[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Re: Benton et al.'s Supertree (long)
Michael Bruce Habib wrote-
> As someone working on supertree construction and analysis
> as we speak, I have to disagree to some extent with your
> comments. I agree that supertrees are not ideal, but they
> offer coverage across large groups not possible in a
> reasonable time-frame (or at all) using 'traditional'
Supertrees are potentially useful, but only if the characters used are not
repeated. Thus, supertrees combining various molecular sequences are good,
for example. Wide coverage doesn't matter if the results are questionable.
> Keep in mind that using multiple trees that include the
> same data implies, in part, that multiple researchers agree
> with that tree topology. That is, this implies some kind
> of consensus between those researchers, and this is
> essentially the point to begin with. That being said,
> there are still some concerns in repeating data as it were,
> and this can be compensated for if one wishes. Regardless,
> it is likely that this repetition does not affect the
> resulting topology. Weighting schemes of that type often
> produce the same supertree (or nearly so).
No, the use of multiple trees with the same data implies that historically
earlier discovered characters will be overweighted in the analysis. This is
not a defendable reason for weighting characters. Any "consensus" seen due
to this is caused by a lack of new data or ideas until reexamination or new
specimens indicated an alternate phylogeny.
> Second, part of the point in putting together a supertree
> is that it is not always known what is a "good" tree or
> "good" phylogenetic information. They are supposed to
> include all opinions, not just the most recent papers. Yes,
> including information from taxonomies or older phylogenies
> may seem a bit strange, but this information could be
> useful. I, for one, am not willing to say that I always
> know what trees are good and which are not; that is one
> area where supertrees come in handy.
How is valuing someone's undefended opinion of a relationship equally to
someone else's result of a 300+ character phylogenetic analysis useful? The
trees extend back to 1984. Do you know how much we've learned since then?
Cladograms from that time are obsolete. Paul (1984) thought that segnosaurs
weren't theropods. Information found since 1993 proves they were, even Paul
now agrees (DA). Yet Paul's earlier study is still used by Benton et al..
How is this useful in any way? Holtz (1994) placed Chirostenotes in the
Arctometatarsalia away from oviraptorids. Subsequent finds (Sues, 1997,
etc.) prove it's an oviraptorosaur and Holtz now (2000) completely agrees.
Holtz's 2000 analysis contains all the valid characters from his 1994
analysis plus many more, and a lot more taxa. Yet both are valued equally
by Benton et al.. How can this be justified? An earlier tree with less
characters and taxa available can only be "bad" compared to a modern
> Lastly, supertrees can, in fact, give valuable information.
Perhaps valuable historical or sociological information, but not valuable
> They give wider coverage (in terms of how much of the given
> group is represented) than most trees; this is required for
> some types of analysis.
But at the cost of accuracy (which is of prime importance I would think).
If you want to do this, either make a huge analysis (like Holtz 2000 did for
theropods) or make lots of separate analyses of included taxa and combine
the results for a single tree (like Sereno 2000, despite his methodologies).
> They show where disagreement is
> rampant, as well as where agreement in the literature is
They show what clades have lasted in the literature the longest since 1984,
not what the current consensus is. I fail to see the phylogenetic use of
such a measurement.
> Finally, supertrees give the result of using
> several types of characters/phylogenetic information.
They CAN, but Benton et al.'s doesn't. It's all morphological. As long as
the characters in each included analysis are separate, I have no problem
with supertrees. If you made a supertree using separate 12S, 16S,
behavioral and morphological data sets, it would be fine. Benton et al.'s
however, does not have this advantage.
> Since supertrees usually include a unique assortment of
> taxa (ie. more than other trees) and present a new
> combination of phylogenetic information (ie. include
> trees based on many different methods or characters), I
> would argue that they present new information on
> phylogenetic relationships.
Since Benton et al.'s supertree sacrificed accuracy for the large number of
taxa, used only morphological characters and included trees formed by only
two methods (traditional parsimony analysis and subjective creation of trees
from one's own opinions), I don't see how it presented new information on