Halfway back to theropods with sauropodomorph Eoraptor

When Eodromaeus was described, much was said about the analysis placing Eoraptor in Sauropodomorpha.  What nobody did was elaborate on why Martinez et al. (2011) placed it there and how strong the evidence is.  So I made a NEXUS file from Martinez et al.'s matrix to extract the character evidence.

The first thing to note is that the characters listed as multistate don't correspond to those with more than two states in the matrix, which also means the characters said to be ordered aren't those that should be.  So I ordered 51, 69, 70, 76 (though nothing is coded with state 2, making ordering useless), 102 and 117.  In addition, the definitions of some multistate characters have problems.  "Premaxillary tooth number: 4 (0); 3 (1); 5 or 6 (2)" should ideally be ordered, but states 0 and 1 need to be switched so that the intermediate state is in the middle.  "Maxillary/dentary crowns, shape: recurved (0); subtriangular (1); lanceolate (2)" artificially separates the morphology of sauropodomorph and ornithischian teeth, which are after all both triangular and lanceolate in both clades.  The current coding doesn't let the reduced curvature group them together.  "Ilium, preacetabular process, shape: tab-shaped (0); strap-shaped (1); subtriangular (2); semicircular (3)" is problematic since the next character deals with preacetabular length, and the only difference between "strap-shaped" and the other choices is length.  "Ilium, bevis[sic] fossa, shape and orientation: broad laterally-open depression (0); shallow groove or absent (1); arched ventrally-opening ovate or parallel-sided depression (2); arched ventrally-opening posteriorly-expanding to a width approximately 50% of its length (3)" combines a lot of independant variables.  There's depth (already partly covered by another character), orientation, proportional width and posterior expansion.  Oddly, lagerpetonids and Marasuchus are coded inapplicable, despite their correct state of "absent" being covered by state 1.  So given those issues, I ran the matrix and found similar results to Martinez et al.- 3 MPTs of 247 steps each, CI of .615 compared to his 3 MPTs 246 steps each with a CI of .618.  The topology is identical.  If you don't constrain the outgroups to be in the order (hypothetical(Lagerpetonidae(Marasuchus(Silesauridae,ingroup)))), then you get a lot more trees, sometimes with non-dinosauriform sauropodomorphs, and/or ornithischian silesaurids and/or Marasuchus, presumably because characters supporting more basal nodes weren't included.

So that's the matrix, but what about Eoraptor's sauropodomorph position?  It's supported by the following characters-

2. external naris size large, expanded narial margin.
The plesiomorphic state specifies a tapered snout, but some taxa coded 0 like Herrerasaurus, Eodromaeus and Ceratosaurus certainly don't have tapered snouts.  If you measure Eoraptor's naris compared to its skull length, then yes it looks large.  But that's because it has a short snout.  If you measure naris length compared to some more neutral value such as skull height at the orbit, its naris is actually smaller than any other taxon in the matrix.  Also, Panphagia does not preserve a premaxilla, so cannot be coded, contra the matrix.

19. nasal posterolateral process present.
This is actually unknown in silesaurids and Tawa (Nesbitt et al., 2009), though Panphagia has it (Martinez and Alcober, 2009).  It is also present in Coelophysis (Downs, 2000), Megapnosaurus (Downs, 2000) and "M." kayentakatae (Tykoski, 1998).

24. squamosal, ventral process a slender prong 3 or more times basal width.
This is actually present in Silesaurus (Dzik and Sulej, 2007), Lesothosaurus (Sereno, 1991), Saturnalia (Langer and Benton, 2006), "M." kayentakatae (Tykoski, 1998) and Coelophysis (Colbert, 1989). It is polymorphic in Sauropodiformes because one of Martinez et al.'s two examplars (Massospondylus- Sues et al., 2004) lacks it.  The condition in Megapnosaurus is unknown (Raath, 1977).

45. dentary tooth 1 inset.
Oddly, silesaurids are coded as inapplicable, even though this is one of their distinctive characters.  If anything ornithischians should be coded unknown, in case their predentary is homologous to the anterior dentary. 

43. maxillary and dentary crowns lanceolate.
As noted above, there's no difference between Martinez et al.'s triangular and lanceolate shapes, as can be seen by the near identical shape of the supposed triangular tooth of Silesaurus (Dzik, 2003 fig. 5H) and the supposed lanceolate tooth of Panphagia (Martinez and Alcober, 2006 fig. 5B). 

69. deltopectoral crest 45% or more of humeral length.
Eoraptor's crest has previously been reported to be 35% of humeral length (Langer and Benton, 2006).  Note that in this and other cases where I claim Martinez et al.'s codings for Eoraptor are incorrect, it's with the caveat that they are redescribing the taxon and may be shown to be right with the publication of their monograph.  Genasaurs' should be polymorphic, since Scutellosaurus (Colbert, 1964) has a short crest.  Ceratosaurus also has a short crest (Galton, 1982).

76. manus phalanx I-1, rotation of axis through distal condyles 45° ventromedially.
Martinez et al.'s own figure only labels Eoraptor's torsion as 35 degrees. 

79. preacetabular process subtriangular.
As noted above "strap-shaped" should be deleted and the taxa coded to reflect their distal shape, not their length.  In that case, Heterodontosaurus' is tab-shaped (Santa Luca, 1980), Lesothosaurus' is broken (Thulborn, 1972; Sereno, 1991) and Genasauria's is rounded (Colbert, 1964; Galton, 1974).  Panphagia's is also broken (Martinez and Alcober, 2009), while Saturnalia's is tab-shaped (Langer, 2003).  Staurikosaurus' is rounded (Bittencourt and Kellner, 2009).  Among theropods, Megapnosaurus' is tab-shaped (Raath, 1990), "M." kayentakatae's is unpreserved (Tykoski, 1998), Dilophosaurus' is tab-shaped (Tykoski, 2005) and Ceratosaurus' is unpreserved (Gilnmore, 1920). 

81. preacetabular process, attachment scar present.
Also present in Marasuchus (Novas, 1996), Saturnalia (Langer, 2003), Herrerasaurus (Novas, 1993), Staurikosaurus (Bittencourt and Kellner, 2009).  Absent in Sauropodiformes (Cooper, 1981; Huene, 1926).  Unknown in "M." kayentakatae (Tykoski, 1998) and Ceratosaurus (Gilmore, 1920).

90. ischial mid shaft cross-sectional shape subtriangular.
This is absent in Panphagia (Martinez and Alcober, 2009).  It is also present in Lesothosaurus (Thulborn, 1972), Herrerasaurus (Langer, 2003) and "M." kayentakatae (Tykoski, 1998). 

122. astragalus fibular facet, primary orientation lateral.
The condition in Lagerpeton (Sereno and Arcucci, 1993) and Lesothosaurus (Thulborn, 1972) is unknown.  Genasauria (Scutellosaurus- Langer and Benton, 2006) and Herrerasaurus (Langer and Benton, 2006) have a laterally oriented facet.  Megapnosaurus (Langer and Benton, 2006) and "M." kayentakatae (Tykoski, 2005) lack it. 

125. astragalus anteromedial corner shape (dorsal view)- anteriorly projecting at least 25% width of the medial side of the astragalus.
Martinez and Alcober (2009) state this is absent in Eoraptor.  It's present in Megapnosaurus (Langer and Benton, 2006), "M." kayentakatae (Tykoski, 2005) and Dilophosaurus (Tykosaki, 2005).  The condition in Lesothosaurus (Thulborn, 1972) is unknown.  

Additionally, Martinez et al. found Eoraptor to be closer to Panphagia and Saturnalia than sauropodiforms based on-

83. ventral acetabular flange of ilium present.
Lesothosaurus lacks a flange (Sereno, 1991).

88. brevis fossa an arched ventrally-opening ovate or parallel-sided depression.
This was discussed above as being a composite character and miscoded in the outgroup (also in Staurikosaurus- Bittencourt and Kellner, 2009).  Regardless, Eodromaeus has the same state as Eoraptor, so the combination is not unexpected in a basal theropod.

95. ischial antitrochanter, anteroposterior length less than adjacent length of the articular surface for the ilium (also in coelophysoids).
This is actually absent in Panphagia (Martinez and Alcober, 2009), Saturnalia (Langer, 2003), Coelophysis (Padian, 1986), Megapnosaurus (Raath, 1990), "M." kayentakatae (Tykoski, 2005) and Dilophosaurus (Tykoski, 2005).  This is also present in Lagerpeton (Sereno and Arcucci, 1993, coded inapplicable by Martinez et al.), Heterodontosaurus (Santa Luca, 1980), Herrerasaurus (Novas, 1993) and Staurikosaurus (Bittencourt and Kellner, 2009).  Makes me wonder if Martinez et al. reversed the states on accident.  Silesaurus (Dzik, 2003), Lesothosaurus (Sereno, 1991), Genasauria (Colbert, 1964) and Tawa (Nesbitt et al., 2009) do not seem to have significant ischial antitrochanters.

When all of the above are changed in the matrix, Eoraptor is the basalmost theropod.  An additional question to ask is how many previously suggested theropod characters of Eoraptor are in the matrix?  In particular, how many of the characters suggested to link it with Tawa+Avepoda are included?  As listed in the Database, the previously suggested characters are as follows.  Those included in Martinez et al.'s matrix are in bold.

Tykoski (2005)- subnarial process of premaxilla narrow and rod-like; maxilla anterodorsally concave; nasal contacts antorbital fossa; lacrimal inverted L-shape; lacrimal at least equal to orbital height and reaches orbit's ventral rim; antorbital fossa exposure on lacrimal large, with triangular fossa on ventral process; vertebra 25 (dorsosacral) incorporated into sacrum (not actually present in Eoraptor, contra Tykoski); humerus twisted; metacarpal I distal condyles strongly asymmetrical; preacetabular process thin and blade-like; postacetabular process longer than acetabulum (not actually present in Eoraptor- Ezcurra, 2010); pubic fenestra (incorrectly coded absent in Eoraptor); distal ischium <3 times minimum width of shaft; subrectangular distal tibia with posterolateral extension (incorrectly coded absent in Eoraptor).
Ezcurra (2006)- lateral surface of premaxillary body pierced by a single neurovascular foramina above the second premaxillary tooth; maxillary antorbital fossa rostral to internal antorbital fenestra broad and rostrocaudally well extended; medial wall of the antorbital fossa extends through the entire ventral border of the internal antorbital fenestra as a very narrow lamina; lateral lamina of bone of the lacrimal with no interruption of the lacrimal antorbital fossa and restricted to the caudal margin of the ventral ramus along its dorsoventral extension; rostral process of jugal excluded from the internal antorbital fenestra, bluntly squared rostrally; ventral process of the squamosal length forms more than half of the caudal border of the infratemporal fenestra (absent in Eoraptor); humerus length subequal or shorter than 0.6 of the length of the femur; extensor pits on the dorsal surface of the distal end of metacarpals deep, well developed; shaft of metacarpal IV in relation to that of metacarpals I-III significantly narrower (incorrectly coded absent in Eoraptor); metacarpal IV and fourth digit: proximal portion set lateral to Mc III (idem Mc V) and with only one or lacking phalanges; supraacetabular crest and lateroventral border of the postacetabular process (lateral brevis shelf) continuous as a weakly developed ridge; cnemial crest on proximal tibia moderately developed; caudal cleft between medial part of the proximal end of the tibia and fibular condyle present.
Nesbitt et al. (2009)- premaxilla-maxilla, subnarial gap between the elements (absent in Eoraptor); nasal possesses a posterolateral process that envelops part of the anterior ramus of the lacrimal; ilium, ischiadic peduncle orientation well expanded posteriorly to the anterior margin of the postacetabular embayment.
Ezcurra (2010)- posterior part of premaxillary alveolar margin edentelous, resulting in an interruption of the upper tooth row (absent in Eoraptor); anterior margin of maxillary antorbital fossa squared; dorsoventrally compressed ridge on lateral surface of maxilla, forming the ventral border of the antorbital fossa (alveolar ridge) (incorrectly coded absent in Eoraptor); orientation of the lacrimal orbital margin erect and close to vertical; supraacetabular crest of ilium flares lateroventrally to form a hood-like overhang that hides anterodorsal half of acetabulum in lateral view (coded absent in Eoraptor); well developed brevis fossa with sharp margins on the ventral surface of the postacetabular process of the ilium present, being directly ventrally facing; well developed tibiofibular crest on distal femur.

Of course some of these are only supportive of a theropodan Eoraptor given the proper outgroup/ingroup arrangement and some are absent in Eoraptor, showing their own matrices have problems.  But of the 31 characters present (as far as I can tell) in Eoraptor, only 11 are included, and of those 11 at least 4 are incorrectly coded as absent.  Enforcing a placement for Eoraptor closer to avepods than herrerasaurids in this modified matrix is 13 steps longer, but the you have to wonder how many of the proposed characters for placing Eoraptor outside Herrerasaurus+Avepoda were included and how correctly they are coded.  It's all outside the scope of this post.

In the end, I don't think the evidence for placing Eoraptor in Sauropodomorpha is strong.  Of the 12 characters, 4 seem absent in Eoraptor, 2-3 of the others seem present basally in avepods, 3 of the others are present in Herrerasaurus, and all have high homoplasy.  Even the toothless dentary tip and lanceolate teeth are complicated by their presence in silesaurids and/or ornithischians.  With so much homoplasy among basal dinosauromorphs, it's necessary to include fragmentary but important taxa like Lewisuchus, Guaibasaurus, Chindesaurus and Pisanosaurus. Only then and when we incorporate all suggested conflicting character data can we hope to have meaningful results.

Martinez, Sereno, Alcober, Columbi, Renne, Montanez and Currie, 2011. A basal dinosaur from the dawn of the dinosaur era in Southwestern Pangaea. Science. 331, 206-210.