Ecological Metadata Language (EML)

Ecological Metadata Language (EML) is an XML-based metadata specification for documenting research datasets — and associated literature, software, or protocol resources — in the earth and environmental sciences. It is maintained as an open, community-governed specification (current stable release EML 2.2.0, 2019) originating from work at the National Center for Ecological Analysis and Synthesis (NCEAS) and the Long Term Ecological Research (LTER) Network, and is developed collaboratively via its public GitHub repository. EML is the metadata standard used across the LTER Network and by the DataONE federation of earth- and environmental-science data repositories, as well as by the Knowledge Network for Biocomplexity (KNB), the Environmental Data Initiative (EDI), and other environmental data repositories.

An EML document is organized as a set of modules rather than one monolithic schema, so a data provider uses only the modules relevant to its resource. Together the modules capture: resource-level identity and citation information (title, creators, publication date); the structure and content of the dataset itself — its constituent data tables, spatial rasters/vectors, or other entities, down to individual attribute names, units, and missing-value codes; the methods and protocols used to collect the data, including sampling design; and its taxonomic, geographic, and temporal coverage. Later EML releases added modules for semantic annotation (linking fields to formal vocabularies), structured funding information, and machine-readable license/rights statements.

How EML differs from Darwin Core

EML and Darwin Core are complementary standards that operate at different levels of the same dataset and are frequently used together, not competing alternatives. EML describes a dataset as a whole — who created it, how it was collected, what its component files/tables contain, and what geographic, temporal, and taxonomic extent it covers. Darwin Core, by contrast, is a vocabulary of terms (scientificName, decimalLatitude, eventDate, and similar) for standardizing individual occurrence, taxon, or sampling-event records within a dataset. The relationship is made explicit in the GBIF/TDWG Darwin Core Archive (DwC-A) format: an archive bundles one or more Darwin Core-formatted data files (e.g. occurrence.txt) with a separate EML document (conventionally eml.xml) that supplies the dataset-level metadata for that archive. In practice, a biodiversity dataset published through GBIF, DataONE, or a similar network typically has exactly one EML record describing the package, wrapped around potentially millions of Darwin Core-formatted rows.

For guidance on choosing between EML, Darwin Core, DataCite’s schema, and other options when documenting a dataset, see How to Choose a Metadata Schema for a Dataset. For a discipline-general (rather than ecology-specific) dataset metadata schema, compare the DataCite metadata schema, which EML does not replace or compete with — the two are typically used for different kinds of repositories and disciplinary contexts.

References

• Ecological Metadata Language specification and schema documentation (eml.ecoinformatics.org), maintained via the NCEAS/eml GitHub repository.
• GBIF, “What is Darwin Core, and why does it matter?” and the GBIF Integrated Publishing Toolkit (IPT) Darwin Core Archive how-to guide, on the role of EML (eml.xml) alongside Darwin Core data files within a Darwin Core Archive.