[tcs-lc] RE: [SEEK-Taxon] RE: LC/TCS - How many schemas?

Richard Pyle deepreef at bishopmuseum.org
Thu Mar 3 13:25:09 PST 2005


> If a taxonomist "moves a species from one genus to
> another" during a revision there may be some conceptual changes
> which have taken place (either in terms of the description or
> circumscription of specimens)

Absolutely!  But the act of placing it in a different hierarchical context
(i.e., changing the genus) and the act of changing the concept
circumscription (size/shape/position of the circle) are two separate acts.
There are many cases where the name combination changes and the concept
circumscription doesn't, as there are many cases where the name stays the
same but the concept circumscription changes.  I don't, off the top of my
head, have any sort of sense that a genus change implies a greater or lesser
indication that the circumscription also simultaneously changed -- so I
think the two acts are mostly unrelated to each other.

> but as the code (as you point out)
> is only interested in the type then we may have a homotypic
> relationship because they have the same type but not necessarily
> in my understanding of things does this imply they will *always*
> be congruent in terms of the concept.

Certainly not!  Which was my basic point.

> not clear what you mean by the same usage.....

Explained below.

> > There really
> >should never be a situation of a "contains/contained in" or "overlaps"
> >between two homotypic synonyms.
> >
> not if homotypic synonym only implies the type

Sorry -- my language was confused there.  I meant "in the same context" --
meaning, according to the same author(s) at the same point in time. Across
different authors and/or within an author team across time. When the
author(s) or the point in time changes, homotypic synonym concepts relate to
each other by the same mechanisms as any other pair of concepts (with the
only real difference being that an "excludes" relationship involving a pair
of homotypic synonyms necessarily involves at least one
"misidentification").

> >Having said that, we should also acknowledge that Author "X"
> >might use name
> >"A" for one concept circumscription, and Author "Y" might use
> >the same name
> >"A" to represent a *different* concept circumscription.  As
> >long as both
> >concept circumscriptions overlap minimally at the primary type
> >specimen,
> >then these are both Code-compliant, subjectively different
> >concepts; neither
> >of which is a misidentification.  While "A SEC. X" should in
> >all cases be a
> >congruent concept with "B SEC. X" (and similarly "A SEC. Y" should be
> >congruent with "B SEC. Y"), it is NOT the case that "A SEC. X" is
> >necessarily congruent with "B SEC. Y".
> >
>
> so an author will never change his opinion as to the concept
> associated with any name?

Sorry -- I didn't mean to suggest that.  I used "Author 'X'" as short-hand
for "Author X, Year n" -- meaning "a particular author team at a particular
point in time"  a.k.a. "publication".

> I'm led to believe again that just as
> different people have different opinions the same person can
> change his/her mind over time as they learn more....

Absolutely!  My bad for not explaining my shorthand notation.

> >> This is one part of what LC is about because the concept
> >doesn't have to
> >be defined.
> >
> >-- or even implied.
>
> but from this discussion and Nico's analysis of "nomenclatural
> relationships" it seems very hard not to infer concepts...

There are strictly nomenclatural assertions that have very significant
impact on nomenclature, but which do not define or even imply any concept
(beyond the primary type specimen).  In other words, LC needs to accomodate
acts that invoke nomenclatural changes, but that do not in any way impact
(or really even involve) concept definitions.  Unless, of course, a "concept
consisting of one and only one specimen" (i.e., the circle circumscribes one
and only one dot)can be thought of as a valid concept -- and I do not
believe it can/should.

> >The only thing special about the "Original Concept" (the
> >concept used by the
> >author who first established/proposed a code-compliant name) is that it
> >anchors the name to a type specimen
>
> AND there is a description (in any good taxonomic work!) of what
> the author thinks that type specimen is a representative of

Yes, but that doesn't make it "special" -- i.e., distinct from any other
subsequent invokation of a name aling with a description of what the author
thinks that type specimen is a representative of.  My point is, from a
nomenclatural perspective, the concept circumscription (circle) defined in
an original description of a name carries no more special importance than a
re-defined concept circumscription in a later taxonomic revision including
that name.  "Original description" are only special (different from any
other taxonomic assertion) for nomenclatural reasons, not for concept
circumscription reasons.  It's status as the code-recognized publication in
which the name was created and a type specimen fixed is extremely important
for the name object, but of little consequence to the concept object.

> - I
> think this is quite important if we are ever to understand legacy
> data to look at how for example the distribution of species (or
> other taxa) has changed over time. As several taxonomists have
> said to me - who goes and looks at a type specimen when doing an
> identification - virtually no-one they rely on the description -
> only the taxonomists doing the classification usually look at the
> type specimens (this is not to say type specimens aren't
> important - I believe they are - but more to say that the
> original description associated with the name is equally
> important to people as (part of) the definition of that name.

It's only part of *one* definition of the name (the first one).  As you say
few people who make taxonomic determinations ever examine the type directly.
I would also say (for older names in particular) few people rely on the
original description of the name to make the determination either.  Most
people turn to the most recent taxonomic revision (which, one would hoped,
was authored by someone who did examine the type specimen).  Sure, there are
many cases where there are no taxonomic revisions after the original
description, in which case they turn to the original description -- but not
*because* it's the original description -- only because it's the *only*
description (of sufficient rigor).

Where the original description (and type specimen) become very important is
cases involving "misidentifications" (as I have defined that term) -- that
is, when a subsequent concept definition (circle) excludes the type-specimen
"dot".  And even then, from a nomenclatural perspective, it's not the
original description per se, but rather the fixation of the type specimen
that is important (which doesn't always happen in an original description --
especially for older names).  As I mentioned before, the one situation (that
I can think of) where the concept circumscription of the original
description truely is more important than any other subsequent concept
circumscription (from a nomenclatural perspective), is when a neotype needs
to be established for a name.

> >So, yes -- all authors of protologues had a concept (circle)
> >in mind when
> >they established a new name (basionym) -- but how the borders
> >of that circle
> >were defined by the basionym author have no bearing whatsoever on
> >Code-mandated rules of nomenclature (except that the primary
> >type specimen
> >"dot" must be somewhere within the circle).
>
> agree but it doesn't mean it's not important and it doesn't mean
> that you have to separate them out and deal with them
> independently.....

Wait a minute....are we arguing the same point here?!?!?  I'm not entirely
sure what your pronouns above point to, but if your statement above can be
correctly transcribed to:

"...it doesn't mean that you have to separate [original-description concept
definitions of a name] out [from other taxonomic concept definitions using
the same name] and deal with them independently....."

...then we are in complete agreement -- that is my point, that concept
circumscriptions defined in original descriptions should *NOT* be treated
independently from other (comparitively robust) subsequent concept
circumscription definitions using the same name.  The differences are
critical for nomenclature, but transparent for concept mapping.

> because as Nico is trying to show there are
> many subjective calls (i.e. concept related issues ) in
> nomenclatural relationships.

Yes, we can often make inferences about "subjective calls" relating to
concept definitions by simply examining the way the names are used -- and
those sorts of inferences fall clearly outside the domain of LC (in my
mind).

> >This is why I have continued to emphasize in various
> >discussion groups that
> >"Aus bus Smith" is NOT the same as "Aus bus Smith SEC. Smith".
> > The former
> >is a name object, whose only connection with the biological
> >world is via the
> >primary type specimen (i.e., its anchored to a dot).  The latter is a
> >concept object, which represents the "circle" that Smith drew
> >around the
> >type-specimen dot (and many other dots that were not the type
> >specimen) for
> >the name that he first established. The former is in the
> >domain of LC, the
> >latter is in the domain of TCS outside of LC.
> >
> BUT your former could not have existed without the latter so I'd
> questioned allowing it to exist on its own (a weak entity if you
> like!)(in some implementations you could do this for efficiency -
> but that's a separate argument)

O.K., I see what you're getting at.  In the vast, vast, vast majority of
cases (exceptions might include things like "Nessiteras rhombopteryx
Scott"), scientific names do not exist without a bona-fide concept
circumscription preceeding it (I think this was Paul's point as well).  I
have absolutely no quarrel with that.  But that, by itself, does not
necessarily mean that the information management structure should not
decouple name-object data from concept-object data (sorry for the
double-negative there). Just to be clear, my arguments are not about
efficiency in implementation.  My arguments are strictly about information
structure.  It is certainly true that the existence of a name implies the
existence of a concept, but that doesn't mean that the differences in
information management needs can be safely ignored.  I can also argue that
the existence of a concept implies the existence of multiple specimens, and
although TCS certainly allows references to voucher specimens, I don't think
you would agree that a concept is a "weak" entity that should be defined
*only* in the context of voucher specimens --- or would you?

> >The LC name object "Aus bus Smith" is a re-usable object for many TCS
> >concept objects, of which "Aus bus Smith SEC. Smith" is but
> >one.  The same
> >name object can be re-used for all of these concept objects:
> >
> >- Aus bus Smith SEC. Smith
> >- Aus bus Smith SEC. Jones
> >- Aus bus Smith SEC. Brown
>
> this is simply one way to model it - and expressed like this of
> course it looks sensible - but there are many other issues to
> consider - including the complexity of the schema if you want to
> model all you might want to say about name objects. (almost as
> complexly as concepts - there would be very little in the TCS you
> wouldn't need to model so the schema would be twice as complex
> and users would need to be very careful about what info they put
> where - I'll address the issue of LC simplicity versus complexity
> in another email)

O.K., I think you are misunderstanding what my feelings are with respect to
how the idea of "treating name objects separately from concept objects", and
what impacts that would have on schema design.  I completely agree with you
that we do not want to duplicate the complexity of the concept schema when
developing a names schema.  In fact, I think overlap should be reduced to as
close to zero as possible. As I have said before, the only real point of
informational overlap is type specimen fixation (the only place where
code-governed name information intersects with biological reality).  But
part of this involves the acknowledgement that these two statements:

"'Aus bus Smith' is the type species of the genus 'Aus Smith'"

 - and -

"The genus concept 'Aus Smith SEC. Smith' includes the species concept 'Aus
bus Smith SEC. Smith'"

--convey different information.  So, if the Names/LC subschema of TCS
includes elements that establish the name-object 'Aus bus Smith' as the type
species of the name-object 'Aus Smith', and a TCS package contains
concept-object instances of 'Aus Smith SEC. Smith' and 'Aus bus Smith SEC.
Smith', with a RelationshipAssertion that establishes an "includes/included
in" relationship between these two concept-object instances, that should not
be seen as duplication of information.

> agree I don't like you way of presenting it ;-)

Having just re-read what I had written [excluded here], I'm not even sure
that *I* understand what the hell I was trying to say.... :-)

> I would say that there is a concept Aus bus Smith SEC Smith
> (which happens to have a "simple name element" Aus bus Smith)

O.K.... so far, no argument.

> we then have a new concept
> Xus bus (Smith) Pyle SEC. Pyle
> which has a relationship to Aus bus SEC Smith declaring it to be
> a new combination
> and whatever other relationships are appropriate.
>
> we then have another new concept
> Xus bus (Smith) Pyle SEC. Kirk
> which has an appropriate relationship to Xus bus (Smith) Pyle SEC. Pyle
>
> anyone interested in following name changes can track back
> through the history of relationships. They can choose to print
> only the name elements if they want and the relationships between them....

O.K., let me try to understand where you're coming from on this by asking
this question:

In your view, would the connection between "Aus bus Smith" as the basionym
for "Xus bus (Smith) Pyle", as represented in a TCS data package,
necessarily require at least one RelationshipAssertion?

At the heart of this debate, I think, is the desire among nomenclaturalists
to have a schema that establishes the association of "Aus bus Smith" as the
basionym for "Xus bus (Smith) Pyle", WITHOUT the need to create a
RelationshipAssertion instance.

> PS: I'm really impressed if you managed to get to the end of this
> - I nearly gave up several times and you're probably wishing I had.... ;-)

I omitted some of the discussion, because I think a lot of it results from
some basic miscommunication problems (including me miscommunicating with
myself!).

As for wishing you had given up -- DEFINITELY not (at least not from me)!!!

Rich





More information about the Tcs-lc mailing list