Cladistics and "Eclecticism"

[Pierre Deleporte]

At 10:44 06/02/2002 +0100, Zdenek Skala wrote:
>Dear all,
>the current paraphyly/holophyly discussion seems to point constantly to 
>one question:
>Is the amount of character change (better of character state difference in 
>an operational context) a core subject of systematics?
>(...)
>  there were several requests within the current thread for criteria used 
> by eclecticists for delimiting paraphyletic taxa. I would like to repeat 
> again that this is *not* the sole problem of paraphyletic taxa; the same 
> problem arises when delimiting the holophyletic taxa. For some of 
> cladists, the answer seems to be that limits of taxa are arbitrary 
> decisions (practical, not scientific) and hence there is no problem. I 
> would disagree and hope that many cladists will, too. If the system of 
> taxa would be an arbitrary split of a cladogram, systematics is losing 
> its merit IMO - the phylogeny was originally intended to be a *tool* how 
> to naturally explain and summarize the organisms' diversity!

Once you have the data matrix and the tree, you have this tool available 
for science. Naming groups is a question of conventions for efficient 
communication.

>Hence, let's discuss the criteria of the taxa limits and leave for a 
>moment if these limits will be used in a holophyletic or paraphyletic 
>solutions.

Nice suggestion.

>  I personally believe that the amount of apomorphies at the respective 
> nodes gives a good starting point for taxa-limits optimization process.

Yes, if you need to deal with such "well-separated groups" for some 
reasons, then you will like to name them.
Now this is not the only possible situation. As a student of social 
evolution, I could like to name a clade characterised by a synapomorphy in 
this respect (social organisation type Y), and also to name a subclade 
characterized by a further change (e.g. reversal to type X), whatever the 
total amount of changes in other respects.
Thus the number of changes in different characters is not the only possible 
criterion, and characters of interest matter (by definition, they are "of 
interest" for the specialist, and differ according to questions at stake...).
This is why naming some taxa (not all taxa) is a practical question, and 
neither "natural", nor properly "scientific" in any non-trivial way.
"Optimization" in this field depends on the centers of interests of the 
users: which groups we need to name for some concerns (quantitative or 
qualitative).
(Naming all taxa is not an issue here, it boils down to "redrawing" the 
complete tree with words. Maybe largely a waste of time)...

>There was made also a separate point that paraphyletic taxa having 
>"origin" and "end" need to have "more arbitrariness" to be defined than 
>the holophyletic taxa.

I can recognise one of my posts on this topic.

>  I believe this is not true: let's imagine a pectinate cladogram of 
> species A,B,C,D,E where clade A+B is supported by 5 synapomorphies, A+B+C 
> by 2 syn., A+B+C+D by 10 syn. and the entire group by 12 syn. Eclecticist 
> will split the cladogram into, say "genera" A+B, C+D and E and families 
> A+B+C+D and E.

You mean that two significant gaps are retained for naming: 5 changes and 
10 changes.

Besides the trivial terminals and complete clade (ABCDE), note that 
restituting complete phylogenetic information requires here 3 genera and 
two families, thus 5 names,
plus complementary information (e.g. Kinman markers) indicating the 
paraphyletic nature of C+D and where do A+B belong,
plus extra marking (or an ordering system) indicating the hierarchical 
nesting of ABC inside ABCD.

>  Cladists, I believe, will use taxa A+B, C, D, E and more inclusive taxa 
> A+B+C+D and A+B+C+D+E.

You mean that they will possibly name only two clades (besides the trivial 
five terminals and complete clade ABCDE), retaining 5 and 10 as significant 
gaps.

Besides the trivial terminals and complete clade (ABCDE), for fair 
comaprison with the above eclectic system, note that for restituting 
complete phylogenetic information, 3 names are required (AB, ABC and ABCD),
with markers (or an ordering system) indicating their hierarchical nesting.
To be complete, naming the paraphyletic group of interest C+D requires 
several words, but no extra name:
C+D is "non-AB  ABCDs".

>  As you can see, the number of taxa limits is the same or lower in the 
> eclectic solution (limits B|C, D|E) - the difference is only in the level 
> at which they are used. Hence, the "amount of arbitrariness" is the same 
> in cladistic solutions as in the eclectic ones - at best.

I don't see at all that the number of limits is lower. It is 5 and 10 
changes judged "significant gap for naming" in both cases (and besides this 
cladistics clearly require fewer names in this case).

Anyway my point on this topic was not at all the criterion for "significant 
gaps" in itself (same problem in the two cases as you notice quite well, no 
more "significant" or "natural", no less). I meant that additional 
arbitrariness is introduced by the requirement of deciding "when to look at 
phylogeny backwards", and not always "upwards" :


The cladist has a unique criterion: clades. He(she) may add the criterion 
"significant gaps" if he refuses to name all the clades. Here he decides to 
consider "more than 4 changes" as significant (qualitative criterion 
possible, see above).
-------End of decisions.--------
Consequently, he names clades A+B (supported by 5 changes), and A+B+C+D 
(supported by 10 changes).


Now, the eclecticist has two incompatible criteria: clades, and possibly 
grades. He may add "significant gaps" if he refuses to name all the clades 
or grades. Here he decides to consider "more than 4 changes" as 
significant, just like the cladist (and qualitative criteria seem to be 
also possible).
******But this is NOT the end of decisions.******
He has now to decide when he will look at the tree "upwards" versus 
"downwards".
Here he decides to look at the tree "upwards" at the 10-changes gap, but 
downwards at the 5-changes gap
(strange enough, but this is your example).
---------End of decisions.---------
Consequently, he names clades A+B (supported by 5 changes), A+B+C+D 
(supported by 10 changes), and grade C+D (supported by 5 changes at the 
"upper bound"? Or supported by its insertion between the 10- an 5-changes 
gaps??).

It simply seems obvious to me that supplementary decisions are involved in 
the eclectic approach, and that they are logically necessitated by the 
initial choice of implementing two incompatible criteria at the same time 
for the unique task of naming.
The supplementary decision concerns when to stop applying one criterion in 
favor of the other, which is clearly not a concern for the cladistic system.

If we add the possibility that no universal priority rule is presently 
available for deciding between the two eclectic criteria, this makes a 
pretty good lot of supplementary room for arbitrariness in applying naming 
conventions.

.....Isn't it?....

Now the good question seems to be: does it matter at all and why? What is 
gained, what is lost, and are there more coherent alternative solutions for 
conveying complex information (markers...)? And the answer should be linked 
to: what naming conventions do we need for smart communication?
Reaching common agreement between scientists (hence at least more 
stability) seems an issue, and minimizing abitrariness may matter in this 
respect.

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