Project Summary
Cactus: using artificial intelligence to better anticipate species extinction risk
The Cactus exploratory research project investigates a new approach to estimating the conservation status of large numbers of species. By combining artificial intelligence, statistical learning, ecological modelling and large biodiversity databases, the project aims to identify species that may be at risk and help experts better target conservation assessments.
The IUCN Red List is the global reference for assessing species extinction risk. However, assessing thousands of species remains a major challenge, particularly when available data are incomplete or unevenly distributed. Cactus therefore explores complementary methods for scaling up assessments, without replacing official IUCN evaluations.
Combining data to better understand species
The project draws on the growing diversity of biodiversity data, including georeferenced observations, digitised natural history collections and herbarium specimens, species traits, climate data, and information on soils, vegetation and land use.
By combining these different sources, models can learn relationships between species and their habitats, estimate species distributions and explore how these distributions may change under different environmental scenarios. These results can then help identify situations that could increase extinction risk.
This approach must nevertheless address several challenges: biodiversity databases contain far more information about species presence than reliable absence, observations are often concentrated in easily accessible areas, and many species remain represented by only a small number of records.
The role of Pl@ntNet
Observations generated through citizen science provide a particularly valuable source of information for Cactus. Each observation shared through Pl@ntNet provides information about the presence of a species at a given location and date.
The project builds on research carried out around Pl@ntNet on modelling species distributions from citizen observations. The millions of observations collected through the platform can help improve knowledge of plant distributions and identify potential changes in species ranges.
Pl@ntNet can therefore be used not only as a plant identification and observation-sharing tool, but also as a valuable data source for research into biodiversity dynamics and conservation.
Towards decision support for conservation assessment
Cactus explores a process in which models first learn about the environmental conditions associated with species and their habitats, and then estimate how their distributions or suitable habitats could change. These results can help identify species requiring particular attention and contribute to prioritising future assessments.
Such predictions remain exploratory and need to be interpreted carefully. The quality and representativeness of available data, together with model uncertainty, must be taken into account.
By bringing together artificial intelligence, ecology, botany, biodiversity data and citizen science, Cactus explores new ways of understanding species distributions and supporting their monitoring and conservation at large scale.
Programme: Inria Exploratory Action, 2020
Fields: biodiversity conservation, artificial intelligence, statistical learning, ecological modelling and citizen science.
Cactus Project News
The Cactus project has an official webpage, which can be accessed HERE.
