Re: SCIENCE AND CLASSIFICATION (long)

["David Marjanovic" <david.marjanovic@gmx.at>]

> Terminology:  repeatable rather than reproducible, please.  Running the
same
> program on the same data a second time is a repetition.

But running another program on the same data, with another algorithm?

> When you have a hypothesis
> about extinct animals, you cannot immediately use the standard test of
> making predictions from the implications of your hypothesis because you
> cannot rerun evolution.  You can't generate relevant observations at will.
> You can, however, predict that a fossil with certain characters will be
> found, dating from a given time period.  Then, of course, you can settle
in
> to wait and see what happens.
> (Historic sciences and some physics all have this problem; the search for
> confirming observations can be difficult.)

Exactly. You've perfectly summed up the situation. Well, as we can't
influence what the next fossil will look like, I do think discovering it is
a test.

> The other problem to keep in mind is that nature is known not to follow
> parsimony (or any other logical principle) in every case.  You don't know
> that parsimony must give you the correct answer.

Indeed. Parsimony analysis is the best of all bad methods :-)

> Therefore, the use of
> parsimony is a choice (albeit a reasonable choice) on the part of the
> analyst. Any given conclusion is no more likely to be correct than a
> conclusion based on a different logical principle,

There are some -- maximum likelihood, minimum evolution,
neighbor-joining --, but, alas, these are all unapplicable to morphology.
:.-( So parsimony is the only thing known that we can use at all. The use of
pasimony is not a real choice when fossils are used.

> unless there is an
> argument that parsimony is the most likely operant principle in the
specific
> circumstances compared to the alternatives.

This situation does occur. Size and size-related characters are usually
excluded from parsimony analyses because they can easily evolve and reverse
convergently.

> In short, a more elaborate approach like cladistic analysis is not
> necessarily more objective or more accurate than an alternative, less
> formulated approach.

A less formulated approach is less reproducible and therefore less
objective, I think.

> Nothing wrong or unreasonable about it, just not the perfect and only
> solution to finding objective and true answers in evolutionary biology.
> Says me...

I agree wholeheartedly. Just give me a better method... :-)

-----------------------------------------------------------

> Do names really have to wait until evolutionary relationships have been
> established beyond a reasonable doubt?

Of course not. Just think of, say, Deinonychosauria.

> If I limit Birds to the feathered, etc. things I can see out my window,
then
> I have sacrificed sophistication about evolutionary relationships for the
> simplicity of a stable name.

But we _want_ to talk about "sophistication about evolutionary
relationships", and about the change in our ideas of this. This is _the very
idea_ of phylogenetic nomenclature.

> A sparrow will always be a bird.  There are a
> whole bunch of strange looking things dating back over 65 my that we're
not
> completely sure about.

Say Aves will be defined as {*Archaeopteryx lithographica* + *Passer
domesticus*} under PhyloCode. Then sparrows will likewise always be birds,
and we won't be able to stop eagles and ostriches from being birds. We can
then discuss about whether *Tyrannosaurus* is a bird -- whether it _is_ a
bird, not whether it _should be called_ one. It either _is_ a member of this
clade _objectively_ or not.

> Given that the sophisticated understanding of evolutionary relationships
is
> itself evolving, seems that keeping it simple, at least for now, is a good
> thing.

"At least for now" sounds like an Austrian provisorium that quickly turns
into "at least forever". Well, suppose the following definitions --

Aves {*A. lithographica* + *P. domesticus*}
Maniraptora {*P. domesticus* > *Ornithomimus velox*}
Eumaniraptora {*Deinonychus antirrhopus* + *P. domesticus*}
Maniraptoriformes {*O. velox* + *P. domesticus*}
Dromaeosauridae {*Dromaeosaurus albertensis* + *Velociraptor mongoliensis*}
Bullatosauria {*O. velox* + *Troodon formosus*}

and the following phylogenetic hypotheses (not all of them generated by
parsimony analyses, all much shortened and "redrawn" by me, so that they
include all the abovementioned species), then here

--+--*Archaeopteryx*
   `--+--*Dromaeosaurus*
       `--+--+--*Velociraptor*
           |    `--*Deinonychus*
           `--+--Oviraptorosauria
               |?--Ornithomimosauria
               `--+--Troodontidae
                   `--+--*Avimimus*
                       `--Pygostylia
(from PDW)

all shown animals (except the root) are Aves, probably Troodontidae,
*Avimimus* and Pygostylia are Maniraptora, *Velociraptor* and all below are
Eumaniraptora -- so Maniraptora is a part of Eumaniraptora --, probably
Oviraptorosauria and all below are Maniraptoriformes = Bullatosauria, and
*Dromaeosaurus* and all below are Dromaeosauridae. But here

--+--*Ornitholestes*
   |--+--+--*Dromaeosaurus*
   |   |    `--+--*Velociraptor*
   |   |        `--*Deinonychus*
   |   `--+--*Archaeopteryx*
   |       `--Pygostylia
   |--+--Oviraptorosauria
   |   `--Therizinosauroidea
   `--+--Tyrannosauridae
       `--+--Ornithomimosauria
           `--Troodontidae
(from Holtz, 1996)

only Archie and Pygostylia are Aves, *Dromaeosaurus* through Pygostylia plus
the root of this clade are Maniraptora, the same without the root are
Eumaniraptora, all except the root are Maniraptoriformes, *Dromaeosaurus*
through *Deinonychus* are Dromaeosauridae, and Ornithomimosauria and
Troodontidae are Bullatosauria. You see, the names are applicable under
totally different hypotheses -- they are stable, even though some can end up
being synonymous, and even though some can get "strange" contents. Yet
another example:

--+--Tyrannosauroidea
   `--+--*Compsognathus*
       |--*Ornitholestes*
       |--Ornithomimosauria
       `--+--+--Segnosauria sensu lato
           |     `--Oviraptorosauria sensu lato
           `--+--Pygostylia
               `--+--*Archaeopteryx*
                   |--Troodontidae
                   `--+--*Dromaeosaurus*
                       `--+--*Velociraptor*
                           `--*Deinonychus*
(from http://www.cmnh.org/fun/dinosaur-archive/2001Nov/msg00033.html, which
is from Longrich, 2001)

Here Pygostylia and all below are Aves = Eumaniraptora, Segnosauria and all
below plus the root of this clade are Maniraptora, *Compsognathus* and all
below are Maniraptoriformes = Bullatosauria, and *Dromaeosaurus* through
*Deinonychus* are Dromaeosauridae. And if I shamelessly use mine (also
shortened), then even under this unusual topology all names are applicable:

--+--+--*Archaeopteryx*
   |    `--+--*Dromaeosaurus*
   |        `--+--*Velociraptor*
   |            `--*Deinonychus*
   `--+--+--Tyrannosauroidea
       |     `--+--Ornithomimosauria
       |         `--Troodontidae
       `--+--+--Oviraptorosauria sensu lato
           |    `--Segnosauria sensu lato
           `--Pygostylia

All except the root are Aves = Eumaniraptora, Oviraptorosauria through
Pygostylia plus the (long-looking) root of this clade are Maniraptora,
Tyrannosauroidea and all below are Maniraptoriformes, *Dromaeosaurus*
through *Deinonychus* are Dromaeosauridae, and Ornithomimosauria and
Troodontidae are Bullatosauria.

While I am at it -- does anything except Dromaeosauridae, *Bambiraptor*,
*Sinornithosaurus* and *Archaeopteryx* have upside-down-T-shaped
quadratojugals? *Confuciusornis* doesn't, neither does the Spanish
nestling...
Has somebody found out who defined Pygostylia when? I haven't.
Many thanks in advance.