Terrestrial Ecosystem Observatories

Overview This dataset contains a desktop-based inventory of existing/operational terrestrial ecosystems observatory infrastructure, conducted in November 2024. The aim of the dataset is to serve as an evidence base for analysing global terrestrial ecosystems coverage, overlaps, gaps and emerging trends in observation and data access/sharing. Definition and scope For the purposes of this exercise, terrestrial ecosystem obervatories were broadly defined as capabilities that collect data on the state or condition of terrestrial ecosystems, or their components, which are matinained for that purpose, and where the collected data is made available for research or broader use. It includes capabilities that directly observe ecosystems on the ground (e.g., sensor networks), as well as organisations that aggregate, curate and publish data specific to terrestrial ecosystems produced by others. The organisations included vary greatly in purpose—from repositories to standards bodies to observatories to policy-facing indicator platforms—so this dataset reflects a mixed ecosystem rather than a single type of infrastructure. Inclusion criteria Platforms or organisations offering observational or survey-based data products, data portals, or harmonised datasets. Also included 'thematic platforms' (focused on specific aspects of terrestrial ecosystems science such as vegetation, soils or freshwater water) and environmental reporting platforms or applications Only national and global infrastructures were included. Sub-national, state or regional observatories were excluded to maintain comparability, though some may still appear where scale was unclear/insufficiently described Types of data activities captured included data collection, aggregation, curation, standardisation initiatives and tools/platforms providing global environmental indicators. Specimen-only collections were excluded unless the collections were clearly tied to structured surveying activities Excluded: generic data repositories or data aggregators (collecting or publishing research data broadly, not limited to terrestrial ecosystems). Method The search was organised by country and used a combination of Google searches, ChatGPT queries, and snowballing. To scope the effort, ChatGPT was first asked to list the top five countries by population and GDP within each global region (e.g., North, Central and South America). Regions were used instead of continents to introduce more nuance and increase the diversity of findings. For each selected country, an initial Google search was conducted using the basic syntax “[Country name] AND [ecosystem observatory]”. Relevant results were retained. Additional rounds then used alternative keywords such as “terrestrial,” “observation,” “infrastructure,” and “data,” continuing until no new relevant results appeared. This flexible keyword approach helped reveal organisations using local or region-specific terminology. Where an organisation’s website mentioned other observatories or platforms, these were followed as additional leads (i.e., snowballing). After a country search was deeemed to be 'exhausted' via Google (not returning new relevant results), ChatGPT was used to uncover potentially missed capabilities. ChatGPT was also applied to produce English translations for the description section, where these were only available in other languages. Each potentially relevant result was evaluated manually (e.g., website visited and assessed for relevance) and, when confirmed, recorded in the spreadsheet including: Capability name English Translation (where applicable) Acronym (where applicable) Weblink Scope (Domestic, International) Region Country Capability type (Observation, Aggregation, Network) Short description. A visual depiction of the data included in this version is available here: https://embed.kumu.io/9f36e0fecf8683a00bd7b60926cc32c8 Biases and limitations Geographic This dataset should not be considered complete. While the intention was to compile a globally representative sample, the current version only covers North America, Central America, South America, East Asia, Southeast Asia and Oceania. Still pending are Europe, Africa and parts of Asia (Western and South). Where records for capabilities in other regions are listed, they either come from previous work (by Alison Specht) or were found opportunistically during other searches. Importantly also, many observatories undoubtedly remain unidentified—particularly in regions not yet surveyed or in countries where information is not readily accessible through English- or Spanish-language searches. Others are encouraged to continue or expand on this exercise. State- and region-level observatories were excluded in favour of national-scale infrastructures. While this improves comparability, some sub-national facilities may have been missed which nonetheless make significant contributions to global terrestrial ecosystems data. This dataset should be interpreted as a landscape map rather than an exhaustive global inventory. It is best suited for analysing structural patterns, identifying areas of duplication or fragmentation, and understanding the diversity of data-related roles across terrestrial ecosystem organisations. It should not be used to evaluate data quality, operational maturity, or completeness of observatories, nor does it rank organisations or assess their scientific performance. Search terms and method General search terms were used rather than domain-specific terms. As a result, broad multidisciplinary platforms were more easily identified than highly specialised, domain-specific observatories. The snowballing method used to identify observatories favours well-connected and widely-referenced organisations. Highly specialised facilities (e.g., domain-specific platforms such as AmphibiaWeb) may be under-represented, particularly in areas where more targeted search terms (e.g., soil, mycology, phenology) would have uncovered additional observatories. Many of the platforms identified are data repositories that accept any dataset submitted to them, while others focus narrowly on data associated with specific research papers. As a result, the level of curation, standardisation, and long-term stewardship varies widely. Language The search was conducted mainly in English, with Spanish used extensively for South and Central American countries (one of the authors is a native Spanish speaker). This introduces a clear language bias. The most likely areas where relevant infrastructures may have been missed are East Asia (e.g., China, Japan, Korea) and Southeast Asia (e.g., Indonesia, Malaysia), where some organisations may not provide English-language web content or are not well indexed by English-language search engines. However, the impact of this bias is partly mitigated by the fact that organisations intending to engage internationally or participate in “network-of-networks” initiatives typically maintain at least an English-language overview page. In this sense, the presence of English-language content can itself be seen as an indicator of readiness or intent to connect globally. If continuing this work in the future, we advise targeted searches are conducted with the help of native speakers in relevant languages, particularly for Asia and Africa. English would suffice for most of Europe, though local-language searches may reveal smaller or more specialised infrastructures. Expanding linguistic coverage would help identify observatories that serve primarily local or national audiences and are not positioned for global visibility. Author commentary Duplication of Functions (Commentary by I Ceron) A notable pattern emerging from the dataset is the extent to which organisations undertake multiple specialised functions along the terrestrial ecosystem data pipeline—often simultaneously. Many platforms act as self-contained, whole-of-pipeline capabilities, engaging in data aggregation, curation, standards development, storage, search and discovery, and analytics. This results in considerable duplication of effort at both national and global levels. Examples of common duplication include: Repository-heavy ecosystem. Many platforms function primarily as broad data repositories, accepting a wide range of submissions, including datasets associated with individual research papers. Competing standards efforts. Numerous organisations develop or promote data standards, often independently and with limited coordination across domains such as biodiversity, hydrology, soils, or vegetation. Multiple global reporting tools. A large number of platforms provide global indicators or status dashboards for specific environmental variables. These differ greatly in data provenance, underlying methods, and update cycles. This fragmentation has several implications. It weakens opportunities for specialisation and collaboration, as organisations that attempt to “do everything” often end up competing rather than partnering. It also leads to a proliferation of standards and a crowded landscape of discovery portals that users must navigate. The broader question for observatories worldwide is therefore not only what roles they should perform, but what roles they might intentionally stop performing in order to specialise and work more effectively with others. Put differently: the current landscape resembles a set of parallel, disconnected capabilities. A more integrated model—closer to a supply chain—could improve coherence, efficiency, and user experience. Harmonisation (Commentary by I Ceron) Terrestrial ecosystem data harmonisation remains limited and uneven. Most harmonisation happens within the boundaries of specific projects or bilateral partnerships rather than being designed into systems from the start. Few initiatives pursue global, cross-domain harmonisation as a default approach. GBIF is a notable exception, but similar global, standard-first efforts are rare outside the taxonomic domain.