Autochthonous particulate organic carbon drives extreme carbon fluxes in river-reservoir systems: Potential role of interactions between picophytoplankton and heterotrophic bacteria.

Reservoir ecosystems are significant natural sources of atmospheric methane (CH4) and carbon dioxide (CO2), while also receiving large amounts of nutrients and particulate organic carbon (POC) from various sources. Ecosystem-level events, such as variations in microbial community composition or abundance, can substantially contribute to the occurrence of extremes in carbon (CH4 and CO2) concentrations and fluxes. The trophic linkage between picophytoplankton (PP) and heterotrophic bacteria (HB) plays a vital role in regulating POC, thereby shaping patterns of carbon emissions in aquatic systems. However, the effects of their interactions on the dynamics of extreme carbon emissions in river-reservoir systems remain poorly understood. In this study, five reservoirs with different trophic states (Trophic level index: 33.91-70.81) in the upper Yangtze River were selected to identify the source of POC driving extreme carbon emissions and to explore PP-HB interactions. CH4 concentrations were 0.00-0.12 and 0.13-0.33 μmol·L-1 in the Normal and Extremely High Groups, respectively; CO2 concentrations were 8.46-18.01, 24.28-64.63, and 71.92-92.74 μmol·L-1 in the Extremely Low, Normal, and Extremely High Groups, respectively. The evidence from isotope analysis showed that the extreme carbon fluxes were strongly influenced by autochthonous rather than allochthonous POC. Network analysis showed that the positive interaction strength between phytoplankton and bacterioplankton was higher in the Extremely High or Extremely Low Groups than in the Normal Groups of CH4 and CO2 concentrations. The results of the structural equation modeling showed that PP-HB interaction strongly drove the extreme CH4 and CO2 concentrations. These findings highlight the importance of PP and HB in extreme carbon emissions, and we hope our study can provide important implications for integrating PP-HB interactions into predicting extreme carbon emissions in river-reservoir ecosystems.