Accession Numbers and Data Standards: Plant Records Best Practices

Ask three plant records officers how accession numbers should work and you may get four opinions. Yet the underlying principles are stable, they are shared across the botanical community, and getting them right early saves years of cleanup later. This post collects the practices that hold up over decades: numbering, qualifiers, and the data standards that make records exchangeable beyond your own garden.

What an accession number is, and is not

An accession represents one taxon, from one source, received at one time. The accession number is the permanent identifier for that event and everything descended from it. Two principles follow:

Accession numbers are meaningless identifiers. The temptation to encode information in the number (a bed code, a family abbreviation, a donor initial) is strong and almost always regretted. Plants move beds, taxonomy reshuffles families, and encoded meaning goes stale while the number itself must never change. Keep meaning in fields, where it can be corrected; keep the number as a stable label.

Accession numbers are never reused. A deaccessioned number is retired, not recycled. Herbarium vouchers, publications, seed-list entries and old labels may reference it forever, and a reused number silently corrupts every one of those references.

The most widely used format is simply year plus sequence: 2026-0147 for the 147th accession of 2026. The year prefix is technically “meaning in the number”, but it is the one exception the community has broadly accepted: receipt year is a fact about the accession event itself, it never changes, and it makes numbers roughly sortable and human-checkable. Zero-pad the sequence to a fixed width so numbers sort correctly everywhere, including in file names and spreadsheets.

Qualifiers: identifying the individual plant

The accession may be planted in several places, and individual specimens need identifying: for inventory, for mapping, for recording that this tree was pruned or measured. The convention is a qualifier appended to the accession number: 2026-0147*A, 2026-0147*B, or a numeric equivalent. Practices that keep qualifiers useful:

• Assign qualifiers when plants are separated (planted out, distributed), not necessarily at receipt; a seed pan does not need twenty qualifiers.
• Like accession numbers, qualifiers are never reused within an accession. If plant *B dies, the next separation becomes *D, not a second *B.
• Record the qualifier’s own history: location moves, condition assessments, measurements and eventual death all attach to the qualified plant, while source and identity attach to the accession.

Provenance is the scientific core

For conservation and research value, the single most important field group is provenance. The established categories (wild origin, from a wild source via cultivation, cultivated origin, and insufficient data) look simple, but discipline in applying them is what makes a collection scientifically citable. Record alongside them the collector, collection number, date, locality and coordinates for wild material, and the source garden or nursery for cultivated material. A modest collection with rigorous provenance data is more valuable to science than a vast one without it.

Names change; records should remember

Identification is an act, not a fact. Best practice is to record determinations as dated, attributed events: who identified the plant, when, as what, and with what confidence. When a later botanist revises the identification, the earlier determination remains in the history. This is how herbaria have worked for two centuries, and living collections deserve the same rigour, not least because the determination history is often the only way to interpret old labels, publications and seed lists that used a former name.

Exchange standards: ITF2 and Darwin Core

Records kept only in a local format are records only your garden can use. Two standards matter most:

ITF2, the International Transfer Format, version 2, was developed for exchanging accession-level data between botanic gardens and with BGCI’s PlantSearch. Its field list (accession number, name parsed into its components, provenance type, collection details, and so on) doubles as a sensible minimum checklist for what a garden should be recording at all. If your records can produce a valid ITF2 file, your core data is in good shape almost by definition.

Darwin Core is the biodiversity-informatics standard used by aggregators such as GBIF. Publishing your collection as Darwin Core occurrence records makes it visible to researchers worldwide. That visibility is increasingly an expectation for gardens that describe themselves as scientific collections, and in some cases a condition of funding or accreditation.

The practical implication: whatever software you use should map its fields to these standards, so participation in the wider network is an export, not a re-keying project.

Small habits that compound

• Define controlled vocabularies for garden areas, condition ratings and source types, and resist free-text where a list will do.
• Date everything, including the record itself: when it was created and last edited, and by whom.
• Inventory on a cycle. A records system is only as credible as its last stock-take; even a rolling inventory of one garden section per month keeps the whole dataset trustworthy.
• Write your conventions down. A one-page records manual covering numbering, qualifiers, vocabularies and who may edit what turns personal knowledge into institutional knowledge.

None of this requires sophisticated tooling to begin. But applied consistently, these practices are what let a collection database do its real job: telling people you will never meet, decades from now, exactly what grew in your garden and how it came to be there.