Diagnosing lake health: a next-generation tool combining environmental DNA and AI

How can we tell if a lake is healthy? Until now, scientists have mainly relied on measuring physico-chemical parameters and identifying bioindicator species¹. INRAE researchers propose an innovative approach based on environmental DNA (eDNA) data² that has not been assigned to a species — a so-called taxonomy-free approach³. The authors of the study tested the hypothesis that the structure of biological networks built from eDNA data could provide relevant indicators for assessing lake water quality.

More complex, more connected and more structured networks in ecologically healthy lakes

As part of the Pôle R&D ECLA ⁴, the researchers analysed nearly 600 phytoplankton eDNA inventories from 186 French lakes, covering a broad eutrophication gradient⁵.

The team built co-occurrence networks from these DNA sequences and analysed small network motifs known as graphlets — a method rarely applied in ecology. Results show that these networks have a distinctive “topological signature” characteristic of ecological status: the least disturbed lakes have more complex, more connected and more structured networks, whereas networks in highly eutrophic lakes become simpler.

By combining eDNA data, network analysis, ecological niche modelling and artificial intelligence, researchers were able to predict the level of anthropogenic pressure affecting lakes — such as pollution and eutrophication — demonstrating the relevance of the proposed tool.

Beyond lakes, the study opens up a new avenue for biomonitoring. It shows that the structure of ecological networks can be used to develop an indicator of ecosystem status. In time, this approach could be applied to a wide range of biological data and adapted to other environments, including rivers and soils, or to other environmental pressures, such as contaminants and climate change.

These findings contribute to realising the vision of next-generation biomonitoring in support of biodiversity management and protection.

_{1. Species whose ecological profiles can indicate the presence and intensity of an environmental disturbance, or its absence. For example, Dinobryon sociale, a eukaryotic microalgae, is found in Lake Annecy, which is in good ecological status, whereas Microcystis aeruginosa, a potentially toxic cyanobacterium, occurs in environments affected by eutrophication.}

_{2. Technique used to identify the species present in a natural environment from the genetic traces — DNA fragments — they leave behind.}

_{3. Approach that does not require assignment to a specific species, thereby overcoming the limitations associated with incomplete DNA reference databases.}

_{4. OFB, INRAE, USMB.}

_{5. Process by which a lake becomes enriched with nutrients, accelerated by human activities. The transition from an oligotrophic state (low in nutrients) to a eutrophic state (high in nutrients) alters the lake’s characteristics, leading to the extensive growth of phytoplankton and adverse ecological effects.}