Some time ago, Michael Mortimer posted on his site the idea that the Dzharakuduk avian Kuszholia mengi might actually be an oviraptorosaur! The similarities in the vertebrae (based mostly on sacrals) were starling. But also convergent. Was it? The Dzharakuduk locality has also produced definite jaws of oviraptorosaurs, so the question of whether these jaws and those vertebrae belonged together was raised.
The Tiniest Oviraptorosaur
Lev Nessov named Kuszholia mengi early on in 1992 as part of his grand description of scrappy central Asian fossil material, much of it found near the small town of Dzharakuduk in Uzbekistan. It’s a fairly large animal for the site, one of its largest (the pterosaurs and ornithischians of the site being much larger, but few intermediate). However, Kuszholia mengi is known only from vertebrae; the holotype is a partial synsacrum, and various bits referred, including another partial synsacrum, dorsal, and some cervical vertebrae, suggest a similar animal. Diagnostically, the centra of these vertebrae were dorsoventrally flattened, much wider than high, and the neural canal as wide as the centra. This is comparable to some small birds, such as Patagopteryx (Zelenkov & Averianov, 1997), but distinctively the sacra bore large — very large — pneumatic pleurocoels. These oval, elongated pleurocoels are simple in form: they are merely deep, deep excavations with rounded margins, like smooth-sided pits falling into the abyss of the vertebrae. Other sacra from Dharakuduk had pleurocoels, but were less elongate, smaller, with abrupt margins, and often surrounded by laminae. They suggested a more complex diverticular system extended along the sides of the sacrum, unlike in Kuszholia mengi.
Not long thereafter, tiny jaws, fused at the mandibular symphysis, were described and named Caenagnathasia martinsoni. Lev Nessov, a coauthor, would die not long afterward. Caenagnathasia martinsoni is considered unquestionably oviraptorosaurian, given the strong similarities to the North american caenagnathid jaws described as Caenagnathus, hence the name: Asian Caenagnathus. But here’s where things get tricky. Almost none of the Dzharakuduk material is articulated. A few bits here and there make it hard to determine, for sure, what really belongs to what.
So it happens that a lot of bones have to be evaluated in isolation. Some of these are likely to belong to other bits, and thus reduce complication in taxonomy. Indeed, since they were discovered, the jaws have left a tantalizing mystery, and a quest to uncover where that would lead. The holotype is a conjoined pair of dentaries, fused at their symphysis. Caenagnathids are unique amongst oviraptorosaurs (and non-avian theropods, or in fact non-avian dinosaurs) for having fully fused mandibular symphyses, so it’s fairly easy to identify them when they pop up. The paratype is one half of such a structure, and the symphyseal surface is missing, the portion of jaw broken and irregular. connecting to the other jaw is rough, rugose. These mandibles are of roughly the same size.
Thus, when a new pair of fused dentaries were discovered after those of the holotype and paratype, it seemed pretty obvious they should be Caenagnathasia martinsoni. Like the first two sets of partial jaws, this new specimen, described by Hans Sues and Alexander Averianov, presents a scooped mandibular symphysis and an array of ridges on the inside perpendicular to the outer edge, and flanked on the inside by another long ridge. Currie et al. were considerate to imply these ridges were remnants of dental alveoli.
Indeed, I was so enamored of this idea I quickly imagined how the dental alveoli of oviraptorosaurs might have been progressively lost. Sadly, it’s a bad dream. Teeth are responsible for developing interdental, alveolar bone, the bone that forms the margins of the socket; take them away, and there’s no alveolus. So what is an alveolus-like structure doing? An initial surmise, one which works for turtles as well, is that these are secondary developments of the lingual margin of the jaw, and because of their intricate relationship with jaw biting and force dissipation, may be taxonomically viable (compare to Currie et al., 1994, who merely assert the viability of the ridges as taxonomic in nature). Median tubercles, lingual grooves, placement of the splenial and prearticular grooves, Meckel’s groove, etc. — a range of other features suggest possible taxonomic variability, but due to the brevity of jaw material, we are left to speculate still.
But Sues and Averianov go a step further. They describe new vertebrae, as well. They also described and refer to Caenagnathasia martinsoni vertebrae that had previously been referred to Kuszholia mengi, including one of the latter’s paratype specimens, a synsacrum. In seems that the authors all synonymized the two taxa (though, if they did, Kuszholia and mengi have priority over Caenagnathasia and martinsoni, respectively). The question turns then, not to what to do with the mandible, but what to do with the vertebrae.
Michael Mortimer rolled in and considered this material to possibly belong to the same taxon. Back when he first made this observation, he wrote:
This sacral/caudal morphology is also found in oviraptorids (e.g. Microvenator, Gigantoraptor, Rinchenia, Conchoraptor, “Ingenia”). Yet Neimingosaurus and caenagnathids (e.g. Shixinggia, RTMP 92.36.53) have rounded articular surfaces like Kuszholia. Adult therizinosaurs are far larger, with even the smallest basal members (e.g. Beipiaosaurus) being four times as big, and the fusion does indicate Kuszholia‘s holotype is from an adult. There are small caenagnathids though, including Caenagnathasia from the same formation. In fact, Caenagnathasia would have comparably sized sacrals to Kuszholia if scaled from other caenagnathoids. If they are synonymous, Kuszholia would have priority. The holotype is generally similar to Chirostenotes, but differs in having much larger pleurocoels (size related?) and narrower postzygapophyses. It also has larger pleurocoels than Shixinggia. Avimimus is unique among oviraptorosaurs in lacking sacral pleurocoels, so is quite different. Kuszholia is thus provisionally referred to the Caenagnathidae.
(He has since updated the text to reflect Sues & Averianov’s arguments, so these comments are now outdated, but preserved here for posterity. The more up to date comments can be found here. There, Mortimer refers Caenagnathasia martinsoni to Kuszholia mengi, though as my comments might show I would be very wary of this referral on material grounds.)
Michael posted the taxon into his formulation for Oviraptorosauria and I left it at that. Michael further noted that the material might, possibly, be troodontid. The ovirptorosaurian-bird clade, called Pennaraptora, is a perfect container for this morphology, where birds, troodontids, and oviraptorosaurs have peculiar and similar sacra, whereas dromaeosaurids are a bit different. So it’s with some surprise that I found that the new Sues and Averianov paper refer two of the Kuszholia mengi synsacra (one of which was a paratype) to Caenagnathasia martinsoni in their new descriptive work. This material represents a partial synsacrum with about 3 vertebrae, and a complete synsacrum with only 4 vertebrae. No indication of further articulated sacrals are present, though presumably there’d be at least one more sacral, but unfused to the rest. The Kuszholia mengi material is also very small, and as implied by Sues and Averianov, and by Mortimer, they are of comparable size to Caenagnathasia martinsoni, but with specimens of this form also larger there may be a two-fold increase in size of known specimens. Was Caenagnathasia martinsoni not actually one of the very tiniest non-avian theropods?
Peculiarly, Sues and Averianov refer postcranial material to Caenagnathaisa martinsoni without comment or reasoning. It is merely done, and the removal of material from the hypodigm of Kuszholia mengi similarly goes without comment. Presumably this is due to the diverse array of synsacra involved: there certainly seem to be two different types, as outlined above: one synsacrum with simple, elongated and oval pleurocoels bearing sloped margins, and very flattened centra, and the other with more complex, nested pleurocoels, laminae and lateral fossae that do not form pleurocoels, and deeper centra. Both have a longitudinal ventral groove (perhaps part of the difficulty in referral). As noted by Zelenkov and Averianov (2011) and by Mortimer, however, these factors are complex amongst maniraptorans, and thus is is unclear exactly which set of synsacra belong to the oviraptorosaur — if any. Sues and Averianov also describe new anterior vertebrae, cervicals and dorsals, which are more clearly similar to oviraptorosaurs in some respects.
Sadly, none of this material is so clearly oviraptorosaurian, due in large part to the fairly generic morphology present. It seems troodontids are the culprit again, and this actually implies that troodontids may also be a component of the Bissekty fauna, along with oviraptorosaurs. Dromaeosaurids, therizinosauroids may also be present, and suggest that the western Asian fauna from the Turonian was not that much distinct from its far eastern and somewhat younger Santonian through Campanian fauna which are far better known. Should this prove the case, then the Bissekty reflects a western extension of the otherwise central Asian terrestrial vertebrate fauna, and may not possess as much endemism as has been implied. The Bissekty represents a much wetter environment than the Wulansuhai or Shine Khuduk formations (the latter one of the youngest portions of the Psittacosaurus fauna, which spanned multiple tens of millions of years), so there would be expected variations: different mammals, birds, aquatic saurians of many stripes, including crocs and turtles.
Aside: Not the Tiniest?
Peculiarly, some of the material from Dzharakuduk represents oviraptorosaur-like animals considerably larger than the speculated size of the jaw material. This suggests that much of what we think of as Caenagnathasia martinsoni, once called one of the tiniest non-avian dinosaurs, wasn’t really that small, and may have been up to twice as large.
But Wait! …
As I was working on this, another paper was made available in Cretaceous Research prior to print publication which has raised another problem, and this is one that I’ve discussed at length before. Specifically, when it comes to Caenagnathidae. Rather than rehash it all over again, read this post, and come back here.
A large amount of otherwise undescribed “caenagnathid” material has arisen from the late Campanian Dinosaur Park Formation since the 1970s. This material includes a large amount of postcrania. But due to the vagaries of taxonomy, we have a lot of caenagnathid taxa from the DPF: Chirostenotes pergracilis, Caenagnathus collinsi, Macrophalangia canadensis, Caenagnathus sternbergi, and Elmisaurus elegans. These taxa are based variously on a pair of hands, a complete and fused pair of jaws, a nearly complete foot, a partial rear of a jaw, and another partial foot, missing the toes, respectively. Additional material was recovered that suggested some of this might all be the same taxon, especially the skeleton RTMP 79.20.1 and what is now the holotype of Epichirostenotes curriei, ROM 43250. Some have been doubtful, raising their concerns as asides in papers over the years.
Phil Currie especially has tried to maintain taxic distinction between the “elmisaurid” and “caenagnathid” groups, for no greater reason than the extreme of a fused tarsometatarsus. This mechanical property, which occurs in the holotype and some referred feet of Elmisaurus elegans, reflects potential evolutionary distinction, with the Asian Elmisaurus rarus the only other oviraptorosaur aside from Avimimus portentosus to have this condition. Otherwise, workers have variously referred elegans to Chirostenotes as a senior name to the small Caenagnathus sternbergi (= Chirostenotes elegans). This little headache is due to the impression some have had that size-based differences account for taxonomic variation, and are thus of systematic value. A few disagree, and prefer to use minimal taxonomy to reflect this.
[Edit:Pes of caenagnathids show little of the extreme size variation to suggest the morphs that might predicate moderate size of the “Macrophalangia” pes in CMN 8538 relative to the “Chirostenotes” pes in RTMP 79.20.1. Ontogenetic variables have not been taken into account — by anyone — in this account, which is unfortunate considering that it should be important in considering the apparent lack of fusion in proximal metatarsals in some adult morphs versus others in which fusion is present, but at smaller (more “gracile”) size.]
Nicholas Longrich and colleagues tackled this problem a short while ago with the creation of Leptorhynchos for a pair of distinct jaws found in Texas along with some associated postcrania. Some of this material was so close it would likely have been treated as a single specimen; however, the material has all been split so that each bone is a distinct specimen. It is likely, though, that some of it belongs to one individual. I cover the topic here.
Longrich et al. proposed some more radical interpretations, including moving elegans out of either Elmisaurus or Chirostenotes and into Leptorhynchos. But split or lump, and variations in between, size isn’t a necessarily useful character when the holotypes of many cover material known only in that taxon. There are a lot of jaws of caenagnathids from the Dinosaur Park Formation, and few have been described in substantive detail. Their variation has hardly been quantified, but at the same time none of it is associated with postcranial material. Indeed, until Anzu wyliei, no other American caenagnathid was known with both a hand and a lower jaw. The speculation that all of the DPF caenagnathid materia may be lumped into as few as one species (Chirostenotes pergracilis) or two (pergracilis + elegans) with size-based discrimination has been popular.
New work by Greg Funston et al. offer another claim: that there is a larger morph in the material, represented by the Caenagnathus collinsi holotype and referred jaws (all of which are smaller). Moreover, Currie et al. (1994) speculated there might be up to three distinct morphs of jaw in the DPF. Description of a new lower jaw referred to Caenagnathus sternbergi by Currie (2005) was completed by Funston and Currie (2014); this specimen, RTMP 2001.12.12, represents a jaw the same size as the holotype of Caenagnathus collinsi (CMN 8776), but with a sharper, upturned bill tip and distinct surface ornamentation of the long mandibular symphysis. This converges one of the Currie et al. (1994) morphologies into the Caenagnathus sternbergi articular morphology. But no work has yet managed to converge either of these jaws with the postcrania. There’s a “gracile” and “robust” morph for feet, jaws, suggesting that merely two taxa need be represented.
Funston et al., however, seem to take an odd road in claiming that a third, “large” morph is present, and that has become Caenagnathus collinsi. Why these postcrania, which are larger than other “middle” (pergracilis) and “small” (elegans) morphs, are placed into collinsi is not explained. It would be enough to claim they represent large specimens, intermediate between the RTMP 79.20.1 “2 meter” morph and the Anzu wyliei “4 meter” morph. One solution is that Funston et al. are setting aside the name Chirostenotes pergracilis because of the limited material involved, and the “less diagnostic” property of its holotype hands, and thus leaving it bereft of all other material; then, they refer the remainder typically considered of Chirostenotes pergracilis into Caenagnathus collinsi, which I suspect is the case.
The material problem at hand is very, very simple:
The framework for referral of taxa based on disparate holotypes (specimens belonging to different parts of the body) cannot be rectified by fiat. You need overlapping material; without it, you need associated material and a strong bias against taxic mixing (such as a monotaxic bonebed). Without these, given the vagaries of history, the names should be retained for what they are best supported as.
We have cranial material, and no cranial material associated with postcrania, for caenagnathids in the DPF. Thus, all cranial-based taxa should remain just that: alone. It sounds harsh, but while it is likely that some of the material found does belong to known taxa, we don’t actually know which. Caenagnathus collinsi also happens to be the right size to fit the hands of Chirostenotes pergracilis and the feet of Macrophalangia canadensis, arguments about relative robustness aside. Animals in nature also tend to vary in size through some rather large extremes, without genetic defect. We have NO information on relative ages of caenagnathids, thus assumptions of stable size of known specimens becomes prey to the ravages of age: some animals die younger, and smaller, than others.
Until then, the following should be true:
1. Caenagnathasia martinsoni and Kuszholia mengi might be the same taxon. Until a specimen arises that contains both jaw and sacrum, you can’t prove this. So they should be retained separately. (Sorry, Mickey.)
2. Chirostenotes pergracilis and Caenagnathus collinsi cannot be compared, so should be kept separate.
3. Macrophalangia canadensis CAN be compared to Chirostenotes pergracilis, and the material as complemented by Elmisaurus rarus and Anzu wyliei indicates they probably are the same taxon.
4. Leptorhynchos is restricted to gaddisi: We can’t affirm the brief pedal material that belongs to it to elegans, which is just a variety of feet from the Campanian to the Maastrichtian that, like other taxa in that range, probably have greater taxonomic variation than being all one species. They might be the same. Then again, northern provincialism suggests otherwise (the northern Montana-Canada province, the middle Utah province, and the southern Mexico-New Mexico-Texas province all appear to have distinct Campanian through Maastrichtian faunae with little overlap).
Partial postcranial material that may be diagnostic to a new taxon and occurs in a formation with other diagnostic remains but are incomparable (different parts of the animal) simply shouldn’t be referred to one another. It might be right — but you cannot prove it. Science is an evidence-based process, and the arguments for referral of postcrania to only-jaws-known taxa like Caenagnathasia and Caenagnathus makes the whole thing more difficult, not less. The listing of who is referring what where grows with every new paper, not all iterations are consistent, and the composition is in flux because everyone wants to dip their toes and affirm some theory or other. This all while the critical process of handling all the material in a systematic, fully descriptive manner, including comparing all the jaws, remains undone (but it’s in the works). Thus, I question all of these referrals, and strive to maintain the taxa we have because, well … they might ALL be valid and distinct taxa, and knowing which name belongs to what bone can be far more problematic when your variation is primarily in your jaws.
[Edit: I failed to extend on this premise at the time I wrote this, but it was my intention to call out the referral of distinct synsacra to the same complex of specimens. The second synsacral type noted above, though more complete, and shown in the first figure, at bottom (the 4-sacral specimen, ZIN PH 806/16) compares well to some “avian” synsacra Nessov referred to birds, such as the type of Lenesornis maltshevskyi in having a complex array of small, abrupt pleurocoels and laminae. This latter morphology compares significantly well to the synsacra of caenagnathids as recovered with Epichirostenotes curriei and Chirostenotes pergracilis (e.g., RTMP 79.20.1).
However, these synsacra lack pleurocoels on their last centrum, whereas oviraptorosaur synsacra typically possess them, as does ZIN PH 802/16, below. The synsacra of Kuszholia mengi typically lack this complex lamination, so while they possess last-synsacral pleurocoels, they lack the morphology expected in a caenagnathid synsacrum, and thus may belong to a distinct oviraptorosaurian taxon. If this is the case, then there are two distinct species of oviraptorosaur at Dzharakuduk; or that the synsacral morphology is extremely plastic; or that there are convergent adaptations to synsacral amongst the small penneraptorans of the Itemir-Dzharakuduk depression: Itemirus medullaris, a possible dromaeosaurid; Kuszholia mengi, a possible troodontid; Caenagnathasia martinsoni, and possibly an additional Caenagnathasia-like taxon represented by distinct jaws which have gone missing (Currie et al., 1994) but might be represented by the new material noted below, etc. Additionally, some may be more avialaean than penneraptoran.
This diversity underscores the problem noted above that it is problematic to refer to taxa disparate parts of animals without some argument of association, especially when there are indications of diversity within the sample that may represent distinct groups of animals deserving taxonomic recognition. The neat and clean method of collapsing as much taxa into one another serves little purpose but to “dust the bins,” as it were, and reduce the potential, but not actual, redundancy of taxa. As you cannot know which may be which (as yet) doing so is tripping over your own feet to reach a finish line that doesn’t exist. Sues & Averianov, 2015, all by refer Caenagnathasia martinsoni to Kuszholia mengi; however, Averianov (in Zelenkov & Averianov, 2011, stress their inference of the latter taxon as a troodontid, and with the presence of Urbacodon itemirensis (Averianov & Sues, 2007) in the sample it may very well be possible that these two taxa are unnecesarily conflated. It is a tactic that Mortimer perpetuates; the cause being noble, but overreaching.]
Averianov, A. O. & Sues, H.-D. 2007. A new troodontid (Dinosauria: Theropoda) from the Cenomanian of Uzbekistan, with a review of troodontid records from the territories of the former Soviet Union. Journal of Vertebrate Paleontology 27 (1): 87-98.
Currie, P. J., Godfrey, S. J. & Nessov, L. A. 1994. New caenagnathid (Dinosauria, Theropoda) specimens from the Upper Cretaceous of North America and Asia. Canadian Journal of Earth Sciences — Revue canadienne des sciences de la Terre 30 (10-11): 2255-2272.
Funston, G. F. & Currie. P. J. 2014. A previously undescribed caenagnathid mandible from the late Campanian of Alberta, and insights into the diet of Chirostenotes pergracilis (Dinosauria: Oviraptorosauria). Canadian Journal of Earth Sciences — Revue canadienne des sciences de la Terre 51 (2): 156-165.
Funston, G. F., Persons, W. S., IV, Bradley, G. J. & Currie, P. J. [2015, in press] New material of the large-bodied caenagnathid Caenagnathus collinsi from the Dinosaur Park Formation of Alberta, Canada. Cretaceous Research.
Longrich, N. R., Barnes, K., Clark, S. & Millar, L. 2013. Caenagnathidae from the Upper Campanian Aguja Formation of west Texas, and a revision of the Caenagnathinae. Bulletin of the Peabody Museum of Natural History 54 (1): 23-49.
Nessov, L. A. 1992. [A review of localities with bird remains in the Mesozoic and Paleogene of the USSR and description of new records.] Русский Oрнитологический Журнал — Russian Ornithological Journal 1 (1): 7-50. [in Russian]
Scapino, R. 1981. Morphological investigation into functions of the jaw symphysis in carnivorans. Journal of Morphology 167: 339-375.
Sues, H.-D. & Averianov, A. O. [2015, in press] New material of Caenagnathasia martinsoni (Dinosauria: Theropoda: Oviraptorosauria) from the Bissekty Formation )Upper Cretaceous: Turonian) of Uzbekistan. Cretaceous Research.
Zelenkov, N. V. & Averianov, A. O. 2011. [Synsacrum of a primitive bird from the Upper Cretaceous of Uzbekistan.] Палеонтологический Зурнал – Paleontological Journal 2011 (3): 75-80. [Published in English edition Paleontological Journal 45 (3): 314-319.]