[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: On the Other Phylogenetic Systematics, Nixon and Carpenter (fwd)

        Apologies ot the List for cross-posting, but I just finished this
little ditty to the PhyloCode mailing list, and since that list is almost
entirely non-overlapping with the dinosaur list, and since some advocacy
of the discussed papers was recently posted on the list, I felt I ought to
send it here as well.
        Note that it is *not* my point to start wholesale discussion of
this topic; this is, after all, the dinosaur list. I also do not have time
to engage in a protracted discussion. Personally I feel that Benton, Nixon
and Carpenter state their cases very well. If you object to my points,
please read their papers, and, if you feel the viewpoint of the authors
does not sufficiently cover your objections to my statements, feel free to
comment. I will endeavor to make time to respond.

The papers referred to below are:

Benton, M. J. 2000. Stems, nodes, crown clades, and rank-free lists: is 
Linnaeus dead? _Biological Reviews_ 75: 633-648.

Nixon, K.C. and Carpenter, J.M. 2000. On the other 'phylogenetic 
systematics.' Cladistics 16: 298-318.

        Note that, for the record, Carl von Linne (Carolus Linneaus) has
been dead for several centuries.

---------- Forwarded message ----------

On Wed, 14 Feb 2001, Janovec, John wrote:
[prologue and some introductory text deleted]

[JJ commenting on Nixon and Carpenter 2000]
> Does it not bring up some very good points to consider regarding
> the proposed Phylocode? 
        It makes the argument recently summarized by Dr. Cantino, although
in a somewhat reactionary way (a common theme of recent "critiques" of
phylogenetic nomenclature). Some workers would prefer a system in which
species are grouped _a priori_ of knowledge of their phylogeny. A name is
tied explicitly to these groups, and their exclusiveness (relative to
other known species) can be tested by analysis. However, instead of
discarding the name if the group if it does not pass the test (which might
make more sense), the group is reconstituted to exclude some members.
Presumably, if too much change is called for, the group is then disbanded.
There are not objective guides for how much change is too much, or which
ones must be kept in the group.
        This approach, considered the "traditional" one, has some
advantages, in that associations of particular species with particular 
taxa are conserved (of course, this is also possible in the phylogenetic
scheme, through careful definition). However, in a nomenclautral system
that exclusively recognizes monphyletic groups, it can be a little
disturbing. Such traditional schemes do not recognize clades explicitly
(since there is no effort to encompass unidentified descendants of the
common ancestor of a group). Further, they do not recognize, nor do they
associate with the group, the most recent common ancestor of the included
forms. On a more practical side, the most recent common ancestor, and
therefore the clade itself, changes regularly with new hypotheses of
phylogeny, as does the content of the clade (in terms of unknown or
poorly known members). Essentially, yuou are constantly changing the clade
which is named, in the name of a stable assocaition of a name with a set
of species.
        While Nixon and Carpenter do make a valid point, that
composition probably changes more in phylogenetic nomenclature, they fail
to appreciate the value of the fact that a phylogenetic definition always
refers to one and only one clade and common ancestor... ALL that changes
in composition... this is valuable, in that *all* that is testable is
clade membership. We cannot objectively test the membership of an _a
priori_ grouping based on a clade that WILL change with the results of our
analysis (because the limits of the group may be changed to accomodate
new hypotheses of phylogeny). This is comparable to trying to solve for
two variables with one equation; the solution changes constantly and by
the fiat of the systmatist.
        It is important to remember, in such discussion, that a clade is
not a group, it is a monophyletic entity, and clades are indicated by an
anlysis, not fully discovered. The Nixon and Carpenter approach to
taxonomy is firmly grounded in the notion of naming groups based on
essentialist characteristics, and associating a name with a "taxon
concept." This perspective is antithetical to the methods espoused by many
advocates of phylogenetic nomenclature, in which the natural ordering of
life (through phylogenetic relationships) is discovered, and clades are
named with the intention of discovering the known descendants of the
specified common ancestor.
        Nixon and Carpenter (and other authors) are obsessed with the
concept of stability of analytical taxonomic content, a concept as elusive
as the name is cumbersome. They skirt the fact that a phylogenetic name
ALWAYS refers to the same clade, and therefore the same included species.
The crux of phylogenetic nomenclature is that the included species are
HYPOTHESIZED (ala the scientific method, and all that...), and not forced
into a group at the expense of a constant common ancestor. Remember, the
common ancestor is the root of a clade...
        Although they effectivley demonstrate that the content
of phylogenetic "taxa" is unstable, these authors do so through an
extremely loaded example. In all cases, when a name is tied to a group of
very specious groups with poorly resolved interrelationships, the content
will change. However, of the X,000 species they show changing membership,
all of these are retaining their membership at the next most exclusive
level. Thus, the bazillion specie of Angiospermae do not necessarily
change their assocation with this name, even though they may change their
association with the name "paleoherbs," a name which the original authors
chose NOT to name (for reasons Nixon and Carpenter make obvious). Of
course, the other point these authors miss is that, under phylogenetic
nomenclature, these taxa eitherc ARE or ARE NOT members of this clade,
their actual relationships do not change, only our hypothesis does.
        So, in a sense, the traditional methodology advocated by Nixon and
Carpenter is rooted only to groups of species associated _a priori_ of
analysis, with the acknowledgement that these groups may change, or even
be abandoned alltogether. I fail to see any real stability in such a
system. We certainly cannot expect nature to respect the groupings we
force upon it (a point readily acknowledged by most traditionalists).
However, some systematists do not see the need for a foundation for
systematics, for any logical referrent. In the old days, this attitude was
combated by the twin demons phenetics and phylogenetic systematics. Now,
with the victory of phylogenetic systematics in providing a reproducable
basis for phylogenetic analysis, many adovocate complacency and
non-reproducability in our nomenclature. You may decide for yourself if
you would like to see some science in your scientific nomenclature.

        In another recent paper, Benton has argued points similar to Nixon
and Carpenter, but from the more pragmatic perspective of constructing a
classification. While he appears to have missed the explicit point of
phylogenetic nomenclature, that it is NOT a classification, and is meant
to replace the use of classification in biological nomenclature, his
points are somewhat more tangible. PN does not necessarily work well as a
classification, in that it abandons the (artificial) structure, and
therefore the "morphological key" of the Linnean System. However, the
traditional method also fails to be a good classification, in that it is
inextricably interwoven with interpretation.
        The "good Linnean taxonomist" of Benton does not arrange his
groups to best classify his subjects for information storage and retrieval
purposes, if he did so, he would be a pheneticist. Instead, he complicates
his classificationo with consideration of phylogeny. Of course, as Benton
points out, this is not done each time new information presents itself,
the "good Linnean taxonomist" waits for an unspecified time, until the new
phylogeny is better established. Thus, even in his attempts to formulate a
deliberately incomplete phylogenetic reference within the classification,
the traditionalist fails. Although Benton claims to favor monphyletic
taxa, he allows for content to remain constant for long periods at the 
expense of phylogenetic information. What Benton advocates is the least
effective of all possible systems, a poor classification, a poor
phlogenetic reference, and either way a poor systematic nomenclature.
        Practitioners of phylogenetic nomenclature have been characterized
as purists (among other things). However, I fail to see what other
alternative there is. Traditional nomenclature fails to formulate 
explicit, scientific, or even useful classifications. By compromising two
of the possible goals of a nomencaltural scheme, those of
effectively capturing phylogenetic and morphological information,
traditional nomenclature fails to be at all useful or scientific. It is
only useful as an information-storage system for either type of data if
the preferences of the authors, with regard to which sort of information
takes priority at each "level," a criterion which does not seem to be
constant even within a single work by a single author. Of course, any
heirarchy based solely on morphological information is inherently
subjective anyway, a point which eventually lead to the demise of
        Anyway, a traditional taxonomy does serve a third goal well; 
it is a good key for recovering inforation from the literature. However,
even in this it regularly fails, as new concepts and new nomenclatural
schemes rewrite taxonomy, and require every graduate student to develop
his own mental map of arcahic and arcane taxonomy (Predentata =
Orntihischia, Lambeosauridae = Lambeosaurinae, Trachodon = Thespesius =
Claosaurus = Anatosaurus = Edmontosaurus = Anatotitan, etc.). Indeed,
although some practitioners convert traiditional taxa with poorly
thought-out phylognetic defintions (often based on one particular tree, and that
tree alone), I advocate (and will do so in print) a more careful approach
to ensuring access to the literature (although I like crown clades,

        Another point concerning Nixon and Carpenter's paper (albeit a
minor one) which I have touched on earlier, is that ANY system claiming to
name only monophyletic taxa is, by definition, a "node-pointing system."
In the case of traditional taxonomists, this point is lost, because
these workers tend not to consider the undiscovered members of a clade
(such as the common ancestor). However, a clade is an ancestor and all of
its descendants, and therefore a node (do not be confused by recent
papers, such as Padian, Hutchinson and Holtz's excellent terhopod
nomenclature article... node-, stem-, and apomoprhy-based definitions all
identify the same class of entity, a clade, they are just different ways
of specifiying *which* clade is being named). Thus, a way of attaching a
name to a monophyetic group is a way of attatching a name to a clade, is a
way of pointing out a clade, is a way of pointing out a node, is a
"node-pointing system." Nixon and Carpenter seem to be objecting to the
fact that Phylognetic Nomenclature does so explicitly... given this
paragraph, I feel they might benefit from some explicitness in their own

        I hope this helps.

        Jonathan R. Wagner

P.S.    I apoligize if this post seems a bit disorganized, I am using an
outdated e-mail interface (since I am on "company" time... the joys of
graduate funding).