[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

A critique of Lu et al.'s (2002) Oviraptorosaurs compared to birds

While at SVP, Lu was nice enough to give me a copy of his paper-
Lu, Dong, Azuma, Barsbold and Tomida, 2002. Oviraptorosaurs compared to
birds. in Zhou and Zhang eds., Proceedings of the 5th Symposium of the
Society of Avian Paleontology and Evolution. Beijing Science Press, pg.
This is another paper advocating avian oviraptorosaurs, along the lines of
Maryanska et al. 2002.  Unfortunately, it has many of the same problems.
The most important aspect of the paper is the illustration of the skull,
sacrum and partial pelvis of IGM 100/2112, an oviraptorid labeled Ingenia
sp., which looks to me more like Citipati.  Jaime Headden confirmed it was
most similar to IGM 100/42, which has been termed Citipati sp. (though still
called Oviraptor philoceratops by Lu et al.).  It was discovered in the
Nemegt Formation of Mongolia in 1996, and described in Lu's 1999 thesis.
Scans of the photos can be sent to those who request them.

Lu et al. detail nine "new characters" shared between oviraptorids and
birds.  I should mention they include oviraptorids, Chirostenotes (as
Caenagnathus, they seem to ignore "elmisaurid" postcrania) and Nomingia in
Oviraptorosauria, but not Caudipteryx.  No mention is made of Microvenator.
1. Premaxillae fused and with elongate frontal process that extends
posteriorly to lacrimal.
Fused premaxillae are found in pygostylians, but their presence in
oviraptorids is not so clear.  Although Clark et al. (2002) describe
Citipati's premaxillae as fused, a clear suture is visible between the
elements on both their ventral and anterior surfaces.  This is the only
oviraptorid skull illustrated/described well enough to determine the
condition in.  However, Osmolska (1976) illustrated the juvenile crestless
skull ZPAL MgD-I/95 with a median suture, as did Barsbold et al. (1990) for
Conchoraptor and ventrally for Citipati sp. (IGM 100/42).
The second character is variable within oviraptorids (eg. absent in Khaan,
present in Citipati sp.), but is not found in most enantiornithines (contra
Lu et al.).  For example, it is absent in Sinornis, Eoenantiornis,
Eocathayornis and the Catalan nestling.  Confuciusornithids, Boluochia and
Gobipteryx have it, as does Yanornis.  Hesperornis seems to lack it though.
I should note contra Lu et al., dromaeosaurids and Compsonathus do not have
contact between the subnarial premaxillary process and the lacrimal.
2. Lacrimal bowed anteriorly.
This is only true in some oviraptorids (eg. Oviraptor lacks it), while most
others show a slight curve.  Lu et al. compare it to Confuciusornis, which
is similar (as is Changchengornis, probably).  Lacrimals are poorly known in
most basal birds, but the Catalan nestling, Yanornis and Hesperornis all
lack bowed lacrimals.
3. Slender rod-like jugal.
First, the jugal robustness of Archaeopteryx is not very different from
pygostylians.  Some ornithomimids, Sinornithosaurus, Shuvuuia and Avimimus
also fall into the avian range.
The actual cross section shape is known in very few taxa.  Lu et al. even
point out it changes from plate-like posteriorly to rod-like anteriorly in
their new specimen.  Taxa compressed onto slabs like most basal birds (eg.
Archaeopteryx) are probably difficult or impossible to judge the state in.
4. Mobile quadrate-quadratojugal articulation.
Though oviraptorids have a condylar articulation between the elements, it is
immobile (Maryanska and Osmolska, 1997).
5. More than eight sacral vertebrae.
This is odd, because they claim their new specimen only has eight sacrals.
Other oviraptorids have six or perhaps seven.  Anyway, only birds more
derived than Protopteryx have eight or more sacral vertebrae.
6. Ilia nearly contact in dorsal view.
Though present in oviraptorids and Nomingia, most basal birds lack the
condition (eg. Confuciusornis, Changchengornis, Longipteryx, Iberomesornis,
Liaoxiornis, Gobipteryx, Patagopteryx, Yanornis).  An enantiornithine pelvis
from the Lecho Formation (PVL -4041-4) has medially contacting ilia, but
this may be due to distortion (Chiappe and Walker, 2002).
7. Thirteen cervical vertebrae.
The new oviraptorid specimen has thirteen cervicals, one more than other
oviraptorids and Chirostenotes.  Lu et al. state only eight were present in
Archaeopteryx, but there were nine (Elzanowski, 2002).  Also, basal birds
had 9 (Catalan nestling) to 11 (Eoenantiornis, Liaoxiornis) cervicals, only
ornithuromorphs had thirteen or more.
8. Cervical ribs fused to centra.
This is only present in some oviraptorids (0- Ingenia, "Rinchenia"; 1-
Conchoraptor, Citipati), and is ontogenetically variable as well.  Though
Iberomesornis and ornithuromorphs have fused ribs, Confuciusornis does not.
The distribution of the character among coelurosaurs is sporadic.
9. Brachial depression on anterodistal humerus.
Apparently present in IGM 100/2112, but surely convergent.  Among birds,
only carinates and perhaps Neuquenornis (Chiappe, 2002; contra Chiappe and
Calvo, 1994) possess this feature.  It is absent in confuciusornithids, most
enantiornithines, Patagopteryx, Apsaravis and others.
As can be seen, none of these characters support a sistergroup relationship
of caenagnathoids and pygostylians, as advocated by Lu et al. (see below).
At best, it shows some oviraptorids developed several characters (frontal
process of premaxilla extends posteriorly to level of lacrimal, lacrimal
bowed anteriorly, eight sacral vertebrae, more than ten cervical vertebrae,
brachial fossa) in parallel to advanced birds.

Lu et al. also performed a cladistic analysis, though it has important
methodological problems.  They say dromaeosaurids are most often considered
the closest relatives of birds, but then state "In order to test the
hypothesis that oviraptorosaurs are more closely related to birds...
oviraptorosaurs and birds were used as the ingroup and Velociraptorinae as
the outgroup."  This is NOT testing the position of oviraptorosaurs and
dromaeosaurids compared to birds, as the outgroup (by definition) must be
excluded from the ingroup.  Furthermore, 70 of their characters are from
Chiappe et al.'s (1996) analysis of bird phylogeny (which had 99 characters
originally), and two more from their 1998 update (the Shuvuuia paper).  They
added three original characters.  This is in effect testing oviraptorosaurs
in an avian matrix, which while interesting, is not testing their position
relative to other maniraptorans.  Segnosaurs were supposedly "excluded
because of their incompleteness", but Alvarezsaurus was included?  Even
pre-1990 segnosaurs were known from a lot more than Alvarezsaurus is.  In
any case, the topology (1 most parsimonious tree of 149 steps; CI .591) is-
            |  `--Mononykus
Note they removed nearly all alvarezsaurid synapomorphies from Chiappe et
al.'s matrix, and did not include any characters to combine Caudipteryx and
caenagnathoids.  So it's no surprise these taxa didn't clade together.
Characters supporting Oviraptorosauria + Pygostylia (technically
Ornithothoraces here) are-
1. Premaxillae fused
Absent in oviraptorids (see above).
2. Frontal process of premaxilla extending posteriorly to level of lacrimal
Lu et al. miscoded oviraptorids (p), enantiornithines (p) and Hesperornis
(0) (see above).
3. Premaxilla toothless.
This is a problem of including too few taxa.  In this topology, it is only a
single reversal at Enantiornithes.  But take Protopteryx, Longipteryx,
Jibeinia, Eocathayornis and yanornithids into account and their absence in
caenagnathoids and confuciusornithids is suddenly much more parsimonious as
a convergence.  Mononykus/Shuvuuia is miscoded (?).
4. Quadratojugal joined to quadrate by ligament.
Only true in juvenile oviraptorids (Maryanska and Osmolska, 1997), this
should have been at the next lowest node (Metornithes), but Lu et al.
miscoded Mononykus as "0".  Technically, it's unknown in Mononykus, but Lu
et al.'s coding for the taxon obviously includes Shuvuuia material.
Archaeopteryx (1) and Confuciusornis (1) are miscoded.
5. Pygostyle present.
Assuming it was lost in oviraptorids, forcing Nomingia to be in the same OTU
would make this a synapomorphy of the clade.  Of course, Jeholornis
complicates things, but was unknown when this paper was written.
Mononykus/Shuvuuia (?) is miscoded.
6. Ulna subequal or longer than humerus.
Although Citipati has elongate forearms, Oviraptor, Ingenia and Khaan do
not, so oviraptorids are miscoded (p).  Also, Confuciusornis (82-90) has a
comparable ratio to Archaeopteryx (86-90).  As Iberomesornis is certainly
more derived than confuciusornithids, this would mean large ulnohumeral
ratios developed in parallel in Citipati and ornithothoracines.
7. Tibiofibular crest on distal femur.
This character is fine, given the distribution in the included taxa.
Confuciusornis is miscoded (1), and an undefined state 2 is scored for
Ichthyornis and Neornithes.
8. Cervical ribs fused to centra.
Lu et al. claim this is not known in any other theropods except
ornithomimosaurs.  However, it is also known in Allosaurus, Coelurus,
Calamosaurus, YPM 1996, derived therizinosauroids, Avimimus, Sinornithoides
and Saurornitholestes, so this is incorrect.  Velociraptorines (p),
Iberomesornis (1) and oviraptorids (p) are miscoded.  See above for more
9. Prominent acromion process.
Contra Lu et al., the acromion of Archaeopteryx is not less developed than
oviraptorids (Paul, 2002).  They miscoded Confuciusornis (0) and
Alvarezsaurus (?).
10. M. cuppedicus fossa absent in ilium.
Though the new specimen is said to lack one, "Rinchenia" and Conchoraptor
have a reduced, but present, cuppedicus fossa (Xu et al., 2002).
Archaeopteryx (p) and oviraptorids (p) are coded incorrectly.  Oddly,
oviraptorids are coded 0, contra what the text says.  The topology would
actually have this develop in Aves or Caudipteryx+Metornithes, and reverse
in a couple clades.
11. Ilia nearly contact in dorsal view.
Archaeopteryx (0), Mononykus/Shuvuuia (1), enantiornithines (0),
Patagopteryx (0) and Confuciusornis (0) are coded incorrectly.

Correcting codings in their data matrix resulted in 3 most parsimonious
trees (127 steps; CI .622)-
         |  `--+--Patagonykus
         |     `--Mononykinae
Looks like that got alvarezsaurids back together, and moved oviraptorids out
of Metornithes.  Iberomesornis is moved up next to other enantiornithines.
The only real difference here is oviraptorids being inside Aves.
The Oviraptoridae + Metornithes node is supported by premaxillary teeth
absent, prominent anterior dorsal hypapophyses and ischiopubic ratio>.66.
Adding more basal birds with premaxillary teeth (eg. Omnivoropteryx,
Protopteryx, Longipteryx, Jibeinia, yanornithids) would destroy the first
character.  The second has a complex distribution, and is going to have
homoplasy regardless of the phylogeny (0- segnosaurs, Microvenator,
Alvarezsaurus, Sinovenator, Microraptor, Archaeopteryx; 1- Ornitholestes,
caenagnathoids, mononykines, Sinornithoides, Saurornitholestes, Rahonavis).
The third would be compromised by including Chirostenotes and Nomingia, with
short ischia.
Enforcing a traditional contraint tree, with oviraptorids sister to
Caudipteryx and outside Aves, added only five steps.
In conclusion, Lu et al. 2002 suffers from a combination of misunderstanding
PAUP/cladistics, not including relevent taxa and miscoding characters.
There is still no well done phylogenetic analysis that supports oviraptorids
as avians.

Mickey Mortimer