[Taxacom] Felsenstein lecture available on-line

[Richard Zander]

The Felsenstein lecture abstract said, in part:
"I will describe some new developments in statistical models for
discrete as well as continuous morphological characters. They can infer
correlations of character change across a phylogeny. However these
inferences must necessarily be very noisy and uncertain. How will
systematists and other evolutionary biologists cope with this
uncertainty, if they cannot look forward to further molecular sequencing
reducing it? Can we make use of the information without ignoring its
uncertainties or overreacting to them? If we cannot hope for an exact
account of actually what happened in evolution, is despair inevitable?"

My response:
I've not heard the Felsenstein lecture, but that should not prevent
strong opinions about it :) Using statistics with morphological traits
is problematic. Take three similar taxa, A, B, and C. A and B share 5
advanced morphological traits, B and C share 3 different advanced
traits. If the traits are independent, then ((AB)C) is the parsimonious
solution and A and B share a more immediate ancestor than B and C. BUT,
look at it this way. If the 5 traits are possibly linked evolutionarily
such that all might be necessary for survival, and the 3 traits could
also be linked evolutionarily and all are necessary for survival, then
we have total uncertainty with 1 potential trait complex versus another
1 potential trait complex, and no parsimonious solution, because we
could have 5 independent traits versus 1 trait complex or 3 independent
traits versus on trait complex. If some traits were neutral traits, then
maybe one can assume independence of these, but how to tell which are
the neutral independently evolving traits and which part of the trait
complex? Better analysis might do this. 

Remember that the association of certain traits with stasis of taxa over
millions of years is a feature of traditional taxonomy. These traits in
many well-studied taxa are pretty certain. Beyond clustering of similar
taxa according to what we think to be homologous traits, how do we judge
descent with modification and make nice evolutionary trees? How
important is detail in doing so, and how far can we go with the data at
hand?

Molecular sequencing, if only indicative of genetic continuity and
isolation events can throw light on which two are more closely related,
say ((BC)A). But if the morphological traits are linked in various
combinations as above, and then A and B might be a somewhat
anagenetically changed (say, two species, A and B) ancestor of C, having
5 independent traits versus C's 1 trait complex. Morphology reveals or
could reveal evolutionary changes, molecular analysis reveals patterns
of genetic continuity.

I think uncertainty in understanding evolution (revealing modification
with descent of taxa through selection and neutral changes) is utterly
to be expected. But why despair about the mists of history? They are
real. Although the barnyard observation that like yields like has
morphed into a whole field of study involving macroevolutionary change,
the idea that the exactitude of mathematics will provide good
evolutionary trees is overly optimistic. There is such a thing as
statistical certainty, but I've not seen it in phylogenetics yet beyond
general principles of phyletic constraint.

*****************************
Richard H. Zander 
Voice: 314-577-0276
Missouri Botanical Garden
PO Box 299
St. Louis, MO 63166-0299 USA
richard.zander at mobot.org
Web sites: http://www.mobot.org/plantscience/resbot/
and http://www.mobot.org/plantscience/bfna/bfnamenu.htm
*****************************


-----Original Message-----
From: taxacom-bounces at mailman.nhm.ku.edu
[mailto:taxacom-bounces at mailman.nhm.ku.edu] On Behalf Of William Baker
Sent: Wednesday, December 03, 2008 6:03 AM
To: taxacom at mailman.nhm.ku.edu
Subject: [Taxacom] Felsenstein lecture available on-line

In June, I posted an advertisement to Taxacom announcing that Joe
Felsenstein would be giving the annual Sir Julian Huxley lecture to the
Systematics Association on 2 July 2008. In response to requests from
Taxacom subscribers, the lecture was recorded and is now available at
http://www.systass.org/events/JH-Lecture-2008.shtml. The title and
abstract are as follows:

Not the fly on the wall: can systematists cope with uncertainty?
Joseph Felsenstein, University of Washington

Abstract:
In reconstruction of molecular phylogenies, biologists have largely
shifted their emphasis from inferring a single phylogeny to being
comfortable with statistical descriptions of the uncertainty about it,
whether by using bootstrapping, jackknifing, or Bayesian posteriors.
Similar developments for morphological and behavioral characters have
been slower in coming, as there have been fewer defensible statistical
models available. The spread of statistical approaches to the
comparative method is changing that. I will describe some new
developments in statistical models for discrete as well as continuous
morphological characters. They can infer correlations of character
change across a phylogeny. However these inferences must necessarily be
very noisy and uncertain. How will systematists and other evolutionary
biologists cope with this uncertainty, if they cannot look forward to
further molecular sequencing reducing it? Can we make use of the
information without ignoring its unc
 ertainties or overreacting to them? If we cannot hope for an exact
account of actually what happened in evolution, is despair inevitable?

~~~~~~~~~~~~~~~~~~~~~~~~~~~
William J. Baker PhD
Royal Botanic Gardens, Kew
Richmond, Surrey, TW9 3AE
Tel: +44 20 8332 5224
Fax: +44 20 8332 5278
www.palmweb.org
~~~~~~~~~~~~~~~~~~~~~~~~~~~

Please note that due to works on the new Herbarium building, there will
be disruption to visitor services and access to the collections.  Please
contact us for details before arranging any visits for 2008/9.

_______________________________________________
Taxacom mailing list
Taxacom at mailman.nhm.ku.edu
http://mailman.nhm.ku.edu/mailman/listinfo/taxacom