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

Morphometrics and Phylogenetic Analysis



I think it is important to note that there are three ways that
morphometrics can interact with phylogenetic analyses and they are
quite distinct.

First, and this is an area I have been arguing for for more than 15
years now, is the use of morphometrics for the better definition of
characters as used classically in phylogenetic analyses. This should
be a no brainer - actually having evidence for separating the
character states used in an analysis rather than just eyeballing it
and saying this character is subovate and this other one is ovate,
etc. For an approach such as cladistics which claims to be very
quantitative and not subjective, this should have been the first thing
done, frankly, providing quantitative means for defending
characters.Otherwise, we are stuck with just meristic characters (e.g.
humber of belly bristles). However, and amazingly, lots of cladists
will absolutely bristle at doing this - like you just threw cold water
on them and ruined the game they are playing. Consequently, many
cladistic analyses are really totally inadequate and, at times, pretty
embarrassing that they go to so much trouble to analyse characters
that are so wanting in support. I think we are making progress with
many others, such as Tom, and the results will be far superior as a
result. 

Actually, neontologists tend to be more resistant than paleo types on
this, which seems to be part of a pattern where neontologists tend not
to be all that interested, really, in the actual beasts they are
studying anymore  and in their analyses sort of drift away from them
as organic entities and just use them as misc. units. This is also
shown by the actual paucity these days in relatively interesting
research done on living beasts. The paper on giraffe necks, a neat one
I just ran into on how a certain snake actually hisses very loudly,
and a few other recent ones are notable exceptions. If R. MacNeill
Alexander's work (and offspring), and the Netherlands functional
morphologists (publishing in Neth. J. Zool. a lot), and a cadre of
functional types, typically associated with Duke and Friday harbor,
were not around, the amount of truly interesting work done on the
biology of most living organisms would be truly amazingly low. I see
this as a failure on the part of our current academic system where the
extra cost of molecular studies - very useful and important, don't get
me wrong - means that universities will get rid of non molecular
people because their cut of grants for molecular work is larger. So we
have many universities where the only real whole animal courses taught
are by paleontologists in geology departments - if there are any.
Pretty depressing and it's time for neontologists to start pulling
their weight in this area.

Back to the subject (sorry, I digressed).

The second area is to actually use morphometric procedures as part of
the development of the reconstruction of the phylogenetic
relationships. I have been one to suggest this probably is not useful
because of their phenetic nature (see my papers in the Dinosaur
Systematics Volume, although one reviewer took me to task for doing
exactly what I said you shouldn't because, as is typical for many
reviewers, he didn't actually bother to read the paper). The point
people on this are Zelditch and Fink in a series of papers. I'm not
totally convinced as of now but I think it is important that the
possibilities not be closed off and that these guys continue their
work rather than not fund it, as some cladists have suggested. Too
many scientists hate disagreement, which is an unscientific way to be,
paradoxically. Miriam and Bill are very sharp and it is great to have
them exploring this area and they have made some compelling points
about using morphometrics to identify character suites that are highly
intercorrelated and really represent the same thing and should be
considered a single character group rather than lots of characters.
It's important to keep this dialog going and I hope it does. It
parallels, to some degree, the long-term discussions on incorporating
stratigraphic data into phylogenetic analyses. Another important
dialog although I have sympathies in both directions. Along this
general vector, I think morphometrics can, at times, be used to help
select the best phylogenetic reconstruction from trees developed
conventionally. If one tree requires far more morphometric
convolutions than another tree of equal length, then I think it really
is more parsimonious to use the data to support the data with the
simpler morphometric solution. I know many will disagree, but such is
life.

Now, the third way is another way I have pushed for more than a
decade, is for morphometrics to be superimposed on the results of more
standard phylogenetic analyses to see how morphometric states have
evolved and note convergences for study relating them to potential
responses to various evolutionaary or ecological contexts. Size and
sexual dimorphism are biggies here and it follows a very Brooksian
approach to extracting data using the phylogenetic structure as a base
for starting the work. I think there is tremendous power here and it
is one reason I fully support standard phylogenetic analyses, despite
the many flaws, often discussed herein, that we need to work toward
fixing.

Finally, although it is true that it is difficult to figure out how to
publish defenses of large characters, I think it is essential because,
after hearing talks at Yale in February where character numbers were
being thrown around in the hundreds without any way of evaluating
their worth, I realized the value and quality of the phone book
cladistic analyses will always be in great question until it's done. I
know Tom well enough to feel he is probably pretty good at his
character definitions. I think Chris does a rather good job as well.
But have absolutely no idea how good Sereno, for example, really is at
it. Paul puts up slides mentioning characters at meetings that seem
designed to keep you from actually gleening any info from them. The
last I saw of Paul's characters was part of a hadrosaur study he did
with Cathy Forester and they seemed pretty well done, actually, but I
know Cathy much better and would expect that from her. We MUST find a
way to do it, even if it is putting the data on the web for access by
those really interested. SVP is talking about the philosophy of what
to do about these huge matrices in SVP right now and that is one
option.

Anyway, morphometrics should be the best friend of the cladist. That
it very frequently isn't, and at times is avoided at all costs by
some, does not speak well for much of that work. If we are to evolve
towards quantification, we must be as good at defining characters as
we are at the philosophy and algorithms we use to do phylogenetic
analyses that generate the trees we are so fond of.


Ralph Chapman, NMNH