Explainable machine learning reveals climate warming increases risk of algal blooms in lakes and reservoirs

Yiming Ma; Jingfu Wang; Shouliang Huo; Dengjun Wang; Yuchun Wang; Jiayang Li; Jingan Chen; Lian Feng

doi:10.1016/j.watres.2025.124460

Explainable machine learning reveals climate warming increases risk of algal blooms in lakes and reservoirs

Water Res. 2025 Dec 1;287(Pt B):124460. doi: 10.1016/j.watres.2025.124460. Epub 2025 Aug 25.

Authors

Yiming Ma¹, Jingfu Wang², Shouliang Huo³, Dengjun Wang⁴, Yuchun Wang⁵, Jiayang Li¹, Jingan Chen¹, Lian Feng⁶

Affiliations

¹ State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China; University of Chinese Academy of Sciences, Beijing, China.
² State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, China; University of Chinese Academy of Sciences, Beijing, China; Guizhou Province Field Scientific Observation and Research Station of Hongfeng Lake Reservoir Ecosystem, Guiyang, China. Electronic address: wangjingfu@vip.skleg.cn.
³ School of Environment, Beijing Normal University, Beijing, China.
⁴ Department of Agricultural and Biological Engineering, University of Florida, Gainesville, FL 32611, USA.
⁵ State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, China Institute of Water Resources and Hydropower Research, Beijing, China.
⁶ School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen, China.

PMID: 40902384
DOI: 10.1016/j.watres.2025.124460

Abstract

In recent decades, China has exerted significant efforts to diminish the influx of exogenous nitrogen (N) and phosphorus (P) into lakes and reservoirs (L&Rs) in an attempt to control algal blooms (ABs). However, climate change has undermined the effectiveness of these measures. Therefore, distinguishing the contributions of climate change and nutrients is crucial for developing effective ABs management strategies, but remains challenging. Here, we combine the onset and end times of ABs with neural network models to study the trends and controlling factors of algal bloom risk periods (ABRPs) of L&Rs in China from 2003 to 2020. Our results showed that over the past 20 years, the ABRPs of lakes and reservoirs increased by 42 days (11.5%) and 54 days (14.8%), respectively. The contribution of air temperature (T_air) (11%) to prolonging ABRPs was much greater than that of P (5.3%) and N (3.5%). T_air's contribution surpasses that of N and P accounted for up to 63% of one year and will continue to extend with global warming. Moreover, the ABRPs for reservoirs showed a faster growth rate compared to lakes. This may be due to the lower T_air threshold in the reservoir (12.6 °C vs. 16.7 °C in the lake) and the higher contribution of temperature (17% vs. 5% in the lake). Overall, our study highlights the potential importance of climate warming on the onset and end times of ABs, and also distinguishes the differences in driving mechanisms between L&Rs ABs, thus providing a scientific basis for the development of ABs prevention and control strategies for L&Rs in China.

Keywords: Algal blooms; Climate warming; Explainable machine learning; Lakes & reservoirs.

MeSH terms

China
Climate Change*
Eutrophication*
Lakes*
Machine Learning*
Neural Networks, Computer
Nitrogen
Phosphorus
Temperature

Substances

Phosphorus
Nitrogen