[Taxacom] PhyloCode & ICZN to "duke it out"?

[Richard Pyle]

I've been fighting the urge to comment on this thread, but the urge
ultimately won out.

Just a few comments on recent posts, followed by some LONG excerpts from a
recent and relevant exchange with colleagues.

Emily wrote:

> Consider a 
> situation in Linnaean taxonomy, where one family is 
> determined to be derived from within another.  In this case, 
> one name or the other has to be changed because there's no 
> convention to accomodate a rank within the same rank.  

There is nothing about the Linnaean nomenclatural system that prevents this,
and indeed there are situations when it is done consiously and with specific
intent.  I can almost hear the collective "gasp" from cladistically-minded
systematists at this notion -- but having just engaged in this exact same
debate with some friends/colleagues a few weeks ago, my position is fresh in
mind (see excerpts below).

Mark wrote:

> Our taxonomy rests on the bedrock of Linnaean and Darwinian 
> concepts; namely common descent.

Well...our way of interpreting living things today rests on the bedrock of
the *Darwinian* concept of common descent.  Linnaeus, of course, had no such
concept when he invented the nomenclatural system we still use today.  It
survived for a century without any concept of evolution, and another century
without tight integration to evolutionary concepts.  It's only been the last
half-century that some biologists (a minority, really) have tried to
re-define its purpose and function from what it had been the previous two
centuries.

Ken wrote:

>       I'll address Emily's post first and then Mark's.  First 
> of all, I would point out that the elimination of Linnaean 
> ranks only appears to promote stability in a deceptively 
> simplistic way.  The problem is that it encourages the 
> explosive growth of such ranks, even though such ranks are 
> not *explicitly* recognized.  McKenna began running into this 
> problem in the 1970's and 80's with Mirorders, Parvorders, 
> Supraorders, ad nauseum.

On this point, I agree with Emily, Mark, and the Phylocoders.  They're not
ranks, and so they do not suffer the problems you attribute to McKenna.  I
think part of the reason so many people seem to dislike the Phylocode
approach is that they (inappropriately) ascribe properties of the Linnaean
system to it.  Don't think of it as hierarchical, or even as a replacement
to the Linnaean system.  Think of it as a less ambiguous way to represent
cladograms in the form of text (names), rather than line-drawings.  Whereas
the Linnaean system was unambiguously NOT designed to represent phylogenies,
the Phylocode system was explicitly designed for that exact purpose.

But I generally agree with Ken on his other points.

Paul wrote:

> It is true that under the ICBN, a family cannot be contained 
> in another family (and a genus not in another genus), but 
> there are no requirements on what constitutes a family.

I include this because it forced me to realize that I should clarify my
response to Emily above.

Nomenclaturally, Paul (and Emily?) are correct that a family name cannot be
included hierarchically within another family name.  However, my response to
Emility was that the Codes do not require that names be assigned to
monophyletic clades.  Indeed, they are not assigned to clades at all -- they
are assigned to type genera/species/specimens.  Paraphyly is allowed in
Linnaean nomenclature. Indeed, so is polyphyly -- although I don't think
that any modern practicing taxonomist would intentionally establish names
that were polyphyletic.

What follows below are some (slightly) edited excerpts from a recent
exchange with colleagues.  Caution: continue reading at your own risk...

Aloha,
Rich


==================================================================

Dear XXXX,

I don't think we're very far apart on this.   However, where we differ, I
think, is that while I agree that evolutionary relationships are easily the
most important metric for classification (bats are mammals, not birds),
they're not the only metric. The only reason we put names on things in a
hierarchical fashion is to be able to effectively communicate with each
other. Linnaeus preceeded Darwin by a century, so a hierarchical system of
names obviously has some communicative value over and above evolution.  The
question, then, is what are we trying to communicate?  In most (almost all)
cases, the different kinds of information we hope to communicate
(evolutionary relationships, gross morphology, ecological parameters, etc.)
are congruent, and so the nomenclatural classification is straigtforward.
The trouble comes when some of these parameters are not congruent.  In most
such cases, the "true" phylogenetic affinities are concealed by overt
morphology (often driven at least in part by ecology), and only revealed
through techniques such as DNA analysis (which is why this debate often
falls in line with morphology vs. molecules -- but the debate is really not
defined by this distinction).

So...in such cases where morphology and/or physiology/ecology/etc. conceals
phylogenetic affinites (Gomphosus, Aves, Rhincodon, Scaridae, etc.), and
where there is a well-defined apomorphically-robust subgroup within a
larger, generally plesiomorphically consistent clade, we have conflicting
rationales for deciding how to represent nature through nomenclature.  The
main categories to consider are:

1) Inferred Evolutionary relationships;
2) Gross morphological and/or ecological cohesiveness;
3) Stability of nomenclature

In the earlier days of cladistic techniques, there was a lot of "noise"
associated with #1, in that someone would present a cladistic classification
one year, then come back the next year and say "forget everything I told you
last year, this is the real story...".  A lot of this was to be expected
(every new approach to things has growing pains). But I'm not so sure it has
been completely eliminated.  That is, I think that modern approaches to
analyzing both morphological and molecular characters can give a false sense
of confidence sometimes.  I discovered this for myself when I first started
playing with cladograms, and seeing them with their clean bifurcating trees
arranged just so and naïvely thinking "Ah hah! So this is how they
evolved!".  Certainly people are much wiser now than they used to be about
understanding how to interpret the results -- but even still there are
conclusions being asserted that seem completely off base.  Another potential
issue with #1 is that, as we have discussed, and as I think is becoming more
recognized (especially in plants and corals), is that evolution doesn't
always follow nice bifurcating patterns.  Introgression may not be as rare
as we generally assume.  This is certainly less of an issue for higher ranks
of classification than it is down in the realm of species, but it is still a
potentially confounding issue.  So, in summary for #1, we can't always be
certain that we really do understand the evolutionary relationships.

Even if we do feel confident that we know the "true" evolutionary history
(Scaridae, Aves, Gomphosus, etc.), I'm not sure it's necessarily true that
evolutionary history should always trump everything else (i.e., #2).  I see
your point about how classifying anomolous species (Gomphosus) or cohesive
groups (Scaridae) stictly in terms of item #1 above can prompt people to ask
questions about evolution -- and I think that's a good thing.  But at what
cost?  In some cases, the cost is confusion, and disconnection with
historical information.  People have other reasons for clustering groups
nomenclaturally besides their evolutionary history.  Flip your example
around:  rather than starting with a person observing "Thalassoma varius"
and wondering why it has the same genus name as these other short-snout
wrasses, instead start with a non-ichthyologist researcher (e.g.,
biogeographer or ecologist) looking at a cladogram of fishes in the context
of some other pattern in nature.  Seeing "Gomphosus varius" nested within a
clade containing mostly Thalassoma might prompt them to wonder why the genus
is different for that one species -- thus leading them to realize the
morphological/ecological divergence.  So, it depends on which direction you
come at it.  If you see the critter first, then the phylogenetic affinity is
hidden ("lost") if the names don't reflect it.  If you see the name first,
the the ecological/morphological uniqueness/cohesion is hidden/lost.  It
works both ways.  As XXXXXX has argued before, recognizing an anomolously
distinct group (e.g., Paranthias, in addition to the other examples already
mentioned) draws attention to the anomaly -- thereby serving a communicative
function.

As for #3, the more the language changes, the harder it is to to connect
information in the present with information in the past.

So I guess the overall point is that #1 should be the main driving
consideration in constructing nomenclature, but it's not the *only*
consideration.  Given the occassional weakness of #1, and the occassional
strength of #s 2 & 3, I believe there are cases where trying to reflect
evolutionary affinities perfectly through Linnaean nomenclature detracts
from the communicative value of the names.

This leads me to a final point, which is that the Linnaean nomenclatural
system sucks as a tool for communicating inferred evolutionary patterns.  As
I already mentioned, it was around 100 years before Darwin came along, and
it's really only been in the last few decades (i.e., post-Hennig) that
taxonomists have gotten more methodical about using Linnaean nomenclature
primarily as a tool to communication evolutionary relationships.  The
Linnaean system is superficially attractive as a tool for communicating
phylogeny because of its hierarchical structure, and that seems conducive to
describing a pattern of branching trees.  But there are two fundamental
flaws with using the Linnaean system to communicate evolution:  1) Ranks;
and 2) Typification.

The "rank" problem is pretty obvious, and doesn't need a lot of elaboration.
I think we can all agree that where to draw the lines for ranks is arbitrary
(and I firmly believe that this is true for species every bit as much as for
higher and lower ranks).  And the Linnaean system is all about ranks.

The typification problem is not discussed as often, but I think it's
actually the larger of the two issues.  All Linnaean names governed by Codes
of Nomenclature are anchored to single-point types
(specimens/species/genera, but in all cases ultimately s single specimen).
As such, they do not describe sets of organisms (e.g., Clades), but instead
represent anchorpoints in nature around which taxonomists will subjectively
circumscribe larger sets of organisms.  There will, therefore, always be
arguments about how broadly or narrowly those circumscriptions should be
drawn (i.e., lumpers vs. splitters).  Nobody seems to like the Phylocode,
but they got one thing right: they abandoned the single-point type system in
favor of a multi-point (really should have been two-point) system of
defining the names.  In other words, the names are defined in terms of
clades, not in terms of arbitrary evoltionary distance from a name-bearing
type (the other thing Phylocode did right was to abandon ranks).  So, if the
definition of a Phylocode name is "all decendants of the most recent common
ancestor of A and B", we now have an objective, testable circumscription of
a set of organisms, that future reasearch will only further clarify.
Contrast this to the never-ending problem of lumpers vs. splitters, which
does not apply to Phylocode names.

I'm not advocating Phylocode per se -- just making the point that the
Linnaean system was not designed for, nor (for most of its existence)
applied as, a tool to communicate evolutionary relationships.  It has served
its purpose extraordinarily well fover the past quarter-millenium, and it
seems unwise to me to try to re-define its function now -- especially when
it's a poor system for communicating evoluton in the first place.  If we
really want to reflect phylogeny through nomenclature in all cases, then we
really should design a better nomenclatural system.

Anyway -- I am certainly happy to agree to disagree on this.  But it is
intellectually invigorating to discuss it!

Aloha (from the Atlanta airport, where I'm on Hour 2 of a 9-hour
layover....)

Rich



*************
Another (slightly edited) message, in the same conversation, responding do a
different friend/colleague:

> My point is simply this: We are at a turning point in systematics 
> where "new" types of data and analytical tools are revolutionizing the 
> field.
> As such, hypotheses will be challenged, tested, and changed or left 
> alone accordingly.  If the process of naming organsisms, per se, is 
> not seen as a hypothetico-deductionist practice we have nothing to 
> fear; it cannot be tested since there is no scientific method without 
> a hypothesis. Thus, organisms may bear whatever name is determined to 
> be the most preferable to the "experts".  However, if we allow 
> taxonomic practices to take on their full potential and reflect 
> hypothesized relationships, they become subject to hypothesis testing 
> and should change as will our understanding of those relationships as 
> new techniques and tools arise.

Agreed!  Which is why it would be utterly foolish to ground our primary
mechanism for communicating these hypotheses about evolutionary
relationships on the Linnaean system of nomenclature.  For crying out loud,
we can sequence Craig Venter's entire genome, but we can't come up with a
palatable system of nomenclature for naming clades that is better than a
system designed 250 years ago by a creationist?  The phylocode guys had the
right basic idea -- but I guess they either implemented it inappropriately,
or pissed too many people off (or both).

[...]

> Are you really arguing that "traditional" taxonomic practices keep a 
> broader perspective on evolutionary relationships,

More than keep a broader perspective -- it is unambiguously the single most
important factor that every taxonomist worth his/her salt uses as the
guiding principle behind creating Linnaean nomenclature.  This is certainly
true for me, XXXX, and every other "traditional" taxonomist I know.  This is
why we are all in full agreement in the VAST majority of cases.  The only
time we come to different conclusions is when we encounter a case where
certain groups (birds, Gomphosus, Paranthias, etc.)  fall into some sort of
ecological/physiological/morphological pathway that leads them way out into
left field in such a way that, even though we acknowledge it is a cohesive
and apomorphically anomolous subclade within a broader, plesiomorphically
homogenous clade, there is justifcation for recognizing its
cohesive/anomolaus state of being through judicious application of ranked
Linnaean nomenclature.  Why? Because this is how it has been done since the
creationist Carl Linnaeus invented the system 250 years ago.

Your agrument should not be against people who use Linnaean nomenclature in
a historically consistent way. Rather, your argument should be (as mine is)
that Linnaean Nomenclature is a piss-poor mechanism for communicating
evolutionary patterns in nature.  It should never have been co-opted for
that exclusive purpose in the first place.  It certainly is convenient that
"traditional" applications of Linnaean nomenclature are most often congruent
with modern phylogenetic analyses, and that exceptions are relatively
uncommon.  And, it's certainly fair to say that nomenclature should be
changed to reflect modern interpretations of evolutionary history, when
doing so carries more overall communicative benefit (or even if it only
exacts a relatively small cost to human communication).  But it is not
correct to say that ranked Linnaean nomenclature must, in all cases,
strictly reflect (inferred) hierarchical patterns of monophyly, to the
exclusion of all other reasons for which we put Linnaean names on organisms
in the first place.  Sometimes, the sum of other reasons should (correctly)
trump the noble (but ultimately impossible) desire for nomenclature to
always adhere to strict monophyly (i.e., never allowing paraphyletic names
in any circumstances).

> or
> simply that that evolutionary relationships have no place in taxonomy?

Good heavens, no!  Evolutionary relationships always should be (and really
mostly have always been -- whether early taxonomists realized it at the time
or not), the number-one factor to consider when coming up with a sensible
nomenclatural classification.  It's just that it's not the *only* one.  

> the ultimate standoff in this argument is whether or not you think 
> that monophyly is important in designating natural groups, and if 
> natural groups should have a nomenclatural designation.

That is not correct.  I DO think that monophyly is important in designating
natural groups.  And I DO think that such groups should have a nomenclatural
designation.  Just not one based on a system designed 250 years ago by a
creationist.  I *wish* the phylogenetic community could either get behind
(and fix) Phylocode, or come up with a better system -- because as I said
before, the Linnaean system is simply not well designed for this purpose.
It is extremely useful for the purpose it has served for the past 250 years.
Just not ideal for this new purpose it has been co-opted for.

> If we can't agree on shared
> ancestry as an important criterion to this end, we will continue this 
> battle until we have expired.

Well, then maybe we don't have to have this battle anymore -- because we all
agree on shared ancestry as an important criterion -- not only for
understanding the biological world in general, but in fact for representing
THE most important criterion in assigning Linnean nomenclature.  The only
battle comes when people forget that it's not the absolute ONLY criterion
for assigning Linnaean nomenclature.

> Is this really just because you don't want to be considered 
> Osteichthy?  :)

Ultimately, we're all Archaea. :-)

Aloha,
Rich

*************
Later, in the same conversation:

> Above the species level,
> there are no real entities, hence to try to extend this argument 
> (familial or generic level reticulated evolution) is mute, and in my 
> mind, a little absurd.

Almost as absurd as the notion that "species" are any more "real" than ranks
above, or below. :-)

They're all equally arbitrary -- at least broadly speaking.  Sure, I could
certainly invent some objective criterion for "species" (e.g., 1.5%
substitution, or whatever), and then objectively draw lines between species.
But I think you know as well as I do (probably better) that what would
result is a system of nomenclature that is much less useful for
communication than what we have now.  All you're doing is moving the point
that is arbitrary from one place to another -- but it's still no less
arbitrary.

> I think the problem
> you have identified, and one that keeps me awake at night, is that we 
> have yet to come up with truly functional species concept.  If you 
> subscribe to the biological species concept, many named species 
> dissapear (including, i'm afraid, many of those pretty Centropyge).  
> If you subscribe to a concept that relies on diagnosibility, we would 
> h ave like have a patronym for everyone on the planet!  I think that 
> the fact that we can identify these reticulated patterns argues that 
> we have a much better understanding of the evolutionary relationships 
> than ever before.

We CERTAINLY have a better understanding of the evolutionary relationships
now than before (at least we think so -- and I would believe it).  And we'll
have an even better understanding 100 years from now.  But we're talking
about two different things:

Evolution is a continuous, unbroken chain of reproductive events all the way
back to the first self-replicating proto-life form.  Billions (trillions?)
of generations, and not a single one interrupted in terms of living things
around today.  Where, along that glass-smooth continuous process are we to
draw "species" lines?  We could point to vicariant events and say "Here!
This is where the species event occurred!"  But to do that, you have to say
the F1 generation was a different species from its parents.  If my daughter
goes to the moon to help start a new colony of humans, that eventually after
thousands of years of non-interbreeding diverges from the Earth-bound humans
such that we would regard them as a separate species (lots of mutagens when
there is no atmosphere), does that mean that Cara and I are different
species, because she and I (and our respective other descendants) were
separated by the vicariant event of her going to the moon?  Or, does the new
lunar species become a full species only later, after the divergence.  But
then which one of Cara's descendants was the last Homo sapiens on the moon?
Was it the first one who was biologically unable to reproduce with any
earth-bound human?  Or was it the last one who was able to?

All of biodiversity is the product of an unbroken chain of reproductive
events, and the reason we have failed to identify a "truly functional"
species concept is because in this unbroken chain of reproductive events,
there have been very few -- if any (and certainly no universally applicable)
inflection points to indicate when one species ends, and the next begins
(this was the basis of my tongue-in-cheek comment in the last email that we
are all ultimately Archaea).  Actually there is one truly functional species
concept -- it is the one that has been in de-facto use for 250 years, and
the one that is still in use for the vast majority (arguably 100%) of cases
today.  Like it or not, the real (and practical) definition of a species
always has been (and I suspect will continue to be for a long time): a
species is what a taxonomist or community of taxonomists says it is.  Not
much predictive value.  Spotty (if any) objectivity.  Not really science.
But damn -- there aren't many practices that have remained more or less
unchanged for two and a half centuries.  There must be *something* of
practical value in there!

> Two last points.  First, it certainly sounds like you enjoy the 
> phylocode for its choice to define "sets" instead of "points" (and 
> thanks for that explanation, it made more sense to me than any other 
> explanation I've yet received about the phylocode!).  However, it 
> seems odd to me that you like a system developed by ardent cladists 
> and that places all of its emphasis on monophyly!

See -- that's where I think you misunderstand me (and XXXXX).  Personally, I
have TREMENDOUS respect for the evolutionary insights that cladistics offers
(see my earlier note to XXXXXX on this), and I also understand the value of
monophyly for interpreting and understanding evolution.  It's not my
personal game -- but not because I think it's useless, or bad science (it's
much better science in some senses than much of the rest of biology).
People like XXXXX and I really only have one beef:  the destabilization of
Linnaean nomenclature for reasons that do not ultimately serve the
over-riding optimality of communication.

And, to be honest, I'm really indifferent to Phylocode.  I would never, ever
use it myself -- not because I hate it or think it's bad -- but as I said
before, it's just not my game. But I do understand where they were coming
from on the "define a clade" approach (contrasted with the single-point
typification of the Linnaean system). 

> Hope your trip is going well...this is just too much fun!

Yes it is, and I absolutely agree!  It would be a bit funner, I think, over
beers and in person -- which I hope we can do soon after I get back!

Gotta get back to the other 70 emails that have come in since I last logged
on...

Aloha,
Rich

*******
And one more -- again from the same conversation, but this time addressed to
someone on my side of the argument:

> I am hoping that the XXXXXidae will survive as a family in spite of the 
> cladistic onslought.

XXXXXXX: don't blame cladistics -- it's a very useful tool for helping to
shed insight on inferred evolutionary relationships.  I think your "beef"
(as is mine), is with the notion that paraphyly (i.e., nomenclaturally
recognizing a well-defined subgroup from a broader and otherwise
monophyletic clade) is somehow illegal for Linnaean nomenclature.  This is
much more a question of personal philosophy and linguistics than it is a
question of science, in my opinion -- but too often people confuse the
science of cladistic analysis as a method to infer phylogentic
relationships, with the more traditional linguistic "art" of Linnaean
nomenclature.  The former is mostly objective, and the latter is largely
subjective.  Cladograms should be generated, tested, and changed
continuously in an effort to come up with a more and more reliable inference
about historical evolutionary patterns.  Linnaean Nomenclature, on the other
hand, is a communication tool used by VASTLY more people than those who are
interested in studying and inferring phylogenies -- and, I think, needs to
be respected more as such.  Life is wonderful for all concerned parties when
the two happen to be congruent (as is the case most of the time). It's just
these oddball cases (Scaridae, Gomphosus, Rhincodon, Aves, etc.) which,
unfortunately, happen to involve groups that people outside of
nomenclatural/phylogenetic science happen to refer to a lot.

Not a comment directed to the CC list -- just feeling like bantering a
little.... :-)

Aloha,
Rich