[Taxacom] Paraphyly/predictivity

[Richard Zander]

I must have been dense not to recognize your (rather oblique) reference to Saussure's syntagm. (You other taxacomers doubtless picked up on this immediately, right?) Although I appreciate your belaboring the point, Peter, I totally reject the analogy.
 
Saussure suggested that, linguistically, nouns that can be totally replaced in sentences may be termed paradigms. My position is that evolutionary taxonomy, phenetics, morphological parsimony analysis, and molecular parsimony/Bayesian analysis are all analytic efforts to reveal facts about evolution, and are complementary and overlap significantly. They are not paradigms. They can all contribute to a classification to the extent they reveal information about evolution.
 
Cladists remove unique traits before creating the, then purified, "data set," and remove autapomorphies after the analysis (traits that do not prove locally synapomorphic). Both unique traits and autapomorphies are evolutionary information.
 
Morphological analysis creates sister-group relationships that are similar but not identical to those from molecular analysis, which means that we do (through consilience) get information on sister-group relationships but such information is lacking detail. Sister-group relationships from morphological information are limited by the fact that traits are treated as independent, yet selection links them. For instance, three traits linked by selection should be less important in parsimony than two unlinked traits, but this is not the case. Also, survival of ancestors (static populations maintained by stabilizing selection) results in polytomies that are randomly resolved by interpretation of the some of the traits of descendants as synapomorphic, and there is no direct provision for identifying surviving ancestors. 
 
Molecular analysis gives you only a tree (of one gene or many) that indicates genetic isolation and coherence of lineages, it does not assure that isolation results in speciation or that one taxon (perhaps of two static populations that each produce daughter species by peripheral isolation) may not exist in two different lineages. 
 
Combining molecular and morphological phylogenetic understanding of what we know about evolution results in a sister-group tree of life that is poorly resolved in detail and sometimes in major aspects. Necessarily so, in view of the present limitations of sister-group analysis. Surely, surely, addition of evolutionary information from autapomorphies and clearly implied ancestor-descendant relationships must make classification better.
 
Information on autapomorphies is a product of cladistic analysis. Even elimination of unique evolutionary information while preparing a data set is part of cladistic analysis. This information is often more important and more assured than sister-group clustering. The so-called paradigms do overlap. 
 
Elimination of ancestor-descendant information in classification is wrong, in part because diagnoses of evolutionarily significant taxa disappear. In some cases, such as recognizing the Cactaceae as a tribe or subtribe of Portulacaceae, a diagnosis with evolutionary information in it survives and can be located by someone interested in evolution. But in other cases, such as the lumping of the moss genera of Ephemeraceae, Splachnobryaceae and Cinclidotaceae into the large family Pottiaceae without discussion beyond enforcement of holophyly or with establishment of subfamilies or tribes for them, is more than disconcerting.
 
_______________________
Richard H. Zander
Missouri Botanical Garden
PO Box 299
St. Louis, MO 63166 U.S.A.
richard.zander at mobot.org
 

________________________________

From: Peter Stevens [mailto:peter.stevens at mobot.org]
Sent: Sat 3/28/2009 4:10 PM
To: Richard Zander
Cc: Mike Dallwitz; TAXACOM
Subject: Re: [Taxacom] Paraphyly/predictivity



Not to belabo(u)r the point, but let me repeat: thinking of common 
nouns and classifications, in folk classifications and in language 
generally, when nouns are members of contrast sets, they are 
generally non-overlapping - that is, one does not include the other. 
It clearly helps in communication.  This might be relevant when we 
are thinking of paraphyly and monophyly - [and autophyly].

As others have mentioned, classifications are for communication, and  
as such they should work within the constraints of language/
communication.