erratum : Oragnutan relationships

pierre deleporte pierre.deleporte at UNIV-RENNES1.FR
Tue Apr 9 15:13:25 CDT 2002

Erratum :

>Date: Tue, 09 Apr 2002 15:00:23 +0200
>From: pierre deleporte <pierre.deleporte at>
>Subject: Re: Oragnutan relationships
>Strange enough... molecular analysis using parsimony doesn't require the
>definition of an ancestral sequence, outgroup rooting is sufficient
>exactly like for morphological data. The same way, morphological data do
>not require a priori statements of polarity. Outgroup rooting has largely
>lowered the requirement for primary homology assumptions whatever the data
>sets you use.

I should have written : "the requirement for primary polarity assumptions"

In fact outgroup rooting globalizes the polarity assumptions which are
implicitly decided by choosing the optimal graph without requiring an a
priori statement for every character of the data matrix.

Of course homology assumptions are crucial (alignment, for molecular data).


>  I think that molecularists are generally less interested in homology
> (not only primary assumptions of putative synapomorphy) particularly at
> the level of sites in sequences, because they generally have no
> evolutionary hypothesis for the presence of a given base rather than
> another one at a given position in a sequence... but the method is the
> same for molecular and morphological data in basic parsimony analysis.
>Approaches involving more elaborated models of character evolution differ
>(e.g. molecular clocks have hardly an equivalent in "morphological
>clocks"...). This question of underlying models is important and may be
>different for molecular and morphological traits, but the question of a
>priori / a posteriori synapomorphy statements seems largeley irrelevant to
>me in this comparison. As stated by Richard Jensen, assessing synapomorphy
>is always a posteriori, with reference to an accepted phylogeny
>("secondary homology" sensu de Pinna). If primary homology statements were
>to be preferred, why perform phylogeny inference at all ? Another question
>is the possible differentiel weighting of characters, but this leads us
>back to the question of assumed models of character evolution...
>A 08:29 09/04/2002 -0400, John Grehan wrote :
>>>Aren't all synapomorphies "identified a posteriori?"  I don't see how you
>>>can id
>>>a character as a synapomorphy unless you already have the phylogeny worked
>>>Richard J. Jensen
>>I was using Schwartz's terminology and I appear to have created confusion
>>by not providing the complete context of his statements which are as follows:
>>As such, molecular synapomorphy is identified a posteriori, after the
>>phylogeny is generated, rather than as a result of broad outgroup
>>comparisons that yield hypotheses of character polarity, some of which may
>>be in conflict. Molecular synapomorphies,however, can take the form of
>>either entire sequences of specific sites in sequences that are identified
>>only as a consequence of the clustering analysis used. Thus, there never is
>>any character conflict within a particular molecular study, only trees of
>>differing lengths. Whereas hypotheses of morphological synapomorphy can be
>>used in attempts to falsify other hypotheses of synapomorphy, no such
>>methodological procedure can be applied to phylogenies generated from
>>molecular data because the guiding principle in the latter endeavor is one
>>of "minimum number of steps" needed to achieve the most parsimonious theory
>>of relatedness.
>>In a morphological cladistic analysis, the issues of determining character
>>polarity are dealt with first and then competing theories of relationship
>>are generated from the alternative combinations of character states.
>>Basically, features are investigated prior to the phylogenetic assessment
>>of taxa. In molecular analysis, polarity is determined by specifying a
>>particular taxon (and thus its molecular attributes) as primitive relative
>>to the taxa of interest and then generating a tree based on nested sets of
>>increasing overall similarity among the non-rooted taxa. Molecular
>>synapomorphy is subsequently defined in the context of the
>>already-determined branching sequence.
>>John Grehan

More information about the Taxacom mailing list