Monday, February 17, 2014

Zhongornis probably isn't scansoriopterygid, contra O'Connor and Sullivan (2014)

Hi everyone.  Today we're looking at O'Connor and Sullivan's (2014) paper reinterpreting Zhongornis.  As the abstract says "The recently described maniraptoran theropod Zhongornis haoae, known from a single juvenile specimen, was originally identified as a bird. However, morphological re-evaluation reveals striking resemblances to both Oviraptorosauria and Scansoriopterygidae."  Well, does it?

No.

This is one of those papers that got worse and worse as I read more of it.  It seems to be written to reach a certain conclusion, but finds a different conclusion that is basically unacknowledged.  There are only a few actual reinterpretations of anatomy, namely the sacrum as having less vertebrae (5-6 instead of 6-7), the tail as having more vertebrae (~20 instead of 13-14), a less elongate coracoid and an ischium without a proximodorsal process (reinterpreted as the ilial peduncle).  Are these better than Gao et al.'s (2008) original interpretations?  It's impossible to tell, since the figures are terribly compressed jpegs that show artifacts even in the line drawings.  I've requested better versions from O'Connor but have yet to receive a reply.  Regardless, let's look at the evidence Zhongornis is closely related to scansoriopterygids and/or oviraptorosaurs.

The stated similarities to scansoriopterygids are as follows-

1. Short and deep skull.  Expected in any juvenile.
2. Short humerus (humerofemoral ratio 104% compared to scansoriopterygids' 98-112%).  Also expected in younger specimens, and a similar distance from adult Confuciusornis (114-127%) as the juvenile Liaoxiornis (108%) is from the adult Cathayornis (117%).
3. Weakly curved manual unguals with low flexor tubercles.  The curvature is actually similar to confuciusornithids and other basal birds.  O'Connor and Sullivan have a simplistic idea of basal birds and deinonychosaurs having more strongly curved manual unguals than oviraptorosaurs or scansoriopterygids, but there is a lot of variation in each. Also Scansoriopteryx and Epidexipteryx actually have large flexor tubercles.
4. Reduced number of caudal vertebrae (~20).  This is only true in Epidexipteryx (16) and admitted to be absent in Scansoriopteryx (~30-35).  This is also true in pygostylians of course (e.g. juvenile Dalingheornis with unfused caudals has ~20).
5. No distinct transition point or elongated distal caudals.  This is untrue in scansoriopterygids, where the distal vertebrae are elongate with reduced neural spines and transverse processes.  It can't really be evaluated in pygostylians with a pygostyle, though juveniles lacking a pygostyle like Dalingheornis are similar to Zhongornis and Epidexipteryx.
6. 'Incipient' pygostyle.  This phrase is used incessantly throughout this paper, but is never defined sufficiently.  The distal caudals in Zhongornis and Epidexipteryx are unfused (Scansoriopteryx's are unpreserved), so are not strictly pygostyles.  O'Connor and Sullivan cite Persons et al. (in  press) as attributing an incipient pygostyle to Caudipteryx, because "the last five vertebrae appear to be tightly integrated into an inflexible unit." But this is true of any coelurosaur, as the last caudals generally have flat articular surfaces and often elongate zygapophyses. Any pygostyle development is at least as true in pygostylians as it is in Epidexipteryx.
7. No obturator process.  Also true in Pygostylia, Jeholornithidae and Omnivoropterygidae.
8. Penultimate pedal phalanges longest.  This is not true in Scansoriopteryx, which has III-3 equal to III-1 and IV-4 equal to IV-1.  It is unpreserved in Epidexipteryx.  II-2 is longer than II-1 in most basal avialans as well, including confuciusornithids.
9. Manual phalanx I-1 longer than metacarpal II.  This is untrue in Zhongornis (94%) and scansoriopterygids (91% in Scansoriopteryx), and is similar to basal avialans (91% in Balaur; ~92% in Confuciusornis) and basal oviraptorosaurs (~93% in Similicaudipteryx; 89-93% in Caudipteryx).
10. Lack of proximodorsal ischial process.  This would be similar to scansoriopterygids, but the poor preservation proximally and great simiarity in shape to taxa with such processes (e.g. juvenile enantiornithine GMV-2158) make this equivocal at least.

Left ischia in lateral view of Zhongornis (top; after O'Connor and Sullivan, 2014), juvenile enantiornithine GMV-2158 (middle; after Chiappe et al., 2007), Scansoriopteryx holotype (bottom ; after Czerkas and Yuan, 2002).  Note the similarity between Zhongornis and GMV-2158, suggesting the process on the upper right is the proximodorsal process and not the iliac peduncle.
So we have no good scansoriopterygid characters in Zhongornis

How about oviraptorosaurian ones?

1. Low number of sacrals.  This could easily be ontogenetic, as juveniles often have less sacral vertebrae than adults.  The enantiornithine GMV-2158 has six sacrals for instance, when adult enantiornithines have seven or eight.  So if O'Connor and Sullivan are right that Zhongornis has 5-6, that would work if the adult had seven sacrals like confuciusornithids.
2. 'Incipient' pygostyle.  The argument used above works here- incipiency is a vague descriptor and any pygostyles are present in pygostylians as well.
3. Concave anterior narial margin formed by premaxilla.  This is true in the vast majority of theropods, pygostylians included.
4. Long frontals.  This is a confusing character to list here, since oviraptorosaurs usually have shorter frontals than most coelurosaurs due to their longer parietals.  Thus frontal length doesn't provide evidence Zhongornis is an oviraptorosaur.
5. Frontals narrow anteriorly and greatly expanded posteriorly.  Another confusing character, since oviraptorosaurs generally have less triangular frontals than most coelurosaurs.
6. Short and robust tail.  The tail is about equally short in juvenile pygostylians (e.g. length 1.59 times femoral length in Dalingheornis), Zhongornis (1.54) and basal oviraptorosaurs (~1.5 in Caudipteryx, 1.36 in Similicaudipteryx).  Oviraptorosaurs' tails are robust due to long chevrons (unpreserved in Zhongornis) and transverse processes (short in Zhongornis).  Thus there is no evidence Zhongornis has a tail that is robust for its length.
7. Robust furcula.  Basal oviraptorosaurs like Protarchaeopteryx, Similicaudipteryx and Caudipteryx actually have gracile furculae, and basal avialans like confuciusornithids have robust furculae.
8. Pointed epicleidia on furcula.  This is similar to some oviraptorids (Oviraptor, Citipati), though lacking in others (Khaan, Conchoraptor, "Ingenia", Jiangxisaurus, Heyuannia).  Unfortunately, other oviraptorosaurs do not preserve epicleidia in anterior/posterior view.
9. Metacarpal I 33% of metacarpal II length.  While shorter in many basal avialans, this is even stated by the authors to be present in Confuciusornis, and is even longer in Changchengornis and Balaur.  
10. Metacarpal I wider than metacarpal II.  Again, this is present in confuciusornithids and Balaur.
11. Manual phalanx I-1 subequal in length to metacarpal II.  As discussed in scansoriopterygid-like character 9 above, this is similar in Zhongornis, Confuciusornis and Balaur.
12. Weakly curved manual unguals.  As noted above, there is a lot of variation in oviraptorosaurs, with e.g. Protarchaeopteryx having more strongly curved unguals than Caudipteryx.  The curvature in Zhongornis isn't more similar to oviraptorosaurs' than pygostylians'.
13. Long nasals.  Another confusing character, since oviraptorosaurs have shorter nasals than most coelurosaurs.
13. ~20 caudal vertebrae.  This is not different from basal pygostylians as e.g. the juvenile Dalingheornis holotype has about 20 caudal vertebrae while lacking a pygostyle.
14. Reduced manual digit III of three phalanges. This may not be true (a short III-1 may be hidden), but the similarity to Caudipteryx's two phalanges noted by O'Connor and Sullivan is problematic for two reasons. First, as they state, Sapeornis has less than four phalanges on digit III (also true in Balaur). More importantly, the authors are not proposing a caudipterid identification for Zhongornis, and more basal oviraptorosaurs like Protarchaeopteryx and Similicaudipteryx have an unreduced digit III.

Only the narrowly pointed epicleidia are more similar to some oviraptorosaurs than to basal pygostylians.  O'Connor and Sullivan go on to compare scansoriopterygids with oviraptorosaurs, listing supposedly shared characters, and stating "We suggest that accumulating morphological information regarding both scansoriopterygids and basal oviraptorosaurs may eventually demonstrate that the former clade is either on the oviraptorosaurian stem or nested within basal Oviraptorosauria (Fig. 3), and convergently evolved ‘avian characteristics’ as a result of adaptation to an arboreal lifestyle."

Analysis and unjustified conclusions

Of course just listing characters isn't that useful, and O'Connor and Sullivan proceed to add Zhongornis to the Xiaotingia version of Senter's TWG matrix.  They recover it as the most basal avialan, followed by Scansoriopterygidae, Jeholornis and Avebrevicauda.  Deinonychosaurs (including archaeopterygids) are sister to Avialae, with oviraptorosaurs and therizinosaurs more basal.  So hypotheses unsupported, right?  You'd think so, but the authors go on as if the opposite had occured.

Strict consensus tree of O'Connor and Sullivan (2014).  Note Aves should either be within Ornithuromorpha or at the Paraves node, depending on the definition used.  Is it suspicious the Paraves clades are flipped so that Zhongornis and scansoriopterygids are next to oviraptorosaurs?
 "Scansoriopterygidae is recovered as the sister taxon to Aves, with the two together forming the clade Avialae."

Since when is Jeholornis+Avebrevicauda Aves?  Aves is either crown birds, or Archaeopteryx plus crown birds, which would be Eumaniraptora here.

"Zhongornis is resolved as sister taxon to Avialae (Scansoriopterygidae + Aves) supporting our hypothesis that Zhongornis is closely related to scansoriopterygids."

First, the scansoriopterygid+jeholornithid+avebrevicaudan clade doesn't correspond to any proposed definition of Avialae.  Either Zhongornis is an avialan because it's closer to Aves than dromaeosaurids and/or troodontids (Gauthier, 1986), because it has wings for powered flight homologous to Aves (Gauthier and de Queiroz, 2001) (moreso than scansoriopterygids seem to at least) or because it's in the Archaeopteryx+Aves clade (Gauthier and Wagner, 2001).
Second, it's only closely related in the Petersian sense of being sister to a clade including scansoriopterygids and another branch.  You could just as validly say Zhongornis is closely related to Jeholornis+avebrevicaudans.

"A relationship between Zhongornis and Scansoriopterygidae is supported by six characters (101, 103, 166, 273, 317, and 325); however, scorings for most of these characters are ambiguous in Zhongornis because of missing data. The only one whose presence can be confirmed in Zhongornis, namely the fact that the minor digit is shorter than the major digit, is absent in Epidendrosaurus."

Er.... a relationship can't be supported by characters that are unknown in one of the two taxa.   Having a supporting character lacking in one taxon is even less sensical.  Obviously character 325 (manual digit III longer than II) can't support this clade if it is absent in Zhongornis.  I'm honestly not sure what this list is supposed to be.  I'd say it's a list of Avialae characters in ACCTRAN (where traits are optimized as evolving as early as possible, so that the basalmost avialan Zhongornis is modeled as having characters unknown for it but present in more derived birds), except 325 should just be a scansoriopterygid character and the actual list of avialan characters should include e.g. obturator process absent.  Alas, O'Connor and Zhou don't include their matrix, so we can't know for now.

We get numerous proclamations such as-
"This study reveals new morphological information that strongly suggests the holotype of Zhongornis is a juvenile scansoriopterygid or close scansoriopterygid relative."
"Based on new morphological data and comparisons with other avian and non-avian taxa, we suggest that the Early Cretaceous ‘bird’ Zhongornis haoae may in fact be either a member or a close relative of the Scansoriopterygidae"
"Revised anatomical interpretation of the tail and more detailed comparisons with non-avian dinosaurs strongly suggest that Zhongornis haoae is not a bird but a member or close relative of the enigmatic maniraptoran clade Scansoriopterygidae."

The analysis didn't even suggest Zhongornis is a scansoriopterygid, let alone strongly suggest it.  Again the authors ignore the equally strong suggestion from their analysis that Zhongornis is a close relative of more derived avialans.

"The apparent scansoriopterygid affinities of Zhongornis would suggest the clade persisted from the Mid-Late Jurassic Daohugou times into Early Cretaceous Jehol times."

Whoah whoah... now the placement of Zhongornis in Scansoriopterygidae is apparent?  

"Zhongornis also bears some similarity to basal oviraptorosaurs, supporting the hypothesis that the Jurassic scansoriopterygids may be stem-group relatives of the Cretaceous Oviraptorosauria."

Apparently it didn't, since scansoriopterygids are avialans in their tree, while oviraptorosaurs are further removed than deinonychosaurs.  And I'm sure Mike Keesey is wondering what the crown-group oviraptorosaurs are. ;)
And tying both MIA hypotheses together...

"The relatively short forelimbs and short hallux in Zhongornis may suggest this taxon is a basal scansoriopterygid, close to the divergence of this clade from basal oviraptorosaurs, although this is inconsistent with its occurrence in the Yixian."

Well, IF Zhongornis were a scansoriopterygid and IF scansoriopterygids were oviraptorosaurs, that might be true.  Though Zhongornis actually has longer forelimbs (84% of hindlimb length, excluding phalanges) than scansoriopterygids (70-81%), and an equally long hallux (I-1 23% of metatarsal II).  So even this doubly hypothetical scenario then doesn't match with the evidence.

Sadly, this paper reminds me most of something from Feduccia or Martin.  O'Connor and Sullivan start with a relationship in mind and list characters to support it, but these are generally incorrect or equally correct for the opposing hypothesis.  They also often compare characters to different taxa in a group, so Zhongornis is like oviraptorids but not Caudipteryx in A and B, and like Caudipteryx but not oviraptorids in X and Y, so therefore is like oviraptorosaurs.  Phylogenetic terms are misapplied, and the result of any analysis is only important in the ways that it agrees with their ideas, not in the ways it disagrees.  One gets the impression the original draft never even included an analysis, as the discussion doesn't take it into account and the authors seem not to know how to evaluate their ideas with that dataset.  What they should have done is constrain scansoriopterygid and oviraptorosaurian Zhongornis to see how unparsimonious those hypotheses are, then constrain oviraptorosaurian scansoriopterygids including or excluding Zhongornis.  I've done that in the Lori analysis, and will say the results are much closer to their cladogram than their written ideas.

References- Czerkas and Yuan, 2002. An arboreal maniraptoran from Northeast China. Feathered Dinosaurs and the Origin of Flight. The Dinosaur Museum Journal. 1, 63-95.

Chiappe, Ji ans Ji, 2007. Juvenile birds from the Early Cretaceous of China: Implications for enantiornithine ontogeny. American Museum Novitates. 3594, 46 pp.

Gao, Chiappe, Meng, O'Conner, Wang, Cheng and Liu, 2008. A new basal lineage of Early Cretaceous birds from China and its implications on the evolution of the avian tail. Palaeontology. 51(4), 775-791.

O'Connor and Sullivan, 2014. Reinterpretation of the Early Cretaceous maniraptoran (Dinosauria: Theropoda) Zhongornis haoae as a scansoriopterygid-like non-avian, and morphological resemblances between scansoriopterygids and basal oviraptorosaurs. Vertebrata PalAsiatica. 52(1), 3-30. 

Persons, Currie and Norell, in press. Oviraptorosaur tail forms and functions. Acta Palaeontologica Polonica. doi: http://dx.doi.org/10.4202/app.2012.0093

1 comment:

  1. "And I'm sure Mike Keesey is wondering what the crown-group oviraptorosaurs are."

    GAHHHHHH!!!!!

    ReplyDelete