Clades, cladons, and "cladifications"
Nico Mario Franz
nmf2 at CORNELL.EDU
Wed Jun 16 00:47:42 CDT 2004
Original post: David Hull in his 1979 paper ("Limits of Cladism"; Syst.
Zool., 28:416-440) [...]
Herb: You are absolutely correct that is exactly what I said (see above)
and I was wrong. You sure can do it! My defense is that it was not what I
meant to say, so let me try again by using an example.
If another taxonomist were to take your data, use your algorithm and
outgroup, there is a high probability of her producing the same cladogram.
(In fact if she didn't get the same tree we would assume that an error
occurred.) However, if she used your cladogram to classify the taxa the
probability is much less of her producing the same classification as
yours, even if you told her that you followed the branching sequence but
used your own judgment to assign ranks. I think that is what Hull meant
when he said, "no methods have been set out thus far which permit the
inclusion of both sorts of information [genealogy and divergence] in a
single classification in such a way that both are retrievable."
The point is that while both kinds of information can be included, both
kinds can not be retrieved.
---NF: too strong a rejection in my feeling. Not *fully* retrieved, yes;
not *at all* retrieved (in the sense of either/or), no.
That is why the other taxonomist can duplicate your cladistic result with
your data, but not necessarily your classification from your cladogram.
The same data (cladogram) can produce different resutls because your
choice of where to cut is not part of the cladogram.
---NF: This is correct, and it wouldn't surprise me if Hull actually meant
that, since he's also taken the lead (eventually with Ghiselin) on the
taxonomists = typologists = essentialists argument. That argument was
based on a faulty historical analysis of practice, and so may this be.
---At this point you're talking about the Linnean system as much as you
are talking about cladistics, I think. When a cladogram is translated and
(sensibly, I'd hope) adjusted to fit the Linnean scheme (better: our
current adoption/evolution of it), some information will get lost. This is
partly because the Linnean nomenclature is ALSO a system to provide a
natural language, not just one to deal with evolutionary complexities. A
compromise in depth and density of meaning has been found for the language
to work among humans in everyday speak. Ranks and classifications convey
some of what cladograms portray, but not all of it. It turns out that
utter precision would make the language unworkable, since every little
change would require the renaming of clades.
---And still, ignoring the bit of side-tracking that happened here, Hull
is wrong, or at least too ambitious. You can for example cherry-pick
informally named clades among the few that you care to name, with the many
others you choose to leave "as is" (unnamed, just displayed) in the
cladogram. By naming informally those that have a neat set of, say,
supposed adaptations, you can gel clades, with ranks, with words, with
evolution (number of derived states, reversals, etc.).
---In a favorable reading, Hennig's point was to me: "when in doubt,
consider phylogeny first, THEN other perhaps significant phenomena." Why?
Because given what we think we know about evolution, phylogeny ought to
generate by-and-large the most predictive classificatory system. This is
of course a highly contentious point, and central to Hennig's argument.
Yet often clades ARE very successful vehicles for testing predictions in
evolutionary biology. Just check the evolution journals. They (the natural
clades) are hard to get at though, and paraphyletic groups aren't
horrendous either to test many hypotheses. But we can strive to do better.
IMO this leaves the door open for a natural phylogenetic system that can
take in and spit out information about divergence almost as much as
I'm [not] condeming any method of classification. I was just trying to
explain what thought Hull was getting at.
More information about the Taxacom