NEWS & EVENTS

The BIOSPACE 2025 conference brought together experts in remote sensing (RS) and biodiversity research to explore the latest technological advancements and their applications for ecosystem monitoring. Members of the GUARDEN project actively contributed to the event, presenting groundbreaking results on AI-driven habitat mapping and biodiversity assessment through Earth Observation (EO) technologies.

One of the key contributions from GUARDEN was the presentation by César Leblanc and Théo Larcher of GeoPl@ntNet, an innovative AI-powered platform for habitat prediction based on species observations. The approach, which requires only species lists to infer habitat distributions, was met with great interest. Researchers from the Joint Research Centre (JRC) and the Thünen Institute (Germany) discussed potential collaborations to integrate GeoPl@ntNet outputs into the production of a custom habitat quality index, reinforcing the relevance of our work for conservation policy applications.

Private sector engagement also emerged, with actors expressing interest in using an API for targeted citizen science observations from GUARDEN tools. Additionally, academic researchers explored the potential of collaborating on AI-based biodiversity modeling, with a specific focus on species prediction using image-text embedding alignment. Throughout the conference, GUARDEN’s work sparked discussions on the scalability and explainability of AI models for biodiversity mapping. While attendees were impressed by the ability to generate large-scale habitat maps, questions arose about the applicability to tropical regions and the need for further model interpretability studies. The event also highlighted the growing role of AI foundation models in EO-based biodiversity research. Several works revealed that while EO foundation models such as DOFA, Prithvi, and SatDINO hold promise, their benefits vary depending on the task. This insight opens new avenues for GUARDEN to explore whether environmental foundation models could enhance habitat prediction accuracy.

The conference underscored the importance of aligning biodiversity research with policy frameworks. The 2024 EU Nature Restoration Law was highlighted as a critical reference, as it defines key biodiversity indicators to be monitored, providing a concrete opportunity for GUARDEN to contribute to policy-driven EO applications.

As the field of EO-based biodiversity monitoring evolves, GUARDEN continues to push the boundaries of AI, remote sensing, and conservation science, fostering interdisciplinary partnerships to enhance ecosystem management and policy implementation.