Discussion around specialised classes of `tern:Sample` and relationships to taxon identification

[edmondchuc]

For tern:Sample, we should stick to the same definition as sosa:Sample. Currently, tern:Sample is the class that represents all physical samples. Create a specialised class tern:MaterialSample to represent physical samples. See Darwin Core Terms MaterialSample.

Depending on the Surveillance protocol, other specialised classes of tern:Sample may be required (if they have specific properties). E.g. tern:AnimalSample?

[dr-shorthair]

Note that sosa:Sample includes non-material samples.

[dr-shorthair]

So if you need to limit it to material samples, then you need a sub-class, such as tern:Sample. What common properties of a tern:Sample are different to a sosa:Sample?

[edmondchuc]

This is what we currently have in feature/material-sample.

[dr-shorthair]

So I wonder if they should just be collapsed together?

[edmondchuc]

Do you mean to collapse tern:Sample and tern:MaterialSample together? I was thinking to keep it separate so that we can represent both material and non-material samples.

[dr-shorthair]

Yes, because you had said above that tern:Sample was physical samples. But

(a) if there is a TERN need for non-material-samples (b) the TERN class has additional attributes or constraints,

then the three step hierarchy is OK.

[edmondchuc]

The three-step hierarchy will remain for now. As discussed in the previous week (about transects and subplots), we do have a need to represent non-material samples. See https://github.com/ternaustralia/ontology_tern/pull/23 for a diagram example.

This has been merged into develop by https://github.com/ternaustralia/ontology_tern/pull/21.

[edmondchuc]

Notes from today's meeting:

• We may require a Specimen class as a subclass of MaterialSample. It will have a property for the accession number.
• MaterialSample already has the feature type property, but as @kitchenprinzessin3880 noted, we may need a material type property as well. At least for now, I don't think we receive the material type information in any of the datasets we have looked at.

[edmondchuc]

Small ontology to capture Identification and Taxon information and relate it back to MaterialSample was started a year ago in tern-sd. tern-sd basically reuses what is defined in Darwin Core Terms already and adds in a few relationships and properties to the classes of interest.

• Documentation - https://ternaustralia.github.io/ontology_tern-sd/
• Repository - https://github.com/ternaustralia/ontology_tern-sd

In the field, a plant voucher is collected from a plant individual (e.g. a tree). This plant voucher (material-sample/specimen) is a sub-sample of the tree. The field species name (observation) is made on the tree. This same name is also associated with the plant voucher. The plant voucher is sent to the local state/territory herbarium for name identification. The identified name is is then matched to a national authoritative source, for plants it is the APNI.

E.g. if the identified name is "Acacia Mill", then the name is matched to the APNI resource https://id.biodiversity.org.au/459697. Dumps of the APNI data can be found at https://biodiversity.org.au/nsl/services/export/index.

tern-sd defines a FieldTaxonName which is a result of an observation. It has a relationship to a StateIdentification (identification made by state/territory herbariuam) which has relationships to a StateTaxon and a NationalTaxon. The StateTaxon is the information determined by the local herbarium and the NationalTaxon is the name matched from the StateTaxon to whatever the national authoritative source is. In the case of vascular plants, the authoritative source is APNI. The StateIdentification does have a relationship to the MaterialSample via tern-sd:identifiedSample.

Next week I will provide some example data in the form of diagrams once I've reviewed it a bit further.

[edmondchuc]

Conclusion from today's meeting:

• Stick with the observations model for modelling taxon identification
• Some Darwin Core classes will continue to be used - MaterialSample and Taxon
• The act of identification is captured as an Observation
• The procedure determines how the observation was made (e.g. in the field, sent to the local herbarium or mapped to an external web service like APNI or AFD)
• Since the procedure determines how the observation was made, the observed property taxon identification can be used for all scenarios
• tern:Taxon is a subclass of tern:Value to cater for the triple tern:Observation sosa:hasResult tern:Value
  • Sometimes the result will be a tern:Taxon (herbarium result) and sometimes it will be a tern:Text (field result)
• @dr-shorthair will investigate how Darwin Core expects users to use the related classes for identification

Related links:

• https://ternaus.atlassian.net/wiki/spaces/DAWE/pages/2254667806/Specimen+taxon+identification
• https://dwc.tdwg.org/terms/
• https://ipt.gbif.org/manual/en/ipt/2.5/dwca-guide

[edmondchuc]

This needs to be documented preferably in the cookbook.