Decadal water level decline suppresses ecosystem metabolism and mitigates net heterotrophy in a floodplain lake

Precipitation and hydrological processes fundamentally reshape aquatic physicochemical properties and biological interactions, thereby regulating lake metabolism and carbon (C) dynamics. This study integrated five years of field data with machine learning to estimate 30 years of ecosystem metabolic rates (gross primary production [GPP], ecosystem respiration [ER], and net ecosystem production [NEP]) in a floodplain lake, elucidating responses to seasonal and long-term water level variations. Seasonally, high water levels and temperatures synergistically increased GPP and ER but decreased NEP. Spatially, sub-lakes exhibited elevated metabolic rates compared to the main lake and channels, attributed to dense aquatic vegetation supporting both production and respiration. However, consistent allochthonous C inputs maintained the lake as a net C source. Over the long term, a 30-year decline in water levels suppressed ecosystem metabolism. Specifically, ER declined more sharply than GPP, leading to a significant rise in NEP and shifting the lake toward a weaker C source. These insights underscore the urgency of adaptive hydrological management to preserve ecosystem metabolic integrity, highlighting that anthropogenic water regulation, rather than climate warming alone, may dictate the future C cycling of large floodplain lakes.