Botanical name modeling

There have been several issues related to modeling botanical nomenclature, and several conversations on Matrix, and I thought I would try to write down the outstanding concerns, as I understand them. A Github issue seems as good a place as any to put this, for future reference. No immediate issue resolution is expected.

Perhaps the best way to start is with the minimal elements of a data model that I see as being necessary:

Create a uninomial name (a) with original parent (A) and original author (X) → A a X (i)
Optionally allow for Y to validly publish an invalidly published original name by X, thus → A a (X ex Y) (ii)
Create any number of new combinations (e.g., species under different genera, or shift of rank within a genus). Each of these is a correct (sensu ICN) name → B a (X ex Y) Z, A b var. a (X ex Y) Q, etc. (iii)
Different taxonomists may prefer to accept different combinations, moving the other combinations into homotypic synonymy, so some method is needed to specify the accepted status, in the opinion of the person running this project in TW → Accepted: B a (X ex Y) Z; ≡ A a (X ex Y), ≡ A b var. a (X ex Y) Q (iv)
Each of these combinations is an effective new name in TW, and can be used to create OTUs. (v)
In an nomenclatural export (e.g., CoL dump), for each combination the original basionym would be given, as well as the status (accepted or homotypic synonym). (vi)

This can be visualized as:

A a X --> A a X ex Y --+---+---------> A a X ex Y             || synonym
                        \   \
                         \   +-------> A b var. a (X ex Y) Q  || synonym
                          \
                           +---------> B a (X ex Y) Z         || accepted

Here is a real world example:

Salix pseudomonticola C.R.Ball (basionym) (i)
Salix barclayi var. pseudomonticola (C.R.Ball) Kelso (new combination) (iii)
Accepted in Flora of North America (2010), by G. W. Argus: Salix pseudomonticola C.R.Ball. In the nomenclatural browser, this would be shown as: Salix pseudomonticola C.R.Ball ≡ Salix barclayi var. pseudomonticola (C.R.Ball) Kelso (iv)
In our work, we use a Taxonomic Concept, aka OTU, of Salix pseudomonticola C.R.Ball sec. Argus 2010 (v)

Based on my exploration of TW - but not a comprehensive understanding of it! - several of the above capabilities seem not to be present:

The existing implementation allows a single preferred (= accepted?) new combination to be chosen as the preferred name, but not the basionym. (cf. (iv))
To indicate an invalidly published name to be validly published (the ex relationship) two identical (uninomial, rank, parent) names must be created and linked with name relationship ‘validly published as’ (cf. (ii), #3315)
Once a combination has been established, in TW one can only create an OTU using the current preferred name, not the basionym. For our Taxon Concept modeling we need to indicate which exact taxonomic name is used (c.f. TDWG’s TCS2). (cf. (v))

In essence, It seems that TW offers a linear path from basionym to one of the new combinations, with a single preferred name to use for OTUs. Currently, in order to model branching a graph such as indicated above, it is not possible to just create a single protonym with combinations, but one must follow one of two approaches (omitting the complication of the ex):

for each name (A a, B a, A b var. a) create a new protonym A a, then create the combinations for B a and A b var. a, and then indicate synonymy with relationship homotypic synonym between A a and B a, and A b var. a and B a, or,
for each name (A a, B a, A b var. a) directly create new protonyms without combinations, using verbatim authorship to make the parentheses ((X) Q, (X) Z), and then indicate synonymy with relationship homotypic synonym between A a and B a, and A b var. a and B a, and is basionym of relationships from B a to A a, and A b var. a to A a.

However the creation of these additional protonyms seems to confuse the combination logic engine and the various independently created protonyms get ‘tangled up’ (demonstration offered, when we meet).

If there were to be no further discussion/development of this, approach (2) would be the way to go, being simpler, and completely avoiding the combination logic engine. However, this seems a shame, and currently does not export basionym relationships in the CoL dump.

None of this is a criticism of TW as it stands - these are just barriers I currently see for implementing a botanical monograph in TW. Thanks as ever for this powerful informatics tool.

SpeciesFileGroup / taxonworks

Botanical name modeling #4088