FWD>True phylogenies

Warren Lamboy warren_lamboy at QMRELAY.MAIL.CORNELL.EDU
Fri Jul 28 07:58:26 CDT 1995


Mail*Link(r) SMTP               FWD>True phylogenies

Neal Evenhuis writes:

--------------------------------------
Date: 27/07/1995 16:37
From: Neal Evenhuis
Warren,

Wow! You've been able to find a "true phylogeny" somewhere to compare with
all those "estimates" that we've been doing for decades now!

Gosh, if you (or Sokal, Sneath, or Nei) had the answers all this time,
why didn't you tell the rest of us so we wouldn't be wasting our time
trying to figure out what the various phylogenies were?!

And to think I'd been taught that we'd never know the "true phylogeny" of
any group unless we lived long enough to see the process of evolution in
the group we were studying. Guess, I'd been taught wrong. The answers were
there the whole time.

Where'd you find the answers? In the back of the book?

Neal L. Evenhuis
-----------------------------------------------------------------------------
I reply:


Dear Neal:

Yes, it is possible to know the true phylogeny, if it is produced by computer
simulation or generated in the lab.  Papers that do this include (partial
list):

Fiala and Sokal.  1985.  Evolution 39:  609-622.
Hillis et al.  1992.  Science 255:  589-592.
Jin and Nei.  1991.  Molec. Biol. and Evol. 8:  356-365.
Lamboy.  1994.  Systematic Botany 19:  489-505.
Nei et al.  1991.  Jour. of Molec. Evol. 19:  153-170.
Rohlf et al. 1990.  Evolution 44:  1671-1684.
Saitou and Imanishi.  1989.  Molec. Biol. and Evol. 6:  514-525.

as well as additional and more recent papers by some of these same workers.


You have identified precisely the problem:  we can never know the true
phylogeny of a group of real organisms (except in a case like Hillis et al.
1992, where the phylogeny was generated in the lab).  Because we can never
know the true phylogeny organisms, it is IMPOSSIBLE to test phylogeny
reconstruction (PR) methods to determine how accurate they are.  Thus, PR
methods wind up being justified using ancillary criteria, e.g., whether they
satisfy the tenets of a particular philosophical viewpoint, whether they yield
phylogenies that are shorter than those provided by other methods, whether
they allow for character state reconstruction, whether they allow taxa at
internal nodes, etc., the list is interminable.  Unfortunately, the question
we would really like to have answered, is how well do they do at finding the
true phylogeny--and this, as you have stated, can never be determined using a
real data set.  Thus, the need for simulation studies.  Simulation studies
show that no method does a particularly good job at finding the true tree   ,
see esp. Rohlf et al. 1990 and Lamboy 1994.

Incidentally, I tried looking in the back of the book, but the answers there
were wrong.


Warren Lamboy




More information about the Taxacom mailing list