Farewell to Species - reticulation

Pierre Deleporte Pierre.Deleporte at UNIV-RENNES1.FR
Tue Feb 1 19:37:20 CST 2000

Comments tu Curtis Clark and Thomas G. Lammers.

Curtis Clark wrote: 

>At 04:27 PM 00.02.01 +0100, Pierre Deleporte wrote:
>>For instance, is homoplasy due to
>>hybridization differentiable from homoplasy due to other processes?
>This is a common misconception. If a species or lineage of hybrid origin
>has inherited features from both parents, those features are *homology*
>("similarity due to common descent"), not homoplasy, no matter how they
>disturb a branching cladogram. Any method other than concordance of
>evidence that separates homology and homoplasy (and yes, I know that some
>argue that there are no such methods) will separate homoplasy from the
>homologies that hybrid lineages share with their parents.

I understand your point, I could have named more precisely these
"homologies (i.e. inherited traits) scatterred on fully resolved
cladograms". I was just restricting "homology" to "homology by descent in
direct lineage". I am not the only one.

The problem arises simply from the fact that contiguous similarities are
easily accepted as homologies (the contiguity criterion for "secondary" or
"confirmed" homology), when scattered similarities are not so easily, and
surely not systematically interpreted as "homologies by hybridization".
In the mind of most phylogeneticians, a distinction is spontaneously made
between "cladistically confirmed homologies" and "possible but scatterred
(misleading) homologies". The construction of reticulograms would clarify
the naming of "homologies by hybridization"... simply we have to construct
them first, which was the topic of my message.

We could fix terms to descript topographic distributions on cladograms or
reticulagrams (e.g. contiguity vs scattering), and terms to qualify the
inferred biological processes at work (descent in direct lineage, descent
by hybridization, parallelism, reversal, horizontal transfer,
convergences... the four first ones at least being some way inherited
traits, and thus "biological homologies", sensu Curtis Clark?). Anyway some
distinction between "homology by descent in direct lineage" and e.g.
"homology by hybridization" seems useful in the phylogenetic debate.

Thomas G. Lammers wrote:

>And it is exactly the sort of misconception that arises when we worry so
>much about making pretty patterns of stick-figures that we lose sight of
>the real live organisms and biological processes that our patterns are
>meant to represent.
>Submitted for mulling over: cladistics is flawed science because it comes
>to a conclusion first ("all evolution is dichotomously divergent, with
>ancestors becoming extinct at the appearance of a pair of daughter-taxa")
>and then seeks data to support that conclusion.  When the data do not
>support that conclusion but suggest that other patterns are operative and
>equally valid (e.g., when we get polytomies or reticulations), they are
>judged to be faulty and dismissed.

>Thomas G. Lammers, Ph.D.

Thomas, I think that many so-called "cladists" DO make these distinctions,
DON'T state that all evolution is dichotomously divergent, and are really
concerned with the possibility of identifying non-dichotomous evolution.
The problem is just: how should we manage to do this, in general or in
particular lineages? And isn't a cladogram a good starting point anyway?
I simply do not know of anyone starting on readily other bases for
phylogeny reconstruction (even "clocks" is not at odds with descent, it
just adds a questionable assumption of regularity of the process). Who
would reject the dichotomous reconstruction if it explains most of the data?
A lack of dichotomous resolution effectively raises problems of
interpretation, but I hardly understand your charge against "cladistics" in
general on that basis... Nobody is obliged to be confident in poorly
supported clades! Non-resolution is an interesting result in cladistics
anyway, however frustrating.

I must disagree when you state that cladistics "is flawed science because
it comes to a conclusion first" ("that all evolution is dichot." etc): in
fact, you are just stating the evolutionary assumptions used in standard
cladistics as a justification of this method for phylogeny reconstruction.
Do you know of methods with alternative or better assumptions, or do you
stand in the illusion that no assumptions are required for reconstructing
the past? Assumptions are not a flaw, they are just a logical necessity for
induction, and of course they require as strong external justification as
possible. Let's evaluate assumptions of alternative methods and their
justifications in the general or in particular cases, but we surely cannot
dispense with assumptions, and by definition... assumptions stand "first",
in cladistics like in any other competing phylogenetic approach.


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

More information about the Taxacom mailing list