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


chris brochu wrote:

>  In either case, you have the
> >_definition_ of the taxon, and as a separate matter you have the
> >_diagnosis_ of
> >the taxon, or how you tell if organism A belongs to this taxon or not.  The
> >diagnosis has a list of features which are considered autapomorphies for the
> >taxon, and if A is
> This really was no different in the Linnean system.  We can consider a
> taxon to have several properties:  name definition, diagnosis, group
> membership, timing of origin.  The Linnean system effectively tried to
> maximize the stability of the diagnosis, but diagnoses were still labile:
> were we to diagnose Dinosauria on the basis of a perforate acetabulum and
> inturned femoral head, and then discover some new fossil with one but not
> the other, the diagnosis would still require modification.

Would it?  If the diagnosis of Dinosauria is as you state, then an animal must 
both those features to be a dinosaur.   Remember, I'm a programmer.  Boolean 
(A AND B) is true if and only if A is true and B is also true.  (I'm sure you
already know that, but an awful lot of stuff I've said lately has been badly
misconstrued, and I want to avoid any more of that.)  If taxon M is diagnosed 
by 22
characteristics, and organism X has 21 of those characteristics, then X isn't a
member of M.

> If you look at any published compendium of croc fossil diversity, you'll
> find Crocodylus from the Cretaceous, Paleocene, and Eocene.  But molecular
> data - primarily protein distances at first, and now mitochondrial
> sequences - were strongly pointing to a divergence among living species
> within the past 3 to 6 million years.  Were the fossils misidentified?
> Were molecular rates within Crocodylus amazingly slow?  Ten years of debate
> raged.
> The problem, though, is that no two people were using the name "Crocodylus"
> in the same way.  The molecular people were using the *crown-group*
> Crocodylus, but paleontologists were using Crocodylus as some sort of
> form-taxon - if it's not a gator or gharial, it's gotta be Crocodylus.  The
> oldest fossil crown-group Crocodylus are about 5 million years old. This
> debate should never have occurred, and resulted from a lack of stability in
> the meaning of a name.

I agree, this is a problem, and any system of taxonomy should have a way of
avoiding misunderstandings like this.  That's why you need standardized 
for terms, and standardized methods for diagnosing membership in taxa.  I've 
objected to the _concept_ of cladistic analysis and phylogenetic taxonomy.  My
objections all concern the way the concepts are put into practice.

> Taxon names are not like computer code.  They are more like personal names.
> Philosophically, a taxon is an individual.  This is why it is inappropriate
> to refer to The Dinosauria - it should be Dinosauria, just as it's Jon
> Woolf, not The Jon Woolf.

A taxon is an individual entity, yes.  So is a variable.  And a field, and a
record, and a file.  In the programming paradigm called Object-Oriented 
(OOP), objects can be organized into "classes," where each class is either a 
class or a subclass of another class.  Each subclass has all the features of its
parent class, and at least one additional characteristic that makes it unique.
Sounds a lot like a typical nested-tree of taxa to me.

>  It's a cardinal rule for me:
> >Don't Screw Around With Names.  Ever.  If a programmer doesn't use naming
> >conventions, it instantly tags them as a sloppy thinker and almost certainly
> >error-prone in other ways too.  And it makes the program a whole lot harder 
> >to
> >follow.
> Again, I don't think computer code is a valid model for taxonomy.  The
> goals and underlying philosophies are very different.

Are they?  My goal when I write a program is twofold.  First, it has to do what
it's supposed to do.  Second, it has to be readable by another programmer --
because I know that someday, some other programmer will have to follow my 
and understand my logic.  To that end, names, and comments, and code structure 
need to be carefully constructed to communicate information efficiently.  
information in minimum time and space.  I honestly don't see why taxon names 
be any different.  Knowledge is useless if it can't be communicated to others,
right?  So who's a better communicator: somebody who uses simple, clear, easily
understood names, or somebody who uses obscure tongue-tangling names that force 
to consult six other papers in four different journals published over the last 
years in order to figure out what they're talking about?

> >No, but there is apparently something that you aren't getting.
> >_Morganucodon_ is
> >a small, somewhat shrewlike fossil from the Early Jurassic.  It has
> >differentiated
> >teeth, semi-erect posture, a dentary-squamosal jaw joint, evidence of
> >endothermy,
> >a bunch of unusual specializations about the ear and braincase -- in
> >short, it has
> >every feature I've ever seen or heard used to diagnose mammalness in fossils.
> >None of the authorities I've read can point to any trait or suite of
> >traits that
> >distinguish crown-group mammals from morganucodonts.  Most are pretty
> >unequivocal
> >in calling _Morganucodon_ a mammal, even though it clearly dates from
> >before the
> >probable LCA of monotremes and therians.  So, how can you possibly tell
> >that it
> >_isn't_ a descendant of that LCA and therefore a member of Mammalia?
> Have a look at Rowe (1988, Jour. Vert. Paleo.) or the papers (esp. that by
> Rowe) in the Mammal Phylogeny volumes for the details.

Love to if I could find 'em, but my resources are limited.  Right now, the best
references I have on early mammals are Hopson's 1994 paper in _Major Features of
Vertebrate Evolution_ and Zhexi Luo's article "Sister-group relationships of
mammals and transformations of diagnostic mammalian characters," from _In the
Shadow of the Dinosaurs: Early Mesozoic Tetrapods_.  Hopson discusses two
definitions of Mammalia: the "classic" definition where a mammal is defined as
anyhing that has all the features considered to be diagnostic of Mammalia; and
Rowe's phylogenetic definition of Mammalia as a stem-based taxon (thx for the
correct term) composed of the LCA of Theria and Monotremata, and all 
descendants of
that ancestor.  Hopson points out that describing that LCA in a useful way is
problematic at best, since much of early mammal classification is based on teeth
and monotremes have few or no teeth.  He then takes the (IMHO) very sensible
position that since it's impossible at this time to tell which Mesozoic taxa fit
into Rowe's Mammalia and which ones don't, he's going to use the classic
character-based definition.  Luo simply side-steps the entire argument, using 
term Mammalia with the classic definition and noting that this is the same as
Rowe's Mammaliaformes.

Incidentally, and jerking this grimly back to the list's ostensible topic <g>, 
appears that Dinosauria has a similar problem, though not as extreme.  Current
taxonomy as given in _The Complete Dinosaur_ says that by the time they appear 
the fossil record, the three major groups of dinosaurs (Theropoda, 
and Ornithischia) are already separate, and there is no known animal that's a 
candidate for their LCA.  So how do you tell what's a dinosaur and what isn't?

> >This is an interesting statement, for a couple of reasons.  First, what
> >makes you
> >think cladistic analyses are any more objective than the old Linnaean 
> >methods?
> I'm not sure "objective" is the right word here.  But, phylogenetic methods
> do have two big, big plusses:
> 1.  When presented properly (and I know they're not always so presented),
> they're explicit.  They come with a matrix.  You don't like my codings?
> Here's the data - have at it.  Redo it, rerun it yourself.  And to the
> criticism that no one really goes through those big matrices - I know of
> two people doing just that with mine, and I've done it with the matrices of
> several others, both on theropods and crocs.

Using the same datapoints and characteristics, or different ones?  One great 
of computer-based analyses is that they're 100% reproducible: given the same 
and the same program, you get the same output every time.  But if the input is
wrong, the output will also be wrong -- the same way, every time.  And if the
program is flawed, the output will be flawed -- the same way, every time.  Been
there, seen that, got the battle-scars.  I don't place blind faith in computer
analyses, and I get skeptical whenever anyone else does.  A computer's only as
smart as its program, and only as accurate as its input.

-- Jon W.