Taxonomic naming authorities from various providers including Catalogue of Life (COL), Global Biodiversity Information Facility (GBIF), International Union for Conservation of Nature (IUCN), and the National Center for Biotechnology Information (NCBI)
licenses:
Taxonomic name tables for 5 naming authorities, normalized to a common Darwin Core schema.
This is an archival release. It republishes taxadb 22.12 exactly as it was, so that analyses which pinned this version keep resolving. The files are byte-identical to the original release; they have deliberately not been corrected. A silently repaired snapshot would return different results to a script written against the real one, which is worse than leaving it as it was.
For current work use the newest version -- see
taxadb::available_versions(). Tables in this release violate schema rules
that postdate it, listed per table in manifest.csv under
rules_violated (7 of 9 tables here violate at least one).
Known issues in this release, since fixed in later ones: ncbi carries no
taxonID on accepted names; col and gbif leave scientificName empty
for many higher taxa; ott synonyms have no taxonomicStatus. See the
taxadb NEWS.
Each table is one or more Parquet files, named
22.12/<schema>_<provider>_part_<n>.parquet.
No special client is needed; anything that reads Parquet will do.
There are two kinds of table. dwc_<provider> is the complete name list:
one row per name, accepted names and synonyms together. common_<provider>
is the vernacular names, one row per name with the language it is in.
A provider that publishes no vernacular names has no common table; ott
is the one such provider here.
Every dwc table carries these columns, with the same meaning throughout:
Providers add their own columns beyond these; see columns in
manifest.csv for exactly what each table holds.
Darwin Core leaves acceptedNameUsageID optional on an accepted name, and
most providers omit it there, reasoning that an accepted name is its own
accepted name. These tables always populate it, repeating the taxonID.
That way resolving any name to its accepted identifier is a single column
read with no special case, whether the name turned out to be a synonym or
not. Relatedly, a name the provider redirects nowhere is its own accepted
name even where the provider hedges about it, so GBIF's doubtful and
COL's provisionally accepted names are self-referencing too.
These are not interchangeable, and should not be combined. Providers
represent independent taxonomic theories: the same name can be accepted by
one and a synonym of something else in another, so two providers can
contradict each other in ways that merging silently discards.
col, gbif and ott are themselves synthesis projects integrating other
checklists; the rest are primary authorities.
Each dataset carries its provider's terms. There is no repository-wide licence, and none of these datasets inherits terms from any other:
CC0-1.0 covers both the CC0 dedications and the works placed in the public
domain by their producing agency (ITIS and NCBI), which impose no conditions
either way. The two FishBase datasets are the only non-commercial ones here;
the other eleven permit commercial use, CC-BY-4.0 ones with attribution.
Attribution means citing the provider -- see Citation below.
manifest.csv records, for every table, the upstream release it was built
from, its row count, its full column list, and a SHA-256 of each Parquet
part.
The provider abbreviation and this release's version do not by themselves say what went in, which is why the upstream release is recorded separately: the GBIF table here is built from the most recent backbone GBIF has published, which is dated 2023-08-28, and is therefore older than this snapshot's version suggests.
Cite the underlying provider, not this redistribution:
For the package and the schema:
vernacularName | One common name, where the provider has one. |
ott | Open Tree Taxonomy | as released in taxadb-cache 22.12 (2022-12-20) | CC0 1.0 |
| yes |
dwc_ncbi | CC0-1.0 | yes |
dwc_ott | CC0-1.0 | yes |
dwc_ott| 6,753,824 |
| 189 MB |
| 34 |