Cladistics and "Eclecticism"

[Pierre Deleporte]

At 19:30 05/02/2002 -0800, Curtis Clarke wrote:

 >Tom is on his
>own when it comes to his views of speciation. I understand them, they make
>logical sense to me, but I don't find them useful for dealing with the
>biological phenomena. Even among the "orthodox", there are differences of
>opinion (suggesting that perhaps the epithet "strict cladist" is undeserved
>or misapplied).

Important precision. I think that population biology may be very little 
concerned by an initial stage of possible population split with no 
differentiation, and sytematics is likely not concerned at all. The 
"eclecticism" debate seems better argued in terms of well differentiated 
taxa. Defining "monophyly" as well (see Turner), though Tom's general 
notion of a "complete" clade makes sense for me (even if concerning the 
undifferentiated "terminal taxon"). I would not apply this inside 
populations anyway.
I think that identifiability of taxa through character analysis matters 
(more below).

>At 07:18 AM 2/5/02, Pierre Deleporte wrote:
>>But in the real life, we did not observe the evolution of the populations.
>>Then we are facing two different taxa, because they carry different
>>character sets.
>     Curtis Clark:
>Questions of classification aside (for the moment), those of us that study
>speciation do in fact want to infer that which we cannot observe, and one
>good piece of evidence for speciation by peripheral isolation is that the
>putative "ancestral" species has no apomorphies that are not also shared by
>its derivative. If enough time passes that the ancestral species develops
>autapomorphies of its own, then the signal is lost, and ordinary Hennigian
>rules are perfectly adequate.

I fear I partly disagree, but I have to specify my point on an example:

1) the hypothetical history :
    outgroups  O1, O2..., sister species  A  and  B,  then B splits into B1 
and B2


         Characters :

01:     000000000...
02:     000000000...
A:      000000123...    (1 = synapomorphy with B, 23 = autapomorphies)
B:      000000145...    (1 = synapomorphy with A, 45 = autapomorphies)


then B splits, B1 does not change, B2 acquires autapomorphies :

01:     000000000...
02:     000000000...
A:      000000123...
B1:     000000145...    (4 = synapomorphy with B2, 5 = symplesiomorphy with 
B2)
B2:     000000146...    (4 = synapomorphy with B1, 6 = autapomorphy of B2)


I understand the above statement of Curtis like this: 5 is no autapomorphy 
of B1 "of its own"; thus 5, compared to 6, is a "piece of evidence" of 
preripheral isolation of B2, because the "ancestor-like" B1 has no 
apomorphies of its own.
But this makes sense only if we know the true ancestor B: 000000145...
In the real systematicians life, we simply don't know B, thus we have no 
"piece of evidence" and cannot decide if character state 5, or 6, or 
neither 5 nor 6, is the ancestral condition for B1 + B2 (hence my recent 
"erratum" posting: the data are not really misleading, they are simply 
ambiguous).

B1 and B2, as they stand, are thus treated by the systematician as quite 
"ordinary" different terminal taxa. This is true even without any 
differentiation of B1 relatively to B.
The plesiomorphic condition of 5 is undetectable. This "particular case" is 
thus in no way particular.
Unless we would arbitrarily "infer what we cannot observe", i.e. decide 
that 5 is plesiomorphic (but why not 6, or none of them, given the data 
really at hand?)

The example is thus not relevant to the "eclecticism" debate. It simply may 
help underlying the possible ambiguous status of character states is sister 
taxa, a well known limit of character state optimisation;
and more generally, the possibly multiple evolutionary interpretations of a 
same caldogram (more nice simple examples, including reticulation, in e.g. 
Skala and Zrzavy, Cladistics 10-3, 1994).

The be complete (maybe boring, sorry...), the following example would be 
relevant to the eclecticism debate:

01:     000000000...0000
02:     000000000...0000
A:      000000123...0000
B1:     000000145...0000
B2:     000000146...789X    (789X = autapomorphies of B2)

Here, evolution "accelerates" in B2, and this is detectable through the 
amount of "true and detectable" autapomorphies 789X, raising the questions 
of significant or not significant "gap" or other criteria for deciding to 
name or not to name a group (be it mono-holo or paraphyletic), and the 
supplementary question of the conventions for possibly naming a 
paraphyletic group (01+02+A+B).

But this has clearly nothing to do with the plesiomorphic status of 5 
versus 6 (undetectable), hence nothing to do with the unchanged descendant 
B1 (undetectable) relative to B (unknown).

Now I fully agree with Curtis that, under a model of peripheral isolation 
evolution with founding effect and acceleration of evolution in the budding 
species ("saltation"), the example above could be interpreted by ecologists 
as B2 likely budding out of ancestral B = strictly identical to unchanged B1.
This is a possibility. But it remains that 5 is merely "different from 6" 
at face value, and 5 = autapomorphy is also a possibility, because B1 could 
have evolved a little bit. Who knows? Peripheral isolation is a vicariance 
event after all... So, where is the point for systematics? Indicating the 
possible alternatives? With markers?
Isn't it sufficient fo the systematician to provide information about the 
number of (possible) autapomorphies in B1 and B2, and everybody using the 
phylogeny will make up one's mind?
Should branch length information be added in the nomenclatural conventions? 
With markers?  Following which criteria?


    Curtis (about conventions for reticulation):

>It's the situations in the middle that "matter", but even there, the
>problem is often avoided. Our bread and pasta wheats are the result of
>intergeneric hybridization between Triticum and Aegilops, except that now
>most botanists put the Aegilops species in Triticum (not to avoid
>"cladistic problems" but rather in an attempt to pay lip service to a
>reproductive species concept). Sometimes species of intergeneric hybrid
>origin are originally given "hybrid names" that properly belong only to
>F1s, and the name sticks (Raphanobrassica comes to mind: the artificial
>alloploid between cabbage and radish).
>
>A genus I study, Encelia of the Asteraceae, has two well-defined clades,
>and one of the species of hybrid origin has parents from each clade. Were I
>to name the clades as subgenera, I would be faced directly with the issue
>of where to put the hybrid species.

Thanks a lot for comments. What would you do?

Anyway it's amazing how the terrific reticulation problem seems to be dealt 
with so quietly by people standing on the front line: botanists.
Is it finally that simple to add a little complementary information to a 
necessarily schematic conventional system when useful, and all is clear for 
everyone concerned?

Cheers
Pierre



Pierre Deleporte
CNRS UMR 6552 - Station Biologique de Paimpont
F-35380 Paimpont   FRANCE
Téléphone : 02 99 61 81 66
Télécopie : 02 99 61 81 88