Place in taxonomy

Tom DiBenedetto tdib at OCEANCONSERVANCY.ORG
Tue Jul 2 13:28:21 CDT 2002

Robert Mesibov wrote:

> Tom obviously agrees with my summary of the current situation, and now I
> have an answer of sorts to my "Why?", namely: "all phylogenetic insight does
> indeed come from the evidence of heritable characters."
> This is an extraordinary statement and I'm hoping Tom or another Taxacom-er
> might like to expand on it. I would have thought there were legions of
> instances where phylogenetic insights were aided and abetted by knowledge of
> where the taxa came from.

Perhaps this gives a hint sometimes as to what may be the most plausable
hypothesis given current understandings about geography, but there should be no
expectation that the most geographically-plausable hypothesis is more likely to be
correct. Spatial distributions are simply a terrible source of phylogenetic evidence
because a "character-match" (two species occupying the same area) can easily arise
by mechanisms other than descent. Matches in heritable character-states must arise
through descent, and thus are useful for discerning the pattern of descent (the tree).
The source of error in phylogenetics is the mistaken assertion of homology -
concluding that two characters are the same when they are not. But spatial
"characters" can mislead phylogeny not only through mistaken perceptions of
sameness, but through inherent problems. Character matches can be correctly
percieved and still indicate the wrong tree
The problem here is, in a sense, a magnification of the problem with R/DNA
evidence, where historically different character states may in fact be chemically
identical. If particular character states can be achieved in a manner which obscures
the history of state changes, the utility of the data-source declines. With genetic
evidence there remains a window in which we can assume that significant historical
signal remains. Beyond that the only recourse is to analyze data in light of the
statistical generalizations arising from evolutionary process models - an approach
that severely limits the scope of the results. Spatial data is far worse because there is
not the huge mass of data that would allow even such a problematical statistical
approach. Thus there really is no way to determine the historical significance of a
particular spatial character state.
We also have many examples of phylogenetic arrangements informing the study of
the relationships of areas. This is what I think that Nelson and Platnick took from
Croizat's work - the notion of earth and life evolving (somewhat) together and that the
distribution of clades can inform the study of the evolution of the earth. It would be
intuitively harmonious to think that there could be a symmetrical relationship here,
but I feel that the nature of the evidence makes this simply untenable.

Tom diBenedetto

More information about the Taxacom mailing list