LinneanCoreSomePrivateConceptDefinitions

GregorHagedorn - Fri Dec 10 2004 - Version 1.7
Parent topic: LinneanCoreDefinitions
Gregor: This is a rough attempt to express my thinking about the different kind of taxon concepts used in biological taxonomy. I am not a concept expert, but I notice that we have difficulty communicating about different kind of concepts - perhaps this attempt of definition can help to find agreement. Other terms than the ones used here may be more appropriate. However, at the moment I cannot correlate these "types-in-my-head" with any of the concept types used by Nico Franz and in TCS (especially the WIKI page http://www.soc.napier.ac.uk/tdwg/index.php?pagename=ModellingNamesAndConcepts and the documentation file associated with TCS 0.8). I am not good at writing concise definitions and usually say too much - please excuse.

I am not claiming I have solutions for the concept debate. My motivation is that I believe it an error to throw the separation between the well understood and analysed type concepts used in the biological nomenclatural codes and the much less well understood and generally not applied hierarchy or circumscriptions concepts over board.

Also I hope nobody reads this :-), since I break my promise given elsewhere to stay away from concept discussion :-). Seriously, this is an attempt to make my thinking understandable to others, and I welcome comments, but not on the expense of getting distracted from the much more important LC core issues.

I'm afraid I couldn't resist reading this - but that's what you wanted really isn't it??? - Trevor -- Sure, your comments are most welcome to help me develop my thoughts and better communicate! - Gregor

Type Concept: A concept defined through a single point in the "biodiversity space", the type specimen. Based on priority of publication, a name string will be fixed to this point, and reversely. The type specimen may be referred to directly or indirectly (e.g. when a name is based on another name as in new combinations, or higher taxa typified through other names). The rules of nomenclature in bacteriology, botany, and zoology are based on this type concept. The concept itself has no value when identifying unknown organisms, but it serves as a basis for the circumscription concepts layered on top of it. The advantage of this concept is that the available identifier name-space for taxon names can be governed by strict rules and some stability in the association between circumscription and name-string-with/without-authors can be achieved. (I believe this is what is sometimes called nomenclatural "name-object".) -- Discussion

Character circumscription Concept: A concept defining a classification method for unidentified biological individuals. The circumscription uses characteristical properties that are inherited (usually genetically) from parents to children. It is therefore valid over extended periods of time. Circumscription Concepts may exist for taxa at all ranks. -- This is the only concept used in identification. This kind of taxon concept is often loosely referred to as a "taxon concept" and contrasted to a name. The result of multiple incongruent taxon circumscriptions on knowledge gathered about organism included in these circumscriptions is difficult to analyze (except where voucher specimens can be reanalyzed under both concepts). Making such comparisons operational is the main point in dealing with taxon concept logic. -- Discussion

Specimen Set Concept: A circumscription may also consist of multiple specimens listed as being studied ("Specimen circumscription", especially studied in the Prometheus I model by Jessie). Non-type specimens other then the type I believe are examples rather than types (alhtough they may have special nomenclatural importance for lecto/neotypification, or illegitimate superfluous names). --
Note: Always having to use "character circumscription concept" and "specimen circumscription concept" seems to be difficult. It may simplify communication if we could agree to use "circumscription concept" as a shorthand for "character circumscription". Since the specimen circumscription concept is a specimen set, I suggest to use "Set concept" when referring to these.

Hierarchy Concept: Or perhaps enumerated concept instead? A tree of concepts based on other concepts. This has no value for identification. If the result of an identification is directly a genus or family, the identification must be based on a circumscription concept, not on an application of the hierarchy knowledge. Different hierarchies superimposed on knowledge associated -- Discussion

-- Main.GregorHagedorn - 19 Nov 2004

Comments: TCS does not have a problem representing these different examples of Concept Definitions, or indeed mixtures of parts of any of them. It is non-prescriptive in that it aims to capture any format or information in a taxon definion. Because of the rules of nomenclature I would have thought that any concept with an atttached scientific name includes minimally a Type as part of its definition (at least implicitly).

I think where you are leading in the following discussion is that you would like to have separate types of concept definitions re-usable and recombinable - to create new emended/composite definitions. (We would definitely see these as still being sec. A New Author.) This is definitely something that would be very useful for future working practice in taxonomy - but not really an issue yet where none of these modular definitions exist in any databases!

'In the meantime' it may be useful for TCS to have additional concept to concept relationships types such as : 'shares character circumscription', 'shares taxon circumscription', 'shares specimen corcumscription' (type specimen relationships are probably already covered in the nomenclatural relationships). Would this be a useful addition to TCS? - so that you could make a new concept Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999 sec. Scholler, Hagedorn, & Rubner 1999, which specifically includes in its definition 'shares character circumscription of' Monacrosporium bembicodes (Drechsler) Subram. 1963. sec. Hagedorn & Scholler 1999; AND 'shares character circumscription of' Monacrosporium bembicodes (Drechsler) Subram. 1963 sec. Rubner 1996;

-- Main.TrevorPaterson 05.12.04

Gregor: I feel my intention well captured when you say "would like to have separate types of concept definitions re-usable and recombinable - to create new emended/composite definitions."

However, I disagree with "very useful for future working practice in taxonomy - but not really an issue yet where none of these modular definitions exist in any databases!". Most importantly, I believe this is what taxonomists do everyday. They freely combine name changes and character circumscription changes and hierarchy changes from different publications without a problem. If I have a good morphological character circumscription for a taxon, someone else moves the name into a different genus (Ustilago violacea to Microbotryum) based on chemical character, and yet someone else places Microbotryum no longer in the smut fungi, but close to the rust fungi based on DNA sequences, I can effortless base my work on combination of all three publications, creating a concept that has never been explicitly published, but is simply the state of scientific knowledge. b) also, you know the other models better than I, but our own
DiversityTaxonomy models do make these distinctions, and the Myxomycete and Lichen Name lists are to be delivered to GBIF from those lists. -- Gregor
- Yes - this is what taxonomists do on a daily basis - but if they need to communicate their new concepts they need to publish them somehow.... TCS could provide a wrapper to do this - especially if it were expanded to haverelationships to inherit specified parts of definitions form other concept. -- Main.TrevorPaterson
- Can you give me an example, where such a use of multiple concept aspects would ever be published - other than in a monograph which would soon be outdated again? I believe it would not be publishable because of lack of novelty- you can not simply publish a paper saying: "there have been 4 pieces of information, and I subscribe to all 4 of them". I think combining hierarchy, nomenclatural change and character circumscription is normally only used (and cited by citing multiple papers) but not published. -- Gregor, 10. Dec. 2004

You talk about "mixtures" of types. Do you mean that the publication is always one atomic concept, and that different aspects according to these types are expressed? Whether atomic or not is probably a question of taste. My point is that TCS fails to express the differences, i.e. it does not express which of these aspects is vouchered by a concept. That is what I understood to be the meaning of the TCS concept type.

We were not thinking of 'typing' concepts on the basis of their content (character definition type, circumscription type) - this would be apparent form what data was represented in a TCS object. The typing of concepts that we have provided is: original, revision, nominal, aggregate, incomplete.
from our documentation:
$ Original: The first representation/original publication of a taxon concept/name use. In other words: concepts where a new name has been created. The definition of the taxon concept will be variable depending on where and when it was published. These concepts act as the foundation on which current taxonomic concepts are based. $ Revision: A concept definition presented as part of a taxonomic revision. Usually the results of a re-classification (e.g. combining or splitting) of existing concepts where a previously used name is redefined in some way. $ Incomplete: An incomplete record of a concept. This should be used if it is known (or suspected) that not all information about the concept is contained in the data source. $ Aggregate: A collection of TaxonConcepts grouped together under a name of utility by a user/author. It needs to include links to all of the TaxonConcepts it covers explicitly stated. $ Nomenclatural(now 'Nominal'): A concept holding purely nomenclatural information, any relationships held will be to similar nomenclatural concepts. It (implicitly) covers all concepts that ever used the particular name. A special case of this is a vernacular concept. We see original concepts as being those of most importance ot nomenclaturists - as they record the first use and definition of a name: when this same name is used subsequently with different definition we represent this with a revision concept. Nominal concepts only hold a name - with no definition - this allows names to be used as fuzzy concepts if no definition is known (or required?!). Incomplete concepts are incomplete revision or original concepts which are made as placeholder because the information is not really available - but we need to create a relationship to this concept. Aggregate concepts wopuld allow someone to make their own useful concepts - such as Bob Peet's view of everything covered by the common name 'Hickory'. -- Main.TrevorPaterson - 10 Dec 2004

* I did read these. As I said in the session in Christchurch, I feel this is mixing different issues. I can not argue very well, however, because I do not understand these types or "fill them with meaning" for me. That is what I meant that a definition is needed (rather than a description). The biggest problem in a way is that many refer to the term concept and name. The way I understand TCS every kind of publication name use is a concept. That however, makes it difficult to understand the difference between original and revision. Perhaps you need to clarify whether you understand name including letter change variants, and including the "sec. author" or not. Original: For botanists a combination is an original nomenclatoral act, for zoologists not. So would a comb. nov. be an Original in botany and a Revision in Zoology? What about a validation of a name - no concept is expressed, yet it is a revision of another earlier and illegitimate (bot.) name (and under TCS, a concept). Revision: Is it a revision if I place a genus into a different family or order? Or only if I place the species into a different Genus? Or only if I change the type specimen through neotypification, but leave the character circumscription fully intact and identical? Incomplete: What is the data source - is it the source database from which TCS xml is generated or is it a publication? If the latter, when would ever be all the information from the publication be present in the nomenclator or taxonomic dataset? Is this not unlikely, not the least due to copyright restrictions? I would need guidelines to say when is something sufficiently incomplete to treat it as a different type. And why can not be an Original, Revision, and Aggregate all be incomplete? This seems to mix different levels into one "typology". -- Aggregate: what makes an aggregate any different from publishing a new genus aggregating multiple species? Nominal: it seems to me that it does not hold any nomenclatural information (place of publication, holotype, basionym) they way you explain it. -- Gregor, 10. Dec. 2004

Regarding these TCS concept types: I believe it would be helpful to define them much better than in the documentation available at Christchurch (0.80 Word document) - to me these types are not intuitive! - and provide a reasoning why they exist (use cases, why is a consumer interested in them, which algorithimic or human reasoning depends on them?). This is part of the problem why I present my "concept types" - which then as you point out are aspects of a name publication - here.
You say "Because of the rules of nomenclature I would have thought that any concept with an atttached scientific name includes minially a Type as part of its definition (at least implicitly)." - I agree fully - can you help me pointing out where I say something that contradicts this? I probably expressed something confusingly. Also, how does this relate to TCS? Is there a mechanism that any TCS concept should have a type? I rather think that some types are not expressed in TCS, especially iconotypes.
- We did ask everyone if they could comment/ammend the list types ( and relationship types ) we dont have iconotype as a type label included in the enumerated list - but could happily add it. [ list is currently: epitype, holotype, isotype, lectotype, neotype, non-type, paratype, syntype, isosyntype, hapantotype, paralectotype, paratype, syntype,type]. ( we did at one time allow a class attribute on specimen to record whether it was a physical specimen, a picture etc - but we have dumped this). -- Main.TrevorPaterson - 10 Dec 2004

Thanks!

-- Gregor Hagedorn, 10. Dec. 2004

(Gregor, continued:) Example scenario to muddle the above:

I take an example from my own work, that also shows my confusion about how to apply the idea that there is a 1:1 correlation between a name and a concept in practice.

Dactylella bembicodes Drechsler 1937 was described using an illustration as an iconotype (i.e. no holotype specimen exists) and with a diagnostic description
Subramaniam recombined it to Monacrosporium bembicodes (Drechsler) Subram. 1963
Mekhtieva recombined it to Golovinia bembicodes (Drechsler) Mekht. 1967
Zachariah & Insell 1979 describe nuclei of the trapping organs, accepting Monacrosporium bembicodes (Drechsler) Subram. 1963, but mentioning synonyms not accepted by other authors
- Character circumscription information can not be compared... I have no idea what to do with this - except that under a fuzzy type concept merging I can merge this property information with other information - while warning users that this may not be exact.
Rubner 1996 accepts Monacrosporium bembicodes (Drechsler) Subram. 1963, and provides an emended description of the species based on Drechsler's original description. She further reproduces the original illustration of Drechsler.

* Is this Monacrosporium bembicodes (Drechsler) Subram. 1963 emend. Rubner 1996? There is both a type concept and a well defined (and thus operationally useful) character circumscription concept present. * Rubner has also a "specimen set concept" with two cultures studied. Although not mentioned it is implicit that the iconotype (the illustration that was even reproduced) also belongs to this set. However, Rubner explicits mentions that the first culture did not result in any information (it neither sporulated nor produced the trapping organs - it could thus have been almost any fungus) and the second produced only spores (it thus could be several species of Monacrosporium and needs not be Monacrosporium bembicodes). As a result another study using the same species may or may not correspond with her concept. * Rubner has also a hierarchy concept (Fig. 19) -- which later turned out to be wrong.

Scholler, Hagedorn, & Rubner publish a molecular phylogenetic analysis of the anamorphic classification. Based on the phylogram they create Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999.
- They provide no diagnostic description of their own. Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999 sensu Scholler, Hagedorn, & Rubner 1999 makes no sense to me - it is a "descriptio nudum" (new term :-)!)
- The nomenclatural rules require that the protologue must be cited. Does this mean that the character circumscription concept should better be cited as Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999 sensu Drechsler 1937?
- Since one of the coauthors is Rubner and Rubner 1996 is cited, it would be reasonable to guess that their real morphological concept is based on Rubner 1996 (and indeed it is!) Does this mean that the character circumscription concept of Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999 sensu Scholler, Hagedorn, & Rubner 1999 should also be cited as Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999 sensu Drechsler 1937
- By basing their nomenclatural results on molecular phylogeny published in Hagedorn & Scholler 1999 they refer to molecular descriptive data. The bembicodes sequence itself was first published and submitted to
  GenBank by Liou & Tzean 1997. By adding many other sequences Hagedorn & Scholler 1999 were able to create a meaningful DNA alignment (esp. the outgroup assumptions in Liou & Tzean 1997 were wrong). One could cite this as "sensu Liou & Tzean 1997 emend. Hagedorn & Scholler 1999".
  - Is this the character circumscription concept of Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999 sensu Scholler, Hagedorn, & Rubner 1999 -- or of -- Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999 sensu Hagedorn & Scholler 1999?
  - Note that according to ICBN a name may not be used prior to its publications, so the name used in Hagedorn & Scholler 1999 is really Monacrosporium bembicodes (Drechsler) Subram. 1963.
Baral, Weber & Hagedorn (manuscript) will treat the species as a member of the teleomorph family Orbiliaceae.

If I am now receiving an unidentified culture and after adding nematodes I find the typical trapping organs of bembicodes, and I then sequence ITS DNA to obtain a confirmation I would end up with:

Type concept under ICBN: Drechslerella bembicodes (Drechsler) Scholler, Hagedorn, & Rubner 1999
Character circumscription concepts:
- Morphology: sensu Drechsler 1937, emend. Rubner 1996
- Molecular: sensu Liou & Tzean 1997 emend. Hagedorn & Scholler 1999
"Specimen set concept" not applicable
Hierarchy concept sensu Baral, Weber & Hagedorn.

Without having different names for the various concepts I can not express that I base an identification morphologically on one publication (Rubner 1996), but know that both the type concept and the hierarchy concept of this publication are to be rejected. Note that within certain bounds the type, character circumscription, and hierarchy concepts can be freely combined with each other without invalidating any of the conclusions of the other concept.

-- Gregor Hagedorn, 26. Nov. 2004