MAMBO: New Technologies to Better Monitor European Biodiversity

The European project MAMBO is developing a new generation of tools to monitor species and their habitats. Artificial intelligence, acoustic sensors, automated cameras, drones, satellites, and LiDAR are combined with citizen science and field surveys. For Pl@ntNet, the project paves the way toward identifying entire plant communities and high-resolution mapping of species and habitats.

Monitoring biodiversity is essential to protect threatened species, restore degraded environments, and assess the effectiveness of environmental policies. However, data across Europe remain highly uneven: some regions and taxonomic groups are well documented, while others are far less so. Traditional field surveys also demand significant time, resources, and taxonomic expertise.

Funded by the Horizon Europe program, MAMBO (Modern Approaches to the Monitoring of Biodiversity) brings together ten European partners across ecology, computer science, remote sensing, citizen science, and the social sciences. Its goal is not to replace naturalists, but to provide them with tools capable of automating routine tasks, analyzing large datasets, and extending monitoring efforts to understudied species and regions.

Monitoring Species Through Image and Sound

Many species can be detected from a single photograph or acoustic recording. Leveraging recent advances in deep learning, MAMBO is developing several complementary technologies:

  • Image recognition models for animals and plants;
  • Acoustic models to identify birds, bats, and certain insect groups;
  • Automated cameras to monitor nocturnal insects and pollinators;
  • Automated image analysis tools for vegetation survey plots (quadrats).

These systems can collect standardized observations over extended periods, including in remote or hard-to-reach areas. They also help document ecological groups that are vital to ecosystem functioning yet notoriously difficult to monitor at scale, such as insect pollinators.

Pl@ntNet at the Core of Automated Vegetation Monitoring

The project builds on Pl@ntNet’s extensive expertise in automated plant identification and citizen science data management. Pl@ntNet contributes at several key levels:

  • Its identification engine is tested under real-world ecological monitoring conditions;
  • Its models and datasets are used to develop new vegetation analysis methods;
  • Its API enables seamless integration of plant recognition features into third-party tools and platforms;
  • Its geolocated occurrence data help model species distributions and map habitats.

Through MAMBO, Pl@ntNet continues its transition from a tool focused on individual plant identification to a robust infrastructure capable of describing plant communities and supporting habitat monitoring.

Recognizing Plant Communities in a Single Photograph

Ecologists often study vegetation using defined sample plots on the ground, known as quadrats. Within these plots, they record the species present, their relative abundance, and their ground cover percentage. Repeated over time, these surveys provide valuable insights into changes occurring in grasslands, sand dunes, or wetlands.

While highly informative, this method requires considerable field time and advanced botanical skills. MAMBO is therefore developing an automated approach to analyze quadrat photographs. Unlike standard Pl@ntNet queries—which typically focus on a single specimen—this new tool evaluates an entire scene containing multiple species to detect the various plants present and estimate cover.

From Plant Observations to Habitat Maps

The co-occurrence of specific plant species can signal a dry grassland, wetland, heathland, or forest habitat. MAMBO leverages plant community composition as a bridge linking field observations to large-scale habitat mapping.

In synergy with the GUARDEN project, experiments conducted within MAMBO contributed to the development of GeoPl@ntNet, a spatial platform for plant biodiversity mapping. These maps represent statistical predictions based on available occurrences and environmental variables. While they do not replace on-the-ground inventories, they help guide field surveys and highlight priority areas where new data collection is most needed.

Demonstrations Across Europe

MAMBO’s tools are being field-tested across diverse European ecosystems and climatic zones:

  • Mediterranean habitats in France;
  • Mols Bjerge National Park in Denmark;
  • Calcareous grasslands of Salisbury Plain in the United Kingdom;
  • Oostvaardersplassen Nature Reserve in the Netherlands;
  • The agricultural site of Friedeburg in Germany;
  • Comino Island in Malta;
  • The Petrohan site in Bulgaria.

At these demonstration sites, automated cameras, light traps, weather stations, and vegetation plots are deployed. Automated approaches are benchmarked against traditional field surveys to evaluate their accuracy, potential biases, cost-efficiency, and operational value.

Open Tools for Research and Conservation

MAMBO adheres strictly to Open Science principles. Datasets, software, machine learning models, and processing pipelines are made publicly available following FAIR principles (Findable, Accessible, Interoperable, and Reusable).

The project directly supports the goals of the EU Biodiversity Strategy for 2030, the Birds and Habitats Directives, Natura 2000 monitoring, and European pollinator initiatives.

For Pl@ntNet, MAMBO marks a significant milestone toward comprehensive plant biodiversity monitoring. By combining visual recognition, community surveys, predictive modeling, and remote sensing, the project transforms individual observations into actionable insights across landscape and continental scales.

  • Duration: September 2022 – August 2026
  • Program: Horizon Europe

To learn more: