[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Unwin's cladogram - long

It's best to talk specifics in an argument. I would have sent this
privately, but David  Unwin seems to prefer a public forum.

Here are a number of specific problems foundin the recent cladistic
analysis of the Pterosauria by David Unwin (2003) in his paper entitled:
On the phylogeny and evolutionary history of pterosaurs, part of the
recent Ecology and Paleobiology of the Pterosauria.

Clade 1. Pterosauria

Clade 2. Macronychoptera: Dimorphodon macronyx > Quetzalcoatlus

1. dentary forms >75% of the length of the mandible.

Unwin considered the plesiomorphic state to exist only among outgroup
taxa, and in his most primitive pterosaur, Preondactylus in which the
dentary forms half of less of the mandible length. Preondactylus has an
atypically gracile mandible, thus is derived from more typical sister
taxa, including nonvolant prolacertiforms. Only the inner mandible shows
a dentary break, at the 0.66 mark. Sharovipteryx appears to share the
plesiomorphic condition, but in  Longisquama the apomorphic condition
also exists with teeth below the orbit. In Dimorphodon, a Preondactylus
sister taxon, the dentary also extends only to the 0.66 mark, which
marks it as a reversal. The surangular plate is missing in this taxon
and the area in question was presumed filled by the dentary in earlier
reconstructions on which Dave must have based his data.

2. Coracoid at least 66% the length of the scapula.

Unwin says thi plesiomorphic state only occurs in Preondactylus and the
various outgroups. Not so. Basal pterosaurs that share the plesiomorphic
condition include ?Peteinosaurus (3359), MCSNB 8950, ?E. rosenfeldi,
Jeholopterus, Campylognathoides zitteli, and Pterorhynchus.
Sharovipteryx has a coracoid exceeding 0.66 of the scapula, but this is
not yet in the literature.

3. Manual phalanges relatively robust compared to pedal phalanges.

Unwin says this only occurs in pterosaurs more derived than
Preondactylus and the various outgroups. Perhaps his confession of
reliance on figures, rather than the specimen itself {in this case only}
caused this problem. My observations of the holotype and the latex mold
do not agree with earlier figures. All pterosaurs, including
Preondactylus, have larger manual unguals than pedal unguals, unless
they are subsequently reversed, as in ctenochasmatids.

4. Forelimb length > 2.5x hindlimb.

Unwin says the plesiomorphic stated only occurs in Preondactylus and the
various nonvolant outgroups. This may be a valid character ? but
according to a matrix with 4x the taxa and 3x the characters, probably a
reversal in Preondactylus.

5. Humerus longer than the femur.

Unwin reported that the plesiomorphic state only occured in
Preondactylus and the various nonvolant outgroups. Later Unwin explained
that in any other "rhamphorhynchoid" the humerus is either the same
length or longer than the femur (with the exception of E. cromptonellus,
but even this is an error in the literature). E. ranzi, E.
comptronellus, R. muensteri, S. crassirostris and the Maxberg specimen
of Scaphognathus all have a humerus equal to the femur, so the character
should have been written: Humerus not shorter than the femur. Even so,
in Preondactylus, the humerus and femur are also subequal in length.
32mm vs. 32.5 mm, according to Dalla Vecchia and my observations agree.

Caelidracones: Anurognathus > Quetzalcoatlus

6. Quadrate inclined anteriorly.

Unwin is not sure about anurognathids, which have not been
reconstructed. My reconstructions indicate a vertical quadrate (except
in Anurognathus), or one that leans forward at the top (Jeholopterus).
Besides, a leaning quadrate is the plesiomorphic state because it shows
up in Longisquama and Sharovipteryx. Please note that the quadrate is
just barely not vertical in the various Campylognathoides,
Eudimorphodon, and in primitive Rhamphorhynchus (i.e. R. intermedius),
also in primitive Dorygnathus (Donau) and Sordes. It leans south in
derived taxa of many clades.

7. Ulna longer than tibia.

Pretty good character. It only reverses, as Unwin notes, in much higher
taxa, but note that Austriadactylus and R. muensteri (the Tischlinger
specimen) reflect the plesiomorphic condition, while MPUM 6009, which is
the most primitive pterosaur in my book, reflects the apomorphic

8. Fibula < 80% tibia.

Here Unwin notes the fusion of the fibula to the tibia in anurognathids,
which is valid for all but Batrachognathus (reversal). Unfortunately for
Dave, some Rhamphorhynchus, Dorygnathus, Sordes and Scaphognathus
reflect the pleisiomorphic condition.

In this category Unwin also notes the brevity of the anterior process of
the ilium in primitive pterosaurs. In Preondactylus the pelvis has a
much different shape than figured, so reliance on the literature again,
is fatal. Plus, most of the ilium lies upon a matrix crack in
Preondactylus. The ilium of Dimorphodon is either hidden beneath other
bones (Unwin 1988) or incompletely excavated (Buckland 1859), so appears
short. Regardless, Longisquama and Sharovipteryx both have incredibly
elongated ilia, as does MPUM 6009, so the relative reduction that occurs
in basal dimorphodontids + eudimorphodon is a common reversal that
occurs numerous other times within the Pterosauria.

Lonchognatha ? Eudimorphodon ranzii > Rhamphorhynchus muensteri (but
strangely also includes Pterodactyloidea, according to the "contents")

9. Rostrum low with straight or concave dorsal outline.

Good character for all non-dimorphodontids -- except that Cosesaurus and
Longisquama also display the apomorphic condition. It goes to show that
the convex rostrum is derived.

10. Posterior process of premaxillae interfingers between frontals.

Unfortunately, the premaxilla does not reach the frontals until later in
pterosaur phylogeny. Eudimorphodon and Campylognathoides (all of them)
display the plesiomorphic condition. And I would not say "interfingers".
It?s more of an overlap when it does happen.

11. External narial opening low and elongate.

Good character for nearly all non-dimorphodontids, except that
Cosesaurus and Longisquama also display the apomorphic condition. It
goes to show that the big naris is derived.

12. Nasal process of maxilla inclined backwards. [= inclined posteriorly
from the base].

Good character for nearly all non-dimorphodontids, except that
Dimorphodon itself, Cosesaurus and Longisquama also display the
apomorphic condition. It goes to show that the vertical ascending
process of the maxilla is derived and pretty much restricted to anuros.

13. Broad maxilla-nasal contact.

Unfortunately, there is no good way to judge this as even a narrow
maxilla ascending process broadens dorsally to provide broad contact
with the nasal. Whatever David was seeing in non-dimorphodontids, he
would see in the higher prolacerts.

14. Orbit larger than antorbital fenestra.

Again, Cosesaurus and Longisquama display the apomorphic condition. So
do Tanystropheus and Megalancosaurus. Dimorphodontids are different
because they are derived. After all you have to develop an AOF _after_
you have an orbit, right? So the AOF has to start small.

Breviquartossa ? Rhamphorhynchus muensteri > Quetzalcoatlus

15. Ventral margin of skull curved downwards caudally.

Well, Dimorphodon has a deep jugal (almost never figured), but it rises
again at the quadrate articulation. Eudimorphodon has a jugal downstep
leading toward the quadrate. The depressed jugal is certainly present in
Rhamphorhynchus, but I don?t see it in some Dorygnathus (although R156
has it and so does Sordes), but none of the Scaphognathus have it. All
in all it pretty much comes and goes.

16. Loss of ?coronoid? eminence on caudal end of mandible. [actually not
the caudal end, but anterior to the articular region, according to the
next sentence].

Not only is the area difficult to see in many taxa, that little bump is
never much of anything to snort at even when present.

17. Development of bony mandibular symphysis [and?]
18. Mandibular symphysis forms more than 30% the length of the mandible.

Unwin notes that reversals of this trait occur in scaphognathines,
Istiodactylus, lonchodectids and Germanodactylus. Rhamphorhynchus
certainly achieves the apomorphic condition. Basal dorygnathids do not,
among those not listed above. It looks like this trait also comes and

19. Loss of heterodonty in the mandibular dentition.

Unwin places Dorygnathus in this clade and this taxon displays the
greatest heterodonty of all. Must have overlooked that one. Nuf sed.

20. Metacarpals I, II and III of equivalent length.

I noted a similar character, but unfortunately it doesn?t start to kick
in until after Scaphognathus crassirostris, when I<II=III. This occurs
three times by convergence. I=II=III occurs five times, and once in the
outgroup, the Anurognathidae, which isn?t supposed to happen according
to Unwin, who scored a zero for this matrix box.

21. Short metacarpal IV. [shorter than mc I-III].

I?ll have to get back to you on this one. I?ve never seen this, unless
it could be attributed to dislocation.

In this character section Unwin muses on the possibility "that
Pterodactyloidea might share a closer relationship with Anurognathidae
than with any other ?rhamphorhynchoid? clade" ? noting pelvic shape
similarities, reduction of cervical ribs and reduction of caudal series.
David, if your analysis can?t get you any closer to the answer than
this, it?s time to dig deeper. Seriously. Wait to publish until you're

Rhamphorhynchidae ? Sordes > Rhamphorhynchus muensteri

22. Less than 11 pairs of teeth in the rostrum.

Unwin includes Angustinaripterus in this clade, and it has more than 11
pairs of rostral teeth. Same with Rhamphorhynchus intermedius.
Otherwise, this clade is internally consistent.

23. Deltopectoral crest tongue-shaped, with necked base.

Good character. No problems, but not sure about the "necked" base in any
but Rhamph.

Pterodactyloidea ? Pteranodon > Quetzalcoatlus

24. Narial and antorbital fenestrae confluent.

I used to believe this old saw until I took a much closer look at the
situation. The naris and AOF are not confluent in the new ctenochasmatid
that Chris Bennett is studying. It?s big enough to clearly discern. Get
closer and you?ll find the same situation in other ctenochasmatids and
azhdarchids. In the ?pterodactyloid? descendents of the various
Scaphognathus specimens [in other words, the rest of the pterdacs]
you?ll find the naris and AOF do merge (as shown in No. 9), but a
secondary naris, already present in Scaphognathus (all of them), takes
over. You?ll always find it between the anterior nasal and jugal, which
stretches anteriorly much further than anyone has shown before.

25. Basipterygoids united to form a median bar of bone.

Not until later, Dave. You?ll find separated basipterygoids on all
Azhdarchids, Cycnorhamphus, all the Pterodactylids and the basal
Germanodactylids, such as No. 12, G. rhamphastinus and GR SOS 21.

26. Reduction of cervical ribs.

Not yet. Look more closely and you?ll find cervical ribs on pretty much
the same list as above.

27. Caudal vertebral series shorter than dorsal series.

One of the outgrowths of the unossified juvenile theory is that some
parts never ossify as adults in certain taxa (i.e. Nyctosaurus fingers).
The caudal series in pterodactyloid-grade pteros is another example.
Often longer than the ossification will show, still not always longer
than the torso.

28. Pteroid long and slender.

I broke this character down as short, long and at least half the radius
in length. There is an obvious elongation of the pteroid, but according
to PAUP it occurred three times with the Maxberg specimen of
Scaphognathus displaying a pteroid about as long, relatively, as that in
Anhanguera or Pterodaustro. This character needs to be better quantified
in all future work.

29. Wing metacarpal (IV) at least 80% the length of the humerus.

Good character. No exceptions.

30. Pes digit V with a single phalanx or entirely absent.

Actually the ungual (sometimes soft) is retained by basal pterosaurs. A
rather large two phalanx pedal digit V is retained by Azhdarchids (and
shows up in their footprints!), and it is retained in the 'size squeeze'
pterosaur descendents of Scaphognathus. Ctenochasmatids seem to retain a
long and short phalanx (see Pterodaustro). Otherwise, a good character,
though a bit more complicated than previously thought. In some cases
where the toe was thought absent, it has been found in soft form.

Ornithocheiroidea ? Istiodactylus > Pteranodon

31. Development of a notarium

To his credit, Unwin reports that the differences in notarial and
scapular design among various taxa suggest convergence in this
character. My analysis also suggests that a notarium evolved three times
and that precursor situations (scapula attach to a single vertebra or
neural spines grow together without fusion) occurred in smaller forms.
Unfortunately in some of these key taxa the beast is on its back, hiding
the goodies! In the end, it becomes a spectrum of blended possibilities.

32. Coracoid larger than scapula

Possible. But probably a convergence. The difference between the
coracoid and scapula is never substantial and sometimes subequal in
Pteranodon, unlike ornthocheirids.

33. Humerus with warped deltopectoral crest.

This character may be related to the above one through engineering in
that both Pteranodon and ornithocheirids are soaring specialists with
deep chests, unlike the related Nyctosaurs, with its short ribs. Here
again, my PAUP is bigger than your PAUP and this character comes out

34. Pneumatopore piercing the anconal surface of the proximal part of
the humerus.

I avoid characters like this, which can only be viewed on a few taxa.
There?s your Mulligan.

35. Distal end of humerus has a triangular outline.

Again, roadkill specimens rarely make this observation possible.

36. Ornithocheiroid carpus

Here again, sounds like you would have to be able to hold it to judge
it. And make sure it?s uncrushed.

37. Reduction of proximal ends of metacarpals [all or some do not reach
the carpus]

Here is a great example of selective ossification in that the
metacarpals only ossify distally, but most certainly maintained some
sort of softer tissue contact with the carpus. Considering the poor
preservation of ornithocheirid fingers and metacarpals (does anyone know
of a complete set, except in Haopterus, Arthurdactylus and
Brasileodactylus?) this is a difficult character to judge. Certainly
true in Nyctos, Pteranos and higher Tapejaridae, but that?s another
clade with mmmmuch longer metacarpals. Contra Bennett 2001, some
azhdarchids (eg. Zhejiangopterus) maintain carpal-metacarpal ossified

38. Femur with stout neck and steeply directed caput.

Homoplasy at work, here and elsewhere.

Euornithocheira ? Ornithocheirus mesembrinus > Pteranodon [same as above
clade, sans Istiodactylus]

39. Concave posterior margin of nasoantorbital fenestra

Also occurs in the outgroup: Scaphognathus, Pterorhynchus,
Cycnorhamphus, Diopecephalus, Thalassodromeus. Absent in the ingroup:
Haopterus, Coloborhynchus spielbergi, Criorhynchus, and possibly the
Field Museum Nyctosaurus. Probably not a great character.

40. Basal region of the orbit infilled.

This is widepsread. It also occurs in Austriadactylus, higher
Anurognathids, Eudimorphodon, higher Dorygnathus, Azhdarchids,
Ctenochasmatids, Pterorhynchus, and Tapejarids. I really don?t see it in

41. Coracoid facets on sternum lateral to each other.

This character rarely shows in roadkill fossils, so I avoid it.

Unwin also notes the character "radius less than half the diameter of
the ulna" which joins Istiodactylus to this clade, but is not found in
Pteranodon and Nyctosaurus. Now that's a good character!

Pteranodontia ? Nyctosaurus > Pteranodon [finally a clade I can believe

42. Tall, narrow frontal crest.

Nyctosaurus exceptions are considered gender differences by Unwin, but
this is a guess. Outgroups, such as Tupuxuara, also have a tall, narrow
frontal crest increased by a contribution from the premaxillae, but
Unwin does not score these. Wishful thinking won't put a crest on some

43. Dentition absent

Actually the anterior premaxillary and dentary teeth form anterior jaw
tip spikes in this clade, but aside from these, this is a good

44. Mandibular rami elevated well above level of lower jaw symphysis

I suppose it?s all in how you look at it, but I see a straight, and only
sometimes curved, line leading from jaw tip to articulation and
declining angle from tip to mid jaw at the symphysis. Azhdarchids have a
medial keel to the same effect (but not scored as such by Unwin) and
tapejarid mandible keels are open posteriorly, homologous to those in
this clade (again not scored). Some Nyctosaurus and Pteranodons do not
have this deep mandible feature, but rather have a straight ventral

45. Pneumatic opening in palmar surface of proximal part of humerus.

As before, I avoid such minor and restrictive characters.

46. Hyper-elongation of the wing metacarpal. [>2x humerus]

Almost a good character. Unwin mentions convergence in Diopecephalus and
azhdarchids, but only scores azhdarchids. The metacarpus of
Diopecephalus is less than 2x humerus. Some Pteranodons (FHSM 11402) do
not exhibit this character. Noripterus, a dsungaripterid does, and I?ll
bet the other dsungaripterids do too [lack of a humerus and a complete
metacarpus prevent me from putting money on it though]. The new
Tupuxuara mounted in a Japanese museum (keyword Tupuxuara in Google)
also has this character, but I don?t know what?s real and what is not.
Any information on this specimen would be rewarded.

Lophocratia ? Pterodaustro > Quetzalcoatlus  [strange choice of name,
considering that a crested clade is the outgroup], also includes

47. Humerus with elongate, rectangular deltopectoral crest [& squared
off free margin]

Frankly, I don?t see it. What I do see is a variety of shapes, mostly
rounded at the free margin.

48. Extensive sagittal cranial crest

Unwin admits that this character is absent in some six ingroup taxa, has
a different morphology in select taxa, then guesses that immaturity or
gender might explain the differences. Then scores for the plesiomorphic
condition (with the proviso that this may be inappropriate). No. You
can?t do this when you build a matrix. Score the character as is and let
PAUP tell you what the relationships are later.  Unwin also notes that
relationships within this clade are difficult to resolve. Heck yeah!
This is a mixed bag, Dave!

Ctenochasmatoidea ? Cycnorhamphus > Pterodaustro, also includes
Pterodactylus, Lonchodectes (a Greensand Fm. Clade known from bits and
pieces evidently) and Huanhepterus

49. Quadrate oriented in subhorizontal position [and?]
50. Squamosal located level with or below the base of the lacrimal
process of the jugal [and?]
51. Occiput faces ventrally

These three characters essentially state the same thing three times. I
wonder though, if the base of the skull keeps the same orientation, but
the rest of the skull droops. Unwin also notes the Zhejiangopterus has a
similar low quadrate, but this is based on examination of a poor tracing
that confuses the maxilla with the jugal, among other things. I divide
this character into > 160 degrees, or ? l60 degrees to the jawline. Even
this gets confused by the posterior depression of the jugal in certain
taxa. Cycnorhamphus has the plesiomorphic condition. So do some, but not
all, Pterodactylus.

Euctenochasmatia ? Pterodactylus kochi > Pterodaustro

52. Neural arch of mid-series cervicals depressed and with low neural

Unwin also notes a similar morphology in azhdarchids. In my study, both
are derived from separate Dorygnathus sequences and Pterodactylus has a
separate derivation from Scaphognathus. Ctenochasma and Pterodaustro
have some high and some low cervical spines. Pterodactylus has lower
ones. Diopecephalus, a sister to Pterodactylus, has rather substantial
neural spines. Convergence is rampant. Diversity is abundant. It?s more
complicated than suggested by this character state and scoring. Unwin
notes that in putative juveniles the plesiomorphic state is present.
That?s because they?re not juveniles. Tiny No. 9 is close to
Pterodactylus and ties it to Scaphognathus, but it is not a juvenile of
either. It's the transitional taxon.

53. Elongated mid-series cervicals [at least 4x the width of the

For this Unwin seems dependent on a drawing in Wellnhofer 1991 that
copies one from Bonaparte 1971. I have little faith in this drawing
since I know the pelvis is imagined and three moderate cervicals are
known, no elongated ones. Still it has inspired research, so more on
this later, including a retraction if necessary. Regardless, Ctenochasma
has no elongated cervicals. But all pterodactylid cervicals are indeed

Unwin also notes here that the concave rostral profile of
Ctenochasmatidae and Cycnorhamphus might have united them, but notes
that Gnathosaurus does not share this concavity. For this latter
assertion Unwin must be relying on a lateral reconstruction made by
Wellnhofer that reinflates a dorsally crushed skull with great
imagination and IMHO, little accuracy. A smaller gnathosaurus (AMNH
5147) is known and preserved in lateral view. The concavity of its skull
can be seen, leaving some cladistic doubt as to the straightness of the
larger rostral profile.

Dsungaripteroidea ? Germanodactylus cristatus > Dsungaripterus

54.   Distal ends of paroccipital processes expanded

Good character. But here, doggone it, Unwin admits to reproducing
Wellnhofer?s image, then in the next sentence describes the expanded
paroccipital processes found on the counterplate ? but fails to
illustrate them~!

55. Dsungaripterid teeth

Good character, but hard to see in any Germanodactylus. Perhaps best
relegated to Dsung. I wonder if this was scored for Germanodactylus just
to keep this character on the table.

56.  Jaw tips toothless, but followed by tooth row.

Unfortunately, all Germanodactyus retain premaxillary teeth (once again,
reliance on Wellnhofer drawings is a mistake, and not sure why Unwin
thinks that only two are known) and Unwin scores Germanodactylus as 0/1,
a polymorphism, to keep this character alive. Here?s a perfect example,
even using Unwin?s data, of creating a chimaera rather than adding taxa
to see the subtle, step-by-step changes between taxa.

57.  Largest teeth occur in the caudal half of the dentition

Here Unwin notes that the character is less pronounced in
Germanodactylus. I?ll say ? it?s not present! Or if Dave sees it,
illustrate it!

58. Limb bones with relatively thick bone walls

This could be a great character, however, here again, a lot of taxa will
go unchecked in this category, because broken bone ends may not always
be visible.

59. Strongly bowed femur

Again, tough to determine in crushed taxa.

Unwin also notes a strong anterior bowing of the wing phalanx 1. Not
sure if this means anterior convex or concave. But in all
Germanodactylus available to me (about 5 specimens) this phalanx is

Azhdarchoidea ? Tapejara > Quetzalcoatlus

60. Orbit located well below level of dorsal margin of nasoantorbital

This is an interesting and much discussed character, but it turns out to
be a convergence under the weight of a suite of competing characters. I
can see how the connection could be visually made, but fortunately there
is PAUP.

Unwin also notes several apomorphic but convergent characters including:
absence of teeth, dorsal expansion of the caudal half of the rostrum
(actually only in some ingroup taxa), humeral fenestra, unossified
proximal metacarpals, and larger manual phalanges than pedal phalanges.

Plus, putative apomorphies: frontal extends anterior to the lacrimal
(not so, confusing a large nasal with the frontal I assume), femur >
1.25x humerus (among tapejarids, only in the Japanese tupuxuarid mount).

Neoazhdarchia ? Tupuxuara > Quetzalcoatlus

Not numbered, only under remarks: Apomorphies include: notarium, loss of
metacarpal contact with distal syncarpal, rostrum forms 88+% of the
total skull length. All of these characters are also shared with some
Pteranodons and Nyctosaurus. The latter character is also shared with
Pterodaustro and Liaoningopterus.

In conclusion ~

With so many internal problems, many recognized by the author, this
report appears to be fatally flawed and essentially useless. Even so,
this analysis has taught me that the next time a cladistic analysis is
proposed, more taxa and more characters have to be employed. No
supra-generic taxa. Only specimens. And strict reliance on the data
without guesstimates regarding maturity and gender differences.

We?ll see what problems develop _afterwards_ without inducing errors
ahead of time by creating chimaerae.

Hope this helps the next cladogram builder.

David Peters
St. Louis