More on the 'cladistics' of sequences

[Richard.Zander at MOBOT.ORG]

Well, not really. Morphological data can be used in the same way as
molecular data to come up with testable, well supported hypotheses. Proof is
that morphological results generally duplicate molecular results. See my:
http://www.phyloinformatics.org/pdf/2.pdf paper for how to estimate
confidence intervals for morphologically based cladogram internodes.

Accuracy is of course the big bugaboo but is a problem for molecular data,
too. Morphological analyses are commonly judged as underparameterized
(uniform priors, uniform everything). Molecular analyses are, IMO, usually
overparameterized in the sense that if there is a way to increase confidence
levels, it will be avidly incorporated.

All weighting increases confidence levels, and is only sometimes justified.
E.g., transition-transversion ratios may be incorporated into analyses with
a step matrix that gives a cost of 2 or even 5. Yet, following the rationale
of my paper in the URL above, a weight of, say, 2 implies increasing the
number of steps on an internode from one to two. This translates into
increasing the probability of a synapomorphy (by chance alone) from 0.34 to
0.89, or about 3 times (2 successes out of 2 trials at probability 1/3).
Given that transitions have twice the ways of occurring as transversions,
you need transitions at 6 times the number of transversions to justify
weighting with a cost of 2. Transversions are more rare than they should be
but weighting them higher is perhaps less often justified than actually
happens. Note that the average ratio is used as a probability in published
papers I've seen; there is no translation from ratio to theoretical
probability of occurence in a cladogram.

Or am I wrong? This stuff is actual math.

-----Original Message-----
From: Richard Jensen [mailto:rjensen at SAINTMARYS.EDU]
Sent: Tuesday, June 08, 2004 7:23 AM
To: TAXACOM at LISTSERV.NHM.KU.EDU
Subject: Re: [TAXACOM] More on the 'cladistics' of sequences

Morphological data are especially difficult to
evaluate and, in fact, are unlikely to provide "accurate" phylogenies (see
Lamboy, 1994. Systematic Botany 19: 489-505). Yes, Lamboy's analyses were
based on simulated data, but the data matrices generated were not unlike
real morphological data matrices.


______________________
Richard H. Zander
Bryology Group
Missouri Botanical Garden
PO Box 299
St. Louis, MO 63166-0299
richard.zander at mobot.org <mailto:richard.zander at mobot.org>
Voice: 314-577-5180
Fax: 314-577-9595
Websites
Bryophyte Volumes of Flora of North America:
http://ridgwaydb.mobot.org/bfna/bfnamenu.htm
Res Botanica:
http://ridgwaydb.mobot.org/resbot/index.htm
Shipping address for UPS, etc.:
Missouri Botanical Garden
4344 Shaw Blvd.
St. Louis, MO 63110