Re: Underlying basis of classification (Was: Re Dinobirds)

[chris brochu <cbrochu@fmppr.fmnh.org>]

Philidor11@aol.com wrote:
> 
> The subject I've been inquiring about is the nature of a cladistic
> hypothesis, and from your response and Mr. Holtz's I think a number of
> observations are reasonable.  Before checking to make sure I've understood
> properly, let me mention a site which makes such deliberations more pleasant:
> <A HREF="www.hamsterdance.com">mammalian movements</A>
> Be sure to let the sound load.
> 

The URL doesn't work - sorry.  I tried changing it to
www.hamsterdance.com/mammalian.movements, and no luck there, either.

> 1. a hypothesis concerns characters shared by a number of animals which may
> have no functional significance to any or all of the animals

More generally, a hypothesis is a proposition whose truth or falsehood
is not initially known. (I got that definition from John Merck - not
sure where he got it.)  Hypotheses describe patterns.  Theories are
statements of process that explain hypotheses, provided the hypotheses
are not falsified.  I can say "Humans are more closely related to squid
than to magnolias" - this is a hypothesis, and we have no a priori
reason to accept it as true or false.

In phylogenetics, hypotheses are usually expressed as branching diagrams
(cladograms) - though that's not the only way to do it.

I agree with the statement you make toward the end, that phylogenetic
hypotheses are not like other hypotheses.  In experimental science, we
know instantly whether a hypothesis is flat-out wrong.  We can't do that
quite so simply in phylogenetics - we can overturn a tree insofar as a
new data set supports something else, but there's always a possibility
that even more data will tip the balance back to the previous tree. 
This is why we usually call trees "recovered results" or "estimates"
rather than hypotheses - and it's also why so much time and paper are
being spent on measures of support and robustness for different kinds of
trees.

I discovered what may be the best published description of phylogenetic
methodology at the evolution meetings, and part of it is available
on-line.  Go to http://www.sinauer.com/Titles/Text/judd.html#chaps and
select "chapter 2:  Principles of Biological Systematics."  There will
be much there on what I'm talking about.




> The example used was characters linked as part of a survival strategy for the
> animals.  This seemed a promising possibility as evolution is based on
> different rates of survival caused by different attributes.

Selection is an important driving force behind evolution.  But it is not
the only one.  One of the central concepts in modern molecular
phylogenetics is neutral evolution - this isn't exactly the same as
random genetic drift, but for our purposes it's close enough.  Features
do not always propagate through a population because they're
advantageous.

[shortened]


> 3. The characters used to infer a relationship may be changed without
> affecting the hypothesis.  This implies that more than one set of characters
> may attain similar levels of validity.
> One example is dinosaurs included among ornithiscian dinosaurs which lack the
> bird-hip which apparently diagnoses the group.  An alternative definition is
> used to associate certain dinosaurs with the group, apparently with equal
> validity.

Actually, the ceratopsian example I gave does not describe what you're
talking about - it describes the reason why a fossil lacking the
diagnostic characters for a group can still be diagnosed within the
group.

But, your point is well-taken.  In phylogeneticspeak, "Multiple samples
from the same data set should preserve the same signal, even though the
specific support at each node may change."



> Mr. Brochu observes:
> <<But the definition of Avialae is the same [despite finding dinosaurs with
> feathers], and the shape of the tree in that part of Theropoda is no
> different.  This is because the definition of Avialae has nothing to do
> whatsoever with the characters diagnosing
> the group - it would include birds and those taxa closer to them than to
> dromaeosaurids regardless of the characters at the root of Avialae.>>
> The characters change, but the avialae group is as valid as ever.  Unless
> someone has insisted on the existence of avialae, new analysis has produced
> characters which apparently were present before the prior, feather-based
> analysis and which come to the same result.
> This is important because it implies that there is not much difference among
> the best results of the cladistic analyses.

I'm not sure what you mean here.  



> 
> In sum,  it seems that the only immediately apparent criteria for choosing
> among potential hypotheses which include all the available data are logical
> (particularly simplicity of explanation [fewest transformations]) rather than
> biological, 

Not true for many molecular analyses, which tend to rely on likelihood
models which are explicitly based on some sort of biological framework
from the get-go.  But if you rely on parsimony, then yes - our primary
tree-selection criterion is logical.



and that whatever the criteria used, they allow a number of
> similarly successful answers.  The form of the hypothesis is that a common
> ancestor of a specified group of animals would produce specified characters
> in the fewest evolutionary steps.


This isn't how we would normally phrase it, but the concept is basically
the same.  




> That type of formulation gets around the definition/diagnosis problem neatly,
> because the hypothesis generating mechanism is looking at all possible common
> ancestors and then seeing which ones generate useful hypotheses.  The truth
> of a posited ancestor is being demonstrated by the finding of a reason to
> believe that the ancestor existed.  In a sense it is asking what characters
> justify the existence of a common ancestor.  I never was good at working in
> mirrors, so it was hard for me to see.


This isn't entirely accurate.  We don't normally "look at all possible
common ancestors."  Rather, we look at all possible hierarchies of
relationship - or more accurately still, we let our computer algorithms
do it for us.  That ancestors "exist" on the nodes is understood, but
the search is for the hierarchy.



> 
> Of course, such logical hypotheses need strong verification, and that seems a
> difficult project. 


Not necessarily.  One of the best kinds of verification is congruence -
different data sets from different sources should preserve the same
signal.  Why should they preserve the same pattern if they don't reflect
the same signal?  And since morphology and molecules are passed on from
ancestor to descendent, what process other than evolution would cause
it?  Morphology and molecules should support very similar trees, and
most of the time, they do.  Stratigraphy and biogeography should
likewise preserve phylogenetic signals, and having a tree match
stratigraphy (or come close to doing so) is regarded as a kind of
congruence.  But having biotic (morphology, molecules) and nonbiotic
data disagree is not (or should not) be viewed as strong evidence
against the tree, as the nonbiotic data are not preserving phylogenetic
signals alone, and we don't know how strong the phylogenetic component
is relative to others.

Of course, I come at this as one who works with the living as well as
the dead, so I have the luxury of being able to use molecules as well as
morphology.




 Most of the evidence such as time when animals lived...
> (by the way, Mr. Brochu observed that
> <<We can make hypotheses of ancestral status on the basis of negative
> evidence - "Fossil A does not occur above fossil B, and does not have
> autapomorphies, and is the sister taxon to fossil B, so fossil A might be the
> ancestor of fossil B."  We could always later discover fossil B below fossil
> A, or autapomorphies in fossil A,
> that would falsify our hypothesis.>>
> I think he was oversimplifying for my benefit; the more primitive form may be
> contemporary with a later form, so fossil B might be found below fossil A.)

Actually, I wasn't oversimplifying at all.  Ancestors should precede
descendents.  

But I agree that later discoveries could reverse the situation and
resuscitate the hypothesis of direct ancestry.




> ...and other types of scientific inquiry seem to be weak or lacking.
> As Mr. Brochu says, the ancestor is also unlikely to be found and confirm the
> hypothesis:
> << We justify the inability to document an ancestor because, frankly, we can
> almost never do it.  The only positive evidence for ancestral status of a
> species would be the discovery that some members are actually closer to
> another species than to other members of its own species, and that requires a
> massive sample and is almost never
> reflected in morphology.>>
> Both Mr. Brochu and Mr. Holtz assert that the hypothesis must be disproved,
> not proved.

It was Karl Popper, actually - Tom and I (along with most of the
scientific community) just follow along.




> Mr. Brochu:
> <<Proven?  Can't be done.  But as scientists, we're not in the business of
> proving anything - its disproof we're after.>>  The disproof comes from new
> discoveries.
> Mr. Holtz:
> <<Chris is refering to a basic aspect of the scientific method: one cannot
> prove an hypothesis, but we can falsify it.  (The classic example: how many
> white swans do you need to see to prove the statement "all swans are white"?
> In contrast, how many black swans do you have to see to falsify the
> statement "all swans are white"?)>>
> The difficulty is that no paleontological hypothesis can be tested.

Sure they can.  As long as we have repeatable observations, hypotheses
can be tested.  The issue is how strongly we can take statements of
"You, sir, are falsified!"  This will vary depending on the hypothesis.



  In terms
> of Mr. Holtz's analogy, all the available swans have been looked at before
> the theory was ever produced.
> Because there are no ways of testing the hypothesis (prediction and
> experiment in the 'classic' scientific method), 

Ah, the old experimental-versus-historical debate.  Even Popper flubbed
on this one once.

The key to testing hypotheses is having repeatable observations.  In the
experimental world, we generate these by replicating the process in the
laboratory.  But we historical scientists can't do that, since the
"experiment" has already been run.  Where philosophers sometimes trip is
with the concept of "repeatable" - it's not the *process* that must be
repeated, but the *observation* - and as long as we continue to resample
from nature, we can repeat our observations.




the hypothesis can be either
> true until disproven or in doubt until some way can be found of proving it.
> I was carrying over from the 'classic' definition the attitude that no
> hypothesis is true until you prove it, that it must be possible to falsify a
> hypothesis immediately.

I agree that phylogenetic hypotheses are not immediately falsifiable -
but falsification still occurs.


chris


-- 
----------------------
Christopher A. Brochu
Department of Geology
Field Museum of Natural History
Roosevelt Road at Lake Shore Drive
Chicago, IL 60605

voice: 312-665-7633  (NEW)
fax: 312-665-7641 (NEW)
electronic:  cbrochu@fmppr.fmnh.org