TDWG2026
Home / Talks / ODIN
talk · Tuesday 22 September · 12:02–⁠12:16 · ODIN

Transitivity Failures in Taxonomic Name Resolution: Implications for Botanical Dataset Construction

Adam Richard-Bollans · Data Quality: From Standard to Practice

Recording time 3:31:26–3:44:33Open on Vimeo ↗

Summaries and key moments are machine-produced. Check the recording for the speaker’s exact words.

The short versionAlways keep original verbatim names so datasets can be re-resolved against the latest taxonomy.

Overview

What this was about

Adam Richard-Bollans shows that resolving names to accepted names across taxonomy versions is not transitive: a name x resolved to y in one version may not resolve to where y resolves in a later version, e.g. because heterotypic synonyms are split. Comparing World Checklist of Vascular Plants versions (2022 vs 2025), about 13,000 of ~1.4 million names (≈1%) disagree, about half at species level, implying ~0.5% species-level error in datasets that kept only accepted names, rising to 0.7% by June 2026. Errors cluster in western and southern Europe (France, Italy, Spain).

Why it matters. Many secondary datasets discard original names after resolution; this quantifies the silent error that introduces and shows it grows over time.

Key ideas

In the room

  • Most aggregated botanical datasets resolve names to accepted names and often discard verbatim names.
  • Transitivity: x→y in v1 and y→z in v2 should imply x→z in v2; heterotypic synonym splits break this.
  • Example: Panicum nitidum resolved to Dichanthelium dichotomum in WCVP v10 (2022), but to Dichanthelium nitidum in v14 (2025).
  • ~13,000 of ~1.4 million names disagree (≈1%); about half at species level, some at genus level.
  • Regional clustering of non-transitive resolutions in western/southern Europe.
  • Error grows with time since the original version (0.5% in 2025 to 0.7% by June 2026 for a 2022 dataset).
  • World Flora Online results are more severe; resolving to a different taxonomy likely increases error further.
Jump in

Notable moments

In their words

Transcript

Automatically generated captions can contain mistakes, especially in names and technical terms. Times are relative to the room recording.

Read the transcript ↓
Loading transcript…
Keep exploring

Related talks.

See the map ↗
talk · 22 Sept · 11:15–⁠11:30 · SAL C

A knowledge organisation system for taxonomic data

Separating taxon data into layers with immutable taxon concepts at the centre gives a practical path to implementing TCS and fixing checklist quality problems.

Niels Klazenga
talk · 24 Sept · 16:15–⁠16:30 · SAL B

From 'Lost in Translation' to Fit-for-Purpose Data: Why Ecologists and Conservation Practice Still Need Groupings and Uncertainties

Biodiversity data systems must be able to record identification uncertainty and taxonomic groupings, or ecologists face a choice between false certainty and lost ecological information.

Bernhard Kløw Askedalen, Natalie
talk · 22 Sept · 16:15–⁠16:25 · SAL A

A closer look at steps needed to extract research-ready biodiversity data from legacy literature

Legacy literature data can only be unlocked by combining better OCR/NLP with taxonomic rigour that accounts for synonymy, shifting concepts, coreference and mixed languages.

Gerwin Kasperek
Programme abstract

Transitivity Failures in Taxonomic Name Resolution: Implications for Botanical Dataset Construction

Adam Richard-Bollans

A key decision when creating datasets about plants is whether to store the scientific names as originally recorded in a data source or to resolve them to currently accepted names. When a scientific name is resolved to an accepted name, and the original name is discarded, information is lost regarding the description and taxonomic classification used in the original identification. In practice, this may have limited impact if the original and accepted names consistently resolve to the same name in future taxonomic databases. However, this depends on the transitivity of name resolution across taxonomic database versions i.e. given names X, Y, Z and taxonomic database versions v1, v2 such that X resolves to Y in v1 and Y resolves to Z in v2, is it the case that X resolves to Z in v2? If this property does not hold in general then creating datasets by storing only resolved, accepted names may create unexpected outcomes with wide-ranging consequences for botany, and biological sciences more broadly, especially in an era where large scale data analyses are becoming increasingly prevalent in addressing many real-world challenges. Using the World Checklist of Vascular Plants and World Flora Online, we demonstrate that this transitivity fails for approximately 1% and 6% of plant names, respectively, and that the number of these discrepancies increases as new taxonomy versions are released. These results highlight the importance of storing verbatim names as found and resolving them to accepted names as close to the point of end-use as possible, using the most recent taxonomic database.

Submitted to the conference programme before the talk, so the talk itself may differ.