The Arguable Identity of Paraxenisaurus

Hi everyone.  In light of Cau's recent post on the supposed new Mexican deinocheirid "Paraxenisaurus normalensis" (Serrano-Brañas et al., 2020), I figured I'd check the taxon out to see what I thought.

The first thing you might note are the quotation marks surrounding its name, as this is yet another example of authors not including an lsid or reference to ZooBank in their electronic descriptions.  ICZN Article 8.5.3. states names published electronically must "be registered in the Official Register of Zoological Nomenclature (ZooBank) (see Article 78.2.4) and contain evidence in the work itself that such registration has occurred", and the pre-print is said to be in preparation for Volume 101 of Journal of South American Earth Sciences, cited as August 2020.  Thus it gets to join the ranks of "Thanos" and "Trierarchuncus"as theropods that will eventually be validly named this year.  But at least it's not stuck in the purgatory of twelve Scientific Reports Mesozoic theropods, which will never be physically published and thus will remain invalid unless outside action is taken.

One of the big takeaways from Cau's blogpost is that "I am doubtful about the possibility of referring these elements [the paratypes] to the same species of the holotype, since there are very few superimposable elements among the three specimens. Therefore, there is a risk that Paraxenisaurus , - understood as the sum of all three specimens - is a chimera."  After reading the paper, Andrea REALLY undersold this critique.  Here are the specimen materials lists, with the overlapping elements highlighted in matching colors-

(BENC 2/2-001; proposed holotype) proximal manual phalanx II-2 or III-3, partial astragalocalcaneum, partial metatarsal II, phalanx II-1 (115 mm), proximal phalanx II-2, partial metatarsal III, proximal phalanx III-3, distal metatarsal IV, phalanx IV-1 (104 mm), phalanx IV-3 (67 mm), phalanx IV-4 (45 mm), partial pedal ungual IV
(BENC 1/2-0054) distal metacarpal I, proximal phalanx I-1, partial manual ungual I, distal metacarpal II, distal phalanx II-2
(BENC 1/2-0091) several proximal caudal central fragments (66, 75, 76 mm), proximal metacarpal II, partial metacarpal III, distal femur (155 mm trans), distal metatarsal IV
(BENC 1/2-0092) several distal caudal vertebrae (70, 71 mm)
(BENC 30/2-001) pedal ungual II, pedal ungual III

As you can see, there's only one strict overlap, with BENC 1/2-0091 sharing a distal metatarsal IV with the proposed holotype, found ~14 kilometers away.  The paper lists no proposed apomorphies or unique combination of characters for distal metatarsal IV, and indeed the description states they preserve largely non-overlapping portions-

"In the holotype, the distal articular surface is fragmented (Figures 11a1 and 11a2); but in the referred specimen (BENC ½-0091), this surface is nicely preserved and has a non-ginglymoid outline (Figures 11b1 and 11b2). The medial condyle is mostly preserved in the holotype (Figure 11a3), but in the referred specimen it is completely broken (Figure 11b3). Conversely, the lateral condyle is broken in the holotype (Figure 11a4), but is well preserved in the referred specimen (Figure 11b4). Collateral ligament fossae are well developed on both condyles and have approximately the same size and depth (Figures 11a3 and 11b4). In cross-section, the shaft of metatarsal IV near the distal end is thicker dorsoventrally than wide."

Needless to say, metatarsal IV has a shaft which is deeper than wide in all ornithomimosaurs, and the preserved ligament fossae are on opposite sides in each specimen (medial in proposed holotype, lateral in 1/2-0091).  Below is a figure comparing the two Mexican specimens with Ornithomimus velox, with 1/2-0091 flipped so that all are comparable as left elements.  I don't see anything the "Paraxenisaurus" specimens have in common that could diagnose a taxon.

Left distal metatarsal IV of (left to right) intended "Paraxenisaurus normalensis" holotype BENC 2/2-001, intended "Paraxenisaurus normalensis" paratype BENC 1/2-0091 (right element flipped), and Ornithomomus velox holotype YPM 542 in (top to bottom) dorsal, lateral, ventral and medial views ("Paraxenisaurus" after Serrano-Brañas et al., 2020; Ornithomimus after Claessens and Loewen, 2016).

While no other elements are exactly matched, referred specimen BENC 30/2-001 does include pedal unguals II and III, while the intended holotype has pedal ungual IV.  These are again from different localities, although closer this time (~2.8 km), and this time we have characters listed in the diagnosis- 

"(9) distinctively broad and ventrally curved pedal unguals that angled downward with respect to the proximal articular surface and depending on the digit, the proximodorsal process becomes slightly enlarge and changes its position from nearly horizontal to mostly vertical, adopting a lipshaped appearance; and (10) pedal unguals with a rounded, large foramen on the medial side* and a deep ventral fossa that surrounds a strongly developed, ridge-like flexor tubercle."

Pedal unguals of (left to right) intended "Paraxenisaurus normalensis" paratype BENC30/2-001 right digit II, right digit III and intended "Paraxenisaurus normalensis" holotype BENC 2/2-001 left digit IV in (top to bottom) right, left, proximal and dorsal views (after Serrano-Brañas et al., 2020).  Green lines point to supposedly natural median foramen

Ventral curvature is plesiomorphic, the unguals of BENC 30/2-001 are not broader than other ornithomimosaurs', and ventral angling with the proximal end held vertically is common in theropods and present in e.g. Garudimimus and Beishanlong.  The proximodorsal process "changing its position" is using a difference between 30/2-001's mostly horizontal processes and the intended holotype's more vertical process as character, which in itself presupposes they are the same taxon.  The ventral fossa surrounding a ridge-like flexor tubercle is also present in Harpymimus, Garudimimus, Beishanlong and large Dinosaur Park unguals (NMC 1349, RTMP 1967.19.145) and is not shown in the intended holotype but is claimed to be "partially broken."  This leaves the medial foramen, which might be a valid character in unguals III and IV (II is damaged in that area), but might also be taphonomic, as there are many other small circular areas of damage (e.g. center of proximal surface of ungual IV).  While the two unguals in 30/2-001 are similar to each other, that of the intended holotype is more strongly curved, has that smaller more dorsally angled proximodorsal process, is wider in proximal view, and lacks the expanded ventral half characteristic of ornithomimosaurs that is present in the other specimen.  But even if these two pedal unguals are correctly referred, they are all that's present in specimen BENC 30/2-001.  So they get us nowhere in determining caudal, manual (besides proximal manual phalanx II-2 or III-3) or femoral morphology.

The final issue I noticed was the emphasis on "Paraxenisaurus" having a first pedal digit.  This would ironically be unlike Deinocheirus, but plesiomorphically shared with Nedcolbertia, "Grusimimus", Garudimimus, Beishanlong, Archaeornithomimus and Sinornithomimus.  The character state is based on metatarsal II, where "a facet on the posterior surface of the distal quarter of this shaft, indicates the presence of an articulation area for metatarsal I." The figure shows a longitudinal groove extending down the posterior center of distal metatarsal II, which as anyone who has scored taxa for Clarke's bird matrix could tell you, is not how non-birds attach their hallux to the metatarsus.  Hattori (2016) for instance writes in Allosaurus "there is no attachment scar corresponding to the metatarsal I fossa on either medial or plantar aspect of MT II" and in Citipati "there is no obvious attachment scar of MT I on either medial or plantar aspect of MT II."  Serrano-Brañas et al. state "in Garudimimus brevipes ... the attachment site is also placed in the same area as in Paraxenisaurus normalensis", but the feature in Garudimimus is a raised scar with sharp medial demarcation from the shaft.  As Middleton (2003) recognized, this scar is for the m. gastrocnemius, specifically the m. gastrocnemius pars medialis (Carrano and Hutchinson, 2002), and I'll note it's present even in Gallimimus which lacks pedal digit I (Osmolska et al., 1972: Plate XLIX Fig. 1b).  "Paraxenisaurus"'s groove is then more likely to be the m. flexor digitorum longus II tendon, which "passed through the ventral groove in its respective metatarsal to insert serially on each of the pedal phalanges" in e.g. Tyrannosaurus (Carrano and Hutchinson, 2002).  

Left metatarsal II in ventral view of (left to right) intended "Paraxenisaurus normalensis" holotype BENC 2/2-001 (after Serrano-Brañas et al., 2020; yellow arrow points to supposed articulation for metatarsal I), Garudimimus brevipes holotype IGM 100/13 (after Kobayashi and Barsbold, 2005; line points to supposed articulation for metatarsal I), Gallimimus bullatus ZPAL MgD-I/94 (after Osmolska et al., 1972), and Tyrannosaurus rex FMNH PR2081 (after Brochu, 2003).

What exactly is "Paraxenisaurus"?  Comparison is hindered by the specimens being figured mixed together, and the figures are not in numerical order in the preprint, being shown in the order of- 1, 10-19, 2, 20-23, 3-9.  In addition, the scale bars vary within the same figure (e.g. phalanx IV-1 is proximally ~61 mm wide in figure 14a but ~93 mm wide in figure 14e) and the listed measurements are different yet (e.g. IV-1 is listed as 83 mm wide).  Thus any composite reconstruction is necessarily approximate.  The supposed manual element is too fragmentary to give much information, but it is of the appropriate size and shape to be a proximal pedal phalanx I-1.  This would make more sense preservationally since the other material preserved in the specimen is all from the tarsus and pes.  It's a shame the astragalocalcaneum is not described better or figured in more views, as the dorsal (= proximal?) perspective has many broken surfaces and edges, so that e.g. the small calcaneum might be preservational.  The fused proximal tarsals are like ceratosaurs, deinocheirids (Deinocheirus plus Hexing), alvarezsaurids and caenagnathids.  Having any sense of the ascending process morphology could tell us much.  Metatarsal II is not obviously deeper than wide, unlike ornithomimosaurs (except Harpymimus; unreported in deinocheirids), but like carcharodontosaurids, therizinosauroids, some oviraptorids and velociraptorines.  The proximal outline of metatarsal III would at first glance appear to be the strangest thing about this material, being reconstructed as strictly dorsoventrally oval unlike all(?) other theropods.  Tilting it and adding a posterior tapered tip results in a close match to Majungasaurus however (see figure below).  If it is an unreduced proximal metatarsal III, tyrannosauroids, most ornithomimosaurs, alvarezsauroids and pennaraptorans would be excluded.  Proximal phalanx II-2 lacks a proximoventral heel, so is not from a deinonychosaur.  The pedal phalanges are too elongate to be therizinosauroid, and the pedal ungual is too broad.  Phalanges are not as dorsoventrally compressed as Mapusaurus, and as noted above they lack the ventrolateral shelves found in ornithomimosaurs.  Abelisaurid phalanges seem similar however.  I wonder if we have a case like Camarillasaurus or probably Dandakosaurus involving misidentified elements making the specimen seem stranger than it really was, with so many edges of supposed metatarsal III dotted to indicate incompleteness that it could actually be metatarsal II or IV.  Certainly nothing connects this specimen with Deinocheirus.  As per the numerous errors illustrated by Hartman et al. (2019) nobody should trust Choiniere et al.'s scorings in any case.  The Lori matrix recovers "Paraxenisaurus" as a ceratosaur closest to Aucasaurus as far as taxa with well preserved feet are concerned, but also doesn't include characters particular to ceratosaurs and isn't great with pedal characters in general.  So I would place the specimen as Neotheropoda incertae sedis (or even indet.) pending a better description of the tarsus and of the real bone surfaces on supposed proximal metatarsal III. 

Pes of "Paraxenisaurus normalensis" holotype (center) in dorsal view compared to Majungasaurus crenatissimus composite (left) and Deinocheirus mirificus referred specimen IGM 100/127 (right).  Colored proximal view of "Paraxenisaurus" is after Serrano-Brañas et al., with reoriented metatarsal III as per my interpretation shown above that.  Note "Paraxenisaurus" elements were scaled using their scale bars, whereas scaling to listed measurements results in different proportions, so those should be seen as approximate.  "Paraxenisaurus" after Serrano-Brañas et al. (2020), Majungasaurus after Carrano (2007) and Deinocheirus after Lee et al. (2014).

References- Osmólska, Roniewicz and Barsbold, 1972. A new dinosaur, Gallimimus bullatus n. gen., n. sp. (Ornithomimidae) from the Upper Cretaceous of Mongolia. Palaeontologica Polonica. 27, 103-143.

Carrano and Hutchinson, 2002. Pelvic and hindlimb musculature of Tyrannosaurus rex (Dinosauria: Theropoda). Journal of Morphology. 253, 207-228.

Brochu, 2003. Osteology of Tyrannosaurus rex: Insights from a nearly complete skeleton and high-resolution computed tomographic analysis of the skull. Society of Vertebrate Paleontology Memoir. 7, 138 pp.

Middleton, 2003. Morphology, evolution, and function of the avian hallux. PhD thesis, Brown University. 147 pp.

Carrano, 2007. The appendicular skeleton of Majungasaurus crenatissimus (Theropoda: Abelisauridae) from the Late Cretaceous of Madagascar. In Sampson and Krause (eds.). Majungasaurus crenatissimus (Theropoda: Abelisauridae) from the Late Cretaceous of Madagascar. SVP Memoir 8, 164-179.

Lee, Barsbold, Currie, Kobayashi, Lee, Godefroit, Escuillie and Tsogtbaatar, 2014. Resolving the long-standing enigmas of a giant ornithomimosaur Deinocheirus mirificus. Nature. 515, 257-260.

Kobayashi and Barsbold, 2005. Reexamination of a primitive ornithomimosaur, Garudimimus brevipes Barsbold, 1981 (Dinosauria: Theropoda), from the Late Cretaceous of Mongolia. Canadian Journal of Earth Sciences. 42(9), 1501-1521.

Claessens and Loewen, 2016 (online 2015). A redescription of Ornithomimus velox Marsh, 1890 (Dinosauria, Theropoda). Journal of Vertebrate Paleontology. 36(1), e1034593.

Hattori, 2016. Evolution of the hallux in non-avian theropod dinosaurs. Journal of Vertebrate Paleontology. 36(4), e1116995.

Hartman, Mortimer, Wahl, Lomax, Lippincott and Lovelace, 2019. A new paravian dinosaur from the Late Jurassic of North America supports a late acquisition of avian flight. PeerJ. 7:e7247.

Cau, 2020 online. http://theropoda.blogspot.com/2020/04/paraxenisaurus-un-deinocheiride.html

Serrano-Brañas, Espinosa-Chávez, Maccracken, Gutiérrez-Blando, de León-Dávila and Ventura, 2020. Paraxenisaurus normalensis, a large deinocheirid ornithomimosaur from the Cerro del Pueblo Formation (Upper Cretaceous), Coahuila, Mexico. Journal of South American Earth Sciences. 101, 102610.