falsifying evolution/Bremer support

[John Trueman]

Richard Zander wrote
>
>I generated five random data sets mimicking 800-bp molecular data sets with
>5 taxa each. All produced single, fully resolved trees. In most, there was
>at least one branch with Bremer support of 4, and in one, there was one
>branch with Bremer support of 10.
---

What this demonstrates is that the absolute value of the Bremer support
statistic is hard to interpret. Without some form of null distribution who
can say whether or not a score of 4, or even 10. is significant? Clearly,
from this experiment the probability of achieving one branch with a score
as large as 10 in a 5 taxon-800bp dataset evolving according to whatever
evolutionary process was used, by chance alone, is of the order of 20%. A
Bremer support of 10 ought not to be considered significant for such data.

The TPTP test is designed for just this sort of situation. For a defined
branch on a tree obtained from real data (or a defined taxon partition
which could have appeared but failed to appear on the tree) the TPTP
compares Bremer support actually achieved against a null model of the
Bremer support which can be achieved in data matrices with that number of
taxa, that number of characters (bps), that state (base) usage, etc., only
WITHOUT the pattern of hierarchic relationship which a shared evolutionary
history would impose.

My forecast is an a-posteriori TPTP test on the group which yielded a
Bremer score 10 will show this pattern to be not significant.

John Trueman


======================================================================
John Trueman
Faculties Research Fellow
Bioinformatics Group
Research School of Biological Sciences
Australian National University
Canberra, ACT 0200,  AUSTRALIA


ph: +61 2 6249 4840
fax: +61 2 6279 8525
email: trueman at rsbs.anu.edu.au

Reason is a tool. Try to remember where you left it.
=======================================================================