falsifying evolution

[P.Hovenkamp]

At 02:04 PM 17-08-99 -0700, Kipling Will wrote:
>Richard Zander wrote:
>
>> The CI of all trees was around .4, similar to that of real data sets.
>> Clearly, real molecular data in published trees usually do support
>> morphological data sufficiently to have drowned out a "random pattern" that
>> is quite distinct, but tests are needed to evaluate molecular data that do
>> not.
>
>If you have more than 60 taxa a CI of .4 is quite good. CI is directly
>effected by number of taxa see the following...
>
>Sanderson, M. J. & Donoghue, M. J. (1989). Patterns of variation in
>levels of homoplasy. Evolution 43, 1781-1795.
>
>de Queiroz, A. & Wimberger, P. H. (1993). The usefulness of behaviour
>for phylogeny estimation: Levels of homoplasy in behavioural and
>morphological characters. Evolution 47, 46-60.

Totally random patterns show a similar correlation to the one found by
Sanderson and Donoghue, rather closely following their line at a lower
level; with values for CI slightly under .40 with 10 taxa, dropping rapidly
to slightly over .10 for 30 taxa, and then trailing off to under .05 at 70
taxa or more (simulation performed with binary data with both states
occurring with approximately the same frequency).

Utilities similar to the one used by Richard Zander must be widespread -
developing one is an excellent first task for budding programmers...

Best wishes,

Peter Hovenkamp
P. Hovenkamp
Rijksherbarium, Leiden
The Netherlands
hovenkamp at rhbcml.leidenuniv.nl
http://rulrhb.leidenuniv.nl/