Angiosperm Families - INTKEY and WWW Packages

John Trueman trueman at RSBS-CENTRAL.ANU.EDU.AU
Wed Apr 2 09:06:03 CST 1997

Dear Mike
In a recent posting regarding a new INTKEY database, Mike Dallwitz wrote:

>Users should note, however, that it is impracticable to account for
>proposed family recircumscriptions when (as is usually the case with
>those reflecting recent cladistic and molecular/cladistic studies) they
>are presented unaccompanied by adequate revised family descriptions.
>How long will journal editors and referees allow this taxonomic
>malpractice to continue?"

Mike, aren't you confusing two quite separate things: the delimitation of
taxa and the means for identifying members of those taxa?  The former quite
properly should be based on phylogenetic relationship, and relationship
alone.  The latter is best served if there are some good, clear, character
states ('autapomorphies', in cladistics jargon) by which all members of a
taxon can be recognised, but sometimes these simply do not exist.

Generally speaking, cladistic and other numerical analyses (eg, molecular
studies) identify taxa by apomorphies.  At times these are obscure things,
such as a 'G' at position 1045 along some molecule.  Items like these do
not make good diagnostic characters.  Nonetheless the groups they delimit
may be well supported and can be genuine monophyletic groups worthy of
formal recognition. A family can quite properly be defined as, say, "all
species more closely related to X than to any other species" but this is
not to say the family itself (ie, all member species and only member
species) can be diagnosed, other than by the obscure thing.  We cannot
necessarily expect every good family to display characters of the sort
which INTKEY seeks when we ask it to place a specimen into it's correct
category. Sometimes we may have to identify to some lower taxonomic level
first, then ask what family that genus (or whatever) has been placed in.

If, also, you are asking that every specimen of a family display some
simple character-state by which it can be identified to family, I think you
confuse character with attribute.  To my way of thinking a character state
describes a distribution of observables within a taxon.  An attribute is a
feature of a specimen, and is the condition of the observable in that
specimen.  For example, 'almost always black' is a good character state for
any of several species of crow, but no individual crow is 'almost always
black', they invariably are black, white, or piebald.  "90% black" might
well be a good cladistic (or phenetic) character state distinguishing crows
from sparrows "<10% black", but it only indirectly addresses the question
whether the black specimen before you is a crow.

I see nothing wrong with defining or redefining taxa in terms of estimated
evolutionary relationships, and I don't see current cladistic or
molecular/cladistic practice as being in any way a taxonomic malpractice in
this regard.

I wonder how other taxacom-ers view this issue?


John Trueman
Bioinformatics Unit
Research School of Biological Sciences
Australian National University        -------------------------------
Canberra, ACT 0200,  AUSTRALIA       | Progress is made whenever a   |
                                     | timid  hypothesis is rejected |
                                     | or a bold one is not.         |
ph: +61 6 249 4840                    -------------------------------
fax: +61 6 279 8525

More information about the Taxacom mailing list