How the calendar
is built.
Fruit phenology of native Hawaiian plants from seed-bank collection records: what the records are, how each one is given a month, an island and a place, when we are willing to call something a season, and what the calendar cannot tell you. Every figure on this page is computed from the records as they stand today.
What the dataset is
The calendar is built from seed-collection accession records contributed by — member banks of the Hawaiʻi Seed Bank Partnership — — records of — taxa. Each record is one collection: a taxon, the bank that holds it, and — where the bank recorded them — a collection date and a standardized reference code naming the island and the area it came from.
These are records of collecting, not of observing. A month on the calendar means seeds of that taxon were actually collected in that month. That is a good proxy for when fruit was ready, and it is also a record of where people went and when they had the access and the funding to go.
Taxon names are kept as the banks recorded them. Records that are identical in every field are loaded once.
How a record gets its month
The month is the month of the recorded collection date. Most records carry a date; the rest have no month and take no part in the calendar, though they still count toward a taxon’s totals. We do not estimate a month from an accession number or a year.
How a record gets its island
An island is assigned by three rules, tried in order of how much we trust them. Each record keeps a note of which rule placed it.
| Rule | What it reads | Records |
|---|---|---|
| Collection code | The island token written into the record’s standardized reference code. | — |
| Single-island bank | The bank holds collections from one island only, so its records are on that island. Banks that collect across islands are excluded from this rule. | — |
| Single-island taxon | Every code-resolved record of the taxon is on one island, so its remaining records are placed there. | — |
| Unresolved | None of the rules applies. The island is left blank; nothing downstream guesses. | — |
The check. The single-island-bank rule can be tested wherever a record has both a code-read island and a single-island bank. On the workbook of August 31, 2026, all 5,580 such records agreed (100.00%). We report this figure rather than assume it, and it is recomputed whenever the records change.
Named areas and mapped regions
The area is the population-reference code in the record’s reference code — a short code for a named place, such as a reserve, a gulch or a management unit. Case variants of the same code are merged. By decision of the Network (September 1, 2026), areas are published at this named-area grain. No collection-point localities are published.
Some areas are drawn on the map as real boundaries: each was matched by name to a feature in the Hawaiʻi Statewide GIS — a reserve, an ahupuaʻa or a watershed partnership — and the map shows, for every region, which kind of match placed it (an exact name match, a code that uniquely prefixes a feature name, or a feature name found inside the area’s label). A match is our reading, offered for partners to check; the authoritative shapes stay with the state. Areas without a match are searchable by name but not drawn. Most unmatched records come from management units whose boundaries are held by the managing program.
When we make a seasonal claim
A taxon needs at least 10 month-resolved records before the calendar says anything about its season. Below it, the records are still shown, as plain counts, with a label saying no seasonal claim is made.
Ready this month lists only taxa above the bar, ranked by the share of each taxon’s month-resolved records that fall in the month — so a taxon with a real seasonal peak ranks above one that is simply collected often. A taxon’s peak month is the month holding its largest share.
Conservation flags (PEPP, Species of Conservation Importance) are carried as the banks recorded them. Climate vulnerability scores are those of Fortini et al. (2013).
What the calendar cannot tell you
- Absence is not evidence. No record in a month does not mean a taxon is unavailable then; it may never have been visited then.
- Effort is uneven. Some islands, and lands with a resident management program, are far more heavily collected than others. The calendar inherits that.
- Seasons move. Fruiting shifts with elevation, with windward and leeward exposure, and from year to year. A statewide month is a starting point for planning a trip, not a forecast.
- Collection follows need. Rare taxa are collected whenever seed can be had, which can flatten a real season; common taxa may be collected only when a project needs them.
- Names are as recorded. We have not reconciled synonyms across banks, so one taxon may appear under more than one name.
Exports
Three tables can be downloaded from the calendar, each as CSV or JSON, each written from the live records at the moment of download and each opening with its attribution, version date and citation: collections by taxon, month and island; a per-taxon summary that carries the seasonal bar; and the named areas.
How to cite
Hawaiʻi Seed Bank Partnership and Laukahi Network (2026). Fruit phenology of native Hawaiian plants from seed-bank collection records [dataset]. Laukahi, The Hawaiʻi Plant Conservation Network. https://laukahi.org/phenology (accessed November 18, 2026).
The version date is the day the records were last loaded; cite it together with the date you accessed the calendar, since both the records and the figures on this page change as banks contribute. The data belong to the contributing banks; please credit the Hawaiʻi Seed Bank Partnership in any use.
Corrections
If a record, an island, an area match or a name looks wrong, write to coordinator@laukahi.org with the taxon and what you expected to see. Corrections are made in the source records and appear here at the next load.