Re: New AVES definition refined (more testable?)

["David Marjanovic" <david.marjanovic@gmx.at>]

> Paraphrasing (I think), you're saying that it is inappropriate
> to make a list of characters and conclude that a species which
> has X number of those characters must be related to another species
> or group of species.

I think he says it's inappropriate to use the possession of _any_ out of a
number of characters as a synapomorphy. As a fictional example, to define a
group by "the possession of either D-sectioned or unserrated pmx teeth, or
both" is certainly not a good idea.

> Your reason is that:
> <As far as I know, some genetic agent that groups a certain number
> of characters together, then expresses a random sampling of them
> in every generation, is not known.>
> [...] isn't it possible to say that (just inventing) someone's
> descendants may be identified by either or both of two uniquely
> shaped birthmarks?  It is theoretically possible to have alternative
> unique connections, no?

The occurrence of such a situation gets difficult to test a posteriori. I
think such a phenomenon would never be used as an apomorphy, it would be
considered individual variation... which is what it is, actually, no?

> Also, and more significantly, as a lineage evolves through time,
> some of the characters which linked a species to its ancestors
> may disappear.  Part of the complexity of cladistic logic is
> that of linking one species to another, which in turn has characters
> which help to identify certain characters as derived/shared from
> the mrca.  Thus, a descendant species need not have many of the
> identifying marks of an ancestral species to be included.
> The list of characters linking two species may not be the same
> as the list of characters linking those species to their mrca,
> no?

Fully correct.

> In that sense, if you drew up a list of characters used in establishing
> a whole hypothesis of relationships, some of those animals would
> have some items on the list and others not.

But you won't get a random pattern. You might get something like:

character        taxon
12345
10000            1
11000            2
11100            3
01100            4
00110            5
10010            6
00011            7
01001            8
10001            9

In this (erm, _very_ simple) case, you'll be able to produce a most
parsimonious cladogram by hand, despite the reversals:

--+--1
   `--A--2
       `--B--3
           `--C--4
               `--D--5
                   `--E--6
                       `--F--7
                           `--G--8
                                `--9

Synapomorphies:
A: acquisition of ch. 2
B: 3
C: loss of 1
D: 4, loss of 2
E: loss of 3
F: 5
G: loss of 4

Autapomorphies:
6: reversal of 1
8: reversal of 2
9: reversal of 1

But actually... I just see that 6 might come out in a trichotomy with 1 and
A. Also, it takes 2 steps to reach 8 from (the metataxon) 7, but also from
(the metataxon) 2. It's not possible without some calculating to find the
MPT here :-] Take-home message for me: I've used too few characters compared
to the number of taxa. Gotta look for more.

> If I'm reading you correctly, your observation is about the question
> 'Are these two species related' as opposed to the question, 'Do
> these two species share a common ancestor?'

Certainly not -- as a cladist, he equates these 2 questions :-)

> Cladistic logic, as I tromp along in its path, has an amazing
> number of conditional assumptions and complex associations.

Which might be why cladistics is a scientific and not a philosophical method
:-)
Though in this case I think you should just read better literature on
cladistics, such as the archives :-)