ClassOccurrence

The dwc:Occurrence class

The Occurrence class represents instances where the presence of an organism at a location is documented. It functions primarily as a node that connects IndividualOrganisms to Events as well as to evidence that supports the record. In DSW, the Occurrence class differs somewhat from the Darwin Core standard (DwC) because some terms in the DwC Occurrence class are placed within the Specimen class of DSW.

Definitions of Occurrence

The Occurrence class is perhaps the most difficult to define of any of the classes included in darwin-sw. Occurrence resources are at the intersection of several other classes and depending on where the boundaries occur, various components may or may not be included in the definition.¹ There is a general consensus that an Occurrence is related to the events, individual organisms, and the evidence used to support the Occurrence. This relationship has resulted in several definitions of Occurrence. An Occurrence can be formally defined as "a tuple of (Individual, Event), and has properties for referring to the Individual and the Event." ² (A tuple is defined as an ordered list of elements and has a technical meaning in the programming community.) An alternative way of expressing this using terminology of the database world would be to say that an Occurrence is a "a many-to-many join between Events and Individuals, with a few additional properties".³ Since darwin-sw is an RDF representation of these relationships, in RDF terminology an Occurrence is a node that is connected by object properties to instances of the Event and IndividualOrganism classes.

¹ http://code.google.com/p/darwincore/issues/detail?id=70

² http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001782.html

³ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001785.html

Scope of Occurrence

Part of the confusion about the meaning of "an occurrence" stems from the historical use of the term to refer both to a primary unit of data gathering (i.e. that an individual of a taxon occurred at a particular time and place) and also to the result of modeling (i.e. we know from aggregate records that a taxon occurs or ever occurred, in a particular geographic area).¹ This second use has been referred to as the "checklist" use of occurrence² (i.e. indicating the presence of a taxon in an area on a list of taxa present) and is different enough from the first type of use that it has actually been described as a separate class: "Distribution".³ However, it has been pointed out⁴ that the "checklist" use of Occurrence is actually related to the first use if the relationship

TaxonConcept <--> Occurrence <--> Event <--> Location

is collapsed to

TaxonConcept <--> Location

The argument has also been made⁴ that the difference between these two views is simply a matter of scale and on that basis it may not make sense to create a distinction between the two.⁵ As a practical matter, there seems to be a fairly clear distinction between properties that should be applied to the first type of use (e.g. dwc:recordedBy) and those that should be applied to the second type (e.g. dwc:occurrenceStatus). For that reason, in darwin-sw instances of the Occurrence class are considered to fall within the first type of use (which is consistent with the definitions provided above), while the second type of use is not included in the current model.

¹ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001631.html

² http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001633.html

³ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001650.html

⁴ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001638.html

⁵ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001642.html

What is included within the boundaries of the Occurrence class?

Although most references to occurrences carry the connotation of a taxon representative documented by evidence at a place and time, there has been considerable variation in which of these aspects of an occurrence are assumed to be included as a part of the Occurrence resource itself.

The definition of Occurrence as a subclass of dctype:Event (http://purl.org/dc/dcmitype/Event) in the Darwin Core type vocabulary¹ carries the connotation that an occurrence resource includes the event at which the resource was recorded/observed/collected.² (Note: in Darwin Core Decision-2011-10-16_6 at http://rs.tdwg.org/dwc/terms/history/decisions/index.htm the subclass declarations of terms in the Dawin Core Type vocabulary have been removed.) However, this understanding does not seem to be widely held and in darwin-sw we make a clear distinction between Events and Occurrences, declaring them to be disjoint classes (i.e. a reasoner would detect an error if a resource were typed as both an Event and an Occurrence).

More widely held is the understanding that an occurrence resource also includes the evidence which serves as the documentation that the occurrence happened.³ In particular, at the time that the Darwin Core standard was ratified specimens were considered to "be" Occurrences, not just to serve as evidence for Occurrences.⁴ This view is reflected by the inclusion of terms which refer to specimens (e.g. dwc:preparations⁵) within the Occurrence class of DwC⁶ and the definition of the Darwin Core type PreservedSpecimen as a subclass of Occurrence.¹ However, this view was not universally held.⁷ Considering Occurrence resources to include the evidence that supports them introduces serious difficulties when one tries to describe them in RDF,⁸ so in darwin-sw we separate resources that represent the record of the presence of an organism at a certain time and place (instances of the dsw:Occurrence class) from the evidence (instances of the dsw:Token class) that supports that record. See the ClassToken wiki page for more information.

¹ http://code.google.com/p/darwincore/source/browse/trunk/rdf/dwctype.rdf

² http://lists.tdwg.org/pipermail/tdwg-content/2010-November/001828.html

³ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001681.html

⁴ http://lists.tdwg.org/pipermail/tdwg-content/2010-November/001836.html and http://lists.tdwg.org/pipermail/tdwg-content/2009-June/000441.html

⁵ http://rs.tdwg.org/dwc/terms/index.htm#preparations

⁶ http://rs.tdwg.org/dwc/terms/index.htm

⁷ http://lists.tdwg.org/pipermail/tdwg-content/2009-October/000291.html

⁸ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001742.html

Relationship of "observations" to Occurrences

In earlier drafts of Darwin Core, separate terms were defined for properties of observations and properties of specimens. The creation of the Occurrence class was in part a recognition that there was a fundamental similarity in metadata which described observations and metadata which described specimens. In the DwC standard, a single term in the Occurrence class, such as dwc:recordedBy, replaces separate terms (e.g. collector¹ or observer) that apply to specimens and observations but which fundamentally represent the same thing. Taking the approach that a Token is separate from an Occurrence leads to the question of the nature of the Token that supports an "observation"². One approach is to define observations as Occurrences that do not have tokens.³ Another is to define a class of Token called something like "Observation" in recognition that even observations have evidence, although that evidence may not be as accessible as an image or a specimen.⁴ ⁵ An advantage of this approach is that it would allow Observation class Tokens to be assigned identifiers and described using existing vocabularies for Observations.⁶

At the present time, darwin-sw does not take a position on defining such a class but simply assumes that a Token be some kind of resource that can be identified by an URI and assigned an rdf:type that is well-known. For example, an observation could be supported by a foaf:Document or dctype:Dataset that is accessible via an HTTP URI.

¹ http://rs.tdwg.org/dwc/terms/history/versions/index.htm

² "Observation" suffers from the same problem as "occurrence" in the sense that it is vague in the absence of a clear definition. In fact, one could take the position that "observations" are essentially what darwin-sw asserts Occurrences to be, i.e. the record that a taxon representative was present at a certain time and place apart from the evidence (Token) that supports that record. See http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001743.html for a historical reference to this outlook. It is also possible to consider an "observation" to be a type of dctype:Event. However, this fails to make a distinction between "observation" as an act (i.e. sensu Event) and "observation" as the product of that act (i.e. sensu Token). See http://lists.tdwg.org/pipermail/tdwg-content/2009-November/000263.html for discussion of this distinction.

³ Biodiversity Informatics 7:17-44. https://journals.ku.edu/index.php/jbi/article/view/3664

⁴ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001743.html

⁵ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001755.html

⁶ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001736.html

Separating the use of Occurrences in documenting species distributions from the declaration of a resource as an instance of the class Occurrence

The use of the term "occurrence" to describe the kinds of things included in the dsw:Occurrence (==dwc:Occurrence) class may lead to the misconception that all instances of the class must serve the purpose of documenting a "valid" occurrence of a taxon in its "natural" environment.¹ However, it has been made clear that there is no expectation that an instance of the Occurrence class must be suitable for modeling species distributions, only that it represent the presence of a particular taxon representative at a particular time and location. It is not necessary that the location, time, or taxon necessarily be known² but the Occurrence cannot refer to a generic representation of a taxon such as a diagram or generic illustration.³ In DwC, the appropriate term for describing whether an Occurrence is a "natural occurrence" (i.e. indicating that the Occurrence is suitable for modeling a taxon's distribution excluding captive or cultivated Occurrences) is dwc:establishmentMeans⁴. However, extensive discussion⁵ of the use of dwc:establishmentMeans revealed that there was substantial disagreement about what should be permissible as the controlled vocabulary for that term and even the DwC class under which the term should be listed⁶. The conclusion (if there actually was one!) was that the term dwc:establishmentMeans implied a property that was associated by complex relationships with several existing DwC classes⁷ ⁸ ⁹. Thus it is probably impossible to assign a domain to dwc:establishmentMeans which is any existing DwC or darwin-sw class.

¹ http://lists.tdwg.org/pipermail/tdwg-content/2009-October/000282.html

² http://lists.tdwg.org/pipermail/tdwg-content/2009-October/000280.html

³ http://lists.tdwg.org/pipermail/tdwg-content/2009-October/000289.html

⁴ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001612.html

⁵ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001611.html and many subsequent posts in that thread.

⁶ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001629.html

⁷ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001644.html

⁸ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001646.html

⁹ http://lists.tdwg.org/pipermail/tdwg-content/2010-October/001731.html