[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Re: Senter 2004 - Renesto and Binelli 2005 (long)
(Experiment: Is it a good idea if I write a DML post after 2 at night? We
will find out.)
(Preliminary result: No, because it took me over an hour.)
Important remark at the beginning: It does happen that badly coded matrices
get published. I am currently going through one published in 2002 which
contains so many mistakes in just 19 ingroup taxa x 41 characters that I get
a quite different tree after having only gone through a bit over 10 of the
characters (and replacing the wishful-thinking all-zero outgroup with two
real fossils -- that alone had almost no effect except for making one of the
characters parsimony-uninformative, though). I didn't even have to add any
taxa or characters. Incidentally, the tree I get now is much closer to
current orthodoxy than the published one... but then that orthodoxy is
itself just a few years older.
----- Original Message -----
From: "David Peters" <davidrpeters@earthlink.net>
Sent: Saturday, June 23, 2007 11:48 PM
Subject: Re: Senter 2004 - Renesto and Binelli 2005 (long)
|--Paleothyris
`--+--Petrolacosaurus
< logic problem here (l.p.h): Coelurosauravus does not have classic
diapsid temporal regions, nor similar manus or pedal patterns.
It does have classic diapsid temporal regions, except that the lower
fenestra is open like in *Claudiosaurus* and the crown-group.
`--+--+--Coelurosauravus
< l.p.h.: highly derived taxa so low on the tree?
|--*Paleothyris*
`--+--*Petrolacosaurus*
`--+--+--------------------------*Coelurosauravus*
Better this way?
"Derived" is in no way connected to "closely related to whatever taxon
happened to survive to the present" or to "top right/right bottom corner of
the tree".
| |--Longisquama
| `--Megalancosaurus
< Logic problem here: L. and M. are nested within rib gliders?
*Coelurosauravus* was not a rib glider.
Again, no common skull, manus, pedes, pelvic or pectoral
patterns can be seen here.
You mean, you can't see any -- presumably just because you haven't seen
Senter's matrix.
`--+--Icarosaurus
`--+--Younginia
`--+--+--Askeptosaurus
< l.p. h. Note we’re going from a gliding taxon to a crawling taxon
to a marine taxon in just two steps. Can we find the intermediaries
so this at least appears logical?
You seem to have misconceptions about what a cladogram is. Don't be fooled
by the above presentation which ignores all branch lengths.
`--+---------*Icarosaurus*
`--+--*Youngina*
`--+--+---------------------*Askeptosaurus*
I suppose you'll like this better.
| `--Palaeopleurosaurus (can’t find this taxon)
It's a pleurosaur -- a sphenodontian.
`--+--+--Monjurosuchus (choristodera)
| `--Hyphalosaurus (choristodera)
< This is interesting because it puts Choristoderes near the base of
the Prolacertiformes, but they belong at the base of the
Archosauriformes.
*Palaeopleurosaurus* is the only lepidosaur in this tree, so the
choristoderes turn up as the basalmost archosauromorphs, as the sister-group
of (Prolacertiformes + Archosauriformes).
I note no Squamates here.
I agree that having an aquatic sphenodontian as the only lepidosaur in the
whole matrix was not a good idea.
< Proterosuchus does belong next to Prolacerta as the sister to
Proterosuchus, so this part is good.
Except that you wrote "*Proterosuchus*" twice. What do you mean.
`--+--Proterosuchus
`--+--Euparkeria
`--+--+--Parasuchus
<l.p.h. Parasuchus nests here evidently by default. Where are the
Proterochampsids, Doswellia, Cerritosaurs and larger Choristoderes
that would ordinarily nest with Parasuchus?
"Ordinarily"? Nowhere except in your analysis, which has a couple of
problems we discussed last year.
<l.p.h. Do Scleromochlus + pterosaurs + dinosaurs arise out of
Parasuchus? What do they have in common?
Nope, *Parasuchus* (as the only included phytosaur) is the sister-group of
the rest of Crurotarsi, and Crurotarsi as a whole is the sister-group of
Ornithodira. This is entirely normal (even though I'm not quite happy with
the arrangement within Crurotarsi). I copy the line with *Parasuchus*
(...isn't that a synonym of *Paleorhinus*, BTW?) below and remove most of
the whitespace so you can see this:
`--+--+--Parasuchus
| `--+--Ornithosuchus
| `--+--+--Saurosuchus
| | `--Postosuchus
| `--+--Gracilisuchus
| `--+--Sphenosuchus
| `--+--Saltoposuchus
| `--+--Dibothrosuchus
| `--+--Alligator
| `--+--Malawis[uch]us
| `--Protosuchus
`--+--+--Scleromochlus
And here is it again, with whitespace distributed such that the tree lines
up on my screen (Outlook Express, so Times New Roman instead of a monospace
font is by default used for plain text):
`--+--+--Parasuchus
| `--+--Ornithosuchus
| `--+--+--Saurosuchus
| | `--Postosuchus
| `--+--Gracilisuchus
| `--+--Sphenosuchus
| `--+--Saltoposuchus
| `--+--Dibothrosuchus
| `--+--Alligator
| `--+--Malawis[uch]us
| `--Protosuchus
`--+--+--Scleromochlus
<l.p.h. same old question: how do hypertrophied digits and low
chevrons arise out of atrophied digits and deep chevrons? Especially
when you have prepubes and uropatagia back in Sharovipteryx.
Low chevrons out of deep chevrons is clearly easy. But why do you think the
4th finger or the 5th toe was atrophied in the MRCA of pterosaurs and
*Scleromochlus*?
I ran the matrix of Renesto and Binelli, based on Senter and the
result was a slightly different tree than published. I guess and hope
that it was because Renesto and Binelli used a different matrix than
they published.
Are you sure you used the exact same settings? Like: treatment of multistate
taxa depending on the brackets or not, same characters ordered, and so on?
1. Petrolacosaurus is scored with a sharp snout
As opposed to?
17. the jaw joint should be coded posterior to the orbit in
Petrolacosaurus,
That's where it was. I'll try to check the other characters on Monday.
58. the acetabulum is coded as open for Petrolacosaurus
66. Petrolacosaurus is coded as having a calcaneal tuber.
These are strange, though.
Coelurosauravus and Longisquama are united by Senter by two characters:
22. Parietals form a crest
23. Parietals are ornamented with bumps
The above may be true in the broadest sense, but looking at the
details, the two taxa have nothing in common in the parietal. In
Coelurosauravus the parietals branch out posterolaterally, leaving a
gap in the center as they meet the large supratemporal/squamosal
forming the lateral shields—pure decoration beyond the brain case. In
Longisquama the parietals form a median crest over the brain. These
are complete opposites morphologically.
You mean, there is no crest in *Coelurosauravus*?
After that it’s clear that
any bumps thereon are non-homologous.
Why?
Why only two characters? And two superficial characters at that?
Why "superficial"? What were those South American birds again where feather
color turned out to contain much more phylogenetic signal than beak shape?
The questions also arise: which Eudimorphodon specimen? Which
Tanystropheus specimen? Which Macrocnemus specimen? They all score
differently in certain matrix boxes.
There are different species of *Tanystropheus*, but the others should only
differ in ontogenetic characters.
4. Eudimorphodon and Longisquma are scored as not having an
antorbital fenestra
Indeed *Longisquama* has no evidence of one. Coding *E.* as lacking it is
strange, though.
27. Mid cervicals are longer than dorsals in more derived
Eudimorphodon ranzii, but not longer than the dorsals in MPUM 6009.
That's because the latter is a baby.
33-35. There are three characters devoted to sacral count. The first
(33) offers choices of 2, 3 or 5. No taxa are observed with 5
sacrals. The second (34) offers choices of 3 or 4. Those previously
observed having two sacrals are here observed having three. The third
(35) offers choices of 3 or 5. Here Archaeopteryx and Sinosauropteryx
are observed having five.
Bizarre. Could you tell us how the characters are described?
41, Coelurosauravus is coded as having a two-headed posterior rib
47. Longisquama is coded as having a furcula.
Which is probably the interclavicle, and thus not a furcula.
74 Pedal digit V is scored as unknown for Eudimorphodon, rather than
long and metapodial as it is in all sister taxa.
Is it known in *E.* itself, or is it not?
75 Yet on this next character toe V is coded as per other basal
pterosaurs in Eudimorphodon.
Bizarre.
Ultimately, by using only published characters and taxa, but making
corrections (as noted above) PAUP produced the following consensus
tree, based on 36 trees.
Under what settings?
Below I have tried to put your extremely space-consuming tree presentation
into the usual e-mail format. Please tell me if I've made any mistakes --
because of all the empty lines and line breaks it was very difficult to
read. Did you copy it directly from the PAUP* log, with Mac line breaks and
all?
,--*Coelurosauravus*
|--*Gephyrosaurus*
|--*Mesosuchus*
|--+--*Youngina*
| `--+--*Paleothyris* (correct spelling, unfortunately)
| `--*Petrolacosaurus*
|--+--*Prolacerta*
| `--+--*Proterosuchus*
| `--+--*Euparkeria*
| |--*Ornithosuchus*
| `--+--*Archaeopteryx
| `--*Sinosauropteryx*
`--+--*Macrocnemus*
|--*Langobardisaurus*
|--*Tanystropheus*
|--+--*Longisquama*
| `--*Eudimorphodon*
`--+--*Hypuronector*
|--*Vallesaurus*
`--+--*Dolabrosaurus*
`--+--*Megalancosaurus*
`--*Drepanosaurus*
You’ll note that Coelurosauravus no longer nests with Longisquama,
Instead, L. nests with Eudimorphodon along with Macrocnemus,
Langobardisaurus, Tanystropheus and drepanosaurs. Prolacerta nests
with Proterosuchus and the Archosauriformes.
I note the phylogenetic bush which even fails to contain Diapsida. Oops,
sorry, that's an illusion -- _you forgot to designate *Paleothyris* as the
outgroup!_ RTFM!
PAUP* cannot guess -- if you don't tell it anything, it will take the top
OTU of the list as the outgroup.
Here's the tree again, correctly rooted on the correct outgroup:
+--*Paleothyris*
`--+--*Petrolacosaurus*
`--+--*Youngina*
|--*Coelurosauravus*
|--*Gephyrosaurus*
|--*Mesosuchus*
|--+--*Prolacerta*
| `--+--*Proterosuchus*
| `--+--*Euparkeria*
| |--*Ornithosuchus*
| `--+--*Archaeopteryx
| `--*Sinosauropteryx*
`--+--*Macrocnemus*
|--*Langobardisaurus*
|--*Tanystropheus*
|--+--*Longisquama*
| `--*Eudimorphodon*
`--+--*Hypuronector*
|--*Vallesaurus*
`--+--*Dolabrosaurus*
`--+--*Megalancosaurus*
`--*Drepanosaurus*
That looks better, but the tree still fails to find things like the diapsid
crown-group.
Because Senter/Renesto and Binelli used taxa from both branches of
the the Reptilia (Archosauromorpha and Lepidosauromorpha)
Of the diapsid crown-group.
Funny how my 2000 paper found support from every part of the taxa,
but Senter gets the nod for two questionable characters: crests and
bumps and otherwise no morphological similarities.
And, importantly, not more than one character that supports a different
topology. If there were two or more, it would show, at least in the strict
consensus.