Wildlife Ontology

Abstract

A simple vocabulary for describing biological species and related taxa. The vocabulary defines terms for describing the names and ranking of taxa, as well as providing support for describing their habitats, conservation status, and behavioural characteristics, etcCopyright © 2010 the British Broadcasting Corporation.

This work is licensed under a Creative Commons Attribution License. This copyright applies to the Wildlife Ontology and accompanying documentation in RDF. This ontology uses W3C's RDF technology, an open Web standard that can be freely used by anyone.

Introduction

The Wildlife Ontology is a simple lightweight ontology for publishing data about all forms of biological taxa, including phyla, families, and species. The terms in this ontology allow data to be published about:

The relationships between taxaTheir associations with specific habitats, their mode of life, as well as their specific behavioursWhether a taxon is endangered according to the IUCN termsTopic relations between web documents and multimedia objects that may feature a taxon...etc

The Wildlife Ontology was originally designed to support the publishing of data from the BBC Wildlife Finder application. This application provides access to a rich set of information and data about biological species, as well as pointers to BBC broadcast output that relate to these topics. The ontology should therefore complement the existing Programmes Ontology for describing TV programmes.

Whilst it originates in a specific BBC use case, the Wildlife Ontology should be applicable to a wide range of biological data publishing use cases. Care has been taken to try and ensure interoperability with more specialised ontologies used in scientific domains such as taxonomy, ecology, environmental science, and bioinformatics.

Rationale

The Wildlife Ontology has been designed with a particular set of priorities, and it is useful to review these to highlight specific decisions that were made during the modelling process and also to highlight where specific modelling was not done and where it is expected that more specialised ontologies will be used. The primary users of data published using the wildlife ontology are not expected to be taxonomic domain experts, so where necessary some trade-offs have been made to simplify naming and modelling to maximise use of data by the non-specialist. However it is hoped that the data published using this ontology can be easily mixed with data from other sources.

Taxonomic Concepts and Taxonomic Names

Individual species, and other tax, are not clearly defined concepts. The notion of what constitutes the definition of a species may change over time. A species may turn out to be simply a variant of another species, or may be promoted to the level of a genus. The defining characteristics of the members of a species may also similarly change over time. Another point of change is the taxonomic hierarchy: different viewpoints will exist as to the hierarchical organisation of taxa into different ranks; the ranks themselves are often sub-divided and re-grouped, reflecting the viewpoints of different disciplines.

From a modelling perspective this means that there can be no single complete universal description; some trade-offs will always be necessary. In creating this model a decision has been made to use only a simplified taxonomic hierarchy that features the primary ranks. Sub-divisions of ranks, e.g. sub-phylums, etc are not represented. The hierarchical relationship between individual taxa has also not be strictly defined, instead each taxa is linked to its higher ranks through dedicated properties. This potentially allows for some rearrangement in structure, but also supports publishing data in circumstances where data is not readily available about each level in the hierarchy.

In taxonomy, a distinction is also often made between a taxonomic concept and its taxonomic name. Taxonomic names have their own relationships and life-cyle which can be somewhat independent of the concepts to which they are applied. From an RDF perspective this means that taxonomic names should also be modelled as resources: names are not just properties of a taxonomic concept. Preserving this distinction means that data published using the Wildlife Ontology should remain interoperable with external sources. This is especially true when attempting to link to data associated with Life Science Identifiers (LSIDs) which are assigned independently to both taxonomic concepts and names.

Species as Classes vs Species as Instances

One perennial problem associated with modelling biological taxonomies using RDF is whether to attempt to model individual species as Classes, or whether to simply model species as instances of a generic Species class. The latter approach is simpler and avoids creating a huge ontology that attempts to model all biological organisms. Existing ontologies have taken different approaches to resolving this issue, some choosing one style, others another. At present there doesn't seem to be a consensus. With this in mind, the Wildlife Ontology adopts the simpler of the two approaches, i.e. modelling species as instances of a Species class, as this maximises interoperability with many of the existing Linked Data sources, particularly dbpedia, which adopt similar approaches.

Existing Work

During the development of the Wildlife Ontology existing work on modelling and publishing RDF data about species descriptions was reviewed. These are summarised below:

• Biological Taxonomy Vocabulary
• The Evolutionary Trees and Natural History Ontology
• Geospecies Ontology
• TDWG Ontology
• Uniprot Core Ontology

These existing vocabularies vary considerably in their approach to modelling species and taxa, particularly around the representation of hierarchies, and differentiation between taxonomic names and concepts. While many of these vocabularies are heavily used within specific research projects, they are in various stages of development and adoption outside of the originating project seems to be low.

The decision was therefore made to create a new lightweight vocabulary, the Wildlife Ontology, to provide a simple, easy to understand vocabulary that could be reused by non-domain experts. This addresses the immediate goals behind opening up the data from the BBC Wildlife Finder application. The approach taken in the design of the vocabulary, as noted in the above rationale, has been to maximise interoperability with these existing vocabularies using Semantic Web infrastructure, e.g. stating equivalencies between classes, properties and instances.

Ontology Diagram

The following diagram illustrates the relationships between the key classes in the ontology. A number of classes, e.g. sub classes of TaxonRank, Habitat and Adaptation have been omitted for clarity.