Deciphering the sources of dissolved organic matter in a eutrophic lake: Insights from the identification of stable molecular markers by FT-ICR MS and assembly processes

Precise source apportionment of dissolved organic matter (DOM) is pivotal for understanding carbon cycling mechanisms and developing effective management strategies in eutrophic lakes. With the rapid advancement of molecular characterization technologies, tracing the provenance of DOM from a molecular perspective warrants profound investigation. Herein, a novel quantitative framework for DOM source apportionment was developed to improve source identification accuracy in a eutrophic urban lake, following a step-wise workflow involving marker screening via molecular transformation networks, validation of marker conservatism using ecological null modeling, and quantification of source contributions with Bayesian mixing models. A total of 292 unique, degradation-resistant molecular markers representing distinct end-member characteristics were screened through potential molecular transformation of DOM. Furthermore, ecological null modeling verified that the assembly processes of these 292 molecules in the lake were dominated by drift (82.2%) within stochastic processes, thereby validating their robustness as molecular tracers. Random Forest and Shapley additive explanations analyses further identified signal intensity and N/C ratio as the most sensitive molecular parameters for quantitative source tracking. The relative contributions of DOM sources were determined as follows: internal sediment release (29.5%) > fertilizers (12.1%) > plant litter (11.1%) > aquaculture effluent (9.8%) > wastewater treatment plant effluent (9.3%) > livestock manure (7.7%) > topsoil (7.5%) > phosphate rock (5.7%). Overall, this integrated framework provides a robust tool for fine-scale discrimination between internal and external DOM sources in eutrophic lakes and demonstrates the strong potential of this multilevel system for accurate DOM source apportionment.