[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Re: Phylogenies, science, tea-time and more...
>> Enough for an introduction. I was reading the discussion following Nick's
>> intervention in favor of getting rid of silly postcranial characters and
>> fragmentary taxa in theropod phylogenetic analysis, and was suddenly
>> cought by the need of telling what I think about all this.
>I hate to spoil all the fun... but apparently someone needs to kill the
>by explaining it: Nick is parodizing the way ornithischian phylogenetics is
>almost always done. By applying it to theropods, he demonstrates what
>nonsense results when one only puts a few skulls into one's matrix. He is
>crying desperately for someone to do a serious phylogenetic analysis of
>Ornithischia. (Probably he'll end up doing it himself, alone.)
Ouch. I really am stupid. Well, at least it seems I was not the only one to
get cought, or all this list is just one vast conspiracy...
>> It certainly was a good thing to
>> nuke the old crumbling linnean tree and get rid of "thecodonts" and
>> "carnosaurs", which were mostly synonyms of ignorance.
>Unfortunately (IMNSHO), the name Carnosauria was _immediately_ reintroduced
>to designate the sistergroup of Coelurosauria (...a name that should have
>been dropped as well...). So, while *Tyrannosaurus* and *Megalosaurus* are
>no longer carnosaurs, *Allosaurus*, *Carcharodontosaurus*, *Sinraptor* and
>*Monolophosaurus* are still carnosaurs.
Yep, that's one thing I know! :-) Anyway I don't think it's a bad thing that
old names are redefined under cladistic rules (you seem to disapprove the
continued usage of the term Carnosauria), and kept in usage, although I admit
respect for historical usage might lead to two quite different directions:
save the name whatever it comes to mean or nuke it whenever it doesn't
correspond to its original meaning... Well, if we must rename every clade
the phylogeny shifts around, we really are in trouble. But that's not really
what I was talking about.
>> I know this also depends on money, research credits, and anyway
>> everydoby wants to be as modern and up to date as possible... But I dare
>> to think
>> this is NOT science. For me, dinosaur paleontology is mostly anatomy,
>> stratigraphy, paleobiology and paleoecology. The first two items are the
>> dry and hard
>> data on which the other two more attractive ones are based.
>> But where does phylogenetic systematics fit in there?
>It is based on anatomy, it is partly based on paleobiology, and
>including -ecology are partially based on it.
Ja! But the problem is that in a large part of current dino research anatomy
is no more than a tool for building some matrix, that, stuffed into some
computer program, will release the new version of a phylogeny, just to be
destroyed two months later. I'm not saying cladistics are coming from nowhere
pointless, just that they seem to be a little bit too overriding these days.
And I do think (but I have no experience, so this could be complete nonsense)
that a deep knowledge of anatomy should be something a human brain can cope
with well enough to identify phylogenetic relationships, first at species
level, and then higher up. After all, most of the important dinosaur clades
identified well before cladistics, on a morphological basis.
Now I know interrelationships among these groups are quite tricky to figure
out, but I'm not convinced that a whole bunch of, say, maniraptor
phylogenetic analyses are better than one thorough anatomical, computerless
study of basal members of the various maniraptoran clades. That's largely
what cladistic analyses want to be, but frankly, saying that A is more closely
related to B than to C because it shares 5 apomorphies with the former and 3
with the latter doesn't look like a reliable method to me. It works very fine
for massive phylogenetic evidence, like "tyrannosaurs are coelurosaurs", but
when we get to more subtle aspects or fragmentary specimens, constant and
massive changes in the results of analyses show that they're not that useful.
>> But after this is done, I can't see the sheer utility of those numerous
>> You add some abscure species known from two
>> teeth and one caudal vertebra and the tree gets completely changed...
>> of some characters you don't like and Corythosaurus might appear as a
>> new ankylosaur!
>If your tree is so unstable, you clearly have far too few characters*, and
>probably too few taxa as well.
>* Get a look at the "Basal Tetanurae" chapter in The Dinosauria II. The
>analysis in there has 638 characters. It will teach you what I mean by "too
Not "my" tree. I don't have any. I'm now old and experienced enough to know
that I'm very far from being competent enough to pretend tackling any serious
phylogenetic work on dinos, although the time may come some day... I'm just
seeing that most new papers imply some new tree, and that there's no real
reason to, at first glance, believe someone in particular is right. A 600-more
characters analysis means that the researcher(s) (I'm still waiting for my
copy of the book, should arrive in a week or two...) behind have real in-depth
knowledge of the anatomy of the beasts they're talking about, which is usually
not the case, just because that research has not been done for most new or
little-known species. Then, in any cladistic tree, if some topology is weakly
supported, then it cannot really be trusted. If it is strongly supported,
then the cladistic method probably wasn't necessary to find it out. I'm
absolutely not against cladistics. I just indended to mean that there should
detailed anatomical study, and less quickly done analyses. A really
meaningful phylogenetic analysis, just like a really thorough anatomical study
dealing with a reasonably large group or reasonably well preserved bunch of
specimens), used to imply a monograph, or at least a long paper. These are, I
think, invaluable, and all the more necessary as we find new species.
>> I think Nick is right at pointing that part of the rampant phylogenetic
>> instability comes from including too fragmentary taxa,
>Ironically, missing data are usually much more harmless than one might
>think. There are theoretical studies on this to supplement my personal
>experience (and doubtless the much richer one of others).
I would naturally tend to think that missing info is harmless, but how many
times have I seen (mainly on that list, I have to admit) trees shifted by the
inclusion of some fragmentary bits and pieces... Anyway, what I meant is
that once cladistics have determined, say, where do tyrannosaurs fit in the
theropod tree, and the general pattern of tyrannosaur interrelationships, if
new species is described from just one incomplete specimen, then a
discussion on its phylogenetic position, based on careful examination of what
(I mean, an actual TEXT about that), should work just as well, or better,
than a computer calculation.
>> but doesn't it come also from inadequate study of those
>> reasonably well preserved specimens we have at hand?
>Of course. My bird matrix
>http://www.unet.univie.ac.at/~a0000265/cladogram.htm is an extreme case of
>this. I haven't seen any of the specimens I've coded there, I have relied
>completely on the literature which is often very superficial. By examining
>the specimens -- or reading good monographs on them, which don't exist -- I
>could perhaps cut the number of question marks in half. (That's the most
>important reason why I don't consider it publishable -- even though I fear
>it is better than any published analysis of Mesozoic birds, for example in
>the sheer number of taxa that aren't included anywhere else).
Seems like we're quite much agreeing in fact...
>> As a student in social sciences, I have got quite a good knowledge of the
>> good and bad points of quantitative and qualitative methods. Cladistics
>> are a
>> quantitative method, everyting being based on the NUMBER of shared
>> and the like. When reading sociology works, I tend to strongly dislike
>> studies based on pure statistics, while I believe interviews of "real"
>> people are
>> often thrilling. Is it that silly to say that we could somewhat turn
>> actually thoroughly interviewing fossils instead of comfortably use some
>> unreliable statistical method?
>I don't think this metaphor works. In sociology you have to cope with
>of phenomena that are not quantifiable to start with. In morphology, those
>are rare... and actually just causes of ignorance. An example is the
>question whether a particular character should be ordered, and if so, in
>which way, or even which characters to include in the analysis at all.
>Learning more about the ontogeny, inheritability, variability etc. of the
>character in question (and on its dependence on other characters!) can
>us to exclude several possibilities, ideally all except one. (Unfortunatel
>papers on phylogenetic analyses rarely explain the choices that were made
Here I definitely don't agree. First, the bearing of quantification and
statistics on sociology is huge, although it's dwindling. But that's not the
point. You can indeed quantify many morphological traits, but 1) I'm not sure
quantification is the best way to approach morphology (that is, to do it well,
you have to know all those other variables you've just cited, and to explain
how you use them), and 2) well, see just below...
>BTW, I wouldn't call parsimony statistics... maximum likelihood and
>especially Bayesian methods do use statistics... and as long as we aren't
>talking about molecular data of taxa with long branches, I wouldn't call
>of them "unreliable"!
The thing I can't grasp about parsimony is why one femur character and
another cranial one should have more weight than one single vertebral one, for
example. I feel cladistics are sometimes a bit like asking "are American
democrats or republicans?", and answering that according to the last elections
most parsimonious to conclude they're republicans. I know that this is a very
rough caricature, but anybody working in cladistics is confronted to just
that kind of problem. By the way, what are "Bayesian methods"?
>> And, more generally, is it really more interesting
>> to know where Torvosaurus fit on the cladistic tree than to
>> know what it looked like and how it lived?
>This is a false dichotomy! If we don't know its place in the tree of life,
>we _cannot_ find out certain aspects of what it looked like and how it
>lived. "Nothing in evolution makes sense without a good phylogeny."
Is it not supposed to be the other way round? How can we figure out where it
fits on the tree if we don't previously know what IT is, independantly of
phylogeny? You don't need to know whether a cat is closer to a dog or to a
mouse to what a cat is, how it behaves, etc... As far as I know, knowing that
tyrannosaurs are closer to birds than to allosaurs did not improve in any way
our understanding of how tyrannosaurs lived. We keep interpreting them just as
if they had kept being "carnosaurs", giant theropod predators, and I think
we're right about it. Evolution is one thing, knowledge about one particular
species or ecosystem at one given time is another.
>The best thing about cladistics is that it has introduced tree-thinking.
Right. But it also tended to nuke down grade thinking, which was also very
informative, not from a phylogenetic point of view, but from a conceptual one.
Herbivory is not a phylogenetic group, but it's a hell of an useful notion
if you study an ecosystem... I do love tree-thinking. I just think it becomes
bad, just like any other way of thinking, when it becomes omnipresent and
Now I wonder how I can write so much silly things to answer a reply that I
largely agree with... Well, maybe it's too late for me here in France...