Skip to content

Batches B4-B8: alternatives, notes, qualifiers, single letters, odd characters, no ? in names - #254

Merged
mdoering merged 16 commits into
mainfrom
fix/issues-b4-b8
Nov 19, 2026
Merged

mdoering merged 16 commits into
mainfrom
fix/issues-b4-b8

Conversation

@mdoering

@mdoering mdoering commented Nov 19, 2026 •

Copy link
Copy Markdown
Member

Triage batches B4–B8, one commit per issue, plus the follow-ups from the review of every pinned cf./aff./nr. name and the phrase work on codes after a species.

Issues

Commit Issue Change ChecklistBank COL XR
12f9abc #235 alternatives with their own authorship: Carduelis carduelis (Linnaeus, 1758) / Serinus canaria (Linnaeus, 1758) is OTHER, one genus is informal 1 row 0
49a2523 #241 a bracket holding = is a synonym note, left unparsed: Pleurosorus (= Asplenium) rutifolius (R.Br.) Fee 246 rows 0
60b40a4 #233 a capitalised epithet after cf./aff./nr. on evidence: Phrynopus cf. Barlenae → species barlenae + cf. 9 rows 0
d48c00d #231 a quoted epithet before a manuscript note: Lepidobolus 'quadratus' (sp. nov. ined. 4 rows 0
2d22a5d #246 c.f., dotless/glued n sp, lone infraspecific letters; new warning single letter genus or epithet 2,930 rows 199
41ae0a9 #252 a qualified epithet before sp. keeps its species: Euxoa nr. idahoensis sp. 1clay 179 names 0
a05df2c, 0495dff #244, #236 new warnings genus or uninomial with digit or unusual character and epithet with digit or unusual character; a GTDB suffix is exempt 9,608 rows genus, 23,452 epithet (10,627 / 26,159 before the GTDB exemption) 556 (epithet)
9fc5f4d #244 a GTDB name is INFORMAL, its suffix kept on its name part at every rank (Bacillus_BF, Pseudomonas_E putida, Streptococcus mitis_I), anchored on the taxon it was split from (Pseudomonas putida); GTDB's unnamed species Pseudomonas_E sp002874965 informal with the assembly as phrase, no longer OTHER 13,897 rows (9,506 from OTHER) 0
e5e7fec #248 NameType::Other documents what an unparsable result tells – –
08a4f04 #243 no ? in a rendered name; doubtful + the epithet qualifier carry it – 37

Review and phrase follow-ups

Commit Change ChecklistBank COL XR
1431c54 vel aff./vel cf. is the qualifier of the last epithet; Septifer bilocularis Brachidontes cf. variabilis → species + unparsed second name; split BIN Bold:Aab, 0925 → phrase 22 rows 0
ccc8d6e codes with full stops/underscores, labelled codes (spore IB.15), DSM 30006 = CIP 81.8, fam. designations, bracketed rank notes after the author ~6,050 rows 0
4e6e35c clade codes (ANME-1, SAR324, NC10, UBA4372 sp) are OTHER; HIV/SIV/HTLV strains with the virus code 3,978 rows 0
5010879 Agabus cf. lutosus griseus/griseipennis → species lutosus + cf. + phrase griseus/griseipennis 0 0

api (gbif/name-parser dev)

The Java binding builds against name-parser-api 6.0.2-SNAPSHOT (851e26c) until the api is released; a binding release needs the api release pinned first.

Golden files

Closes #235, closes #241, closes #233, closes #231, closes #246, closes #244, closes #236, closes #243. Refs #252, #248.

`Carduelis carduelis (Linnaeus, 1758) / Serinus canaria (Linnaeus, 1758)` became Carduelis
carduelis with the author "Serinus canaria", and `Chionodoxa forbesii BAKER / luciliae BOISS.`
left the alternative unparsed: the alternatives rules of #212/#225 wanted the slash right after
the epithet. An authorship may now stand between them, so the same rules apply: two genera are
no name, one genus is informal with the alternatives as its phrase.

Only an authorship of capitalised names, years, `&`, `et`, `ex` and `in` counts, every alternative
after it must be a whole epithet or species with or without its own authorship, and the slash
stands apart on one side at least. So alternative authors (`Aus bus Jarocki / Schinz`), strains
(`indica PD1998/pus`), pages (`pl. 222 / figs 1 – 7`), hosts (`ex Ficus popenoei/nymphaefolia`)
and a word holding a slash (`Girraween/Polblue variant`) stay as they were.

ChecklistBank: 1 row of 190,347 with a slash or pipe (`Gymnophryxe pretoriensis Bezzi, 1911/
Archiclops pretoriensis Bezzi, 1911`, now OTHER); COL XR: none.

Closes #235
`Pleurosorus (= Asplenium) rutifolius (R.Br.) Fee` took the bracket for the basionym author or the
subgenus and lost the epithet, `Tillandsia schreiteri Lillo & Castellanos (=T.sphaerocephala
Baker?)` made the synonym an author. Only a trailing `[= Grislea L. 1753]` was set aside. Now any
round or square bracket holding `=` is: it is left unparsed, the name is doubtful as before, and
the rest parses (`Pleurosorus rutifolius (R.Br.) Fee`). After an indet marker the bracket is the
voucher of the unnamed species and stays its phrase (`Euodia sp. (= RFK/3921)`). A synonym
reference alone (`=ixioides?`) is OTHER.

ChecklistBank: 246 of 695 rows with `=` in brackets; COL XR: none.

Closes #241
`Phrynopus cf. Barlenae` and `Teucrium cf. Cyprium Boiss.` kept the capitalised word in the phrase
`cf. Barlenae`, as a genus compared with. The capitalised epithet rule (`Ormosia Bolivarensis`)
now looks past a cf., aff. or nr. too, and the qualifier doubts the epithet: species `barlenae`,
qualifier `cf.`, with the warning "capitalised epithet put in lower case".

After a qualifier the word is mostly a genus or higher taxon compared with, so only a Latin epithet
ending or an abbreviated author counts as evidence, in the name or given apart. A rank hint does
not, nor does an author with a year, which is mostly the anchor genus' (`Dicranoplia cf. Anomala`
+ `Reitter 1903`), nor a family-group name. Without evidence the word stays the phrase:
`Pristimantis cf. Ventrimarmoratus`, `Onthophagus cf. Aphodius`.

ChecklistBank: 9 of 1,492 rows with a capitalised word after cf./aff./nr. (`Lingula cf.
Ioulanensis`, `Picalgoides aff. Pteroglossorum`); `Halictus cf. Nomioides`, `Metacrias cf.
Tineldae` and `Necatoridi cf. Necatoreae` follow the Latin ending as they would without the
qualifier. COL XR: none.

Closes #233
`Lepidobolus 'quadratus' (sp. nov. ined.` made "'quadratus' sp.nov." an author of the uninomial
Lepidobolus. A quoted epithet followed by a note of a new or unpublished species (`sp. nov.`,
`sp. nov. ined.`, `ined.`, `ms.`), in brackets, its bracket left open or bare, is a manuscript
name and no cultivar: the species `quadratus`, flagged manuscript, with the note as nomenclatural
note. A quoted word alone stays a cultivar (`Lepidobolus 'quadratus'`).

ChecklistBank: 4 rows (`Aspidella 'noddy' ined.`, `Celmisia 'pulchella' ms`, `Coccomyces
"lignicola" ined.`); COL XR: none.

Closes #231
Most single-letter genera and epithets were misreadings. `Dracaena c.f. afromontana Mildbr` took
the `c` of a two-dot cf. as the epithet: a lower-case `c.f.` before an epithet is now cf. (the
initials of `C.F. Liang` stay initials). `Forcipata n sp. ancantha` and `Emblyna n. sp1` took the
`n` of n. sp. as the epithet: a dotless n and a marker glued to its number now open the phrase of a
new species as `n. sp.` does, the phrase kept as written, and the formatter adds no `sp.` before
it. A lone letter in the infraspecific slot without a dot is a designation, so the name is
informal as with the species slot (`Polypedates leucomystax b`); with a dot it stays an abbreviated
epithet (`Diacamma rugosum r.`).

What is left gets the new warning "single letter genus or epithet": `M mus`, `Galium spurium var.
L.`, `Collettea a`, the Greek-lettered and lettered varieties (`var. ß`, `var. b`). Not the `?` of
a missing genus, and not an abbreviated genus, which keeps its dot and its own warning (`M. mus`).
22 golden rows gain the warning.

ChecklistBank: of 49,312 candidate rows 2,651 gain only the warning, 256 become informal and 23
other n sp. names become phrases; COL XR: 199 rows, nearly all the warning.

Closes #246
A qualifier only doubts the species it stands before, it does not take it away:
`Euxoa nr. idahoensis sp. 1clay`, `Prionospio aff. iberian sp. c` and
`Ramazzottius aff. szeptycki sp. n. 1 ALL-2022a` are the species with its
qualifier and the unnamed one's phrase, as `Gemmula cf. cosmoi NP-2008` is.
Several unqualified words before the marker still hang the phrase off the genus
(`Prionospio iberian canyons sp. c`). The Euxoa golden rows are back to what
they were before #252.

ChecklistBank: 179 names, mostly `Acropora aff. ... sp. 1 JTL-2012`; COL XR: none.

Refs #252
From a review of every pinned cf./aff./nr. name:

- `vel aff.` or `vel cf.` behind a name, "this or one close to it", is the
  qualifier of the last epithet as a whole: `Garcinia riparia vel aff.`,
  `Pisonia macrocarpa Miranda vel aff.`, `Dalbergia densa var. australis vel
  aff.`. `vel` was read as an infraspecific epithet or an author, `aff` as an
  epithet. Also in a separate authorship (`Miranda vel aff.`).
- A capitalised word with a qualified epithet after a species is a second name,
  no author takes a qualifier: `Septifer bilocularis Brachidontes cf.
  variabilis` is the species, the second name left unparsed. An autonym after
  its author stays one (`Swartzia flaemingii Raddi cf. flaemingii`).
- A BOLD BIN split by a comma, `Lacanobia nr. subjuncta Bold:Aab, 0925`, is the
  phrase of the doubted species, as written; it was the author "Bold".

ChecklistBank: 22 rows (12 names), all of them `vel aff.` but Septifer and
`Sumatrina (?) turbida Streblus aff. longissima`, which no longer loses
`Streblus aff.`; COL XR: none. One golden row (Lacanobia) re-baselined.
- A code after a species may hold full stops and underscores: `Anabaena bergii
  PMC215.03`, `Escherichia coli STEC_H.I.8`, `Botrylloides cf. lenis SMRC_3.1`.
  Split at the full stop, the code became an author and its last digits were
  left over. A word labelling the code goes with it into the phrase (`spore
  IB.15`, `type A14`, `clone CK051`, `isolate Kol-2`), and so does a code glued
  to the epithet by a full stop (`Halimeda cuneata.Africa.1`). An abbreviated
  author glued to a year by a full stop (`Mill.1768`) is no code.
- A culture accession takes the accessions of the same strain in other
  collections after `=`: `DSM 30006 = CIP 81.8`; the acronyms were a team of
  authors, often COMPLETE with nothing left over.
- `fam.` and a letter, a code or `gen.` after a taxon designate an unnamed
  family in it, as `gen.` does for a genus: `Digenea fam. B sp. MAT.CG.2`.
- A rank and an epithet in brackets after the author, and what follows, are
  left unparsed: `Camelina sativa (L.) Crantz (f. annua), f.1`. A cultivar after
  it stays the name's (`(f. biennis), cv. Annushka`).

ChecklistBank: about 6,050 rows (2,962 dotted codes, 1,423 accessions with
equivalents, 1,198 underscore codes, 406 labelled codes, 36 rank notes, 25
families); COL XR: none.
A code of capitals and digits in the genus's place labels a clade:
`ANME-1 cluster archaeon GoMg3.2`, `SAR324 cluster bacterium`, `NC10
bacterium`, `UBA4372 sp`, `TM7 phylum`. It was title-cased into a genus
(`Anme-1`, `Sar324`) and the rest read as epithets and authors; now OTHER. The
strains of the retroviruses (`HIV-1 06TG.HT064`, `HTLV-1 ...`) are OTHER with
the virus code.

ChecklistBank: 3,978 rows, 3,090 of them HIV strains; COL XR: none.
`Agabus cf. lutosus griseus/griseipennis` doubts the species lutosus, the
alternatives lie below it: species lutosus with the qualifier cf., the
alternatives its phrase, as for an unqualified species (`Tachybaptus
ruficollis tricolor/vulcanorum`). Alternatives of the species itself keep
only the genus: `Agabus cf. lutosus/griseipennis`.
…, #236)

Two warnings tell which part is broken, so that a genus can still be
trusted when only an epithet is not: UNUSUAL_GENUS for a genus or
uninomial (`Imsh8ia aleurites`), UNUSUAL_EPITHET for an infrageneric,
specific or infraspecific epithet (`Imshia aleu3ites`,
`Lepidostoma unidentified_sp1_ab9tf03`, `Ephedra przewalskiiß`,
`Coccinella 6maculata`, `var. β`). A word with a digit on its own stays
OTHER (`Lepidostoma1`).

Latin letters with accents (`Cephaëlis`, `Pardøsa`), æ and œ, hyphens,
apostrophes and dots (`o'donelli`, `st.-johnii`, `M. mus`) are no odd
characters, nor are the digits of a virus species
(`Simplexvirus humanalpha1`) or the GTDB suffix of a genus under the
bacterial code (`Bacillus_BF`). The cultivar epithet is not looked at.

30 golden rows gain the warning.

Closes #244
Closes #236
An unparsable result carries its type and code, nothing else: the doc
of NameType::Other lists what it stands for, a virus with the virus
code among them. Warnings stay on parsed names.
The `?` qualifier of a doubted name part stays out of the name: the
doubtful flag and the epithet qualifier carry it, as the warning
"question marks removed" says. `Metacnephia sommermanae ?` renders
`Metacnephia sommermanae`, `Ferganoconcha? oblonga` `Ferganoconcha
oblonga`, `Glenea (? Macroglenea) anulicornis` `Glenea (Macroglenea)
anulicornis`. cf., aff. and the other qualifiers stay. The api's
NameFormatter does the same on dev.

6 golden format rows re-baselined.

Closes #243
The capital suffix GTDB gives a split genus or species is a naming
convention the parser keeps on purpose, given a code or not:
`Pseudomonas_E putida`, `Streptococcus mitis_I`, `Haemophilus_A
parahaemolyticus_B`. Only under the bacterial code it was exempt, so
about one in nine ChecklistBank names flagged were GTDB names. A
lower-case label after an underscore is still flagged
(`Culicoides clavipalpis group_sp08`).
The api's NameFormatter keeps the `?` of a doubted name part out of the
rendered name (#243) and reads the dotless `n sp` marker (#246); its
Warnings has the new UNUSUAL_GENUS and UNUSUAL_EPITHET (#244) and the
constants the parser already emitted. The binding builds against the
snapshot until the api is released; a binding release needs the api
release pinned first.
…plit from (#244)

The suffix GTDB gives a taxon it splits, an underscore and one to three
capitals, stays on its name part at every rank: the uninomial
`Bacillus_BF` (was Bacillus with the phrase BF, unless under the
bacterial code), the genus of `Pseudomonas_E putida`, the epithet of
`Streptococcus mitis_I`. A phrase cannot take it, it is always appended.
Such a name is no regular scientific name: it is informal, whatever the
code, since the same suffixes are used elsewhere in the tree, and the
odd-character warning skips the suffix. Its anchor drops the suffixes:
`Pseudomonas_E putida` hangs off Pseudomonas putida, `Bacillus_BF` off
Bacillus.

The unnamed species GTDB places in a genus, `Pseudomonas_E sp002874965`
or `Acidaminococcus sp000437815`, was OTHER: it is informal now, the
genome assembly its phrase. After a code in the genus' place it stays
OTHER (`UBA1847 sp013042375`).

ChecklistBank: 13,897 rows, 9,506 of them rescued from OTHER, 3,545
binomials now informal, 845 uninomials keep the suffix
(`Paenibacillus A` -> `Paenibacillus_A`); COL XR: none. Two golden rows
re-baselined (`Desulfobacterota_B`, `Blattellinae_SB`). The api's
ParsedName.anchor() does the same on dev.
@mdoering
mdoering merged commit bbf9c22 into main Nov 19, 2026
10 of 11 checks passed
@mdoering
mdoering deleted the fix/issues-b4-b8 branch November 19, 2026 15:51
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment