Monday, June 22, 2015

Chilesaurus brings out the BANDit in me

In the last post, I detailed what Novas et al.'s analyses actually indicated regarding Chilesaurus' relationships.  Far from all indicating a basal tetanurine status, there was good support for positions in Sauropodomorpha, Coelurosauria or sister to Avepoda.  What if we dig a bit deeper?

What kind of megalosaur looks like Chilesaurus?!

Chilesaurus (holotype SNGM-1935; premaxilla in medial view) and Monolophosaurus (IVPP V84019) snouts scaled to same premaxillary alveolar length (modified after Novas et al., 2015 and Brusatte et al., 2010 respectively).

If you look at Chilesaurus, there are basically no characters similar to monolophosaur-piatnitzkysaur grade tetanurines.  And I mean nothing.  This is from the person who came up with a list of characters connecting ornithomimosaurs to ceratosaurs and ornithischians to theropods, so that's really saying something.  The best I can do here is to note the scapula, humerus, radius and ulna wouldn't be out of place for a basal tetanurine, oh and it's coded as having a femoral extensor groove.  I must admit I feel rather Feduccian saying this, but the result of placing Chilesaurus in Carrano et al.'s analysis looks to me like a case of cladistic failure where it has to land somewhere but the conclusion is unrealistic, being least terrible as opposed to most parsimonious.  It's rather like when Chatterjee included Protoavis in his bird analyses- it emerged in a certain place (above Archaeopteryx but below ornithothoracines), but much of the morphology just didn't make sense there, from the unfused clavicles and plesiomorphic manus to the very derived pelvis.  Another example is when alvarezsaurids were placed in Avialae, with their short coracoids, elongate chevrons, long ischia, etc..

Chilesaurus has no interdental plates, leaf-like teeth with marked wear facets, no postaxial epipophyses, no hyposphene-hypantrum articulations, no gastralia, a short manual phalanx II-2, one phalanx on manual digit III, no obturator process or foramen on its ischium, opisthopuby, separate pubic apices, no fibular crest, a tiny astragalar ascending process, huge pedal digit I which may contact the tarsus, completely unreduced proximal metatarsal III, etc..  You'll see some of these in some tetanurines, but never close to so many at the same time (e.g. parvicursorines have up to 8, therizinosaurids have 4, ornithothoracines have up to 9).  A few aren't even known to reverse (e.g. ascending process, fibular crest) regardless of the animal's ecotype.  Even when the characters are common in theropods, they're not found at the megalosaur grade- e.g. elongate presacral centra with low neural spines, no metacarpal IV, no supracetabular crest, antitrochanter, supratrochanteric crest, small pubic peduncle, cylindrical anterior trochanter, etc..  So immediately I'm suspicious.

Chilesaurus tested in Coelurosauria

Chilesaurus (holotype SNGM-1935) and Piatnitzkysaurus (holotype PVL 4073; flipped horizontally) fourth cervical vertebrae in right lateral view, scaled to same centrum length minus anterior ball if present (modified after Novas et al., 2015 and Bonaparte, 1986 respectively).

Noting most of these characters are found in coelurosaurs, which were not strongly rejected in the only matrix to include a vaguely representative sample (Smith et al.'s needed only 3 more steps to place it there), I added Chilesaurus to the Lori analysis (704 characters).  Monolophosaurus is the outgroup, with some carnosaurs and basal coelurosaurs, and a complete sampling of maniraptoriforms minus derived ornithuromorphs. If it's really a basal tetanurine, Chilesaurus should end up down by Monolophosaurus due to the lack of characters present in various coelurosaurian groups.

Without getting into the details of the topology, Chilesaurus emerged as a non-alvarezsaurid alvarezsauroid in a tree where Haplocheirus was closer to compsognathids as in several recent analyses.  This makes a lot of sense compared to being a basal tetanurine, as it has the following characters shared with alvarezsaurids or more inclusive clades- *basally constricted teeth; *reduced tooth curvature; *no interdental plates; *no hyposphene-hypantra; *transversely compressed axial neural spine; *low presacral neural spines; *lower postaxial epipophyses; *amphicoelous cervical centra (as in Alvarezsaurus); *metacarpal I more than 66% mcII length; *manual phalanx II-2 shorter than II-1; reduced manual digit III; metacarpal IV absent; *low manual flexor tubercles; *poorly curved manual ungual I; pubic peduncle dorsoventrally shorter than ischial peduncle; *pubic peduncle short anteroposteriorly compared to ischial peduncle; *reduced supracetabular crest; *antitrochanter; *at least mesopuby; *no obturator foramen in pubis; *pubic apron reduced in length; *very small pubic expansion; *distal pubic expansions not combined medially (as in Patagonykus); *no ischial obturator process or foramen; *greater trochanter anteroposteriorly broader than femoral head; *cylindrical anterior trochanter; *proximally extensive anterior trochanter; *no anteromedial femoral crest distally; *cnemial crest with low anterior angle (as in patagonykines).

Constraining Chilesaurus to be by Monolophosaurus is 13 steps longer, which is fairly well rejected.  And the biostratigraphy fits, as alvarezsauroids diverged in the Middle Jurassic or earlier and (as Haplocheirus isn't one here) emerged in South America.  Thus IF Chilesaurus is an avepod, I'd say a basal alvarezsauroid position is most likely.

Chilesaurus tested in Ornithischia

Chilesaurus (paratype SNGM-1936) and Monolophosaurus (IVPP V84019; flipped horizontally) pelvis in right lateral view, scaled to same ilial length (modified after Novas et al., 2015 and Zhao and Currie, 1994 respectively).

My second test is one Novas et al. should have done- code Chilesaurus in an ornithischian analysis.  Most of the characters listed above as shared with alvarezsaurids are actually common in basal ornithischians, as indicated by asterisks.  This brings to mind the hypothesis of Alifanov and Barsbold (2009) and Alifanov and Saveliev (2011) that alvarezsaurids aren't theropods and are more like ornithischians.  As noted in my theropod-ornithischian post, the latter clade also has members with some of the supposed avepod characters of Chilesaurus- strap-like scapula, elongate preacetabular process, distal tibia backs fibula, etc..  The jaws, manus, pelvis and pes of Chilesaurus all certainly look more ornithischian to me at a general level.  I chose the newest version of Butler's ornithischian analysis, Han et al.'s (2012; 227 characters) with Kulindadromeus added as in Godefroit et al. (2014).  In addition to adding Chilesaurus, I added Monolophosaurus to test Novas et al.'s hypothesis, along with Allosaurus as a complete non-coelurosaur tetanurine and standard theropod (the original matrix's only potential theropod is Herrerasaurus), Parvicursorinae and Patagonykus to test my above hypothesis, and Haplocheirus as an example of the kind of theropod alvarezsaurids evolved from.  This also coincidentally tests Alifanov's idea.

When run, Chilesaurus emerges as an iguanodont more derived than Anabisetia, but less than rhabdodontids, Tenontosaurus and dryomorphs.  Alvarezsaurids emerge as theropods.  But it only takes two more steps to make Chilesaurus sister to other ornithischians or a basal marginocephalian instead, and only three more to make it a basal cerapodan.  Forcing it to be theropod takes 4 more steps, so still isn't too bad, and it comes out by alvarezsaurids.  Forcing it to be by Monolophosaurus takes ten more steps, so seems unlikely.  This might seem to indicate that Chilesaurus is plausibly an ornithischian, though without an obvious place in the clade (except probably not thyreophoran, where it takes 8 more steps).  The taxa it usually ends up closest to are Yueosaurus and Pisanosaurus, the latter which is often found to be the most basal ornithischian.  However, alvarezsaurids take only two steps to place sister to other ornithischians, and only five steps to place within Ornithischia (as thyreophorans or marginocephalians).  Assuming alvarezsaurids are actually theropods, this suggests that the results of placing a theropod in an ornithischian analysis don't mean much.  So while this test didn't reject an ornithischian Chilesaurus, I don't think it rejected an alvarezsauroid relationship strongly either. 

We also have the same problem as we did placing Chilesaurus in Alvarezsauroidea- it has features that are out of place for a non-basalmost Ornithischia and never or rarely seem to reverse.  Premaxillary teeth extend to tip of element, no predentary, no cingulum (also in derived heterodontosaurines and derived iguanodonts), pleurocoels, no fourth manual digit, deep preacetabular process (also in some eurypodans), no prepubic process (also in Pisanosaurus), non-pendent fourth trochanter (also in pachycephalosaurs), unreduced fibular facet on astragalus, no ossified tendons along vertebrae (also in Kulindadromeus, Koreanosaurus and Yueosaurus).  If Pisanosaurus is the basalmost ornithischian, it is more derived than Chilesaurus in having a reduced fibular facet on its astragalus yet less derived in lacking opisthopuby.  Notably, Butler's matrix lacks any silesaurs, which may be relevant to basal Ornithischia if Langer and Ferigolo are right.

Chilesaurus REtested in Sauropodomorpha

Chilesaurus (paratype SNGM-1937; missing proximal ends of metatarsals I and II, distal digit IV, and metatarsal V) and Dilophosaurus wetherilli (holotype UCMP 37302; missing digit I and distal digit IV) pedes in proximal (top) and anterior (bottom) view, scaled to the same metatarsal III length (after Novas et al., 2015 and Welles, 1984 respectively).

I next looked over Chilesaurus' codings in the Otero and Pol analysis that's based on Yates' sauropodomorph matrix.  I found an appalling amount of un-coded and miscoded characters (45 of 361, or 12%).  I list them here so that readers can see just how obvious many are, and that even if my interpretations of some are wrong, there are WAY too many to be excused.

- Chilesaurus has a 'lateral plate' on its dentary (character 2), which is just to say the labial edge is much higher than the lingual edge, though it was coded as lacking one.
- It was left uncoded for its large subnarial premaxillary process (character 7).
- And its lack of a posteromedial premaxillary process (character 9).
- And the anterior narial edge being posterior to the center of the premaxillary ventral edge (character 18).
- And having the posterior narial border at least posterior to the premaxilla-maxilla suture (character 19).
- And lacking a strong inflection on the maxilla to form an obvious anterior ramus (character 25).
- And having a first dentary tooth adjacent to the symphysis (character 100).
- And having less than five premaxillary teeth (character 107).
- And having procumbant maxillary teeth (character 110).
- It was miscoded as lacking longitudinal labial grooves on its teeth (character 119), visible in the dentary teeth.
- Since Avepoda was coded as state 2 for character 147 ("Lateral surfaces of the dorsal centra: with invasive, sharp-rimmed pleurocoels") despite basally only having dorsal pleurocoels in anterior centra, and Chilesaurus is stated to have anterior dorsal pleurocoels, it is recoded to be state 2 as well.
- Chilesaurus was oddly miscoded as lacking septate cervical pleurocoels (character 148), despite the emphasis on them in the figure and text.
- If the Nesbitt et al. matrix is correct, Chilesaurus is miscoded here as having hyposphene-hypantrum articulations (character 157).
- Chilesaurus has transversely compressed dorsal neural spine cross sections (state 0 for character 169).
- The in situ photo of the holotype shows the proximal chevrons are more than twice as long as their corresponding centra (character 195).
- Character 205 was oddly coded inapplicable, but codes for humerofemoral length, which is 69% and thus falls into state 2.
- The deltopectoral crest apex is about 42% down the humerus (state 1), not over 50% that it was coded as in character 207.
- Like 205, Chilesaurus was coded inapplicable for its radiohumeral ratio, which is 70%, so state 1 in character 213.
- Since distal carpal II is unpreserved, it isn't known whether it abutted or overlapped distal carpal I (character 219).
- Similarly, its extent over metacarpal II is unknown (character 220).
- Chilesaurus was coded as nonexistent state 2 for character 221, referring to the presence of distal carpal V, which is unknown in that taxon.
- The manus is about 43% of humerus+radius length (state 1), not over 45% (state 2), for character 222.
- Metacarpal I is miscoded as being proximally narrower than metacarpal II (character 224).
- Character 232 codes for metacarpal V length, but with no character coding for metacarpal V presence, Yates coded other taxa lacking the bone as having a short mcV.  Thus Novas et al.'s coding of inapplicable for Chilesaurus and Tawa should be changed to state 0.
- Character 242 is uncoded, though Chilesaurus has manual ungual II less than 70% of manual ungual I's length (state 2).
- Character 243 was coded inapplicable, but is polymorphic, as manual digit II has three phalanges (state 0) and digit III has less than four phalanges (state 1).
- Character 244 was also coded inapplicable, but should be state 1, as there are no manual digits IV or V.
- Character 252 is miscoded, as the pubic peduncle of the ilium is not twice as dorsoventrally deep as the distal end is anteroposteriorly long (state 0).
- Character 254 is also miscoded, as the ischial peduncle is not much shorter than the pubic peduncle (state 0).
- Character 255 was miscoded as having a postacetabular process longer than the distance subtended by the peduncles, when the actual value is 75% (state 0).
- Chilesaurus was coded nonexistent state 2 for character 256, as was Chromogisaurus (a mistake originally made by Otero and Pol).  Both should be coded state 1, as they have a deep brevis fossa.
- Character 262 is miscoded, as Chilesaurus has a minimal pubic apron width of more than 40% of the width across the ilium's pubic peduncles (state 0).
- Character 267 is miscoded, as Chilesaurus does have a small distal pubic expansion (state 1).
- Character 278 is miscoded, as the distal ischia are narrower than deep (state 1).
- As restored, character 279 is miscoded because the hindlimb is longer than the trunk (state 0).
- Character 281 is miscoded, as the femur has a near circular section (state 0).
- Chilesaurus was coded inapplicable for anterior trochanter shape, and while it is neither a tubercle (state 0) or a ridge (state 1), it's certainly not absent (state 2).  Thus it is coded 0/1 here.
- Character 291 is miscoded, as the anterior trochanter would be visible in posterior view (state 1).
- Another character oddly coded inapplicable, character 299 should be coded 0 (tibia longer than femur).
- Character 302 was also coded inapplicable, but should be 0 (tallest point of cnemial crest proximally located).
- Character 314 (position of iliofibularis tubercle) should be inapplicable, as the authors state Chilesaurus lacks one.
- There seems to be a significant fibular facet on the astragalus, making character 317 miscoded (state 0).
- Another basic limb proportion left uncoded, the metatarsotibial ratio of 62% makes character 336 coded 0.
- Character 344 seems to be miscoded, as pedal ungual I looks longer than all other non-ungual phalanges (state 1).
- Character 350 also seems to be miscoded, as pedal ungual III is less than 85% of ungual II's length (state 1).
- The final basic measurement left uncoded, the femur is between 200 and 399 mm, if the holotype's size is doubled as the author states some specimens are.

The resulting tree is more resolved by Chilesaurus than before, with it sister to Avepoda, and successively less closely related to Tawa, Chindesaurus and Agnosphitys.  Forcing it to be sauropodomorphan is now 4 steps longer instead of 5, and forcing it to be ornithischian is only 5 steps longer instead of 11.  Thus while Chilesaurus had basically the same most parsimonious position, the miscodings made it seem less similar to ornithischians.  Deleting Avepoda to test for convergence leaves Chilesaurus sister to Tawa, but now it moves to Sauropodomorpha in only one more step (Ornithischia in 4). So if Chilesaurus isn't an avepod, it seems basically as likely to be a sauropodomorph as a theropod, and only slightly less likely to be an ornithischian.

Chilesaurus REtested as a basal tetanurine

Chilesaurus (composite using SNGM-1887, 1935 and 1937; missing apparently ultimate phalanx III-1) and Allosaurus fragilis (proposed neotype USNM 4734) manus scaled to same metacarpal II length (modified after Novas et al., 2015 and Gilmore, 1915 respectively).

A fourth test involves seeing how Carrano et al.'s coelurosaur miscodings affected Chilesaurus' position.  The file used is one I've been working on that recodes the included coelurosaurs (Compsognathus, Proceratosaurus and Ornitholestes), adds more members (Bicentenaria, Sciurumimus, Zuolong, Juratyrant, Eotyrannus, Aorun, Coelurus, Scipionyx) and corrected megaraptorans including the new Megaraptor skull, in addition to correcting some fragmentary taxa (Poekilopleuron, Lourinhanosaurus) and ordering a few characters correctly.  In this test, the unstable Poekilopleuron, Xuanhanosaurus and Pivetaeusaurus were deleted a priori to speed up analysis time.  None were suggested to be close to Chilesaurus, so it shouldn't matter.  Again, an amazing amount of miscodings (48 of 351, or 14%) were found for Chilesaurus-

- Chilesaurus was left uncoded for its seemingly unfused premaxillae (character 1).
- And the moderately low premaxillary body (character 2).
- And the low angled anterior premaxillary margin (character 6).
- And the lack of a premaxillary diastema posteriorly (character 7).
- And the lack of a prominent premaxillary palatal process (character 9).
- And the seemingly non-interlocking premaxilla-maxilla articulation (character 11).
- And the at least not highly elongate anterior maxillary ramus (character 12).
- And the procumbant anterior maxillary teeth (character 13).
- And the ridge at the ventral antorbital fossa margin as coded in other matrices (character 22).
- And the lack of a supraorbital 'palpebral' (character 61).
- And the lack of postorbital-lacrimal contact (character 62).
- And the stated deep basisphenoid recess (character 96).
- There is no paradental groove due to lacking interdental plates, so character 123 is inapplicable.
- There is no lateral dentary groove, so character 124 is miscoded.
- Chilesaurus is left uncoded for its three premaxillary teeth (character 150).
- And its even premaxillary tooth spacing (character 151).
- And the first premaxillary tooth being subequal to the rest in size (character 152).
- And the mid-maxillary tooth spacing being adjacent (character 154).
- And the dentary teeth being subequal in size to maxillary teeth (character 155).
- Since there are no postaxial epipophyses or character coding for epipophysis presence, character 177 (epipophysis size) should be coded as the smallest state instead of inapplicable.
- Character 184 should be coded 1, as the text states Chilesaurus has large anterior dorsal hypapophyses.
- The cervical ribs seem to be unfused, so Chilesaurus should be coded for characater 208.
- Character 223 is miscoded, as the scapula is 6.9 times longer than wide (state 0).
- Character 233 is also miscoded, as the deltopectoral crest apex is ~40% down the element (state 0/1).
- The distal carpal is very comparable in shape and extent to distal carpal I of Xuanhanosaurus and Guanlong, so should be coded state 0 for character 247 (not fused to distal carpal II).
- Character264 is miscoded, as Chilesaurus has a narrow brevis fossa (Salgado et al., 2008).
- Character 267 is mistyped as an 8, not an actual state.  It should be state 1 "reduced [supracetabular] shelf."
- Similarly, Chilesaurus has a transversely concave acetabular pubic peduncle surface (Salgado et al., 2008), so should be recoded 1 for character 269.
- Chilesaurus is miscoded as having a pubic peduncle much larger than the ischial peduncle (character 270)
- Character 271 is miscoded, as Chilesaurus does not have an acuminate ischial peduncle.
- Chilesaurus has a pubic peduncle over twice as long as wide distally (Salgado et al., 2008),so was miscoded for character 272.
- Character 276 is also miscoded, as Chilesaurus has a ventral preacetabular lobe (state 1).
- Character 281 is miscoded, as Chilesaurus lacks a fenestra in its pubis.
- Character 284 is miscoded as well, as the pubic apices don't contact (state 0).
- As there is no obturator foramen, character 287 should be coded 0.
- While the authors code Chilesaurus as inapplicable for character 290 "Pubis, shape of boot in ventral view", Carrano et al. code taxa with no substantial boot based on their distal proportion.  So given Chilesaurus' wide distal end, it should be coded 0 as well.
- Character 297 is miscoded, as Chilesaurus has an ischial apron.
- Chilesaurus lacks a dorsomedian ischial ridge (character 298).
- It is miscoded for character 307, as Chilesaurus has a prominent fourth trochanter.
- Similar to character 290, as Chilesaurus lacks a mediodistal femoral crest, it should be coded as the less developed state of character 310, not inapplicable.
- Characters 323 and 324 are miscoded, as Chilesaurus lacks a fibular crest so would be inapplicable for thickness and connection to lateral condyle.
- In the same way, the fibula is stated to lack a proximomedial fossa, so should be coded inapplicable for character 325 and 326 describing its morphology.
- The authors state Chilesaurus lacks an iliofibularis tubercle on the fibula, so it should be coded 0 for character 327, not unknown.
-  The authors didn't notice the mistake in Carrano et al.'s matrix where character 331 ("Astragalus, ascending process morphology: absent (0), blocky (1), laminar (2).") is coded with only two states, blocky as 0 and laminar as 1.  Thus Chilesaurus' 1 coding should be 0.
- Character 333 is miscoded, as the astragalar ascending process is lower than the astragalar body (state 0).
- Character 335 is also miscoded, as the astragalar fossa in Chilesaurus is posterior to the ascending process, so it has no fossa anterior to the process (0).
- Chilesaurus seems to have a wide fibular fossa on its astragalus, so would be recoded 0 for character 336.
- Metatarsal I is over 50% of metatarsal II's length (character 340).

Once these rather numerous miscodings and uncodings are corrected for, Chilesaurus emerges as the most basal megalosaurid.  As it only took two steps to place in Megalosauroidea before the corrections, this isn't much of a change.  What IS a big change though is that it now takes only two more steps to place within Coelurosauria (in a clade with Aorun and Sciurumimus) or outside Avepoda.  Before the corrections, the number of steps needed was 7 and 13 respectively, so the miscodings heavily weighted it toward the basal tetanurine position.  This shows Chilesaurus isn't strongly supported as a basal tetanurine.

Note I didn't bother recoding Chilesaurus in Smith et al.'s matrix because my prior work on it (Mortimer and Marjanovic, in prep.) indicates there are large amounts of uncoded characters for many taxa.  But recall from the last post that only two steps were needed to move Chilesaurus outside Avepoda, and that at least two miscodings were found that would do that.  

Chilesaurus REtested as a dinosaur

Chilesaurus (paratype SNGM-1936) and Eustreptospondylus (holotype OUM J13558) astragali in anterior view scaled to same width (modified after Novas et al., 2015 and Sadlier et al., 2008 respectively).

My fifth and final test used Nesbitt et al.'s matrix, but with Daemonosaurus added as in Sues et al. (2011) and changes from Langer and Ferigolo (2013).  I also added the characters from Nesbitt (2011) that vary in Dinosauromorpha (22, 58, 85, 113, 127, 129, 255), Lutungutali (codings from Peecook et al., 2013), Pseudolagosuchus (codings from Nesbitt, 2011), Lewisuchus (codings from Bittencourt et al., 2014), Asilisaurus (codings from Nesbitt, 2011), Diodorus (codings from Kammerer et al., 2012), Panphagia, Pampadromaeus (codings partly from Cabreira et al., 2011), Chromogisaurus, Agnosphitys, Guaibasaurus (codings from Ezcurra, 2010), Saltopus (codings from Langer and Ferigolo, 2013), Teyuwasu, Nyasasaurus (codings from Nesbitt et al., 2013), Eodromeus and the new characters from the Daemonosaurus, Lutungutali, Diodorus, Nyasasaurus and Guaibasaurus (Ezcurra, 2010) analyses.  Velociraptor (75 miscodings) and Eoraptor (91 miscodings) were comprehensibly checked for accuracy, and I ended up finding a lot of other theropod miscodings too. In the case of Chilesaurus, 38 of 315 or 12% of characters were miscoded.

- Chilesaurus was left uncoded for character 1, the proportions of the subnarial premaxilla, which are state 1.
- Based on the anterior body's slope, the premaxilla's dorsal process would have state 1 for character 3 instead of ?.
- The premaxilla has a long subnarial process, so is state 1 for character 4 instead of ?.
- Chilesaurus was miscoded as unknown for character 5, when the authors state its premaxilla has a "prominent plate-like postnarial process."
- Chilesaurus was left uncoded for the absence of a posteroventral premaxillary process (character 7, state 0).
- Chilesaurus is left uncoded for premaxillary tooth number (character 9), which was coded as being 3 in the Smith et al. matrix.
- It was left uncoded for character 11, though it clearly has state 0 of premaxillary teeth present anteriorly.
- Character 50 was left uncoded, but the skeletal indicates Chilesaurus lacks pseudosuchian-like ventrally extensive postorbital-squamosal contact.
- The authors stare Chilesaurus has a deep basisphenoid recess, so should be coded 1 for character 69, not ?.
- Character 85 should be a no-brainer to code (skull less than 50% of presacral length), but was inexcusably left uncoded.
- Again, the orbit is clearly round based on the frontal, so character 89 should be coded.
- Character 111 is miscoded, as Chilesaurus lacks "apicobasally tall and blade-like" teeth.
- The figured cervical 4 seems to lack epipophyses, and it's coded as having"absent or poorly developed" postaxial epipophyses in the Smith et al. matrix and none in the Yates matrix, which would make character 126 miscoded.
- Chilesaurus has a forelimb/hindlimb ratio of 58%, so should be coded 0 for character 150.
- The coracoid is stated to lack a biceps tuber by the authors, contra their coding of character 158.
- 161 should be coded 0 (humerofemoral ratio over 59%) , as Chilesaurus' ratio is 69%.
- The humerus has a head separated from the deltopectoral crest by a narrow ridge, so Chilesaurus should be coded 1 for character 163.
- The metacarpometatarsal ratio is .37 based on the measurement table and manus figure, so should be coded 1 for character 172, not 0.
- Character 179 should be coded 0 ("Digit I with metacarpal: longer than the ungual") instead of ?.
- Character 188 should be coded 2, as manual digit V is absent or "reduced to a tiny nubbin".  This showed nothing was coded as state 2, and indeed Allosaurus, Velociraptor, Ceratosaurus, Dilophosaurus, Coelophysis, Tawa, Eudimorphodon and Dimorphodon should be coded that way too.
- As Chilesaurus lacks a supracetabular crest, it cannot be coded for that crest's orientation (character 189).
- Chilesaurus has a basically squared anterior preacetabular margin, so should be coded 1 for character 193, not ?.
- Similarly, character 194 should be coded 0 ("posterior margin of the postacetabular process in lateral view: straight or
convex"), not ?.
- Character 202 is miscoded, as Chilesaurus doesn't have a posteriorly hooked ischial peduncle.
- Character 203 is also miscoded according to the measurement table, as Chilesaurus has a pubofemoral ratio of less than 70%.
- In a miscoding that makes Chilesaurus actually less like ornithischians, it doesn't seem to have a prepubis (character 205).
- Chilesaurus is miscoded as lacking a pubic boot (character 207).
- The pubic boot is less than 33% of pubic length (character 209).
- Character 210 is miscoded, as the figure clearly shows the pubic articular surfaces for ilium and ischium are separated by a gap.
- The pubes are distally narrower than proximally, so character 213 should be recoded 1.
- In the same way, character 219 should be coded 2 ("Ischium, proximal articular surfaces: articular surfaces with the ilium and the pubis separated by a large concave surface"), not 0.
- Based on the prominent anterior trochanter, it's near certain Chilesaurus was miscoded as having a silesaur-like flat "anterolateral" proximal femoral surface (character 225).
- Chilesaurus lacks a dorsolateral femoral trochanter (state 0), so is miscoded for character 230.
- Character 265 is miscoded, as the authors state the proximomedial fibular face has no concavity
- Character 273 is miscoded, as Chilesaurus has an astragalar ascending process shorter than the astragalar body.
- Similarly, the ascending process is blocky, not laminar.  So 274 is miscoded.
- Chilesaurus lacks osteoderms, so should be coded 0 for character 308.
- Finally, Chilesaurus is coded in the Smith et al. matrix as lacking gastralia, so should be coded 2 for character 315.

Chilesaurus still emerges as the sister group of the only included coelurosaur, Velociraptor.  But now that both are coded correctly, Chilesaurus needs 9 more steps to group with basal tetanurines. So the matrix doesn't just generically support a neotheropod assignment for Chilesaurus, it only appeared to because Velociraptor was wrongly coded the same as Allosaurus and Ceratosaurus for so many characters.  Forcing Chilesaurus outside of Avepoda takes 11 more steps, and it becomes the sister group of Avepoda, though the clade reverses its polarity so that Velociraptor is the first to branch off and coelophysids are 'most derived'.  Forcing it to either Sauropodomorpha or Ornithischia takes 17 more steps.  These sound like easily rejected alternatives, except that most of the steps are needed to get Chilesaurus away from Velociraptor itself, not Neotheropoda as a whole.  We can tell this by deleting Velociraptor and rerunning things.  Chilesaurus remains closest to neotheropods, but now it only takes 7 more steps to place it in Sauropodomorpha or Ornithischia.  Furthermore, if we force it out of Avepoda, it takes one of these positions instead of still being a theropod, since the artificially 'basal' Velociraptor isn't there to attract it.

|--Marasuchus
`--+--Teyuwasu
   `--Dinosauria
      |--Ornithischia
      |  |--Lewisuchus
      |  `--+--+--Pseudolagosuchus
      |     |  `--Asilisaurus
      |     `--+--Silesauridae
      |        |  |--Eucoelophysis
      |        |  `--+--+--Lutungutali
      |        |     |  `--Silesaurus
      |        |     `--+--Diodorus
      |        |        `--Sacisaurus
      |        `--Genasauria
      |           |--Scutellosaurus
      |           `--+--Eocursor
      |              `--+--Lesothosaurus
      |                 `--+--Pisanosaurus
      |                    `--Heterodontosaurus
      `--Saurischia
         |*-Saltopus
         |--Herrerasauridae
         |  |--Staurikosaurus
         |  `--+--Chindesaurus
         |     `--Herrerasaurus
         `--Eusaurischia
            |--Sauropodomorpha
            |  |--Guaibasaurus
            |  `--+--+--Efraasia
            |     |  `--Plateosaurus
            |     `--+--+--Panphagia
            |        |  `--Pampadromaeus
            |        `--Saturnaliinae
            |           |--Chromogisaurus
            |           `--Saturnalia
            `--Theropoda
               |--Agnosphitys
               `--+--Eoraptor
                  `--+--Nyasasaurus
                     `--+--Eodromaeus
                        `--+--+--Chilesaurus
                           |  `--Daemonosaurus
                           `--+--Tawa
                              `--Avepoda
                                 |--Coelophysoidea
                                 |  |--Liliensternus
                                 |  `--+--kayentakatae
                                 |     `--Coelophysis
                                 `--+--Zupaysaurus
                                    `--+--Cryolophosaurus
                                       `--+--Dilophosaurus
                                          `--Neotheropoda
                                             |--Ceratosaurus
                                             `--Tetanurae
                                                |--Piatnitzkysaurus
                                                `--Allosaurus

One most parsimonious tree of modified Nesbitt et al. matrix after deletion of Saltopus and Velociraptor, using only verifiable codings for Chilesaurus (see below). 

So many miscodings

I gotta say that I'm surprised each matrix has at least 12-14% of Chilesaurus' characters miscoded.  That's likely to be an underestimate since the taxon was described in a tabloid (Nature), so e.g. the vertebrae are almost completely undescribed and unillustrated.  Indeed, some of the coded characters that cannot be checked would be very odd for a dinosaur if true.  For instance, Chilesaurus is coded as having a hooked metatarsal V in the Nesbitt et al. matrix.  Yet this isn't found in any known dinosauromorph.  This makes me wonder if Chilesaurus is falling out in so many different positions because its actual morphology isn't being tested.  So I recoded Chilesaurus for only those characters that could be determined from descriptions or illustrations, not merely the authors' codings.

In the Carrano et al. matrix, verifiable Chilesaurus emerges as the most basal tetanurine, or just closer to birds than Chuandongocoelurus and sinensis, or in Coelurosauria (as the sister of Aorun).  Only a single extra step is needed place it outside Avepoda now.

In the Otero and Pol matrix, verifiable Chilesaurus emerges in a trichotomy with Tawa and Avepoda.   It now only takes two more steps to place in Sauropodomorpha (emerges as the basalmost), and four more in Ornithischia (down from 4 and 5 respectively). 

In the Nesbitt et al. matrix, verifiable Chilesaurus still emerges sister to Velociraptor.  Now 9 steps are needed to exclude it from Avepoda (emerges as sister to Daemonosaurus), but these are all due to Velociraptor.  For if Velociraptor is deleted while using verifiable Chilesaurus, the latter emerges as either sister to the two tetanurines or sister to Daemonosaurus outside of Avepoda.  Using verifiable Chilesaurus really helps it group with ornithischians, taking only 7 more steps instead of 17, which doesn't change much (8 more steps vs. 7) if Velociraptor is deleted.  This shows the steps that grouped Chilesaurus with Velociraptor to the exclusion of ornithischians were largely those that cannot be verified in Chilesaurus given its description.  On the other hand, using only verifiable Chilesaurus characters made it slightly less likely to be sauropodomorphan (23 more steps vs. 17), but if Velociraptor is deleted this falls to 7 more steps just like if even the questionable Chilesaurus characters are retained.

What's it mean?

Chilesaurus (after Novas et al., 2015; lower left) compared to its suggested relatives- Monolophosaurus (copyright Scott Hartman; upper left), Shuvuuia (copyright Ville Sinkkonen; upper right), and Tawa (copyright Scott Hartman; lower right).  Scaled to same femoral length.


My conclusion here (finally!) is that using pure parsimony, Chilesaurus will clade with maniraptoriforms and more precisely alvarezsaurids.  It emerged strongly supported in that position in the Lori matrix and Nesbitt et al.'s once only verified codings were used, took only 4 steps to move there without any theropod characters in Butler's ornithischian matrix, and is one of the most parsimonious possibilities in Carrano et al.'s matrix if only its verifiable codings are used.  A basal tetanurine position is strongly rejected if alvarezsaurids are included (13 more steps in the Lori matrix; 10 more in Butler's matrix). Yet I think both of these placements are unlikely due to the incongruities caused by inserting Chilesaurus there, so that even though it's not most parsimonious, I find convergence between Chilesaurus and tetanurines more likely than reversals in Chilesaurus.  The fact verifiable Chilesaurus emerges outside Avepoda in one of the most parsimonious trees when Velociraptor is excluded from Nesbitt et al.'s matrix and only takes 1 more step to place there in Carrano et al.'s matrix thus makes sense to me.  This would allow many characters to be plesiomorphic (e.g. no fibular crest, short astragalar ascending process, large pedal digit I), and also agrees with its placement in Otero and Pol's trees.  I admit I might be proven wrong if this is a repeat of 1984-1992 Segnosauria and we find a proto-chilesaur with more megalosauroid or alvarezsaurid characters, but for now this seems more realistic to me (see my preferred cladogram above).  Is it possibly not theropod?  Verifiable Chilesaurus falls out in Sauropodomorpha with 2-7 more steps, and in Ornithischia with 4-8 more.  So it seems possible it's either one, and combining the Yates and Nesbitt matrix would give us better numbers.  But what we really need before a much better conclusion can be reached is an osteology of Chilesaurus.  Please say one's in development.  The thing's simply too weird, with too many coding issues for a tabloid description to work long term. 

References- Gilmore, 1915. On the fore limb of Allosaurus fragilis. Proceedings of the United States National Museum. 49, 501-513.

Welles, 1984. Dilophosaurus wetherilli (Dinosauria, Theropoda): Osteology and comparisons. Palaeontographica, Abteilung A. 185, 85-180.

Bonaparte, 1986. Les Dinosaures (Carnosaures, Allosauridés, Sauropodes, Cétiosauridés) du Jurassique moyen de Cerro Cóndor (Chubut, Argentine). Annales de Paléontologie. 72, 247-289.

Zhao and Currie, 1994. A large crested theropod from the Jurassic of Xinjiang, People's Republic of China. Canadian Journal of Earth Sciences. 30(10), 2027-2036.

Sadlier, Barrett and Powell, 2008. The anatomy and systematics of Eustreptospondylus oxoniensis, a theropod dinosaur from the Middle Jurassic from Oxfordshire, England. Monograph of the Palaeontological Society. 1-82.

Salgado, Cruz, Suarez, Fernandez, Gasparini, Palma-Heldt and Fanning, 2008. First Late Jurassic dinosaur bones from Chile. Journal of Vertebrate Paleontology. 28(2), 529-534.

Alifanov and Barsbold, 2009. Ceratonykus oculatus gen. et sp. nov., a new dinosaur (?Theropoda, Alvarezsauria) from the Late Cretaceous of Mongolia. Paleontological Journal (English edition). 43(1), 94-106.

Brusatte, Benson, Currie and Zhao, 2010. The skull of Monolophosaurus jiangi (Dinosauria: Theropoda) and its implications for early theropod phylogeny and evolution. Zoological Journal of the Linnean Society. 158(3), 573-607.

Ezcurra, 2010. A new early dinosaur (Saurischia: Sauropodomorpha) from the Late Triassic of Argentina: A reassessment of dinosaur origin and phylogeny. Journal of Systematic Palaeontology. 8(3), 371-425. 

Alifanov and Saveliev, 2011. Brain structure and neurobiology of alvarezsaurians (Dinosauria), exemplified by Ceratonykus oculatus (Parvicursoridae) from the Late Cretaceous of Mongolia. Paleontological Journal. 45(2), 183-190.

Cabreira, Schultz, Bittencourt, Soares, Fortier, Silva and Langer, 2011. New stem-sauropodomorph (Dinosauria, Saurischia) from the Triassic of Brazil. Naturwissenschaften. 98(12), 1035-1040.

Nesbitt, 2011. The early evolution of archosaurs: Relationships and the origin of major clades. 352, 292 pp.

Sues, Nesbitt, Berman and Henrici, 2011. A late-surviving basal theropod dinosaur from the latest Triassic of North America. Proceedings of the Royal Society B. 278(1723), 3459-3464.

Han, Barrett, Butler and Xu, 2012. Postcranial anatomy of Jeholosaurus shangyuanensis (Dinosauria, Ornithischia) from the Lower Cretaceous Yixian Formation of China. Journal of Vertebrate Paleontology. 32(6), 1370-1395.

Kammerer, Nesbitt and Shubin, 2012. The first basal dinosauriform (Silesauridae) from the Late Triassic of Morocco. Acta Palaeontologica Polonica. 57(2), 277-284.

Langer and Ferigolo, 2013. The Late Triassic dinosauromorph Sacisaurus agudoensis (Caturrita Formation; Rio Grande do Sul, Brazil): Anatomy and affinities. In Nesbitt, Desojo and Irmis (eds.). Anatomy, Phylogeny and Palaeobiology of Early
Archosaurs and their Kin. Geological Society, London, Special Publicataions. 379, 353-392.

Nesbitt, Barrett, Werning, Sidor and Charig, 2013. The oldest dinosaur? A Middle Triassic dinosauriform from Tanzania. Biology Letters. 9(1), 20120949.

Peecook, Sidor, Nesbitt, Smith, Steyer and Angielczyk, 2013. A new silesaurid from the upper Ntawere Formation of Zambia (Middle Triassic) demonstrates the rapid diversification of Silesauridae (Avemetatarsalia, Dinosauriformes). Journal of Vertebrate Paleontology. 33(5), 1127-1137.

Bittencourt, Arcucci, Marsicano and Langer, 2014. Osteology of the Middle Triassic archosaur Lewisuchus admixtus Romer (Chañares Formation, Argentina), its inclusivity, and relationships amongst early dinosauromorphs. Journal of Systematic Palaeontology. 13(3), 189-219.

Godefroit, Sinitsa, Dhouailly, Bolotsky, Sizov, McNamara, Benton and Spagna, 2014. A Jurassic ornithischian dinosaur from
Siberia with both feathers and scales. Science. 345(6195), 451-455.

Novas, Salgado, Suarez, Agnolın, Ezcurra, Chimento, Cruz, Isasi, Vargas and Rubilar-Rogers, 2015. An enigmatic plant-eating theropod from the Late Jurassic period of Chile. Nature. 522, 331-334.