2025-2029 MASSIF

Le suivi de l’hyperdiversité entomologique se heurte à des verrous qui entravent son déploiement à large échelle. Compte tenu de l’importance de la diversité taxonomique et fonctionnelle des insectes forestiers et des pressions associées au changement global, il est impératif de déployer des systèmes de monitoring couplant le suivi de la biodiversité à long terme à des dispositifs de biovigilance assurant la détection précoce des espèces émergentes ou envahissantes. En complément des données des PC X-RISKS et MONITOR, le projet MASSIF développera des méthodes d'identification moléculaire à haut débit, ainsi que des outils optiques et photographiques pour le suivi des insectes forestiers, en s'appuyant sur les avancées récentes en intelligence artificielle (IA) embarquée ou non. Le projet sera organisé en 4 WP :

WP0 : gouvernance et diffusion des résultats et des connaissances ;
WP1 : développement de deux bases de données de référence : barcodes ADN et photographies de référence ;
WP2 : sélectionner, entrainer et évaluer des algorithmes d'IA pour la détection, l'identification et le dénombrement des espèces, en laboratoire et dans des systèmes embarqués sur le terrain ;
WP3 : développement du prototype du système d'acquisition d'images et d'analyse embarquée ;
WP4 : déploiement des prototypes sur le terrain, comparaison des performances des méthodes d'identification (expertise humaine, moléculaire et IA), développement d'outils d'analyse ;
WP5 : gestion sécurisée des données, développement d'outils numériques garantissant l'interopérabilité, diffusion et maintenance des bases de données et des outils statistiques.

Partenaires : INRAE CBGP, INRAE URZF, INRAE EFNO, INRAE DYNAFOR, iEES Sorbonne Université, MNHN MECADEV, CNRS SETE, INP-Toulouse LAAS, ONF LNEF
Financements : PEPR FORESTT, APR 2024

English translation
Project title
Automated Monitoring and Intelligent Surveillance Systems for Insect Biodiversity in French Forest Ecosystems

Summary
Forest ecosystems face several major threats that jeopardize their resilience and the maintenance of their functionalities. Establishing long-term monitoring observatories, both for taxonomic and functional biodiversity, is therefore crucial. This is especially true given that forests serve as habitats or refuges for numerous species. Insects, which represent the majority of biological diversity, are major contributors to numerous ecosystem services (pollination, organic matter cycling, trophic networks, pest regulation). The observed collapse of insect diversity in forests thus severely threatens the resilience of these ecosystems. Conversely, certain insects, whether native pests or invasive exotic species, pose an increasing threat to forest health under changing climatic conditions (ecosystem disservices). However, monitoring entomological hyperdiversity faces significant challenges that hinder its large-scale implementation. Given the importance of the taxonomic and functional diversity of forest insects and the pressures associated with global change, it is now imperative to deploy monitoring systems that combine long-term biodiversity tracking with biovigilance frameworks to ensure the
early detection of emerging or invasive species. Innovative technologies have potential to monitor insects at large spatial scales. This project will develop high-throughput molecular identification, as well as optical and photographic tools for
monitoring forest insects by leveraging recent advances for device-embedded and non-embedded artificial intelligence (AI). We will focus on functionally and patrimonially significant groups that are abundant and diverse, including guilds involved in ecosystem services or disservices. The proposed consortium, interdisciplinary in nature, involves partners with complementary expertise.
To achieve its objectives, the project will be organized into four thematic work packages (WPs) and two transversal WPs.

The first WP focuses on developing two reference databases essential to the project: a DNA barcode library and a photographic database centered on expertly identified specimens.
WP2 aims to select, train, and evaluate AI algorithms for species detection, identification, and counting, both under laboratory conditions and in field-embedded systems.
WP3 is dedicated to instrumentation and aims to deploy efficient traps and develop the prototype for image acquisition and embedded analysis (development of autonomous, non-destructive, and energy-efficient multi-sensor devices).
WP4 involves deploying these prototypes in the field, comparing the performance of identification methods (human expertise, molecular, and AI), developing analytical tools, and optimizing large-scale monitoring schemes (Renecofor plots, ONF networks).

The project will also include two transversal WPs.
WP0 will focus on governance and the dissemination of results and knowledge.
Linked to the NUM-DATA FP, WP5 will address secure data management, the development of digital tools ensuring interoperability with other national or international databases, the dissemination of databases and statistical tools, and the long-term maintenance of this infrastructure. Data sharing will adhere to FAIR principles. This project will provide the necessary tools to allow the monitoring of forest entomofauna and the surveillance of forest health, thus complementing the data collected within the frameworks of the X-RISKS and MONITOR FP.