold (2009).  Chilesaurus emerges as an alvarezsauroid.  Note the true number is probably much higher since neither included alvarezsaurid has cranial material.

30 steps- Ceratosaurian ornithomimosaurs, as in my half-joking post.  They don't even come out by Limusaurus or Elaphrosaurus, instead Ornithomimosauria (including Zuolong and Gualicho) are sister to other ceratosaurs.  How disappointing.

35 steps- Sauropodomorphan BeipiaosaurusFalcarius stays by oviraptorosaurs, while Beipiaosaurus ends up sister to Guaibasaurus.

35 steps- Phytodinosaurian BeipiaosaurusFalcarius stays by oviraptorosaurs, while Beipiaosaurus is sister to Chilesaurus as an ornithischian.

36 steps- Ornithomimosaurian ElaphrosaurusGualicho follows, and ornithomimosaurs move stemward to be sister to Tyrannoraptora.

36 steps- Megalosaurid abelisauroids, as in Paul (1988).  Megalosaurus moves to Abelisauria.

36 steps- Dromaeosaurid Ornitholestes, as in Makovicky (1995).  Fukuivenator emerges as the most basal dromaeosaurid, and dromaeosaurids are the most basal pennaraptorans with oviraptorosaurs, scansoriopterygids, anchiornithids and troodontids successively closer to birds.  This is equivalent to my old post about getting dromaeosaurid evolution backwards.

37 steps- Bullatosauria.  This actually moves ornithomimosaurs plus alvarezsauroids into Avialae to be sister to troodontids.  I had to specify both Zanabazar and Sinornithoides as troodontids, because specifying Zanabazar alone moves it into Ornithomimosauria without the other troodontids at a lower cost of 21 steps.

37 steps- Oviraptorosaurian Sapeornis, as in Paul (2010).  Oviraptorosaurs move to just closer to Aves than Archaeopteryx, with scansoriopterygids closer to core oviraptorosaurs than Sapeornis.

Phylogram from Huene (1923) showing his idea of what were carnosaurs vs. coelurosaurs.

39 steps- Huene's (1923) Carnosauria vs. Coelurosauria dichotomy.  For such an archaic concept, this works surprisingly well.  The trick is that Huene's and Cau's Carnosauria are basically the same.  By 1923, Huene had moved Ceratosaurus to Coelurosauria and placed tyrannosaurids and Elaphrosaurus there as well.  His carnosaurs are Megalosaurus, Eustreptospondylus and Allosaurus.  Even looking at the taxa not included in Cau's analysis, most shake out right- coelurosaurian Sarcosaurus, Halticosaurus, Procompsognathus, Podokesaurus, Betasuchus, Genyodectes, Proceratosaurus and Thecocoelurus vs. carnosaurian Magnosaurus, Poekilopleuron, Spinosaurus and Metriacanthosaurus.  There's a load of non-theropods in there and Huene got Sarcosaurus? andrewsi, Dryptosaurus and Valdoraptor wrong, but still impressive.  So this basically how many steps it takes to force carnosaurs stemward of Ceratosauria sensu lato.

43 steps- Sauriurine enantiornithines, as in Martin (1983).  Basically constraining enantiornithines (Bohaiornis and Cruralispennia here) as closer to Archaeopteryx than to Aves.  Rahonavis and Balaur emerge as sauriurines, but surprisingly scansoriopterygids, Sapeornis, jeholornithids, Confuciusornis, Zhongjianornis and Protopteryx remain as closer to Aves ('Ornithurae' in BANDit terminology).

57 steps- Sauriurine enantiornithines and Confuciusornis, as in Hou et al. (1995).  This is more in line with BANDit thought, as not only enantiornithines, Archaeopteryx and Confuciusornis fall out as sauriurines, but also Protopteryx, Sapeornis, jeholornithids (Martin, 2004),  Vorona (Kurochkin, 2006), Rahonavis and Xiaotingia (those two as archaeopterygids).  Unlike the plus 43 step tree, scansoriopterygids are outside Sauriurae plus 'Ornithurae' similar to Czerkas' hypothesis.

58 steps- Abelisaurid Piatnitzkysaurus, as in Currie and Zhao (1994).  Condorraptor follows, Eoabelisaurus also becomes an abelisaurid.

75 steps- 'Carnosauria' vs. 'Oviraptorosauria' of Russell and Dong (1994).  The Alxasaurus description is my most nostalgic technical paper, because it was the first I tracked down that wasn't in Science or Nature.  The authors presented a strange new analysis of theropods, where tetanurines fell into 'Carnosauria' (Baryonyx, Yangchuanosaurus, Allosaurus, dromaeosaurids and tyrannosaurids in successive order) and 'Oviraptorosauria' (ornithomimosaurs, therizinosauroids, oviraptorosaurs and troodontids in successive order).  Needless to say, it doesn't hold up, even when topology within each clade is allowed to vary like it is here.  Birds end up in 'Oviraptorosauria', so that would be Coelurosauria under current nomenclature.

78 steps- "Pneumatocrania", Holtz's (1994) concept combining oviraptorids, 'elmisaurids', tyrannosaurids, troodontids and ornithomimosaurs to the exclusion of dromaeosaurids and birds.  Cau's matrix doesn't result in anything close to Holtz's topology for this clade, with troodontids sister to oviraptorosaurs (including scansoriopterygids), and ornithomimosaurs sister to tyrannosauroids with alvarezsauroids and therizinosaurs in a trichotomy with Tyrann+Ornithom.  As with Bullatosauria, Sinornithoides had to be specified as well.

81 steps- Arctometatarsalia sensu Holtz (1994).  This is like "Pneumatocrania" except it excludes oviraptorids.  In Cau's analysis, this results in most oviraptorosaurs (including scansoriopterygids) being maniraptorans, but 'elmisaurid' Anzu being sister to Zanabazar deep within Troodontidae.  Again unlike Holtz's topology, tyrannosauroids, ornithomimosaurs, alvarezsauroids, therizinosaurians and Fukuivenator are successively closer to troodontids.

84 steps- Huene's Pachypodosauria, where his carnosaurs are closer to sauropodomorphs than his coelurosaurs.  Pachypodosauria ends up containing sauropodomorphs and ceratosaurs plus Cau's expanded Carnosauria, with every other theropod a coelurosaur.

[Edit] 100 steps- Conservative BANDit topology.  Here I specified taxa with stage III or IV feathers as birds, and retained Heterodontosaurus, Plateosaurus, Herrerasaurus, Coelophysis, Majungasaurus, Megalosaurus, Allosaurus and Sinosauropteryx as dinosaurs theropodsForcing birds outside Saurischia, Dinosauria, Dracohors, Dinosauriformes, etc. is difficult as it is actually easier to get Teleocrater up in basal Coelurosauria than to break up those clades.  But just breaking up Theropoda into these two clades takes 83 steps.  Tyrannosauroids, ornithomimosaurs, alvarezsauroids and therizinosaurs group with birds.  

140 steps- Coelophysoid birds, as in Raath (1985).  Other theropods known at the time were constrained as monophyletic relative to a Coelophysoidea containing rhodesiensis, Archaeopteryx, Hesperornis, Ichthyornis and MeleagrisBicentenaria, Fukuivenator, halszkaraptorids, scansoriopterygids and paraphyletic anchiornithids end up bridging the gap between coelophysoids and birds.

Paul's (1984) cladogram of predatory dinosaurs, after Raath (1990).  It differs from Paul's actual printed cladogram where compsognathids are better interpreted as in an unresolved trichotomy between 'allosaurs' and other coelurosaurs, ornithomimids merely extend with a '?' falling between tyrannosaurids and Archaeopteryx, and oviraptorids extend with a '?' that falls on the troodontid plus bird branch.  Note this differs from the PDW phylogeny in that megalosaurids are down by Procompsognathus and tyrannosaurids are sister to Protoavia.

191+ steps- Kurochkin's (2006) diphyletic birds.  BANDit Kurochkin had a weird hypothesis that Archaeopteryx and enantiornithines were theropods, but Confuciusornis, Patagopteryx, Ichthyornis, Hesperornis and of course Aves are birds.  These diverged at the typically vague BANDit level of Archosauromorpha or Archosauria, with no comment on where crocodylians, ornithischians, sauropodomorphs, etc. go.  It's again very hard to constrain in TNT since the program doesn't let the outgroup (Euparkeria) be specified, so if you constrain the next closest taxon (Teleocrater) to be outside of dinosauromorphs on one side and 'Ornithurae' on another, it's more parsimonious for TNT to force Teleocrater into Ornithothorces than to make that basal divergence.  But even a weak version of Kurochkin's hypothesis where Coelophysis and Allosaurus are still theropods and lead to enantiornithines but are more closely related to 'Ornithurae' than Teleocrater, Lagerpeton, Marasuchus, Heterodontosaurus and Plateosaurus results in 191 more steps.  Fukuivenator, halszkaraptorids, scansoriopterygids, Balaur and Zhongjianornis end up on the 'ornithurine' line.

Ignoring maniraptoromorphs (so as not to hint at Lori's topology) I'm most surprised by the robusticity of Cau's expanded Carnosauria, the rootward mobility of compsognathid-grade taxa, how unparsimonious carnosaurian megaraptorans are, and how parsimonious ceratosaurian megalosaurids and arctometatarsalian tyrannosaurids are.  I think Gualicho is the most interesting theropod right now in terms of just what it is, since its remains are decent but it can pretty easily move between ceratosaurs, tyrannosauroids and ornithomimosaurs.  Bicentenaria also finds its way into a suprisingly large number of hypotheses, so deserves a more detailed description.

References- Huene, 1923. Carnivorous Saurischia in Europe since the Triassic. Bulletin of the Geological Society of America. 34, 449-458.

Martin, 1983. The origin and early radiation of birds. in Brush and Clark, (eds.). Perspectives in Ornithology. 291-338.

Paul, 1984. The archosaurs: A phylogenetic study. In Reif and Westphal (eds.). Third Symposium on Mesozoic Terrestrial Ecosystems, Short Papers. 175-180.

Raath, 1985. The theropod Syntarsus and its bearing on the origin of birds. In Hecht, Ostrom, Viohl and Wellnhofer (eds.). The Beginnings of Birds. Freunde des Jura-Museums Eichstätt, Eichstätt. 219-227.

Bakker, Williams and Currie, 1988. Nanotyrannus, a new genus of pygmy tyrannosaur, from the latest Cretaceous of Montana. Hunteria. 1, 1-30.

Paul, 1988. Predatory Dinosaurs of the World. Simon & Schuster, New York. 464 pp.

Raath, 1990. Morphological variation in small theropods and its meaning in systematics: Evidence from Syntarsus rhodesiensis. In Carpenter and Currie (eds.). Dinosaur Systematics: Approaches and Perspectives. Cambridge University Press, Cambridge. 91-105.

Britt, 1991. Theropods of Dry Mesa Quarry (Morrison Formation, Late Jurassic), Colorado, with emphasis on the osteology of Torvosaurus tanneri. Brigham Young University Geology Studies. 37, 1-72.

Olshevsky, 1991. A Revision of the Parainfraclass Archosauria Cope, 1869, Excluding the Advanced Crocodylia. Mesozoic Meanderings. 2, 196 pp.

Novas, 1992. La evolucion de los dinosaurios carnivoros. In Sanz and Buscalioni (eds.). Los Dinosaurios y Su Entorno Biotico: Actas del Segundo Curso de Paleontologia in Cuenca. Instituto "Juan Valdez", Cuenca, Argentina. 126-163.

Currie and Zhao, 1994. A new carnosaur (Dinosauria, Theropoda) from the Jurassic of Xinjiang, People's Republic of China. Canadian Journal of Earth Sciences. 30(10), 2037-2081.

Holtz, 1994. The phylogenetic position of the Tyrannosauridae: Implications for theropod systematics. Journal of Paleontology. 68(5), 1100-1117.

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(10), 2107-2127.

Currie, 1995. Phylogeny and systematics of theropods (Dinosauria). Journal of Vertebrate Paleontology. 15(3, 25A.

Hou, Zhou, Gu and Zhang, 1995. Confuciusornis sanctus, a new Late Jurassic sauriurine bird from China. Chinese Science Bulletin. 40(18), 1545-1551.

Forster, Sampson, Chiappe and Krause, 1998. The theropod ancestry of birds: New evidence from the Late Cretaceous of Madagascar. Science. 279, 1915-1919.

Sereno, 1999. The evolution of dinosaurs. Science. 284, 2137-2147.

Holtz, 2000. A new phylogeny of the carnivorous dinosaurs. GAIA. 15, 5-61.

Kischlat, 2000. Tecodoncios: A aurora dos Arcosaurios no Triassico. in Holz and De Rose (eds.). Paleontologia do Rio Grande do Sul. 273-316.

Longrich, 2002. Systematics of Sinosauropteryx. Journal of Vertebrate Paleontology. 22(3), 80A.

Maryanska, Osmolska and Wolsan, 2002. Avialan status for Oviraptorosauria. Acta Palaeontologica Polonica. 47(1), 97-116.

Paul, 2002. Dinosaurs of the Air. The Johns Hopkins University Press, Baltimore. 460 pp.

Rauhut, 2003. The interrelationships and evolution of basal theropod dinosaurs. Special Papers in Palaeontology. 69, 1-213.

Martin, 2004. A basal archosaurian origin for birds. Acta Zoologica Sinica. 50(6), 978-990.

Carr, 2006. Is Guanlong a tyrannosauroid or a subadult Monolophosaurus? Journal of Vertebrate Paleontology. 26(3), 48A.

Kurochkin, 2006. Parallel evolution of theropod dinosaurs and birds.  Entomological Review. 86 (Supp. 1),  S45-S58.

Smith, Makovicky, Hammer and Currie, 2007. Osteology of Cryolophosaurus ellioti (Dinosauria: Theropoda) from the Early Jurassic of Antarctica and implications for early theropod evolution. Zoological Journal of the Linnean Society. 151, 377-421.

Alifanov and Barsbold, 2009. Ceratonykus oculatus gen. et sp. nov., a new dinosaur (?Theropoda, Alvarezsauria) from the Late Cretaceous of Mongolia. Paleontological Journal. 43(1), 94-106.

Paul, 2010. The Princeton Field Guide to Dinosaurs. Princeton University Press. 320 pp.

Alifanov and Saveliev, 2011. Brain structure and neurobiology of alvarezsaurians (Dinosauria), exemplified by Ceratonykus oculatus (Parvicursoridae) from the Late Cretaceous of Mongolia. Paleontological Journal. 45(2), 183-190.

Lee and Worthy, 2011. Likelihood reinstates Archaeopteryx as a primitive bird. Biology Letters. 8(2),

Carrano, Benson and Sampson, 2012. The phylogeny of Tetanurae (Dinosauria: Theropoda). Journal of Systematic Palaeontology. 10(2), 211-300.

Senter, Kirkland, DeBlieux, Madsen and Toth, 2012. New dromaeosaurids (Dinosauria: Theropoda) from the Lower Cretaceous of Utah, and the evolution of the dromaeosaurid tail. PLoS ONE. 7(5), e36790.

Agnolin and Novas, 2013. Avian ancestors: A review of the phylogenetic relationships of the theropods Unenlagiidae, Microraptoria, Anchiornis and Scansoriopterygidae. Springer Netherlands. 96 pp.

Novas, Agnolin, Ezcurra, Porfiri and Canale, 2013. Evolution of the carnivorous dinosaurs during the Cretaceous: The evidence from Patagonia. Cretaceous Research. 45, 174-215.

Foth, Tischlinger and Rauhut, 2014. New specimen of Archaeopteryx provides insights into the evolution of pennaceous feathers. Nature. 511, 79-82.

Baron, 2017. Pisanosaurus mertii and the Triassic ornithischian crisis: Could phylogeny offer a solution? Historical Biology. DOI: 10.1080/08912963.2017.1410705 

Cau, 2018. The assembly of the avian body plan: A 160-million-year long process. Bollettino della Società Paleontologica Italiana. 57(1), 1-25.


  1. 9 steps- Carnosaurian Tyrannosaurus, which brings megaraptorans, Gualicho and Bicentenaria to form the sister group of Allosauroidea (including Monolophosaurus). I bet this is more parsimonious than most readers would assume given published topologies over the past two decades.

    *is data point*

    11 steps- Alvarezsauroid Nqwebasaurus. Alvarezsauroids emerge sister to ornithomimosaurs, with Haplocheirus and Jianchangosaurus forming a basal avemetatarsalian clade.

    Wait. What? Do you mean arctometatarsalian?

    It's again very hard to constrain in TNT since the program doesn't let the outgroup (Euparkeria) be specified, so if you constrain the next closest taxon (Teleocrater) to be outside of dinosauromorphs on one side and 'Ornithurae' on another, it's more parsimonious for TNT to force Teleocrater into Ornithothorces than to make that basal divergence.

    And you can't include Euparkeria in the constraint? What do constraints look like in TNT?

    Kischlat, 2000. [...] Paleontologia do Rio Grande do Sol.


    1. Correct re: Sul and arctometatarsalians.

      Regarding TNT constraints,I think I've worked that out now and will update.

  2. Thanks for listening (and accepting) my suggestion: D

  3. Thanks for the test, Mickey.
    The discrepancy in number of MPTs and thus in the strict consensus final topology is due to an error I made in search setting used in the original analysis (done about one year ago, now largely outdated even comparing it with the result of another test using the same taxon sample), so I think the strict shown in the paper is based on a subset of the actual trees found. Sorry, I tested so many versions of that matrix, to obtain a reduced tree not much different from that of the total matrix, that probably saved and then used not the exhaustive output.
    A lesson for future versions of the analysis :-)

    1. There are enough alternative Lori TNT/NEXUS files, so I completely understand. ;)

  4. The reason why so many finds Carnosauria sensu lato (Allosauroidea-Megalosauroidea) so unexpected is that most of us are "coelurosaurists" and assume that carnosaurs are the plesiomorphic condition of tetanurans. But if you look at coelophysoids and elaphrosaur-grade ceratosaurs, you should conclude that basal tetanurans were more compsognathid-grade than allosaur-grade.

    In any case, Carnosauria is almost constantly recovered in all iterations of my full matrix. And I suspect it will be even stronger once the new Patagonian tetanurine of Rauhut is published.

    1. That's a good point. I know when I was scoring Neovenator and Elaphrosaurus for the Lori matrix I was surprised by how coelurosaury they were. Even coelophysids can be surprising, with long distal caudal prezygapophyses and hollow paroccipital processes.

      For readers, the Patagonian taxon Andrea's talking about is-

      undescribed tetanurine (Rauhut and Diego, 2012)
      Callovian, Middle Jurassic
      Canadon Asfalto Formation, Chubut, Argentina
      Material- incomplete skull (~800 mm), presacral column, scapulae, coracoids, complete forelimbs, partial pubis, partial hindlimbs including tibia, fibula and metatarsal IV
      Comments- Pol and Rauhut (2012) note this as a large partially articulated basal tetanurine. Rauhut and Diego (2012) report tetanurine characters (maxillary fenestra; single pair of cervical pleurocoels; fibular flange on tibia that is offset from proximal end; well-developed posteromedial process of proximal end of metatarsal IV) as well as characters more basal than most tetanurines (apneumatic anterior dorsal vertebrae; short and distally strongly expanded scapula). Within Tetanurae, it shares characters with both megalosauroids (medially closed maxillary fenestra; U-shaped ventral postorbital process; broad groove on basioccipital below occipital condyle; lack of medial depression on proximal fibula) and carnosaurs (large antorbital fossa with associated foramina in nasal; raised lateral rims the nasal; well-developed lacrimal horn with two pneumatic recesses; pneumatic foramen in jugal; antarticular; metacarpal III <35% of metacarpal II width). They suggested this could indicate carnosaurian megalosauroids or high amounts of homoplasy.
      Reference- Pol and Rauhut, 2012. A Middle Jurassic abelisaurid from Patagonia and the early diversification of theropod dinosaurs. Proceedings of the Royal Society B. 279(1741), 3170-3175.
      Rauhut and Diego, 2012. A new basal tetanuran theropod from the Early Middle Jurassic of Patagonia, Argentina. Journal of Vertebrate Paleontology. Program and Abstracts 2012, 160.

      I don't know if anything has been published since 2012...

  5. So what would you currently say is the most feasible identification for Gualicho? Obviously it likes to jump around a bit in analyses and could logically fit in a couple of clades, but I still think a ceratosaurian identity is most likely.

    1. I honestly don't have an opinion I feel is well informed enough to have any weight behind it. Preliminary Lori analyses including all theropods found it as a ceratosaur, tyrannosauroid (with some megaraptorans) and ornithomimosaur at various times, but were not the most parsimonious trees and lacked many characters for use outside Maniraptoromorpha anyway. I do think Deltadromeus and it are closely related, whatever they are.

  6. Gualicho (and Deltadromeus) are among the most unstable taxa in my full matrix. They may be noasaur-grade ceratosaurians, basal coelurosaurs or megaraptoran tyrannosauroids.
    I'd love so much that a skull for one of them is found!

  7. In "4 steps- Ornithischian silesaurids. Lewisuchus is outside Dinosauria, Saurischia exists, and Asilisaurus and Pisanosaurus form a silesaur grade to either side of Silesauridae."
    What did you meant by "either side of Silesauridae"?

    1. I mean the ornithischian topology was ( Asili ( ( Sile , Saci ) (Pisano, other ornithischians ) ) )

    2. Now I understand, thanks.

  8. How many extra steps are needed for these topologies?

    1. Arctometatarsalian oviraptorosaurs, as in Osborn (1924).
    2. Coelurid Ornitholestes, as in Romer (1956).
    3. The traditional post-Huene carnosaur/coelurosaur dichotomy, as in Colbert (1964).
    4. Megalosaurus outside Avepoda, as in Paul (1984).
    5. Dromaeosaurid Archaeopteryx, as in Paul (1988).
    6. Birds & oviraptorosaurs outside Dinosauria with deinonychosaurs staying with theropods, as in Olson (2002).

    1. 1. 10 steps. Arctometatarsalia is made of a therizinosaur-oviraptorosaur clade and an ornithomimosaur-alvarezsaur clade.
      2. 5 steps. Nothing else falls into Coeluridae, which become the earliest diverging maniraptoromorphs.
      3. 85 steps. Well, technically Colbert (1964) had carnosaurs evolving from coelurosaurs, with 'podokesaurs' outside an ornithomimid-coelurid-carnosaur clade. But I get what you mean- dividing theropods into what were coelurosaurs vs. carnosaurs in our simplistic childrens books. A clade of coelophysoids and small ceratosaurs (I specified Elaphrosaurus and Masiakasaurus as coelurosaurs) form the first coelurosaur branch, while large ceratosaurs, megalosaurids, piatnitzkysaurids, Monolophosaurus, Sinraptor, Allosaurus and Acrocanthosaurus are successively closer to Tyrannosaurus plus Megaraptora. I didn't constrain Dilophosaurus, and Gualicho falls out as a coelurosaur which probably wouldn't have been believed in the 60s, but eh...
      4. 22 steps. Megalosaurus clades with Chilesaurus as the sister group to Avepoda, while Eustreptospondylus (constrained as an avepod) is sometimes sister to Avetheropoda.
      5. 24 steps. Similar to his allosaurid Ornitholestes, while Paul assigns dromaeosaurs to Archaeoptergidae, he thinks of that family as paraphyletic and has Archaeopteryx as the first branching 'protoavian'. Forcing Archaeopteryx to be closer to Velociraptor than troodontids or Meleagris takes 24 more steps, with anchiornithines being closer to eudromaeosaurs than Archaeopteryx. Forcing Archaeopteryx to be outside Maniraptoriformes is 59 more steps.
      6. 174 steps. Chilesaurus and scansoriopterygids are the first diverging birds, while sinovenatorids and anchiornithids (neither constrained) are closer to Archaeopteryx and other birds than oviraptorosaurs are.

    2. Thanks for answering! I definitely expected dromaeosaurid Archaeopteryx to be easier.

      If you don’t mind me asking, how many steps are needed for these (which I forgot to include in my previous comment)?

      1. Ornithischian birds, as in Galton (1970).
      2. Guanlong sister to Ornitholestes, based on Paul (1988) putting Proceratosaurus there.
      3. Abelisauroid Berberosaurus, as in Allain et al. (2007).
      4. Megalosauroid megaraptorans, as in Smith et al. (2008).

    3. 1. 156 steps, with Herrerasaurus constrained as a saurischian. Daemonosaurus is sister to ornithischians plus birds, and the following are successively closer to birds than Ornithischia- Chilesaurus, scansoriopterygids, halszkaraptorids, anchiornithids, troodontids minus Zanabazar (jinfengopterygids I suppose) and microraptorids (including Xiaotingia). What's funny about Galton (1970) is that basically the whole paper is correctly discounting arguments for why birds can't be closely related to ornithischians, but supporting evidence besides the pubic orientation is almost absent. It's just that, high sacral number (incorrectly claiming six for Archaeopteryx) and anteriorly toothless jaws (admitting Archaeopteryx doesn't show this).
      2. 18 steps. The two form the sister group to Maniraptoriformes. Higher than would have guessed.
      3. 2 steps.
      4. 25 steps. Bicentenaria and Gualicho are basal megaraptorans. Though note in Smith et al.'s topology, Megaraptor was sister to spinosaurids within Megalosauroidea, so the absence of spinosaurids in Cau's matrix probably makes this less parsimonious than it would otherwise be.

    4. Included Baryonyx and tested its sister-taxon relationships with megaraptorans: +1 step, with megalosaurs paraphyletic to Avetheropoda.
      Also included Lesothosaurus (but not the updated version of Buriolestes, which is instead in the full matrix): sauropodomorphs + herrerasaurs are sister taxa of silesaurids + ornithoscelidans.

      I am sincerely evaluating to stop giving credit to strict parsimony as basis for phylogeny...

    5. I am sincerely evaluating to stop giving credit to strict parsimony as basis for phylogeny...

      Actually, the behavior of Bayesian inference with paleontological datasets (lots of missing data that is distributed neither randomly nor by rate of evolution) has never been investigated. See the appropriate section in the Discussion of my latest preprint.

    6. Bayesian inference is not the only alternative to strict parsimony.

    7. If you mean parsimony with implied weights, that's discussed there, too... if you mean maximum likelihood, that's been tested even less than Bayesian inference, but is known to perform worse than Bayesian with certain distributions of missing data at least:

      Simmons MP. 2011a (printed 2012). Misleading results of likelihood-based phylogenetic analyses in the presence of missing data. Cladistics 28:208–222. DOI 10.1111/j.1096-0031.2011.00375.x
      Simmons MP. 2011b (printed 2012). Radical instability and spurious branch support by likelihood when applied to matrices with non-random distributions of missing data. Molecular Phylogenetics and Evolution 62:472–484. DOI 10.1016/j.ympev.2011.10.017

  9. My philosophy is to compare the results of different methods and in particular to include stratigraphy as additional source of data for tree inference.
    Madzia D., Cau A. 2017 - Inferring ‘weak spots’ in phylogenetic trees: application to mosasauroid nomenclature. PeerJ 5:e3782. DOI:10.7717/peerj.3782

    More importantly, we should move from naive parsimony focusing on just step count to a more mature evolutionary approach including testable inference derived from competing hypotheses as criterion for tree selection.

    1. Completely agree here. Step count is just an easy and quick metric to get a rough idea of how likely some topology is. As Jenner has argued in multiple papers, the current cladistic climate is great at Stage 2 of the process- amassing a ton of data. But it's lacking in Stage 1 (ensuring the data is good- a focus of the Lori analysis) and Stage 3 (usefully engaging with the data, which Andrea emphasizes here). From what I've seen in matrices I've used, my George Dubya gut instinct likes parts of parsimony results and parts of Bayesian results. Don't have enough experience with weighted results yet.

    2. If by weighting you mean implied weighting, see my preprint: it works if its assumptions are met, and often they're not.

  10. Interesting how Chilesaurus becomes a bird when the latter are constrained to be outside Theropoda…

    What about these topologies?

    1. Teleocrater sister to theropods, as suggested by Appleby et al. (1967).
    2. Hesperornis outside Ornithothoraces, as in Cracraft (1986).
    3. Megalosaurid Ceratosaurus, as suggested by Bakker et al. (1992).
    4. Abelisaurids as carnosaurs outside megalosauroids + allosauroids, as in Currie & Spinar (1994).
    5. Allosauroid Cryolophosaurus, as in Sereno et al. (1996).
    6. Enantiornithe Patagopteryx, as in Feduccia (1999).
    7. Patagopteryx outside Avialae, as suggested by Steadman (2004).
    8. Paravian Protarchaeopteryx, as in Holtz et al. (2004).
    9. Megalosauroid Cryolophosaurus, as suggested by Holtz (2007).
    10. Jeholornithid Rahonavis, as in Godefroit et al. (2013).