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

Re: A bit of cladobabble was Re: Cearadactylus - long, but with sausages

David Marjanovic (david.marjanovic@gmx.at) wrote:

<<How do we know some characters are homoplastic? Because the animals are
in other clades. How do we know the animals are in the other clades?
Because putting them elsewhere would require much more homoplasy than we
could destroy that way...>>

Brian Philidor (philidor11@snet.net) wrote:

<When you say 'much more' you could be referring to characters by simple
raw count or to characters significant in distinguishing related taxa.  I
think you mean raw count, as in the last sentences of your para: "That's
what's called parsimony, and it's why cladistic analyses are done with
such huge amounts of characters."

And yet, we're reading (with interest) about a dispute where certain
characters are stated to be more significant than others diagnostically.
It's this contrast between total number of characters and asserted
significant characters that creates the apparent contradiction. Characters
which are not significant create the homoplasy, but how do we know they
are not significant?  This is not an example working simply from the
tree-with-the-fewest-steps model. I'm sure there is an answer, but since I
didn't see it, I asked.>

  I think there might be a tad confusion here. A phylogeny without
homoplasy is a myth, and attempts to remove homoplasy from a data set is
futile without "making" the set without homoplasy. To demonstrate, give
ten taxa three characters in a set, with each character coded the same for
each taxon. This is a tree that will have 100% homoplasy. There is no
resolution. This reduces when synapomorphies between any volume of taxa
_save one_ or more are introduced, and reduces steadily until there are no
synapomorphies for any clade that are not also seen elsewhere. In other
words, the perfect tree without homoplasy lacks convergences; this, as may
be plain looking at distantly related groups with similar features, is
evolutionarily impossible, and the real science is _figuring out which are
homoplastic and which are validly synapomorphic_ and yes, the best way to
do this is add tons of characters. This can be done a variety of ways:
take regions of morphology and scientifically (and still subjectively)
isolate staes and conditions of this region that are shared between at
least two forms; find genetic loci that show a reasonable length of
sequence that matches another in the same gene. One cannot actually remove
homoplasy from a matrix without creating a false tree, such as one offered
by Hou et al., in 1996, on the coding of *Cathayornis* in Nature. This
tree was erroneously coded and found to be "perfect."

        | 1 2 3 4 5 
taxon A | 0 0 0 0 1
taxon B | 0 0 0 1 1
taxon C | 0 0 1 1 1
taxon D | 0 1 1 1 1

  This is an example of a "perfect" matrix. One can do the same thing with
a gene tree. As you can see, it ignores convergence, and states in no
uncertain terms that the selection is absolute and cannot in any manner
show that a taxon can be possibly related to another form. This ignores
data, and negates homoplasy. Introducing a true data set, such as that of
Brochu, 1999, or Holtz, 2000, one can then observe a selection of regions
and states of those regions as found in nature, and there is plenty of
homoplasy. Some fo the best trees I've seen have had at least between
35-50% homoplasy, and this may be resolved later -- with the addition of
more character, refinement of earlier characters, and so forth.


Jaime A. Headden

  Little steps are often the hardest to take.  We are too used to making leaps 
in the face of adversity, that a simple skip is so hard to do.  We should all 
learn to walk soft, walk small, see the world around us rather than zoom by it.

"Innocent, unbiased observation is a myth." --- P.B. Medawar (1969)

Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.