Message 2000-10-0026: Re: Hybrid specifiers

Mon, 23 Oct 2000 15:42:32 -0700 (PDT)

[Previous by date - Re: Hybrid specifiers]
[Next by date - Defining species]
[Previous by subject - Re: Hybrid specifiers]
[Next by subject - Re: IGNORE THAT LAST MESSAGE: Fwd: PROPOSED ARTICLE X - aut=]

Date: Mon, 23 Oct 2000 15:42:32 -0700 (PDT)
From: Nathan Wilson <>
Subject: Re: Hybrid specifiers

On Fri, 20 Oct 2000, Kevin de Queiroz wrote:

> Nathan stated:
> I'm not sure how relevant this is to the issue we've been discussing,
> but I would like to know why Nathan thinks there is no good definition
> of the term "species" (since I have argued to the contrary in several
> papers). 

I'm afraid you have caught me without doing all my homework.  I have not
had the chance to read your arguments in this regard.  If you could give
me the references to your arguments, I'll try to look them up the next
week or two.  One of the most frustrating things for me is that I do not
have easy access to scientific journals.  I pretty much have to set aside
half a day to go to UCLA and hope that I can find the journal in question. 
I wish all this information was on-line :(.

This issue *is* at the center of my thinking about the problem so I think
it is worthwhile trying to get some clarity about it in this forum.  I
will directly answer this question in a following message.

> I'm also uncertain about whether Nathan ever answered my question about
> the reference of the name Alpha.  I get the impression from his last
> message that he is viewing the clade as stemming from both Y and (Z). 


> If so, he is indeed "changing what is meant by the term clade."  The
> definition of the term "clade" is a common ancestor (or ancestral
> species) and all of its descendants (not Nathan's "set of all
> descendants of a particular hypothetical species," since this definition
> excludes the ancestral species from the clade).  If this more-or-less
> standard definition is adopted, then there are two clades involved in
> Nathan's example, one stemming from Y, the other stemming from (Z), and
> in this particular case those clades are partially overlapping (have
> species 3 and 4 in common).  If Nathan wants to apply a name to the
> combination of these two clades, that's OK, but we probably should call
> it something other than a clade (such as a set of two partially
> overlapping clades).

I understand the distinction and was trying to point it out myself in
my last note.  I would be perfectly happy adopting a new term, but don't
have a good idea what that term should be.  The point of view I'm taking
is, given the information used in the Phylocode to define clades, what
precise groupings can you create.  That is what sets of individuals
can be precisely determined assuming you had complete knowledge of the
ancestry of every individual.

> It also occured to me that under one of the ways of stating a node-based
> definition, the application of the name might not be ambiguous even in
> this confusing case.  That is, if Alpha is defined as the clade stemming
> from the most recent common ancestor of 3 and 4 (as opposed to the least
> inclusive clade containing 3 and 4), then unless the splits at Y and (Z)
> occurred at the same instant (highly unlikely), one of the ancestors (Y
> or (Z)) is more recent than the other, and thus the name refers to only
> one of the two clades (though in practice it might not be easy to
> determine which one). 

Interesting idea, but I'm concerned, as you are, that there would be no
easy way to determine which ancestor takes priority.  This could get
particularly ugly if the rate of speciation or time between generations
within the species are different along the different paths from Y and (Z)
to 3 and 4. 

I also had an interesting realization about the Michel Laurin's idea of
changing the specifiers to deal with hybrid specifiers.  The simple
realization is that adding a specifier either leaves the clade unchanged
or makes it more specific, but replacing a specifier can expand the clade. 
In fact the result of having both specifiers is the intersection of the
clades defined using just hybrid and the one using just the replacement. 
This means it is better to simply add the new specifier, rather than
replace the hybrid specifier with the new specifier. 

More precisely, if you use the definition that a most recent common
ancestor is an ancestor of all the specifiers which has no descendant
which is an ancestor of all the specifiers, then simply adding additional
specifiers to the definition is more likely to result in a unique most
recent common ancestors than replacing the hybrid specifier. 

Michel's goal in replacing the hybrid specifier was to improve the chances
that there is a unique most recent ancestor.  The expectation is that the
new specifier is not a hybrid, so simply adding it to the list of
specifiers eliminates the extra ancestor even if you leave the hybrid as a
specifier.  Furthermore, if you leave the hybrid in, then you eliminate
the chance that the set of other specifiers have some other most recent
common ancestor which the hybrid previously eliminated. 

Best regards,


Feedback to <> is welcome!