[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

AW: Notarium question

> If you score them differently on the assumption that they
> are separate
> apomorphies, then you're assuming what you're
> setting out to prove.

On the other hand, if you score simply "notarium present (y/n)?", you may not 
be able to show they are separate apomorphies *even if they were* because your 
scoring assumes that the notarium is a monolithic entity - which any cladistic 
analysis will try to optimize towards a single origin. 

(This is the mistake that led Cracraft to "prove" that Hesperornithes, loons 
and grebes are a clade to the exclusion of all other birds living and extinct. 
In the case of the notarium, its presence was used to argue for the monophyly 
of "Metaves" - which, as it seems now, are almost certainly a pseudoclade. For 
a similar error, consider the fallacy "everything with feathery integument is a 
member of Aves".)

You can usually get around this by de-emphasizing the trait as a whole and 
concentrating on particular sub-traits - by considering the *structure* of a 
feature rather than its *existence*. In determining character choice & scoring, 
it helps that we have robust grounds to suspect a particular overall polarity: 
notarium fusion is apomorphic vs vertebral nonfusion, as far as can be told.

This need not yield "better" results (i.e. better-supported and less 
polytomies). But uncertainty that is correct is preferrable to certainty that 
is wrong, and in some cases - when the potentially misleading traits carry much 
weight in the scope of your analysis - this may be the only way to get results 
that are not outright wrong.

There are some more ways to avoid this problem; Kevin McCracken et al did a 
nice study of the issue some years ago: