[Previous by date - RE: Re: RE: RE: Nathan Wilson's question]
[Next by date - Re: Moore's hybrid example (was Nathan Wilson's question)]
[Previous by subject - Re: Monotypic genera and the PhyloCode [was: Re: Lumping Spi=]
[Next by subject - Re: Moore's hybrid example (was Nathan Wilson's question)]
Date: Tue, 17 Oct 2000 16:42:40 -0700 (PDT)
From: Nathan Wilson <velosa@cinenet.net>
To: PhyloCode mailing list <PhyloCode@ouvaxa.cats.ohiou.edu>
Subject: Re: Moore's hybrid example (was Nathan Wilson's question)
First off, let me say that I'm very impressed by the amount of discussion
that my concerns have generated. I was out of the country for two weeks
and have only now caught up with the discussion. I particularly
appreciate Gerry continuing to argue that there is an unaddressed concern
(and for the rather flattering comparison to a revolutionary war hero :).
With regard to the clade diagrams that Gerry and Kevin have been
discussing (reproduced below), I agree with Kevin that his (Z) clade
provides the best interpretation in the case of a single hybrid. As I've
mentioned, I'm primarily a computer scientist so please forgive me if my
terminology is a little off. In particular, I'm not sure what to call the
branch points in the clade diagram, but as I understand it, they are
hypothesized ancestral species of the species that occur underneath them.
Using this terminology, Kevin's solution is consistent with the node-based
clade definition given in the Phylocode if you define a `most recent
common ancestor' as any ancestor which is an ancestor for all the given
specifiers that has no descendent which is also an ancestor for all the
given specifiers. In Gerry's solution the top most ancestor, Z, does not
meet this definition since the top of Kevin's (Z) clade is a descendent
which is an ancestor for all the given specifiers.
Gerry's solution can work if you define the most recent common ancestor as
the ancestor of all the given specifiers through which all paths to all
the specifiers from the origin of life pass and which has no descendent
through which all such paths pass. Comparing the two definitions, the
first definition has the advantage of always being a subset of Gerry's.
This means it is less radically effected by the discovery of hybrids.
However, the second solution (Gerry's) is nice because it does necessitate
that the most recent common ancestor is unique. The first definition does
not. To see this, I start with the diagram Kevin provided of Gerry's
problem:
3
/ / / / / / / / / / 7 12 9 / 10 5 13 6 1 4 \ 2 8 11
\ / \ / \ / \ / \ / \ / \ /
\/ / \/ \/ \/ A\ \/
\ / / / \ \ /
\ / / / \ \ /
\ / / \ /
\ / / \ /
\ / / \ /
B\/ / \/(Z)
\ / /
\ / /
\/ /
\ /
\ /
\ /
\ /
\ /
\ /
\ /
\ /
\/Z
If we now discover that 4 is also a hybrid, but of different species in
clades A and B than 3, we end up with something like:
4
/\ 3
/ \ / / \ / / \ / / X / / \ / / \ / / \ / / \ / / \ / / \ 7 / 12 9 / 10 5 13 6 \ 1 \ 2 8 11
\/ / \ / \ / \ / \ / \ / \ /
\ / / \/ \/ \/ A\ \/
\/ / / / \ \ /
\ / / / \ \ /
\ / / / \ /
Y\ / / \ /
\ / / \ /
\ / / \/(Z)
B\/ / /
\ / /
\ / /
\/ /
\ /
\ /
\ /
\ /
\ /
\ /
\ /
\ /
\/Z
Using the definition I created to support Kevin's interpretation, both Y
and (Z) are most recent common ancestors of 3 and 4. Using the definition
I created to support Gerry's interpretation, Z remains as the most recent
common ancestor. So the two possibilities I see are to allow for multiple
most recent common ancestors in node-based clades (my proposal) or to
accept the Gerry's solution which adds a rather disturbing level of
instability whenever a hybrid is discovered.
This brings me to the the solution proposed by Michel (Laurin). He
suggested changing a clade specifier to a closely related one when it is
discovered to be a hybrid. The first problem I see with this solution is
that there may be no good alternative (non-hybrid) candidate to change
the specifier to.
The other problem is it forces an even stronger reliance on the concept of
species by requiring a clear definition of what a hybrid is. As I
mentioned before, expanding the definition of a node-based clade to allow
for multiple most recent common ancestors has the added benefit of
eliminating the reliance on a species concept. In cases where species are
clearly defined and do not hybridize everything will have unique most
recent common ancestors. At the other extreme, within a traditional
species you can end up with a fair number of most recent common ancestors,
but it provides a valuable way to describe clades that correspond to races
and other concepts within populations that only have weak barriers to
reproduction.
Nathan Wilson
Computer Scientist
Collective Source