GBIF at 25: The Open Data Machine Behind Global Biodiversity Science

Twenty-five years ago, a handful of governments bet that the fastest route to saving life on Earth ran through something unglamorous: standardized, freely shared data about where species live. That bet became the Global Biodiversity Information Facility, or GBIF, an intergovernmental organization proposed in 1999 by the OECD’s Megascience Forum and launched in 2001 with 21 country participants. Today, by August 2026, the network spans data providers in more than 120 countries and offers free access to over 3.8 billion occurrence records, each one a digital testimony that at least one individual organism was recorded at a specific place and time, by a specific person or institution. A new overview published in the Biodiversity Data Journal traces how this distributed machine was built, how it works, and why its infrastructure needs are now larger than at any point in its history.

The founding logic was born of frustration. The 1992 Convention on Biological Diversity had formalized international commitments to conserve and sustainably use biodiversity based on scientific knowledge, yet the data needed to honor those commitments were scattered, concentrated in a small number of wealthy institutions, and maintained under a patchwork of pre-digital protocols. The OECD’s 29 national parties concluded that an international mechanism was needed to make biodiversity data accessible worldwide, envisioning a facility that would be distributed while encouraging cooperation, global in scale though implemented nationally, and open to all countries while funded mainly by those with the greatest capacity. GBIF’s 2026 reality bears a striking resemblance to that vision, but the organization has also deepened its operations far beyond what the founders could have anticipated.

What makes GBIF unusual is its federated structure. It is not directed by any single overarching organization; the collective community is its own highest authority. Seventy Participant countries and 42 organizational Participants govern through a Governing Board, advised by standing committees including a Science Committee and a Nodes Committee. The operational engine sits in national nodes, typically hosted by existing biodiversity institutes, which coordinate data mobilization, support publishers, encourage research use, and act as knowledge hubs. A Secretariat of 35 staff, housed at the University of Copenhagen, operates the informatics and coordinates the network, though helpdesk services and development are increasingly distributed to nodes and regional contractors. Much of the work also depends on structured volunteers, from translators who render the GBIF portal in twelve languages to mentors and reviewers recognized annually with digital badges.

The technical core is data mediation. Institutions holding biodiversity data, from natural history museums to citizen science platforms to private companies, publish datasets through pathways such as GBIF’s open-source Integrated Publishing Toolkit, which creates Darwin Core Archive files and registers them with the GBIF Registry. Harvesting infrastructure then continuously retrieves these datasets, applies taxonomic interpretation, geographic validation, and temporal normalization, and stores the results in a scalable warehouse built on Apache Hadoop and Apache Spark following a medallion architecture that preserves original records while generating increasingly standardized representations. Search indexes based on Elasticsearch expose the products in real time through open RESTful APIs, powering the GBIF web portal, hosted national portals, and client libraries such as rgbif for R and pygbif for Python. Every month, GBIF also snapshots occurrence data to three cloud repositories for large-scale analytical queries.

Interoperability across billions of heterogeneous records depends on community-driven data standards, chiefly Darwin Core, maintained by the nonprofit Biodiversity Information Standards, known as TDWG. Because GBIF is often the only place many researchers encounter the concept of data standards, it plays a critical role in testing, maintaining, and developing them. A major milestone arrived around 2016 with sampling event datasets, which invert the occurrence paradigm by documenting the survey design and sampling effort, enabling inference of organismal absence and interpretation of abundance. This capability, built on the Darwin Core Event Class and extended in 2024 by the Humboldt Extension, is now central to GBIF’s future, supported by the new Darwin Core Data Package designed to reflect the natural complexity of ecological data. GBIF plans to implement infrastructural support for this model incrementally over the coming years.

The data sources themselves tell the history of the network in chronological layers. Nearly half a billion digitized natural history records form the oldest layer, though digitized specimens still represent only a fraction of the world’s collections, and GBIF’s Global Registry of Scientific Collections now links over 270 million specimens to their institutions. Participatory science platforms emerged in the 2010s as a dominant contributor, with estimates suggesting around half of all GBIF-mediated data may be tied to amateur observers, a substantial proportion from bird networks. DNA-derived data, from barcoding to environmental DNA metabarcoding, became a serious focus in the mid-2010s, with a Metabarcoding Data Toolkit piloted in 2024 and a new eDNA integration initiative launched in 2026. Survey and monitoring data, including camera traps and passive acoustic sensors, represent the most active development area today, supported by a 2025 publishing guide that better represents survey design.

Quality control follows a library analogy: GBIF sets standards, provides guidance, and endorses publishers, but does not edit each dataset, much as a library does not edit each book. Instead, it requires every publisher to maintain a responsive point of contact, flags several dozen common data quality issues during processing, from geospatial errors to taxonomic mismatches, and promotes a framework of contextual fitness-for-use rather than absolute quality. Users are asked to report concerns, and a data validator service helps publishers resolve issues before publication. Persistent geographic and taxonomic asymmetries remain the network’s central challenge: underrepresentation of highly biodiverse Global South regions, overrepresentation of charismatic birds, and scarcity of marine and cryptic taxa. These gaps are larger than GBIF can resolve alone, but capacity development programs aim to narrow them.

The measurable impact is remarkable. GBIF-mediated data underpin a body of over 15,000 peer-reviewed publications, growing at more than eight new articles per day, with research reaching nearly all the world’s countries. Downloads average 10 billion records per day, accessed by over a quarter of a million unique email addresses. In a 2022 user survey, 47 percent of users said their research would have been impossible without GBIF, and a further 41 percent said it would have required significantly greater time and effort. The annual value of researcher time saved was estimated at 35 million euros in 2021, and the overall economic impact works out to roughly 12 euros in compounding societal benefits for every euro invested. At the science-policy interface, GBIF data underpin 25 percent of IUCN Red List assessments, feed IPBES reports, and support multiple indicators of the Kunming-Montreal Global Biodiversity Framework, including the headline indicator for invasive alien species under Target 6.

The network’s reach now extends into the private sector. Under the Global Biodiversity Framework’s Target 15, nations must assess and disclose biodiversity risks and impacts by businesses, and by the end of 2025 nearly 150 private sector companies had contributed over 14 million occurrences to GBIF, mainly from impact assessment data. The Taskforce on Nature-related Financial Disclosures explicitly calls on businesses to publish biodiversity data via GBIF in its Recommendation 5 on incentivizing corporate nature data exchange. Meanwhile, capacity development programs such as Biodiversity Information for Development, funded primarily by the European Union, have mobilized 703 datasets from Africa, Latin America and the Caribbean, and the Pacific islands in their first eight years, downloaded 1.8 million times and generating 715 publications and 34 Red List assessments.

As GBIF enters its next quarter century, the organization faces a paradox of success: the infrastructure needs are larger now than they were in 2001, driven by growing data volumes, novel data types, and a monitoring framework that places biodiversity data at the center of international policy. The 2023-27 Strategic Framework emphasizes growth in the community and expansion to new data types, informatics services, and data uses. The authors of the new overview hope that a historically informed understanding will strengthen the community’s capacity to respond to biodiversity data needs in the next twenty-five years and beyond, sustaining the vision of a world in which the best possible biodiversity data underpins research, policy, and decisions. For a generation of scientists who have taken GBIF’s services as a given, the message is clear: the open data machine behind global biodiversity science is not a static utility but a dynamic, growing network whose impact relies on the ongoing contributions of everyone who uses it.

Subject of Research: Biodiversity informatics and open data infrastructure

Article Title: The Global Biodiversity Information Facility at 25

Article References: Miller, J., Mandeville, C., Schigel, D., Gamboa Martinez, J., Nielsen, A. M., Sheldon, S., Hahn, A., Raymond, M., Bundgaard-Jensen, S., Bagard Laursen, M., Copas, K., Blissett, M., Høfft, M., Méndez Hernández, F., Noesgaard, D., Rodrigues, A., Russell, L., Stjernegaard Jeppesen, T., Elkjær Ørum-Kristensen, A., … Robertson, T. (2026). The Global Biodiversity Information Facility at 25. Biodiversity Data Journal, 14, Article e208528. https://doi.org/10.3897/bdj.14.e208528

Image Credits: AI Generated

DOI: 10.3897/BDJ.14.e208528

Keywords: GBIF, biodiversity data, open science, data standards, Darwin Core, Kunming-Montreal Global Biodiversity Framework, biodiversity informatics, Global, Biodiversity, Information, Facility, scientific research