BINs (Barcode Index Numbers) and OTUs (Operational Taxonomic Units) as hypotheses

[sharifX]

Following up on the DiSSCo Transition Workshop discussions, I have created two examples to explore how BINs can be described as species hypotheses within the TCS framework. Once a formal taxonomic assignment is made, we can simply update the taxonName and add an interpretation link. This approach allows for tracking historical changes and sharing updated classifications without losing past data.

Two examples:

1. BOLD:ACZ4050 has no species level identification, BOLD:AAH4793 is its nearest neighbour.
2. BOLD:AAA1039 has species level identification ("Dasyvesica lophotalis"), BOLD:ACZ4050 is its nearest neighbour.

New Considerations

• Should there be a "SpeciesHypotheses" type (for BINs with no species level identification), or should these be described as TaxonConcepts?
• Introduced a new property: interpretedAs
• Is isIncludedIn make sense here?

Example BOLD:ACZ4050

{
  "@context": "https://raw.githubusercontent.com/tdwg/tcs2/master/examples/context.jsonld",
  "type": "SpeciesHypotheses",
  "dcterms:title": "Pococera sec. BOLD:ACZ4050",
  "accordingTo": "https://portal.boldsystems.org/bin/BOLD:ACZ4050",
  "taxonName": {
    "id": "https://portal.boldsystems.org/bin/BOLD:ACZ4050",
    "type": "TaxonName",
    "taxonNameString": "Pococera",
    "dwc:taxonRank": "genus",
    "dwc:higherClassification": "Arthropoda | Insecta | Lepidoptera | Pyralidae | Epipaschiinae"
  },
  "nearestNeighbors": [
    {
      "type": "TaxonConcept",
      "dcterms:title": "Nearest neighbor BOLD:AAH4793",
      "accordingTo": "https://portal.boldsystems.org/bin/BOLD:AAH4793",
      "dwc:relationshipOfResource": "nearest neighbor by barcode clustering"
      "taxonName": {
        "id": "https://portal.boldsystems.org/bin/BOLD:AAH4793",
        "type": "TaxonName",
        "taxonNameString": "Pococera",
        "dwc:taxonRank": "genus",
        "dwc:higherClassification": "Arthropoda | Insecta | Lepidoptera | Pyralidae | Epipaschiinae"
      }
    }
  ],
  "interpretedAs": [
    {
      "type": "TaxonConcept",
      "dcterms:title": "Pococera sp. sec. GBIF",
      "accordingTo": "https://www.gbif.org/species/10434557",
      "taxonName": {
        "id": "https://www.gbif.org/species/10434557",
        "type": "TaxonName",
        "taxonNameString": "Pococera sp.",
        "dwc:taxonRank": "UNRANKED"
      }
    }
   
  ]
}

Example BOLD:AAA1039

{
  "@context": "https://raw.githubusercontent.com/tdwg/tcs2/master/examples/context.jsonld",
  "type": "TaxonConcept",
  "dcterms:title": "Dasyvesica lophotalis sec. BOLD:AAA1039",
  "accordingTo": "https://portal.boldsystems.org/bin/BOLD:AAA1039",
  "taxonName": {
    "id": "https://portal.boldsystems.org/bin/BOLD:AAA1039",
    "type": "TaxonName",
    "taxonNameString": "Dasyvesica lophotalis",
    "dwc:taxonRank": "species",
    "dwc:higherClassification": "Arthropoda | Insecta | Lepidoptera | Pyralidae | Epipaschiinae"
  },
  "interpretedAs": [
    {
      "type": "TaxonConcept",
      "dcterms:title": "Pococera sp. sec. GBIF",
      "accordingTo": "https://www.gbif.org/species/1870580",
      "taxonName": {
        "id": "https://www.gbif.org/species/1870580",
        "type": "TaxonName",
        "taxonNameString": "Dasyvesica lophotalis",
        "dwc:scientificNameAuthorship": "Hampson, 1906",
        "dwc:taxonRank": "species"
      }
    }
  ],
  "isIncludedIn": {
    "id": "https://www.catalogueoflife.org/data/taxon/6CDYB",
    "type": "TaxonConcept",
    "accordingTo": "https://www.catalogueoflife.org",
    "taxonName": {
      "id": "https://www.catalogueoflife.org/data/taxon/6CDYB",
      "type": "TaxonName",
      "taxonNameString": "Dasyvesica lophotalis",
      "dwc:scientificNameAuthorship": "Hampson, 1906",
      "dwc:namePublishedIn": "Catalogue of the Lepidoptera Phalaenae in the British Museum. Volume 4",
      "dwc:namePublishedInYear": "1906"
    },
    "dwc:taxonRank": "species",
    "dwc:higherClassification": "Arthropoda | Insecta | Lepidoptera | Pyralidae | Epipaschiinae | Dasyvesica"
  },
  "nearestNeighbors": [
    {
      "type": "TaxonConcept",
      "dcterms:title": "Nearest neighbor BOLD:ACZ4050",
      "accordingTo": "https://portal.boldsystems.org/bin/BOLD:ACZ4050",
      "dwc:relationshipOfResource": "nearest neighbor by barcode clustering",
      "taxonName": {
        "id": "https://portal.boldsystems.org/bin/BOLD:ACZ4050",
        "type": "TaxonName",
        "taxonNameString": "Pococera",
        "dwc:taxonRank": "genus",
        "dwc:higherClassification": "Arthropoda | Insecta | Lepidoptera | Pyralidae | Epipaschiinae"
      }
    }
  ]
}

[nielsklazenga]

Thanks for thinking with us @sharifX. This is great.

Regarding your question whether there should be a 'SpeciesHypothesis' type or whether Species Hypotheses are Taxon Concepts, I think we can have our cake and eat it. Species Hypotheses are Taxon Concepts, but we can look into categories (or types) of Taxon Concepts. See Franz & Peet, 2009, Table 1. There is a 'type' attribute on the <TaxonConcept/> element in TCS 1, but some members of the Task Group were not keen on (or not interested in I should say) this, so we have left it out for now. I think that having an entirely different type of taxon concept like the Species Hypothesis (the other different types could all be categorised as 'Relational Concepts') is a good reason to bring it back.

In the standard itself these categories would be SKOS Concepts, but in RDF they would be subtypes of the TaxonConcept (at least that would make sense to me), so your example would stay the same.

As for your other two questions, I have used isIncludedIn myself in TaxonConcept-example-6, so yes, I think that makes sense. Likewise, interpretedAs could be mapped to isCongruentWith and nearestNeighbors to some other mapping property. However, I am all for having terms with more specific meaning, as long as they can be strictly defined, which people can map to the more general mapping terms if they choose to do so (and we can suggest those mappings).

Are you happy for me to convert this issue to a Discussion? Then we can have issues for proposed terms that come out of it.

Incidentally, an IRI equivalent for dwc:higherClassification is on my wish list for TCS. It is not in TCS 1 though, so we'll add that later.

[nielsklazenga]

Regarding 'Operational Taxonomic Unit' (OTU), I know that in OpenBiodiv-O (Senderov & al., 2018) OperationalTaxonomicUnit is a superclass of TaxonomicConcept, but in my mind – and I think according to the original definition of 'Operational Taxonomic Unit' (which was in Sokal & Sneath, 1963, I think) – OTUs and Taxon Concepts are different things, although it is not always clear where one ends and the other begins.

For example, in Urmas's presentation, which was between our two presentations at the DiSSCo Transition Workshop, the Individuals he talked about first are OTUs, but Taxon Hypotheses and Species Hypotheses are Taxon Concepts. If it had stayed with Individuals and those Individuals would have made it into GBIF (or into your examples), the Individuals would be Taxon Concepts too (at least in TCS).

[sharifX]

Hi @nielsklazenga, thanks for the reply. Yes, I’m happy to continue this as Discussion.

You make a good point regarding the boundary between OTUs and Taxon Concepts. Perhaps this distinction can be clarified in the context of how BINs are used and implemented.

Here’s how I’m currently thinking about this:

• OTUs represent clusters of entities grouped based on shared characteristics, often in the absence of formal taxonomic classification.

• BINs are similar to OTUs in that they group DNA barcodes based on clustering. However, BINs do not claim to represent formal taxonomic entities. That said, they can be linked to such entities, and in this sense, they can also be interpreted as Taxon Concepts?

We likely need input from the BOLD/iBOL community to refine this understanding further.