
Introduction
For several decades, cyclical desert locust invasions have seriously threatened the livelihoods of millions of people in the arid and semi-arid regions of West Africa. Capable of migrating thousands of kilometers and devastating entire crops in just a few hours, the Desert Locust (Schistocerca gregaria) is recognized as one of the most dangerous transboundary pests for agricultural and agro-pastoral resources. To combat this scourge, countries in West and Northwest Africa, united under the Commission for Controlling the Desert Locust in the Western Region (CLCPRO)—a commission of the Food and Agriculture Organization of the United Nations (FAO)—have adopted a preventive control strategy, which requires precise, responsive, and continuous ecological monitoring capabilities.
As part of this effort, a regional project funded by the French Development Agency (AFD) and the FAO, titled “Strengthening the Foundations of the Preventive Control Strategy and Developing Operational Research on the Desert Locust in the Western Region”, aims to enhance the tools available to surveyors, improve locust risk analysis, and propose more reliable and harmonized intervention criteria.
Among the key actions of this project, strengthening botanical identification skills in the field has emerged as a vital priority. Locust risk assessment largely depends on the observation of habitats favorable to Desert Locust reproduction, characterized by specific plant communities. However, targeted intervention zones—often remote, impoverished, or affected by conflict—rarely have the technical or human resources necessary to maintain continuous botanical expertise. Therefore, equipping field surveyors with modern, offline-capable, reliable, and educational tools has become essential.
This is where the long-standing collaboration between CIRAD and CLCPRO comes into play. With its development capabilities, the Pl@ntNet consortium led a four-year initiative to adapt Pl@ntNet to the needs of the CLCPRO project.
The goal: To deploy a Pl@ntNet micro-project dedicated to the flora of Desert Locust habitats, based on the reference botanical work by Duranton, Foucart, and Gay (2012). This initiative produced an illustrated database of 237 species, trained an image recognition model combining performance and robustness, and provided field surveyors with a contextualized tool usable with or without internet access, on mobile or tablet.
Naturally aligned with citizen science, this initiative also helped spark a collective learning dynamic, local reappropriation of botanical knowledge, and field image capitalisation, with several hundred validated observations submitted by surveyors across the target countries. Training modules and tutorial videos facilitated regional uptake of the project. A look back at four years of work marked by collaborative AI, participatory science, and real on-the-ground impact.

1. A dedicated platform for a strategic flora
Objective: Adapt Pl@ntNet to the 237 plant species listed in the “Florule des biotopes du Criquet pèlerin” (Duranton et al., 2012), which are crucial for the reproduction and gregarization of the pest.
Achievements:
- Creation of a dedicated Pl@ntNet micro-project including 111,114 images and 82,901 botanical observations.
- Integration of a cutting-edge AI model trained on a wide range of global plant species.
Development of contextualized web and mobile interfaces, including taxonomic filters, distribution maps, and enriched ecological descriptions. Explore the micro-project: https://identify.plantnet.org/fr/clcpro/species
2. Duranton et al. (2012)’s Flora revisited and accessible online
To make the most of this reference work, the team:
- Structured species descriptions on the Pl@ntUse wiki: https://identify.plantnet.org/fr/clcpro/identify
- Added key metadata: IUCN statuses, traditional uses (based on the GRIN-USDA database), and multilingual common names.
- Enriched the visual database with historical illustrations and new field observations. Example: Androcymbium gramineum (Cav.) J.F.Macbr.: https://identify.plantnet.org/fr/clcpro/observations/1012373642
3. Offline identification: An asset for isolated areas
Aware of the logistical challenges in desert zones, the project innovated by deploying:
- An embedded mobile model that works offline (size: 507 MB), downloadable by all users with an account.
- Text and image data downloadable for all species in the flora.
Nearly 5,435 identification queries performed by 438 users since the project’s launch.
4. Training and field deployment
To support adoption of the tools developed, key actions were undertaken:
- 4 tutorial videos in French and English presenting the web and mobile versions were produced (available on YouTube:
https://www.youtube.com/watch?v=QxbGLPSGTzw&t=6s
https://www.youtube.com/watch?v=WlyIlnINUss&t=21s
https://www.youtube.com/watch?v=ZXlT8_cvWyA&t=115s
https://www.youtube.com/watch?v=3YI-ouhJ_Fg&t=19s
- An educational guide for field surveyors was produced.
- Several workshops were held, including a field training in Mauritania (October 2024) involving 16 surveyors from 14 countries, generating nearly 207 validated observations.
Pl@ntNet has transformed how habitats are mapped. By combining speed and reliability, it is helping evolve methods toward real-time decision-making.
Conclusion
This work, supported by FAO, illustrates the power of partnerships between citizen science and entomological expertise. By transforming botanical heritage into an interactive digital resource, Pl@ntNet and CLCPRO offer a reproducible model for managing arid ecosystems. Explore the micro-project and contribute too!
Access the “Desert Locust Habitats” micro-project: https://identify.plantnet.org/fr/clcpro/species
Watch the tutorial videos: https://www.youtube.com/@plantnetapp
Contacts and operational partners:
Cyril Piou – cyril.piou@cirad.fr, Scientific and Operational Lead of the project
Antoine Foucart – antoine.foucart@cirad.fr, Entomologist
Pierre-Emmanuel Gay – pierre-emmanuel.gay@cirad.fr, Cartographer, Co-authors of the reference work
Together, let’s continue cultivating innovation for resilient ecosystems.