Comments by Mickey Mortimer

Etymology

"primitive hollow-tail from Chuandong", after Chuandong, where the remains were found.


Holotype

(CCG 20010, subadult) (2.4 m, 12 kg) third cervical vertebra (61 mm), tenth cervical vertebra (69 mm), third dorsal centrum (65 mm), fourth dorsal vertebra (58 mm), proximal caudal vertebra (60 mm), partial scapula (282 mm), partial ilium, proximal pubis, proximal ischium, femur (201 mm), tibia (231 mm), astragalus, calcaneum, incomplete metatarsal II, phalanx II-1 (30 mm), metatarsal III (122 mm), metatarsal IV (114 mm), phalanx IV-1 (16.5 mm), phalanx IV-2 (9 mm), phalanx IV-3 (10 mm), phalanx IV-4 (12 mm), pedal ungual IV (18 mm)

Diagnosis

Lateral depression of third cervical extends over 60% of central length; small anterodorsally projecting process anterior to proximal caudal neural spines (also in some carnosaurs); scapula less than 5.5 times longer than wide; base of preacetabular process less than half acetabular height; prominent anteriorly projecting process lateral to anterior trochantor on femur; medial condyle much larger than lateral condyle on tibia.

Description

Although described by He Xinlu back in 1984, Chuandongocoelurus has been largely ignored in the literature.  Comparing the femoral length to Elaphrosaurus, it can be estimated to have measured 2.4 meters in length.  Scaling from Paul's (1988) estimated mass of Elaphrosaurus results in a weight of 12 kilograms.  The unfused neurocentral sutures in the dorsal vertebrae show this was a subadult, so this was not its maximum size.

Of the two cervical vertebrae preserved, one is clearly from a very anterior position due to it's slender and short centrum.  Glut (1997) refers to an axis, but the strong parapophyses suggest it was a third cervical instead, as Carnotaurus and Dilophosaurus have very reduced axial parapophyses (the latter has none).  The centrum is perhaps flat anteriorly and slightly convex posteriorly, with an anterior face 38% wider than tall.  There is a deep lateral depression (pleurocoel?) extending from the posterior border of the centrum to beneath the diapophyses.  The diapophyses extend ventrolaterally to almost contact the parapophyses.  The prezygopophyses are broken off, but thir bases show they were massive.  The posyzygopophyses are also broken and were more slender.  There is a large postzygopophyseal-central choana that has a step on the ventral edge.  The neural spine is not well preserved, but was low and rounded.

The other cervical is from the posterior portion of the series as seen by the postzygopophyses extending past the centrum and the craniocaudally short neural spine.  Comparison with Elaphrosaurus suggests it is the tenth.  The centrum is again elongate, with a slight lateral depression.  The anterior face is perhaps slightly convex and 38% wider than tall, while the posterior face is perhaps flat or concave.  The parapophyses are massive and positioned on the anteroventral corners.  There is a circular neural canal, a bit less than 40% as tall as the central face.  The prezygopophyses are quite massive compared to Elaphrosaurus, though broken at their tips.  The ventrolaterally projecting diapophyses end much further from the parapophyses than in the third cervical.  The cervical ribs were apparently unfused to the vertebrae.  The neural spine is low and short craniocaudally, although only its base remains.  There is a large postzygopophyseal-central choana again, but no step this time.  There is no evidence of epipophyses.

A centrum is probably from the third dorsal, as the parapophyses are partially on the centrum and partially on the neural arch, as seen in the third dorsals of Elaphrosaurus and Dilophosaurus. There is a slight lateral depression, the anterior face is 14% wider than tall and it looks slightly opisthocoelous.

The other dorsal has a parapophysis placed on the neural arch, but not at the dorsal edge of the prezygopophysis.  This is seen in the fourth dorsal of Elaphrosaurus. The centrum is again rather elongate, with a small lateral depression and indeterminate face convexity.  The ventral edge is more strongly concave than Elaphrosaurus.  The diapophyses project laterally and appear backswept.  The prezygopophyses are very short, but still more massive than Elaphrosaurus.  A large postzygopophyseal-central choana is still present, with a step like the third cervical.  There are large postzygopophyses and a moderate sized rectangular neural spine, with ventral margins sloping towards the zygopophyses, especially the postzygopophyses.

The caudal vertebra is probably from around the fifth position, judging by elongation.  The centrum has no lateral depression and a moderately concave ventral surface.  The anterior face is perhaps concave while the posterior is slightly convex.  There are prominent transverse processes and the bases of large prezygopophyses.  The neural spine is craniocaudally expansive and there looks to be a small anterior spine medial to the prezygopophyses.  Postzygopophyses are broken off.

The scapula is very broad for a theropod (~5.1 times longer than broad), which is more than even abelisaurs and megalosaurs (~5.9 times).  However, it lacks the expanded distal end of coelophysoids, Dilophosaurus and basal non-theropod dinosaurs and is thus still strap-like.  Most of the anterior edge is lacking, but it is generally similar to Carnotaurus, differing in the slightly concave posterior margin.  An extensive posteriorly facing glenoid is present.

The pelvis is unfused and propubic, with the pubis about 27 degrees from the vertical.  The preacetabular process is broken, but extended past the pubic peduncle and is dorsoventrally narrow, less than half of acetabular height.  The dorsal edge of the ilium is fairly straight and there is no vertical ridge or other ornamentation laterally.  The large pubic peduncle projects anteroventrally and is slightly expanded distally.  There is an extensive supracetabular shelf that ends halfway through the ischial peduncle.  The ischial peduncle is craniocaudally 42% as long as the pubic peduncle.  The postacetabular process is broken off, but could not have been very tall.

Only the proximal section of the pubis is preserved.  The dorsal margin of an obturator fenestra can be seen, although with the ventral margin incomplete, it could be an obturator notch.  There is no room for a pubic fenestra below it.

Only the most proximal section of the ilium is preserved.  Oddly, the ilial peduncle is much wider than the pubic peduncle, contrasting with the peduncles they attach to.  No conclusion regarding obturator processes or flanges can be reached.

The femur is hollow and sigmoid in medial view.  The head is anteroposteriorly narrow, taking up less than half the proximal end in medial view.  The anterior trochantor is a bit more massive than Dilophosaurus in proximal view and is hooked medially.  There seems to be a smaller process directly lateral to the anterior trochantor.  A trochanteric shelf is apparently absent.  The fourth trochantor is smaller than Dilophosaurus, but much larger than the reduced process in Elaphrosaurus.  Distally, the femur shows a very slight extensor groove and a deep rounded flexor groove without a cruciate ridge.  A mediodistal crest is present, as in neoceratosaurs, but its proximal extent is hard to judge.

The tibia is quite elongate (15% longer than the femur) and slightly bowed laterally.  The proximal end is craniocaudally elongate, has a single cranial cnemial crest and the lateral condyle is smaller than the medial condyle.  The fibular crest cannot be distinguished.   Distally, the astragalus backed the tibia.  In distal view, the tibia is very anteroposteriorly narrow and roughly triangular.

The tibia, astragalus and calcaneum were unfused to each other.  In the astragalus, the medial condyle is much larger, there seems to be no transverse groove extending across the condyles and the condyles are separated from the ascending process by a groove or excavation.  The extent of the ascending process is uncertain, as it is broken proximally.  However, it was obviously more prominent than ornithopods (contra Norman, 1990) and a bit more extensive than Dilophosaurus, possibly making the astragalus 83-93% as tall as wide.  The anterior concavity of the astragalus in distal view is not as developed as coelurosaurs.  The calcaneum was large and rectangular.

The metatarsus, though elongate (61% of femoral length), is not arctometatarsalian.  Indeed, the third metatarsal is wider proximally than distally.  I am wary of He's pedal reconstruction however, as metatarsal IV's distal end seems to be in lateral/medial view, as might metatarsal III's.  Also, metatarsal II's broken distal end appears to be in posterior view.  All this in a reconstruction of anterior view.  Maybe they were twisted due to post-burial deformation or illustrated inaccurately.  The proximal ends of metatarsals II and IV are flared outward.  In proximal view, metatarsal II is more tapered anteriorly, metatarsal III more narrow, with a less expanded posterior end, and metatarsal IV is wider and more wedge-shaped than in Elaphrosaurus. A phalanx, identified as II-1, is shown backwards as its ginglymoid articulation is facing proximally.  This phalanx probably is II-1, as it is nearly identical in comparative size and shape to that element in Elaphrosaurus. A series of apparently articulated phalanges are identified as digit IV.  IV-1 is the largest and IV-2 is the shortest.  They are all fairly similar in morphology, with lateral ligament pits and ginglymoid distal articulations where can be seen.  The fact that IV-2 is the shortest leads to doubt they are articulated correctly, as this is never seen in theropods.  An ungual phalanx is also preserved.  It is short and gently curved, with no obvious flexor tubercle.

Relationships

Not many authors have attempted to classify Chuandongocoelurus.  He (1984) apparently assigned it to the Coeluridae, though I cannot read the Chinese text.  Such an assignment is obviously based on small size, as the Coeluridae was the diminutive equivalent of the Megalosauridae at the time.  In actuality, Chuandongocoelurus is much more primitive than Coelurus, although the fragmentary remains of the former and poor description of the latter make this hard to demonstrate directly.  The only other author to hypothesize as to Chuandongocoelurus' relationships is Norman, who referred it to Theropoda indet., while noting the primitively broad scapula, low ascending process and uncompressed third metatarsal.  As noted above, although broad, the scapula is still strap-like.  Additionally, it is much broader than any other theropod or basal dinosaur and is thus an autapomorphy of the genus.  The low ascending process is actually broken, though it would be of ceratosauroid level if complete (after Welles and Long, 1974).  The third metatarsal is compressed in proximal view, leading to an hourglass shape seen in tetanurans and neoceratosaurs. 

All material of Chuandongocoelurus is obviously theropod, as seen by the hollow centra and femur, strap-like scapula, tall ascending process, compressed third metatarsal and other characters.  There are three competing hypotheses for basal theropod phylogeny-

(((Coelop, Diloph) (Elaph (Cerato, Abeli))) Tetan)        Holtz 2000, Sereno 1999

(Coelop (Diloph (Tetan (Cerato (Elaph, Abeli)))))         Rauhut 1999

(Coelop (Abeli (Cerato, Tetan))))                               Carrano and Sampson 1999

The latter two are unpublished, but Rauhut kindly sent me his thesis, so I can use it to test Chuandongocoelurus' phylogenetic relationships.

Using Holtz's (2000) phylogeny, Chuandongocoelurus is a ceratosaur based on- ilial supracetabular shelf; metatarsal III proximal area larger than II or IV.  It would lack the ceratosaurian characters of- ischial antitrochantor large; trochanteric shelf of femur well developed.  It also has the neoceratosaurian character of metatarsal III proximal surface dumbbell shaped.  It lacks the (Ceratosaurus + Abelisauroidea) characters- cervical centra markedly opisthocoelous; fourth trochantor little developed.  However, it has the abelisauroid characters- mid-cervical centra anterior faces more than 20% wider than tall; dorsal centra wider than high; scapular blade at least four times longer than midshaft width; iliac-ischial articulation smaller than iliac-pubic articulation.  It also has the tetanurine characters of- scapular blade at least four times longer than midshaft width; iliac-ischial articulation smaller than iliac-pubic articulation; femoral extensor groove present; trochanteric shelf absent; crista fibularis present.  It lacks the tetanurine character of an allosauroid tarsus.

The supracetabular shelf may be a valid ceratosaur synapomorphy, though it is also present in Herrerasaurus.  Metatarsal III has a proximal area greater than II and IV in basal tetanurans as well (Britt 1991, Dong and Currie 1993), so this is not a valid ceratosaurian character.  Chuandongocoelurus does appear to lack an ischial antitrochantor, although this is very possibly due to the illustration quality.  The absence of a trochanteric shelf could easily mean this was a gracile individual, as known individuals of Dilophosaurus for instance lack trochanteric shelves.  The presence of a dumbbell-shaped proximal metatarsal III with anterior and posterior ends expanded to slightly overlap the surfaces of metatarsals II and IV is a complex character. 

Both Ceratosaurus and Elaphrosaurus show strongly expanded posterior ends that overlap II and IV, but the anterior ends are expanded very slightly in Elaphrosaurus and vary from expanded moderately (IV) to the opposite condition (II) in Ceratosaurus.  Chuandongocoelurus has more of a dumbbell shape than Ceratosaurus and resembles Elaphrosaurus except the lack of a posterior expansion behind IV.  Further examination of this character is beyond the scope of this post. 

The fourth trochantor of Elaphrosaurus is reduced (Janensch, 1925), but not that of Ceratosaurus (Madsen and Welles, 2000), so I am confused by this character defining a (Ceratosaurus + Abelisauroidea) clade excluding Elaphrosaurus.  As Elaphrosaurus has all cervical centra at least 20% wider than tall (except the seventh- 11%), all dorsal centra wider than tall (except the fourth- 97%) and a craniocaudally longer pubic peduncle than ischial peduncle (the scapula is unknown), I think its lack of opisthocoelous cervicals is more likely to be a reversal. 

The tetanurine characters present in Chuandongocoelurus are also abelisauroid synapomorphies (discussed above), known in neoceratosaurians (extensor groove- Rauhut, 1999), gracile ceratosaurs (trochanteric shelf absent) or are present in at least some ceratosaurs (crista tibiofibularis present).  Thus, although somewhat problematic, Holtz's analysis supports Chuandongocoelurus as an abelisauroid, assuming Elaphrosaurus is in this clade (as it was in Holtz, 1994).

In Sereno's (1999) phylogeny, Chuandongocoelurus would not be a ceratosaur because- pelvis not fused; ischial antitrochantor smaller than articular surface for ilium; femoral trochanteric shelf not trough-shaped; astragalocalcaneum unfused.  It is not coelophysid, as it has dorsal centra less than 2.5 times as long as tall, and is not coelophysine, as it has subrectangular dorsal neural spines.  It has the tetanurine characters of an iliopubic articulation anteroposteriorly longer than the ilioischial articulation and a third metatarsal with proximal width less than metatarsals II or IV at its narrowest point.  It lacks the neotetanurine character of prominent hypapophyses on anterior dorsals.  It has the coelurosaurian characters postaxial cervical centra not strongly opisthocoelous and no transverse groove across astragalar condyles, but lacks a medial ascending process edge at a high angle.

The lack of fusion could be due to the subadult age of Chuandongocoelurus, and the trochanteric shelf and ischial antitrochantor were discussed above.  The large iliopubic articulation is also seen in abelisauroids, as mentioned above.  The narrowness of metatarsal III proximally is also seen in Dilophosaurus, so this is not unheard of in ceratosaurs.  Elaphrosaurus has strongly amphicoelous cervicals and Dilophosaurus lacks a transverse astragalar groove, so these characters are not unique to coelurosaurs.  Thus, Sereno's analysis presents weak evidence against a coelophysid or neotetanurine identity.

In Rauhut's (1999) phylogeny, Chuandongocoelurus would be a member of the (Dilophosaurus + Ceratosauria + Tetanurae) clade based on- anterior faces of cervical centra much wider than tall.  It would be a member of the (Ceratosauria + Tetanurae) clade based on- femoral distal condyles well rounded in lateral view; distal tibia strongly expanded mediolaterally and braodly triangular.  However, it lacks a substantial femoral extensor groove, unlike this clade.  It is a ceratosaur (equivalent to traditional Ceratosauroidea or Neoceratosauria) based on- scapula not expanded distally.  It differs from ceratosaurs in lacking a transverse groove across the astragalar condyles.  It shares the following characters with Elaphrosaurus- dorsal pleurocoels absent; mid-caudal neural spines subrectangular and sheet-like.  It is not tetanurine based on- fibular condyle of tibia not offset from cnemial crest; ascending process lower than astragalar body.

The presence of extensor grooves in theropods seems to be sporadic.  They are absent in Liliensternus, Dilophosaurus and spinosaurids (as well as some coelurosaurs), but present in Segisaurus, Syntarsus, ceratosaurs and most basal tetanurans.  Thus, I don't think we can conclusively say its presence is a synapomorphy of a Ceratosauria + Tetanurae clade.  Although unlike other ceratosaurs in lacking a strong extensor groove, the difference between Liliensternus and Syntarsus for this character shows it can vary within closely related taxa. 

The absence of a transverse astragalar groove may be due to the illustration quality, as only a few horizontal lines are used to indicate the texture in this area.  Rauhut's characters also leave many questions, though Chuandongocoelurus seems close to Elaphrosaurus.  Entering Chuandongocoelurus into Rauhut's matrix results in 22176 trees, with Chuandongocoelurus a member of the (Ceratosauria + Tetanurae) clade, but excluded from the Carnosauria and (Coelurus + Compsognathidae + Tyrannoraptora) clade.  As carnosaurs include megalosaurs and spinosaurs in Rauhut's trees, this leaves the Ceratosauria, extremely basal Tetanurae (along with Piatnitzkysaurus and "Szechuanosaurus" zigongensis) and extremely basal Coelurosauria (along with Proceratosauria).

It seems there are still some problems to work out with basal theropod relationships (work in progress...), but the above analyses all seem to place Chuandongocoelurus in the vacinity of abelisauroids.

Abelisauroid characters- cervical neural spines project only slightly above neural arch; preacetabular process less than 1.5 times as high as acetabulum; on ilium, ischial peduncle much narrower craniocaudally than pubic peduncle.

Compared to Ceratosaurus and Carnotaurus, Elaphrosaurus and Chuandongocoelurus share the following synapomorphies- elongate anterior dorsal centra (posterior central face height <65% of central length); anterior cervicals with low rounded neural spines; proximal caudal neural spines elongate anteroposteriorly, extending over ~2/3 of central length; femoral shaft strongly sigmoid; mediodistal crest of femur extends posterodistally across medial surface, not along anteromedial edge; deep groove posteriorly between lateral and medial tibial condyles; laterally hooked tip of cnemial crest.

Xenotarsosaurus also lacks the hindlimb characters.  Coelophysids developed elongate anterior dorsal centra and very low rounded cervical neural spines in parallel, as they are absent in Herrerasaurus, Dilophosaurus and basal tetanurines.  Herrerasaurids also developed long proximal caudal neural spines.  The last two characters are also developed in tetanurines.

Based on other evidence, Ligabueino, Masiakasaurus, Noasaurus, Velocisaurus and probably Elaphrosaurus form a monophyletic group, perhaps best termed the Noasauridae.  The lack of comparable material in many taxa makes their interrelationships difficult to establish, but characters shared by some include the hyperextendable second pedal digit of Ligabueino and Noasaurus, the tapered preacetabular process of Elaphrosaurus and Ligabueino, and especially the reduced lateral metatarsals of Masiakasaurus, Noasaurus and Velocisaurus (also seen to a lesser degree in the fourth of Elaphrosaurus).  Their poor preservation and description makes it difficult to evauluate if they share any of the Chuandongocoelurus + Elaphrosaurus synapomorphies, but Masiakasaurus has low rounded anterior cervical neural spines and elongate anterior dorsal centra.  Ligabueino lacks a strongly sigmoid femur and the derived mediodistal crest morphology.

Chuandongocoelurus differs from Elaphrosaurus in several ways- large postzygopophyseal-central choana in presacral vertebrae; step in postzygopophyseal-central choana of anterior cervicals and anterior dorsals; lateral depression of third cervical extends over 60% of central length; posteroventral border of anterior cervical centra not convex; larger prezygopophyses on posterior cervicals to proximal caudals; small anterodorsally projecting process anterior to proximal caudal neural spines; preacetabular process less than half acetabular height; pubic peduncle expanded distally; postacetabular process narrower; obturator notch (or foramen) in pubis; ilial peduncle of ischium larger than pubic peduncle; long dorsoventral axis of femoral head angled less anterodorsally; prominent fourth trochantor; medial condyle larger than lateral condyle on tibia; fibular crest indistinct; no transverse groove across astragalar condyles; outer edges of metatarsals II and IV flared priximally; proximal end of metatarsal III wider than distal end; metatarsal IV broader than II proximally; metatarsal III without lateroplantar expansion to back IV.

Many of these are plesiomorphic or found in other theropods, but several are unique to Chuandongocoelurus and are listed above in the diagnosis.  Thus, Chuandongocoelurus is not indeterminate, as Norman (1990 and Rauhut (1999) have suggested.

In conclusion, Chuandongocoelurus is a neoceratosaur most closely related to Elaphrosaurus, although how closely related other noasaurids are to either of them is difficult to determine.  Perhaps Chuandongocoelurus and Elaphrosaurus should be classified as noasaurids, however I would like more work done on abelisauroid relationships before such taxonomic decisions are made.  The presence of early neoceratosaurs in Asia in not entirely unexpected, as Dandakosaurus and Lukousaurus may also be early representatives.

References

He, 1984. The vertebrate fossils of Sichuan: Sichuan Scientific and Technological Publishing House, 168 pgs.

Norman, David B. 1990.  Problematic Theropoda: ``Coelurosaurs''.  p. 280-305 in David B. Weishampel, et al. (eds.), The Dinosauria. University of California Press, Berkeley, Los Angeles, Oxford.