Often considered as our “last frontier”, the oceans cover about 71% of our planet and concentrate an essential part of the biodiversity. Studying biodiversity and its evolution is therefore a major scientific and societal challenge to meet the challenges posed by climate change and contribute to the United Nations Sustainable Development Goals (SDGs).

The Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) and the Intergovernmental Panel on Climate Change (IPCC) believe that data structuring remains one of the major obstacles to knowledge of biodiversity. Given the complexity of interactions between ecosystems and the heterogeneity of data, current indicators remain limited in predicting changes in biodiversity, particularly in the marine environment.

AI has a significant potential to address this challenge. By proposing new solutions for collecting, structuring and using data, AI can provide new resources for drawing up a current picture of the state of biodiversity and predicting its evolution. When applied to the marine environment, these innovations can help advance the fourteenth SDG, which promotes the conservation and sustainable use of marine and coastal ecosystems.

Based on this observation, a call for projects “Challenge IA-Biodiv” was launched in March 2021. This call is intended for the AI and biodiversity scientific communities to pool their expertise. Projects will have to meet three objectives:

– Optimize AI methods to improve marine biodiversity research,

– design innovative predictive models and indicators,

– develop hybrid AI methods to enhance our knowledge of marine environments.

This international challenge is part of the dynamics of the “research” component of the national strategy for artificial intelligence (SNIA) launched in 2018 to strengthen France’s position in this sector. It responds to a real challenge of shared innovation by spurring logics of co-construction of new tools.

THE THREE WINNING PROJECTS OF THE IA-BIODIV CHALLENGE

Following the evaluation and selection process mobilizing a panel of independent international researchers, three projects led by multidisciplinary and international teams of researchers specialized in the field of AI and biodiversity have been selected:

• The AIME (Artificial Intelligence for Marine Ecosystems) project, funded by Expertise France, is being carried out by a Franco-African consortium: Cadi Ayyad University (Morocco), the University of Yaoundé (Cameroon), the Laboratory of Medical Imaging and Bioinformatics and the Gaston Berger University of Saint-Louis (Senegal), the UMR Tropical Marine Ecology in the Pacific and Indian Oceans (New Caledonia, France), the Institute of Research for Development (IRD), and the Marine Biodiversity, Exploitation, and Conservation Institute (France)
  This project aims to address three scientific challenges: (1) combining or hybridizing AI techniques to improve the accuracy and precision of biodiversity indicators; (2) developing indicators, at multiple scales, that capture various aspects of the health and pressures on marine ecosystems; and (3) integrating them into an AI model capable of explaining and predicting the spatiotemporal dynamics of marine biodiversity in case studies. The AIME project will provide valuable tools to support decision making in coastal marine ecosystem management strategies.

• The SMART-BIODIV project (Artificial Intelligence Technologies for Biodiversity Research) is led by a French consortium: the Georgia Tech-CNRS International Joint Unit in Lorraine, the Central Supelec Loria, the Interdisciplinary Laboratory of Continental Environments and the Villefranche Oceanography Laboratory. It will develop new methods for managing and integrating data on the biodiversity of coastal marine areas using machine learning algorithms to fill in missing data and build indicators adapted to assess the biodiversity of the areas observed. This project also proposes to make available to the scientific community large datasets consisting of millions of images of plankton organisms.

• The FISH-PREDICT project (Predicting the Biodiversity of Reef Fishes) is led by a French consortium: the Marine biodiversity exploitations and conservations, the Montpellier Laboratory of Computer Science, Robotics and Microelectronics, the Center for Functional and Evolutionary Ecology, the Laboratory of Technical Sciences of Information, Communication and Knowledge, as well as by the Alpine Ecology Laboratory. This project aims to generate ecological indicators and predictive models of the biodiversity of disturbed ecosystems by combining artificial intelligence methods with known evaluative approaches. It will thus create the first knowledge base of marine biodiversity and, subsequently, develop prediction and interpretation models. It also aims to unveil intelligent solutions for nature to ensure the sustainability of coastal socio-ecological systems.

THE CHALLENGE, AN ORIGINAL MODALITY OF COLLABORATIVE RESEARCH

At a time when global challenges are becoming more and more numerous and complex, it is essential to innovate and encourage the sharing of knowledge in order to find appropriate answers. The Challenge, a specific funding instrument of the French National Research Agency (ANR), makes it possible to respond to these challenges by simultaneously exploring different scientific or technological approaches to a single problem and by pooling solutions. It compares and shares the approaches and work of the funded consortia, and encourages the establishment of common references between scientific communities from different backgrounds.

The IA-BIODIV Challenge will be led by the operational consortium (COpé), composed of the Laboratoire national de métrologie et d’essais (LNE), pilot of the COpé, the Foundation for Research on Biodiversity (FRB) and the Muséum national d’Histoire naturelle (MNHN) via the infrastructure of the National Biodiversity Data Pole (PNDB)

Thus, the three selected research teams will work on the elaboration and development of their project in a collaborative manner and will meet at the start of the Challenge on February 23, 2022. Their work, planned by Copé over 4 years, will focus on the coastal marine environment in the Mediterranean Sea during the first two years and will concentrate on the Pacific Ocean during the last two. Based on common data sets, the teams will each address specific and complementary issues on marine biodiversity.

The COpé will play a central role in the organization and the smooth running of the challenge: as a trusted third party, it will ensure the scientific animation, the evaluation of AI systems, the constitution of datasets and their access via the “AI-BiodivNet” environment. The three research consortia will collaboratively exploit and enrich this digital environment throughout the Challenge. They will have access to the intensive computing resources of GENCI (Grand équipement national de calcul intensif). The platform will eventually be accessible to all scientists and researchers and will contribute to the dissemination of sustainable data sets for future scientific work in AI and in the field of biodiversity.