Online taxonomic resources provide information about the taxonomic status of a taxon as well as synonymous relations, they facilitate the taxonomic data capture, help input data control processes and integrate information in other databases and infrastructures (Triebel et al. 2012). Despite almost all these taxonomic resources containing information about all living things, each has its own data domains, providers, scope of contents and user communities. For instance: occurrences and records in GBIF, taxonomic checklists and classifications in CoL, names in the Global Names Usage Bank (GNUB, http://www.globalnames.org/GNUB), taxonomy in marine species in WoRMS, zoological publications and nomenclatures in The Official Registry of Zoological Nomenclature (ZooBank, http://zoobank.org), knowledge data and multimedia objects in EoL, DNA barcoding sequences in International Barcode of Life (iBOL, http://ibol.org), etc.

Scientific names and taxonomies are of essential interest in major biodiversity platforms (Boyle et al. 2013, Metzger et al. 2013, Patterson et al. 2010) and are typically fed by individual scientists and institutions researching data. However, they may alternatively be supplied by primary data collected from other databases (Nozères et al. 2012) which creates a graph, in which taxonomic resources are the nodes and data flows the connectors, creating dependences between them. For example, EoL contains information from several taxonomic hierarchies like GBIF, GNA, iBOL, International Nucleotide Sequence Database Collaboration (INSDC), JSTOR Plant Science and Biodiversity Heritage Library (BHL), The International Plant Names Index (IPNI, http://www.ipni.org), etc.

In the future, data flows will be even more complicated due to the growth in the number of initiatives, infrastructures and collaborations that cover taxonomy and classification challenges. The new biology based on the big data world is envisaged as a discipline with a strong data-centric character and a growing role for informatics. The responsibility for managing data from many sources will probably carried out by modules or nodes that serve specified subdisciplines. The nodes will aggregate heterogeneous content within a particular subdomain, making it discoverable and available to end users (Patterson 2014, Thesen and Patterson 2011). An example that follows this model is the Global Names Architecture, which index, organize and interconnect on-line information about organisms and their names.

Today, one of the main challenges in bioinformatics is the implementation of the interoperability in an environment where interdisciplinary cooperation is key to scientific understanding (Berendsohn et al. 2011, Fielding 2000, Remsen et al. 2006. Data sharing is essential to fostering the collaboration and large-scale analysis needed for the successful treatment of the initiatives connected with biodiversity (Costello et al. 2013, Hine et al. 2003). Currently, there are two main proposals to achieve the interoperability desired across such systems and data; federating and brokering. In the federating solution (Heimbigner et al. 1985), the interoperability is achieved by a set of service buses based on the SOA architecture binding the user and the provider. Initiatives that follow this solution are based on federal specifications, covering data and metadata models, predefined vocabularies and ontologies, service interfaces and binding protocols. The Infrastructure for Spatial Information in the Europe (INSPIRE, http://inspire.ec.europa.eu) directive or the National Spatial Data Infrastructure (http://www.fgdc.gov/nsdi/nsdi.html) are two examples of such a federated approach.

However, in the brokering approach (Nativi et al. 2011) the heterogeneity is addressed by focusing on mediation rather than standardization. It proposes a User-Broker-Producer model by which the Broker consists of multiple support components facilitating the discovery of, semantic and natural language mediation, data access services, workflow process, and publishing. The brokering model has been adopted by the European approach to Global Earth Observation System of Systems (EuroGEOSS, http://www.eurogeoss.eu) project. Following the functional requirements, the brokering option is selected as all data providers use a different specification to interoperate. Therefore, a middleware (composed of brokers) interconnect clients and online taxonomic resources in a common infrastructure.

In the last years, some tools have already been developed for the exploration of distributed information about taxonomies. New advances in taxonomic publication processes are designed to speed information automatically to diverse users. One method dealt with a solution for special citation of taxonomic work when used in wiki pages by combining both the original non-wiki source and the respective wiki page (Penev et al. 2011). However, another approach sets out a workflow that describes the assembly of elements from a Scratchpad taxon page (http://scratchpads.eu) to export a structured XML file (Blagoderov et al. 2010). Methods of semantic tagging and semantic enhancements, text and data processing, publishing and dissemination in taxonomy have provided an increase in open access literature and journals aiming at rapid publication of taxonomic treatments, including new publication models such as semantically enhanced information (Penev et al. 2010). Also, the software package DELTA includes a data-basing program which stores morphological data for export in different forms and acts as a manager of taxonomic research (Coleman et al. 2010).

Recently, many organizations have developed different software tools to harvest, publish and share data (Katsanevakis et al. 2012, Kennedy et al. 2006, Wu et al. 2009). GBIF offers community tools to enable the integration of biodiversity data from heterogeneous sources using standards and protocols; an example is the Integrated Publishing Toolkit (IPT, http://ipt.gbif.org) which enables the publication of content in databases or text files using open standards Darwin Core (DwC, http://rs.tdwg.org/dwc) and the Ecological Metadata Language (EML, http://knb.ecoinformatics.org/software/eml/). The GBIF Spreadsheet Processor is a similar tool intended for smaller occurrence datasets stored in excel files. The limited number of concepts in Darwin Core offer a strength in usability but is weak for observational descriptions (Hill et al. 2010). The Biological Collection Access Service for Europe (BioCASE, http://www.biocase.org) has developed BioCASE Provider Software (BPS, http://www.biocase.org/products/provider_software), a middleware that facilitates the connection and mapping of data into XML schemes such as the Access to Biological Collections Data standard (ABCD, http://www.tdwg.org/standards/115/download/ABCD_v206.html). The main advantage of ABCD is the large set of concepts available relative to a species observation. This however, can be challenging to use owing to concept ambiguity. Biodiversity Information Standards (TDWG) is the organisation that works on defining standards in the field of biodiversity informatics. The most widely deployed formats for occurrence data are DwC and ABCD. The TCS (Taxonomic Concept Transfer Schema) was also developed for exchange taxonomic data but it defines only the taxonomic backbone.

A graphical representation of the design of the method is presented in Figure 2. It can be noticed that follows the steps of the brokering approach. Layers and modules are shown, which are necessary for the access, extraction, integration, analysis and representation of the data devolving from online taxonomic resources of the LifeWatch architecture (illustrated in Figure 1).

Figure 2. 

General structure of the service in correspondence with the LifeWatch infrastructure layers.

The information is retrieved from data providers and flows through the system from one layer to another. In each layer the information is filtered, selected and formatted to finally facilitate scientists in the analysis of specific species information.

The Data layer contains the distributed taxonomic resources that feed the application with their information. Each resource contains a specification of interoperable services which describe how to access the data.

The ICT-Core Middleware layer includes the Data Wrapper module which accesses the taxonomic resource, queries the information about the requested taxon and extracts the specific fields that the interface shows. Features such as the specification, metadata, request and response are in different resources. Thus, when a new taxonomic resource is added to the system, the module is modified and new ETL (Extract, Transform and Load) solutions are created to obtain the required information.

The Virtual Labs Middleware layer contains two modules. First, once the data from all online taxonomic resources is obtained, the information is joined together in the Data Integration module. This information facilitates the data management and data representation by the broker and the analytic tools respectively.

Second, the Brokering module manages all the data flow in the application using a broker, defining species concepts which are shown to the user. The broker converges disparate vocabularies and enables uniformity of search and access in divergent online taxonomic resources. It receives the name of a taxon from the user interface and calls the Data Integration module to query the taxon in all the taxonomic resources. Subsequently, the result is passed to the next layer for analytical purposes. Finally, the species information generated is sent to the graphical user interface.

The Virtual Labs Application layer includes all the analytical tools to support the visualization of the extracted data from online taxonomic resources using reports, graphs, tabs, rows, colors, etc., improving the exploration and the information driven-decisions.

Lastly, a graphical user interface shows the results with different options to analyze and download the information. The management of the taxonomic resources in the system is flexible, which means that the addition or deletion of a taxonomic resource only supposes the modification of the Data Wrapper module. To incorporate a new online taxonomic resource it is necessary to map the information retrieval from the resource to the specific concepts that the system manages (taxonomies, synonyms, valid and accepted names, etc.). Hence, there is an abstraction layer between the ICT-Core Middleware layer and the Virtual Labs Middleware layer where the implementation details of the taxonomic resources are not relevant to the rest of the design.

Following the previous design a taxonomic tool has been implemented to facilitate the exploration of taxonomies, accepted names and synonyms, using the information from five online taxonomic services; GBIF, CoL and WoRMS, ITIS and GNI. The last resource, GNI, is a compilation of all the various namestrings that have been used as scientific names, whether correctly or not, with variant spellings and mis-spellings. In this sense, GNI cannot be considered a source of taxonomic information as CoL or WoRMS.

The structure of the solution is represented in Figure 3, where three main elements are distinguished: the online taxonomic resources, the server containing the application that access and shows the information and the user who provides inputting species names.

Figure 3. 

Implementation of the service with the used technologies.

The server is the principal element in the figure as it provides fundamental infrastructure services. It accommodates the services implemented according to the LifeWatch requirements described in Section 2. These services are listed in four categories: Core Basic Services, Supporting Basic Services, Supporting Thematic Services and Specific Thematic Services. A more detailed list of services and its descriptions can be found in the LifeWatch Reference Model (Giddy et al. 2009). The presented tool is an approximation of the Taxonomy Access Service placed in the Supporting Thematic Services category. An example of such an application is Liferay web portal that contains the infrastructure website which is the service access point. In Liferay, applications are deployed in the form of portlets. We have developed a portlet that implements the search of taxonomic information using the previous design and a business intelligence solution called Pentaho (http://www.pentaho.com).

The three online taxonomic resources make up the current Data layer. They offer interoperable web services and interfaces that facilitate data queries from external applications and integrate such information into other systems. Due to the web services’ specification in any taxonomic resource differs, a new wrapper solution has to be designed to consult each of them. In the portlet, the Data Wrapper module is implemented by the Pentaho Data Integration tool (also known as Kettle). This tool permits the design of transformations, enabling ETL capabilities to form requests, process responses and locate information in each taxonomic resource. Three separate transformations are defined because both ways to query data and the XML structure in responses are different in these taxonomic resources.

The modules contained in the Virtual Labs Middleware Layer are implemented using different software tools. First, the Pentaho Data Integration is again used for the development of the Data Integration module. Once the information of each taxonomic resource is available separately, a new transformation is designed using this tool to join them together in a common XML file.

In the Brokering module, a Java program implements the broker using libraries. The program provides the link between the Analytic tool, the Data Integration, and the Data wrapper modules. When a user introduces the name of a species in the application, the broker sends the request to the Data wrapper module. To do this, it uses the libraries contained in the Pentaho Data Integration SDK to call the transformations which in turn implements the wrappers from the Java program. Subsequently, the Broker sends the information to the Data Integration module. As a result, a XML file contains the data from the three taxonomic resources using common labels.

Finally, the broker sends this file to the Analytic tool, through the libraries defined by the the Pentaho Reporting SDK, which consequently generates a report with the information.

The Analytic tool module is implemented using a report temfig designed by the tool Pentaho Report Design Wizard. This temfig produces a report in which the information is organized using dynamic tables (available in pdf, html and xls). The report is progressed to the graphical interface that illustrating the final results and available for download.

The application also has a data exportation option in compliance with the Darwin Core standard. Darwin Core is an internationally recognized standard for biodiversity data exchange, used by GBIF and other organizations to encode data related to organism names, taxonomies, references, etc. This option provides sufficient flexibility to support specific tasks, allowing advanced users to build custom applications tailored to particular needs (Chapman 2005).

The application is used by some researchers based in the Ecology Unit at the University of Salento (Lecce, Italy). The work is focused on experimental research in aquatic ecosystems. The implemented system has many benefits and enables the reconciliation of species information in different online taxonomic resources.

Firstly, the time that scientists spend researching a taxon has been drastically reduced. Currently, CoL, GBIF, WoRMS, ITIS and GNI all have different websites, interfaces and tools (which the reseacher would need to use). With the proposed system, users don’t need to consider the range of research methods as the search is combined including all online taxonomic resources. Furthermore, the application permits users to introduce a list of taxa showing the results in a crosstab report.

Secondly, we noticed that almost all the scientists base their research on two or more online taxonomic resources. The same taxon cannot appear in a taxonomic resource but can be classified with various synonyms in another. In some cases, especially in old species lists, the same taxon appears with various synonyms provoking confusion. This application permits the resolution of names and synonyms, consequently reducing the size of the lists and avoiding false results and conclusions.

Finally, the method is useful to scientists who work with new or recently-discovered species. In these cases, accepted names and synonyms change frequently. The application helps to find divergences in taxonomies and accepted names between online taxonomic resources.