[sci.bio] Is DNA hybridization phenetic?

rising@utzoo.UUCP (Jim Rising) (10/30/87)

Joe,

I'd be interested in your reasoning for saying that DNA hybridization
is not phenetic.  It would seem to me that a commparison of the unweighted
overall similarity in the DNA base sequences between two species could be
the ultimate phenetic method--depending, I suppose on how it was analysed.
Of course, I doubth that this is what DNA hybridization studies actually 
do, but don't they purport to do so?

--Jim Rising
-- 
Name:   Jim Rising
Mail:   Dept. Zoology, Univ. Toronto
        Toronto, Ontario, Canada    M5S 1A1
UUCP:   {allegra,ihnp4,linus,decvax}!utzoo!rising

felsenst@uw-entropy.UUCP (11/02/87)

In article <8858@utzoo.UUCP> rising@utzoo.UUCP (Jim Rising) writes:
>Joe,
>
>I'd be interested in your reasoning for saying that DNA hybridization
>is not phenetic.  It would seem to me that a commparison of the unweighted
>overall similarity in the DNA base sequences between two species could be
>the ultimate phenetic method--depending, I suppose on how it was analysed.
>Of course, I doubth that this is what DNA hybridization studies actually 
>do, but don't they purport to do so?
>
>--Jim Rising

People calling themselves cladists generally hold that use of the full
character data (such as DNA sequences) is "cladistic" and that methods 
that first reduce the data to a table of species-by-species similarity
(or distance) measures are "phenetic", and that the latter loses large
amounts of information (many believe: all the useable information) and is
invalid.

No doubt it loses some information, but certainly NOT all the useable
information.  Certifiably "cladistic" methods such as parsimony, which 
ignore all information from supposedly "phylogenetically uninformative"
sites also lose some information, but that doesn't cause them to get
labelled with this unfortunate useage, and I wouldn't dismiss them out
of hand for that reason.

My point is simply that "cladistic" and "phenetic" refer to approaches
to classification, and that labelling a method intended to infer a
phylogeny as "phenetic" wrongly implies that it is irrelevant to the
task.  The algorithmic similarity to methods of cluster analysis
used to create phenetic classifications is irrelevant -- you have to
ask how good these data and the analysis methods are for the task at hand.

I have done a crude calculation (for publication in a paper in Journal
of Molecular Evolution early in 1988) that Sibley and Ahlquist's data
have about as much statistical power as 4.4 KB of DNA sequences.  That
doesn't sound like a qualitative distinction to me, especially since
even with much less sequence a "cladist" would call sequence 
data "cladistic".

The (usual) confusion between classification and inferring phylogenies
is not helped by this useage.

Joe Felsenstein, Dept. of Genetics SK-50, Univ. of Washington, Seattle WA 98195
 BITNET:    uw-entropy!uw-evolution!joe%uw-beaver.ARPA@UWAVM
       or   uw-entropy!uw-evolution!joe%uw-beaver.ARPA@UWAVM.ACC.WASHINGTON.EDU
       or   uw-entropy!uw-evolution!joe%uw-beaver.ARPA@WISCVM.WISC.EDU  
 ARPANET:   uw-entropy!uw-evolution!joe@beaver.cs.washington.edu
       or   uw-evolution!joe%entropy.ms@beaver.cs.washington.edu
 UUCP:      ... uw-beaver!uw-entropy!uw-evolution!joe