Machine learning reveals global drivers of dissolved nitrogen dynamics in aquatic ecosystems: Implications for N₂O...

Dissolved organic nitrogen (DON) and dissolved inorganic nitrogen (DIN) in aquatic ecosystems significantly influence the global nitrogen cycle, leading to eutrophication, harmful algal blooms, and greenhouse gas emissions. DON and DIN are controlled by multiple environmental drivers, but the relative importance of various drivers remains poorly understood at the continental scale. In this study, we analyzed 703 lake and 1061 river samples, compared DON and DIN concentrations and examined the associated driving factors. Machine learning models were developed using climate, hydrology and human activity data to explore impact mechanisms and to establish a global prediction model. Our study found that nitrogen concentrations in Asia and North America were higher than other regions, with DIN concentrations in Asian lakes even ten times those of other regions. Redundancy analysis and feature importance analysis indicated that nitrogen concentrations in Asia waters were dominated by climatic factors such as temperature and precipitation, as well as agricultural activities, whereas in developed regions such as North America, GDP and land use patterns emerged as the dominant influences. Additionally, linear regression showed that the DIN/DON ratio, as a potential indicator for greenhouse gas nitrous oxide emission assessment (P < 0.0001), could more comprehensively reflect nitrous oxide production potential than DON and DIN alone. This research highlights the need for region-specific water pollution control strategies in order to mitigate the negative impacts of excessive nitrogen loading.