TDWG 2016

In previous work we implemented a prototype of an ontology-based semantic enrichment and mediation system for flower-visiting data digitized from labels of flower-visiting insect specimens in natural history collections. This system transformed database records documenting physical specimens into enriched records of ecological events e.g. a FlowerUtilizingEvent or a FlowerProductUtilizingEvent. In subsequent work we created a probabilistic model (Bayesian Network) of the causal knowledge that an expert implicitly uses to interpret individual specimen records e.g. to assert that an insect was probably foraging for pollen or probably foraging for nectar.
The objective of the present work is to link interpretations of individual organisms behaviour (i.e. output from the Bayesian Network) to aggregated records of behavioral interactions. These aggregations are population samples which allow the data to be interpreted at a higher level of abstraction (population and community level) i.e. in terms of ecological relationships between population samples of different species. Using the widely adopted modelling construct of the Interaction Network (represented by an ontology), we modelled an ecological community as a network of interacting populations of different species. Each node in the interaction network is a population sample of a different species. Nodes are connected by edges representing ecological interactions, of which there are several types.
We envisage a system that will allow a user to filter database records by spatio-temporal extent so as to realistically model a network of co-existing populations. The user will then be able to adjust the level of precision of the visualised ecological interactions (e.g. ecologicalInteraction > foragingEcologicalInteraction > nectarForagingEcologicalInteraction).
Explicit semantics could bring a degree of standardization to the construction of interaction networks and the interpretation of flower-visiting data.

Several South African research institutes operate equipment to record marine underwater video footage for ecological monitoring of e.g. coastal fish stocks. We designed a local process to upload fish length and count data output from a Stereo Baited Remote Underwater Video (Stereo-BRUV) camera. Data were uploaded to a Specify database using the Specify Workbench, which required us to design a standard Workbench Template to be used by participating institutes. Working more broadly we mapped the fields in the Specify database to terms in the Darwin Core schema and Audubon Core schema to describe the data with a view to publishing the standardised data. We evaluated the richness of the standardised metadata obtained through the use of these vocabularies. Specific terms to describe BRUV data are unavailable. The development of such terms will aid the discovery, integration and interpretation of BRUV data.

In the current context of Biodiversity loss and climate change, it is more than ever necessary to adapt and develop our scientific practices to face these present-day global issues. Scientists, particularly biologists, have to define new protocols to optimize the tremendous amount of new data being generated and to analyse them.
Monitoring Biodiversity is a complex problem because of its multiple facets and cross-domains links. The creation and use of ontologies to conceptualize those different aspects of Biodiversity is an efficient means for key stakeholders and policies makers to promote consistency and reliability of systems.
For this purpose, the Environment Ontology (ENVO; http://www.environmentontology.org) is a community ontology for the concise and controlled description of environments. It is interoperating with other domain ontologies closely linked to the representation of biodiversity in order to better interface with efforts such as Darwin Core and initiatives to promote the achievement of the United Nations’ Sustainable Development Goals (SDGs).
As part of the ENVO consortium, BiGAEOn is an ontology for biogeographic areas specifically. Biogeographic areas are the basic units used in Comparative Biogeography to produce classifications of biogeographic areas, here, bioregionalisation. BiGAEOn model describe and harmonize biogeographic entities (e.g. areas of endemism, endemic areas…) as well as their relationships. Hence, it provides a rigorous and simple framework that improves biogeographic analyses and interoperability between systems.
In particular, BiGAEOn integrates formal descriptions of WWF ecoregions (http://www.worldwildlife.org). In this presentation, we will illustrate how our ontology fits current debates with a case study on Australia, since it’s the actual scene of the bioregionalisation revival.

Biodiversity data are complex and abundantly available and spread out over a multitude of repositories. These data can be classified as semi-structured and are organized differently, depending on the elicitor or the expert who generated the knowledge. This constitutes the problem of biodiversity data interoperability. To mitigate such problems and to improve knowledge acquisition, OntoBio was developed.
The methodology adopted for the development of OntoBio, uses explicit knowledge to define the ontological schema of the domain. Thus, the tacit knowledge of the domain is not considered during modeling and it is observed that much could be inferred and the scope of the modeled schema would be amplified if it were considered during the process of formalization. The incorporation of tacit knowledge to ontological schemas has the purpose of increasing the expressiveness of ontologies. This purpose has guided the development of a conceptual framework to incorporate semantics to formal ontologies through tacit knowledge.
The conceptual framework consists of the following steps: (1) knowledge elicitation; (2) knowledge formalization; (3) ontology matching; (4) recommendations for ontology evolution and; (5) analysis of the recommendations for ontology evolution. The application of the framework to OntoBio has produced two main outputs: (a) recommendations for evolution of the underlying ontology from the domain Expert Mental Model (EMM). In this research, EMMs refer to more specific fact situations, rather than more general phenomena. To each new elicited and formalized EMM, new recommendations for change are available. Ontology becomes a dynamic instrument of knowledge representation; and (b) to each EMM applied to the framework, a Progressive Formalization Schema (the knowledge of a domain may be presented at different levels of formalization, from text documents to explicit rules) is generated, allowing ontology engineers to revisit the elicited and formalized knowledge for further use. Also, it allows access to knowledge at different levels of granularity and minimizes the semantic losses that may occur at different levels of knowledge representation.
The steps (3) and (4) are under development and the tests carried out so far have covered the ichthyology domain. The next phase of the research, includes the design of an experiment to elicit scientific knowledge of strategic research groups at Instituto Nacional de Pesquisas da Amazônia (INPA), for example ornithology, and disseminate the EMMs. This implies that any new EMM should be mapped to OntoBio, resulting in improvements to the ontology.

Plant phenology — the timing of plant life-cycle events, such as flowering or leafing-out — has cascading effects on multiple levels of biological organization, from individuals to ecosystems, and is crucial for understanding the links between climate and biological communities.  Today, thanks to data digitization and aggregation initiatives, phenology monitoring networks, and the efforts of citizen scientists, more phenologically relevant data is available than ever before.  Unfortunately, combining these data in large-scale analyses remains prohibitively difficult, mostly because the organizations producing phenological data are using non-standardized terminologies and metrics during data collection and data processing.  The Plant Phenology Ontology (PPO) is a collaborative effort to help solve this problem by developing the standardized terminology, definitions, and term relationships that are needed for large-scale data integration.  In this talk, I will give an overview of the PPO, including the high-level design of the ontology, examples with real phenological data, and future development efforts.