Why Thulborn's ideas on dinosaur polyphyly are wrong

[Michael Mortimer <mickey_mortimer111@msn.com>]

Thulborn, 2006. On the tracks of the earliest dinosaurs: implications for 
the hypothesis of dinosaurian monophyly. Alcheringa 30, 273-311.

Thulborn's second recent paper is an investigation of dinosaur monophyly 
using ichnological data.

Although he notes early on that Carrano and Wilson (2001) found that 
dinosauroid tracks could be made by non-dinosaurian dinosauromorphs (e.g. 
Marasuchus), this is soon forgotten and Ladinian dinosauroid tracks are 
attributed to quadrupedal theropods.

Thulborn reprises his confused and archaic ideas on herrerasaurids (see my 
earlier post on his digital homology paper), using the same Padian (1986) 
quote to introduce them.  After this muddled grasp of their phylogenetic 
affinities, the author tries hard to make Herrerasaurus chirotheroid instead 
of dinosauroid.  He illustrates and discusses the partial holotype pes, 
claiming that perhaps the first ungual was longest and that the fourth digit 
was much longer than II and subequal to III (these areas are unpreserved in 
that specimen).  Yet Novas (1993) described a complete pes of Herrerasaurus, 
which has a short first pedal ungual and fourth digit (unpreserved in the 
illustrated right foot, but preserved on the left).  So any statements based 
on Thulborn's fictional chirotheroid Herrerasaurus such as "the pes would 
have been rather prosauropod-like, which seems reasonable considering that 
Herrerasaurus has sometimes been likened to, or even classified among, the 
prosauropods (e.g. Colbert 1970, van Heerden 1978,Cooper 1980,1981)" or "it 
implies that any herrerasaurian footprints that do exist might have been 
classified among chirotherioid ichnotaxa, such as Brachychirotherium, or 
allied with problematical ichnogenera such as Batrachopus, along with tracks 
of some prosauropods, thecodontians and 'crocodiliornorph' reptiles" are 
obsolete and incorrect.  His concluding sentence "In view of these chronic 
uncertainties about the anatomy of the herrerasaurian pes (theropod-like or 
prosauropod-like?) it seems impossible to identify the tracks of 
herrerasaurs with any reasonable degree of confidence." is particularly 
laughable.  Amusingly, Thulborn then notes the complete pes described by 
Novas, but incorrectly states its distal phalanges are unknown (they are 
even listed in the measurement table on the same page) and that it "does not 
resolve uncertainties surrounding the form of the herrerasaurian pes."  
Using the given measurements, we can see that Herrerasaurus has a mesaxonic 
pes, where digit IV is intermediate in length between II and III, just like 
the dinosauroid tracks illustrated in Thuborn's figure 4.  Herrerasaurids 
would have left dinosauroid footprints, just like Eoraptor, silesaurs and 
Marasuchus, while Lagerpeton and Scleromochlus would have left ectaxonic 
prints due to their elongate digit IV.

Thulborn reminds us that these ideas are better left in the 1970's by 
referring to coelophysoids as podokesaurids (or procompsognathids).  Staying 
behind the times, he notes "The Ornithosuchidae is often identified as the 
sister-group of a monophyletic taxon Dinosauria (e.g. Paul 1984, p. 176) or 
as the sistergroup of a taxon comprising Dinosauria, near dinosaur 
thecodontians such as Lagosuchus and, sometimes, the Pterosauria (i.e. the 
clade 'Ornithodira' sensu Gauthier 1986; Benton & Clark 1988, fig. 8.1; 
Benton 1990, fig. 1.1).", despite the fact all recent analyses have agreed 
ornithosuchids are crurotarsans.  He considers the ichnogenus Sphingopus to 
be ancestral to the dinosauroid type.  With a distally placed pedal ungual I 
and large pedal digit V impression, these are possibly referrable to 
poposaurs like Effigia.  Effigia would make a track with a dinosauroid-like 
cross axis, as shown by its metatarsal proportions.  This would also explain 
the pronated manus with large lateral digits.  Thulborn concludes "In 
summary, early dinosauroid tracks are almost certainly the work of theropod 
dinosaurs." without ever considering basal dinosauriformes.

The author next discusses sauropodomorphs.  He uses Blikanasaurus as an 
example of an early sauropodomorph, but we now know it was a basal sauropod 
and its enlarged pedal digit I and possible ectaxony are not primitive for 
sauropodomorphs.  Saturnalia (Langer, 2003), Pantydraco and Efraasia show 
that basal sauropodomorphs had short pedal digit I and were mesaxonic.  The 
transformation to a Blikanasaurus-like pes can be followed via plateosaurs 
and Anchisaurus.  Furthermore, the subdigitigrady of Blikanasaurus is indeed 
a reversal from the digitigrade condition seen in basal sauropodomorphs 
(Saturnalia, etc.), despite how unlikely it seems to Thulborn.  Citing 
Charig et al. (1965) as support for primitively subdigitigrade 
sauropodomorphs is completely worthless, given the recent discovery of basal 
taxa (e.g. Pantydraco, Saturnalia) and the unresolved relationships of 
sauropodomorphs through the 1990's, let alone the 1960's.  Thulborn agrees 
with Olshevsky that the enlargement of pedal digits I and V in sauropods are 
"implausible reversals", yet again without any evidence save personal 
incredulity.  He then claims the enlarged first manual ungual and robust 
first manual digit were slight modifications of the condition seen in 
crurotarsans, yet basal sauropodomorphs such as "Thecodontosaurus" YPM 2195 
show that these features were less developed than in prosauropods and 
sauropods.  Thulborn uses Tetrasauropus as an example of a basal 
sauropodomorph track, while doubting the sauropodomorph identity of Otozoum. 
 Yet the most basal sauropodomorphs were mostly bipedal (Langer, 2003) and 
even prosauropods couldn't walk quadrupedally with a pronated manus (Bonnan 
and Senter, 2007).  Indeed, the latter authors concluded Otozoum was left by 
prosauropods while Tetrasauropus was left by basal sauropods.  Contra 
Thulborn, basal sauropodomorphs would have left tracks very similar to the 
dinosauroid type but with a more prominant pedal digit I.  Indeed, the pes 
of Saturnalia would make indistinguishable tracks from Herrerasaurus.

Thulborn then discusses ornithischians.  He uses Charig (1982) as support 
that Pisanosaurus has no ornithischian synapomorphies, contrary to 
Weishampel and Witmer (1990) and Sereno (1991, 1999), as well as more recent 
references such as Langer (2004), Langer and Benton (2006) and Irmis et al. 
(2006).  Again the use of such an outdated reference is misleading and 
inappropriate.  Thulborn suspects Pisanosaurus may be a herrerasaurid(!) 
because "there are some definite resemblances between the hindlimb bones of 
Pisanosaurus (Bonaparte 1976, figs 5,7) and those of Herrerasaurus 
ischigualastensis", and Bonaparte (1976) judged some characters to be 
intermediate between sauropodomorphs and "thecodonts". While the hindlimb 
elements do exhibit plesiomorphic dinosauriform characters, Irmis et al. 
identified several ornithischian characters in the jaws- coronoid process; 
constricted tooth roots; emarginated tooth row.  These are not seen in 
herrerasaurids, and Langer (2004) and Langer and Benton (2006) found 
Pisanosaurus to clade with ornithischians (not herrerasaurids) with good 
support.  Even if Pisanosaurus turns out to be a basal dinosauriform with 
ornithischian-like jaws (as in Silesaurus; though it didn't clade with 
Silesaurus in Langer and Benton's study either), it's certainly no 
herrerasaur.  Thulborn views the elongate saurischian-like manus of 
heterodontosaurids as autapomorphic, but it may be plesiomorphic based on 
Butler's (2005) unpublished analysis.  Then the "more generalized form" with 
short digits and IV longer than II (as seen in Lesothosaurus) would be a 
genasaurian morphology.  In fact, heterodontosaurids would have probably 
made dinosauroid tracks, as their first digit did not contact the ground, 
and their pes was slender with mesaxonic digits and claw-like unguals.  The 
same could be said of lesothosaurs.  Thulborn uses both Otozoum and 
Batrachopus as examples of early ornithischians, yet the former is 
prosauropod (Bonan and Senter, 2007) and the latter crocodylomorph (Lockley 
and Meyer, 2004).  Furthermore, he refers Atreipus to the Theropoda because 
"the manus ... is matched in theropods (Thulborn 1993) and the well-defined 
digital nodes of the pes resemble those in Grallator."  Yet the manus with 
its elongate digit IV is not matched in theropods, whereas it does resemble 
those of lesothosaurs.  Saurischians (except sauropods) cannot pronate their 
hands anyway, so Atreipus cannot be theropod.  Safran and Rainforth (2004) 
agree it is ornithischian and distinguish it from Grallator based on a few 
characters.  It's possibly a lesothosaur or related taxon and shows 
ornithischians evolving from the dinosauroid track type in the Carnian.  
Another possibility is that it is a silesaur, which have theropod-like pes, 
quadrupedal stance and possibly pronated manus.  Silesaurs are known from 
the Late Triassic of North America and Europe, matching the distribution of 
Atreipus.

Based on the above, Thulborn finds the chance of sauropods and 
ornithischians being secondarily quadrupedal to be implausible, due to the 
number of changes that would have to occur (e.g. enlarged pedal digit I in 
sauropodomorphs, redevelopment of pedal digit V in sauropods, 
semiplantigrade posture).  But as we've seen, basal sauropodomorphs and 
ornithischians actually had pes which would have left tracks basically 
indistinguishable from the dinosauroid type which characterized theropods 
and basal dinosauriforms.  Thulborn cites several papers in support of 
dinosaurian polyphyly (Thulbom 1975, Charig 1976, 1982, Maryanska & Osmolska 
1983, Zhao 1983, Welles 1986, van Heerden 1997), but none are relevent 
today.  For instance, Zhao (1983) suggests prosauropods, sauropods, 
carnosaurs, coelurosaurs and ornithischians each had a separate 
pseudosuchian ancestor.  No evidence was given, no particular 
non-dinosaurian archosaurs were suggested as relatives to any dinosaur 
clade, and the entire idea is preposterous in an age where we know Zhao's 
"prosauropods" are a grade leading to sauropods, while his "carnosaurs" and 
"coelurosaurs" are artificial divisions of theropods.  Welles (1986) 
concerns the origin of theropods, where he concludes that "proterosuchians" 
must have been their ancestors, since the crurotarsan ankle of 
"pseudosuchians" is too derived.  Yet this says nothing of other dinosaurs, 
for which such a conclusion would be equally true.  Van Heerden (1997) 
wasn't even a technical article, but rather a chapter in the laymans book 
"The Complete Dinosaur".  Any study from the 1970's or even early 1980's 
would be plagued with fictional ideas like pseudosuchians sensu lato, 
teratosaurs, palaeosaurs and the (pre-Gauthier) carnosaur-coelurosaur 
dichotomy.  Not to mention the absence of cladistic methodology and the 
numerous new specimens and taxa discovered since.  In short, such papers are 
useless today.

Thulborn postulates each dinosaurian clade had an independant origin from a 
chirotheroid "thecodont" based on ichnological trends, but as we saw above, 
his identifications are full of errors.  His figure 8 is a phylogeny of 
sorts showing the proposed transformations.  At the base is a chirotheroid, 
probably a crurotarsan.  The theropod lineage starts with Sphingopus 
(possibly a poposaur), then goes through Atreipus (a lesothosaur or 
silesaur), and ends at Eubrontes (a theropod) but with the manus of 
Atreipus.  Thus Thulborn's idea that theropods retained digits II-III-IV is 
based on non-theropod tracks.  His sauropodomorph lineage begins with 
Navahopus (a basal sauropod) and goes through Pseudotetrasauropus and 
Brontopodus (more derived sauropods).  It completely skips prosauropods and 
basal taxa like Saturnalia.  Finally, his ornithischian lineage begins with 
Otozoum (a prosauropod) and continues with Anomoepus and Moyenisauropus 
(ornithischians).  These errors invalidate Thulborn's problems with deriving 
sauropodomorphs and ornithischians from dinosauroid trackmakers since the 
morphologies he views as basal for each clade are usually not members of 
that clade at all, and are not basal members even when they are properly 
assigned.

He states that each dinosaur clade developed an interlocking tibioastragalar 
articulation independantly, as shown by different mechanisms.  Yet these 
morphologies (tall ascending process in derived theropods; posterodistal 
tibial notch in derived sauropods) are not present in basal members of each 
clade (coelophysoids, Guaibasaurus, Saturnalia).  Contra Thulborn, all basal 
dinosauromorphs have homologous ascending processes, not just theropods.  
His figure 9 comparing Allosaurus, Diplodocus and Hypsilophodon to 
illustrate the differences is misleading, and would be ineffective if basal 
members were used.

For his discussion of the osteological evidence for dinosaur monophyly, 
Thulborn lists 38 previously proposed dinosaurian synapomorphies.  He notes 
that most are postcranial, "even though cranial structures are 
conventionally regarded as more conservative than postcranial ones and, 
thus, more trustworthy as indicators of phylogenetic affinity." Yet until 
the discovery of Silesaurus (published after anything Thulborn cites on the 
issue), non-dinosaurian dinosauromorph crania were poorly known, and 
conventional wisdom regarding conservatism in various skeletal regions is 
not based on empirical data.  Thulborn first concentrates on the absent 
postfrontal, which is indeed also known in proterochampsids, crocodylomorphs 
and poposaurids.  Yet the fact a character exhibits homoplasy elsewhere in a 
phylogeny is not a reason to doubt its validity, unless Thulborn is 
proposing to unite a dinosaur subclade with one of those other three groups 
of archosauriformes.  He then states the expression of the supratemporal 
fossa on the frontal is a consequence of the absent postfrontal, so should 
not be considered a separate character.  Yet pterosaurs have the former but 
not the latter, showing he is incorrect.  As the supratemporal fossae are 
exposed on the frontals of pterosaurs and Silesaurus, this is an 
ornithodiran character, so Thulborn is right to reject it as a dinosaurian 
synapomorphy, but for the wrong reasons.  Of the other cranial 
synapomorphies noted by Thulborn as being called equivocal by Novas (1996)-
- an ectoptergoid which dorsally overlaps the pterygoid is a dinosaurian 
synapomorphy (absent in pterosaurs and crurotarsans).
- a laterally exposed quadrate head has a wider distribution, being present 
in Lewisuchus, some crurotarsans and Euparkeria.
- a reduced posttemporal opening is a dinosaurian synapomorphy (absent in 
Silesaurus, Marasuchus and pterosaurs) otherwise present in proterochampsids 
and crocodylomorphs.

Thulborn remarks that Novas found many proposed dinosaurian synapomorphies 
to be invalid, and yet this is misleading.  Most of the rejected characters 
were rejected because they are also present in dinosaurian outgroups such as 
Pseudolagosuchus, Marasuchus, Lagerpeton and pterosaurs.  These are 
perfectly valid to use when arguing against dinosaurian polyphyly as 
advocated by Thulborn, as he states the dinosaurian common ancestor would be 
plantigrade or semiplantigrade, have a functionally pentadactyl pes and a 
quadrupedal gait.  Ornithodiran, dinosauromorph and dinosauriform characters 
all support a dinosaurian common ancestor which is a bipedal digitigrade 
animal with a tetradactyl (though functionally tridactyl) pes.  They support 
dinosaurian monophyly exclusive of the crurotarsan or basal archosauriform 
taxa Thulborn advocates theropods, sauropodomorphs and ornithischians 
separately evolved from.  Of the ten postcranial characters Thulborn claims 
Novas "found to be equivocal or of doubtful validity"-
- postaxial cervical epipophyses are a dinosaurian synapomorphy (absent in 
Silesaurus, Pseudolagosuchus and Marasuchus) otherwise present in pterosaurs 
and teratosaurids.
- an elongate deltopectoral crest is a dinosaurian synapomorphy (absent in 
Silesaurus, Marasuchus, most pterosaurs and Scleromochlus) otherwise present 
in campylognathoidids.
- fewer than four phalanges on manual digit IV is a dinosaurian synapomorphy 
(absent in pterosaurs and crurotarsans) reversed in derived sauropodomorphs.
- a brevis fossa is a Dinosauria+Silesaurus synapomorphy (absent in 
Marasuchus, Lagerpeton, Scleromochlus and pterosaurs) reversed in 
herrerasaurids.  An analogous condition is present in teratosaurs and 
poposaurs.
- an ischium with a shallow medioventral lamina is a Dinosauria+Silesaurus 
synapomorphy (absent in Marasuchus, Lagerpeton, Scleromochlus and 
pterosaurs) otherwise present in poposaurids.
- a reduced medial tuberosity of the femur is a Dinosauria+Silesaurus 
synapomorphy (absent in Pseudolagosuchus, some Marasuchus and Lagerpeton), 
that is also present in some Marasuchus specimens, and it absent in some 
Plateosaurus and ornithopods.
- a prominant anterior trochanter is a Dinosauria+Silesaurus synapomorphy 
(absent in Pseudolagosuchus, Marasuchus, Lagerpeton, Scleromochlus and 
pterosaurs) reversed in Staurikosaurus.
- a tibial descending process which fits caudal to the astragalar ascending 
process is a Dinosauria+Silesaurus synapomorphy (absent in Pseudolagosuchus, 
Marasuchus, Scleromochlus and pterosaurs).
- a flat to concave proximal calcaneum is a Dinosauria+Silesaurus 
synapomorphy (absent in Pseudolagosuchus and Marasuchus) otherwise present 
in Lagerpeton.
- a proximally flat distal tarsal IV is a Dinosauria synapomorphy (absent in 
Marasuchus, Lagerpeton and pterosaurs).

Claiming that "Eventually Novas (1996) found that dinosaurian monophyly 
could be supported by only two or three seemingly unequivocal 
synapomorphies" is simply a lie, as Novas supported Dinosauria with 
seventeen characters, six of which were unequivocal.  Oddly, these six 
unequivocal characters include only one of the "two or three" listed by 
Thulborn (dorsosacral added to sacrum), while another (perforated 
acetabulum) was equivocal due to uncertainty in Pseudolagosuchus, and the 
third (mesotarsal ankle) was never used by Novas (1996) as a dinosaurian 
synapomorphy.
- a dorsosacral added to the ancestral two sacral vertebrae is present in 
Eoraptor, theropods, Efraasia, basal prosauropods, sauropods and 
ornithischians.  It is absent in herrerasaurids, Saturnalia, 
Thecodontosaurus(?), Plateosaurus+Sellosaurus and immediate outgroups 
(Silesaurus, Pseudolagosuchus, Marasuchus, Lagerpeton).  Poposaurs, 
Pterosaurs and Scleromochlus all have a dorsosacral as well.  It is thus 
equally parsimonious as a dinosaurian synapomorphy reversed in a few taxa or 
a character which convergently developed in several dinosaurian clades.
- a perforated acetabulum is a dinosaurian synapomorphy (absent in 
Silesaurus, Marasuchus, Lagerpeton, Scleromochlus and pterosaurs) otherwise 
present in poposaurids and ornithosuchids.
- a mesotarsal ankle joint is generally agreed to be an avemetatarsalian 
synapomorphy instead.

Far from being equivocal or of doubtful validity, nearly all of these 
characters unequivocally support Dinosauria or Dinosauria+Silesaurus.  A few 
cannot be determined among basal dinosauromorphs, but they still support a 
monophyletic origin of dinosaurs separate from chirotheroids.  Numerous 
additional characters diagnose nodes between Avemetatarsalia and Dinosauria 
(Novas, 1996; Benton, 1999; Langer and Benton, 2006).  Also important is 
that the many characters diagnosing Saurischia and Eusaurischia support a 
Dinosauria more monophyletic than advocated by Thulborn, where 
sauropodomorphs and theropods have separate origins.  The occasional 
reversals and convergences are of little consequence, as no one dinosaur 
clade exhibits many reversals and the only non-dinosaurian taxa that share 
more than a few of the characters are poposaurids (e.g. Effigia).  As usual 
for 1970's dinosaur polyphyly proposals (and BAND proposals), Thulborn never 
addresses which non-dinosaurian taxa are most closely related to theropods, 
sauropodomorphs or ornithischians.  This leaves his hypothesis untestable.  
His statement "In short, the evidence for dinosaurian monophyly is weak, 
inconclusive and no more compelling than it was 20 years ago (Charig 1982)." 
is of course completely inaccurate.  To the contrary, newly discovered basal 
theropods and sauropodomorphs (Guaibiasaurus, Saturnalia) are extremely 
similar to each other, while more complete remains of herrerasaurids show 
them to have many saurischian characters.

In conclusion, Thulborn's ideas on dinosaur polyphyly are based on 
misidentification of tracks, personal incredulity regarding character 
transformations, a near complete disregard for basal dinosauromorphs and 
basal sauropodomorphs, and a dependance on outdated sources.  Furthermore, 
he is extremely misleading in regard to herrerasaurid relationships and 
pedal morphology, Pisanosaurus' relationships and morphology, dinosaur 
tarsal morphology, and the number and quality of proposed synapomorphies 
supporting dinosaurian monophyly.  Many of his ideas are disturbingly 
archaic, and like Feduccia and Martin, he cites outdated articles and those 
which tangentially deal with a topic as evidence of controversy or validity 
when in actuality all modern articles dealing with that topic in depth have 
reached a consensus.  Just like his digital homology paper, this one belongs 
back in the 1970's.

Mickey Mortimer