[Tcs-lc] concepts of Higher taxa

[Richard Pyle]

Martin wrote: 

> While on the whole I think we are all in agreement - namely
>  
> 1) The creation of new taxa cannot change existing higher taxon concepts
> 2) Because of the way nomenclature is implemented it is not and 
> should not be possible to identify the concept of  higher rank 
> that the author was placing the new taxon in.

I agree with the above two statements, as long as you repalce the word "possible" with "required".  I think it is certainly possible that the describer of a new species may reference a previously-existing concept for the parent genus, but it doesn't happen often, and it will not likely be required by the Codes.

> I think there are two terms that we cold coin here that may be useful
> 1) Nomenclatural placement
> 2) Conceptual placement

Yes!  I think these are very useful terms.  One of the points I've been trying to make for a long time is that the nomenclatural connection between a GenusName and a SpeciesName carries with it different information than the conceptual connection between a GenusCircumscription and a SpeciesCircumsctription; even though they both look the same.

> We the describe the publication of a new taxon - eg a species as 
> being in a genus this will be precise in its nomenclatural 
> placement but at best vague in terms of its conceptual placement.

I think at "best" it will not be vague, because it's entirely possible that when Jones describes his new species "xus" and places it in the genus "Aus", Jones may well happen to explicitly state, "I follow Smith's concept of the genus Aus in placing this new species within it."  This may happen if, for example, Smith recently revised the genus Aus, and Jones took one of Smith's species concepts and split it into two.

The problem, of course, is that this "best" case scenario probably doesn't happen often.  When a new species is described within the context of a revision or review of the entire genus, then we're probably OK, because there's no question that the author of the speicies is also the definer of a concept for the genus.  But I suspect more often there will be cases that do not rise to this level.
 
> Rich's suggestion is that the nominal concept is a good way of 
> handling the conceptual vagueness of higher taxa, I unfortunatley 
> have to disagree. As far as I can see the "nominal concept" only  
> muddies the distinction between nomenclature and classification 
> which is one of the things we are trying to avoid. To my mind the 
> use of the nominal concept in this way runs the risk of 
> overemphasising its importance by potentially giving the 
> impression of it being a coherent whole when in reality it would 
> be just an ad-hoc collection of entities that have no formal 
> placement in a higher ranking concept. 

O.K., so your concern here is that if we have the new species Aus xus Jones, and Jones did not explicitly define which concept definition for "Aus" that he intended to include his new species within (and did not define his own concept instance for Aus), then if we default to mapping Jones' xus to the Nominal TaxonConcept instance of "Aus", then we're effectively stuck in terms of placing it within family, etc.

That was worded in an awfully convoluted way....what I mean is, by linking a new species to a nominal genus-rank TaxonCocept, we may be unable to establish classification at ranks higher than family? If so, I understand this concern, and agree that it is a real problem. I'll have to think on this some more.  This would be the problematic example:

Jones describes the new species Aus bus, and includes it within the family Aiidae.  Jones provides a full enumeration of all the genera that are included within the family Aiidae, and therefore adequately defines or references a particular TaxonConcept instance for the family Aiidae.  But Jones fails to enumerate all of the species within the genus Aus, and therfore fails to provide an adequate concept definition for the genus Aus. The question is, how to reflect the classification of Jones: Aiidae-->Aus-->bus, when he provided explicit Concepts only for Aiidae and bus?

I think it is still possible to represent this via a Nominal TaxonConcept References in TCS, but I'm not yet sure exactly how.  I'll have to play with some instance documents.

> Why not accept the fact 
> that when a new taxon is published the associaton of the new 
> concept with a name of  higher rank is just that - a relationship 
> to a name not a concept. We can then represent this relationship 
> in the schema by a "link" to a name and not a concept (nominal or 
> otherwise). Now that it is agreed that names will be represented 
> in the schema as "top level" objects this should be quite 
> possible to do. 

Yes, if names are indeed represented as top-level objects, then I completely agree that this would be the way to handle it.  I'm not entirely sure that names are best treated as top-level objects, though.  I have not given up on defining them as Nominal TaxonConcepts.  When someone provides a taxon name that can be distinguished from homonyms, but is otherwise vacant of any specific concept circumscription definition; I think it is fair and appropriate to refer to the Nominal TaxonConcept instance for that name.  Absent evidence to the contrary, the use of an identifiable taxonomic name suggests strongly that the concept definition falls within the "sum of all concepts" for which the name has been applied.

> What is more because all the type infomation 
> should be held in instances of name objects and not concept 
> objects the default association between a member taxon and the 
> type of the higher name with which it is associated will be 
> maintained, but with the advantage that the type information need 
> not be repeated for each and every concept that is to be 
> represented. Although such a change will make the schema slightly 
> less elegant conceptually it will be much cleaner.

This is one of the fundamental points I've been trying to make all along:  The difference between name-name relationships and circumscription-circumscription relationships (and also, the difference between name-specimen relationships [=typification] vs. circumscription-specimen relationships [=SpecimenCircumscriptions]).

> >This sounds better to me, but I'm not clear of what you mean by 
> "revision".
> >Does that basically mean that concepts are created only when all 
> senior and
> >junior synonyms of all child taxa are explicitly addressed 
> within the same
> >publication?
> 
> This is where the binomial naming scheme trips me up again and I 
> have to mix concepts and nomenclature -sorry - 
> So the answer to your question is Yes - it is a question of 
> intent - in effect the placement of a new taxon in a higher 
> "thing" (nomenclatural placement)  requires the new taxon to be 
> "identified" within the context of an existing higher taxon 
> concept although the concept is not usually specified when the 
> name is published. Therefore, although the addition of the new 
> taxon to a higher taxon effectively creates a new higher taxon 
> concept, it was not the intention of the author of the new taxon 
> to do so nor is it clear what the circumscription of the modified 
> higher taxon should be. 

O.K., I understand and agree with all of the above.

> It is therefore better to leave the new 
> taxon without any association with any existsing higher concept.

...unless the author of the new taxon makes it clear to which previously-defined higher concept the new taxon is being placed.  In most cases, this information will not be provided.  Exceptions include:

- Naming of new species within the context of a genus revision.

- Simultaneous naming of a new species within a new (monotypic) genus (a case where the relationship between the new species concept and the new genus concept is "congruent" -- which breaks the assumption that "is child of" necessarily implies "is included in"...)

- Splitting of a previous species concept into two smaller species concepts, and explicitly refering to the same previous genus concept that had included the original (larger) species concept.

(I'm sure there are other examples...)

> Only when a publication clearly states that its contents 
> represent the results of a re-eaxmination of the entirety of an 
> existing concept, or a geographically restricted subset of such a 
> concept should we recognise, for the purposes of TCS, the 
> creation of a modified concept. 

I think I mostly agree...but the "geographically restricted subset" part opens a door for some ambiguity.

> At the time that such a revision 
> is undertaken all "free floating" concepts associated with the 
> name of the higher taxon (eg. species concepts that are only 
> linked to the name of the genus being revised) should be taken 
> into consideratoin and be either explicitly included or excluded 
> from the new version of the higher taxon concept.

I agree that this is the "cleanest" way to anchor miscellaneous species descriptions to an explicit genus concept; but I don't think it necessarily forces us to abandon the approach that the authors of miscellaneous species decriptions had in mind a genus concept, but simply failed to provide adequate information to unambiguously map it to other, previous concepts of the same name. In such cases, my preference would be to create a new TaxonConcept instance for the genus name AccordingTo the author of the new species name.  The definitive Relationships of this genus concept would be "includes" the new species.  If the authors of the new species failed to provide information about which specific genus concept they used, then by default a RelationshipAssertion could be created to map the Genus AccordingTo SpeciesAuthor TC instance to the Nominal TC instance for that genus name.

Something like this:

<Dataset>
  <Publications>
    <Publication id="P1">
      <PublicationSimple>Smith 1950</PublicationSimple>
    </Publication>
    <Publication id="P2">
      <PublicationSimple>Jones 1970</PublicationSimple>
    </Publication>
  </Publications>
  <TaxonConcepts>
    <TaxonConcept id="TCo1" type="original">
      <Name type="scientific">
        <NameSimple>Aus Smith 1950</NameSimple>
        <NameDetailed id="TCn1"/>
      </Name>
      <AccordingTo>
        <AccordingToSimple>Smith 1950</AccordingToSimple>
        <AccordingToDetailed>
          <AuthorTeam>Smith</AuthorTeam>
          <Date>1950</Date>
          <PublishedIn ref="P1"/>
        </AccordingToDetailed>
      </AccordingTo>
      <Relationships>
        [Definitive relationships of Aus Smith SEC. Smith]
      </Relationships>
    </TaxonConcept>
    <TaxonConcept id="TCn1" type="nominal">
      <Name type="scientific">
        <NameSimple>Aus Smith 1950</NameSimple>
        <NameDetailed>
          <Label>Aus Smith 1950</Label>
          <NomenclaturalCode>ICZN</NomenclaturalCode>
          <Rank>genus</Rank>
          [etc...]
        </NameDetailed>
      </Name>
    </TaxonConcept>
    <TaxonConcept id="TCi2" type="incomplete">
      <Name type="scientific">
        <NameSimple>Aus Smith 1950</NameSimple>
        <NameDetailed id="TCn1"/>
      </Name>
      <AccordingTo>
        <AccordingToSimple>Jones 1970</AccordingToSimple>
        <AccordingToDetailed>
          <AuthorTeam>Jones</AuthorTeam>
          <Date>1970</Date>
          <PublishedIn ref="P2"/>
        </AccordingToDetailed>
      </AccordingTo>
      <Relationships>
        <Relationship type="includes">
          <ToTaxonConcept ref="TCo3"/>
        </Relationship>        
      </Relationships>
    </TaxonConcept>
    <TaxonConcept id="TCo3" type="original">
      <Name type="scientific">
        <NameSimple>Aus xus Jones 1970</NameSimple>
        <NameDetailed id="TCn3"/>
      </Name>
      <AccordingTo>
        <AccordingToSimple>Jones 1970</AccordingToSimple>
        <AccordingToDetailed>
          <AuthorTeam>Jones</AuthorTeam>
          <Date>1970</Date>
          <PublishedIn ref="P2"/>
        </AccordingToDetailed>
      </AccordingTo>
      <Relationships>
        <Relationship type="is included in">
          <ToTaxonConcept ref="TCi2"/>
        </Relationship>
      </Relationships>
    </TaxonConcept>
    <TaxonConcept id="TCn3" type="nominal">
      <Name type="scientific">
        <NameSimple>Aus xus Jones 1970</NameSimple>
        <NameDetailed>
          <Label>Aus xus Jones 1970</Label>
          <NomenclaturalCode>ICZN</NomenclaturalCode>
          <Rank>species</Rank>
          [etc...]
        </NameDetailed>
      </Name>
    </TaxonConcept>
  </TaxonConcepts>
  <RelationshipAssertions>
    <RelationshipAssertion type="congruent" id="RA1">
      <AccordingTo>
        <AccordingToSimple>Pyle 2004</AccordingToSimple>
      </AccordingTo>
      <FromTaxonConcept ref="TCi2"/>
      <ToTaxonConcept ref="TCn1"/>
    </RelationshipAssertion>
    <RelationshipAssertion type="congruent" id="RA2">
      <AccordingTo>
        <AccordingToSimple>Pullan 2005</AccordingToSimple>
      </AccordingTo>
      <FromTaxonConcept ref="TCi2"/>
      <ToTaxonConcept ref="TCo1"/>
    </RelationshipAssertion>
  </RelationshipAssertions>
<Dataset>

Summary:

TaxonConcepts:
TCo1=Aus Smith SEC. Smith (original)
TCn1=Aus Smith SEC. [Nobody] (Nominal)
TCi2=Aus Smith SEC. Jones (incomplete)
TCo3=Aus xus Jones SEC. Jones (original)
TCn3=Aus xus Jones SEC. [Nobody] (Nominal)

RelationshipAssertions:
RA1=Pyle asserts that Aus SEC. Jones is un-defined (name-only)
RA2= Pullan asserts that Aus SEC. Jones is congruent to Aus SEC. Smith

This shows how name-objects could be represented as Nominal-type concepts, and how an ambiguously defined genus circumscriptions (e.g., Jones' use of the genus "Aus" when describing the new species "Aus xus") can be established as an implied genus concept, which Pyle felt was undefinable (RA1), but Pullan felt could be confidently mapped as congruent to Aus Sec. Smith (RA2).

I don't think I used the "incomplete" TC type correctly here.  I believe that type was intended for cases where Jones *did* provide a good defnition for his concept of Aus, but that the definition was simply not captured by the TCS data provider. This got me thinking about the six defined types of concept instances.  They are:

1. Original
2. Revision
3. Incomplete
4. Aggergate
5. Nominal (=Nomneclatural)
6. Empty

I think 3-6 are fine as they were originally defined.  However, 1 & 2 were established in the paradigm that names were not to be treated as separate objects from concepts.  Thus, "Original" roughly equates to "Concept as defined in the original description of a new name", and "Revision" roughly equates to "Subsequent use of a previously-created name for a novel defined Concept".

If, indeed, TCS ends up treating names as distinct objects (either as top-level objects, or as Nominal-type TaxonConcept instances), then I would propose replacing "Original" and "Revision" with "Defined" and "Undefined".

Thus, in the instance example above, TCo1 and TCo3 would be of type "defined", and TCi2 would be more properly represented as "undefined" (rather than "incomplete").

> I think that makes sense ;-)

I only wish I could make the same claim about my own response; but on re-reading, I cannot truthfully do so. :-)

> BTW I can't remember but does the current TCS have a means of 
> flagging the geographical scope of a classification? eg. for 
> representing a classification presented in a flora.

I don't think so...but it might be useful. Also, this comment reminded me of something else that I've wrestled with: Using geographic distributions as another mechanism for defining concept circumscriptions.  Essentially, it's a fourth way of defining the boundaries of a TaxonConcept (the first three being Relationships, SpecimenCircumscriptions, and CharacterCircumscriptions.  But then again, the practice of using biogeography as a factor in defining taxon  circumscriptions (as I did conveniently in my example involving Hawaii & Johnston Atoll populations) might be viewed by taxonomic purists as circular reasoning.

I should now get back to work...

Aloha,
Rich