Characterization of Dissolved Organic Matter Released from Decomposing Wood 

in Denitrifying Bioreactors: An FT-ICR MS Study 

 

 

 

 

A Thesis 

Presented to the Faculty of the Graduate School 

of Cornell University 

In Partial Fulfillment of the Requirements for the Degree of 

Master of Science 

 

 

 

 

 

 

by 

Megan Bonite 

December 2024 



 

 

 

 

 

 

 

 

 

 

 

 

© 2024 Megan Bonite



 

ABSTRACT 

While several studies have employed Fourier transform ion cyclotron resonance mass 

spectrometry (FT-ICR MS) for analysis of complex dissolved organic matter (DOM), using FT-

ICR MS to analyze molecular-level changes in labile carbon over time remains a largely 

unexplored area of woodchip bioreactor (WBR) design and optimization. In this study, a variety 

of FT-ICR MS visualization and ordination methods, such as heteroatom class distribution plots, 

van Krevelen diagrams, and principal component analysis, were used to explore correlations 

between molecular DOM properties and key WBR functions. Comparisons between upstream 

and downstream flow were made to elucidate possible spatial effects on lability, motivated by 

the potential effect of oxygen on wood-derived dissolved organic carbon. The effectiveness of 

periodic drying-rewetting cycles to stimulate both wood degradation by aerobic microbes and 

denitrification by anaerobic microbes was then explored. Analysis of DOM by FT-ICR MS 

presents a new way to discover how to increase the efficiency of WBRs for nitrate removal from 

contaminated, nitrogen-rich water sources.  

 

 



iii 

 

BIOGRAPHICAL SKETCH 

Megan Bonite, born and raised in Old Bridge, New Jersey, earned her Bachelor of Arts in 

Chemistry from Rutgers University New Brunswick in May of 2023. In August of the same year, 

she joined the Department of Chemistry and Chemical Biology at Cornell University in pursuit 

of her Master of Science, working under the guidance of Professor Matthew C. Reid. 

 

 

 

 

 

 

 

 

 

 



iv 

 

ACKNOWLEDGMENTS 

I would like to firstly extend my gratitude to my advisor, Professor Matthew C. Reid, for his 

unwavering support during my time at Cornell University. The steadfast dedication Professor 

Reid shows each of his students inspired me to carve out my own path for my future in the world 

of research and academia. My time under his mentorship will always be cherished and looked 

back upon fondly. I also would like to extend my thanks to my lab mates and to Yi Sang and Iva 

Petrovic, whose preliminary work was integral in the development of this thesis. 

  

Next, I would like to express my thanks to Professor David B. Zax, Director of CCB’s M.S. 

Graduate Program. Since my first day in Ithaca, Professor Zax has always been a reliable source 

of guidance and support through both good and bad days. He has always kept my special 

interests in environmental research in mind, wanting me to have the best, most fulfilling 

experience possible at Cornell and encouraging me to branch out to seek what I am truly 

passionate about. 

  

To my dearest and closest friends, and my partner Michael - I am eternally grateful for your love, 

friendship, and emotional support, which kept myself grounded miles away from home. 

  

Lastly, for Mom and Dad, I hope that I will always continue to make you proud, so that you can 

always look back to your decision with pride knowing both of your children have never taken 

your sacrifices for granted. 

 

  



v 

 

TABLE OF CONTENTS 

ABSTRACT ................................................................................................................... i 

BIOGRAPHICAL SKETCH ..................................................................................... iii 

ACKNOWLEDGMENTS .......................................................................................... iv 

LIST OF FIGURES ................................................................................................... vii 

LIST OF TABLES ...................................................................................................... ix 

LIST OF ABBREVIATIONS ...................................................................................... x 

CHAPTER 1: Introduction .......................................................................................... 1 

CHAPTER 2: Methods ................................................................................................. 8 

2.1 Site Description and WBR Design ....................................................................... 8 

2.2 Sample Collection and Preparation ....................................................................... 9 

2.2.1 Solid Phase Extraction (SPE) ...................................................................... 10 

2.3 Data Analysis ...................................................................................................... 11 

2.3.1 Biogeochemical Properties of Woodchips ................................................... 11 

2.3.2 Negative-Ion Electrospray Ionization (ESI) FT-ICR MS ............................. 13 

2.3.3 Calculation of Parameters ........................................................................... 14 

2.3.4 Statistical Analysis ....................................................................................... 15 

CHAPTER 3: Results ................................................................................................. 17 

3.1 Comparison of Inlet / Outlet Flow WBR ............................................................ 17 

3.1.1 FT-ICR MS Spectra and DOM Composition of Inlet / Outlet Flow WBR ... 18 

3.1.2 Distribution Patterns of Oxygen, Carbon, and DBE ................................... 21 

3.1.3 Modified van Krevelen Plots of Nitrogen-Containing Groups .................... 23 

3.1.5 Determining Amount of Labile Carbon in Inlet and Outlet WBR Samples . 25 

3.1.6 Calculated DOM Parameters and Principal Component Analysis (PCA) .. 26 

3.2 Effect of Oxic-Anoxic Cycling on DOM ............................................................ 29 

3.2.1 FT-ICR MS Spectra and DOM Composition of Oxic-Anoxic and Anoxic-Only 

Conditions ............................................................................................................. 30 

3.2.2 Distribution Patterns of Carbon, Oxygen, and DBE ................................... 34 

3.2.3 Modified van Krevelen Plots of Nitrogen-Containing Compounds ............. 39 

3.2.4 Application of the MLB to Determine Levels of Labile Carbon .................. 40 

3.2.5 Calculated Parameters and Principal Component Analysis ....................... 41 

CHAPTER 4: Discussion ............................................................................................ 45 

4.1 Impact of Flow on DOM Composition and Lability .......................................... 45 

4.1.1 Spatial Influences on Microbial Lignin Degradation .................................. 45 



vi 

 

4.1.2 Assessment of Lability Confirms Increased Degradation Upstream ........... 48 

4.1.3 Biogeochemical Measurements Reflect FT-ICR MS Findings .................... 48 

4.2 Oxic-Anoxic Cycling May Improve Longevity of Denitrifying WBRs ............. 49 

4.2.1 DOM From Anoxic Conditions Reflect Energetic Constraints on Lignin Degradation

............................................................................................................................... 49 

4.2.2 Oxic-Anoxic Cycling Leads to a Dynamic Equilibrium of Labile Carbon Release and 

Consumption ......................................................................................................... 52 

4.2.3 Correlations between Biogeochemical Measurements and FT-ICR MS Data52 

CHAPTER 5: Conclusion .......................................................................................... 55 

CHAPTER 6: Future Work ....................................................................................... 56 

REFERENCES ............................................................................................................ 57 

 



vii 

 

LIST OF FIGURES 

Figure 1. Schematic of the three major components of wood and their chemical structures. 

Chains of cellulose, made of linear glycosidic linkages, arrange to form microfibrils (Zhao et al., 

2022). Each microfibril is surrounded by hemicellulose (e.g., xylan), a branched 

heteropolysaccharide that covalently connects to a rigid lignin outer shell consisting of ρ-

hydroxyphenyl (H-type), guaiacyl (G-type), and syringyl (S-type) subunits (Malherbe & Cloete, 

2002).  

Figure 2. Schematic of lignin degradation by white-rot fungi modified from Cragg et al. (2015). 

Laccase and lignin peroxidase are excreted through fungal hyphae. Laccases use oxygen as an 

electron acceptor to attack phenolic moieties and produce free radicals while lignin peroxidase 

use hydrogen peroxide (H2O2) as electron acceptor to depolymerize lignin into monomers 

(Burlacu et al., 2018). 

Figure 3. Satellite image of the site of the WBR in Freeville, New York. 

Figure 4. Schematic of the 10-year-old denitrifying woodchip bioreactor from which samples 

were collected from upstream (Port1), midstream (Port2), and downstream (Port3) locations.  

Figure 5. Sample collection for WBRs subjected to oxic-anoxic and anoxic-only conditions. 

Each grey line dictates a day starting from April 18 to May 6. On April 24, the media was 

changed. April 27, April 30, and May 3 are referred to as Day 3, Day 6, and Day 9, respectively, 

indicating how many days have passed after changing media. 

Figure 6. Nitrate removal batch experiments with woodchips collected from the inlet and outlet 

port of WBRs. At time 0, batch reactors were replaced with fresh nutrient media and were 

incubated anoxically. 

Figure 7. Measurements of A) SUVA254 and B) Dissolved organic carbon in the Anoxic Inlet 

and Cycle Inlet WBRs over time. 

Figure 8. ‘CHO’-reconstructed mass spectra of inlet A, B, C and outlet D, E, F triplicate 

samples. 

Figure 9. Class-separated van Krevelen diagrams of triplicate inlet samples A, B, C and 

triplicate outlet flow samples D, E, F. Plot points are classified based on regions of H/C and O/C 

ratio dictated by Zhou et al., 2024: (1) lipids- 1.7 ≥ H/C ≤ 2.2, 0 ≥ O/C ≤ 0.2 (2) protein and 

amino sugars, 1.7 ≥ H/C ≤ 2.2, 0.2 ≥ O/C ≤ 0.6, (3) carbohydrates, 1.5 ≥ H/C ≤ 2.2, 0.6 ≥ O/C ≤ 

1.0, (4) unsaturated hydrocarbons, 0.7 ≥ H/C ≤ 1.5, 0 ≥ O/C ≤ 0.1, (5) lignin 0.6 ≥ H/C ≤ 1.7, 0.1 

≥ O/C ≤ 0.6, (6) tannins, 0.5 ≥ H/C ≤ 1.5, 0.6 ≥ O/C ≤ 1.0, (7) condensed aromatics, 0.3 ≥ H/C ≤ 

0.7, 0 ≥ O/C ≤ 1.0. 

Figure 10. Box plots showing A) oxygen, B) carbon, and C) DBE distributions of the inlet and 

outlet samples. Red solid line represents median value. Blue dashed line represents mean value. 



viii 

 

Figure 11. Modified 3D van Krevelen plots of nitrogen-containing compounds in inlet A, B, C 

and outlet D, E, F samples. Bubble size reflects abundance, with larger size indicating higher 

occurrence of compounds associated with a specific H/C : N/C ratio. 

Figure 12. Average percentage of labile formulas in the inlet and outlet according to application 

of the molecular lability boundary coined by D’Andrilli et al. (2015). 

Figure 13. Distribution of A) aromaticity index and B) NOSC for inlet and outlet samples. 

Figure 14. Biplot from principal component analysis of inlet and outlet WBR samples using the 

top 20% of formulas ranked by relative abundance. 

Figure 15. Comparison of ‘CHO’-reconstructed mass spectra of inlet samples from oxic-anoxic 

cycling “Cycle Inlet” and anoxic-only “Anoxic Inlet” conditions, A) 3 days B) 6 days and C) 9 

days following changed media. 

Figure 16. Class-separated van Krevelen plots of A) Cycle Inlet and B) Anoxic Inlet samples 

from i) 3 days, ii) 6 days, and iii) 9 days after changed media. Each plot is separated into regions 

based on H/C and O/C ratio as dictated by Zhou et al., 2024: (1) lipids- 1.7 ≥ H/C ≤ 2.2, 0 ≥ O/C 

≤ 0.2 (2) protein and amino sugars, 1.7 ≥ H/C ≤ 2.2, 0.2 ≥ O/C ≤ 0.6, (3) carbohydrates, 1.5 ≥ 

H/C ≤ 2.2, 0.6 ≥ O/C ≤ 1.0, (4) unsaturated hydrocarbons, 0.7 ≥ H/C ≤ 1.5, 0 ≥ O/C ≤ 0.1, (5) 

lignin 0.6 ≥ H/C ≤ 1.7, 0.1 ≥ O/C ≤ 0.6, (6) tannins, 0.5 ≥ H/C ≤ 1.5, 0.6 ≥ O/C ≤ 1.0, (7) 

condensed aromatics, 0.3 ≥ H/C ≤ 0.7, 0 ≥ O/C ≤ 1.0. 

Figure 17. DBE distribution over time for Cycle Inlet and Anoxic Inlet samples A) 3 days, B) 6 

days, and C) 9 days after changed media. 

Figure 18. Carbon number distribution of ‘CHO’-containing compounds time for Cycle Inlet 

and Anoxic Inlet samples A) 3 days, B) 6 days, and C) 9 days after changed media. 

Figure 19. Oxygen number distribution of ‘CHO’-containing compounds over time for Cycle 

Inlet and Anoxic Inlet samples A) 3 days, B) 6 days, and C) 9 days after changed media. 

Figure 20. Modified van Krevelen plot of nitrogen-containing compounds in A) Cycle Inlet and 

B) Anoxic Inlet samples from i) 3 days, ii) 6 days, and iii) 9 days after changed media. Bubble 

size reflects abundance, with larger size indicating higher occurrence of compounds associated 

with a specific H/C : N/C ratio. 

Figure 21. Percentage of labile formulas over time in the Cycle Inlet and Anoxic Inlet samples. 

Figure 22. Average A) aromaticity index and B) nominal oxidation state of carbon for Cycle 

Inlet and Anoxic Inlet samples over time. 

Figure 23. Date-separated PCA biplot generated from Cycle Inlet and Anoxic Inlet samples 

using top 20% of formulas ranked by relative abundance. 

 



ix 

 

LIST OF TABLES 

Table 1. Summary of biogeochemical properties of inlet and outlet WBR. 

Table 2. Number of identified compounds through FT-ICR MS of samples from inlet and outlet 

flow WBR. 

Table 3. N/C ratio distribution of nitrogen-containing compounds in the inlet and outlet samples. 

Table 4. Summary of calculated parameters for PCA using top 20% of formulas ranked by 

relative abundance. 

Table 5. Identified compounds from Cycle and Anoxic samples. 

Table 6. N/C ratio distribution of nitrogen-containing compounds in the Cycle Inlet and Anoxic 

Inlet samples. 

Table 7. Summary of calculated parameters of top 20% of peaks ranked by relative abundance. 

 

 

 

 



x 

 

LIST OF ABBREVIATIONS 

NPS Nonpoint source 

N Nitrogen 

P Phosphorus 

WBR Woodchip bioreactor 

DOC Dissolved organic carbon 

DRW Drying-rewetting 

DOM Dissolved organic matter 

FT-ICR MS Fourier transform ion cyclotron resonance mass spectrometry 

ESI Electrospray ionization 

MLB Molecular lability boundary 

NOSC Nominal oxidation state of carbon 

DBE Double bond equivalent 

PCA Principal component analysis 

SUVA254 Specific ultraviolet absorbance 

AI Aromaticity Index 

AImod Modified aromaticity index 

SPE Solid phase extraction 

APPI Atmospheric Pressure Photoionization 

%MLBL Molecular lability index 

PC Principal component 

MnP Manganese peroxidase 

NPS Nonpoint source 

 

  



1 

 

CHAPTER 1: Introduction 

Nonpoint source (NPS) pollution from agricultural runoff is a primary driver of water quality 

impairment in the United States, threatening aquatic ecosystem and water resources through 

contamination from sediments, pathogens, and nutrients such as nitrogen (N) and phosphorus (P) 

(O’Geen et al., 2010; Zhang et al., 2024). Anthropogenic inputs of N and P are responsible for 

eutrophication that contributes to the growth of harmful algal blooms and dead zones that 

damage the economic and recreational viability of coastal areas (Malone & Newton, 2020). Key 

results from the 2015 National Coastal Condition Assessment revealed that eutrophication is a 

major problem in about 67% of total estuarine area, with the Gulf of Mexico in particular 

suffering due to nutrient loading from the intensive agricultural activity in the Upper Mississippi 

River Basin (U.S. Environmental Protection Agency, 2021). Although investments into more 

passive treatment of NPS pollution has led to some local improvements in smaller watersheds in 

the Mississippi River Basin, these efforts have not resulted in significant load reductions in the 

Gulf (Crawford et al., 2019), prompting research into more advanced yet sustainable treatment 

technologies that may be better suited to regulate NPS pollution. 

 

One method to manage NPS nitrate entering coastal waters involves enhancing microbial 

denitrification through subsurface biofilter systems such as woodchip bioreactors (WBRs) 

(McGuire et al., 2023; Schipper et al., 2010, Christianson et al., 2016). During denitrification, 

facultative anaerobic microbes that colonize woodchip surfaces convert NO3
- through 

dissimilatory reduction into non-reactive dinitrogen (N2) and the greenhouse gas nitrous oxide 

(N2O) (Warneke et al., 2011). Dissolved organic carbon released from decomposing wood serves 

as an electron donor and carbon source for denitrifying microorganisms. There are high levels of 



2 

 

NO3
- in agricultural runoff originating from substantial use of commercial fertilizer products, 

including manure, which will easily leach through soil into drainage water through soil macro-

pores (Mardani et al., 2020) due to low soil retention and high water solubility (Puckett, 1994). 

Thus, the critical challenges of managing N-contaminated waters concern developing cost-

effective sustainable systems to supply ample labile carbon to power microbial denitrification, 

enhancing NO3
- removal while limiting accumulation of N2O. Standard agricultural tile drainage 

systems contain low concentrations of ammonium (NH4
+) and labile carbon (McGuire et al., 

2023). WBRs employ lignocellulosic media (e.g. woodchips) as a renewable carbon source as a 

means for NPS-N management. The success of this method has led to an increase in their 

popularity in the past decade (Cameron & Schipper, 2010; Lopez-Ponnada et al., 2017; McGuire 

et al., 2023; Warneke et al., 2011). 

 

Wood is a lignocellulosic biopolymer composed of three main components (Figure 1): cellulose 

(45-55%), hemicellulose (24-40%), and lignin (18-35%) (Malherbe & Cloete, 2002; Lopez-

Ponnada et al., 2017). Lignin is a stable, phenolic heterogeneous polymer that entangles with 

cellulose and hemicellulose via ester and ether linkages, shielding more reactive sites from 

enzymatic hydrolysis (Zhao et al., 2022; De Bhowmick et al., 2018). Unlike cellulose and 

hemicellulose, lignin contains aromatic groups that are essential in creating the structural 

integrity of cell walls, also constructing pathways to facilitate fluid transport and defend against 

unwanted pathogens (Lankiewicz et al., 2023).  

 



3 

 

Figure 1. Schematic of the three major components of wood and their chemical structures. 

Chains of cellulose, made of linear glycosidic linkages, arrange to form microfibrils (Zhao et al., 

2022). Each microfibril is surrounded by hemicellulose (e.g., xylan), a branched 

heteropolysaccharide that covalently connects to a rigid lignin outer shell consisting of ρ-

hydroxyphenyl (H-type), guaiacyl (G-type), and syringyl (S-type) subunits (Malherbe & Cloete, 

2002).  

 



4 

 

 
Figure 2. Schematic of lignin degradation by white-rot fungi modified from Cragg et al. (2015). 

Laccase and lignin peroxidase are excreted through fungal hyphae. Laccases use oxygen as an 

electron acceptor to attack phenolic moieties and produce free radicals while lignin peroxidase 

use hydrogen peroxide (H2O2) as electron acceptor to depolymerize lignin into monomers 

(Burlacu et al., 2018). 

 

Lignin degradation (Figure 2) is traditionally performed by white-rot and brown-rot fungi and 

saprophytic bacteria that secrete extracellular oxygen-dependent enzymes (e.g. peroxidases and 

laccases) capable of depolymerizing phenolic and non-phenolic components (Zhao et al., 2022). 

Anaerobic lignin deconstruction is generally slow, although it can be facilitated through select 

fungal species (Lankiewicz et al., 2023) that rely on hydrolytic enzymes to catalyze oxidative 

depolymerization reactions (Anderson et al., 2024). WBRs are placed under continuous 

saturation to generate ideal anoxic conditions for denitrification, therefore, systems become 

increasingly carbon-limited with time due to the relatively slow rate of lignin degradation 



5 

 

(Korner & Zumft, 1989; McGuire et al., 2023) exacerbated by decreasing concentrations of 

dissolved organic carbon (DOC) with age and weathering (McGuire et al., 2021).  

 

Periodic drying-rewetting (DRW) cycles in WBRs may enhance nitrate removal rates by 

introducing aerobic periods for the carbon substrate, increasing both lignin degradation rate and 

availability of labile carbon for microbial use upon resaturation (Maxwell et al., 2020). While 

DRW cycles are typically associated with enhanced N2O emissions due to the anoxic inhibition 

of N2O (McGuire et al., 2021), some studies suggest that greater carbon availability may 

outweigh O2 restrictions, leading to reduced N2O accumulation during oxic-anoxic transitions 

(Feyereisen et al., 2020; McGuire et al., 2021). This system of oxic-anoxic cycling, hence, may 

enhance denitrification by accelerating the decomposition of recalcitrant, lignocellulosic material 

into labile carbon during oxic periods and stimulating activity of denitrifying microbes during 

anoxic conditions (Maxwell et al., 2020). While prior studies have shown that oxic-anoxic 

cycling can impact the quantity of dissolved organic carbon released from woodchips as well as 

bulk measures of carbon quality (e.g., SUVA254) (McGuire et al., 2021; Zhang et al., 2024), there 

is a lack of molecular-level data on how wood-derived DOC is impacted by oxygen exposure.   

 

There has been limited previous work presenting enzymatic molecular-level characterization of 

lignin due to the complexity of dissolved organic matter (DOM), which can be comprised of 

thousands of individual components of varying relative abundances. However, the introduction 

of Fourier transform ion cyclotron mass spectrometry (FT-ICR MS) by Cumisarow & Marshall 

(1974) presented a new way to interpret large sets of chemical information from complex 

systems. FT-ICR MS achieves high resolution, sensitivity, and mass accuracy over a wide mass 



6 

 

range, making it adept to characterizing dynamic data sets, especially in conjunction with other 

techniques such as electrospray ionization (ESI) (Bahureksa et al., 2021). Over the past twenty 

years, several studies have been conducted to standardize the interpretation of DOM over a 

diverse range of environments (Hur et al., 2010) and abiotic and biotic factors (D’Andrilli et al., 

2023). FT-ICR MS can sort detected formulas in a DOM sample as lipid-, protein-, amino sugar-

, cellulose-, lignin-, tannin-, and hydrocarbon-like by linking chemical identity with levels of 

hydrogenation and oxygenation as determined by H/C and O/C ratios (D’Andrilli et al., 2015). 

 

A common method of FT-ICR MS visualization is the van Krevelen diagram- plots of H/C 

versus O/C ratio (D’Andrilli et al., 2015) that allow identification of chemical reaction pathways 

to provide more information on oxygen saturation, microbial influence, and carbon saturation 

within a system (D’Andrilli et al., 2015). In 2015, D’Andrilli et al. proposed the introduction of a 

molecular lability boundary (MLB) that linked the lability of organic carbon with hydrogen 

saturation. Molecular formulas with H/C ≥ 1.5 are considered “more labile”, containing protein 

and amino acid-like compounds. A recent 2023 study by the authors further specified that lability 

is viewed from the perspective of heterotrophic and autotrophic microbes, while recalcitrance 

describes material that is either unchanged by biotic and abiotic factors, or an end-product of 

heterotrophic degradation (D’Andrilli et al., 2023). 

 

This work employs FT-ICR MS for the characterization of DOM from woodchip bioreactors to 

examine transformations of labile carbon in WBRs. First, to elucidate spatial effects on labile 

carbon, samples taken from the upstream “inlet” and downstream “outlet” at medium depth of a 

WBR were investigated for compositional changes along the flow path. Then, a time-series 



7 

 

experiment was conducted to compare samples that had undergone a series of oxic-anoxic 

cycling to a control anoxic-only group over time. These experiments sought to test the capability 

of FT-ICR MS to 1) elucidate DOM changes due to oxic-anoxic cycling and 2) visualize 

temporal and spatial influences in DOM composition. FT-ICR MS analysis will guide discussion 

on how the lability of DOC in WBRs may be affected by both enzymatic and nonenzymatic 

pathways for organic matter degradation. Parameters to be assessed through FT-ICR MS include 

the molecular weight distribution, carbon number distribution, oxygen number distribution, 

modified aromaticity index (AImod) (Koch & Dittmar, 2006), nominal oxidation state of carbon 

(NOSC), and double bond equivalent (DBE) distribution. Principal component analysis (PCA) 

was used to elucidate correlations between woodchip biogeochemical properties such as protein 

content, SUVA254, and DOC concentration, and calculated parameters from FT-ICR MS data. 

 

The goals of this body of work are to: 1) Evaluate DOM leached from woodchips in the 

bioreactor at the inlet or outlet, motivated by differences in dissolved oxygen at these two 

locations; and 2) Determine effects of fluctuating oxic-anoxic conditions on DOM leached from 

wood over a period of time, motivated by studies indicating that deliberate oxic-anoxic cycling 

can be an effective approach to boost carbon bioavailability in WBRs.  



8 

 

CHAPTER 2: Methods 

2.1 Site Description and WBR Design 

Woodchips were sourced from a 10-year-old WBR situated in Freeville, New York (Figure 3). A 

simple schematic of the inside of the WBR is depicted in Figure 4. The WBR covers an area of 

6.1 m by 3.1 m with a depth of 69 cm, buried beneath 36 cm of soil. The bioreactor was filled 

with ash woodchips and enclosed in an impermeable polyethylene sheet. The downstream 

section of the reactor is equipped with a V-notch weir, designed to maintain water levels at a 

minimum depth of 49 cm in the reactor. In June 2021, the bioreactor was intentionally drained 

and then excavated at two locations—inlet (Port1), and outlet (Port3)—to collect woodchip 

samples from three distinct depth intervals within the bioreactor: 65-70 cm (“deep”), 30-35 cm 

(“middle”), and 0-5 cm (“surface”) below the woodchip surface.  

 
Figure 3. Satellite image of the site of the WBR in Freeville, New York. 

 

 



9 

 

 
Figure 4. Schematic of the 10-year-old denitrifying woodchip bioreactor from which samples 

were collected from upstream (Port1), midstream (Port2), and downstream (Port3) locations.  

 

2.2 Sample Collection and Preparation 

The collected woodchip samples were subsequently frozen at -20°C prior to analysis. Samples 

from middle depth were used to elucidate compositional differences in DOM collected from the 

inlet and outlet ports. 

 

A new batch reactor series was designed to explore the effects of oxic-anoxic cycling on the 

efficiency of lignin degradation and subsequent nitrate removal. Woodchip samples were 

collected over a fifteen-day period from both the inlet and outlet flow of the batch reactors from 

April 22 to May 6, 2022, as displayed in Figure 5. For purposes of this work, only inlet samples 

taken from April 27 to May 3 were considered for further analysis through FT-ICR MS.  

 

Woodchips were thawed, rinsed, and pre-incubated in a synthetic bioreactor media. 



10 

 

 
Figure 5. Sample collection for WBRs subjected to oxic-anoxic and anoxic-only conditions. 

Each grey line dictates a day starting from April 18 to May 6. On April 24, the media was 

changed. April 27, April 30, and May 3 are referred to as Day 3, Day 6, and Day 9, respectively, 

indicating how many days have passed after changing media. 

 

2.2.1 Solid Phase Extraction (SPE) 

 

All samples were purified by solid phase extraction (SPE) to isolate DOM from porewater. 

During SPE, moderate hydrophobic to hydrophobic molecules are retained on the solid phase as 

the sample flows through a cartridge before the molecules are eluted by a selected solvent 

(Minor et al., 2014). The styrene divinyl benzene polymer PPL was selected as sorbent due to its 

highly efficient recovery of DOC from a range of highly polar to nonpolar substances (Dittmar et 

al., 2008). 

 

Samples were filtered and acidified in glass tubes to pH 2 with trace metal grade concentrated 

hydrochloric acid. The PPL cartridge was rinsed with methanol and acidified Milli-Q water to 

clean and prime prior to adding the samples. After loading, the PPL cartridge was rinsed with 15 

mL of pH 2 acidified water to remove salts. The cartridge was then dried under a gentle flow of 



11 

 

ultrapure nitrogen gas. Samples were then eluted using methanol. Sample extracts were placed in 

a freezer held at -20°C. 

 

2.3 Data Analysis 

2.3.1 Biogeochemical Properties of Woodchips 

Preliminary experiments for data concerning the inlet-outlet flow WBR were conducted by Iva 

Petrovic and Yi Sang (Sang et al., 2024). The measurements for the oxic-anoxic and anoxic-only 

time series experiments were performed by Yi Sang. All sets of preliminary data have been 

approved for the further analysis of the samples through FT-ICR MS. Measured biogeochemical 

properties of the triplicate inlet (labelled “IMA”, “IMB”, “IMC”) and outlet (labelled “OMA”, 

“OMB”, “OMC”) WBR samples are outlined in Table 1 and Figure 6. Figure 7 represents only 

single measurements taken at different time points as there was an insufficient number of 

samples collected to generate average measurements per data point. 

Table 1. Summary of biogeochemical properties of inlet and outlet WBR. 
Port Sample SUVA254 

(L/mg C-m) 

[DOC] 

(mg/L) 

pH Protein (mg/kg) C/N ratio 

Inlet IMA 1.041 17.48 6.58  

709.08 ± 32.62 

 

132.23 ± 36.35 IMB 1.065 19.45 6.53 

IMC 0.698 27.84 6.57 

Average 0.934 ± 0.205 21.59 ± 5.50 6.56 ± 0.03 

Outlet OMA 1.700 3.67 7.05  

247.31 ± 70.61 

 

350.70 ± 44.82 OMB 1.961 4.05 6.98 

OMC 0.459 7.28 7.07 

Average 1.374 ± 0.803 5.00 ± 1.98 7.03 ± 0.05 

 



12 

 

 
Figure 6. Nitrate removal batch experiments with woodchips collected from the inlet and outlet 

port of WBRs. At time 0, batch reactors were replaced with fresh nutrient media and were 

incubated anoxically. 

 

 
Figure 7. Measurements of A) SUVA254 and B) Dissolved organic carbon in the Anoxic Inlet and Cycle 

Inlet WBRs over time. 

 

 



13 

 

2.3.2 Negative-Ion Electrospray Ionization (ESI) FT-ICR MS 

All samples were ionized with negative ion mode ESI. ESI was chosen over other methods such 

as atmospheric pressure photoionization (APPI) due to the abundance of hydroxyl, carboxyl, and 

phenolic groups in lignin that are more easily deprotonated via ESI (Qi et al., 2016). 

FT-ICR MS data was obtained in 2022 with 21 Tesla FT-ICR mass spectrometer at the National 

High Magnetic Field Laboratory in Tallahassee, Florida USA. The mass spectrometer achieves 

high mass-resolving power (m/Δm50% > 2,700,000 at m/z 400) and mass accuracy (80 parts per 

billion (Smith et al., 2018) enabling high mass resolution of closely spaced peaks (Bahureksa et 

al., 2021). Molecular formulas were assigned by PetroOrg© software (Corilo et al., 2013). 

Formula assignments by PetroOrg© included combinations of C, H, N, O, and S within the 

ranges 12C1-100, 
1H1-150, 

14N0-2, 
16O1-30, and 34S0-2. Contributions from 13C isotopes were excluded. 

Peaks corresponding to SO3 and SO4-containing formulas were discarded, as they are widely 

observed chemical noise peaks resulting from plastic and soap contamination introduced to the 

samples during collection and preparation. Surfactants readily ionize due to their amphiphilic 

nature and ionizable head groups, so their existence within FT-ICR MS spectra may lead to 

incorrect assignments (Bahureksa et al., 2021). Molecules were grouped into compound classes 

based on the number of existing heteroatoms N, O, and S. Classes of CçHhOo, CçHhOoNn, 

CçHhOoNnSs, and CçHhOoSs were dominant across all samples and used for further analysis. 

Molecules were further classified by H/C and O/C ratios according to recent literature also aimed 

at characterizing DOM samples through (-) ESI FT-ICR MS (Zhuo et al., 2024): (1) lipids- 1.7 ≥ 

H/C ≤ 2.2, 0 ≥ O/C ≤ 0.2 (2) protein and amino sugars, 1.7 ≥ H/C ≤ 2.2, 0.2 ≥ O/C ≤ 0.6, (3) 

carbohydrates, 1.5 ≥ H/C ≤ 2.2, 0.6 ≥ O/C ≤ 1.0, (4) unsaturated hydrocarbons, 0.7 ≥ H/C ≤ 1.5, 



14 

 

0 ≥ O/C ≤ 0.1, (5) lignin 0.6 ≥ H/C ≤ 1.7, 0.1 ≥ O/C ≤ 0.6, (6) tannins, 0.5 ≥ H/C ≤ 1.5, 0.6 ≥ 

O/C ≤ 1.0, (7) condensed aromatics, 0.3 ≥ H/C ≤ 0.7, 0 ≥ O/C ≤ 1.0.  

 

ESI-FT-ICR MS is not an inherently quantitative technique, however, because each sample was 

measured on the same instrument with identical parameters, it is possible to compare peak 

magnitudes across a sample set. However, relating relative abundance to concentrations of 

detected compounds in the sample is not allowed due to varying ionization efficiencies of 

polyfunctional components in organic matter, influenced by factors such as acidity, 

hydrophobicity, and molecular weight of the detected analytes (Bahureksa et al., 2021; Sleighter 

et al., 2010). Therefore, data visualization methods such as van Krevelen diagrams, class 

distribution plots, and statistical methods are better suited to study patterns in DOM. 

 

2.3.3 Calculation of Parameters 

A series of magnitude-weighted parameters were determined for each sample based on their 

molecular formulas. The weighted total carbon number (C#
w), hydrogen to carbon ratio (H/Cw), 

oxygen to carbon ratio (O/Cw), nitrogen to carbon ratio (N/Cw), double bond equivalent (DBEw), 

modified aromaticity index (AImod,w) as described by Koch & Dittmar (2006), and nominal 

oxidation state of carbon (NOSCw) are calculated as follows: 

𝐷𝐵𝐸 = 𝐶 + 1 −
𝐻

2
+

𝑁

2
 

𝑁𝑂𝑆𝐶 = 4 −
4𝐶 + 𝐻 − 3𝑁 − 2𝑂 − 2𝑆

𝐶
 

𝐴𝐼𝑚𝑜𝑑 =
[1 + 𝐶 − 0.5𝑂 − 𝑆 − 0.5(𝑁 + 𝐻)]

𝐶 − 0.5𝑂 − 𝑁 − 𝑆
 



15 

 

𝐶𝑤 = 𝛴 (𝐶𝑖  × 𝑀𝑖) 𝐷𝐵𝐸𝑤 = 𝛴 (𝐷𝐵𝐸/𝐶𝑖  × 𝑀𝑖) 

𝐻/𝐶𝑤 = 𝛴 (𝐻/𝐶𝑖  × 𝑀𝑖) 𝐴𝐼𝑚𝑜𝑑,𝑤 = 𝛴 (𝐴𝐼𝑚𝑜𝑑,𝑖  × 𝑀𝑖) 

𝑂/𝐶𝑤 = 𝛴 (𝑂/𝐶𝑖  × 𝑀𝑖) 𝑁𝑂𝑆𝐶𝑤 = 𝛴 (𝑁𝑂𝑆𝐶 × 𝑀𝑖) 

𝑁/𝐶𝑤 = 𝛴 (𝑁/𝐶𝑖  × 𝑀𝑖) 𝑀𝑖 = 𝐼𝑖/𝛴𝐼𝑖 

where C, H, N, O, and S represent the number of carbon, hydrogen, nitrogen, oxygen, and sulfur 

atoms. Mi, the relative abundance of each individual peak i divided by the total relative 

abundance in the sample, is used to calculate the set of magnitude-weighted parameters, signified 

by w. The molecular lability index (%MLBL) of each sample was also calculated based on the 

methods detailed in D’Andrilli et al. (2015): 

%𝑀𝐿𝐵𝐿 = (
# 𝑜𝑓 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑓𝑜𝑟𝑚𝑢𝑙𝑎𝑠 𝑤𝑖𝑡ℎ

𝐻
𝐶 ≥ 1.5)

𝑡𝑜𝑡𝑎𝑙 # 𝑜𝑓 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑓𝑜𝑟𝑚𝑢𝑙𝑎𝑠
) 

 

The Python library pandas was used for data transformation and analysis. The Matplotlib library 

was used for data visualization. 

 

2.3.4 Statistical Analysis 

A principal component analysis (PCA) was performed on each set of data. PCA reduces 

multidimensional sample space with the first principal component, PC1, explaining most of the 

variance among the data set, and the second component, PC2, explaining residual variance 

(Sleighter et al., 2010) to show how samples may cluster based on how similar they are. 

“Loadings” refer to the projection of each variable in the dataset onto each PC, indicating how 

much that variable influences each PC, while “scores” represent the projection of each sample 

onto each PC (Xue et al., 2011; Sleighter et al., 2010). Magnitude-weighted parameters were 

calculated as detailed in Section 2.3.3. To minimize the contributions of low abundant peaks near 



16 

 

the noise level, it was decided to take into consideration only the top twenty percent of unique 

individual formulas to represent each sample. Then, PCA was applied to the matrix utilizing the 

relative magnitudes of the formulas in each sample. 

 

Independent t-tests were conducted to determine the existence or absence of significant 

differences between the means of the inlet and outlet data. The calculated t-value quantifies true 

differences between the averages, calculated as follows: 

𝑡 − 𝑣𝑎𝑙𝑢𝑒 =  
𝑥1̅̅̅ − 𝑥2̅̅ ̅

√(
𝑠1

2

𝑛1
+

𝑠2
2

𝑛2
)

 

where 𝑥1̅̅̅ and 𝑥2̅̅ ̅ represent the means of groups marked “1” and “2”, s1 and s2 represent the 

standard deviations of the two groups, and n1 and n2 representing each group’s sample size. The 

scripy-stats package for Python was used for calculation of the associated p-values that dictate 

the probability of occurrence of results.  



17 

 

 CHAPTER 3: Results 

3.1 Comparison of Inlet / Outlet Flow WBR 

In evaluating the results of FT-ICR MS analysis for spatial effects across the WBR, we assume 

that the success of nitrate removal over time as shown in Figure 6 will be align with the DOM 

composition at each location. We anticipate more microbial activity in the inlet due to exposure 

to oxygen entering the WBR, enhancing wood degradation and hence, carbon availability. Using 

nitrogen content as a proxy for microbial makeup (i.e., as amino acids and proteins), key 

expectations include higher average N/C ratio and higher number of CHON compounds in the 

inlet. Carbon, oxygen, and DBE distributions will point towards differing degrees of degradation 

of lignocellulosic material, with the inlet expected to be skewed toward lower carbon, oxygen, 

DBE, as well as AImod, values, reflecting greater lignin depolymerization and breakdown of 

aromaticity. NOSC should be higher in the inlet, consistent with the existence of more oxidized, 

rather than reduced, material, which implies active degradation. Sample clustering in PCA 

should demonstrate that outlet samples display less consistency due to energetic constraints on 

lignin degradation leading to more heterogeneity amongst organic material. Lastly, we expect 

that the inlet will have higher percentage of labile formulas as dictated by the MLB due to 

oxygen-stimulated microbial activity. 

 

Observations that would suggest suboptimal WBR performance include non-significant 

differences in CHO content, indicating limited lignin degradation, and in CHON and CHONS 

content, which would contradict the hypothesis microbial makeup is distinct between the inlet 

and outlet. Similarly, minimal differences in carbon, oxygen, DBE, and AImod that would 

indicate ineffective lignin degradation, and thus, suboptimal denitrification. 



18 

 

3.1.1 FT-ICR MS Spectra and DOM Composition of Inlet / Outlet Flow WBR 

Unique formulas, ranging from 9,000 to 12,000 per sample after removal of soap contamination 

and contributions from 13C isotopes, were assigned using PetroOrg©. The inlet averaged 9,327 

assigned formulas. The outlet averaged 10,635 assigned formulas. Formulas were categorized 

depending on heteroatom composition into classes CHO, CHON, CHONS, and CHOS (Table 2). 

Table 2. Number of identified compounds through FT-ICR MS of samples from inlet and outlet 

flow WBR. 
  Number of Detected Formulas 

Port Sample CHO CHON CHOS CHONS Total 

Inlet A 4790 1638 1846 465 9016 

B 5004 1546 1886 389 9058 

C 4988 2437 1723 527 9908 

Average 4927±119 1874±490 1818±85 460±69 9327±503 

Outlet A 6198 903 1510 87 9240 

B 6491 1091 2001 146 10012 

C 8356 1318 2261 114 12653 

Average 7015±1171 1104±208 1924±381 116±30 10635±1790 

An independent t-test was conducted to confirm that there was no significant difference in the 

total average number of detected formulas between the inlet and outlet samples (t(4)= -1.218; p= 

0.29) and the detected CHOS formulas (t(4)=-0.468; p=0.66), given the p-value associated with 

the calculated t-values is less than the 0.05 threshold, supporting acceptance of the null 

hypothesis. Statistically significant differences were observed in the number of CHO formulas 

(t(4)= -3.073, p=0.03) and CHONS formulas (t(4)= 7.942; p=0.00), as the associated p-values are 

less than 0.05. Additionally, while the associated p-value for the difference between CHON 

formulas (t(4)= 2.504; p=0.066) was not within the 95% confidence interval, it is close enough to 

claim there is some probable difference between the two means. 

 

Because lignin is entirely composed of the three elements carbon, hydrogen, and oxygen, CHO-

containing compounds were used to reconstruct the mass spectra of each sample (Figure 8).  



19 

 

 
Figure 8. ‘CHO’-reconstructed mass spectra of inlet A, B, C and outlet D, E, F triplicate 

samples. 

 

Peaks in the spectra span an m/z range of 200 to 1000, which aligns with similar studies also 

analyzing water-sourced DOM samples (Minor et al., 2014). The mass spectra are bimodal with 

peaks centered around 325 and 425. Spectra between the inlet and outlet are highly similar, with 

the outlet samples achieving slightly higher relative abundances across the entire m/z range. 

Major differences in relative abundance of detected ions are observed across the entire m/z 

range, with the outlet averaging higher relative abundance. Because differing ionization 

efficiencies exist for polyfunctional DOM species, quantification of FT-ICR MS data is less 

sound and data should preferably be interpreted qualitatively. Diving more into the visualization 

of compositional trends provides a safer way of interpreting differences in FT-ICR MS spectra 



20 

 

rather than relating peak magnitudes back to concentrations of components in the original 

samples, as each peak in the mass spectrum is normalized to the most abundant peak. 

 
Figure 9. Class-separated van Krevelen diagrams of triplicate inlet samples A, B, C and 

triplicate outlet flow samples D, E, F. Plot points are classified based on regions of H/C and O/C 

ratio dictated by Zhou et al., 2024: (1) lipids- 1.7 ≥ H/C ≤ 2.2, 0 ≥ O/C ≤ 0.2 (2) protein and 

amino sugars, 1.7 ≥ H/C ≤ 2.2, 0.2 ≥ O/C ≤ 0.6, (3) carbohydrates, 1.5 ≥ H/C ≤ 2.2, 0.6 ≥ O/C ≤ 

1.0, (4) unsaturated hydrocarbons, 0.7 ≥ H/C ≤ 1.5, 0 ≥ O/C ≤ 0.1, (5) lignin 0.6 ≥ H/C ≤ 1.7, 0.1 

≥ O/C ≤ 0.6, (6) tannins, 0.5 ≥ H/C ≤ 1.5, 0.6 ≥ O/C ≤ 1.0, (7) condensed aromatics, 0.3 ≥ H/C ≤ 

0.7, 0 ≥ O/C ≤ 1.0. 



21 

 

Class-separated van Krevelen diagrams (Figure 9) reveal most molecules exist as heterogeneous 

lignin- and tannin-like. CHON1S1 and CHON2S1 species dominate in the inlet of the bioreactor, 

with “CHO” referring to compounds with carbon, hydrogen, and oxygen atoms within the ranges 

C1-100, H1-150, and O1-30, and subscripts denoting the number of nitrogen and sulfur heteroatoms 

present in the species- as previously confirmed by Table 2. These species concentrate within the 

lignin-like region of the diagram. While ranges for H/C and O/C ratios below the MLB (H/C = 

1.5) do not vary, above the MLB, detected compounds extend to higher O/C ratios in the inlet. 

These compounds are mostly carbohydrate-like. Across all samples, the compounds that 

dominate above the MLB are of the classes CHO and CHOS1. There are no substantial 

differences in the amount of detected lipid-like compounds. However, the outlet samples have a 

greater abundance of compounds with low degrees of oxygenation and hydrogenation which are 

identified as condensed aromatic species. 

 

3.1.2 Distribution Patterns of Oxygen, Carbon, and DBE 

Oxygen, carbon, and DBE distributions were plotted to probe further evidence for increased 

lignin degradation in the inlet samples (Figure 10). Carbon and DBE distributions were plotted 

using all identified compounds previously quantified in Table 2. Oxygen distribution were 

plotted using only CHO-containing compounds in order to focus on solely lignin-related 

material. 

 



22 

 

 
Figure 10. Box plots showing A) oxygen, B) carbon, and C) DBE distributions of the inlet and 

outlet samples. Red solid line represents median value. Blue dashed line represents mean value.  

 

The oxygen distribution graph (Figure 10A) reveals that the inlet contains compounds with 

oxygen numbers in the interquartile range (IQR) from O8 to O15. The outlet has a slightly wider 

range, from O8 to O18. The mean oxygen number for CHO compounds in the inlet was 12 with a 

median of 11. In the outlet, the mean and median both round to 13. Results of the t-test (t(4)=-

2.65, p=0.058) showed that there is some plausible difference between means. 

 



23 

 

The carbon number distribution plot (Figure 10B) shows that carbon-containing species in the 

inlet IQR span from C18 to C27, with a mean and median of 23. The outlet exhibits a wider range, 

from C19 to C31, and a larger mean and median at 26 and 24, respectively. Additionally, the outlet 

contains higher-numbered outliers. A t-test was performed on the average carbon numbers (t(4) = 

-4.34, p=0.01) to ultimately confirm that there was a significant difference between the inlet and 

outlet. 

 

Figure 10C shows the average distribution of double bond equivalents of all compounds detected 

in the inlet and outlet triplicate samples. DBE, referring to the level of unsaturation of a 

compound as measured by aromatic rings and double bonds, reflects sample aromaticity and 

complexity. The inlet IQR spans from 9 to 16, with the outlet slightly wider from 9 to 18. The 

mean and median of the inlet, 12.20 and 12.0, are moderately lower than that of the outlet, 13.7 

and 13.0. Results of the t-test (t(4)= -4.66, p= 0.01) confirmed that these differences are 

statistically significant. 

 

In summary, the inlet exhibits lower average and median oxygen number, carbon number, and 

DBE. Additionally, the inlet IQR across all plots is considerably narrower and tends to skew to 

lower values. 

 

3.1.3 Modified van Krevelen Plots of Nitrogen-Containing Groups 

The difference in the number of N- and S-containing compounds between the inlet and outlet 

samples prompted further analysis through a modified version of a van Krevelen graph plotting 

H/C and N/C ratios with bubble size reflecting relative abundance (Wang et al., 2024). 



24 

 

 
Figure 11. Modified 3D van Krevelen plots of nitrogen-containing compounds in inlet A, B, C 

and outlet D, E, F samples. Bubble size reflects abundance, with larger size indicating higher 

occurrence of compounds associated with a specific H/C : N/C ratio. 

 

Figure 11 confirms increased levels of compounds of type CHON1, CHON2, and CHON1S1 

upstream in the bioreactor. There are no detected compounds of formula CHON2S1 in the outlet. 

Compounds of formula CHON2 show an extended range of N/C ratios in the inlet samples while 

being in a similar H/C range to the outlet samples. The majority of CHON2 compounds were 



25 

 

found centered at H/C = 1.0 and N/C = 0.1. CHON1S1 compounds in the inlet displayed a 

broader H/C range than that in the outlet. Quantification of compounds corresponding to 

different N/C ratios is given in Table 3. 

Table 3. N/C ratio distribution of nitrogen-containing compounds in the inlet and outlet samples. 
  N/C Ratio 

Port Sample 0.025 < N/C 

≤ 0.033 

0.033 < N/C 

≤ 0.05 

0.05 < N/C 

≤ 0.067 

0.67 < N/C 

≤ 0.1 

0.1 < N/C 

≤ 0.2 

Inlet A 55 1082 524 328 114 

B 58 1017 456 296 108 

C 250 1392 613 494 204 

Average 121 ± 112 1163 ± 200 531 ± 79 373 ± 206 142 ± 54 

Outlet A 14 622 276 62 16 

B 58 727 306 109 37 

C 102 894 312 100 24 

Average 58 ± 44 748 ± 137 298 ± 19 90 ± 25 26 ± 11 

An independent t-test was performed to determine whether there is sufficient evidence to say that 

the average amounts of compounds in each N/C range are different between the two sample 

groups. While there is no real difference found for compounds corresponding to 0.025 < N/C ≤ 

0.033 (t(4)= 0.909; p=0.41), it is confirmed that the inlet has a higher number of nitrogen-

containing compounds of N/C > 0.033 than the outlet, and that the difference between the two is 

significant (0.033 < N/C ≤ 0.05: t(4)= 2.97; p= 0.04, 0.05< N/C ≤ 0.067: t(4)= 4.98; p= 0.01, 

0.067 < N/C ≤ 0.1: t(4)= 4.48; p= 0.01, 0.1 < N/C ≤ 0.2: t(4)= 3.68; p= 0.02), which is 

constituent with expectations. 

 

3.1.5 Determining Amount of Labile Carbon in Inlet and Outlet WBR Samples 

 

The percentage of formulas in each sample deemed “more labile” through application of the 

molecular lability boundary was calculated for both series. 



26 

 

 
Figure 12. Average percentage of labile formulas in the inlet and outlet according to application 

of the molecular lability boundary coined by D’Andrilli et al. (2015). 

 

The average number of labile formulas detected were highly similar but with higher standard 

deviation amongst the outlet triplicate samples. An independent t-test was performed, confirming 

that the difference in means was most likely not statistically significant (t(4)=0.217, p=0.84). 

However, the average percentage of labile formulas in the inlet and outlet was drastically 

different (t(4)= 6.22, p=0.003), as shown in Figure 12.  

 

3.1.6 Calculated DOM Parameters and Principal Component Analysis (PCA) 

 

The modified AI (Figure 13A) and NOSC (Figure 13B) were calculated using all identiifed 

formulas. AI allows for identification of aromatic and condensed aromatic species in organic 

matter, with the modified approach assuming that half of the detected oxygen are bound in sigma 

bonds rather than pi bonds as carboxyl groups (Koch & Dittmar, 2006). Koch & Dittmar (2006) 



27 

 

proposed that AI > 0.5 is a threshold for aromatic structures, with AI ≥ 0.67 indicated condensed 

aromatic species. 

 
Figure 13. Distribution of A) aromaticity index and B) NOSC for inlet and outlet samples. 

 

The inlet averaged marginally lower AI at 0.3623 (t(4)=-3.4311, p=0.0265), which is below the 

threshold for aromaticity, which is to be expected for active breakdown of complex aromatic 

material. Although the inlet averaged higher NOSC (t(4)=1.2466, p=0.2806), this was not proven 

to be a significant difference, which was not expected. However, interestingly, positive NOSC 

values, typically indicative of oxidized carbon, are only seen in the inlet samples. More negative 

NOSC values, consistent with more reduced carbon, are seen in the outlet samples. 

 

Next, a set of parameters were calculated using only the top 20% of molecular formulas ranked 

by their relative abundance per sample to minimize any contributions from low abundant 

formulas that may be mistaken for spectral noise (Table 4). These parameters, in conjunction 

with the series of bulk measurements described in Section 2.3.3 and calculations of %MLBL, 

were used to perform a PCA to ultimately determine whether the inlet samples exhibited more 



28 

 

consistency in the chosen variables than outlet samples. Creation of a PCA biplot (Figure 14) 

also enables analysis of possible relationships between variables themselves.  

Table 4. Summary of calculated parameters for PCA using top 20% of formulas ranked by 

relative abundance. 
Port Sample C#

w H/Cw O/Cw N/Cw DBEw AImod,w NOSCw %MLBL 
Inlet A 12.0 0.77 0.26 3.08ꞏ10-4 5.48 0.181 -0.207 6.69 

B 11.4 0.76 0.28 2.94ꞏ10-4 5.36 0.181 -0.165 6.97 

C 11.4 0.74 0.27 1.77ꞏ10-3 5.31 0.175 -0.166 6.36 

Average 11.6

±0.3 

0.76 

±0.01 

0.27 

±0.01 

7.90ꞏ10-4  

±8.47ꞏ10-4   

5.39±

0.09 

0.179± 

0.003 

-0.179± 

0.024 

6.67± 

0.30 

Outlet A 13.1 0.81 0.26 1.52ꞏ10-5 5.85 0.182 -0.254 4.11 

B 12.7 0.75 0.29 3.86ꞏ10-5 6.07 0.192 -0.154 4.85 

C 13.7 0.69 0.30 0 7.05 0.218 -0.075 3.62 

Average 13.1

±0.5 

0.75 

±0.06 

0.28 

±0.02 

1.79ꞏ10-5  

±1.94ꞏ10-5  

6.32±

0.64 

0.197± 

0.019 

-0.161 

±0.090 

4.19± 

0.62 

 

 
Figure 14. Biplot from principal component analysis of inlet and outlet WBR samples using the 

top 20% of formulas ranked by relative abundance. 

 

The first two principal components explain 89.82% of the variance amongst these samples. PC1 

explains nearly 60% of the variance and is related mostly to aromaticity; samples that have 

higher aromaticity (i.e., high DBE, high AImod,w) have more positive PC1 scores. Samples that 



29 

 

have more negative PC1 scores exhibit less aromaticity. Interestingly, main indicators of sample 

aromaticity, DBEw and AImod,w, are not highly correlated with SUVA254. Samples that have 

higher SUVA254 measurements exhibit more positive PC2 scores. Tighter clustering is observed 

in the inlet triplicate samples. These samples are highly correlated to %MLBL, [DOC], and N/Cw 

and inversely correlated with pH and C#
w. An inverse relationship between NOSCw and H/Cw is 

also observed. Weaker correlations exist between the variables and the outlet samples.  

 

3.2 Effect of Oxic-Anoxic Cycling on DOM 

The effects of oxic-anoxic cycling would manifest in distinguishable differences between DOM 

composition in the immediate aftermath of the oxygen exposure that would then diminish over 

time as anoxic conditions dominate inside the WBR. We would anticipate that oxic-anoxic 

cycling would lead to more nitrogen-containing residues, reflected in higher average N/C ratio 

and number of detected nitrogen-containing CHON and CHONS species, due to the stimulation 

of both aerobic and anaerobic microbes. This would be consistent with higher degradation of 

lignin at the start of the anoxic incubation, which could possibly be due to more efficient 

degradation by the microbial community. Degradation of lignin exposes the layers of cellulose 

and hemicellulose that would are more bioavailable forms of carbon for microbial usage. Thus, 

the WBR that had undergone oxic-anoxic cycling would see carbon, oxygen, and DBE 

distribution skewed to lower and narrower ranges, higher average NOSC, and greater percentage 

of labile carbon. The decreasing effect of oxic-anoxic cycling would be evident through tighter 

clustering of samples in PCA over time regardless of experimental conditions. Conversely, we 

expect anoxic-only conditions to start off with some microbial activity that decreases due to 



30 

 

limited oxygen and nutrient availability as internal conditions stabilize inside the reactor over 

time. 

 

If oxic-anoxic cycling was not successful in accelerating the rate of lignin degradation, and 

hence, was not contributing to substantial nitrate removal over time, analysis through FT-ICR 

MS will reveal mostly non-significant differences in carbon, oxygen, and DBE distribution 

between the oxic-anoxic samples and the anoxic-only samples, as well as average NOSC and 

AImod, indicating that the system is dominated by solely anaerobic activity. 

 

3.2.1 FT-ICR MS Spectra and DOM Composition of Oxic-Anoxic and Anoxic-Only Conditions 

The number of unique formulas ranged from 8,000 to 12,000 per sample after removal of soap 

contamination and contributions from 13C isotopes. Detected formulas were quantified and sorted 

into classes as shown in Table 5. Mass spectra of inlet samples from oxic-anoxic (‘Cycle Inlet’) 

and anoxic-only conditions (‘Anoxic Inlet’) over the period from Day 3 to Day 9 were 

reconstructed using only the ‘CHO’ class of compounds. Peaks across all samples were detected 

within an m/z range of 200-1000 Da, which is commonly observed in DOM spectra (Minor et 

al., 2014). A comparison of spectra at each time point is provided in Figure 15. 

Table 5. Identified compounds from Cycle and Anoxic samples. 
  Number of Detected Formulas 

Sample Days after 

Changed Meda 

CHO CHON CHOS CHONS Total 

Cycle 3 7514 572 1074 4 9406 

6 7461 508 1721 24 10251 

9 6916 578 1779 18 9876 

Anoxic 3 6703 1872 1754 215 11863 

6 6650 1349 2185 195 11731 

9 6347 389 1575 5 8773 



31 

 

In both series, there is a gradual decrease in the number of CHO compounds detected over time. 

In the Cycle Inlet series, there is an increase in CHOS compounds. In the Anoxic Inlet series, 

there is a reduction of CHON and CHONS compounds as well as the total number of compounds 

over time. 

 
Figure 15. Comparison of ‘CHO’-reconstructed mass spectra of inlet samples from oxic-anoxic 

cycling “Cycle Inlet” and anoxic-only “Anoxic Inlet” conditions, A) 3 days B) 6 days and C) 9 

days following changed media. 



32 

 

Formulas vary the most across the two sample groups around the m/z range 200 to 500, with 

anoxic-only samples from Day 3 and Day 6 exhibiting a higher abundance of lower molecular 

weight compounds. The mass spectra across all samples appear to be bimodal with clear peaks 

around 325 and 425 m/z. On Day 9, it appears that the molecular composition of the oxic-anoxic 

and anoxic-only samples eventually converge (Figure 15C). Cycle Inlet samples consist of 

mostly peaks of a lower relative abundance compared to Anoxic Inlet samples. Both groups 

demonstrate a characteristic peak around 350 m/z. Due to the complexity of components found 

in heterogeneous lignin, along with concerns of differing ionization efficiencies, it is difficult to 

both elucidate a distinct structure for each compound through FT-ICR MS and tie peak 

abundances to quantifiable amounts of sample. Thus, analysis through visualization of 

molecular-level trends is necessary for full understanding of the spectral data. 

 

Elemental H/C and O/C ratios were calculated to project formulas onto a van Krevelen plot 

(Figure 16) which helps visualize possible chemical reaction pathways occurring during the 

experiment that may influence the characterization of DOM due to organic matter sources, 

oxygen saturation, and microbial influence. 



33 

 

 
Figure 16. Class-separated van Krevelen plots of A) Cycle Inlet and B) Anoxic Inlet samples 

from i) 3 days, ii) 6 days, and iii) 9 days after changed media. Each plot is separated into regions 

based on H/C and O/C ratio as dictated by Zhou et al., 2024: (1) lipids- 1.7 ≥ H/C ≤ 2.2, 0 ≥ O/C 

≤ 0.2 (2) protein and amino sugars, 1.7 ≥ H/C ≤ 2.2, 0.2 ≥ O/C ≤ 0.6, (3) carbohydrates, 1.5 ≥ 

H/C ≤ 2.2, 0.6 ≥ O/C ≤ 1.0, (4) unsaturated hydrocarbons, 0.7 ≥ H/C ≤ 1.5, 0 ≥ O/C ≤ 0.1, (5) 

lignin 0.6 ≥ H/C ≤ 1.7, 0.1 ≥ O/C ≤ 0.6, (6) tannins, 0.5 ≥ H/C ≤ 1.5, 0.6 ≥ O/C ≤ 1.0, (7) 

condensed aromatics, 0.3 ≥ H/C ≤ 0.7, 0 ≥ O/C ≤ 1.0. 

 

Anoxic Inlet samples show higher initial abundance of heterogeneous carbon sources composed 

of both nitrogen and sulfur, classified as CHON1S1 and CHON2S1, that decrease with time, as 



34 

 

previously shown in the data in Table 5. This finding is surprising giving the initial hypothesis 

that the Cycle Inlet samples will have a higher initial abundance of nitrogen-containing 

compounds. However, interestingly, decreases are observed in the abundance of CHON1 and 

CHON2 compounds over time in the Anoxic series, whereas these classes are relatively constant 

in abundance over time after oxic-anoxic cycling. Additionally, Cycle Inlet samples show an 

increasing abundance of lignin-like CHOS1. Condensed aromatic-like CHO compounds of low 

O/C and H/C ratio seem to slightly decrease over time in the Anoxic series. There are visible 

shifts in the number of compounds with high H/C ratios above 1.5, particularly those that are 

protein and amino sugar-like and carbohydrate-like. 

 

3.2.2 Distribution Patterns of Carbon, Oxygen, and DBE 

Figure 17 illustrates changes in the distribution of compounds with certain numbers of double 

bond equivalents over time. 



35 

 

 
Figure 17. DBE distribution over time for Cycle Inlet and Anoxic Inlet samples A) 3 days, B) 6 

days, and C) 9 days after changed media. 

 

Cycle Inlet samples initially exhibit more compounds of higher DBE (12 to 27) compared to 

Anoxic Inlet samples, which contain more of lower DBE (0 to 11). On Day 6, there is minimal 

variation. The Cycle Inlet samples maintain two peaks at 10 and 14 over time. On Day 9, the 



36 

 

distribution of DBE values in both samples begins to converge. The Cycle Inlet sample has a 

slightly higher percentage relative abundance of compounds of lower DBE (6 to 12). Both 

samples now appear bimodal with two peaks at DBE values 10 and 14. The Anoxic Inlet sample 

has a slightly higher abundance of compounds of DBE > 13.  

 
Figure 18. Carbon number distribution of ‘CHO’-containing compounds time for Cycle Inlet 

and Anoxic Inlet samples A) 3 days, B) 6 days, and C) 9 days after changed media. 



37 

 

Carbon number distribution plots (Figure 18) also depict the convergence of molecular data over 

time. A higher carbon number would typically be associated with more intact, complex organic 

matter. Initially, the Cycle Inlet sample contains more higher-numbered compounds. There are 

no substantial differences between Days 3 and 6. The Anoxic Inlet samples show a decrease in 

compounds C8 to C21 from Day 6 to 9, but the percentage relative abundance of higher-numbered 

compounds remains relatively constant. Comparatively, the carbon number distribution of the 

Cycle Inlet samples changes minimally over time, with the most distinguishable change being a 

slight increase of compounds C9 to C18 from Day 6 to 9. Across the duration of the experiment, 

both sample sets observe a maximum at C18. 



38 

 

 
Figure 19. Oxygen number distribution of ‘CHO’-containing compounds over time for Cycle 

Inlet and Anoxic Inlet samples A) 3 days, B) 6 days, and C) 9 days after changed media. 

 

The distribution of CHO classes was also plotted (Figure 17). Cycle Inlet samples show numbers 

ranging from O3 to O30, while Anoxic Inlet displays a range from O1 to O25. Contributions from 

formulas with O1 and O2 were minimal. Figure 17 demonstrates a trend identical to that observed 

in the spectral composition (Figure 13), DBE distribution (Figure 15), and carbon number 



39 

 

distribution (Figure 14), with a convergence of the data by the end of the experimental period. 

The Anoxic Inlet sample initially exhibits more lower-numbered species from Day 3 to 6. On 

Day 9, besides a distinct peak at O3, the percent relative abundance of compounds from O4 to O11 

decreases significantly. The Cycle Inlet samples do not vary much from Day 6 to 9. There are 

increases in the abundance of compounds from O3 to O11 observed on Day 9. 

 

3.2.3 Modified van Krevelen Plots of Nitrogen-Containing Compounds 

 

 
Figure 20. Modified van Krevelen plot of nitrogen-containing compounds in A) Cycle Inlet and 

B) Anoxic Inlet samples from i) 3 days, ii) 6 days, and iii) 9 days after changed media. Bubble 

size reflects abundance, with larger size indicating higher occurrence of compounds associated 

with a specific H/C : N/C ratio. 



40 

 

Modified van Krevelen plots show that the Anoxic Inlet samples from Day 6 to 9 have a greater 

range of both H/C and N/C ratios compared to Cycle Inlet samples. In general, there are more 

nitrogen-containing compounds in the Anoxic Inlet samples at these time points. The range of 

H/C ratios is constant over time for the Cycle Inlet samples, while it gradually decreases for the 

Anoxic Inlet samples. There is an almost entire reduction of CHON1S1 compounds from Day 6 

to 9. During this period, there is a noticeable reduction in the number of nitrogen-containing 

compounds with higher N/C ratios greater than 0.125. 

Table 6. N/C ratio distribution of nitrogen-containing compounds in the Cycle Inlet and Anoxic 

Inlet samples. 
  N/C Ratio 

Sample Date 0.025 < N/C 

≤ 0.033 

0.033 < N/C 

≤ 0.05 

0.05 < N/C 

≤ 0.067 

0.67 < N/C 

≤ 0.1 

0.1 < N/C 

≤ 0.2 

Cycle April 27 12 401 128 24 11 

April 30 7 384 117 23 1 

May 3 1 304 198 76 17 

Anoxic April 27 5 837 546 420 279 

April 30 0 599 421 292 212 

May 3 5 243 92 29 20 

Table 6 confirms that while there are initially more compounds of N/C > 0.033 in anoxic 

conditions, this amount becomes increasingly more comparable with the distribution seen in 

oxic-anoxic conditions, particularly on Day 9. 

 

3.2.4 Application of the MLB to Determine Levels of Labile Carbon 

The percentage of more “labile” and more “recalcitrant” formulas, based on the idea of the 

molecular lability boundary theorized by D’Andrilli et al. (2015), was calculated for each sample 

(Figure 21). 

 

 



41 

 

 

 
Figure 21. Percentage of labile formulas over time in the Cycle Inlet and Anoxic Inlet samples. 

 

Figure 21 depicts an interesting difference between the Cycle Inlet and Anoxic Inlet groups. It 

appears that the samples that underwent oxic-anoxic cycling exhibit percentages of labile carbon 

according to the molecular lability boundary. The percentage of “labile” formulas is consistent 

across the time series with a slight increase on Day 9. Whether the increase observed on Day 9 is 

substantial or simply a result of fluctuating conditions in the inlet would be a point of future 

study. The Anoxic Inlet samples start with a high number of labile formulas on Day 3 that 

quickly decreases as the experiment progresses.  

 

3.2.5 Calculated Parameters and Principal Component Analysis 

 

A total of 50,131 formulas across all samples were identified after exclusion of soap 

contamination and contributions from 13C isotopes. AImod (Figure 22A) and NOSC (Figure 22B) 

were then calculated for each sample. 



42 

 

 
Figure 22. Average A) aromaticity index and B) nominal oxidation state of carbon for Cycle 

Inlet and Anoxic Inlet samples over time. 

 

Interestingly, Figure 22A shows that the AI was more or less consistent in the Cycle Inlet series 

over time. The average was also higher on Days 3 and 6 than that of the Anoxic Inlet series. 

However, aromaticity index for the Anoxic Inlet series increases from 0.328 to 0.373 over time, 



43 

 

which may suggest some accumulation of organic material or decreasing lignin degradation 

capacity. 

 

When lignin degrades, NOSC is expected to increase as carbon atoms become more oxidized. 

Figure 22B shows that average NOSC was positive over the entire duration of the time-series for 

the Cycle Inlet series. The initial average NOSC for the Anoxic Inlet sample on Day 3 was 

negative, meaning a majority of carbon was reduced. On Days 6 and 9, the Cycle Inlet sample 

had a more positive average NOSC which means lignin is more oxidized, with a maximum value 

of 0.0781 on Day 6. Notably, the range of NOSC values is extremely narrow and near zero. 

 

Following the removal of duplicate formulas, 10,285 unique formulas remain across the sample 

set. The top 20% of dominant peaks in each sample were taken to represent each sample to 

minimize contribution from peaks of low relative abundance that may be mistaken for spectral 

noise (Sleighter et al., 2010). Parameters detailed in Table 7 were calculated using only the 

detected peaks chosen for principal component analysis. Preliminary measurements such as 

SUVA254 and [DOC], listed in Section 2.3.1, were also employed as variables for PCA. 

Table 7. Summary of calculated parameters of top 20% of peaks ranked by relative abundance. 
Sample Days after 

Changed 

Media 

C#
w H/Cw O/Cw N/Cw DBEw AImod,w NOSCw 

Cycle 3 14.44 0.65 0.30 1.32×10-4 7.40 0.204 -0.0453 

6 14.06 0.60 0.29 0 7.37 0.210 -0.0095 

9 13.17 0.66 0.30 1.32×10-5 6.82 0.216 -0.0566 

Anoxic 3 13.37 0.82 0.32 3.15×10-4 6.37 0.220 -0.1740 

6 12.83 0.71 0.32 1.26×10-4 6.61 0.238 -0.0630 

9 13.39 0.64 0.27 3.77×10-5 6.80 0.205 -0.0850 

 



44 

 

 
Figure 23. Date-separated PCA biplot generated from Cycle Inlet and Anoxic Inlet samples 

using top 20% of formulas ranked by relative abundance. 

 

The first two principal components account for 86.45% of the total variance amongst the 

samples. PC1 is most directly correlated to SUVA254, so samples with more positive PC1 scores 

will generally have more aromatic content. [DOC] is negatively directed along PC2, so samples 

with more negative PC2 scores are expected to have higher concentration of dissolved organic 

carbon. [DOC] is highly correlated with Cycle and Anoxic samples from the end of the anoxic 

incubation. The Day 3 Anoxic Inlet sample is highly correlated with N/Cw, O/Cw, and H/Cw. 

NOSCw exhibits an inverse relationship with H/Cw. Interestingly, DBEw and AImod,w are not 

highly correlated with each other despite both being typical measures of aromaticity. Other 

variables do not display clear trends with each other or the samples. Results of the PCA show 

increasingly tight clustering of samples over time, regardless of sample type, which is consistent 

with previous results.  



45 

 

CHAPTER 4: Discussion 

4.1 Impact of Flow on DOM Composition and Lability 

4.1.1 Spatial Influences on Microbial Lignin Degradation 

Due to exposure to oxygen entering the bioreactor, the inlet will temporarily experience more 

“oxic-like” conditions which decrease across both length and depth (Jéglot et al., 2021). 

Increased activity from aerobic and anaerobic microbes diminishes across the length of the 

bioreactor, where bioavailable carbon becomes increasingly limited thereby controlling 

microbial activity and diversity in near-anoxic conditions. More easily degradable materials such 

as hemicellulose and cellulose will be first converted into labile components, with more 

recalcitrant compounds later migrating downstream to a less ideal environment for degradation. 

Similar studies exploring laccase-treated lignin samples through FT-ICR MS noted that 

enzymatic activity may reduce chemical diversity which results in decreasing abundance of 

detected ions in the mass spectrum with time (Towle et al., 2024; Echavarri-Bravo et al., 2019), 

supporting the difference between the inlet and outlet spectra in Figure 8. Higher relative 

abundance observed in the outlet spectra over the entire m/z range suggests a lower degree of 

degradation. Additionally, there may be some condensation or accumulation of lignin 

monomeric components, such as aromatic aldehydes vanillin, vanillic acid, and ferulic acid, and 

nonaromatic acetovanillone, syringaldehyde, and acetosyringone (Podgorski et al., 2012) across 

the length of the bioreactor because of rate-limiting oxidative polymerization that prevents 

generation of compounds necessary for microbial metabolism into labile components (Anderson 

et al., 2024). Oxygen-poor and nonhydrolyzable compounds of higher molecular weight may 

begin to accumulate. This may explain the existence of condensed aromatic species in the outlet 

(Figure 9).  



46 

 

Box plots were created to elucidate trends in molecular composition. In the carbon distribution 

plot (Figure 10B), the outlet most likely contains a wider range of carbon numbers due to the 

heterogeneity of organic matter - consisting of recalcitrant lignocellulose, accumulated 

byproducts, and labile material inflowing from upstream and generated by some limited 

microbial activity. The inlet had a lower average and range of detected carbon number, which 

supports the idea that there is a greater number lignin-derived products formed at this location of 

the WBR. The carbon pool is generally more degraded, meaning that the compounds will have 

lower molecular weights. The oxygen distribution graph (Figure 10A) also demonstrates a 

similar trend, with the outlet exhibiting a slightly wider range of oxygen-numbered compounds. 

Narrower diversity in the inlet may reflect more fresh, labile biomass (Podgorski et al., 2012) as 

microbes preferentially decompose labile substrates, leaving behind more complex components 

like residual and partially oxidized lignin and cellulose thus enriching the diversity of oxygen-

containing compounds. Higher abundance of compounds with lower oxygen numbers suggests 

high concentration of simpler, assimilable molecules resulting from enzymatic degradation. This 

finding is supported by the protein and C/N measurements shown in Table 1. The outlet exhibits 

higher amounts of compounds of higher oxygen number, representing residual lignocellulosic 

material and accumulated nonhydrolyzable matter. The results of the DBE distribution graph 

(Figure 10C) further confirm the validity of the observations from Figures 10A and 10B that the 

outlet is accumulating more complex and recalcitrant lignocellulosic material (higher average 

DBE), while the inlet experiences more breakdown of aromatic components into simpler 

structures (lower average DBE). Higher DBE and AI most likely reflect recalcitrant lignin 

residues that have been solubilized. 



47 

 

Further investigation into nitrogen-containing compounds was conducted (Figure 11). The 

CHON1 class changed the least between the two sample sets. The CHON1S1 class reached a 

larger range of H/C values in the inlet. The CHON2 class reached higher degrees of nitrogenation 

in the inlet. These findings suggest that the CHON1S1 and CHON2 classes may be markers for 

increased microbial activity in the inlet of the bioreactor as they probably reflect amino acids or 

proteins that are degraded or utilized as soluble microbial products. The organic nitrogen is most 

likely due to more microbial biomass breakdown presumably caused to oxygen exposure and 

woodchip breakdown in the inlet, thus the appearance of compounds with higher N/C. The broad 

H/C range of CHON1S1 compounds in the inlet may speak to the bioavailability of organic 

matter, as compounds of high saturation, such as carbohydrates, tend to be more readily 

degraded by bacteria. One 2020 study looking at the transformation of DOM from a soil leachate 

found that oxygen-containing functional groups from aromatic and aliphatic groups, particularly 

carbohydrate-like species, were preferentially degraded by microbes, followed by oxygenated 

lignin- and tannin-like molecules (Hofmann et al., 2020). The tightening of H/C values from 

inlet to outlet possibly indicate that more saturated CHON1S1 compounds are either consumed or 

degraded in the inlet, while more more condensed material of lower H/C exist in the outlet. Table 

3 revealed that while differences in the number of compounds with lower N/C were minimal, 

compounds with N/C > 0.033 were found to dominate upstream. Compounds of lower N/C ratio 

are most likely to be lipid-like, carbohydrate-like and other highly aliphatic, carbon-dense 

compounds that contain minimal nitrogen-containing functional groups. As N/C increases to 

values greater than 0.1, it is more likely that these compounds represent amino acid-like and 

protein-like residues. 

 



48 

 

4.1.2 Assessment of Lability Confirms Increased Degradation Upstream 

The ability of the concept of the molecular lability boundary, initially proposed by D’Andrilli et 

al. (2015), to account for spatial influences in labile carbon was assessed. It was concluded that 

there was no significant difference between the average number of more labile formulas as 

dictated by the MLB, but when looking at the percentage of labile formulas relative to the total 

number of formulas in the inlet and outlet, it is confirmed that the inlet has a higher average 

percentage of labile formulas. Upstream, exposure to incoming oxygen accelerates microbial 

degradation of lignin into stocks of labile material which is then used up for denitrification.  

 

4.1.3 Biogeochemical Measurements Reflect FT-ICR MS Findings 

Preliminary batch reactor measurements (Table 2) supported FT-ICR MS findings. The 

differences in SUVA254 and AI between the inlet and outlet were not found to be large, which 

could explain how SUVA254, DBEw and AImod,w- all common markers of aromaticity- were found 

to not be highly correlated through PCA. [DOC] and protein content in the inlet were 

significantly higher than in the outlet, probably reflecting stimulation of fungal enzymes due to 

greater exposure to oxygen entering the bioreactor, leading to greater release of DOC. There was 

an inverse relationship between concentration of protein and C/N ratio in the wood solid phase, 

confirming that higher biofilm concentrations coincided with more wood degradation. This is 

reflected in calculations of the average NOSC at both locations, with the inlet averaging a 

slightly more positive value suggesting the existence of more oxidized material. These findings 

support observations of a decrease in dissolved oxygen along length of a bioreactor, consistent 

with microbial metabolism energetics (Wrightwood et al., 2022). Expectedly, the inlet samples 

were highly correlated to %MLBL, [DOC], and N/Cw through PCA, further supporting this 



49 

 

hypothesis. Nitrate removal rates (Figure 6) strongly reflect trends in [DOC], C/N ratio, and 

protein content. There is greater variability in nitrate concentration over time in the outlet 

samples. Over time, as pathways in the bioreactor diverge, small changes that may stem from 

variations in woodchip mass and surface area may intensify leading to less consistency amongst 

replicate measurements. Similar studies suggest that woodchips taken from upstream tend to 

show signs of decomposition and age in their physical appearance, most noticeably through 

darker coloring, which would influence overall nitrate reduction rates (Mardani et al., 2020).  

 

Inverse correlation of these variables with pH and %MLBL also confirms higher degrees of 

degradation in the upstream samples. The inlet averaged a slightly lower pH than the outlet, 

possibly due to increased concentration of organic acids or dissolved CO2 formed from 

degradation processes. Thus, lower pH would coincide with higher %MLBL. The inverse 

relationship between NOSCw and H/Cw is also to be expected. Higher H/Cw ratio indicates a 

higher degree of saturation. NOSCw decreases because more saturated, aliphatic structures are 

less susceptible to oxidation. Tighter clustering of inlet samples is observed in PCA (Figure 14) 

which suggests greater homogeneity in DOM composition and biogeochemical properties.  

 

4.2 Oxic-Anoxic Cycling May Improve Longevity of Denitrifying WBRs 

4.2.1 DOM From Anoxic Conditions Reflect Energetic Constraints on Lignin Degradation 

FT-ICR MS revealed that DOM composition becomes increasingly similar between the Anoxic 

Inlet and Cycle Inlet series with time. Comparison of mass spectra (Figure 15) revealed that most 

molecular changes occurred in the m/z region from 200 to 500 for the Anoxic Inlet samples. The 

Cycle Inlet samples saw consistently low abundance of ions over time. As previously mentioned, 

enzymatic activity may reduce chemical diversity resulting in the decreasing abundance of ions 



50 

 

over time (Towle et al., 2024; Echavarri-Bravo et al., 2019). Thus, it appears that our samples 

did not capture the bulk of the immediate effects of the oxic-anoxic transition, as our organic C 

pool seems already sufficiently broken down. Potentially, fungal enzymes initially targeted 

higher m/z compounds for lytic depolymerization into smaller species first. Previous studies 

suggest that species within the mass range 250-400 Da may correspond to dimeric lignin-based 

molecules that are not largely altered by enzymatic treatment (Echavarri-Bravo et al., 2019), and 

therefore, require more time to efficiently degrade. With time, internal conditions within the 

bioreactor will begin to stabilize and oxygen and necessary nutrients will become increasingly 

limited, therefore the processes responsible for degradation are largely unfavored. Thus, it is 

possible that the condensed aromatic-like species that are visually decreasing in the van Krevelen 

plots (Figure 16) are related to the these recalcitrant species. 

 

van Krevelen plots (Figure 16) also showed increases in lignin-like CHOS compounds in the 

Cycle Inlet, which has been previously observed (Mardani et al., 2020; Jéglot et al., 2021; 

Echavarri-Bravo et al., 2019) and is thought to be a function of interactions between DOM and 

sulfide ions made possible by dissimilatory sulfate-reducing anaerobic microbes. Sulfate may be 

sourced from agricultural fertilizers, enriched in nutrients like nitrogen that expedite the 

formation of sulfides. Typically, the WBR is not sulfidic due to the presence of NO3
-, but during 

stagnant periods, the reactor can become deeply reducing and sulfidic, creating an opportunity 

for sulfurization reactions to occur (Poulin et al., 2017).  Oxic-anoxic cycling may stimulate the 

production of intermediates such as reactive oxygen species via Fenton reactions (McGuire et al., 

2021) that would enhance rates of organic matter turnover and contribute to the formation and 

accumulation of CHOS compounds. As oxygen and nitrate are gradually depleted with 



51 

 

increasing time in anoxic incubation, microbes may preferentially turn to sulfate as an electron 

acceptor for mineralization of organic matter, leading to declining CHOS production as the 

reactor becomes sulfate-reducing during low-flow periods (Zhang et al., 2024). 

 

The decrease of CHON1, CHON2, CHON1S1, and CHON2S1 compounds (Table 5, Figure 16) in 

the Anoxic series supports that microbial activity slows in anoxic-only conditions, most likely as 

the recalcitrant species are being broken down. Figure 17-19 depicted trends consistent with this 

observation. Initially, the Anoxic Inlet accumulates more compounds of lower DBE and C# due 

to initial faster degradation. By the end of the experiment, the distributions across the two 

conditions converge. The Anoxic Inlet also consists of lower-numbered oxygen-containing 

compounds, potentially resulting from hemicellulose, cellulose and degraded lignin-related 

components (Podgorski et al., 2012). Although it seems contradictory that the anoxic-control 

group experienced more activity, the previous results of the mass spectrum (Figure 15) seemed 

to suggest that the bulk of the activity in oxic-anoxic cycling was not captured through our 

samples, and more interesting findings most likely occurred beforehand. 

 

These observations are reflected in abundance of its nitrogen-containing compounds with N/C 

ratios > 0.033 (Figure 20). Lower N/C ratios most likely represent carbon-dense compounds like 

lipids and carbohydrates, while ratios greater than or equal to 0.1 are potentially amino acids and 

protein-like molecules which reflect microbial activity. Over time, more saturated (higher H/C) 

compounds are utilized or degraded, particularly CHON1 and CHON1S1 species.  

 



52 

 

4.2.2 Oxic-Anoxic Cycling Leads to a Dynamic Equilibrium of Labile Carbon Release and 

Consumption 

 

Assessment of labile carbon through the definition provided by D’Andrilli et al (2015) revealed 

an interesting finding in the temporal effects of oxic-anoxic cycling on lignin degradation 

(Figure 21). After oxic-anoxic cycling, stocks of labile carbon are relatively constant. In anoxic-

only conditions, there is a fast depletion of labile formulae. Microbial activity and diversity 

depend largely on environmental conditions within the bioreactor. Oxic-anoxic cycling 

contributes to a dynamic equilibrium of labile carbon stemming from the constant cycle of 

degradation and consumption of lignin performed by both anaerobic and aerobic microbes. 

Trends in the Anoxic Inlet samples suggest that anaerobic microbes quickly use up labile carbon 

stocks without substantial replenishment due to the lack of oxidative depolymerization, 

ultimately contributing to less substantial nitrate removal in the WBR. More active nitrate 

removal in oxic-anoxic cycling has been previously reported and is thought to result from greater 

carbon availability for denitrification or dissimilatory nitrate reduction to ammonium (DNRA) in 

anoxic and hypoxic zones within the bioreactor (Zhang et al., 2024). Additionally, oxic-anoxic 

cycling has been shown to improve upon energetic constraints on microbial respiration due to 

oxygen limitations (Anderson et al., 2024). 

 

4.2.3 Correlations between Biogeochemical Measurements and FT-ICR MS Data 

Preliminary measurements of SUVA254 and [DOC] of the batch reactors supported FT-ICR MS 

results. SUVA254 decreases linearly in the Cycle Inlet series, while it decreases then increases in 

the Anoxic Inlet series. This suggests a steady breakdown of aromaticity of the DOM pool 

because of lignin degradation, enhanced by oxic-anoxic cycling, further confirmed by the 

accelerating rate of [DOC] release. Although it appears the immediate effects of oxic-anoxic 



53 

 

cycling have worn off, this finding validates that there is still active breakdown of organic 

matter. The slight increase in SUVA254 at the end of the time-series in the Anoxic Inlet series 

may imply accumulation of particulate organic material or rapid activity leading to clogging (von 

Ahnen et al., 2018), which could possibly explain the increasing aromaticity index (Figure 22A) 

over time. Oxic-anoxic cycling may improve WBR longevity by encouraging steady enzymatic 

activity through a dynamic equilibrium of production and release following the bulk of 

immediate oxic-anoxic impacts. In comparison, anoxic-only conditions are energetically 

hindered from quickly regenerating labile carbon for microbial use, therefore the bioreactor may 

be more prone to a buildup of recalcitrant organic material. The NOSC values (Figure 22B) pose 

a somewhat contradictory result, as they suggest that initially, most carbon found at the Anoxic 

Inlet is reduced (negative NOSC), while lignin degradation would ideally be associated with 

oxidized carbon (positive NOSC). However, this may reflect how thermodynamically more 

favorable substrates of higher NOSC are first being used, leaving behind compounds of lower 

NOSC that are more unfavorable for microbial energetics (Boye et al., 2017). 

 

The PCA (Figure 20) elucidated trends amongst measured variables. The Day 3 Cycle Inlet 

sample is strongly correlated with C#
w, consistent with more intact organic matter at the 

beginning of incubation. Day 3 Anoxic Inlet sample is highly correlated with N/Cw, which may 

suggest high initial microbial activity. NOSCw is expected to increase with time, leading to lower 

aromaticity due to greater oxidative ring-opening (McGuire et al., 2021). The inverse 

relationship between NOSCw and H/Cw is expected since more saturated, aliphatic structures 

(higher H/Cw) are less susceptible to oxidation (lower NOSCw). [DOC], negatively directed 

along PC2, is highly correlated with Cycle and Anoxic samples from the end of the anoxic 



54 

 

incubation, consistent with degradation of organic matter. Other variables that do not display 

clear trends with each other or with the given samples may be a result of the complexity of 

lignocellulosic DOM systems leading to statistical ambiguity.  



55 

 

CHAPTER 5: Conclusion 

This study investigated the capabilities of FT-ICR MS to confidently characterize complex 

lignocellulosic-based dissolved organic matter from woodchip bioreactors. Data comparing 

samples from the upstream and downstream of the bioreactor revealed spatial differences that are 

most likely caused by the exposure of the inlet to oxygen, which stimulates both aerobic and 

anaerobic microbial activity, as evidenced by trends in nitrogen-containing compounds, DBE, 

carbon number, oxygen number, labile formulas, and measured biogeochemical properties. This 

oxygen effect is exacerbated through drying and rewetting periods that subject the WBR system 

to altering periods of oxic and anoxic incubation. FT-ICR MS analysis discovered that oxic-

anoxic cycling, rather than outright generating more labile carbon compared to anoxic-only 

conditions, seems to generate a dynamic equilibrium of consumption and release that guaranteed 

relatively constant concentrations of labile material for microbial respiration. Meanwhile, DOM 

composition over time in anoxic-only conditions shows that energetic restraints on lignin-

degrading microbes lead to increased initial effort to depolymerize enzyme-resistant compounds, 

reflected in initially high percentages of labile carbon, supported by trends in carbon number, 

oxygen number, nitrogen-containing compounds, and DBE. Microbial activity and lignin 

degradation gradually fell off over the span of nine days from the start of the long anoxic 

incubation. These results help guide discourse on FT-ICR MS and its potential to identify 

evidence of fungal processing of lignin through characterizing detected molecular compounds, to 

ultimately inform innovative strategies that can optimize WBR performance and longevity for 

denitrification.  



56 

 

CHAPTER 6: Future Work 

Future study would include conducting a longer time-series experiment with increased frequency 

of sampling prior to the change of media. Samples from the Cycle series should be taken during 

both oxic and anoxic periods to determine immediate effects on composition and lability. Other 

conditions within the bioreactors that should be measured include temperature and pH, both of 

which can have a large effect on denitrification. Varying the length and frequency of incubation 

periods can help discover how to further optimize oxic-anoxic cycling. Additional studies can 

delve into assessing labile carbon generated from other organic sources such as sugar cane, 

wheat straw, and maize, and different wood types such as eucalyptus (Cameron & Schipper, 

2010; De Bhowmick et al., 2018; Feyereisen et al., 2020). FT-ICR MS analysis can also be 

supported with other techniques such as GC-MS which has been shown to successfully detect 

dominant aerobic microbial metabolites in reducing conditions (Anderson et al., 2024). 

There is also potential to use FT-ICR MS to analyze mineral effects on lignin degradation. 

Brown rot fungi can perform the Fenton reaction (Cragg et al., 2015) in which FeII reacts with 

hydrogen peroxide, producing hydroxyl radicals that oxidize and break down recalcitrant lignin 

at oxic-anoxic interfaces (Jones et al., 2020). White rot fungi produce enzymes such as oxidases 

and manganese peroxidases (MnP) that destabilize organic carbon through the oxidation of MnII 

to MnIII (Jones et al., 2020). MnIII is a reactive intermediate that can readily break down the 

phenolic subunits of lignin (Cragg et al., 2015). Several strains of bacteria and fungi such as 

Pseudomonas (Hansel, 2017) are known to oxidize MnII through nonenzymatic pathways as well 

(Jones et al., 2020). Thus, a point of future study would be to conduct experiments on Mn- and 

Fe-enriched WBRs to better understand the roles that these mineral processes have in 

transformations of organic matter. 



57 

 

REFERENCES 

Anderson, C. G., Tfaily, M. M., Chu, R. K., Tolić, N., Fox, P. M., Nico, P. S., Fendorf, S., Keiluweit, 

M. (2024). Seasonal Controls on Microbial Depolymerization and Oxidation of Organic Matter 

in Floodplain Soils. Environmental Science & Technology, 58(38), 16815-16823. 

https://doi.org/10.1021/acs.est.4c05109.  

Bahureksa, W., Tfaily, M. M., Boiteau, R. M., Young, R. B., Logan, M. N., McKenna, A. M., & 

Borch, T. (2021). Soil Organic Matter Characterization by Fourier Transform Ion Cyclotron 

Resonance Mass Spectrometry (FTICR MS): A Critical Review of Sample Preparation, 

Analysis, and Data Interpretation. Environmental Science & Technology, 55(14), 9637–9656. 

https://doi.org/10.1021/acs.est.1c01135.  

Burlacu, A., Israel-Roming, F., Cornea, C. P. (2018). Depolymerization of kraft lignin with laccase 

and peroxidase, A review. Scientific Bulletin Series F. Biotechnologies, 22, 172-179. 

Boye, K., Noël, V., Tfaily, M. M., Bone, S. E., Williams, K. H., Barger, J. R., Fendorf, S. (2017). 

Thermodynamically controlled preservation of organic carbon in floodplains. Nature 

Geoscience, 10, 415-419. https://doi.org/10.1038/ngeo2940.  

Cameron, S. G., & Schipper, L. A. (2010). Nitrate removal and hydraulic performance of organic 

carbon for use in denitrification beds. Ecological Engineering, 36(11), 1588–1595. 

https://doi.org/10.1016/j.ecoleng.2010.03.010.  

Christianson, L. E., Schipper L. A. (2016) Moving Denitrifying Bioreactors beyond Proof of Concept: 

Introduction to the Special Section. Journal of Environmental Quality, 45(3), 757-761. 

https://doi.org/10.2134/jeq2016.01.0013.  

Corilo, Y. E., Podgorski, D. C., McKenna, A. M., Lemkau, K. L., Reddy, C. M., Marshall, A. G., & 

Rodgers, R. P. (2013). Oil spill source identification by principal component analysis of 

electrospray ionization fourier transform ion cyclotron resonance mass spectra. Analytical 

Chemistry, 85(19), 9064–9069. https://doi.org/10.1021/ac401604u.  

Cragg, S. M., Beckham, G. T., Bruce, N. C., Bugg, T. D. H., Distel, D. L., Dupree, P., Etxabe, A. G., 

Goodell, B. S., Jellison, J., McGeehan, J. E., McQueen-Mason, S. J., Schnorr, K., Walton, P. H., 

Watts, J. E. M., & Zimmer, M. (2015). Lignocellulose degradation mechanisms across the Tree 

of Life. Current Opinion in Chemical Biology, 29, 108–119. 

https://doi.org/10.1016/j.cbpa.2015.10.018.  

Crawford, J. T., Stets, E. G., & Sprague, L. A. (2019). Network Controls on Mean and Variance of 

Nitrate Loads from the Mississippi River to the Gulf of Mexico. Journal of Environmental 

Quality, 48(6), 1789–1799. https://doi.org/10.2134/jeq2018.12.0435.  

Cumisarow, M. B., & Marshall, A. G. (1974). Fourier transform ion cyclotron resonance 

spectroscopy. Chemical Physical Letters, 25(2), 282-283. https://doi.org/10.1016/0009-

2614(74)89137-2.  

D’Andrilli, J., Cooper, W. T., Foreman, C. M., & Marshall, A. G. (2015). An ultrahigh-resolution 

mass spectrometry index to estimate natural organic matter lability. Rapid Communications in 

Mass Spectrometry, 29(24), 2385–2401. https://doi.org/10.1002/rcm.7400.  

D’Andrilli, J., Romero, C. M., Zito, P., Podgorski, D. C., Payn, R. A., Sebestyen, S. D., Zimmerman, 

A. R., & Rosario-Ortiz, F. L. (2023). Advancing chemical lability assessments of organic matter 

using a synthesis of FT-ICR MS data across diverse environments and experiments. Organic 

Geochemistry, 184, 104667. https://doi.org/10.1016/j.orggeochem.2023.104667.  

https://doi.org/10.1021/acs.est.4c05109
https://doi.org/10.1021/acs.est.1c01135
https://doi.org/10.1038/ngeo2940
https://doi.org/10.1016/j.ecoleng.2010.03.010
https://doi.org/10.2134/jeq2016.01.0013
https://doi.org/10.1021/ac401604u
https://doi.org/10.1016/j.cbpa.2015.10.018
https://doi.org/10.2134/jeq2018.12.0435
https://doi.org/10.1016/0009-2614(74)89137-2
https://doi.org/10.1016/0009-2614(74)89137-2
https://doi.org/10.1002/rcm.7400
https://doi.org/10.1016/j.orggeochem.2023.104667


58 

 

De Bhowmick, G., Sarmah, A. K., & Sen, R. (2018). Lignocellulosic biorefinery as a model for 

sustainable development of biofuels and value added products. Bioresource Technology, 247, 

1144–1154. https://doi.org/10.1016/j.biortech.2017.09.163.  

Dittmar, T., Koch, B., Hertkorn, N., & Kattner, G. (2008). A simple and efficient method for the 

solid-phase extraction of dissolved organic matter (SPE-DOM) from seawater. Limnology and 

Oceanography: Methods, 6(6), 230–235. https://doi.org/10.4319/lom.2008.6.230.  

Echavarri-Bravo, V., Tinzl, M., Kew, W., Cruickshank, F., Logan Mackay, C., Clarke, D. J., & 

Horsfall, L. E. (2019). High resolution fourier transform ion cyclotron resonance mass 

spectrometry (FT-ICR MS) for the characterisation of enzymatic processing of commercial 

lignin. New Biotechnology, 52, 1–8. https://doi.org/10.1016/j.nbt.2019.03.001.  

Feyereisen, G. W., Spokas, K. A., Strock, J. S., Mulla, D. J., Ranaivoson, A. Z., & Coulter, J. A. 

(2020). Nitrate removal and nitrous oxide production from upflow and downflow column 

woodchip bioreactors. Agricultural & Environmental Letters, 5(1), e20024. 

https://doi.org/10.1002/ael2.20024.  

Hansel, C. M. (2017). Manganese in Marine Microbiology. Advances in Microbial Physiology (Vol. 

70, pp. 37–83). Academic Press. https://doi.org/10.1016/bs.ampbs.2017.01.005.  

Hofmann, R., Uhl, J., Hertkorn, N., Griebler, C. (2020). Linkage Between Dissolved Organic Matter 

Transformation, Bacterial Carbon Production, and Diversity in a Shallow Oligotrophic Aquifer: 

Results from Flow-Through Sediment Microcosm Experiments. Frontiers in Microbiology, 11. 

https://doi.org/10.3389/fmicb.2020.543567.  

Hur, M., Yeo, I., Park, E., Kim, Y. H., Yoo, J., Kim, E., No, M. H., Koh, J., & Kim, S. (2010). 

Combination of statistical methods and Fourier transform ion cyclotron resonance mass 

spectrometry for more comprehensive, molecular-level interpretations of petroleum samples. 

Analytical Chemistry, 82(1), 211–218. https://doi.org/10.1021/ac901748c.  

Jéglot, A., Audet, J., Sørensen, S. R., Schnorr, K., Plauborg, F., & Elsgaard, L. (2021). Microbiome 

Structure and Function in Woodchip Bioreactors for Nitrate Removal in Agricultural Drainage 

Water. Frontiers in Microbiology, 12. https://doi.org/10.3389/fmicb.2021.678448.  

Jones, M. E., Lacroix, R. E., Zeigler, J., Ying, S. C., Nico, P. S., & Keiluweit, M. (2020). Enzymes, 

Manganese, or Iron? Drivers of Oxidative Organic Matter Decomposition in Soils. 

Environmental Science & Technology, 54(21), 14114–14123. 

https://doi.org/10.1021/acs.est.0c04212.  

Koch, B. P., & Dittmar, T. (2006). From mass to structure: An aromaticity index for high-resolution 

mass data of natural organic matter. Rapid Communications in Mass Spectrometry, 20(5), 926–

932. https://doi.org/10.1002/rcm.2386.  

Korner, H., & Zumft, W. G. (1989). Expression of Denitrification Enzymes in Response to the 

Dissolved Oxygen Level and Respiratory Substrate in Continuous Culture of Pseudomonas 

stutzeri. Applied and Environmental Microbiology, 55(7), 1670-

1676.  https://doi.org/10.1128/aem.55.7.1670-1676.1989.  

Lankiewicz, T. S., Choudhary, H., Gao, Y., Amer, B., Lillington, S. P., Leggieri, P. A., Brown, J. L., 

Swift, C. L., Lipzen, A., Na, H., Amirebrahimi, M., Theodorou, M. K., Baidoo, E. E. K., Barry, 

K., Grigoriev, I. V., Timokhin, V. I., Gladden, J., Singh, S., Mortimer, J. C., … O’Malley, M. A. 

(2023). Lignin deconstruction by anaerobic fungi. Nature Microbiology, 8, 596–610. 

https://doi.org/10.1038/s41564-023-01336-8.  

Lopez-Ponnada, E. V., Lynn, T. J., Peterson, M., Ergas, S. J., & Mihelcic, J. R. (2017). Application of 

denitrifying wood chip bioreactors for management of residential non-point sources of nitrogen. 

Journal of Biological Engineering, 11, 16. https://doi.org/10.1186/s13036-017-0057-4.  

https://doi.org/10.1016/j.biortech.2017.09.163
https://doi.org/10.4319/lom.2008.6.230
https://doi.org/10.1016/j.nbt.2019.03.001
https://doi.org/10.1002/ael2.20024
https://doi.org/10.1016/bs.ampbs.2017.01.005
https://doi.org/10.3389/fmicb.2020.543567
https://doi.org/10.1021/ac901748c
https://doi.org/10.3389/fmicb.2021.678448
https://doi.org/10.1021/acs.est.0c04212
https://doi.org/10.1002/rcm.2386
https://doi.org/10.1128/aem.55.7.1670-1676.1989
https://doi.org/10.1038/s41564-023-01336-8
https://doi.org/10.1186/s13036-017-0057-4


59 

 

Malherbe, S., & Cloete, T. E. (2002). Lignocellulose biodegradation: Fundamentals and applications. 

Re/Views in Environmental Science and Bio/Technology, 1, 105-114. 

https://doi.org/10.1023/A:1020858910646.  

Malone, T. C., & Newton, A. (2020). The Globalization of Cultural Eutrophication in the Coastal 

Ocean: Causes and Consequences. Frontiers in Marine Science, 7, 670. 

https://doi.org/10.3389/fmars.2020.00670.  

Mardani, S., McDaniel, R., Bleakley, B. H., Hamilton, T. L., Salam, S., & Amegbletor, L. (2020). The 

effect of woodchip bioreactors on microbial concentration in subsurface drainage water and the 

associated risk of antibiotic resistance dissemination. Ecological Engineering, 143, 100017. 

https://doi.org/10.1016/j.ecoena.2020.100017.  

Maxwell, B. M., Diáz-Garciá, C., Martínez-Sánchez, J. J., Birgand, F., & Álvarez-Rogel, J. (2020). 

Temperature sensitivity of nitrate removal in woodchip bioreactors increases with woodchip age 

and following drying-rewetting cycles. Environmental Science: Water Research & Technology, 

6(10), 2752–2765. https://doi.org/10.1039/d0ew00507j.  

McGuire, P. M., Butkevich, N., Saksena, A. V., Todd Walter, M., Shapleigh, J. P., Reid, M. C. 

(2023). Oxic-anoxic cycling promotes coupling between complex carbon metabolism and 

denitrification in woodchip bioreactors. Environmental Microbiology, 25(9), 1696-1712. 

https://doi.org/10.1111/1462-2920.16387.  

McGuire, P. M., Dai, V., Walter, M. T., & Reid, M. C. (2021). Labile carbon release from oxic-

anoxic cycling in woodchip bioreactors enhances nitrate removal without increasing nitrous 

oxide accumulation. Environmental Science: Water Research and Technology, 7(12), 2357–

2371. https://doi.org/10.1039/d1ew00446h.  

Minor, E. C., Swenson, M. M., Mattson, B. M., & Oyler, A. R. (2014). Structural characterization of 

dissolved organic matter: A review of current techniques for isolation and analysis. 

Environmental Sciences: Processes and Impacts, 16(9), 2064-2079. 

https://doi.org/10.1039/c4em00062e.  

O’Geen, A. T., Budd, R., Gan, J., Maynard, J. J., Parikh, S. J., & Dahlgren, R. A. (2010). Mitigating 

nonpoint source pollution in agriculture with constructed and restored wetlands. Advances in 

Agronomy (Vol. 108, Issue C). Academic Press. 

https://doi.org/10.1016/S0065-2113(10)08001-6.  

Podgorski, D. C., Hamdan, R., McKenna, A. M., Nyadong, L., Rodgers, R. P., Marshall, A. G., & 

Cooper, W. T. (2012). Characterization of pyrogenic black carbon by desorption atmospheric 

pressure photoionization fourier transform ion cyclotron resonance mass spectrometry. 

Analytical Chemistry, 84(3), 1281–1287. https://doi.org/10.1021/ac202166x.  

Poulin, B. A., Ryan, J. N., Nagy, K. L., Stubbins, A., Dittmar, T., Orem, W., Krabbenhoft, D. P., 

Aiken, G. R. (2017). Spatial Dependence of Reduced Sulfur in Everglades Dissolved Organic 

Matter Controlled by Sulfate Enrichment. Environmental Science & Technology, 51(7), 3630-

3639. https://doi.org/10.1021/acs.est.6b04142.  

Puckett, L. J. (1994). Nonpoint and point sources of nitrogen in major watersheds of the United 

States. U.S. Dept. of the Interior, U.S. Geological Survey. 

https://pubs.usgs.gov/wri/wri944001/pdf/wri94-4001.pdf.  

Qi, Y., Hempelmann, R., & Volmer, D. A. (2016). Shedding light on the structures of lignin 

compounds: photo-oxidation under artificial UV light and characterization by high resolution 

mass spectrometry. Analytical and Bioanalytical Chemistry, 408, 8203–8210. 

https://doi.org/10.1007/s00216-016-9928-7.  

https://doi.org/10.1023/A:1020858910646
https://doi.org/10.3389/fmars.2020.00670
https://doi.org/10.1016/j.ecoena.2020.100017
https://doi.org/10.1039/d0ew00507j
https://doi.org/10.1111/1462-2920.16387
https://doi.org/10.1039/d1ew00446h
https://doi.org/10.1039/c4em00062e
https://doi.org/10.1016/S0065-2113(10)08001-6
https://doi.org/10.1021/ac202166x
https://doi.org/10.1021/acs.est.6b04142
https://pubs.usgs.gov/wri/wri944001/pdf/wri94-4001.pdf
https://doi.org/10.1007/s00216-016-9928-7


60 

 

Sang, Y., Petrovic, I., DuPlooy, L. E., Vega, M., Zhang, Z., Yang, L., Ge, M., Reid, M. C. (2024). 

Biogeochemical Controls on Mobilization of Labile Carbon from Wood to Denitrifying 

Bioreactors: Insights from Multi-Modal Synchrotron Analysis of Wood Degradation. 

[Manuscript in preparation]. School of Civil and Environmental Engineering, Cornell University. 

Schipper, L. A., Robertson, W. D., Gold, A. J., Jaynes, D. B., Cameron, S. C. (2010). Denitrifying 

bioreactors- An approach for reducing nitrate loads to receiving waters. Ecological Engineering, 

36(11), 1532-1543. https://doi.org/10.1016/j.ecoleng.2010.04.008.  

Sleighter, R. L., Liu, Z., Xue, J., & Hatcher, P. G. (2010). Multivariate statistical approaches for the 

characterization of dissolved organic matter analyzed by ultrahigh resolution mass spectrometry. 

Environmental Science & Technology, 44(19), 7576–7582. https://doi.org/10.1021/es1002204.  

Smith, D. F., Podgorski, D. C., Rodgers, R. P., Blakney, G. T., Hendrickson, C. L. (2018). 21 Tesla 

FT-ICR Mass Spectrometer for Ultrahigh-Resolution Analysis of Complex Organic Mixtures.  

Analytical Chemistry, 90(3), 2041-2047. https://doi.org/10.1021/acs.analchem.7b04159. 

Towle, Z., Cruickshank, F., Mackay, C. L., Clarke, D. J., & Horsfall, L. E. (2024). Utilising Fourier 

transform ion cyclotron resonance mass spectrometry (FT-ICR MS) to track the oxidation of 

lignin by an alkaliphilic laccase. Analyst, 149(8), 2399–2411. 

https://doi.org/10.1039/d4an00124a.  

U.S. Environmental Protection Agency. (2021). National Coastal Condition Assessment: A 

Collaborative Survey of the Nation’s Estuaries and Great Lakes Nearshore Waters. U.S. 

Environmental Protection Agency. https://www.epa.gov/system/files/documents/2021-

09/nccareport_final_2021-09-01.pdf.  

von Ahnen, M., Pedersen, P. B., & Dalsgaard, J. (2018). Performance of full-scale woodchip 

bioreactors treating effluents from commercial RAS. Aquacultural Engineering, 83, 130–137. 

https://doi.org/10.1016/j.aquaeng.2018.10.004.  

Wang, D., Li, M., Chen, J., Chen, H., & Shi, Q. (2024). A new ternary diagram to decipher the 

evolution of maturity and biodegradation of crude oil using ESI FT-ICR MS. Fuel, 359, 130499. 

https://doi.org/10.1016/j.fuel.2023.130499.  

Warneke, S., Schipper, L. A., Matiasek, M. G., Scow, K. M., Cameron, S., Bruesewitz, D. A., & 

McDonald, I. R. (2011). Nitrate removal, communities of denitrifiers and adverse effects in 

different carbon substrates for use in denitrification beds. Water Research, 45(17), 5463–5475. 

https://doi.org/10.1016/j.watres.2011.08.007.  

Wrightwood, O. M., Hattaway, M. E., Young, T. M., & Bischel, H. N. (2022). Assessment of 

Woodchip Bioreactor Characteristics and Their Influences on Joint Nitrate and Pesticide 

Removal. ACS ES&T Water, 2(1), 106–116. https://doi.org/10.1021/acsestwater.1c00277.  

Xue, J., Lee, C., Wakeham, S. G., & Armstrong, R. A. (2011). Using principal components analysis 

(PCA) with cluster analysis to study the organic geochemistry of sinking particles in the ocean. 

Organic Geochemistry, 42(4), 356–367. https://doi.org/10.1016/j.orggeochem.2011.01.012.  

Zhao, L., Zhang, J., Zhao, D., Jia, L., Qin, B., Cao, X., Zang, L., Lu, F., & Liu, F. (2022). Biological 

degradation of lignin: A critical review on progress and perspectives. In Industrial Crops and 

Products, 188, 115715. https://doi.org/10.1016/j.indcrop.2022.115715.  

Zhang, Z., McGuire, P. M., Richardson, R. E., Gu, A. Z., Shalpeigh, J. P., Reid, M. C. (2024). 

Orthophosphate uptake onto woodchips in denitrying bioreactors is enhanced by anoxic-

(anaerobic-)oxic cycling. Science of the Total Environment, 951, 175515. 

https://doi.org/10.1016/j.scitotenv.2024.175515.  

https://doi.org/10.1016/j.ecoleng.2010.04.008
https://doi.org/10.1021/es1002204
https://doi.org/10.1021/acs.analchem.7b04159
https://doi.org/10.1039/d4an00124a
https://www.epa.gov/system/files/documents/2021-09/nccareport_final_2021-09-01.pdf
https://www.epa.gov/system/files/documents/2021-09/nccareport_final_2021-09-01.pdf
https://doi.org/10.1016/j.aquaeng.2018.10.004
https://doi.org/10.1016/j.fuel.2023.130499
https://doi.org/10.1016/j.watres.2011.08.007
https://doi.org/10.1021/acsestwater.1c00277
https://doi.org/10.1016/j.orggeochem.2011.01.012
https://doi.org/10.1016/j.indcrop.2022.115715
https://doi.org/10.1016/j.scitotenv.2024.175515


61 

 

Zhuo, X., He, C., Cai, R., & Shi, Q. (2024). Effect of salinity on molecular characterization of 

dissolved organic matter using ESI FT-ICR MS. Talanta, 266, 125005. 

https://doi.org/10.1016/j.talanta.2023.125005.  

 

https://doi.org/10.1016/j.talanta.2023.125005