Revolution

[James Lyons-Weiler]

On Sun, 29 Sep 1996, ROGER HYAM wrote:

> Dear All,
>
> James Lyons-Weiler made an interesting point in reply to the Eye Eye
> problem.
>
> He said:
> "Kuhn's take on the matter would be that progress in taxonomy can only
> occur when a critical mass of common thought renders an old paradigm
> obsolete, and ushers in a new one.  I don't foresee a successful
> revolution in taxonomy as a process anytime soon."
>
> I have argued for some time that we are approaching a revolutionary
> paradigm shi(f)t in taxonomy. It is  comparable to the Copernican
> paradigm shift. Prior to the acceptance of Copernicus' views the
> explanations for the movements of planitary bodies were getting more
> and more complex with an increasing number of  anomalies. Is this
> not similar to what is happening in taxonomy? Certainly it is in
> higher plants. The exceptions to the rule are growing all the time.
> Surely if the diversity was "atomisable" as we are presuming we would
> have finished with vascular plants by now. (Could all be having fun
> with phylogenies in stead of fretting over what our terminal taxa
> should be?!)
>
> When ever I mention this to people who know about paradigm shifts
> (but not necessarily taxonomy) they imply (very politely) that I am
> talking out of my hat.
>

The complexities in modern taxonomic science (in which I am throwing all
of systematic biology) result primarily from discordance between
independent sources of evidence for natural phylogenetic relationships.
Congruence has held a special place in taxonomic science; when new sources
of data are developed, congruence is almost universally taken as an
indicator of corroboration.  However, discordance has no ubiquitously
accepted significance.  Taking trees from new evidence as indicating that
old tree are wrong is dangerous, because then the conclusions drawn simply
depend on which data are generated first.  Therefore, congruence does not
provide a critical test by which to gauge the accuracy of phylogenies..

An area that has been almost completely ignored is the study of ways to
define when a particular set of characters is likely to be misleading.
Questions about why discordance exists between disparate sources of data
are usually answered in an ad-hoc fashion, filling page after page of
journals with guesses about why the new data do not fit the old trees, and
vice versa.  What is needed are ways to tell when specific aspects of a
given set of characters is apt to bring about erroneous phylogenies, hence
discordance.

A good deal of current ongoing research includes the development of
measures of heterogeneity, ways of ffinding long branches (which can wreck
an otherwise perfectly accurate phylogeny), and ways of performing data
exploration.  However, unless people developing these methods take the
time to make them generally understood, easy to implement, and tie them in
with the existing paradigm, change will be slow.  Taxonomists should take
some time to explore some of the new ways of looking at old data, because
almost invraiably they are rewarded with phylogenies that begin to make a
great deal of sense.

That may sound like a shameless plug.  If it is, it is.

James Lyons-Weiler