[Taxacom] Procedure request

David Campbell pleuronaia at gmail.com
Mon Feb 4 15:41:27 CST 2013

> I thought (perhaps erronously) that homoplasy could be recognized only after
> the 'best' tree is selected and so I am at a loss to see how restricting
> morphological traits to those that are unique within the ingroup has
> anything to do with the amount of resulting homoplasy.

It is possible to tell that homoplasy is present by observing that
different characters support different relationships.  Of course, you
don't know which one is homoplasous and which (if any isn't), but you
know that something is going on.  However, if there are some features
that are highly variable (particularly within what you consider to be
single taxonomic units), or features for which there are strong
reasons to suspect convergence, such features commonly are not
selected for inclusion in a phylogenetic analysis.  Except for
phenetics, most approaches assume that some data are more useful for
the question at hand than others and focus in on those apparently
better characters.  For example, it would be quite easy to come up
with a number of characters to use in an analysis that indicates that
female mammals of all species are more closely related to each other
than to any male mammal.  Of course, any sensible analysis would treat
reproductive features of either sex as different characters of a
particular species rather than as alternate character states, but that
judgement is based on the knowledge that sex-related features vary
within many species.

If I'm doing a phylogenetic analysis of the Bivalvia, I'm not going to
include color pattern-there are uniformly colored clams,
radially-striped clams, etc. present here and there in most major
groups, so I dismiss it as too variable and leave it out.  On the
other hand, for a species-level analysis it might be useful to
include.  Similarly, size is generally not very useful as a high-level
phylogenetic marker, particularly if there's any possibility that
ontogenetic variation may confound your analysis.

Molecular data are not free of selection bias-what you sequence
reflects what genes you think will tell something useful about your
question.  But typically the entire sequence goes into the analysis,
whereas morphological analyses are generally rather more selective.

Not that including more or excluding more probably homoplasous data is
definitely better or worse, but simply that the different typical
level of selectivity in including molecular versus morphological data
will have some impact on the statistics.

Dr. David Campbell
Visiting Professor
Department of Natural Sciences
Gardner-Webb University
Boiling Springs NC 28017

More information about the Taxacom mailing list