Persistent Identifiers

Statement of Purpose

This page was originally created to expand the scope of content from the section on Persistent Identifiers as they relate to Numbering Natural History Collections.

Introduction

As we increasingly digitize specimens and share digital data about our collections, Persistent IDentifiers (PIDs) are often assigned to digital specimen records and can also be used to reference other elements of collections data, such as people or taxa. PIDs are foundational elements of data infrastructure because they enable automated and semi-automated linking between concepts, and also help make data FAIR, or "findable, accessible, interoperable and reusable" (see FAIR Guiding Principles F1 and A1).

What is a Persistent Identifier?

When we talk about Persistent Identifiers (PIDs) we assume that they are:

1. Unique. Unlike a catalog number, which may be locally unique, a PID must be unique on the global scale in order to ensure that the object it identifies can be unambiguously referenced. The need for uniqueness means that PIDs must be generated programmatically rather than created by human logic.
2. Persistent. Once assigned to an object, a PID should never change. PIDs also should not be deleted or reassigned, although in some circumstances a PID may refer to an object that no longer exists. "Never" is still relative; the current systems we use to manage PIDs are expected to have a lifespan of anywhere from decades to centuries.
3. Computer readable. PIDs are designed primarily for use by computers, not humans, although some PID schemes do have components that are meaningful to humans.
4. Resolvable. Generally, we expect that a PID can be reliably resolved to a meaningful information about the identified object, e.g., the PID "http://n2t.net/ark:/65665/3af2b96d2-a8a1-47c5-9895-b0af03b21674" is an actionable URL that redirects to a specimen record on an institutional web portal.

A Universally Unique IDentifier (UUID)–more generically known as a Globally Unique IDentifier (GUID)–is an identifier that is unique, persistent, and computer readable but not resolvable. For example, "3af2b96d2-a8a1-47c5-9895-b0af03b21674" is a UUID and "http://n2t.net/ark:/65665/3af2b96d2-a8a1-47c5-9895-b0af03b21674" is a PID. UUIDs/GUIDs have been widely used by natural history collections to identify digital specimen records in the Darwin Core term dwc:occurrenceID because of how easy they are to acquire via online tools such as https://www.uuidgenerator.net. Although UUIDs/GUIDs will continue to be used and can persist as useful identifiers, the community is moving towards a preference for using true PIDs to reference digital objects such as specimen records. See below for a list of commonly used PID formats and their applications in our domain.

Truly reliable, long term resolvability can be a difficult quality to achieve. Registration agencies are the social infrastructure that govern and maintain resolvability for various PID schemes. For example, DataCite is a registration agency that mints Digital Object Identifiers (DOIs). See Hardisty et al. 2021 for a thorough discussion on what resolvability means and an example of how the European DiSCCo project evaluated PID options for use by its member institutions.

Types of Identifiers

PIDs (and other identifiers) can be assigned to different types of objects within the realm of natural history collections. Although we often think of them in relation to digital specimen records (see Numbering Natural History Collections), PIDs are also useful when assigned to people, organizations, taxonomic concepts and names, geographic places, etc. See the table below for examples of what types of identifiers are most commonly used where.

Table listing identifiers, both persistent and not, that are commonly used in the natural history collections community.

Identifier name	Usage	Registration/resolution agency	Format & Example value	Comments
DOI (Digital Object Identifier)	any digital object	e.g. DataCite or CrossRef	Handle (DOI) https://doi.org/10.5962/p.304567	DOIs are a widely used identifier format with multiple registration and resolution agencies. Learn more at the International DOI Foundation (https://www.doi.org/index.html). Specimens could be assigned DOIs, but the most common applications for DOIs in our community are digital publications and other documents.
ARK (Archival Resource Key)	any physical or digital object	ARK Alliance	[URI]+"ark"+[namespace]+[alphanumeric object identifier] http://n2t.net/ark:/65665/3af2b96d2-a8a1-47c5-9895-b0af03b21674	ARKs are decentralized, persistent identifiers with the aim of being resolvable via a separate service, N2T (https://n2t.net). ARKs are a popular solution for identifying collection objects in the museum world. The ARK Alliance is not a true registration agency, as ARKs are created in a decentralized system. Many ARK users are larger libraries or museums with the capacity to mint their own ARKs, although EZID (https://ezid.cdlib.org) is one example of a registration service provider. Often used as a value in the standard terms dwc:occurrenceID or abcd:unitGUID.
IGSN (International Generic Sample Number)	physical samples	DataCite	Handle (DOI) https://doi.org/10.21384/AU1234	IGSNs are functionally DOIs that are registered in a namespace dedicated to physical samples. Learn more at https://support.datacite.org/docs/about-igsn-ids-for-material-samples. Often used as a value in the standard term dwc:materialSampleID.
GUID (Globally Unique Identifier)	any object	N/A	[alphanumeric object identifier] fb51580402c14e03bfe0c69437e180ec	A GUID is just a 128-bit integer number generated by an algorithm that is unique enough to make the risk of collision (i.e. generating the same GUID twice) null. A GUID is more generic version of a UUID. Anyone can generate GUIDs for free via online tools such as https://www.guidgenerator.com. A GUID is not inherently resolvable. Often used as a value in the standard terms dwc:occurrenceID or abcd:unitGUID.
Universally Unique IDentifier (UUID)	any object	N/A	[alphanumeric object identifier] fb515804-02c1-4e03-bfe0-c69437e180ec	A UUID is a type of GUID where the format is a string of 32 hexadecimal digits displayed in five groups separated by hyphens. UUIDs should be Version 4. Anyone can generate UUIDs for free via online tools such as https://www.uuidgenerator.net. A UUID is not inherently resolvable. Often used as a value in the standard terms dwc:occurrenceID or abcd:unitGUID.
CETAF Stable Identifier	specimens	Consortium of European Taxonomic Facilities (CETAF)	[URI]+[alphanumeric object identifier] http://herbarium.bgbm.org/object/B100277113 or http://www.botanicalcollections.be/specimen/BR0000005516339 or http://data.rbge.org.uk/herb/E00421509	CETAF makes recommendations on the identifier format and qualities, but is not a true registration or resolution agency. Learn more at https://cetafidentifiers.biowikifarm.net/wiki/Main_Page. Often used as a value in the standard terms dwc:occurrenceID or abcd:unitGUID.
Wikidata QID (Q IDentifier)	general purpose	Wikidata	[URI]+[numeric object identifier] https://www.wikidata.org/wiki/Q43649390	Wikidata QIDs can be useful to reference concepts that do not have another, more authoritative source, e.g. a person who collected specimens in the early 1900s but never published anything. Anyone can create or edit Wikidata records, and the data associated with a QID should be expected to constantly evolve. Often used as a value in the standard terms dwc:recordedByID, dwc:identifiedByID, and/or dwc:georeferencedByID.
ORCID (Open Researcher and Contributor ID)	people	ORCID	[URI]+[numeric object identifier] https://orcid.org/0000-0001-6514-963X	ORCIDs use the same format as ISNIs and are used in similar circumstances, except in regards to their creation: whereas asn ISNI may be created on behalf of someone else, an ORCID should only be created by the individual it refers to. If you do not already have an ORCID, you should create one for yourself! Often used as a value in the standard terms dwc:recordedByID, dwc:identifiedByID, and/or dwc:georeferencedByID.
ROR (Research Organization Registry identifier)	organizations	Research Organization Registry	[URI]+[alphanumeric object identifier] https://ror.org/03pnyy777	ROR is the next generation of a similar identifier type called GRID (Global Research Identifier Database). RORs are designed to identify the top level institution, e.g. a university, and so can be difficult to apply to collections which may be, e.g. a department within a university.
ISNI (International Standard Name Identifier)	people, organizations	e.g. British Library	[URI]+[numeric object identifier] https://isni.org/isni/0000000028856450	ISNIs are a widely used identifier format for people associated with creative works, and there are multiple ISNI registration and resolution agencies. Learn more from the ISNI International Agency, (https://isni.org). Often used as a value in the standard terms dwc:recordedByID, dwc:identifiedByID, and/or dwc:georeferencedByID.
MRGID (Marine Regions Geographic IDentifier)	geography	Marine Regions	[URI]+[numeric object identifier] http://marineregions.org/mrgid/48539	Often used as a value in the standard term dwc:higherGeographyID.
Thesaurus of Geographic Names Identifier	geography	Getty Thesaurus of Geographic Names	[URI]+[numeric object identifier] http://vocab.getty.edu/tgn/7007923	Getty Vocabularies includes several subject areas, of which the Thesaurus of Geographic Names is most relevant to natural history collections. Often used as a value in the standard term dwc:higherGeographyID.
GeoNames ID	geography	Geonames	[URI]+[numeric object identifier] https://www.geonames.org/6697173	Unclear how stable these identifiers are. Often used as a value in the standard term dwc:higherGeographyID.
N/A	stratigraphy	Macrostrat	[URI]+[numeric object identifier] https://macrostrat.org/sift/#/strat_name/490	Unclear how stable these identifiers are. The numeric object identifiers appear to be correlated with those also used by the U.S. Geologic Names Lexicon. Often used as a value in the standard term dwc:geologicalContextID.
LSID (Life Science IDentifier)	life sciences	N/A	[authority]+[namespace]+[numeric object identifier] urn:lsid:example.org:specimen:12921	LSIDs were designed to provide a mechanism for the life sciences community to create identifiers that did not rely on any one protocol (e.g. HTTP) for resolution, and that any entity could create (i.e. no need for a registration agency). Although existing LSIDs are still in use, they have been widely deprecated as a format and new usages are not recommended.
AphiaID	taxonomy	World Registry of Marine Species (WoRMS)	LSID (Life Science IDentifier) urn:lsid:marinespecies.org:taxname:527882	LSIDs are not actionable but can be resolved through a predictable URL translation, e.g. https://www.marinespecies.org/aphia.php?p=taxdetails&id=527882. Often used as a value in the standard term dwc:taxonID.
AphiaID	taxonomy	Interim Register of Marine and Nonmarine Genera (IRMNG)	LSID (Life Science IDentifier) urn:lsid:irmng.org:taxname:1166601	LSIDs are not actionable but can be resolved through a predictable URL translation, e.g. https://www.irmng.org/aphia.php?p=taxdetails&id=1166601. Often used as a value in the standard term dwc:taxonID.

Assigning PIDs

How to assign PIDs is a decision that should require careful consideration. See Richards et al. 2011 for a practical discussion on assigning PIDs that continues to be relevant even a decade past its writing (in particular, Section 5. Checklist for Implementing Persistent Identifiers).

It is best practice for PIDs to be assigned by the authoritative source, e.g. the institution who created and will manage in perpetuity a digital specimen record, or an individual themselves for a PID referencing a human. It is also important to consider what the identifier is actually representing, e.g. a physical object vs. its digital surrogate, and to document this decision. For example, the European DiSCCo project has determined that in their context PIDs will be assigned to identify the "digital representations of physical specimens", and a report on PIDs from the Research Data Alliance also recommends assigning them to digital rather than physical objects.

Keep in mind that you can also make use of PIDs that you did not assign. A common example of this is the use of ORCIDs to identify people, like collectors or identifiers, associated with your specimen data.

Contributors

Erica Krimmel

Further Reading

McMurry JA, Juty N, Blomberg N, Burdett T, Conlin T, Conte N, et al. 2017. Identifiers for the 21st century: How to design, provision, and reuse persistent identifiers to maximize utility and impact of life science data. PLoS Biol 15(6): e2001414. https://doi.org/10.1371/journal.pbio.2001414