Assertions and Circumscriptions (Was: names and numbers)

Richard Pyle deepreef at BISHOPMUSEUM.ORG
Sun Oct 21 15:46:08 CDT 2001


A new thread has begun....

> The big question is, would you model and build them differently, say
> as separate entities, in a database sense...

Good question, and I think the answer depends on how you want to use
circumscriptions (see below).

> The determination assertion (specimen identification) is obviously quite
> a different beast to a taxon assertion (taxon circumscription or
> concept), although the latter can be thought of as the totality of the
> former...

I see a "Determination" as the intersection between a specimen and an
assertion.  The source of Determination assertions could be assertions
created specifically for the purpose of the determination (e.g., expert
comes to museum, examines specimens, assigns taxon names to specimens), or
they can be assertions gleaned from other published or unpublished sources
(e.g., Author of published reference lists a series of Museum specimen
catalog numbers under a particular taxon name). In either case, the
relationship between Assertions and Specimens is many-to-many.  A single
assertion can serve to determine many specimens (e.g., expert visits museum
and within a narrow period of time identifies many different specimens as
belonging to a given species name; or a publication lists many specimen
catalog numbers under a given species name); and a single specimen may have
been determined multiple times by mutiple assertions (e.g., identification
history of a specimen).

I see this as fundamentally different from the relationship between
Assertions and Circumscriptions, which can either be one-to-one, or
many-to-one, depending on how flexible you want to be in defining
circumscriptions (again, see below).

> Both are subjective and can only be expressed objectively in the context
> of a reference to a person making an assertion in a particular place and
> at a particular time...

Absolutely, and I will take your phrase, "reference to a person making an
assertion in a particular place and at a particular time" as my core
definition of the concept of a "Reference"  (I usually define a Reference as
simply "time-stamped Agent(s)"...but I think we're driving at the same
point). I agree that the "objective" manifestation of both Determinations
and Circumscriptions *must* exist in the context of a "Reference" (as
defined in the previous sentence), and so I think that both can be
structured (both conceptually, and as database entities) with direct
association to an "Assertion".

As I've repeatedly suggested, I've come at this from the "bottom-up"
approach, so I haven't really given much thought to how the notion of a
"Circumscription" is used. It seems that we all agree that it represents the
boundaries of a "taxon concept", and that the most common practical use of
Circumscriptions is to map equivalencies among different use of nomenclature
(e.g., Smith's 1992 survey of the flora of Sibera referred to species name
Aaa bbb; and Jones' 1999 survey of the flora of Sibera referred to the
species name Xxxx yyyy to mean the same taxon concept, so we can map the
equivalency of
"[Aaa bbb sensu Smith 1992] = [Xxxx yyyy sensu Jones 1999]").

Following this notion of a "Circumscription", I'll now try to articulat how
I imagine the concept fitting into a database, with respect to the notion of
an "Assertion".

First of all, to be clear, I am defining an "Assertion" very broadly as
simply any instance of a Reference citing a taxon name.  One will obviously
find a vast range in the scientific "quality" of such broadly-defined
"assertions", ranging from original descriptions, taxonomic revisions,
non-taxonomic articles, etc. -- all the way down to grey literature, popular
books and articles, and even newspaper articles. Also, Assertions are
entities of nomenclature; not necessarily entities of biological assemblages
of "taxa". In a very "liberal" sense, all of these assertions could
potentially represent Circumscriptions; but except for the implied
biological relationships among primary type specimens, Assertions don't
neccessarily define the boundaries of taxon conecepts with much meaningful
precision.

So how are circumscriptions definied with any meaningful precision?  I know
of three basic ways, at three levels of precision (resolution):  Taxon Names
(Primary Types); Populations (Biogeography), an individual specimens.  The
second of these (Populations) is a bit nebulous, so I'll skip if for this
discussion, and concetrate on the other two (Taxon Names, and individual
specimens).

At the level of Taxon Names, an Assertion can only serve as a meaningful
representation of a Circumscription in the context of other Assertions by
the same Reference about related taxon names.  For example, suppose we have
the three species names:

Aaa bbbb
Aaa cccc
Aaa dddd

For simplicity, I'll keep it all within a single genus and only talk about
species-level circumscriptions, although the same concept extrapolates
easily upwards to all taxonomic ranks. We'll also assume that these are the
only three species names ever assigned to the genus Aaa.

We'll also work with the three references:

Smith, 1967
Jones, 1990
Pyle, 2000

Suppose, then, that we've got the following records in the Assertion table:

ID  Reference      Name  Rank   Valid   Parent
----------------------------------------------
1   Smith, 1967    bbbb  Sp.    bbbb    Aaa
2   Smith, 1967    cccc  Sp.    cccc    Aaa
3   Smith, 1967    dddd  Sp.    dddd    Aaa
4   Jones, 1990    bbbb  Sp.    bbbb    Aaa
5   Jones, 1990    cccc  Sp.    cccc    Aaa
6   Jones, 1990    dddd  Sp.    cccc    Aaa
7   Pyle, 2000     bbbb  Sp.    bbbb    Aaa
8   Pyle, 2000     cccc  Sp.    cccc    Aaa

>From this we know that Smith (1967) treated all three species as valid and
distinct (because the all have Valid=Name). We also know that Jones (1990)
regarded bbbb and cccc as valid species, but treated dddd as a junior
synonym of cccc.  Also, Pyle (2000) treated bbbb and cccc as valid, but
being the sloppy taxonomist that he is, made no mention of what he felt the
status of species dddd should be.

No single record in the above Assertions table constitutes a complete
Circumscription by itself, without knowledge of other assertions made by the
same Reference.  For example, Jones' concept of the taxon represented by the
name cccc represents the combinations of Smith's concepts of the taxa
represented by the names cccc *and* dddd.  We can only know this if we know
that Jones regarded cccc as valid *and* that he regarded the primary type of
dddd and all its kin to be part of the same taxon concept represented by the
name cccc.  If we did not know about Jones' treatment of dddd, we could not
represent a complete circumscription of Jones' concept of cccc.  Similarly,
we can't really know what Pyle had in mind for his concept of cccc, because
he never told us whether he would have included all things previously
assinged to dddd within his cccc circumscription. So, is [Aaa cccc sensu
Pyle, 2000] equivalent to [Aaa cccc sensu Smith, 1967], or is it equivalent
to [Aaa cccc sensu Jones, 1990]?  Without Pyle's assertion about dddd, we
have no way of knowing what Pyle's concept of cccc really is.

The significance of this is that, while Pyle probably had a concept for cccc
(and for dddd in the context of cccc), and thus Assertion record ID# 8
*could* represent a circumscription; we have no way of delineating the scope
of that circumscription based on the incormation contained within Pyle,
2000.  In this way, Assertion record ID#'s 1-6 are all well-defined
Circumscriptions, because they exist in full context, but Assertion ID# 8 is
not.

What about Assertion ID# 7?  In one sense, we could assume that Pyle
regarded bbbb in the same way that all other References treated bbbb (i.e.,
bbbb sensu Pyle as equivalent to bbbb sensu Smith and also equivalent to
bbbb sensu Jones) -- but we can't do that objectively, because for all we
know, Pyle thought of dddd as a junior synonym of bbbb, in which case his
concept of bbbb would be broader than either Jones' or Smith's.

What's the point?  The point is that each Assertion record represents a
"potential" Circumscription, but not all of them can be used as "defined"
circumscriptions.

This is why I said in my last post that I see Circumscriptions as a subset
of Assertions (that is, all Circumscriptions are based on Assertions, but
not all Assertions can be used as Circumscriptions).  Getting back to your
original question, Jim, my simple approach would be to treat
Circumscriptions and Assertions as the same database entity (same primary
key pool), but  that only a subset of assertions constitute meaningful
Circumscriptions. This could be dealt with either by splitting
Circumscriptiopns into a separate table with a one-to-one relationship to
Assertions (linked on their respective Primary Keys); or keep it all within
the Assertions table and flag the Circumscriptions accordingly. The
difference is purely a technical issue.

However, I think we might be restricting ourselves too severely in this
scheme, so an alternative approach would be to establish a many-to-one
relationship between Assertions and Circumscriptions.  For clarity, this
wouldn't be a many-to-one relationship in the classical sense, but rather in
a contextual sense.  For example, suppose there is a reference for Pyle in
2001 where he declared that dddd should be treated as a junior synonym of
bbbb.  Thus, if we expand the scope of Pyle's assertions to treat his 2000
reference and 2001 references as contextually equivalent, then all of a
sudden we can use his 2000 assertion about cccc as a Circumscription,
because it is elaborated in full context by the combination of both the 2000
reference and the 2001 reference. While this would technically leave is with
a one-to-one relationship between Assertions and Cicumscriptions, we would
need an additional layer in the database scheme to map which set of
Assertions can be combined to represent a contextual group.

Now is not the time to explore that further (I'll probably have to break
this post up into 2 parts as it is), but the point is that Circumscriptions
involve considerably more complexity (particularly in the sense that they
are sensative to the context of other assertions), than simple assertions
do.  There's also a whole 'nother tome to write about with respect to
definitions of Circumscriptions in association with specimen
Determinations...but I'll simply address that by saying that I think these
specimen-based circumscriptions can be accomodated via assertions, as long
as the assertions are the foundations of specimen Determinations.

Overall, I'd say that References, Names, and Assertions all conform well to
the notion of centralized, universal ID numbers, because they can achieve
maximal usefulness within their purely objective contexts; whereas
Circumscriptions may not yet be ready for universal centralizations, because
subjectivity may be required to decide which Assertions can represent valid
Circumscriptions, and perhaps also which set of References can be grouped
together for the perposes of defining a full Circumscription context.

Whew....I think I spent more time on that than I probably should have....

Aloha,
Rich

Richard L. Pyle
Ichthyology, Bishop Museum
1525 Bernice St., Honolulu, HI 96817
Ph: (808)848-4115, Fax: (808)847-8252
email: deepreef at bishopmuseum.org
http://www.bishopmuseum.org/bishop/HBS/pylerichard.html
"The views expressed are the author's, and not necessarily those of Bishop
Museum."




More information about the Taxacom mailing list