AI Traces Agricultural Chemicals’ Deadly Path Through UK Freshwater Networks

Cutting-edge AI analysis of 52 UK lakes uncovers how agricultural chemicals and heavy metals devastate aquatic life—and maps a path to smarter conservation.

​​​​​​​Research: Unveiling Landscape-Level Drivers of Freshwater Biodiversity Dynamics. ​​​​​​​Image Credit: Daniel_Kay / Shutterstock

A new study published in the journal Environmental DNA has revealed that scientists can now identify the most harmful pollutants in UK waters that are having the most significant impact on biodiversity. This is possible thanks to the pioneering AI technology developed at the University of Birmingham.

The new technology allowed the team of scientists to analyze water and biofilm samples from 52 freshwater lakes across the country. They efficiently and effectively sifted through reams of complex data to find key links between the presence of pollutants and biodiversity loss. The data concluded that insecticides and fungicides were the main factors affecting biodiversity, along with 43 physicochemical factors (e.g., heavy metals, alkalinity) and lake typology (e.g., depth, size), though climate change was not included in the analysis.

Lead author of the study, Dr. Niamh Eastwood, explained: "Up until now, DNA-based methods have been used to estimate changes in specific species groups or indicator taxa (e.g., diatoms), but have tended to focus on individual environmental factors or single stressors, overlooking the complex interaction between biodiversity and environmental change. This narrow approach is now insufficient to address the complexities of a world facing multiple stressors and rapidly emerging threats to water and wildlife. The results from our study highlighted the severe impact that insecticides and fungicides from agricultural runoff have on aquatic ecosystems. It is clear that these chemicals are harming many more species than they are intended for, making them of great concern."

Senior author Professor Luisa Orsini added: "Protecting biodiversity is more important than ever. Effective conservation goes beyond looking at how single environmental factors affect individual species. Instead, it requires understanding of how these factors interact with other environmental drivers, such as land use and nutrient inputs, to drive overall biodiversity loss. Our innovative, data-driven approach embraces the complexity of natural systems, while providing actionable targets for regulators. By analyzing vast amounts of data, we can uncover which environmental factors have the greatest impact on sensitive species. This insight is key to developing targeted, effective conservation strategies that can address the root causes of biodiversity decline and help preserve our planet's ecosystems. With this approach, we aim to pave the way for smarter, science-backed conservation efforts that safeguard the natural world for future generations."

​​​​​​​Sankey diagram showing significant links between genera identified by the four gene markers (18SV1V2 and 18SV8V9 are combined) in water and biofilm and factors within drivers (plant protection products [PPP] and physico-chemical parameters, no significant typology factors). Significant links are supported by a Pearson's correlation coefficient larger than 0.5 and an FDR-adjusted p-value < 0.05 following 5000 permutations. Left flow (factor) color indicates PPP group or physicochemical parameter, central flow (sample type and driver) color indicates sample type and all right flow (genera) are uncoloured.

Dr. Jiarui Zhou, a senior study author, highlighted the transformative power of artificial intelligence in tackling environmental challenges. "This study utilizes advanced statistical learning to integrate complex multimodal datasets, showcasing how AI-powered approaches can revolutionize environmental science," Dr. Zhou explained. "By enabling the prioritization of species for conservation and identifying the chemicals most harmful to biodiversity, this approach opens new pathways for protecting our natural world. This breakthrough showcases how cutting-edge technology can drive practical solutions in conservation and environmental protection, setting the stage for a healthier, more sustainable planet."

Arron Watson, co-author of the study, emphasized the practical implications of the research, stating: "Our study highlighted the harmful effects of chemicals banned shortly after our study, providing confidence in the approach to uncover harmful substances. This approach could also be used to detect chemicals that still cause harm to biodiversity even after their use is discontinued, due to their persistence in the environment."

This groundbreaking work underscores the importance of proactive measures in chemical regulation and demonstrates the long-lasting impact harmful substances can have on ecosystems. By identifying and addressing these threats, this research supports stronger, data-driven strategies for safeguarding biodiversity and protecting the environment.