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

cladistics rants (was Re: Syntarsus [longish])



I don't have time to get into a protracted debate, but some things
bother me so much I can't pass them along unanswered...

In a message dated Wed, 11 Sep 1996 11:50:26 -0500 (EST),
    Dinogeorge@aol.com (George Olshevsky) wrote:

> The three major problems of cladistic analysis are these: (1) the assumption
> that the most parsimonious distribution of characters gives the correct
> phylogeny is not testable; 

Nobody hides that fact.  Nobody claims that cladistics gives the
correct phylogeny.  It's only claimed that parsimony is a reasonably
good assumption to make, and that the more characters used in the
analysis, the more likely it is that parsimony will give a close
approximation to the correct phylogeny.  Some characters most
definitely are not parsimoniously distributed.  However, it seems to
be a good assumption to make that most of them are.  Can you give a
reason to believe otherwise?

> (2) the way morphological characters are defined is inherently
> subjective;

Why do you claim this is a problem for cladistics as though it isn't
a problem for any other method of phylogenetic analysis?

> and (3) no reliable method exists to weigh morphological characters
> against one another (for example, how many skull characters is a
> postcranial character worth? one? one-half? ten?).

And so rather than trying to treat all characters as equally weighted,
you'd prefer that we used intuition to decide which characters are
worth more?  If, as you say, there is "no reliable method" for
assigning weights to various characters, then why do you implicitly
assign weights to characters when you analyze characters for the
development of phylogenies, George?  How much is a four-toed foot
worth to you, George?

> There is simply no way around these problems, so most cladists
> ignore them--and happily continue to consume mass quantities of
> grant money producing endless streams of cladograms with their
> computers.

The above is nothing but incendiary.  Next time, George, please try to
provide a little more light with your heat.  Discussions about your
personal problems with the way that governments and private agencies
allocate funds have no place on this list.  I'm saying this publicly
in order to squelch any potential off-topic flame fests that the above
(and similar statements) might engender.  As moderator, I will reject
messages if they stray too far in following you along this path.

> If at most one cladogram for any group is the correct one, virtually
> all the cladograms so generated >must< be wrong. How does one pick
> the right needle from this huge haystack? What statistical
> manipulation of 66 cladograms, at least 65 of which are wrong, will
> miraculously produce the right one?

If the assumptions going into the development of the cladograms hold
up, then a consensus will build up.  George is splitting here by
defining all cladograms as either "right" or "wrong".  In point of
fact, a cladogram may accurately depict some relationships while
inaccurately depicting others.  As more characters are added to the
mix, the proportion of accurately depicted to inaccurately depicted
relationships should increase.  As such, as you add more characters to
your analysis, your inferred phylogenies should become more and more
accurate and hence more and more similar to each other.  If the
addition of characters results in wildly different inferred
phylogenies, then no real conclusion can be reached based on those
analyses (other than that they are problematic -- e.g. few of the
characters chosen are parsimoniously distributed, the characters are
not accurately scored etc.).  If the addition of characters causes no
change in the inferred phylogeny, then it's safe to conclude that
you've got the best you can out of cladistics.  That doesn't mean it's
the correct phylogeny.  This is science, after all.

Since I'm pretty sure that George will respond to this, let me state
up front that I'm not likely to respond to his response.  Please don't
draw any conclusions from my silence other than that I have other
things to which I need to attend.

--
Mickey Rowe     (mrowe@indiana.edu)