GBIF

A male Cannabis sativa sheet collected in northern Thailand in 1997 and published through GBIF as a preserved specimen (Maxwell 97-797 (1997)).

A male Cannabis sativa sheet collected in northern Thailand in 1997 and published through GBIF as a preserved specimen (Maxwell 97-797 (1997)).

GBIF

Global Biodiversity Information Facility

Database

Kept by	GBIF
First published	2001
What it holds	Specimens and occurrences Genebank accessions
Covers	Worldwide

Entries

Entries	3,893,699,971
Cannabis entries	35,440
Counted	13 November 2026

Use

Data licence	Varies by datasetCC0, CC BY 4.0 or CC BY-NC 4.0, chosen by each publisher
Access	Open
Website	gbif.org

GBIF, the Global Biodiversity Information Facility, is an intergovernmental network and data infrastructure that gathers records of where and when organisms have been found and makes them openly available through a single portal, GBIF.org. Its Secretariat is in Copenhagen, Denmark. Its sources include museum and herbarium specimens, some gathered in the 18th century, DNA barcodes and photographs taken with phones. On 13 November 2026 its index held about 3.89 billion occurrence records. Of these, 35,440 were filed under the genus Cannabis, among them 13,033 preserved specimens.

Origins and organisation

GBIF grew out of a 1999 recommendation by the Biodiversity Informatics Subgroup of the Megascience Forum of the Organisation for Economic Co-operation and Development (OECD). The subgroup called for an international mechanism to make biodiversity data accessible worldwide. Science ministers of the OECD member states endorsed the proposal. In 2001 participating governments established GBIF by a memorandum of understanding. The facility was planned as a framework for digitising biodiversity data and for making databases scattered around the world work together.

The countries and organisations that take part work through participant nodes. A Governing Board approves an annual work programme for the Secretariat within a five-year strategic framework. Under GBIF's data user agreement the Secretariat is neither owner nor custodian of the data it serves and does not guarantee their quality or completeness.

The founding aim centred on specimens, then kept almost entirely in museums. Observations later outgrew them. Records linked to a physical specimen made up 25 per cent of the data in 2007 and 14 per cent in 2020, while datasets gathered mainly by volunteers rose from 11 to 65 per cent over the same years. An estimate published in 2021 put the share of the world's natural history specimens available through GBIF at 10 to 20 per cent.

How records reach GBIF

Institutions that hold data, among them herbaria, museums and gene banks, publish through GBIF by registering their datasets with its registry, from which GBIF harvests and indexes them. Most records use Darwin Core, a set of terms first drawn up around 1999 for the place and date at which an organism occurred and for specimens kept in collections. The Taxonomic Databases Working Group ratified it as a standard in October 2009. A germplasm extension, modelled on the FAO/IPGRI Multi-Crop Passport Descriptors, carries the passport data of genebank accessions in the same format. GBIF's Integrated Publishing Toolkit, open-source software, packages a dataset as a Darwin Core Archive with a public web page and registers it for harvesting.

GBIF crawls each registered dataset, preferring Darwin Core Archives. Every record receives a stable GBIF identifier. Where a publisher supplies no occurrence identifier, the identifier is built from the institution code, collection code and catalogue number. Coordinates given in another datum are converted to WGS84. All coordinates are checked against the country the publisher gives. A record with zero or out-of-range coordinates, or with coordinates outside its stated country, is flagged. Latitude and longitude that appear to be swapped are put back and flagged. Each scientific name is matched to the GBIF Backbone Taxonomy, which is rebuilt several times a year, mainly from the Catalogue of Life. The publisher's unprocessed values remain available in downloads.

Licences and citation

In 2014, after a consultation across the network, the Governing Board decided that every occurrence dataset must carry one of three Creative Commons licences, chosen by its publisher: CC0, which leaves the data free for any use; CC BY, which requires attribution; or CC BY-NC, which also excludes commercial use. Since August 2016 every open-access occurrence dataset on GBIF.org has carried one of them in machine-readable form. GBIF holds that restrictive readings of "non-commercial" run counter to open access, and asks users of CC BY-NC data to act in good faith. Of the 35,440 Cannabis records on 13 November 2026, 17,172 were under CC BY-NC 4.0, 13,099 under CC BY 4.0 and 5,169 under CC0.

GBIF's data user agreement requires users to credit the publishers whose data they use, where appropriate by a digital object identifier (DOI). Users must also keep to each publisher's licence, which prevails over the agreement where the two conflict. The agreement is governed by Danish law. GBIF has issued a DOI with every download since 2016. In 2019, 38 per cent of the studies that used its data cited one. A download's landing page lists the datasets it draws on. A single record is cited by its dataset's citation together with the record's own address on GBIF.org.

Cannabis records

An 1834 sheet from Annaba in the Paris herbarium, one of the museum's Cannabis specimens published through GBIF (Steinheil s.n., Bône (1834)).

The 35,440 Cannabis records held on 13 November 2026 came from 887 datasets and 544 publishers. Observations by people numbered 21,038 and preserved specimens 13,033. Most of the remaining 1,369 were material samples (538) and records of material cited in publications (493). Russia had the most records (5,397), followed by the United States (4,740), Germany (2,687), the United Kingdom (2,209), Sweden (2,130) and China (1,951). The largest single dataset was the research-grade observations of iNaturalist, with 10,061 records under CC BY-NC.

The 13,033 preserved specimens came from 128 countries, most from the United States (2,881), China (1,624) and Canada (834). Only 5,787 of them carried coordinates. The largest sources of specimen records were EURISCO, the European genebank catalogue (1,368), a 2023 compilation of plant specimen data from China (946), the New York Botanical Garden (472), the Deam Herbarium of Indiana University (418), the Naturalis Biodiversity Center (387) and the Muséum national d’Histoire naturelle (307). EURISCO describes genebank accessions held in Europe, from data sent to it by national inventories. GBIF files all 1,368 of its Cannabis records as preserved specimens.

Use in cannabis research

Tori Ford and colleagues (2024) built ecological niche models for feral Cannabis sativa in 12 states of the Midwestern United States. They gathered herbarium and occurrence records through the programming interfaces of eight online repositories, GBIF among them. Collaborators at three universities supplied more. Cleaning removed duplicates, outliers, invalid coordinates and coordinates belonging to institutions. After cleaning, restriction to the region and thinning, 620 of the 680 records drawn from GBIF were kept, out of 1,193 in the final set.

Anna Halpin-McCormick and colleagues (2025) modelled the distribution of Cannabis sativa from a GBIF download of iNaturalist observations. Inspecting the photographs of 416 records, they judged 302 to show wild or escaped plants. After duplicates were removed 234 remained. They kept the 137 east of the prime meridian because they took Eurasia to be the centre of origin of Cannabis.

See also

• All accessions
• Gene bank
• Muséum national d’Histoire naturelle
• Royal Botanic Garden Edinburgh