Cladistic hypotheses

[Richard Jensen]

I disagree with Kirk's comments that no cladogram can be used to test any other cladogram.  I use data set A to construct a cladogram.  I then use an alternative dataset to "test" the first cladogram  (and, simultaneously, test the cladogram derived from the alternative data set).  If two independent data sets produce identical (toplogically identical) cladograms, then this agreement can be viewed as a test of the hypothesis that the pattern revealed by one set of characters will be supported by a second set of characters.

Of course, given that I don't know the true cladogram, I can't use data set A to "test" data set B (or vice-versa).  But, I can test the hypothesis that the two data sets agree (i.e., that the two cladograms are the same).

As for G.B.'s comment that data are not hypotheses, I question this as well.  I measure the petioles of a set of leaves from species A and the petioles from a set of leaves from species B.  In order to conduct a valid comparison, I have to hypothesize that the two sets of measurements are commensurable, i.e., that the data have been derived from homologous structures.  Similarly, when I code species A as having a rotate corolla, I am making a hypothesis for the species based on the observations that I have made.  Any individual observation, in and of itself (e.g., this flower has a rotate corolla) is not a hypothesis, but once I begin the work of the comparative morphologist, my observations (data) become hypotheses.

A happy Thanksgiving to all who enjoy this truly American (sensu lato) holiday.

Dick J.

----- Original Message -----
From: "J. Kirk Fitzhugh" <kfitzhug at NHM.ORG>
Date: Wednesday, November 23, 2005 3:39 pm
Subject: Re: Cladistic hypotheses

> At 10:17 AM 11/23/2005 -1000, you wrote:
> >I didn't entirely understand Kirk's message that prompted this,
> but as I
> >see it, you test the cladogram by running an analysis and
> determining if
> >it is the best one available according to whatever criteria
> (parsimony,>ML, Bayesian, etc.) you are using.  Of course this is
> all automated so
> >that the "hypotheses" are generated automatically.  The only
> thing that
> >seems unusual (in terms of deviating from typical scientific
> theory) is
> >that a hypothesis (i.e. cladogram) can't really be falsified on
> its own
> >but only when you find another one that has better support.
>
> Unfortunately, there is nothing within the inference of cladograms,
> regardless of what software one uses or mental inclination one
> has, that
> has anything at all to do with testing.  The long-standing notion
> that one
> cladogram can test/falsify another cladogram is mistaken.  One
> cladogrambased on data set a+b+c cannot be evaluated against a
> cladogram based on
> data set a+b+c+d (or even mutually exclusive data sets).  The two
> hypotheses stand independent of one another as entirely different
> causalexplanations for different sets of causal questions.
>
> Kirk
>
> -----------------------------------------------------
> J. Kirk Fitzhugh, Ph.D.
> Curator of Polychaetes
> Invertebrate Zoology Section
> Research & Collections Branch
> Los Angeles County Museum of Natural History
> 900 Exposition Blvd
> Los Angeles CA 90007
>
> Phone:   213-763-3233
> FAX:     213-746-2999
> e-mail:  kfitzhug at nhm.org
> http://www.nhm.org/research/annelida/staff.html
> http://www.nhm.org/research/annelida/index.html
> ----------------------------------------------------
>