Species list → Darwin Core Archive (Taxon core)
What you are seeing — you have a species LIST — a site inventory, a park's flora, a survey checklist — and every instruction you can find assumes occurrence records, with dates and coordinates your list never had.
A checklist is not a thin occurrence dataset; it is a different dataset class with its own core. GBIF ingests it as a Darwin Core Archive whose table is the Taxon core — one row per name, with taxonID, scientificName and taxonRank — and whose meta.xml declares that row type. The occurrence fields you keep being asked for (eventDate, decimalLatitude, basisOfRecord) simply do not apply.
The trap is forcing a list through an occurrence template anyway. An occurrence asserts that somebody observed this taxon at this place on this date. Your checklist asserts none of that — so every date and coordinate added to make a template validate is a small fabrication, and it will be indexed as if it were a field observation. A list published as occurrences pollutes exactly the record it was meant to enrich.
Three things decide whether a checklist archive loads cleanly. The names: scientific names, one per row — common names belong in their own column, never mixed into scientificName. The rank: a list that mixes families, genera and species must say which is which, or every downstream match is a guess. The identifiers: each row needs a taxonID that is stable within the file, because that is what the archive's internal links hang off.
This corridor is measured, not described: we built a checklist archive this way and put it through GBIF's own validator. It finished, reported the archive indexable — their word, their bar — indexed 426 records, and raised zero issues. Being indexable is the structural gate, not publication itself: getting a dataset onto gbif.org still requires a registered, endorsed publisher account, which no file tool can shortcut and we do not claim to.
taxonRank column is what stops a matcher treating them as siblings.kingdom, family): homonyms are real, and a genus name alone can match two kingdoms.Upload the spreadsheet or CSV as it is. We read the names, confirm each column from its values, refuse what the values cannot support, and emit the Taxon-core Darwin Core Archive — the same construction GBIF's validator finished on with 426 records indexed and zero issues raised. Free until 1 September 2026, 00:00 UTC · no account · your rows are discarded after the check.
Check the file you have →Measured on a file GBIF had already accepted and served — their validator asks will this load, and it had. Every figure is read from ops/data.json, never typed here. We issue no rating and no seal — each finding names the rule it applied.