NEWS & EVENTS

On May 13–14, 2025, the DeepSDM conference brought together researchers, data scientists, and biodiversity experts at Agropolis Fondation in Montpellier for two rich days of exchange focused on deep learning applications in species distribution modeling. Organized as a joint event by the GUARDEN and MAMBO Horizon Europe projects, the event attracted over 65 in-person participants and a total audience of more than 120, including those attending remotely via Zoom.

After the symposium opening by Alexis Joly, the stage was set for a full day of high-level presentations and engaging discussions. The conference explored the potential of Deep Species Distribution Models, a new generation of tools that apply deep learning techniques to predict species distributions based on complex environmental and occurrence data. These models go beyond traditional methods by capturing non-linear interactions, offering promising avenues for biodiversity monitoring and conservation.

The morning sessions featured talks by leading researchers such as Devis Tuia from EPFL and Florian Hartig from the University of Regensburg, who provided a strong conceptual and methodological foundation. This was followed by a second session chaired by Pierre Bonnet (Cirad, Pl@ntNet), with presentations from Philipp Brun and Benjamin Deneu (WSL), and Damaris Zurell (University of Potsdam), the latter joining via videoconference.

In the afternoon, the program continued with contributions from Sarah Si-Moussi and Selene Cerna (LECA), followed by a joint presentation from Alexis Joly and Diego Marcos (Inria, University of Montpellier), who shared insights from ongoing research and tools developed within the GUARDEN and MAMBO projects.

The day concluded with a dynamic round table discussion titled “What can deep learning offer to species distribution models for biodiversity knowledge and its conservation?” Moderated by Christophe Botella and Diego Marcos, the session combined a short series of case study presentations with over an hour of open dialogue. Panelists included Clovis Grinand (Nitidae), David Mouillot (University of Montpellier), Alexis Joly (Inria), Philipp Brun (WSL), Laura Manocci (IRD), and Olivier Gimenez (CNRS). The exchange with the audience was particularly rich.

The second day, held on May 14, was dedicated to a collaborative study day where a small group of participants explored tools, datasets, and modeling approaches in more detail.

With its strong attendance, stimulating discussions, and the high quality of presentations, this first edition of DeepSDM was a clear success. It demonstrated the growing interest in AI-based methods for biodiversity research and reinforced the value of interdisciplinary collaboration. Recordings and materials from the event will be made available soon on our communication channels.