Exploring the Complexities of Dissolved Organic Matter Photochemistry from the Molecular Level by Using Machine Learning Approaches

Environmental drivers of dissolved organic matter composition across central European aquatic systems: A novel correlation-based machine learning and FT-ICR MS approach

Dissolved organic matter (DOM) present in surface aquatic systems is a heterogeneous mixture of organic compounds reflecting its allochthonous and autochthonous organic matter (OM) sources. The composition of DOM is determined by environmental factors like land use, water chemistry, and climate, which influence its release, movement, and turnover in the ecosystem. However, studying the impact of these environmental factors on DOM composition is challenging due to the dynamic nature of the system and the complex interactions of multiple environmental factors involved. Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) enables detailed molecular-level analysis of DOM, allowing the identification of thousands of individual molecular formulas potentially representing unique markers for its "molecular history". The combination of FT-ICR MS with machine-learning techniques is promising to unravel DOM-environment interactions owing to their capacity to capture complex non-linear relationships. We present a novel unsupervised multi-variant machine-learning approach, aiming to model correlation coefficients as robust indicators of how changes in environmental factors (e.g., the concentration of nutrients or the land use) result in changes in the molecular formula descriptors of DOM (i.e., aromaticity index or hydrogen to carbon ratio). We applied this approach to an environmental data set collected from 84 sites across central Europe exhibiting a broad range of water chemistry and land uses. Our model revealed an increase in molecular mass and aromaticity of DOM in densely forested regions as compared to open urban areas, where DOM was characterized by higher concentrations of dissolved ions and increased microbial degradation, leading to smaller and more aliphatic DOM. Our findings highlight the substantial human impact on climate change, as evidenced by the accelerated photochemical and microbial degradation of DOM, which consequently enhances greenhouse gas emissions and exacerbates global warming.

Predicting the Kinetics of Cu and Cd Release from Diverse Soil Dissolved Organic Matter: A Novel Hybrid Model Integrating Machine Learning with Mechanistic Kinetics Model

Kinetic release of trace metals from soil dissolved organic matter (DOM) to solution is the key process controlling the mobility and bioavailability of trace metals in soil environment. However, due to the complexity of soil DOM, predicting the reaction rates of trace metals with soil DOM from different sources remains challenging. In this study, we developed a novel hybrid model integrating machine learning with mechanistic kinetics model, which can quantitatively predict the release rates of Cu and Cd from diverse soil DOM based on their compositions and properties. Our model quantitatively demonstrated that the molecular compositions of DOM controlled metal release rates, which had more profound impact on Cu than Cd. Our modeling results also identified two key factors affecting metal release rates, in which high concentrations of Ca and Mg ions in DOM significantly decreased the release rates of Cu and Cd, and the reassociation reactions of metal ions with DOM became more significant with the release of metals from DOM. This work has provided a unified kinetic modeling framework combining both mechanistic and data-driven approaches, which offers a new perspective for developing predictive kinetics models and can be applied to different metals and DOM in dynamic environments.

Molecular dynamics and factors governing recalcitrance of dissolved organic matter: Insights from laboratory incubation and ultra-high resolution mass spectrometry

The oceanic dissolved organic matter (DOM) reservoir is one of Earth's largest carbon pools, yet the factors contributing to its recalcitrance and persistence remain poorly understood. Here, we employed ultra-high resolution mass spectrometry (UHRMS) to examine the molecular dynamics of DOM from terrestrial, marine and mixed sources during bio-incubation over weekly, monthly, and one year time spans. Using Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS), we classified DOM into three distinct categories (Consumed, Resistant and Product) based on their presence or absence at the start and end of the incubation. Our results show that molecular properties, such as hydrogen to carbon ratio (H/C), modified aromaticity index (AImod), and nominal oxidation state of carbon (NOSC), strongly influence DOM lability and its biogeochemical cycling. Interestingly, Product formulas identified in the short-term incubations were often reclassified as Consumed formulas in longer-term incubations, underscoring the importance of incubation time in determining the persistence of DOM formulas. Further, we introduced a Change Ratio (CR) to identify formulas with significantly altered relative abundances. The molecular characteristics of these Increase or Decrease formulas exhibited notable differences, reinforcing their role in determining lability. In seawater samples, Decrease formulas were more abundant than Increase formulas, supporting the dilution hypothesis, which suggests low concentrations contribute to biological recalcitrance. However, the instability of relative abundance differences between Increase and Decrease formulas when CR thresholds were altered, coupled with the robustness of AImod differences, highlights the dominance of molecular properties over concentration in determining DOM lability. Furthermore, the AImod distribution of these Increase and Decrease formulas mirrored deep-enriched and surface-enriched formulas in the open ocean, validating our incubation results with field investigations. Overall, our study demonstrates that combining laboratory incubation with UHRMS advances our molecular-level understanding of DOM recalcitrance and thus global carbon cycling.

Network-Based Methods for Deciphering the Oxidizability of Complex Leachate DOM with •OH/O3 via Molecular Signatures

In landfill leachates containing complex dissolved organic matter (DOM), the link between individual DOM constituents and their inherent oxidizability is unclear. Here, we resolved the molecular signatures of DOM oxidized by •OH/O3 using FT-ICR MS, thereby elucidating their oxidizability and resistance in concentrated leachates. The comprehensive gradual fragmentation of complex leachate DOM was then revealed through a modified machine-learning framework based on 43 key pathways during ozonation. Specifically, humic substances like humic acid (HA) and fulvic acid (FA) were measured to be the dominant DOM fractions in concentrated leachates, accounting for 35.9-51.7% of the total organic carbon, which was consistent with the observation by three-dimensional fluorescence spectroscopy. According to FT-ICR MS, carboxyl-rich alicyclic molecules (CRAMs) or lignin-like substances were the most abundant components, comprising 40.2-54.5% of all substances. The machine learning modeling showed that molecular weight was the most important structural factor for DOM resistance to •OH and O3 degradation (SHAP value 0.84), followed by (DBE-O)/C (0.32), S/C (0.31), and H/C (0.08). During •OH and O3 attacking, unsaturated and reduced compounds were the dominant precursors. For the molecular transformation of CRAMs-DOM, oxygen addition reactions were found to be the predominant O3-attacking process, along with the dealkyl and carboxylic acid reactions during •OH oxidation that often resulted in more complete degradation of DOM. This study proposed a new framework integrating molecular signatures and machine learning for unraveling DOM's inherent reactivity in complexity, which informs strategies for managing concentrated leachates.

Advancing micro-nano supramolecular assembly mechanisms of natural organic matter by machine learning for unveiling environmental geochemical processes

The nano-self-assembly of natural organic matter (NOM) profoundly influences the occurrence and fate of NOM and pollutants in large-scale complex environments. Machine learning (ML) offers a promising and robust tool for interpreting and predicting the processes, structures and environmental effects of NOM self-assembly. This review seeks to provide a tutorial-like compilation of data source determination, algorithm selection, model construction, interpretability analyses, applications and challenges for big-data-based ML aiming at elucidating NOM self-assembly mechanisms in environments. The results from advanced nano-submicron-scale spatial chemical analytical technologies are suggested as input data which provide the combined information of molecular interactions and structural visualization. The existing ML algorithms need to handle multi-scale and multi-modal data, necessitating the development of new algorithmic frameworks. Interpretable supervised models are crucial owing to their strong capacity of quantifying the structure-property-effect relationships and bridging the gap between simply data-driven ML and complicated NOM assembly practice. Then, the necessity and challenges are discussed and emphasized on adopting ML to understand the geochemical behaviors and bioavailability of pollutants as well as the elemental cycling processes in environments resulting from the NOM self-assembly patterns. Finally, a research framework integrating ML, experiments and theoretical simulation is proposed for comprehensively and efficiently understanding the NOM self-assembly-involved environmental issues.

Novel Photoelectron-Assisted Microbial Reduction of Arsenate Driven by Photosensitive Dissolved Organic Matter in Mine Stream Sediments

The microbial reduction of arsenate (As(V)) significantly contributes to arsenic migration in mine stream sediment, primarily driven by heterotrophic microorganisms using dissolved organic matter (DOM) as a carbon source. This study reveals a novel reduction pathway in sediments that photosensitive DOM generates photoelectrons to stimulate diverse nonphototrophic microorganisms to reduce As(V). This microbial photoelectrophic As(V) reduction (PEAsR) was investigated using microcosm incubation, which showed the transfer of photoelectrons from DOM to indigenous sediment microorganisms, thereby leading to a 50% higher microbial reduction rate of As(V). The abundance of two marker genes for As(V) reduction, arrA and arsC, increased substantially, confirming the microbial nature of PEAsR rather than a photoelectrochemical process. Photoelectron ion is unlikely to stimulate photolithoautotrophic growth. Instead, diverse nonphototrophic genera, e.g., Cupriavidus, Sphingopyxis, Mycobacterium, and Bradyrhizobium, spanning 13 orders became enriched by 10-50 folds. Metagenomic binning revealed their genetic potential to mediate the photoelectron-assisted reduction of As(V). These microorganisms contain essential genes involved in respiratory As(V) reduction, detoxification As(V) reduction, dimethyl sulfoxide reductase family, c-type cytochromes, and multiple heavy-metal resistance but lack a complete photosynthesis system. The novel microbial PEAsR pathway offers new insights into the interaction between photoelectron utilization and nonphototrophic As(V)-reducing microorganisms, which may have profound implications for arsenic pollution transportation in mine stream sediment.

Elucidating per- and polyfluoroalkyl substances (PFASs) soil-water partitioning behavior through explainable machine learning models

In this study, an optimized random forest (RF) model was employed to better understand the soil-water partitioning behavior of per- and polyfluoroalkyl substances (PFASs). The model demonstrated strong predictive performance, achieving an R2 of 0.93 and an RMSE of 0.86. Moreover, it required only 11 easily obtainable features, with molecular weight and soil pH being the predominant factors. Using three-dimensional interaction analyses identified specific conditions associated with varying soil-water partitioning coefficients (Kd). Results showed that soils with high organic carbon (OC) content, cation exchange capacity (CEC), and lower soil pH, especially when combined with PFASs of higher molecular weight, were linked to higher Kd values, indicating stronger adsorption. Conversely, low Kd values (< 2.8 L/kg) typically observed in soils with higher pH (8.0), but lower CEC (8 cmol+/kg), lesser OC content (1 %), and lighter molecular weight (380 g/mol), suggested weaker adsorption capacities and a heightened potential for environmental migration. Furthermore, the model was used to predict Kd values for 142 novel PFASs in diverse soil conditions. Our research provides essential insights into the factors governing PFASs partitioning in soil and highlights the significant role of machine learning models in enhancing the understanding of environmental distribution and migration of PFASs.

Deciphering the Role of Microbial Extracellular and Intracellular Organic Matter in Antibiotic Photodissipation: Molecular and Fluorescent Profiling under Natural Radiation

This study addresses existing gaps in understanding the specific involvement of dissolved organic matter (DOM) fractions in antibiotic photolysis, particularly under natural conditions and during DOM photobleaching. Employing fluorescent, chemical, and molecular analysis techniques, it explores the impact of extracellular and intracellular organic matter (EOM and IOM) on the photodissipation of multiclass antibiotics, coupled with DOM photobleaching under natural solar radiation. Key findings underscore the selective photobleaching of DOM fractions, propelled by distinct chemical profiles, influencing DOM-mediated antibiotic photolysis. Notably, lipid-like substances dominate in the IOM, while lignin-like substances prevail in the EOM, each uniquely responding to sunlight and exhibiting selective photobleaching. Sunlight primarily targets fulvic acid-like lignin components in EOM, contrasting the initial changes observed in tryptophan-like lipid substances in IOM. The lower photolability of EOM, attributed to its rich unsaturated compounds, contributes to an enhanced rate of indirect antibiotic photolysis (0.339-1.402 h-1) through reactive intermediates. Conversely, the abundance of aliphatic compounds in IOM, despite it being highly photolabile, exhibits a lower mediation of antibiotic photolysis (0.067-1.111 h-1). The triplet state excited 3DOM* plays a pivotal role in the phototransformation and toxicity decrease of antibiotics, highlighting microbial EOM's essential role as a natural aquatic photosensitizer for water self-purification. These findings enhance our understanding of DOM dynamics in aquatic systems, particularly in mitigating antibiotic risks, and introduce innovative strategies in environmental management and water treatment technologies.

Insights into the transformation of natural organic matter during UV/peroxydisulfate treatment by FT-ICR MS and machine learning: Non-negligible formation of organosulfates

Natural organic matter (NOM) is a major sink of radicals in advanced oxidation processes (AOPs) and understanding the transformation of NOM is important in water treatment. By using Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) in conjunction with machine learning, we comprehensively investigated the reactivity and transformation of NOM, and the formation of organosulfates during the UV/peroxydisulfate (PDS) process. After 60 min UV/PDS treatment, the CHO formula number and dissolved organic carbon concentration significantly decreased by 83.4 % and 74.8 %, respectively. Concurrently, the CHOS formula number increased substantially from 0.7 % to 20.5 %. Machine learning identifies DBE and AImod as the critical characteristics determining the reactivity of NOM during UV/PDS treatment. Furthermore, linkage analysis suggests that decarboxylation and dealkylation reactions are dominant transformation pathways, while the additions of SO3 and SO4 are also non-negligible. According to SHAP analysis, the m/z, number of oxygens, DBE and O/C of NOM were positively correlated with the formation of organosulfates in UV/PDS process. 92 organosulfates were screened out by precursor ion scan of HPLC-MS/MS and verified by UPLC-Q-TOF-MS, among which, 7 organosufates were quantified by authentic standards with the highest concentrations ranging from 2.1 to 203.0 ng L‒1. In addition, the cytotoxicity of NOM to Chinese Hamster Ovary (CHO) cells increased by 13.8 % after 30 min UV/PDS treatment, likely responsible for the formation of organosulfates. This is the first study to employ FT-ICR MS combined with machine learning to identify the dominant NOM properties affecting its reactivity and confirmed the formation of organosulfates from sulfate radical oxidation of NOM.

Reassessing the Quantum Yield and Reactivity of Triplet-State Dissolved Organic Matter via Global Kinetic Modeling

Measuring the quantum yield and reactivity of triplet-state dissolved organic matter (3DOM*) is essential for assessing the impact of DOM on aquatic photochemical processes. However, current 3DOM* quantification methods require multiple fitting steps and rely on steady-state approximations under stringent application criteria, which may introduce certain inaccuracies in the estimation of DOM photoreactivity parameters. Here, we developed a global kinetic model to simulate the reaction kinetics of the hv/DOM system using four DOM types and 2,4,6-trimethylphenol as the probe for 3DOM*. Analyses of residuals and the root-mean-square error validated the exceptional precision of the new model compared to conventional methods. 3DOM* in the global kinetic model consistently displayed a lower quantum yield and higher reactivity than those in local regression models, indicating that the generation and reactivity of 3DOM* have often been overestimated and underestimated, respectively. The global kinetic model derives parameters by simultaneously fitting probe degradation kinetics under different conditions and considers the temporally increasing concentrations of the involved reactive species. It minimizes error propagation and offers insights into the interactions of different species, thereby providing advantages in accuracy, robustness, and interpretability. This study significantly advances the understanding of 3DOM* behavior and provides a valuable kinetic model for aquatic photochemistry research.

Recent advances in the role of dissolved organic matter during antibiotics photodegradation in the aquatic environment

The presence of residual antibiotics in the environment is a prominent issue. Photodegradation behavior is an important way of antibiotics reduction, which is closely related to dissolved organic matter (DOM) in water. The review provides an overview of the latest advancements in the field. Classification, characterization of DOM, and the dominant mechanisms for antibiotic photodegradation were discussed. Furthermore, it summarized and compared the effects of DOM on different antibiotics photodegradation. Moreover, the review comprehensively considered the factors influencing the photodegradation of antibiotics in the aquatic environment, including the characteristics of light, temperature, dosage of DOM, concentration of antibiotics, solution pH, and the presence of coexisting ions. Finally, potential directions were proposed for the development of predictive models for the photodegradation of antibiotics. Based on the review of existing literature, this paper also considered several pathways for the future study of antibiotic photodegradation. This study allows for a better understanding of the DOM's environmental role and provides important new insights into the photochemical fate of antibiotics in the aquatic environment.

Dissolved organic matter composition and characteristics during extreme flood events in the Yangtze River Estuary

Understanding the molecular composition and fate of dissolved organic matter (DOM) during transport in estuaries is essential for gaining a comprehensive understanding of its role within the global biogeochemical cycle. In 2020, a catastrophic flood occurred in the Yangtze River basin. It is currently unknown whether differences in hydrologic conditions due to extreme flooding will significantly impact the estuarine to oceanic DOM cycle. We determined the DOM composition in the Yangtze River estuary (YRE) to the East China Sea by using Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) during the high discharge and the flood period (monthly average discharge was 1.2 times higher) on the same trajectory. Our study found that the composition of DOM is more diverse, and more DOM molecules were introduced to the YRE during the flood, especially in the freshwater end member. The result revealed that the DOM was significantly labile and unstable during the flood period. A total of 1840 unique molecular formulas were identified during the flood period, most of which were CHON, CHONS, and CHOS compounds, most likely resulting from anthropogenic inputs from upstream. Only 194 of these molecules were detected in the seawater end member after transporting to the sea, suggesting that the YRE served as a 'filter' of DOM. However, the flood enhances the transport of a group of terrigenous DOM, that is resistant to photodegradation and biodegradation. As a result, YRE experienced ~1.6 times higher terrigenous DOC flux than high discharge period. Considering the increased frequency of future floods, our study provides a preliminary basis for further research on how floods affect the composition and characteristics of estuarine DOM. With the help of the FT-ICR MS technique, we can now better understand the dynamic of DOM composition and characteristics in large river estuaries.

Deciphering DOM-metal binding using EEM-PARAFAC: Mechanisms, challenges, and perspectives

Dissolved organic matter (DOM) is a pivotal component of the biogeochemical cycles and can combine with metal ions through chelation or complexation. Understanding this process is crucial for tracing metal solubility, mobility, and bioavailability. Fluorescence excitation emission matrix (EEM) and parallel factor analysis (PARAFAC) has emerged as a popular tool in deciphering DOM-metal interactions. In this review, we primarily discuss the advantages of EEM-PARAFAC compared with other algorithms and its main limitations in studying DOM-metal binding, including restrictions in spectral considerations, mathematical assumptions, and experimental procedures, as well as how to overcome these constraints and shortcomings. We summarize the principles of EEM to uncover DOM-metal association, including why fluorescence gets quenched and some potential mechanisms that affect the accuracy of fluorescence quenching. Lastly, we review some significant and innovative research, including the application of 2D-COS in DOM-metal binding analysis, hoping to provide a fresh perspective for possible future hotspots of study. We argue the expansion of EEM applications to a broader range of areas related to natural organic matter. This extension would facilitate our exploration of the mobility and fate of metals in the environment.