Re: "Common ancestor" in cladistics

[Michael Habib <mbh3q@virginia.edu>]

At the risk of making things more confusing...

> That confused me, too.   Studies of phylogenetic relations of birds 
> use a
> related confusing term of basal, ie, some groups of birds are "basal" 
> to
> others.   (Unfortunately which groups are basal to which varies 
> completely
> by the study.)

I actually like the terms basal and derived, but perhaps I am unusual.  
Basal simply means being closer to the ancestral character set, and 
derived is further from the ancestral character set (reversals are 
derivations, they do not make taxa more basal...though they may appear 
so).  Trees do vary a great deal, so what is ancestral, basal, derived, 
sister-to, etc. all varies.  Oh well, so it goes...

> Methodologically we rarely have DNA from any common ancestors at all, 
> so
> what is done is to chart or measure genetic differences between 
> existing
> groups, on the theory that the groups with the most genetic 
> differences are
> teh most distantly related.

True, although genetic differences are not the only characters that can 
be used to determine trees.  Also, distance is only one metric.  
Parsimony is commonly used, which tracks the implied changes, rather 
than total difference.  There are a number of other methods as well.

> I know part of it is that researchers assume that the greatest genetic
> differences took the most time to develop - but I'm not convinced that 
> that
> actually tells us which group happened first if we have no ancestral 
> DNA to
> compare it to.

Again, good point.  This is where outgroup comparisons come in.  They 
are used to "set the zero".  That is, they define the root of the tree 
you're building, and then you can track the changes from this point.  
(That's a gross over-simplification, but generally speaking...maybe 
someone with more expertise than I can chime in to help explain).

There are a couple of generally aspects of phylogenetics worth keeping 
in mind that may help.  One is that trees do not really map taxa; they 
map characters.  We call particular groupings of characters species, 
genera, etc., and so use these maps to estimate how organisms are 
related, but ultimately the computers are crunching characters, not 
animals.

Another thing to remember is that trees map divergences (at least those 
created by pure cladistic methods).  This means that the generally 
assumption is that all nodes should have two branches; more (a 
polytomy) are said to imply a lack of agreement/resolution.  Also, as 
already mentioned, this divergence pattern assumes that the ancestors 
will never be both found and identified as an ancestor.  This is a 
reasonable assumption, because it is very likely, but strict cladistics 
assumes that it does not happen at all.  The common ancestors can only 
be estimated (though their character sets can be partially 
reconstructed).

Cheers,
--Mike Habib