Keywords

11.1 An Emerging Culture of Data Sharing, Publication and Citation

It has been widely accepted that the future usability and availability of research outputs, and specifically data, will be enhanced by proper description of these outputs using standardised metadata schemes, supplemented by deposit of the data in trusted repositories. Despite this, such outputs continue to be poorly described in practice. In addition, it is also commonly reported that the data supporting scholarly publication quickly becomes inaccessible or lost (Vines et al. 2014; Goddard et al. 2011). This disparity between what is seen as desirable behaviour, and reality is about to change, due to three significant drivers:

  • Data publication and citation is gaining momentum (Chavan and Penev 2011). For a comprehensive review, see the report by a CoDATAFootnote 1 Task Group (Socha 2013).

  • Funders are increasingly demanding the preservation of and continued open access to tax-funded research outputs.Footnote 2,Footnote 3,Footnote 4

  • Controversy in respect of reproducibility of scientific claimsFootnote 5 have led to insistence by journalsFootnote 6 that the data underpinning articles should be made available.

We believe these drivers will rapidly increase the availability of well-described, well-preserved, and sometimes standardised data services in the future.

11.1.1 Research Infrastructures

The drive towards data publication and citation requires support, hence the growth and proliferation of Research Data Infrastructures. These are supplemented strongly by voluntary, community-driven initiatives, and by member-funded bodies that support standardisation and interoperability.

Infrastructure operates on several levels: it provides governance and collaboration infrastructure (for example, the Belmont ForumFootnote 7 and Future EarthFootnote 8), architecture and standards infrastructure (e.g., Research Data AllianceFootnote 9—RDA, TDWG,Footnote 10 OGC,Footnote 11 GEOFootnote 12), and physical, centralised or federated infrastructure (GBIF,Footnote 13 EUDAT,Footnote 14 and GEOSSFootnote 15). Some global and regional initiatives span all of these (for example, the ICSU World Data System,Footnote 16 and GEO itself), and some are focused more narrowly on regional or domain-specific infrastructures (for example, DataOne,Footnote 17 EU BON,Footnote 18 Lifewatch,Footnote 19 and others).

It is worth noting that one of the motivations for the Research Data Alliance is to provide a cross-disciplinary, global exchange to minimise duplication of effort and divergence. Hence the landscape is at once characterised by divergent initiatives resulting from the nature of competitive grant funding and efforts to converge the impacts of funding these efforts. This is necessary, since divergence results in multiplicity of approaches, standards, protocols, and vocabularies—not supportive of interoperability.

11.1.2 Persistent Identifiers and Linked Open Data

Establishment of access to research outputs, either directly or via standardised services, requires a critical element: the ability to reliably find such objects in the web. This implies a persistent identifier, and several mechanisms are available to achieve this.

The biodiversity informatics community requires an identifier architecture that is capable of resolving two overlapping requirements—that of permanently identifying resources (data, services, and other web-based resources), and that of permanently identifying concepts (taxons, biomes, etc.).

There are several services available for either hosting or providing a minting framework for persistent identifiers (PIDs). Services that are general in nature, and allow hosting of PIDs on behalf of anyone, include the foundational Handle System.Footnote 20 This service can be used directly, but is also packaged and mediated, for example by the members of the International Digital Object Identifier (DOI) ConsortiumFootnote 21—allowing value-added services. DOI-based services that are important to our community include DataCite (linking published data sets and meta-data through DOIs to journal articles for purposes of citation tracking) and, CrossRef (more focused on linking DOI-based references across different journals), and GBIF (allocating DOIs for all published datasets and for search results). Several other biodiversity-focused initiatives exist, and these are discussed in the section on ‘Specific Implementation Guidelines’ (Barcode of Life,Footnote 22 Life Sciences Identifier, and similar, with identifiers.orgFootnote 23 providing an aggregation of such services).

The availability of persistent identifiers assists the construction of Linked Open DataFootnote 24 (LOD) networks—making a significant contribution to the Semantic Web.Footnote 25

11.1.3 Free and Open Data: Licensing and Policy

Delivering interoperable, open access to data and services involves (1) the implementation of applicable policies and (2) appropriate supporting licenses.

There are likely to be as many policies as there are data custodians and providers, but this is not really an issue as long as there is general compliance with the principles of free and open access—as documented by various global programmes such as the ICSU World Data System,Footnote 26 GEO,Footnote 27 and others.

Licenses, however, do need to be standardised, since machine-readability is a prerequisite for automated processing of data and services in the web. The most widely adopted candidates for this are the Creative CommonsFootnote 28 family of licenses. These have been tested in multiple jurisdictions. Note that issues still under discussion include:

  • ‘Legal Interoperability’ (how different licenses combine in automated processes, and what the resulting license is) (Uhlir 2013),

  • Conditions or exceptions to be added to licenses to address legitimate concerns in respect of privacy, ethics, publication embargoes, endangered species, and similar.

GEO BON, being part of GEO, will adopt the GEOSS Data Sharing Principles (currently under review and likely to be modified slightly). In short, these are:

  • There will be full and open exchange of data, metadata and products shared within GEOSS, recognising relevant international instruments and national policies and legislation;

  • All shared data, metadata and products will be made available with minimum time delay and at minimum cost;

  • All shared data, metadata and products being free of charge or no more than cost of reproduction will be encouraged for research and education.

11.1.4 Data Citation and Publication

Many of the institutional, technical, and legal hurdles that impeded the growth of data citation and publication have been addressed, and there is a broad consensus amongst journal publishers, data centres, and scientists in general on implementation (Socha 2013). CoDATAFootnote 29 and RDAFootnote 30 have played (and continue to play) a significant enabling role in this process.

Scientists should note that future research would be subject to:

  • Planning for deposit and description (through metadata) of research output in a Trusted Digital RepositoryFootnote 31—increasingly required by funders;

  • Allocating persistent identifiers to such outputs, as appropriate.

Global coordinated research programmes, such as Future Earth, also attempt to align their funded outputs with the requirements of free and open access, and to promote a culture supportive of data publication and citation.

11.1.5 Big Data, Citizen Science, Crowdsourcing, and Proliferating Sensors

The field of biodiversity observation and monitoring is subject to rapid change both in regard to the variety of sources and to the volume size of the data that needs to be described, visualised, understood, preserved, and processed. This is due to a number of interrelated factors:

  • Growing Diversity and Productivity of Observation Channels: Increasing availability of sensor channels lead to larger volumes of usable data. Traditional channels (remote sensing, gene sequencing, field observation) are increasingly supplemented by crowd-sourced observations, and the rapidly growing number of connected smart devices in the internet (Hugo et al. 2011).

  • Methods using automated markup for metadata and data mining of existing or future publications contribute to increasing volumes (Agosti and Egloff 2009).

  • Storing Observations: It is becoming increasingly affordable to store and process large volumes of data.

  • Less Expensive Platforms: It is becoming very affordable to deploy observation platforms such as aerial dronesFootnote 32 and underwater guided cameras, leading to large, multidimensional data sets at low cost of acquisition. Similarly, cost reductions are set to deliver significant and growing volumes of environmental genomic data addressing aspects of biodiversity which until now have been inadequately recorded.

These factors all combine to put pressure on the traditional architecture, standards, and infrastructure arrangements that have evolved to deal with a less demanding situation. The implications of this growth need to be accommodated in requirements for a scalable architecture.

11.2 The Network of the Future

GEO BON is by definition a network, and it is important to recognise that the concept of a network applies on multiple levels: on an institutional and personal level; as a collaboration network; and with the support of an infrastructure network. This infrastructure includes networks defined physically through protocols, schematically and syntactically through registries and catalogues, and semantically in emergent knowledge networks, ontologies, and vocabularies.

Any future networks, and resulting research data infrastructure, will likely be a combination of all of these and require governance, best practice conventions, standards, and reference implementations to work.

11.2.1 A Vision for Future Data and Services

The vision for a future network extends work done earlier by GEO BON (Scholes et al. 2012), and includes ideas about the generic use cases that it should support. This is summarised largely in the GEO BON ManifestoFootnote 33 (Hugo et al. 2013), which highlights a set of functions that are expected to be available. These, in turn, influence architecture and standards that are required to support such a network. The GEO BON Data Working Group (Working Group 8) has focused on these, and on developing a working implementation demonstrating the generic use case.

The Manifesto, as set out in updated form below, addresses description, discovery, assessment, access, analysis, and application or reporting, by stating that it is the interest of any specific community to do the following:

  • Ensure that scientific data and services are described properly, preserved properly, and discoverable;

  • Once discovered, the utility, quality, and scope of data can be understood, even if the data sets are large;

  • Once understood; the data can be accessed freely and openly;

  • Once accessed, the data can be included within distributed processes, and collated—preferably automatically (Hernandez et al. 2009a, b), and on large scales (the ‘Model Web’) (Nativi et al. 2013);

  • Once processed, the associated mediations and annotations, usefulness, and knowledge gathered can be re-used.

All of this needs to be implemented against the backdrop of:

  • Due recognition to the creators of the data, models, and services;

  • The push to extend formal metadata with Linked Open Data and persistent identifiers;

  • The increased availability of crowd-sourced and citizen contributions;

  • A proliferation of devices and sensors; and

  • The construction of knowledge networks.

11.2.2 The Role of Standards and Specifications

Standards and specifications are intended, from a formal systems engineering perspective, to reduce the risk of failure. The basic aim of this approach is ‘Predictable Assembly from Certifiable Components’ (Wallnau 2003). The risk of failure is lowered because assembly is made from components certified to meet the specifications and standards. In the type of scalable, open architecture envisaged for GEO BON, the ability of third parties to assemble larger systems from components using well-defined interfaces is critical as a contributor to the goal of interoperability and scalability.

Data standards in biodiversity are primarily defined by the Biodiversity Informatics Standards organisation. It is better known by its earlier name ‘Taxonomic Databases Working Group’Footnote 34 (TDWG). TDWG works with other standards bodies, such as Open Geospatial Consortium (OGC), and has been recognised by them.

11.2.3 A Scalable, Interoperable Architecture

A realistic, shorter-term expression of the goals implied by the manifesto can be summarised as follows (Saarenmaa et al. 2014):

  • Allow for data flow from observations through various aggregation and processing/modelling services, supporting evaluation of EBVs and derived indicators;

  • Automated and streamlined, as appropriate;

  • Using a plug-and-play (service-oriented) approach, supported by robust service provider organisations;

  • Coordinated through a GEO BON registry system and linked to the GEOSS Common Infrastructure;

  • Transparent to users through multiple channels, portals and applications.

11.2.3.1 General Requirements for a Biodiversity Information Architecture

Scalability, access, security, user concurrency and data reliability must be considered. For scalability, it is expected that tens of thousands of data sources will ultimately be integrated through GEO BON. They will be hosted in a smaller number of data repositories. Additionally:

  • The infrastructure must incorporate a federated architecture which will allow many data centres, initiatives, and infrastructures to co-exist and participate;

  • While a minimum set of standards is desirable, pragmatism and reliance on brokering and mediation will be the norm for a considerable time to come;

  • Human resource, financial, scalability, and institutional constraints will necessitate building the infrastructure using many small contributions in addition to a few large, global ones.

The main components in the information architecture can be divided into three main functions, corresponding to the tasks of (i) data publishing, (ii) data discovery, and (iii) data access. As a fourth function, various applications and uses can be envisaged, and for all functions mediation may be required between services and clients in cases where standardisation of services and vocabularies are not perfect.

There are two options for interoperability architecture, both essentially ‘service-oriented’, with varying degrees of rigour required for implementation. Firstly, the model proposed by EU BON and others, is based on an Enterprise Service Bus (ESB), and allows automation of asynchronous workflow and distributed processing as envisaged by the Model Web. Secondly, one can serve a significant proportion of needs with less complex synchronous orchestration, using mostly RESTful Services. These architectures are not mutually exclusive and are likely to co-exist within a systems-of-systems environment.

11.2.3.2 Option 1: SOA and ESB

The Service Oriented Architecture (SOA) is a model, which has achieved ‘best practice’ status within the Open Geospatial Consortium (OGC). Building on SOA has been recommended also for GEO BON (Ó Tuama et al. 2010) and EU BON (Saarenmaa et al. 2014). In an SOA, different functionalities are packaged as component services that can be orchestrated for specific tasks. An Enterprise Service Bus (ESB), which is a virtual private connector over the Internet, would connect external data sources using various SOA standards (WSDL,Footnote 35 SOAP,Footnote 36 RESTFootnote 37 and BPEL,Footnote 38 among others). The use of an ESB facilitates the interactions among data sources, working in a message-centred interaction and providing the ability to orchestrate web services through the use of workflow handling technology (e.g., Kepler,Footnote 39 TavernaFootnote 40).

11.2.3.3 Option 2: Synchronous, RESTful Services

Some applications do not require orchestration of services to take account of long-running, asynchronous processes, and may not require authentication if data services are in the public domain. In these cases, RESTful HTTP calls, stored in OGC Web Context Documents (XML files defining a collection of RESTful services and their roles) should be adequate to collate information in support of a user requirement. The role that each service plays to achieve the collective outcome will have to be captured for future use, and can potentially be stored in OGC Web Context Documents (XML files defining a collection of RESTful services and their roles), but other methods may also be used.

11.3 Considerations in Respect of Best Practice

11.3.1 Sources of Data and Its Classification

11.3.1.1 Essential Biodiversity Variables

The Essential Biodiversity Variables (EBVs) (Pereira et al. 2013), under development by GEO BON, provide a critical use case for determining requirements for information systems. An EBV is defined as ‘a measurement required for study, reporting, and management of biodiversity change’. EBVs provide focus in two important ways:

  • promote harmonised monitoring by stipulating how variables should be sampled and measured;

  • facilitate integration of data by acting as an abstraction layer between the primary biodiversity observations and the indicators.

For example (Fig. 11.1), we could build up an aggregated population trend indicator (for multiple species and locations) from an EBV which estimates population abundances for a group of species at a particular place and which, in turn, is derived from the primary, raw data which can involve different sampling events and methodologies.

Fig. 11.1
figure 1

An EBV acts as an intermediate layer between raw data and indicators

Full size image

GEO BON has identified six EBV classes. These are listed in Table 11.1 with some candidate EBV examples. By analysing the variables/measurements associated with each EBV, appropriate data standards can be proposed or recommended, or new and enhanced standards proposed. Of particular relevance are the EBV definitions and how an EBV is measured. For example, the three EBVs listed for the Species Populations class, can be broken down as illustrated in Table 11.2. In fact, the Species Population class EBVs are possibly the most tractable given the current status of biodiversity informatics, and could act as the initial test case.

Table 11.1 EBV classes with examples
Full size table
Table 11.2 The three EBVs of class species populations with their definitions and variables/measurements
Full size table

In addition to suitable data exchange standards, there is a need to identify appropriate communication protocols for messaging and data flow between systems, and, as part of the architecture design, how to automate the data flows for the EBVs.

The EBV on abundances and distributions would need to be measured using ‘counts or presence surveys for groups of species easy to monitor or important for ecosystem services, over an extensive network of sites, complemented with incidental data’. Such an EBV would be updated at intervals from 1 to 10 years. EBVs have not yet been implemented, but need to be piloted.

Implementation of these specific EBVs calls for integration of data from sites such as those of LTER, and other regular surveys, and from historical and recent data published through GBIF. Integration implies processing services that would compute abundance trends and changes in distribution for these two types of data: surveys and incidental. These are shown in Fig. 11.2 as ‘ecological’ and ‘occurrence’ domains. Software tools and web services are available to do these computations, for instance from the TRIM,Footnote 41 BioVeL,Footnote 42 and EUBrazilOpenBioFootnote 43 projects. Recent developments within GBIF include support for additional core data elements from survey data,Footnote 44 indicating the possibility of incorporating all of these data sources within a single access infrastructure.

Fig. 11.2
figure 2

The GEO BON vision of automated, streamlined data flow, end-to-end, from observations to Essential Biodiversity Variables (EBVs), using a plug-and-play service-oriented approach, coordinated through the GEO BON registry system and linked to the GEOSS Common Infrastructure, and transparent to users through portals. Source Hugo et al. (2013); modified by Hoffman et al. (2014)

Full size image

The computation of an EBV of this class involves data cleansing and normalisation and interpolation of values to offer a modelled data surface. Such EBVs could be visualised in a portal, which would allow selecting the data sources and species in question, showing the intermediate steps, and presenting the trend and change of distribution for individual species or whole groups of organisms.

11.3.1.2 Protocols for Observation

The two largest domains of biodiversity observation are specimen occurrences and biological (natural resource) surveys. The former is frequently based on sporadic, opportunistic collection or observation activity, while the latter consists of repeated sampling at known sites, locations and follows a known protocol from which quantitative estimates of abundance, and at times additional information, can be derived. Hence, the latter method is most appropriate for observing change, but the former can also be used, if the observations sets are large enough and sampling biases can be eliminated by computation (Ariño 2010). Data potentially available through both of these domains are very large. GBIF, which already represents the occurrence domain, currently has mobilised more than 15,000 data sets and is expanding to index and integrate data from survey datasets. ILTER, which represents the ecosystem monitoring domain, has 25,000 data sets. Both have the potential of growing at least ten-fold. In particular, for ecosystem monitoring, much data exists in government agencies for the environment, forestry, fisheries, and agriculture, which in many cases have not yet started any data sharing activities.

Biodiversity observation is unique in that for species occurrence, most observations are made by volunteers. The EUMON projectFootnote 45 estimates that 80 % of biodiversity monitoring data comes from volunteers. In Finland, for example, there are 60 different biodiversity monitoring programmes in which 250 person years are spent annually, and 70 % of this is voluntary work. This pattern is similar to some extent many other countries—a summary prepared based on a listing of such volunteer programmes is shown in Fig. 11.3. In the top 15 topics, only astronomy is unrelated to biodiversity.

Fig. 11.3
figure 3

Topics covered most often by Citizen Science Programmes (https://en.wikipedia.org/wiki/List_of_citizen_science_projects#Active_citizen_science_projects). Inset—distribution of GBIF observation data, a large proportion of which originates from volunteer contributions (http://www.gbif.org/occurrence)

Full size image

Volunteer contributions pose a special challenge in respect of introduction of observer bias and strict adherence to observation protocols, and may be used in special circumstances to derive additional EBVs (Kery et al. 2010; Hui and McGeogh 2014).

11.3.1.3 Generic Data Families

The GEO BON working group on data integration and interoperability has developed a classification of generic data families and their interoperability requirements (Fig. 11.4). Data families are grouped according to variations in their spatial, temporal and semantic coverages with each unique combination of these, supported by a vocabulary/ontology, is considered a generic data family. As an example: occurrence, genome, and ecosystem data families all include a reference to a particular place and time, but differ in that occurrence data also references a taxon, genome data references a sequence and ecosystem data references biological phenomena.

Fig. 11.4
figure 4

Example generic data families and interoperability requirements. The abbreviations are: S-DB: spatial database; WxS: OGC (Open Geospatial Consortium web services); O&M: OGC Observations and Measurements model; SOS: OGC Sensor Observation Service; CSV: comma separated value; DwC: Darwin Core. The leftmost boxes represent typical data families and their dimensions, the centre shows typical data storage technology, and the rightmost boxes typical services whereby such data is exchanged. Some data storage technologies support multiple service standards. Source Hugo et al. (2013)

Full size image

The different types of coverage (spatial, temporal and semantic) and their attributes are:

  • Spatial Coverage: XYZ

  • Temporal Coverage: T (continuous or near-continuous); t (discrete)

  • Topic or Semantic/Ontological Coverage

    • P: Phenomenon

      • mostly physical, chemical, or other contextual data

    • B: Biological

    • Tx: Species and Taxonomy (with some extensions)

    • Al: Allele/Genome/Phylogenetic.

The dimension of a sampling event or specimen applies to all data families.

11.3.2 Published Advice and Guidance

The recommendations from published material discussed here have been incorporated into the ‘Specific Implementation Guidance’ later in the chapter, as appropriate.

Recent advances in the availability of standards include the development of ‘Biological Collections Ontology’ (BCO) and the ‘Population and Community Ontology’ (PCO) (Walls et al. 2014)—bridging a gap in the availability of vocabularies derived from formal ontology to describe the collection of biodiversity data, and to formulate more complex relationships between primary data elements such as evolutionary processes, organismal interactions, and ecological experiments.

11.3.2.1 Research Data Alliance (RDA)

The Research Data Alliance (RDA) produces community consensus on important aspects of research data infrastructure in general, and includes representation from biodiversity and ecosystem data infrastructures.Footnote 46 This interest group envisages work in respect of name (vocabulary) services standardisation, with a focus on taxonomy, and the support of improved interoperability. In more general terms, RDA has recently endorsed its first sets of formal outputs, and some of these have a bearing on biodiversity informatics:

  • The Data Citation Working GroupFootnote 47 has produced a clear set of guidelines in respect of implementation of persistent identifiers for data sets.

  • The Data Type Registries Working GroupFootnote 48 aims to standardise the description of complex data types—which in principle includes the ‘data families’ that can be identified for GEO BON EBVs. This enables processes, visualisations, and other tools to reliably be linked to data services.

  • The Metadata Standards Catalog Working GroupFootnote 49 has produced a set of principles, and aim in future to develop a canonical set of metadata elements that can serve as a broker between different metadata schemas in use by communities.

  • The Practical Policies Working GroupFootnote 50 has published its first recommendations in respect of 11 important practical policies for repository management, based on a survey of the research repository community.

11.3.2.2 Global Biodiversity Informatics Conference (GBIC)

The Global Biodiversity Informatics Conference (Copenhagen, 2012)Footnote 51 assessed the state of Biodiversity Informatics across four focus areas (Understanding, Evidence, Data, and Culture), and provided a community consensus on the desirable futures for the elements in each of these focus areas (Hobern et al. 2012).

11.3.2.3 GEO Data Management Principles

The GEO Data Management PrinciplesFootnote 52 were adopted in short form by the organisation in April 2015, and in full form by the GEO Plenary in November 2015. The 10 principles deal with aspects of discoverability, accessibility, usability, preservation, and curation.

11.3.2.4 EU BON

EU BON published a review and guidelines for its proposed architecture (Saarenmaa et al. 2014) that contains a portfolio of recommendations. These recommendations (39 in all) are strongly supportive of existing projects and initiatives (Lifewatch, BioVEL, EBONE, INSPIRE, LTER, GBIF, to name a few) and provide guidance in respect of service-bus type implementation in a service-oriented architecture.

11.3.2.5 CReATIVE-B and GLOBIS-B

The CReATIVE-B projectFootnote 53 (2011–2014) dealt with the ‘Coordination of Research e-Infrastructures Activities Toward an International Virtual Environment for Biodiversity’. CReATIVE-B enabled collaboration between the European LifeWatch/ESFRI Research Infrastructure and other large-scale research infrastructures on biodiversity and ecosystems in other parts of the world. The project published an integrated Roadmap in 2014 and this serves as high-level guidance in respect of biodiversity infrastructure and data management activities.

GLOBIS-B has as its main aim the definition of research needs and infrastructure services required to calculate EBVs, and will do so by fostering collaboration between scientists, global infrastructure operators, and legal interoperability experts. GLOBIS-B has produced its first outputs, and a recent publication (Kissling et al. 2015) details thoughts on interoperability in support of EBVs. GLOBIS-B correctly identifies a scientific challenge (definition of EBVs) and a technical one (legal and information technology considerations) that need to be addressed.

11.3.2.6 EarthCube and DataONE

These are primarily US-based initiatives, though DataONE has participating data providers from outside the US, and EarthCube has formal collaboration with EU partners. DataONE publishes and maintains best practice in respect of data management,Footnote 54 which was reviewed for inclusion into our guidance, and EarthCube has recently published a roadmapFootnote 55 and a supporting architectureFootnote 56 that also contributed input by way of principles.

11.4 Specific Implementation Guidelines

References quoted in the following sections are available in the supplementary materials on the Springer Website. Supplementary materials are also hosted and maintained on the GEO BON website at http://dataintegration.geobon.org/guidance.

11.4.1 Recommended Data Management Approaches

This section proposes guidelines for biodiversity data management from three perspectives: that of (1) individual researchers, (2) institutions, projects, or initiatives (such as regional BONs), and (3) from the broader community and GEO BON’s perspective. It focuses on the information technology aspects of the challenge to provide an infrastructure in support of EBV calculation. The guidelines support both architectures described above.

For all of these end user categories, we recommend that

  • General guidelines in respect of data management be followed (Section A below, and elaborated in supplementary materials), with indications of deficiencies that may exist;

  • Specific guidelines to foster semantic interoperability are followed (Section B below). These are also supplemented by online materials and deficiencies are highlighted;

  • As a first choice, data be shared in global repositories that serve a specific data family and is well established (Section C below);

  • Other data be published and catalogued using widely adopted interoperable service standards and content schema—while recognizing that the community, and especially GEO BON, should play a role in extending such content schema where deficiencies exist (Section D).

Content schema and vocabularies in support of specific EBVs will be required once the community has adopted definitions—GEO BON has a critical role in developing these, and the GLOBIS-B project will make a direct contribution to this effort.

11.4.2 Section A: General Considerations

These considerations apply to all research data infrastructures (Table 11.3).

Table 11.3 General guidelines applicable to all research infrastructures
Full size table

11.4.3 Section B: Semantic Interoperability

Guidelines in respect of the use of name services (vocabularies, ontologies, and persistent identifiers), and development of a knowledge network as it applies primarily to biodiversity informatics (Table 11.4).

Table 11.4 Guidelines in respect of semantic interoperability
Full size table

11.4.4 Section C: Specialised Global Infrastructure

For some data types and families, it is best practice to publish data and make it available via established global infrastructures (Table 11.5).

Table 11.5 Guidelines applicable to data families for which global infrastructures exist
Full size table

11.4.5 Section D: Aggregators and Open Federated Infrastructures

The data families and types listed below are best published in a federated manner, using standardised service protocols and content standards, with reliance on aggregation of standard metadata implementations to improve accessibility. GEO BON might consider hosting its own metadata aggregator as a component of the GEOSS Common Infrastructure (Table 11.6).

Table 11.6 Guidelines applicable to data families for which distributed systems and federated access will apply
Full size table

11.5 Conclusions

Biodiversity informatics is inherently a global initiative. With a multitude of organisations from different countries publishing biodiversity data, the foremost challenge is to make the diverse and distributed participating systems interoperable in order to support discovery and access to data. A common exchange technology, e.g. the widely used XML or JSON over HTTP, may allow the syntactic exchange of data blocks, but participating systems also need to understand the schema and semantics of the data being delivered in order to process it meaningfully. Unless the data share a common reference model, the exchange implies brokering, mediation, or other semantic processing.

The challenge, then, from the perspective of GEO BON, is largely one of agreeing appropriate content (schematic and semantic) standards for the main data families appropriate to each EBV. This will not address all requirements, but should go a long way towards creating successful interoperability precedents and simplify the broadening of the scope of application.

11.5.1 What Is Already Achievable?

Researchers, institutions, and regional or global infrastructures or initiatives that follow the guidelines published in the chapter will already make an immense contribution to the components of an interoperable, federated system of systems as envisaged by GEO.

11.5.2 What Needs to Be Improved?

The guidance has indicated for each aspect what role GEO BON can play in coordinating the solutions to non-ideal situations and development of community-endorsed standards, and in general this remains a significant requirement.

If one considers the more specific goal of EBV interoperability: the majority of EBVs still need to be defined by the GEO BON community, and guidance in respect of interoperability standards and software to support these is dependent on these definitions. In practical terms, the tasks at hand are:

  • Review the guidance presented here as more EBVs are formalised;

  • Identify the main deficiencies in respect of the available interoperability standards that can be used for GEO BON supported EBVs across data families;

  • Define extended content standards for the major data exchange service protocols (IPT, OGC WxS, NetCDF, Sensor Observation Services), using patterns and resources that already exist;

  • Build mediation tools for mapping of non-standardised data sets, such as those found routinely in MetaCAT and PlantNet repositories, to services that are schematically and semantically interoperable; and

  • Build schematic translation tools to serve any content standard over any service syntax.

It remains unclear how large data sets will be made available and included into an interoperable, orchestrated workflow in an open, free environment—the costs and time involved in sub-setting and processing the data may prove to be prohibitive, and it should be appreciated that the concept of having a suite of EBVs available within a distributed, interoperable global system of systems is constrained in many countries by availability of data sets and resources to gather and maintain such data sets.

Despite these constraints, GEO BON hopes to make steady progress in respect of extending the scope of content standards and services that implement them—leading to a set of EBVs available to a variety of end users from a variety of distributed contributors.