Preferential nitrogen retention characterizes current eutrophication in United States lakes.

The relative retention of nitrogen (N) and phosphorus (P) in lakes and their roles in nutrient cycling remain debated. Although the United States has implemented long-term P control measures in lakes, inexplicable eutrophication exacerbation still reoccurred over the past two decades. Here we geographically track and analyze the nutrient retention changes of 121 lakes in the USA during 2012-2022 using a combination of machine learning and Bayesian hierarchical methods. We find that the preferential retention of P in 70.3% lakes in 2012 shifted to that of N in 85.1% lakes in 2017, and 92.6% lakes in 2022. Four key lake indicators exhibit flickering-like dynamics, which may help identify the nutrient metabolic disorder-like state in the aquatic ecosystem. Overall, our results show that eutrophication in U.S. lakes during 2012-2022 is advancing on the basis of preferential and net nutrient retention shift from P to N, flickering-like dynamics, and associated occurrence of the nutrient metabolic disorder-like state that are all predominantly driven by internal cycling instead of external loading. These patterns are consistent with the possibility of a regime shift rather than simple eutrophication recurrence, and provide insights relevant to lake management and water quality improvement strategies.