[tdwg-content] Taxon and Name

[Richard Pyle]

Hi Nico, 

>     As a matter of clarification, perhaps also to the group - 
> the "definition" I provided for a taxonomic concept has a bit 
> of a normative quality (agenda is too grand a word). The 
> thinking behind it is that concept taxonomy with fairly 
> rapidly dissolve into name taxonomy if the distinction 
> between acts of authoring (even if congruently), citing, and 
> identifying to, concepts is not maintained with some consistency.

Agreed (I think...if I understand you correctly).

>     Sure, the Catalogue of Life (as just one example) 
> purports to present some authoritative (mix of) taxonomic 
> view(s). An informal name on a museum specimen by a late 
> expert of the group probably translates into a concept in the 
> mind of a student familiar with the group.

Yes, they're all implied concepts.  It's just that only a tiny, tiny subset
of them are well-defined.

>     I just think that there's this other taxonomy out there 
> in the future, where we taxonomists think and act more like 
> we care for others (incl. computers) to understand our 
> classifications, where the parts come from, what's congruent 
> and what has changed, how to precisely reconcile with 
> previous views, etc. And for that future to become more real, 
> perhaps a high threshold for identifying new concepts (in the 
> sense of authoring anew [versus citing], not necessarily a 
> new meaning) is needed.

Yes, I can see that (and we've had this conversation before).  But first we
need to come up with a common understanding of what a "taxon concept" is,
and how to articulate it with some precision.  One way to define a concept
is as a clade; that is, something like "all descendents of the most recent
common ancestor of these two organisms".  This is one of the ways that
Phylocode establishes the definition of a clade.  It's not perfect (nothing
is ever perfect), because with such definitions you will often have
organisms that belong to two different taxon concepts simultaneously
(hybridization, introgression, etc.) But that's not intrinsically a bad
thing, as long as it's understood and accomodated.  The real problem, of
course, is that we're still in our relative infancy in our ability to
discirn whether or not a paricular organism is, or is not, a descendant of
the most recent common ancestor of two other organisms.  Also, there is the
problem of mapping to centuries of legacy information.

So I think the most practical way to define the cirumscription boundaries of
a taxon cocnept at this point in history (and the one that is most likely to
leverage historical content) is via type specimens (proxied by heterotypic
synonyms).  It's much fuzzier and less precise than the mechanism described
in the previous paragraph, but far more practical.

What you're advocating is, I think, represents a reasonable path forward
towards more robustly defined, and objectively articulated, taxon cocnepts.

>     In other contexts, possibly including the representation 
> of identification events in museums, the bar for calling 
> something a concept need not be that high (informal names, 
> names outside of publications, local checklists, etc.). In 
> any case, it's a matter of where one puts the emphasis, and 
> hopefully I've pointed out where I would set it and why.

I think the key distinction that should be made is the distinction between a
"defined" concept, and an "implied" concept.  Almost every Taxon Name Usage
(sensu lato) instance carries with it an implied taxon concept, but as I
said, the vast, vast majority of those (especially if you include Museum
specimen identifications) are extremely anemic on deails for understanding
the boundaries of the implied taxon concept, and therefore it's difficult or
impossible to reliably map the congurnecy (or not) with other implied or
defined taxon concepts.  On the other hand, what we should be really
striving for is recognition of taxon concepts that are well "defined". These
are also rooted in TNU's, but carry with them robust information for
inferring the boundaries of the circumscribed concept (full synonymy, robust
mateial examined, robust descriptions of morphological and/or genetic
characters, etc.)  I think this distinction is important to make ("defined",
vs. merely "implied"), because what we'd ultimately like to do is find a way
to map implied concepts to well-defined concepts.

Getting back to DwC, one area of direct relevance to this is the
Identification class. 

Most datasets out there simply slap a taxon name to a specimen or
observation.  Some of them go so far as to say who identified it to that
name, and when.  But most do not take the final step and anchor the
identification to a particular well-defined concept (or, indeed, any TNU).
All specimen identifications represent an action that places the specimen
within the boundaries of a circumscribed taxon cocnept (whether the person
making the identification realizes this or not).  What we should be striving
for is a mechanim to tie specimen identifications to particular TNUs that
represent reasonably well-defined taxon cocnepts.  The statement should be:

"On this date, this person asserted that this specimen falls within the
taxon concept circumscription of Aus bus (L.) sec. Smith 1990".

If an Identification is a tuple of a Taxon instance (in the DWC sense) and
an Occurrence instance (or an Individual instance, if that class becomes
established), with associated metadata, then I think DwC is already primed
to make the quoted statement above (i.e., to anchor Identifications to
particular usage instances, because "taxonID" can represent a specific usage
instance -- assuming the right attributes ae included).  Retsated in DwC
terms, this would be:

"On [dwc:dateIdentified], [dwc:identifiedBy] asserted that
[dwc:occurenceID|dwc:individualID] falls within the taxon concept
circumscription of [taxonID], which represents a taxon name usage for
[dwc:scientificName] according to [dwc:nameAccordingToID]"

In the future, if Peter DeVries is successful with his ambitions, then this
could be simplied to:

"On [dwc:dateIdentified], [dwc:identifiedBy] asserted that
[dwc:occurenceID|dwc:individualID] falls within the taxon concept
circumscription of [taxonID], which represents [dwc:taxonConceptID]"

Or maybe even:

"On [dwc:dateIdentified], [dwc:identifiedBy] asserted that
[dwc:occurenceID|dwc:individualID] falls within the taxon concept
circumscription of [taxonConceptID]"

In the latter two examples, taxonConceptID would represent an array or set
of TNUs with congruent taxon concept definitions.

I'm still not 100% sure how best to use dwc:identificationReferences, other
than perhaps as a method for aggregating multiple TNUs (assumed to refer to
congruent, or at least overlapping concept circumscriptions), which were
used by someone when making an identification.

Aloha,
Rich