Air quality, preventing risks from plants, and conserving tropical forests: how Pl@ntNet collaborates with partners around the world

Posted on by

We often say it: Pl@ntNet is much more than a simple plant identification tool. Around the world, and in collaboration with professionals from a wide range of fields, we work to preserve biodiversity, prevent risks to human health, and develop useful and innovative approaches for research. Today, we would like to share three examples of these collaborations!

ATMO Occitanie: preventing allergy risks

Atmo Occitanie and Pl@ntNet collaborate on monitoring allergenic pollen and improving air quality information in the Occitanie region. This initiative is part of the missions entrusted to France’s approved air quality monitoring associations (AASQA), which are key organizations responsible for monitoring air pollution and, since 2016, pollen in France. The aim is to improve knowledge of the flowering periods of allergenic plants and better anticipate situations that may pose risks to exposed populations.

To achieve this, Atmo Occitanie makes use of citizen observations collected through Pl@ntNet, particularly observations of flowering plants. Combining botanical identification with the date and location of observations makes it possible to track the development periods of the species concerned and obtain localized information on their presence and flowering. This data source complements traditional pollen monitoring systems and the other information used by Atmo Occitanie to produce its forecasts.

The initiative covers 22 plant species and groups identified by allergists as presenting specific concerns in Occitanie, including plane trees and cocksfoot grasses.

The collaboration has notably led to the creation of the “Flowering Observatory”, which encourages citizens to photograph allergenic plants when they are in flower. Observations collected through Pl@ntNet can be used by Atmo Occitanie to monitor flowering patterns across the region and gradually enrich pollen forecasts. The resulting information is then made available to the public, in particular through the regional interactive map showing allergenic pollen forecasts and air quality, available on the Atmo Occitanie website for the current day and the following day, as well as through a dedicated newsletter providing daily pollen forecasts.

The collaboration is therefore based on the complementary expertise and data mobilized by the two organizations. Pl@ntNet provides large-scale citizen observation capabilities and geolocated data on plant species and their developmental stages, while Atmo Occitanie makes use of these data within its expertise in air quality and pollen monitoring, as well as environmental and health information. The use of these data therefore helps improve knowledge of allergenic plants in Occitanie and enhance the information provided to exposed populations.

FREDON: detecting and raising awareness about plants that pose risks

FREDON Bretagne is an association recognized by public authorities as a Regional Plant Health Organization. Established in 1985, it operates at the regional level in the fields of plant health, environmental protection and human health, following a “One Health” approach—that is, an approach that considers human, animal and environmental health as interconnected. Its missions include monitoring and managing health risks associated with plants, as well as supporting local stakeholders.

As part of its work, notably in collaboration with the Regional Health Agency, FREDON Bretagne monitors several introduced plant species that have become invasive and may have impacts on the environment and human health. These include common ragweed, giant hogweed, jimsonweed and American pokeweed.

It was in this context that FREDON Bretagne integrated Pl@ntNet data into its monitoring systems. In 2024, Pl@ntNet provided FREDON teams with dedicated access to its data, giving them an additional source of observations of species presenting health-related concerns. This access was accompanied by a significant increase in the number of reports, leading FREDON Bretagne in particular to develop a dedicated tool, “AlertEspèce”, to organize and monitor these reports.

This experience illustrates the value of participatory science data in complementing existing monitoring systems and facilitating the detection of occurrences that may require intervention.

This approach is particularly valuable for species for which early detection can facilitate management efforts. In 2023, for example, FREDON Bretagne recorded 40 new common ragweed sites, 2 new giant hogweed sites, 21 new American pokeweed sites and 12 new jimsonweed sites!

AlertEspèce now enables individuals and professionals to consult known sites across Brittany and submit new reports from a smartphone by providing the species, a photograph and the location of the observation.

The experience with FREDON Bretagne illustrates a broader approach that we are seeking to develop at Pl@ntNet: promoting the operational use of our data and tools. By drawing on observations produced by our users and on systems that make it possible to use them in professional contexts, Pl@ntNet helps bring participatory science closer to concrete needs—whether in support of biodiversity conservation, botanical research, the development of ethical uses of AI, or, as in this case, the monitoring and management of risks associated with plants.

Tropical forests: improving our knowledge of and protecting these biodiversity treasures

Pl@ntNet’s work is not limited to issues and territories in France. For many years, the platform has been part of an international dynamic, collaborating with scientific and institutional partners as well as biodiversity managers in different countries. This international dimension is a fundamental part of Pl@ntNet and has resulted in long-standing collaborations addressing biodiversity knowledge, monitoring and conservation challenges.

These collaborations are particularly important in the field of tropical forests, which are major reservoirs of biodiversity while also concentrating some of the most complex challenges in terms of knowledge and conservation. Several hundred, or even several thousand, plant species can coexist within relatively small areas. Improving our knowledge of this diversity is nevertheless essential in the face of today’s and tomorrow’s environmental challenges.

Yet observing and monitoring these environments on a large scale remains particularly difficult, notably because of their accessibility, complexity and the considerable volume of data that needs to be processed.

These challenges are compounded by a “taxonomic bottleneck”: put simply, the reliable identification of a large number of plant species still depends heavily on highly specialized expertise that is available to only a very limited number of people. Experts capable of carrying out these identifications are therefore in very short supply and cannot process all the data generated in the field or by new data acquisition technologies such as drones on their own.

In this context, protecting these forests requires monitoring systems that are not only accurate, but above all scalable. Pl@ntNet is working to evolve its tools so that they can process large volumes of data and meet the operational needs of people working in the field. The collaborations around Barro Colorado Island in Panama, the Area de ConservaciĂłn Guanacaste in Costa Rica, and in Central Africa through the One Forest Vision (OFVi) initiative are part of this effort.

The aim is not simply to provide another identification tool, but to co-develop, with researchers and managers of these remarkable biodiversity areas, image analysis processes adapted to tropical forest monitoring.

This approach is based on the complementarity of several levels of observation. Field data provide reference observations, while drone data make it possible to observe ecosystems at much larger scales. AI models are a key solution for connecting these two levels, using knowledge acquired in the field to progressively automate the identification of a much larger number of images: a true transition to scale!

As part of the OFVi project, Pl@ntNet is working in particular on identifying tropical trees from drone imagery. This approach aims to facilitate the processing of large volumes of images and generate information on forest composition at large scales. Work around Barro Colorado Island, meanwhile, explores the connection between canopy observation and artificial intelligence.

Of course, artificial intelligence does not replace the expertise of botanists; rather, it makes it possible to leverage and multiply that expertise. Identifications carried out by specialists provide an essential foundation for training, evaluating and improving AI models. This complementarity between human expertise, field data and AI models therefore makes it possible to leverage the knowledge produced by specialists to process much larger volumes of data.

Through these collaborations, Pl@ntNet seeks to work with research and conservation stakeholders to build robust systems for monitoring tropical biodiversity. The challenge is to move beyond the limitations of occasional inventories and take advantage of new observation capabilities, particularly those offered by drones. By connecting field data, remote sensing and artificial intelligence, this work aims to make more systematic and accurate vegetation monitoring possible, in order to improve our understanding of ecosystems and support conservation and restoration strategies in the face of the biodiversity crisis.

Front cover phots credits: