Are Scansoriopterygids Oviraptorosaurs?

Scansoriopterygidae represent one of those bizarre groups of animals that seem to defy simplistic evaluation; there’s always something about them that says “You should compare with that that other group” whenever you look at a part. It doesn’t help that two specimens are from relatively juvenile-seeming animals, there are only three specimens and each is the holotype of its own species, and that they share a mix of derived and seeming basal characters when it comes to maniraptoran theropod dinosaurs. Of particular interest, though, has been the relatively short skull, extremely long arms with even longer hands and a gigantic third finger, short legs and toes but a long hallux, and strange integument. The latest attempt at figuring out what these yokels are comes from Drs. Federico Agnolín and Fernando Novas, who published their study in an eBook through SpringerVerlag that came out last week. This attempt seems to peg their affinities with oviraptorosaurs, a proposition that intrigues me as it has others. What follows below is an analysis of their justifications for the relationship.

First, the particulars:

Scansoriopterygidae, including all referred taxa. Scansoriopteryx heilmanni is often considered synonymous with Epidendrosaurus ningchengensis, both are probably juveniles, and they may in turn be synonymous with Epidexipteryx hui (in which case Scansoriopteryx heilmanni has priority above both other taxa).

Scansoriopterygidae, including all referred taxa. Scansoriopteryx heilmanni is often considered synonymous with Epidendrosaurus ningchengensis, both are probably juveniles, and they may in turn be synonymous with Epidexipteryx hui (in which case Scansoriopteryx heilmanni has priority above both other taxa).

Agnolín & Novas (2013) produce a new phylogeny for Coelurosauria, comprised of taxa ranging from Allosaurus fragilis into basal birds, such as the uncontroverisan Protopteryx fengningensis. Using 429 character 88 OTUs, including several undescribed species but also several “lumped” taxa such as Sinraptor and Caudipteryx (for which purported variation has been noted), the authors arrived at 50 MPTs of a minimum step length of 1841, which is fairly good (fewer trees with a high volume of data is a good goal, as long as the characters are correctly formulated and that which are multistate are treated correctly when serial or variable (ordered vs. unordered). Ignoring much of the analysis, which Andrea Cau has his eyes on, I focused on oviraptorosaurs (surprise!).

This analysis purports a clade comprising Oviraptorosauria and Scansoriopterygidae. The latter clade is comprised of three taxa: Epidendrosaurus ningchengensis (Zhang et al., 2002), Scansoriopteryx heilmanni (Czerkas & Yuan, 2002) — commonly considered a sernior synonym to the first taxon, but by Agnolín & Novas (2013) it is a junior synonym instead — and Epidexipteryx hui (Zhang et al., 2008); the authors only code for the first and the third, including the second in the first. The skulls of these taxa have been broadly compared to oviraptorosaurs, and for many reasons the similarities are quite striking. In some respects. In others, not so much. Agnolín & Novas find a clade comprising oviraptorosaurs and scansoriopterygids and substantiate it on twelve characters in the main discussion (for numbers following points, characters and followed by states separated by a period):

1. 20.0 – Maxilla and lacrimal contact, forming subnarial margin to external bony naris;
2. 33.1 – Posterior ramus of jugal “rodlike”;
3. 39.1 – Enlarged lateral foramen of the lacrimal;
4. 68.1 – Coronoid process of the mandible;
5. 69.1 – Dentary with caudodorsal process dorsal to external mandibular fenestra;
6. 74.1 – Internal mandibular fenestra large and rounded;
7. 79.1 – Retroarticular process elongated and slender;
8. 255.1 – Dentary with caudoventral process ventral to external mandibular fenestra;
9. 269.1 – Acromion process of scapula does not contact coracoid;
10. 288.1 – Manual phalanx mdII-2 less than 1.2 times the length of mdII-1;
11. 427.1 – Main axis of external bony naris subvertical;
12. 428.1 – Parietal lacks transverse “nuchal” crest.

Of these, only 4 and 7 are substantially supported through optimization. But let me put these characters into context. First, however, let me note that Agnolín & Novas (2013) are not the first to support an oviraptorosaur + scansoriopterygid clade (which despite supporting no author has coined a name for, the same being true for the oviraptorosaur + therizinosauroid clade, which is fortunate as most recent analyses do not support an exclusive relationship among these taxa). Paul (2010) also suggested that scansoriopterygids were oviraptorosaurs, along with omnivoropterygids (sapeornithids).

Before we continue, though, a note:

Agnolín & Novas (2013) discuss “the Oviraptorosauria + Scansoriopteryigdae clade.” A clade comprising Oviraptorosauria + Scansoriopterygidae would be named “Oviraptorosauria,” by almost any given formulation of the former clade: defined as the most inclusive clade containing Oviraptor philoceratops (Osborn, 1924) but not Passer domesticus (Linnaeus, 1758) [Maryańska et al., 2002, after Currie & Padian, 1997 -- vide Barsbold, 1997 -- replacing "birds" with a specific species]. This branch-based definition means that Oviraptorosauria sucks up as much as it can that is closer to one particular internal taxon (in this case, the clade’s namesake Oviraptor philoceratops) than the other (a bird); were this a node-based clade, with another possible oviraptorosaur as an internal specifier but without an external specifier, Agnolín & Novas (2013) might have something, and as such the clade would be useful. But we’ve seen this problem before, and it was dealt with in a particular way, with regards to Therizinosauroidea. As early as 1994, with Russell and Dong’s description and analysis of Alxasaurus elesitaiensis, they produced a clade comprising of therizinosauroids and oviraptorosaurs. This concept has since fallen by the way side as it seems therizinosauroids may be a node to two more basal relative to birds than are oviraptorosaurs; but before this occurred, such a clade would also have been called “Oviraptorosauria.” Mickey Mortimer suggests using a modification of Hu et al.‘s 2009 definition, which adds to the above taxa the additional external specifier Therizinosaurus cheloniformis (Maleyev, 1954); but Sereno (2005, Taxon Search) has also suggested an incresed definition, one which increases the external specifiers further. This structure prevents therizinosauroids from falling into Oviraptorosauria should a phylogeny ever find them as sister taxa (as some older analyses have). This solution may work for scansoriopterygids, if further analysis supported this clade, but if so the definition of Oviraptorosauria would need to be redefined to retain its formal structure (as the authors clearly mean). One wonders, as I know others do, what stability is had if we keep playing around with definitions? Fewer specifiers, clearest intention, as early as possible for definitions is best, and for that, I am perfectly happy with the Maryańska et al. definition. Let the pieces fall where they may.

1. Maxilla/lacrimal contact

A contact between the caudal, subnarial process of the premaxilla and the rostral process of the lacrimal is diagnostic of Oviraptoridae. These bones also approach one another in other “short-faced” dinosaurs and archosaurs, such as Daemonosaurus chauliodus (Sues et al., 2011) and some notosuchians (e.g., Adamantinasuchus navae Nobre & Carvalho, 2006), but rarely do they contact. The two processes approach one another above the maxilla in Incisivosaurus gauthieri (Xu et al., 2002), so there is no contact though the rostral process of the lacrimal may be broken off, and thus not apparent in the holotype (it is further unknown in Protarchaeopteryx robusta Ji & Ji, 1998, for which surface material is mostly obliterated and suture margins certainly difficult to ascertain), but are well-separated from one another in any skull of Caudipteryx (zoui, dongi, or sp.) (see here for their skulls in comparison), in which they are in their plesiomorphic, far-removed positions. The condition of this feature in Similicaudipteryx yixianensis is ambiguous, as the only skull attributed to this species (STM22-6, Xu et al., 2010:Fig.1d), and thus I cannot assess it. Some ornithischians possess a premaxilla/lacrimal contact, especially ceratopsians (e.g., Auroraceratops rugosus You et al., 2005), the heterodontosaurs Heterodontosaurus tucki Crompton & Charig, 1962 and Tianyulong confuciusi Zheng et al., 2009, Psittacosaurus spp. (Sereno, 1986, 1988). This contact is superficial in taxa which have it, lying above the surfaces of the underlying nasal (in ornithischians) and maxilla (in oviraptorids and ornithischians).

Skull of IVPP V15471, holotype of Epidexipteryx hui Zhang et al., 2008. "B" and "C" including insets "B'" and "C'" are from Zhang et al., 2008, while at right "B" is rotated to the right and I've overlain it with colored layers and labels depicting bones, and teeth in bright red.

Skull of IVPP V15471, holotype of Epidexipteryx hui Zhang et al., 2008. “B” and “C” including insets “B’” and “C’” are from Zhang et al., 2008, while at right “B” is rotated to the right and I’ve overlain it with colored layers and labels depicting bones, and teeth in bright red.

When it comes to scansoriopterygids, this feature cannot be assessed in either type specimens of Epidendrosaurus ningchengensis and Scansoriopteryx heilmanni as the relevant regions are not preserved. The former preserves the lacrimal, but the anterior process is not complete, and does not reveal a contact with the premaxilla. In Epidexipteryx hui, the skull is preserved so that the cortical bone surface has separated from the underlying cancellous bone and preserved on each side of a split slab, so that the skull is preserved so that sections of the left side of the skull are preserved on both sides of the holotype. Furthermore, a split in the slab follows down through the external bony naris and bisects the maxilla and nasal, and thus makes the contact between bones around the naris hard to ascertain. A small section of the bone does, however, suggest that the premaxilla and nasal contact one another beneath the naris, while beneath them the maxilla and lacrimal contact one another; the exact form of this contact is obscure, and in the renderings above come down on the side of an absence of contact between premaxilla and lacrimal. This, then would suggest that while oviraptorosaurs may basally preserve the contact, this contact is weak and may vary from basal to basal taxon, is ambiguous in oviraptorosaurs generally or absent until oviraptorids, unknown in caenagnathids, and only certain in oviraptorids.

Further, it seems to me that increased contact of rostral bones of the skull to bones not normally contacted by them (intervened by the nasal and maxilla) is a result of facial shortening, as it occurs in various taxa across craniate animals. Indeed, regardless whether actinopterygian “fish,” lissamphibian, theropsid or sauroposid amniote, the bones of the rostralmost snout gain contact with bones further back normally not contacted as the face becomes abbreviated, a feature in common with skulls suited to producing high forces at the rostral tip of the snout.

Verdict: This character does not support an oviraptorosaur clade inclusive of or sister to scansoriopterygids. If coded for scansoriopterygids, this character should be recoded.

2. “Rod-like” posterior ramus of jugal

The caudal process of the jugal that contacts the quadratojugal is slender and rod-like, roughly rounded in section and very shallow, in several oviraptorids (e.g., Citipati osmolskae; Clark et al., 2002) but not in others (e.g., Khaan mckennai and other “conchoraptorines”; Balanoff & Norell, 2012), in which the process is very deep and in fact expands somewhat distally into a rhomboid shape of sorts. However, in other oviraptorosaurs, the condition is more gracile, and the process begins deep and triangular but shallows drastically into a rod-like aspect at the contact for the quadratojugal. The rostral process of the quadratojugal is very long in oviraptorosaurs, and comes fairly close to the corpus of the jugal (near to the convergence of ascending, dorsal process and caudal process; it remains “rod-like” for its length in most, if not all, oviraptorosaurs. Despite this, some oviraptorosaurs have shallow, “rod-like” caudal processes rostral to the jugal-quadratojugal contact, including Avimimus portentosus (Kurzanov, 1983; a possbile caenagnathid, as per Senter, 2007, and following analyses), Caudipteryx zoui (Ji et al., 1998, holotype and paratype, composited), but not Incisivosaurus gauthieri or a referred specimen of Caudipteryx, possibly a new species (Caudipteryx sp., BPM 0001, Zhou et al., 2000). Thus, the condition has variable expression within oviraptorosaurs, and not even the most basal taxon.

Epidendrosaurus ningchengensis lacks detailed information of this portion of the skull, preserving as it does seemingly only the mandible in alveolar (dorsal) view and the dorsal skull roof in dorsal view as a negative impression — as most of the skeleton is preserved in a negative impression and few real bones. Scansoriopteryx heilmanni and Epidexipteryx hui, however, preserved this region fairly well, and in both, the rostral process of the quadratojugal is slender and shallow, but the caudal process of the jugal is deep, rather than shallow, and as in some oviraptorids, in Scansoriopteryx heilmanni deepens caudally towards the contact with the quadratojugal. The quadratojugal is missing in Epidexipteryx hui, but the jugal is well-preserved and demonstrates a deep, caudally forked contact for the quadratojugal, and does not present a rod-like structure.

Verdict: This character does not support an oviraptorosaur clade inclusive of or sister to scansoriopterygids. If coded for scansoriopterygids, this character should be recoded.

3. Enlarged lateral foramen of the lacrimal

Several oviraptorosaurs (Incisivosaurus gauthieri, oviraptorids) have this large foramen, but others do not (Caudipteryx spp.), and is unknown in caenagnathids or Avimimus portentosus. The lacrimal normally presents a smaller foramen, as in dromaeosaurids, troodontids, even therizinosauroids, while in Scansoriopteryx heilmanni and Epidexipteryx hui, the forament is fairly large. In “basal paravians” such as Xiaotingia zhengi, Anchiornis huxleyi and Archaeopteryx lithographica, the foramen is small, slot like, or potentially absent. This is even the case in therizinosauroids, ornithomimosaurs, and tyrannosaurs;  While basally variable, the variability is low enough to place a reasonable inferrence on the foramen’s presence in species lacking a lacrimal (caenagnathids, Avimimus portentosus), regardless of Caudipteryx spp. Now, I am making an asusmption here about the nature of this “foramen” in oviraptorids, which seems to be a complex of smaller foramina that penetrate a broader “fossa”-like feature on the lateral surface of the lacrimal, and penetrating the bone from the rostral end into the body of the bony, and may have an ill-defined rostral margin. It does not resemble the clean, circular foramen with even margins present in scansoriopterygids, but rather seems an extension of the antorbital air sac that penetrates the lacrimal, forming a fossa across the ventral margin of the anterior process of the lacrimal and no clean ventral margin of the “foramen.” Thus, it would seem that the condition in oviraptorids is a variation of the morphology of the foramen that penetrates the lacrimal laterally other than just size, as Agnolín & Novas (2013) seem to suggest.

Verdict: On its face, this character supports an oviraptorosaur clade inclusive of or sister to scansoriopterygids. I would increase the variation of the character by adding a new one stressing the shape of the margin of the foramen, in which oviraptorids would generally differ even from basal oviraptorosaurs like Incisivosaurus gauthieri.

4. Coronoid process of the mandible

This character, unfortunately, suffers from lack of qualification. There is not merely a “coronoid process”; there is, instead, a range of features on the dorsal margin of the mandible that corresponds to the “coronoid” region of the mandible, or is even not even connected to the site of the coronoid bone (if present). This makes quantification of this feature problematic: it is not a mere binary issue. The expansion of the “coronoid region” can occur at the dentary/surangular suture, or even further caudally, sometimes 1/3 of the way back along the length of the surangular, or even halfway back. When the structure is associated with the coronoid bone, the eminence may mark the rostral position of the m. adductor mandibular externalis posterior on the dorsolateral surface of the surangular and the m. psuedotemporalis superficialis medially; when the “eminence” occurs further caudally, and potentially at the caudal end of the coronoid region, I have my doubts about this, and it seems the structure marks anterior end of the attahcment for the m. adductor mandibular externalis medialis, which is in line with the mAMEP, due in large part to the far caudal position in relation to the dentary-surangular contact, at which point mAMEP and mPstS should attach, and to which the coronoid bone, if present, is restricted. Thus, the structure doesn’t necessarily correspond to the “coronoid region,” and may not be homologous with it. Coding it as if it were may be an error.

Instead, I qualify this “character” thusly:

morphotype A. No extension dorsally.
morphotype B. Small nubbin” on dorsal margin of surangular; this structure typically doesn’t involve the coronoid bone itself.
morphotype C. Dorsal margin of the surangular has a “rounded margin” that forms a “hump” or there is a bit of a corner, such that the dorsal margin is divided into two “faces” at an angle, which itself can develop a “nubbin”; this structure typically doesn’t involve the coronoid bone itself.
morphotype D. Instead, the “rounded margin” is a flange, which is labiolingually compressed and may or may not be medially inflected; this structure typically doesn’t involve the coronoid bone itself.
morphotype E. Rather than restricted to the dorsal margin of the surangular, a “corner” forming an angle of two “faces” of the mandible occurs between the dentary and the surangular, and involves the actual coronoid bone (if present).
morphotype F. Elongated process or raised “eminence” of the surangular also involving the dentary, and quite laterally separated from the alveolar margin, such that occasionally the alveolar row passes *medially* to the process; this process usually involves the coronoid bone, but generally only basally.

To gain insight into this “structure”, I tabulated various taxa for this feature:

coronoid shape distribution

Table describing presence of various morphotypes of “coronoid” morphologies. Lavender shading indicates oviraptorosaurs, pink shading indicated scansoriopterygids.

For the two scansoriopterygid taxa that preserve surangulars and dentaries (the former is absent in Epidendrosaurus ningchengensis, and the latter preserved in a manner not conducive to examining for this feature), they both possess morphotype C, which is shared with some basal oviraptorosaurs, but not others, and those others either develop the “nubbin” of morphotype B dromaeosaurids and tyrannosauroids (set well back from the “coronoid” region, as in Avimimus portentosus) or a distinct flange (caenagnathids and oviraptorids). It is also questionable whether the feature is ontogenetic in nature, that the “nubbin” is an indication of developed muscle attachment and has no systematic significance, or otherwise. Not all dromaeosaurids have the nubbin, as indeed the “famous” MPC-D 100/25 skull of Velociraptor mongoliensis shows a dorsally curved surangular not unlike that of Caudipteryx zoui.

Verdict: This character supports an oviraptorosaur clade inclusive of or sister to scansoriopterygids. However, I would quantify this character and probably break it up into general morphology of the dorsal margin and relationship of structures to the coronoid bone (if present).

5. Caudodorsal process of dentary overlies EMF &
8. Caudoventral process of dentary underlies EMF

These characters describe the structure of the external mandibular fenestra (EMF) to the dentary, especially as the latter forks around the former — or rather, the extension of the former into the caudal margin of the latter. The typical morphology of the dorsal and ventral processes in theropod dinosaurs and basal birds is with the caudal margin of the dentary developing an extension of the ventral margin caudally, but this usually involves a sloped surangular/angular margin in lateral view. If an EMF is present, almost no change in this general arrangement occurs. I hesitate to use the term caudoventral process for the dentary in the ventral half, in such taxa as Tyrannosauroidea, or even when no EMF is present as is the case in Archaeopteryx lithographica. Instead, this should be a process that extends ventral to a portion of the EMF by fact of the caudal margin of the dentary becoming somewhat caudally concave. In this case, this is a continuous character describing the relationship of indentation of that margin by the EMF, which may be best quantified by morphometric shape analysis. If so, troodontids and dromaeosaurids are notable for possessing a very short or no caudoventral process, the ventral margin of the EMF comprised almost entirely of the angular or angular/splenial instead. The caudodorsal process, on the other hand, doesn’t seem to correspond to the EMF as particularly does the caudoventral process; this may be conflating the sometimes interdigitating structure of the dentary and surangular, which should form as a series of processes that extend from each bone and contact one another in a sinuous scarf join: the surangular is slotted medially to the dentary, but lateral to the coronoid bone (if present) and/or a medial process of the dentary, thus “holding” the surangular. This process may extend further onto the lateral surface of the surangular, as it does in tyrannosaurs, caenagnathids and oviraptorosaurs. Ventral to this, the surangular may send a small extension of itself to slightly overlap the dentary laterally, as it does in troodontids and tyrannosauroids, but also (amazingly) in caenagnathids and oviraptorids. This condition in the latter two indicates a strong interdigitation that restricted movement, even mediolateral splay during dorsoventral compression.

The elongation of these processes is related to the extension of the EMF rostrally, and the longer the processes the greater the length of the EMF; the two go hand-in-hand. Thus when a large EMF occurs, at least one of these processes should be present. Indeed, save for a few taxa (troodontids, Dromaeosaurus albertensis, Shuvuuia deserti and crocodilians, which lack the caudoventral process; or Ceratosaurus nasicornis, Erlikosaurus andrewsi and some dromaeosaurids, along with many basal avialaeans such as Xiaotingia zhengi or Archaeopteryx lithographica, which lack the caudodorsal) the two are always present together. These processes, likewise, co-occur and are quite long, in a range of taxa with elongated EMF relative to their jaw length, especially taxa with a lenticular (lens- or “kidney bean”-shaped) aspect to their EMF: Herrerasaurus ischigualastensis (in which the caudoventral process is longer than the caudodorsal), Eoraptor lunensis (even), Carnotaurus sastrei, Incisivosaurus gauthier, Caudipteryx sp., Caudipteryx zoui (where the EMF is shorter and more rhomboid), Caengnathus collinsi, Oviraptoridae (where the EMF assumes a more “cardiac” or heart shape), probably Avimimus portentosus, Scansoriopterygidae, Confuciusornithidae, Omnivoropterygidae (=Sapeornithidae), and purportedly the basal oviraptorosaur Similicaudipteryx yixianensis.

Verdict: This character supports an oviraptorosaur clade inclusive of or sister to scansoriopterygids. But then, it seems to be present in a lot of other taxa with short skulls (see character 1 for discussion) and huge EMF.

6. Internal mandibular fenestra large and rounded

This character is tied to two factors: relative height to length of the surangular, and whether the prearticular sends a process upwards to the coronoid region. The IMF is, for example, “large and rounded” in tyrannosauroids based on the former factor, in caenagnathids based on the latter, and oviraptorids because of both. The medial structure of the jaws in scansoriopterygids is unknown: the medial mandible of Epidexipteryx hui is partially visible, but it is impossible to determine the presence of this feature as the bones may not be fully visible. It is not possible, then, to determine whether the prearticular is exposed, or merely laying flat along the splenial as in oviraptorids and caenagnathids.

Verdict: This character does not support an oviraptorosaur clade inclusive of or sister to scansoriopterygids.

7. Retroarticular process elongated and slender

The retroarticular process begins rostrally at the caudal end of the articular fossa for the quadrate, and acts to extend the moment of the m. depressor mandibulae, which opens the jaw. mDM attaches normally to the dorsal and/or the medial surface of this process, which is normally comprised of the articular and/or surangular, but may be limited to the distal end. Normally among theropods the process is relatively short, being about as high as rostrocaudally long or shorter than that, and being roughly triangular. Oviraptorosaurs present three different morphologies: in Oviraptoridae, the process is very long and dorsoventrally compressed; in Caenagnathidae, it is long, somewhat ventrally deflected at its base, and mediolaterally compressed; and in Caudipteryx spp., Incisivosaurus gauthieri and Avimimus portentosus, it is quadrangular and relatively even in depth to width and barely longer than either dimension. However, the process’s structure in comparison to Scansoriopterygidae is lacking: in Scansoriopteryx heilmanni‘s holotype, the process is broken off, apparently lacking the articular portion, and may be of either the short, triangular form, the long quadrangular form, or the caenagnathid form; in Epidexipteryx hui, it may be short and blunted, and I’ve restored the process in Scansoriopteryx heilmanni to suit.

Verdict: This character does not support an oviraptorosaur clade inclusive of or sister to scansoriopterygids.

9. Acromion process of scapula does not contact coracoid

In Epidendrosaurus ningchengensis, the scapula is a dumbell-shaped bone with relatively evenly deep proximal and distal ends; in Scansoriopterys, the scapula is only partially preserved, and I’ve restored it as a proximal end with seeming “facets” representing the humeral glenoid, suture with the coracoid, and a blunt acromion process; in Epidexipteryx hui, hower, it seems likely that my identification of the Scansoriopteryx heilmanni scapula as a proximal end may be in error, as it also resembles the distal end of the more recently-described taxon. Indeed, the proximal end of the scapula in Epidexipteryx hui presents a tall, narrow acromion which lies alongside the coracoid for nearly their entire length, resembling the condition in Sinosauropteryx prima, Tyrannosauroidea, Ornithomimosauria, and Alvarezsauria (especially Haplocheirus sollers; Choinier et al., 2010) but rather distinct from that of the shallow proximal condition in oviraptorosaurs. In Protarchaeopteryx robusta (Ji & Ji, 1998) and Avimimus portentosus, the acromion is a a blunt, triangular process set back some distance from the coracoid, which agrees with this character, but if so, the condition in Epidexipteryx hui does not; and thus potentially also the condition in the other scansoriopterygids, suggesting many things: that their scapulae are not complete, immature and not developed in this feature, that the character is variable in scansoriopterygids, or that the clade is paraphyletic and that characters are constrained by phylogeny.

Verdict: If going by the morphology in the only reasonably complete (and only well-articulated) specimen, this character does not support an oviraptorosaur clade inclusive of or sister to scansoriopterygids. However, if going by the inclusion of the other taxa and taking their morphology as given, the condition is ambiguous or variable in Scansoriopterygidae (assuming it is monophyletic); excluding Epidexipteryx hui indicates the character is useful and supports this clade.

10. Manual phalanx mdII-2 less than 1.2 times the length of mdII-1

A survey of several coelurosaurians shows this character is not so succinct, and indeed presents a continuous and highly variable character. It also does not support the authors use of it. However, first, the data:

Distribution along a continuous series, describing

Distribution along a continuous series of the ratio manual phalanx mdII-2/mdII-1, where numbers closer to 0 representmuch shorter mdII-2 than II-1, and towards 2 much longer mdII-2s. Basal Avialae (orange, at top), showing specimens attributed to Shenzhouraptor sinensis and Archaeopteryx lithographica crossing a broad range, oviraptorosaurs in blue, therizinosauroids in red, scansoriopterygids in light orange, and basal paravians in light blue.

AL, Archaeopteryx lithographica; AH, Anchiornis huxleyi; D+T, dromaeosaurids and troodontids; MZ, Microraptor zhaoianus; SC, Scansoriopterygidae; SS, Shenzhouraptor sinensis; XT, Xiaotingia zhengi; YL, Yixianosaurus longimanus.

Scansoriopterygids, in light orange, range at the top end of the oviraptorosaur range, and despite their skew in sizes show a remarkable constraint in ratios: 1.63-1.68. Oviraptorosaurs, in red, instead skew across a broader range, from 0.81 to 1.68. The low end of the range, showing longer mdII-1s than mdII-2s, are caenagnathids (0.81-1), while the top end of the range is occupied by basal taxa (1.35-1.68); oviraptorids are scattered from 0.89 (Khaan mckennai) to 1.24 (Oviraptor philoceratops) and barely edge into basal range, suggesting a severe shift and shortening of the manual proportions with size, while the much longer mdII-2s in basal taxa are out of keeping with the variable size of their manus: Caudipteryx spp. have short hands, two functional manual digits, while Protarchaeopteryx robusta has larger hands, three functional digits, and longer arms. If taken as a scale, then it would seem that the very long arms and hands of scansoriopterygids may lead to the shorter hands and arms of oviraptorosaurns, and the transition is smooth (1.6-1.4), whereas the transition from within oviraptorosaurs is much more lumpy.

Verdict: This character supports an oviraptorosaur clade inclusive of or sister to scansoriopterygids. That is, if the continuous character holds out. I’d wonder if the relative length of digit to this porportion is viable as well, as that may better characterize the relative proportion of phalanges (that is, that the porportion MEANS something).

11. Main axis of external bony naris subvertical

The external naris’s long axis shifts around from horizontal to vertical largely dependant on the orientation and shape of the premaxilla. If this is true, then the shape of the premaxilla, or its subnarial depth to oral marginal length in lateral view (note: not alveolar length!) should give a clue as to the orientation of the external bony naris. It also depends on the orientation of the skull. Oviraptorosaurs demonstrate a ventrally-rotated preorbital region of the skull, and this influences the orientation to the “baseline”; change the baseline, and the orientation of the naris changes. Presumably, the “basal skull length” is used, as illustrated by Agnolín & Novas (2013) — thus, a line drawn from the distal end of the quadrate to the tip of the rostrum, or teeth. After correcting for orientation (using the method described here), I measured a host of skulls for three values:

1, their oral basal length (OBL), which is parallel to the adjusted long axis of the skull after re-orientation;
2, the subnarial height (SNH), which is perpendicular to OBL, and is measured regardless of depth of premaxilla, but measures the height from the ventralmost point of the external narial fenestra (ENF) to the oral margin; and
3, the angle of the long axis of the enf relative to OBL, measured as the left side of the skull and a measure of 0° as rostrally-oriented horizontal (thus, 90° is vertical, and measurements caudally increase from there, so °180 is caudally-oriented horizontal).

While doing this, I also determined the proportion of SNH/OBL, which measures relative height of the “premaxillary body” by proxy, or rather the relative aspect of the block of bone defined by these two measures, and a method for determining “verticality” of the rostrum.

Scatter plot of distribution of coelurosaurs (and some carnosaurs) based on the ratio of the subnarial height/oral basal length beneath the external narial fenestra, and against the angle of the ENF, described in the text. Below, the snout of Incisivosaurus gauthieri describes how the measurements are derived.

Scatter plot of distribution of coelurosaurs (and some carnosaurs) based on the ratio of the subnarial height/oral basal length beneath the external narial fenestra, and against the angle of the ENF, described in the text. Below, the snout of Incisivosaurus gauthieri describes how the measurements are derived.

As can be seen, Epidexipteryx hui (the only scansoriopterygid which could be measured, estimating extent of premaxillary bone) falls within three ranges: that of paravians (dromaeosaurids, troodontids and basal avialaeans), that of oviraptorosaurs (most of them oviraptorids), and that of all other maniraptoriforms included (Erlikosaurus andrewsi, Ornitholestes hermanni, Shuvuuia deserti and Gallimimus bullatus). The proportions and the angle of the ENF suggest equal affinity with each group, though within the oviraptorosaur plot the angle of the ENF (139°) measures closely with that of basal taxa (Incisivosaurus gauthieri, 146°; Caudipteryx sp., 138.5° ave.). Conchoraptorines, such as Big Beak but not Khaan mckennai, have extremely low values (94-112°), which comes close to the concept of “subvertcial”; whereas the “other” maniraptoriforms have extreme high values (132-157.5°), close to horizontal, and thus qualify as “subhorizontal,” though this is also true of most paravians in the sample, with Velociraptor mongoliensis (168°, for MPC-D 100/25) having the highest (i.e., most horizontal) value. Thus, if the range of values higher than 135° qualify as “subhorizontal,” and values less than that as “subvertical,” Epidexipteryx hui lacks a subvertical ENF, but so too would basal oviraptorosaurs. Agnolín & Novas (2013) do not qualify this character in this respect, but it would seem to me that the character may need recoding.

Verdict: This character supports an oviraptorosaur clade inclusive of or sister to scansoriopterygids, but only if it is qualified as a continuous character, or with more precise variation in coding. Otherwise, neither scansoriopterygids nor basal oviraptorosaurs possess a subvertical ENF.

12. Parietal lacks transverse “nuchal” crest

I am a bit stumped by this one, though. It’s the last character used. Presumably, it means the caudal margin of the parietal is not raised into a “lip,” a structure that forms the caudal wall to the supratemporal fossa as they pass over the dorsal parietal and converge near the midline. If this character is indeed qualified this way, then this condition is present in most, if not all ovirapoptorosaurs (but possibly not Avimimus portentosus), Epidexipteryx hui and Scansoriopteryx heilmanni, but also many avialaeans; it is absent (that is, there is a parietal “nuchal” crest) in Troodontidae proper (Saurornithoides mongoliensis, Zanabazar junior; Norell et al., 2009), Dromaeosauridae, and possibly even Archaeopteryx lithographica. However, this feature is absent in more derived (crown-ward) stem-birds. This condition is also present in outgroups to the proposed clade, such as Shuvuuia deserti, ornithomimosaurs, and Erlikosaurus andrewsi. It seems peculiar then, unless this character can be qualified another way.

Verdict: I would hazard that this character supports an oviraptorosaur clade inclusive of or sister to scansoriopterygids, but the character is not defined to such a point to take any variation in morphology into consideration: there is not merely a crest there. Indeed, a “raised lip” of the parietal/supraoccipital margin, an expansion only of the supraoccipital above the parietal, the formation of the thin lip not on the dorsal margin but rather on the lateral margins (a “split” crest), may all be viable options. Is the slight raising of the edges in Avimimus portentosus and Erlikosaurus andrewsi useful? or does the crest have to resemble tyrannosauroids or troodontids to count?

In addition, Agnolín & Novas (2013) found six more characters to support their clade, however the distribution of these are not optimal, as in some cases basal oviraptorosaurs or even scansoriopterygids do not possess the character states supported by the analysis:

13. 22.1 – Crenulate margin of buccal edge of premaxilla.
14. 58.1 – Paroccipital processes pendant, curving ventrally.
15. 85.3 – Absence of dentary teeth.
16. 263.1 – Cervical prezygapophyses flexed.
17. 355.1 – Palatine-pterygoid-ectopterygoid bar arches below ventral “cheek” margin.
18. 416.1 – Antorbital fossa rostrocaudal axis shorter than dorsoventral axis (higher than long).

As Agnolín & Novas (2013) did not consider these characters in their discussion on this clade, I will cover these only briefly.

13) A crenulate margin of the premaxilla occurs in both probable species of Caudipteryx (zoui and sp. — no skull is known for dongi), is absent in Incisivosaurus gauthieri, present in Avimimus portentosus, and variable in Oviraptoridae, as well as being unknown in Caenagnathidae. 14) The pendancy or not of the paroccipital processes of the opisthotics are unknown in scansoriopterygids, and the skull of Incisivosaurus gauthieri is mediolaterally crushed and lacks the ventral bowing of the ventral margin of the process as occurs in oviraptorids, caenagnathids and Avimimus portentosus. 15) Incisivosaurus gauthieri has dentary teeth, as indeed is purported for Similicaudipteryx yixianensis, suggesting the condition is plesiomorphic for Oviraptorosauria. 16) I covered this subject somewhat here when discussing another potential oviraptorosaur, and remain affixed to the conclusion that it is ambiguous; it is also not possible to confirm the condition in Epidexipteryx hui, for which the only known decent cervical vertebrae are present, as the specimens are split down the side and broken. 17) There is a little bit of bone between the preserved right side of the skull of Epidexipteryx hui and the jugal/quadratojugal, however this seems to match the distance with another strip of bone visible in the EMF of the mandible to the right mandible itself, suggesting this element is the opposite mandible; if so, the palatal structure cannot be determined in this specimen aside from a long series of blobs, colored light blue in the second figure above. It is further ambiguous whether the palatine descends below the oral margin in non-oviraptorid oviraptorosaurs, as it is further unclear if it does in caenganathids as the only known maxilla suggests the palate may be level, not everted. Paul (2010) presents a skull reconstruction for Caudipterux zoui and Sapeornis chaoyangensis (pp.148 and 145, respectively), but these are not based on clean, undistorted specimens, and the palate is not fully visible in any Caudipteryx skull. 18) This one has a germ of viability to it, as the AOF is higher than long in Epidexipteryx hui, and is about even in Incisivosaurus gauthieri, though it is much longer than high in Caudipteryx spp. and in Oviraptor philoceratops, suggesting it may be longer than high for much of oviraptorosaurs.

1. Not supported
2. Not supported
3. Supported
4. supported
5. Supported – qualified
6. Not supported
7. Not supported
8. Supported – qualified
9. Not supported – qualified
10. Supported
11. Supported
12. Ambiguous
13. Not supported
14. Not supported
15. Not supported
16. Not supported
17. Not supported
18. Probably not supported

So, my tally sits at 6 for, 5 against, 1 ambiguous for the main characters, and 6 for, 10 against, 1 probably against, and 1 ambiguous for the total characters; but of those “for”, two are qualified (heavily convergent character), and of those against one is qualified, which brings the “safe” votes to 4/4. I cannot, with just these characters, dismiss the argument of Agnolín & Novas (2013), though I remain very cautious in character coding and formulation. Not only are continuous characters being coded as binary (or multistate, but still), but it seems that qualification of these states are not very advanced. It is my eventual desire to produce, for the sake of itself and the sanity of researchers, an atlas of character codings used in oviraptorosaur analyses — and thus by extension most coelurosaur/basal avialaean analyses. If I did this, it would take years. But, that’s peace of mind for you. Indeed, Senter (2007) is one of the few researchers to regularly produce figures detailing character codings, but not every character or even state can be coded easily. Qualification of characters and their states, including continous ones, a justification of transforming continuous characters into multistate and a formula that helps this, and quantification of these by setting standards for measurement and deriving data, would be the ultimate goal, and this has uses outside of phylogenetic analysis. But what it could do is put us all on the same page.

Neoflightless Oviraptorosaurs?

But Agnolín & Novas (2013) aren’t the only ones to suggest this clade, as Paul (2010) dis so as well, as part of his argument for neoflightlessness in paravians: It was Paul’s assumption that the oviraptorosaur-like skulls of scansoriopterygids and omnivoropterygids represented a basal oviraptorosaur morphology, and that these taxa represented outgroups to the larger and more terrestrial oviraptorosaurs we’re all more familiar with. For Paul, the lineage was (in order):

OmnivoropterygidaeSapeornis chaoyangensis, inclusive of Omnivoropteryx sinousaorum and Didactylornis jii
ProtarchaeopterygidaeProtarchaeopteryx gauthieri (=Incisivosaurus) and robusta (following Senter, 2007)
Epidexipterygidae - Epidexipteryx hui (excluding Scansoriopterygidae)
Caudipterygidae (=Caudipteridae) – Caudipteryx zoui and yixianensis (=Similicaudipteryx)
AvimimidaeAvimimus portentosus
CaenagnathidaeMicrovenator celer, Hagryphus giganteus, Caenagnathus collinsi, Chirostenotes pergracilis, Elmisaurus rarus, Shangyangosaurus niupanggouensis (for more on this subject, see here), Caenagnathasia martinsoni, Nomingia gobiensis
OviraptoridaeGigantoraptor erlianensis, Oviraptor philoceratops, Citipati … well, I go into this bit in more detail here.

Paul’s argument for inclusion of omnivoropterygians with oviraptorosaurs came with a caveat: “Alternatively, the two groups were not closely related, and the heads and hands evolved in a convergent manner, or the omnivoropterygians were the flying descendants of oviraptorosaurs.” Nonetheless, ignoring the fact that without decent phylogenetic data and a massive step away from phyletic characterization of which characters might be “important” or not, Paul groups these taxa on the basis of

“Head not large, short and deep, sides of back of head made of slender struts, many bones including lower jaws fused together and extra joint absent, jaw joint highly mobile to allow chewing motions, teeth reduced or absent. Neck fairly long. Tail short. Arm short to very long, fingers three or two. Leg short to very long.”

Additionally, Paul suggests that the “[p]resence of large sternal plates, ossified sternal ribs, ossified uncinate processes, and short tail in most examples, and reduction of outer fingers in some examples” are useful. The reduction to only manual digits I and II (avian digits II and III, alular and major) with reduction of digit III (IV, minor) to two proximal and short phalanges occurs in Caudipteryx spp. and Sapeornis spp., but no other taxa, but despite this, the arms and their morphology are otherwise extremely different, while the two taxa show exteme differences in their pelvis, hindlimb, and caudal morphology — including the presence of a pygostyle longer than all free caudals in the latter. None of Paul’s broad features are useful in this respect save generalizations of cranial morphology, which incidentally leads to several features Agnolín & Novas (2013) described, and are discussed here.

From the Start?

Zhang et al. (2008) in their original description originally compared Epidexiptery hui to oviraptorosaurs, in total considering six characteristics before relying on their phylogenetic analysis to place the taxa within Paraves closer to birds than were dromaeosaurids:

1) a short and high skull, 2) ENF positioned high on skull, 3) parietal proportionately long, 4) rostral teeth much larger than caudal teeth, 5) symphysis of mandible inflected ventrally and dorsal margin convex, 6) large EMF.

Of these, I am particularly intrigued by #3, but not so much the others. For example, large EMFs are discussed above, and the presence of a large, robust premaxilla typically results in a high lateral position for the ENF, implying these factors are related, and because they are present in a large diversity of taxa — along with a “short and high skull” — they may have little phylogenetic informativeness. Ventral inflection of the mandibular symphysis does not occur in Incisivosaurus gauthieri (Xu et al., 2002) and a mandible lacking either lingual inflection and ventral deflection of the mandibular symphysis is plesiomorphic for oviraptorosaurs. When it comes to the length of the parietal, I am willing to note that this is useful, and influences a range of features, including the length of the supratemporal fossa/fenestra, the adductor chamber, all factors . However, ultimately I feel this may be related to craniofacial shortening, as the relative lengths of the other cranial bones (especially of the frontal and nasal) are shortened in relation to the lateral facial bones such as the jugal, postorbital, etc. Thus, ultimately, several of these characteristics are ultimately related to similar, but not identical, facial shortening.


Ultimately, I think scansoriopterygids are not oviraptorosaurs, and that even were phylogenetic analysis to bring them close to them, it may be related to the fact that some of the included taxa and their proportions are related to juvenile characteristics (including the overall large size of the skull). The organization of the shoulder, and perhaps even of the somewhat mesopubic pelvis, may suggest that scansoriopteryigds present (at their size or relative age) a basal or pre-maniraptoran morphology, similar to alvarezsaurids; alternately, they may instead be relatively bird-like near-avialaeans. But I do not think there is enough data to suggest that they are oviraptorosaurs, or really near-oviraptorosaurs.

Agnolín, F. L. & Novas, F. E. 2013. Avian Ancestors – A Review of the Phylogenetic Relationships of the Theropods Unenlagiidae, Microraptoria, Anchiornis and Scansoriopterygidae. SpringerBriefs in Earth System Sciences, SpringerVerlag.
Balanoff, A. M. & Norell, M. A. 2012. Osteology of Khaan mckennai (Oviraptorosauria: Theropoda). Bulletin of the American Museum of Natural History 372:1-77.
Barsbold R. 1997. Oviraptorosauria. pp.505-509 in Currie, P. J. & Padian, K. (eds.) Encyclopedia of Dinosaurs. Academic Press, San Diego.
Choiniere, J. N., Xu X., Clark, J. M., Forster, C. A., Guo Y. & Han F.-l. 2010. A basal alvarezsauroid theropod from the Early Late Jurassic of Xinjiang, China. Science 327:571-574.
Clark, J. M., Norell, M. A. & Rowe, T. 2002. Cranial osteology of Citipati osmolskae (Theropoda, Oviraptorosauria) and a reinterpretation of the holotype of Oviraptor philoceratops. American Museum Novitates 3364:1-24.
Crompton, A. W. & Charig, A. J. 1962. A new ornithischian from the Upper Triassic of South Africa. Nature 196:1074-1077.
Czerkas, S. A. & and Yuan C.-q. 2002. An arboreal maniraptoran from northeast China. in Czerkas, S. J. (ed.) Feathered Dinosaurs and the Origin of Flight. The Dinosaur Museum Journal 1:63-95.
Hu D.-y., Hou L.-h., Zhang L.-j. & Xu X. 2009. A pre-Archaeopteryx troodontid theropod from China with long feathers on the metatarsus. Nature 461:640-643.
Ji Q. & Ji S.-’a. 1997. Protarchaeopterygid bird (Protarchaeopteryx gen. nov.)–fossil remains of archaeopterygids from China. Chinese Geology 238:38-41.
Kurzanov, S. M. 1985. Строение черепа динозавра Avimimus [The skull structure of the dinosaur Avimimus]. Палеонтологический Журнал — Palyeontological Zhurnal 19:92-99. [in Russian]
Maleyev, Y. A. 1954. Новые черепообразные ящер Монголий [A new turtle-like reptile from Mongolia]. Priroda 1954:106-108.
Maryańska, T., Osmólska, H. & Wolsan, M. 2002. Avialan status for Oviraptorosauria. Acta Palaeontologica Polonica 47(1):97-116.
Nobre, P. H. & Carvalho, I. de S. 2006. Adamantinasuchus navae: A new Gondwanan Crocodylomorpha (Mesoeucrocodylia) from the Late Cretaceous of Brazil. Gondwana Research 10(3-4):370–378.
Norell, M. A., Makovicky, P. J., Bever, G. S., Balanoff, A. M., Clark, J. M., Barsbold R. & Rowe, T. 2009. A review of the Mongolian Cretaceous dinosaur Saurornithoides (Troodontidae: Theropoda). American Museum Novitates 3654:1-63.
Osborn, H. F. 1924. Three new Theropoda, Protoceratops Zone, central Mongolia. American Museum Novitates 144:1-12.
Paul, G. S. 2010. The Princeton Field Guide to Dinosaurs. Princeton University Press, Princeton, NJ (USA).
Russell, D. A. & Dong Z.−m. 1994. The affinities of a new theropod from the Alxa Desert, Inner Mongolia, People’s Republic of China. in Currie, P. J. (ed.) Results from the Sino−Canadian Dinosaur Project. Revue canadienne de sciences de la Terre — Canadian Journal of Earth Sciences 30(10–11):2107–2127. [dated 1993, but released in January 1994]
Senter, P. 2007. A new look at the phylogeny of Coelurosauria (Dinosauria: Theropoda). Journal of Systematic Palaeontology 5(4):429-463.
Sereno, P. C. 1986. Phylogeny of the bird-hipped dinosaurs. National Geographic Research 2:234-256.
Sereno, P. C. & Chao, S. 1988. Psittacosaurus xinjiangensis (Ornithischia: Ceratopsia), a new psittacosaur from the Lower Cretaceous of northwestern China. Journal of Vertebrate Paleontology 8(2):353-365.
Xu X., Zheng X.-t. & You H.-l. 2010. Exceptional dinosaur fossils show ontogenetic development of early feathers. Nature 464:1338-1341.
You H.-l., Li D., Ji Q., Lamanna, M. C. & Dodson, P. 2005. On a new genus of basal neoceratopsian dinosaur from the Early Cretaceous of Gansu Province, China. Acta Geologica Sinica 79(5):593-597.
Zhang F.-c., Zhou Z.-h., Xu X. & Wang X.-l. 2002. A juvenile coelurosaurian theropod from China indicates arboreal habits. Naturwissenschaften 89(9):394-398.
Zhang, F.-c., Zhou Z.-g., Xu X., Wang X.-l. & Sullivan, C. 2008. A bizarre Jurassic maniraptoran from China with elongate ribbon-like feathers. Nature 455:1105-1108.
Zheng X.-t., You H.-l., Xu X. & Dong Z.-m. 2009. An Early Cretaceous heterodontosaurid dinosaur with filamentous integumentary structures. Nature 458:333-336.
Zhou Z.-h., Wang X.-l., Zhang F.-c. & Xu X. 2000. Important features of Caudipteryx — evidence from two nearly complete new specimens. Vertebrata PalAsiatica 38(4):241-254.

About these ads
This entry was posted in Morphometrics, Paleontology and tagged , , , . Bookmark the permalink.

10 Responses to Are Scansoriopterygids Oviraptorosaurs?

  1. Alessio says:

    Frankly, i always assumed they were some kind of oviraptorosaurs, but maybe you’re right and they’re only related… Didn’t know they haven’t found an adult specimen yet, though.

    • Epidexipteryx hui is probably an adult, or subadult at the least as the authors suggest, and supports some info regarding the “adult” morphology … but it’s a little ambiguous. The holotype is a slab split down the middle of one side, so the bones are imperfectly exposed, and that makes it difficult even gauge surface morphology.

  2. Andrea Cau says:

    Very good post! I agree that an “oviraptorosaur-scansoriopterygid” clade (termed in my blog with the horrible name “Magnoviraptorosauria” – although I agree that the current definition of Oviraptorosauria makes the latter the right name) is not the most parsimonious interpretation of the known data. I tested it with my analysis, and result very unparsimonious. So, their similarities are probably due to convergence or are maniraptoran symplesiomorphies. As I mentioned in my post on the Agnolin & Novas (2013) analysis, their published data matrix does not produce an “oviraptorosaurian-scansoriopterygid” clade when is tested in TNT or PAUP: instead, a “scansoriopterygid-alvarezsaurid” node results.

  3. IanC says:

    Great analysis, and as with a fair few of your posts, you should publish this!

    Second – you mention a possible atlas of Oviraptor characters. Have you thought about using MorphoBank for this – if you publish a phylogenetic analysis, all the character data can be recorded there both as the data matrix, and photos/drawings showing the states for each character/taxon…

    • Yes, but this would require me to acquire visual elements of a host of oviraptorids, something I cannot do until I get out to the various musea they are in, especially Mongolia and New York. When that happens, I will definitely do this.

  4. Scansoriopterygids have been paired with just abut all non-ornithurine maniraptor clades I can think off, from therizinosaurs to enantiornithes, the latter being my favourite interpretation thus far.

    So much mystery and work for a clade that probably only lasted two million years or so.

  5. Pingback: The Fisherman & the Sinosauropteryx | The Bite Stuff

  6. Pingback: The Changeling – The Analysis | The Bite Stuff

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s