[Taxacom] Rational holophyly and ideological extinction

[Curtis Clark]

On 2009-03-17 06:59, Richard Zander wrote:
> It is uncommon for sister-group relationships of species to be well
> supported in detail (general clustering of species is well supported) as
> an advance or new knowledge over what is obvious from omnispection,
> although gene trees may be quite detailed. Molecular trees demonstrate
> only genetic continuity, not necessarily speciation events unless one
> invokes uniform gradualist speciation and the biological species
> concept, in which case anagenetic change in both sister groups then
> generate different species. Second, molecular paraphyly of species is
> not extremely rare since many cladograms in the literature with lots of
> exemplars of the same species demonstrate ancestor-descendant
> relationships. This goes for higher taxa as well, of course, but
> knowledge is presently somewhat limited by the general lack of multiple
> exemplars of the same taxon in any one study.

This is all a limitation of using cladistic techniques at the species 
and population level. Molecular information is useful at this level, but 
using it to build trees is about as useful as making trees out of human 
genealogies. In their best use, cladograms are trees "of species" (IMO), 
no matter what some molecular cladists say. Speciation is best 
understood using the biosystematic approaches you mention below, along 
with standard population genetics. And to say (not that you necessarily 
are, Richard, but others have) that its inability to deal with species 
renders cladistics of limited value makes about as much sense as saying 
that its inability to deal with quarks and gluons renders quantum 
electrodynamics of limited value.

> Reply: One of the points I commonly make on Taxacom is that given the
> data we have, including that of DNA, we cannot resolve many sister-group
> relationships reliably and it is even more difficult to resolve
> ancestor-descendant relationships though many are obvious as
> paraphyletic-autophyletic pairs.

If by this you mean pairs of species, I think this is a misuse of 
cladistics.

> Reply: This seems a quibble. Of course fossils may represent extinct
> lines.

Any organism can be an exemplar of anything you want it to be. Its 
usefulness as an exemplar remains in any given case to be demonstrated.

> One can say that dinosaurs are the ancestors of birds

It has always amazed me that the word "ancestor" can be stretched to 
encompass an entire group of organisms, some percentage of which left no 
progeny at all, and most of which left no progeny alive today. It is 
even more amazing when such a statement comes from someone (again, I 
don't mean you, Richard, but I have heard it from others) who doesn't 
believe that higher taxa are real.

> I think
> that it is okay to say that a paraphyletic species (e.g. brown bear) is
> the ancestor of its autophyletic descendant (polar bear) given what we
> know even though any given brown bear fossil may not be the ancestor of
> anything alive today.

Again, that's a misuse of cladistics.

> Reply: ". . . an artifact of the way a taxonomist chooses to draw up the
> taxa" seems a criticism dependant on cladists' use of holophyly to
> delimit taxa under the assumption that it is more scientific. 

Not at all. My point is that a formulation of taxa that is based on a 
theoretical framework of holophyly cannot admit in theory to one higher 
taxon being the ancestor of another. Ken might even use this as a 
critique of antidismonophyletarianism. If one *wants* 
ancestor-descendant relationships among higher taxa, paraphyly is, 
AFAICT, the *only* way to achieve it.

> I'm afraid I don't understand the irreducible paraphyly bit,
> and how this might support ignoring ancestor-descendant relationships in
> classification. Please explain.

Brown bears are irreducibly parphyletic with respect to polar bears; 
although, given a large sample of all the brown and polar bears that 
ever existed, one could draw up gene trees in an attempt to break up 
Ursus arctos into monophyletic units, this is doomed to fail, because 
relationships within species are tokogenetic, not phylogenetic. Taxa 
that are groups of species are never irreducibly paraphyletic except to 
the extent that they are rooted in species that are irreducibly 
paraphyletic, and even then they can be (and often are, because of 
limitations in the data) depicted as an "unresolved" polytomy.

> There are many techniques
> for pinpointing ancestors, mostly involving that old field
> biosystematics, now fallen into shadow. Chromosome numbers, crossing
> experiments, and biogeography come immediately to mind: of two closely
> related species, the tetraploid may be hypothesized as derived from the
> diploid, particularly if it is local and the diploid is widespread, and
> particularly if one can generate an identical tetraploid from the
> diploid experimentally with colchicine or whatnot.

I was raised in that tradition, and I don't find it incompatible with 
cladistics, although it does seem to be incompatible with many cladists. 
I believe Bruce Baldwin (with whom I share a major professor) wrote an 
article about the cladistic viewpoint implicit in much of 
biosystematics. I seem to remember that for many of the tarweed genera, 
molecular phylogenies closely matched chromosome number. The mistake was 
in using molecular data not to corroborate other data, but to replace it.

> In addition, in my opinion, any paraphyly or even polyphyly of
> traditional groups on a molecular tree is evidence of a shared ancestor
> (at some taxonomic level) deep in the tree.

Unless I'm totally misreading you, well, duh! There's always a shared 
ancestor somewhere; that's what synapomorphy tells anyone who is not a 
pattern cladist. But knowing that there is an ancestor is not the same 
as pinpointing that ancestor to any given organism, morphotype of known 
organisms, or taxon, "natural" or arbitrary.

>  They are not "devilishly
> hard to pinpoint", but are obvious as the "massive homoplasy" that
> cladists vigorously suppress by insisting on strict holophyly and
> backing it up by making new combinations and new names that change
> nomenclature such that only sister-group relationships are recognized.

Please explain. I'm not following you here.

> This is ideological extinction of a major class of evolutionary
> information in classification. The only way to corroborate splitting
> molecularly paraphyletic or polyphyletic lineages (which may be
> evolutionarily monophyletic because they share a deep ancestor) is to
> demonstrate that the products of the split (separate lineages) are
> significantly better circumscribed by new circumscriptions by expressed
> traits than they were previously. This is because splitting in any way
> of a large taxon into smaller taxa will commonly result in some kind of
> description for each of the smaller taxa, which is a problem in
> statistics called multiple comparisons). That new description must be
> better than any previous description for a larger (or smaller) grouping
> at that rank to be corroborative.

If I'm following this, you are arguing (as you have in the past) that 
cladistic techniques don't necessarily tell us what we think they are 
telling us. I don't necessarily disagree with that. But that is 
different from saying that there really is no "evolutionary tree" in any 
sense that we might understand it.

And, following along this path in hoping that I'm understanding you, I 
don't think that shoot-from-the-hip reclassifications by some (or even 
many) cladists by itself discredits cladistics as a methodology. 
Excesses are nothing new; two names that spring to mind are Merriam and 
Greene.

> Reply: I stipulate and asseverate that phylogenetic analysis resulting
> in clarified sister-group relationships is a good thing, an advance, a
> revelation of exciting details (as hypotheses) about evolution (using a
> non-ultrametric clustering method based on a simple model of evolution
> called parsimony), and a good use of time.

Cool! Agreed.

> I also think that wrenching
> all ancestor-descendant information from classifications as a blood
> sacrifice in honor of this advance, and relegating the study of
> ancestor-descendant relationships to the same intellectual level as
> voodoo and metaphysics (pace voodoo doctors and metaphysicians among us)
> is a kind of latter-day Lysenkoism.

Were that the only rationale, I would agree. But the evidence says to me 
that, above the level of species, ancestor-descendant relationships are 
(1) unnecessary, in holophyletic taxa, (2) more trouble than they are 
worth (biosystematics is rather unproductive at that level--I remember a 
presentation in the 1970s basing subgenera and genera in a plant group 
on percent pollen fertility of the hybrids), and especially (3) given to 
unproductive statements (dinosaurs are ancestors of birds) that leave 
nonspecialists thinking they understand things better than they do.

And let's say that acquired characteristics *can* be inherited. Would 
that make Lysenko any better a scientist?

> Mine is not an extreme position,

Nor is mine.

> since (as I repeat from elsewhere) classifications and taxonomic
> treatments following strict phylogenetic monophyly have damaged
> evolution more strongly than creationism or intelligent design will ever
> do.

I have to strongly disagree. Bad and hasty classification damages the 
study of evolution (I assume you don't mean damage to evolution itself), 
but the assertion that time-honored holophyletic groups such as mammals 
and flowering plants are quite enough, and that any additional 
holophyletic groups are damaging, just doesn't make sense to me (I might 
even say it is extreme).


-- 
Curtis Clark                  http://www.csupomona.edu/~jcclark/
Director, I&IT Web Development                   +1 909 979 6371
University Web Coordinator, Cal Poly Pomona