nitrate as an electron acceptor in microbial decomposition ...cj82st421/fulltext.pdf1 nitrate as an...

1

Nitrate as an electron acceptor in microbial decomposition of salt marsh sediment organic matter

and implications for carbon storage

by Ashley Bulseco-McKim

B.S. in Marine Science, University of Hawaii at Hilo

M.S. in Marine Science and Technology, University of Massachusetts Boston

A dissertation submitted to

The Faculty of

the College of Science of

Northeastern University

in partial fulfillment of the requirements

for the degree of Doctor of Philosophy

July 16, 2018

Dissertation directed by

Jennifer L. Bowen

Associate Professor of Marine and Environmental Sciences

2

Dedication

To my parents,

for showing me how to be relentless.

To my brothers,

for teaching me how to follow my heart.

And to my husband,

for getting me to laugh more than I ever have before.

3

Acknowledgements

I could not have had more supportive, encouraging advisors throughout my academic

career thus far. First and foremost, I thank Jennifer Bowen for teaching me how to be resilient,

both as a scientist and as a human. Her ability to think critically and quickly are skills I will

always strive to match. More importantly, Jen has helped me persevere through times of extreme

anxiety and self-doubt, and for that I will be forever grateful. I also thank my co-advisor, Anne

Giblin, whose kindness and willingness to help are unmatched by any other individual I have

ever met. She is an exceptional role model for women in science everywhere, and helped to

shape the clarity of my work. Thank you to the rest of my committee at Northeastern University,

Randall Hughes, Aron Stubbins, and Amy Mueller, who contributed insightful thought to my

dissertation. I would also like to thank my UMass committee, including Bob Chen and Crystal

Schaaf, who both played a critical role in the initial development of my proposal and critical

thinking. Lastly, a sincere mahalo to my undergraduate advisor, Tracy Wiegner, for being my

biggest cheerleader throughout the years.

This journey would not have been possible without the constant support from the Bowen

Lab family, including Annie Murphy, Chris Lynum, Andrea Unzueta Martinez, Joe Vineis,

Kerry McNally, Kenly Hiller-Bittrolff, Sarah Feinman, John Angell, Patrick Kearns, and Brian

Donnelly. In particular, I’d like to thank Annie Murphy for always being my voice of reason

both in and out of the laboratory, and Chris Lynum for knowing exactly what to say without the

need for words. Throughout my graduate school career, I’ve had the pleasure to meet and work

with brilliant students, including Michael Greenwood, Khang Tran, Sean Osborne, Ross

Ackerman, Itxaso Garay, Emma Riccardi, and Matthew Smith. I also thank former members of

the Hannigan Lab, who were always there for me during my early days of graduate school,

4

including Alan Stebbins, Katie Flanders, Eric Wilcox-Freeburg, Jeremy Williams, Bryanna

Broadaway, Alex Eisen-Cuadra, Steve Nye, Nicole Henderson, and Aaron Honig.

I would like to express my gratitude to the immense amount of people who have helped

me during this journey. I thank several individuals from the Marine Biological Laboratory,

including Jane Tucker, Sam Kelsey, Inke Forbrich, Joe Vallino, Tyler Messerschmidt, Suzanne

Thomas, Hap Garritt, Rich McHorney, Marshall Otter, Zoe Cardon, and others who, without

question, never hesitated to train me on various analyses or tolerate my taking up space in the

laboratory. In particular, I would like to thank Jane for her immeasurable patience and support as

I anxiously navigated my doctoral studies. After leaving science in question of my career path, it

was involvement with the TIDE project and constant encouragement from Linda Deegan, David

Johnson, and Jimmy Nelson, that rekindled my passion for research. I am still here because of

these individuals, along with other TIDE members that have helped along the way, including

David Behringer, Bethany Williams, Hillary Sullivan, Caitlin Bauer, and Serina Wittyngham.

I am lucky to have worked with and crossed paths with such influential people. My

involvement with the Coastal & Estuarine Research Federation as the student member-at-large

has provided me with invaluable experience that I believe has shaped me into the scientist I am

today. Throughout my internships and undergraduate study, mentors such as Alan Shanks, Erin

“Ezzy” Cooper, Stephanie Schroeder, Jan Hodder, Chris Langdon, Matthew Gray, Itchung

Cheung, and Jon Sun, all influenced my form of thought. Relationships formed during my time

in Hawaii with individuals, such as Barb Bruno from C-MORE, Sherwood Maynard from the

MOP program, and classmates from UH Hilo, continue to shape me today. I would also like to

thank several brilliant women in science, who inspire me daily, perhaps without even knowing:

Robyn Hannigan, Ellen Douglas, Torrie Hanley, Jessica Carilli, Cascade Sorte, Amanda Glazier,

5

Jamie Dombach, Casey Lyons, Kristin Osborne, Bonnie Blalock, and so many others. Lastly,

thanks to my closest of friends, Nick and Barbarajean Fountoulakis, who have been by my side

through thick and thin.

Words cannot describe how thankful I am to have such a supportive family. My parents,

Dylan and Georgeen, faced such adversity to provide me with every opportunity possible and I

would not be here today without their hard work and relentless encouragement. My brothers,

Brandon and Connor, continue to inspire me daily as they pursue their individual passions. My

step son, Caleb, is one of the strongest kids I know, preserving through constant challenges. I

thank the Marvells, who instilled in me curiosity and healthy skepticism early on in life. And

lastly, I would like to thank my husband, Shaun, whose support and kindness have been

unwavering since we first met. His love keeps me grounded in this crazy world, and I am certain

this journey would have been much less enjoyable without him.

6

Abstract of Dissertation

Atmospheric carbon dioxide (CO2) concentrations continue to rise as a result of fossil

fuel burning and land-use changes, thereby contributing to increases in global temperature, ocean

acidification, and sea level rise. Sequestering some of this excess CO2 in blue carbon habitats,

such as salt marshes, mangroves, and seagrasses, has been proposed as a mitigation strategy due

to their ability to efficiently store carbon. Salt marshes, in particular, store carbon at rates that are

orders of magnitude greater than terrestrial forests due to large inputs of organic matter (OM)

from primary production concurrent with slow decomposition rates; the balance between the two

ultimately determines the burial of OM and carbon storage over time. As nitrogen loading to

coastal waters continues to rise, primarily in the form of nitrate (NO3-), it is unclear what effect it

will have on carbon storage capacity of these systems. This uncertainty is largely driven by the

dual role NO3- can play in biological processes, where it can either serve as a nutrient for primary

production or a powerful electron acceptor fueling heterotrophic microbial metabolism.

Distinguishing between the two is critical, since the former could promote carbon storage by

enhancing fixation, while the latter could potentially deplete this service by stimulating microbial

decomposition.

Using a combination of controlled flow through experiments and field surveys, my

dissertation sought to: 1) determine the importance of NO3- as an electron acceptor in OM

decomposition across different sediment depths, 2) assess whether chronic NO3- enrichment

affected OM burial, and, since microbes are largely responsible for controlling long term carbon

storage, 3) examine microbial community diversity, structure, activity, and assembly of deep salt

marsh sediments spanning over 3000 years of accretion between two sites: an experimentally

enriched marsh and its paired reference marsh. To carry out these objectives, I applied a

7

comprehensive set of tools, including 1) biogeochemical measurements of dissolved inorganic

carbon and nutrient concentrations, 2) OM quality measurements, such as % carbon, % nitrogen,

lipid biomarkers concentration, and Fourier Transform-Infrared Spectroscopy, as well as 3)

sequencing of the 16S rRNA gene, its product 16S rRNA, and shotgun metagenomics.

In controlled flow through experiments where I exposed sediment of varying depths and

OM lability to 500 µM NO3-, I observed a 40-45% increase in OM decomposition in response to

NO3- when compared to a seawater control. This pattern persisted at sediment depths typically

considered to be less labile. NO3- altered both the microbial community and its associated

functional potential, selecting for taxa belonging to groups known to reduce NO3- and oxidize

more complex forms of OM, and increasing the abundance of nitrogen cycling genes.

Stimulation in OM decomposition in response to NO3- was not as pronounced in sediments from

sites that had been chronically exposed to NO3-, with the lowest effect size occurring at a site

exposed to sewage effluent for 40 years, suggesting the effect of NO3- on OM decomposition is

limited. These results demonstrate that NO3- can serve as an electron acceptor in microbial

metabolism and may expand the OM pool available to microbial oxidation, effectively reducing

overall carbon storage potential in salt marsh systems, however, OM that is buried under high

NO3- conditions may be more stable over time.

In a field survey examining microbial community diversity, structure, activity, and

assembly of deep salt marsh sediments spanning over 3000 years of accretion between an

experimentally enriched marsh and its paired reference marsh, I found that both microbial

diversity and gene abundance decreased with depth, potentially due to resource limitation, and

observed high rates of inactivity in deeper sediments. Depth and associated changes in OM

explained changes in microbial community structure in shallow (0-50 cm) sediments, but this

8

pattern became much less apparent in deeper sediments beyond the rooting zone (60+ cm), likely

due to more stochastic assembly at depth. The only difference between the reference and

enriched marshes occurred in deeper sediments, suggesting that the effect of nutrient enrichment

is not detectable over longer time scales of carbon storage; instead, these differences may be

attributed to stochastic processes resulting from energy limitation in deep subsurface marsh

sediments. Overall, my dissertation highlights the role of NO3- as an electron acceptor in OM

decomposition, and underscores the need to better understand the microbes mediating carbon

storage and how they will respond to nutrient enrichment.

9

Table of Contents

Dedication 2

Acknowledgements 3

Abstract of Dissertation 6

Table of Contents 9

List of Figures 11

List of Tables 13

Introduction: 14

Chapter 1: Nitrate addition stimulates microbial decomposition of organic matter 30

in salt marsh sediments

Abstract 30

Introduction 31

Materials and Methods 34

Results 42

Discussion 46

Tables 66

Figures 68

Supplemental Material 77

Chapter 2: Chronic exposure to nutrient enrichment lessens the effect of additional 86

nitrate on organic matter decomposition despite changes to

microbial community structure and activity

Abstract 86

Introduction 86


Results 101

10

Discussion 106

Tables 125

Figures 128


Chapter 3: Stochastic processes shape microbial communities in deep salt marsh 145

sediments

Abstract 145

Introduction 146


Results 158

Discussion 164

Tables 183

Figures 185


Appendix: Nitrate reduction pathways and functional potential in response to 220

nutrient enrichment

11

List of Figures

Chapter 1

Figure 1. Dissolved inorganic carbon production over time across three depths 68

Figure 2. Cumulative dissolved inorganic carbon in response to treatment and depth 69

Figure 3. Nitrate reduction rates and cumulative nitrate reduction across depths 70

Figure 4. Sulfide production rates and cumulative production, sulfur storage, and 71

cumulative sulfate reduction across three depths

Figure 5. Ammonium production rates and ammonium production across three depths 72

Figure 6. Ratio of dissolved inorganic carbon to ammonium production in response to 73

nitrate across three depths

Figure 7. Fourier Transform-Infrared Spectra examining organic matter functional 74

groups in response to nitrate across three depths

Figure 8. Microbial community structure and diversity in response to nitrate by depth 75

Figure 9. Heatmap of order-level taxa relative abundance in response to nitrate 76

Figure S1. Schematic of flow through reactor 82

Figure S2. Bromide breakthrough curve 83

Figure S3. Quantitative PCR of 16S rRNA gene by depth and treatment 84

Figure S4. Relative abundance of 20 bacterial orders present across all samples 85

Chapter 2

Figure 1. Map of sites from a gradient of prior nutrient enrichment 128

Figure 2. Boxplot of Index II values and mid-IR spectra of each site along nutrient 129

enrichment gradient

Figure 3. Microbial community structure and order-level relative abundance of 130

taxa most different among sites from nutrient enrichment gradient

Figure 4. Dissolved inorganic carbon production rate over time in response to nitrate 131

Figure 5. Cumulative dissolved inorganic carbon production in response to nitrate 132

Figure 6. Nitrate and sulfate reduction across sites 133

Figure 7. Mid-IR spectra and total dissolved inorganic carbon as a function of Index II 134

Figure 8. Weighted UniFrac similarity across sites in response to nitrate 135

Figure 9. Order-level relative abundance of taxa most different between treatments 136

Figure 10. Order-level activity assessed by 16S rRNA/16S rRNA gene in response to 137

per site

Figure S1. Bromide breakthrough curve 143

Figure S2. Ammonium production over time from sites across a nutrient enrichment 144

gradient

Chapter 3

Figure 1. Map of core locations within reference and enriched sites 185

Figure 2. Organic matter characteristics and age of reference and enriched sites 186

Figure 3. Shannon diversity and 16S rRNA gene/16S rRNA by depth 187

Figure 4. Weighted UniFrac of total microbial community by depth 188

Figure 5. Relative abundance of top 100 ASVs in shallow sediments at class-level 189

Figure 6. Order-level relative abundance of total microbial community by depth 190

Figure 7. Order-level relative abundance of active microbial community by depth 191

12

Figure 8. Relative abundance of top 100 ASVs in deep sediments at class-level 192

Figure 9. Standardized effect size of mean pairwise distances by depth and site 193

Figure S1. Weighted UniFrac of active microbial community by depth 219

Appendix

Figure 1. Denitrification and dissimilatory nitrate reduction rates over time per depth 238

Figure 2. Cumulative denitrification and dissimilatory nitrate reduction per depth 239

Figure 3. Relative contribution of denitrification and dissimilatory nitrate reduction to 240

total nitrate consumption rates per depth

Figure 4. Non-metric multidimensional scaling plot of subsystems-level functional 241

annotation by treatment and depth

Figure 5. Heatmap of N-cycling gene abundance in response to nitrate 242

Figure 6. Boxplot of total fatty acids and sub-classes per treatment and depth 243

13

List of Tables

Chapter 1

Table 1. Functional group assignments for Fourier Transform-Infrared Spectroscopy 66

Table 2. Organic matter characteristics before and after flow through experiment 67

Table S1. Flow property characteristics of flow through experiment 79

Table S2. Top 30 bacterial orders important in discriminating between treatments 80

Chapter 2

Table 1. Environmental characteristics of three sites along a nutrient enrichment gradient 125

Table 2. Denitrification and dissimilatory nitrate reduction rates per site 126

Table 3. Organic matter characteristics and gene abundance in response to nitrate 127

Table S1. Functional group assignments for Fourier Transform-Infrared Spectroscopy 138

Table S2. Flow property characteristics of flow through experiment 139

Table S3. Relative abundance of order-level taxa most important in discriminating 140

between treatments

Table S4. Taxonomic information for groups that exhibited change in activity 141

Chapter 3

Table 1. Organic matter characteristics from reference and enriched marshes 183

Table 2. Radiocarbon dating for cores from reference and enriched marshes 184

Table S1. Linear mixed effects model results examining the effect of site and depth 194

on organic matter characteristics

Table S2. Linear mixed effects model results examining effect of site, depth, and 195

organic matter characteristics on Shannon diversity


organic matter characteristics on total microbial community structure


organic matter characteristics on active microbial community structure

Table S5. Top 100 16S rRNA gene ASVs from shallow sediments for both reference 198

and enriched sites

Table S6. Top 100 16S rRNA ASVs from shallow sediments for both reference 201

and enriched sites

Table S7. Taxonomic information for ASVs from 16S rRNA gene most important in 204

explaining patterns by depth in shallow sediments

Table S8. Taxonomic information for ASVs from 16S rRNA most important in 205

explaining patterns by depth in shallow sediments

Table S9. Top 100 16S rRNA gene ASVs from deep reference sediments 207

Table S10. Top 100 16S rRNA gene ASVs from deep enriched sediments 210

Table S11. Top 100 16S rRNA ASVs from deep reference sediments 213

Table S12. Top 100 16S rRNA ASVs from deep enriched sediments 216

Appendix

Table 1. Fatty acid sub-class source and lipid number 236

Table 2. Number of basepairs and quality-filtered sequences from MG-RAST 237

14

Introduction

General Introduction

Salt marshes are coastal and estuarine wetlands situated at the intersection between the

land and sea that thrive in protected areas (Redfield 1972) with low wave energy (Allen 2000).

These systems are vital components of coastal systems and provide a number of ecosystem

services, such as shoreline protection (Costanza et al. 2008), nutrient filtration (Valiela & Teal

1974), and support of marine and coastal food webs (Teal 1962). One important ecosystem

service is their ability to store carbon at an average rate of 218 ± 24 g C m-2 yr-1, which is orders

of magnitude greater than terrestrial forests (Mcleod et al. 2011). The ability of salt marshes to

sequester carbon is due to a combination of high rates of primary production and slow

decomposition of anoxic sediments (Reddy & Patrick Jr. 1975). The balance between the two

ultimately determines the rate of organic matter burial. Consequently, salt marshes, and other

“blue carbon” systems (seagrasses and mangroves; Nelleman et al. 2009) have become a major

focus of coastal restoration projects (Warren et al. 2002, Macreadie et al. 2017) as a strategy to

mitigate increasing concentrations of atmospheric carbon dioxide that could otherwise worsen

climate change.

Despite providing an efficient means to store carbon, salt marshes face a number of

anthropogenic-driven threats that can diminish, and potentially reverse, this ecosystem service.

Nitrogen (N) inputs, for instance, have been increasing at an alarming rate (Galloway et al. 2008,

2017), leading to the alteration of coastal biogeochemical cycles via increased primary

production and associated symptoms of eutrophication (Nixon 1995, Rabalais et al. 2009). Salt

marshes have been traditionally viewed as more resilient to the adverse effects of excess N

loading due to their high capacity for removing nutrients (Valiela & Cole 2002), either by

15

assimilation into plant biomass (Valiela & Teal 1974) or conversion to gaseous products (NO,

N2, or N2O) via denitrification (Kaplan et al. 1979, Seitzinger 1988, Hopkinson & Giblin 2008)

or anaerobic ammonium oxidation (anammox; Dalsgaard et al. 2005, Koop-Jakobsen & Giblin

2009a); however, we do not fully understand how much N loading salt marshes can withstand,

and whether this can affect the carbon storage capacity of these highly productive systems.

Nutrient enrichment can increase aboveground biomass (Valiela et al. 1975, Kaplan et al.

1979, Morris 1991, Langley et al. 2013), which may facilitate sediment accretion (Morris et al.

2002), augment carbon sink potential, and consequently help salt marshes keep pace with sea

level rise (Kirwan & Megonigal 2013). In other systems, however, nutrient addition can decrease

belowground biomass, potentially leading to loss of sediment stability (Darby & Turner 2008).

There are two mechanisms by which this could occur that are not necessarily mutually exclusive.

The first mechanism is that wetland plants decrease allocation of root material needed to forage

for nutrients as ambient nutrient supply increases, thus allowing them to divert energy away from

belowground production towards aboveground production and photosynthesis (Levin et al.

1989). Consequently, less root biomass is present to stabilize sediments. The second mechanism

is that increased N concentrations leads to increased microbial respiration (Wigand et al. 2009)

and decomposition of belowground organic matter (Deegan et al. 2012). If the addition of N

stimulates subsurface microbial respiration, which oxidizes organic matter to fuel heterotrophic

processes, then the carbon storage capacity of salt marshes could drastically diminish,

highlighting the need to understand these two mechanisms in response to nutrient enrichment.

One factor that may influence which mechanism underlies responses to experimental

marsh fertilization may be the form of N that is applied. Many studies apply N in its reduced

form, as ammonium (NH4+) or urea, or a mix of oxidized and reduced forms (NH4NO3);

16

although the primary form of N delivery to the coasts is as nitrate (NO3-), not NH4

+ (Cloern

2001, Galloway et al. 2008). NO3- , however, plays a dual role in nature, either serving as a

nutrient in support of primary production or by acting as an energetically favorable electron

acceptor in microbial oxidation of organic matter through various anaerobic respiration

processes, including denitrification (Kaplan et al. 1979, Seitzinger 1988, Hamersley & Howes

2005) and dissimilatory nitrate reduction to ammonium (An & Gardner 2002, Gardner et al.

2006, Giblin et al. 2013). Distinguishing between the two is critical, since the former could

promote carbon storage by enhancing fixation, and the latter could potentially destroy this

service by stimulating microbial decomposition.

In a long-term, multi-investigator nutrient enrichment experiment conducted in Plum

Island Sound in northeastern Massachusetts, USA (Deegan et al. 2007, 2012), researchers

applied N in the form of NO3- into flooding waters (as opposed to dry application on the marsh

surface, as is typically done in marsh fertilization experiments) to closely mimic realistic nutrient

delivery to salt marsh systems. In contrast to previous salt marsh enrichment studies, the plant

community demonstrated only a mild response to nutrients (Johnson et al. 2016). There was no

significant response of either aboveground biomass or shifts in plant species composition due to

nutrient enrichment, which the authors attributed to the more realistic enrichment conditions and

the form of N used. There may be more energetic costs for plants associated with assimilating

NO3- when compared to NH4

+ (Lambers et al. 1998), resulting in lower uptake kinetics

(Mendelssohn & Morris 2000) and rates (Mozdzer et al. 2011), despite elevated porewater NO3-

concentrations in this enriched system (Johnson et al. 2016). It is possible that instead, the

subsurface microbial community is using much of this excess NO3- to fuel heterotrophic

metabolisms in the absence of oxygen.

17

There are several lines of evidence supporting the stimulation of microbial respiration in

response to NO3-. In a creek-scale isotope (15NO3

-) addition experiment comparing N-cycling

between an experimentally enriched marsh and its paired reference marsh, only 7 and 1.5% of

added NO3- was being assimilated by plants and benthic microalgae, respectively, indicating that

< 10% of added NO3- was being used to support primary production (Drake et al. 2009). This

suggests that this excess NO3- must be contributing to some other process. In the same

experimental marshes, Koop-Jakobsen & Giblin (2010) applied a novel push-pull method in

conjunction with isotope pairing (Koop-Jakobsen & Giblin 2009b) to measure N-cycling in the

sediment rhizosphere and found that nutrient addition stimulated denitrification and dissimilatory

nitrate reduction to ammonium, both of which use NO3- as an electron acceptor. Lastly, Deegan

et al. (2012) found that, due to a combination of lower belowground biomass and increased

microbial respiration, nutrient enrichment weakened geomorphic stability, resulting in creek

bank collapse. This degradation may not only decrease the ability of salt marshes to store carbon,

but may also release this carbon that would otherwise be stored for decades (Deegan et al. 2012),

contributing to greenhouse gas release and climate change in unknown ways. These studies

emphasize the role of NO3- in organic matter decomposition, however, in these broad field

surveys, a mechanistic understanding of the role NO3- plays in marsh carbon and nitrogen cycling

cannot be explicitly tested due to the challenge of disentangling competing processes in situ. The

goal of my dissertation, therefore, was to test how NO3- addition affected microbial

decomposition of salt marsh organic matter.

Decomposition of Organic Matter in Salt Marsh Ecosystems:

Belowground decomposition plays a major role in salt marsh ecosystems by recycling

nutrients and controlling carbon dynamics through mineralization of organic material (Howarth

18

& Hobbie 1982), ultimately determining rates of marsh accretion and carbon storage. The

microbial degradation of organic matter in salt marshes, however, is slow due to frequent

inundation, since compounds tend to diffuse ten thousand times slower through water than air.

Thus, when microbes rapidly utilize oxygen in the surface sediments, slower diffusion rates lead

to its insufficient replacement and subsequent depletion. This imbalance results in anoxia below

the top few millimeters of salt marsh sediments (Teal & Kanwisher 1961). Once organisms

deplete oxygen, microbial catabolic activity switches to anaerobic respiration, which uses

alternative electron acceptors that are generally less efficient at oxidizing some types of organic

matter (Reddy & Patrick 1975). As a result of these factors, under typical conditions,

decomposition processes tend to be slower, relative to primary production, and organic matter

accumulates, contributing to salt marsh maintenance (DeLaune & Patrick 1980). However,

whether this continues to be true under high nutrient conditions remains unclear.

The microbial decomposition of organic matter depends on the redox potential and

energy yield of the reaction, availability of electron acceptors, quality of organic carbon supply

(electron donator), and other physiochemical parameters (e.g. temperature, water-table level, and

pH; (Brinson et al. 1981, McLatchey & Reddy 1998). Redox potential (Eh) describes the

tendency of a pair of chemical species to undergo a transfer of electrons through a redox

reaction, with one accepting (reduction) and one donating electrons (oxidation). When ordering

electron acceptor half reactions by redox potential, oxygen is at the top with an Eh of +1.27 V

and carbon dioxide is at the bottom with an Eh of +0.21. The higher the Eh value, the stronger

the electron acceptor, and the greater the potential for electron transfer; this pattern explains the

order in which chemical species will reduce available electron acceptors (Berner 1980, Stumm &

Morgan 1996, Canfield et al. 2005). The electron tower, however, does not sufficiently explain

19

why organisms use one electron acceptor over another. In reality, the order in which microbes

reduce alternative electrons once anoxia occurs depends partly on the energy yield, or Gibb’s

free energy change (ΔG˚), evolved during the redox reaction. Microbes preferentially reduce

electron acceptors that yield the most energy, leading to a predictable sequence of utilization –

starting with oxygen, and proceeding through manganese, nitrate, iron, sulfate, and carbon

dioxide (Canfield 1993). In salt marsh ecosystems, microbes use carbon rich sediments to fuel

their heterotrophic metabolisms in the absence of oxygen by reducing various electron acceptors,

with nitrate (Kaplan et al. 1979, Hopkinson & Giblin 2008) and sulfate (Jorgensen 1977,

Howarth & Giblin 1983) being the most prominent.

Nitrate and sulfate are more important than manganese and iron in salt marsh sediments

because electron acceptor concentration and carbon quality also control oxidation of organic

matter. Sulfate reduction acts as the dominant metabolic strategy in salt marshes (> 50%;

Jorgensen 1977, Howarth 1984) regardless of its low energy yield (Stumm & Morgan 1996) due

to its large supply from tidal flushing (Howarth & Teal 1979) as well as high proportions of

sulfide that are oxidized back to sulfate (Jorgensen 1982). Nitrate, on the other hand, is more

thermodynamically favorable, releasing more free energy per mole of carbon oxidized (ΔG°H2 =

-420 kJ) than reducing SO42- (ΔG°H2 = -98.9 kJ) (Canfield et al. 2005). Further, this NO3

- can

have multiple fates as a respiratory substrate. In marshes, the two most common fates are

denitrification, where the NO3- is converted to a gaseous end-product that leaves the system, or

dissimilatory nitrate reduction to ammonia (DNRA), where the NO3- is converted to another

bioavailable form, ammonia, and remains in the system (Kaplan et al. 1979, Seitzinger 1988, An

& Gardner 2002, Giblin et al. 2013). Increased NO3- availability, which is typically limiting in

20

coastal systems (Ryther & Dunstan 1971), may therefore affect the microorganisms using these

resources, and consequently alter essential carbon and nitrogen cycle processes.

Salt marsh microbial communities

To assess how carbon storage responds to NO3- loading, we need to gain a better

understanding of the microbes that inhabit salt marsh sediments. Recent molecular advances

have significantly improved our ability to understand biogeochemical processes in salt marsh

ecosystems by providing a novel link between the microbial community and the processes they

mediate (Zak et al. 2006). These methods have allowed us to recognize that microbes are the key

organisms responsible for the cycling of carbon and nitrogen in marsh sediments (Benner et al.

1984, Falkowski et al. 2008). We know virtually nothing, however, about which microbes are

performing what services. This is, in part, due to high rates of dormancy in salt marsh sediments.

Previous work (Kearns et al. 2016) found that while N fertilization had no effect on total

microbial diversity (i.e. “who is there”), nutrient enriched marshes demonstrated a significant

loss in potentially active microbial diversity (i.e. “who is actively performing functions”). These

significant changes in the active microbial community, emphasize the need to move beyond

simply identifying microbial taxa through analysis of DNA. We need to examine both the active

community and explore the presence of protein coding genes through metagenomics to more

directly draw linkages between microbial community structure and ecosystem function.

Dissertation overview

In the first two chapters of my dissertation, I used controlled flow through reactor (FTR)

experiments modified from Pallud et al. (2006, 2007) that are uniquely designed to make rate

measurements at steady-state. In contrast to sediment slurries, which are commonly used in

biogeochemical flux measurements, these FTRs allow for the extraction of meaningful kinetic

21

rate measurements that can be used to quantitatively describe microbial processes in response to

environmental perturbation, such as nutrient enrichment. The system itself consists of two

Plexiglas® plates sealed with O-rings to prevent leakage, between which I loaded homogenized

salt marsh sediment and manipulated pore water conditions using a peristaltic pump system. By

carefully maintaining a uniform flow rate at steady state, I was able to infer rates of microbial

respiration by monitoring the change in outflow concentrations.

In my first chapter, I used these FTRs to investigate the effect of NO3- exposure on

microbial community structure and decomposition of organic matter at three different depths: 0-5

cm (recently deposited organic matter), 10-15 cm (within the rooting zone), and 20-25 cm

(beyond the rooting zone) in salt marsh sediments (Valiela & Teal 1974, Valiela et al. 1976). My

objective was to determine if the addition of NO3- would stimulate decomposition of salt marsh

sediments due to its role as an energetically favorable electron acceptor in heterotrophic

microbial respiration, with the hypothesis that microbial respiration would increase in response

to NO3-, but to a lesser extent in deeper sediments that contain less labile organic matter. Results

indicated a significant increase in microbial respiration, particularly denitrification, in response

to nitrate, with the most pronounced effects occurring in sediments that ranged from 150 to 200

years old. This is significant because organic matter at this depth is typically thought to be more

resistant to microbial decomposition, suggesting that the capacity of salt marshes to store carbon

may decrease with eutrophication. Additionally, NO3- addition significantly altered the microbial

community and decreased alpha diversity, selecting for taxa belonging to groups known to

reduce nitrate and oxidize more complex forms of organic matter. Together, these results suggest

the presence of a pool of organic matter that microbes can respire only with NO3- present, and

demonstrate that carbon stored for decades in salt marsh sediments are still vulnerable to

22

microbial degradation under high nutrient enrichment. This work contributes to our knowledge

base by providing mechanistic evidence for the importance of NO3- as an electron acceptor in

systems with high N loading.

The second chapter of my dissertation built on the first by investigating whether chronic

nutrient enrichment exerted any influence on the effect of additional NO3- exposure on organic

matter decomposition by either causing shifts in the microbial community, altering organic

matter chemistry, or some combination of the two. I hypothesized that sites already receiving

high NO3- inputs for long durations would not demonstrate as much stimulation in organic matter

decomposition in response to further NO3- exposure, because whatever carbon compounds

remain will be relatively resistant to degradation. Any available organic matter will have already

been oxidized due to the consistent availability of NO3- as an electron acceptor fueling

heterotrophic respiration. To test my hypothesis, I exposed sediments from three salt marshes

varying in time and intensity of prior NO3- exposure to additional NO3

- and monitored changes in

biogeochemical parameters and microbial community structure and activity. I found that, while

NO3- addition stimulated decomposition in all sediments when compared to a seawater control,

the effect was smaller at a chronically enriched site receiving sewage effluent for approximately

40 years, despite significant changes to both microbial community structure and activity. This

work suggests that long term nutrient enrichment may lead to less overall carbon storage;

however, the fraction of organic matter that does become buried is more stable when compared

to less eutrophic systems as evidenced by lower microbial respiration rates in response to

additional NO3-. These results highlight the need to consider the effects of chronic nutrient

enrichment when quantifying carbon storage potential in salt marsh systems, especially when

determining effective strategies for effective management and restoration.

23

Finally, my third chapter aimed to explore the microbes present in deep salt marsh

sediments at a larger spatial scale. Salt marshes are critical for storing carbon at rates that are

orders of magnitude greater than their terrestrial counter parts, yet, we know very little about the

microbes that mediate this ecosystem service at depth. There is also considerable evidence that

nutrient enrichment promotes dormancy in salt marsh bacteria (Kearns et al. 2016) and enhances

fungal diversity and abundance (Kearns et al. 2018) in the surface, however it is unclear what

effect NO3- addition has in deeper sediments where long term carbon storage occurs. To address

these knowledge gaps, I collected deep salt marsh cores from two marshes, one that has been

experimentally nutrient-enriched, and its paired reference site, and analyzed organic matter

characteristics and microbial community structure, abundance, and diversity. I found that both

microbial diversity and gene abundance decreased with depth, potentially due to resource

limitation, with evidence for significant rates of inactivity in deeper sediments. Depth and

associated changes in organic matter explained a large portion of microbial community structure

in shallower sediments and was driven by shifts in rare taxa. However, in deeper sediments

beyond the rooting zone, changes to the community could no longer be attributed to parameters I

measured. This pattern was likely due to more stochastic assembly at depth. Overall, this work

provides novel information on the microbes mediating carbon cycling in these critical

ecosystems and highlights the need to further examine these highly diverse microbial

communities at depth.

In conclusion, my dissertation work contributes three major findings to the literature: 1)

In addition to serving as a nutrient, NO3- can serve as an electron acceptor in metabolism and

may expand the organic matter pool available to microbial oxidation, effectively reducing overall

carbon storage potential in salt marsh systems 2) the effect of NO3- on organic matter

24

decomposition is limited, however, because any accessible forms of organic matter that become

available under high nutrient enrichment will eventually be depleted and 3) the effect of nutrient

enrichment is not detectable over longer time scales of carbon storage. Overall, my work

highlights the need to further understand the genetic machinery behind the microbes mediating

carbon storage in salt marsh sediments, how they respond to NO3- addition, and what this means

for the critical ecosystem services salt marshes provide.

25

References

Allen J (2000) Morphodynamics of Holocene salt marshes: a review sketch from the Atlantic and

Southern North Sea coasts of Europe. Quat Sci Rev 19:1155–1231

An S, Gardner WS (2002) Dissimilatory nitrate reduction to ammonium (DNRA) as a nitrogen

link, versus denitrification as a sink in a shallow estuary (Laguna Madre/Baffin Bay,

Texas). Mar Ecol Prog Ser 237:41–50

Benner R, Newell SY, Maccubbin AE, Hodson RE (1984) Relative contributions of bacteria and

fungi to rates of degradation of lignocellulosic detritus in salt-marsh sediments. Appl

Environ Microbiol 48:36–40

Berner R (1980) Early Diagenesis. Princeton University Press, Princeton, NJ

Brinson MM, Lugo AE, Place J (1981) Primary productivity, decomposition, and consumer

activity in freshwater wetlands. Annu Rev Ecol Syst 12:123–161

Canfield D (1993) Organic matter oxidation in marine sediments. In: Wollast R., Mackenzie

F.T., Chou L. (eds), Interactions of C, N, P, S Biogeohemical Cycles and Global

Change. NATO ASI Series (Series I: Global Environmental Change), vol 4. Springer,

Berlin, Heidelberg

Canfield DE, Thamdrup B, Kristensen E (2005) Aquatic Geomicrobiology. Elsevier Academic

Press, Boston, MA

Cloern JF (2001) Our evolving conceptual model of the coastal eutrophication problem. Mar

Ecol Prog Ser 210:223–253

Costanza R, Perez-maqueo O, Martinez ML, Sutton P (2008) The value of coastal

wetlands for hurricane protection. Ambio 37:241-248

Dalsgaard T, Thamdrup B, Canfield DE (2005) Anaerobic ammonium oxidation (anammox) in

the marine environment. Res Microbiol 156:457–64

Darby FA, Turner RE (2008) Effects of eutrophication on salt marsh root and rhizome biomass

accumulation. Mar Ecol Prog Ser 363:63–70

Deegan LA, Bowen JL, Drake D, Fleeger JW, Friedrichs CT, Galván KA, Hobbie JE, Hopkinson

C (2007) Susceptibility of salt marshes to nutrient enrichment and predation removal.

Ecol Appl 17:42–63

Deegan LA, Johnson DS, Warren RS, Peterson BJ, Fleeger JW, Fagherazzi S, Wollheim WM

(2012) Coastal eutrophication as a driver of salt marsh loss. Nature 490:388–392

DeLaune R, Patrick Jr W (1980) Rate of sedimentation and its role in nutrient cycling in a

26

Lousiana salt marsh. Estuar Wetl Process 11:401–412

Drake ADC, Peterson BJ, Galván KA, Deegan LA, Hopkinson C, Johnson M, Lemay LE, Picard

C, Peterson J, Deegan A, Drake C, Michael J (2015) Salt marsh ecosystem

biogeochemical responses to nutrient enrichment: A paired 15N tracer study. Ecology

90:2535–2546

Falkowski PG, Fenchel T, Delong EF (2008) The microbial engines that drive Earth’s

biogeochemical cycles. Science 320:1034–1039

Galloway JN, Leach AM, Erisman JW, Bleeker A (2017) Nitrogen: the historical progression

from ignorance to knowledge with a view to future solutions. Soil Res 55:417–424

Galloway JN, Townsend AR, Erisman JW, Bekunda M, Cai Z, Freney JR, Martinelli LA,

Seitzinger SP, Sutton MA (2008) Transformation of the nitrogen cycle: Recent trends,

questions, and potential solutions. Science 320:889-892

Gardner WS, McCarthy MJ, An S, Sobolev D, Sell KS, Brock D (2006) Nitrogen fixation and

dissimilatory nitrate reduction to ammonium (DNRA) support nitrogen dynamics in

Texas estuaries. Limnol Oceanogr 51:558–568

Giblin A, Tobias C, Song B, Weston N, Banta G, Rivera-Monroy V (2013) The importance of

dissimilatory nitrate reduction to ammonium (DNRA) in the nitrogen cycle of coastal

ecosystems. Oceanography 26:124–131

Hamersley MR, Howes BL (2005) Coupled nitrification–denitrification measured in situ in a

Spartina alterniflora marsh with a 15NH4+ tracer. Mar Ecol Prog Ser 299:123–135

Hopkinson CS, Giblin AE (2008) Nitrogen dynamics of coastal salt marshes. In: Capone D,

Bronk D, Mulholland M, Carpenter E (eds) Nitrogen in the Marine Environment, 2nd

edition. Academic Press, Burlington, MA, p 991–1036

Howarth RW (1984) The ecological significance of sulfur in the energy dynamics of salt

marsh and coastal marine sediments. Biogeochemistry 1:5–27

Howarth RW, Giblin A (1983) Sulfate reduction in the salt marshes at Sapelo Island, Georgia.

Limnol Oceanogr 28:70–82

Howarth R, Hobbie J (1982) The regulation of decomposition and heterotrophic microbial

activity in salt marsh soils: A review. In: VS K (ed) Estuarine Comparisons. Academic

Press, Inc, New York, NY, p 183–208

Howarth RW, Teal JM (1979) Sulfate reduction in a New England salt marsh. Limnol Oceanogr

24:999–1013

Johnson DS, Warren RS, Deegan LA, Mozdzer TJ (2016) Saltmarsh plant responses to

27

eutrophication. Ecol Appl 26:2647–2659

Jorgensen BB (1977) The sulfur cycle of a coastal marine sediment (Limfjorden, Denmark).


Jorgensen BB (1982) Mineralization of organic matter in the sea bed the role of sulfate

reduction. Nature 296:643–645

Kaplan W, Valiela I, Teal JM (1979) Denitrification in a salt marsh ecosystem. Limnol Oceanogr

24:726–734

Kearns PJ, Angell JH, Howard E, Deegan LA, Stanley RH, Bowen JL (2016) Nutrient

enrichment induces high rates of dormancy and decreases diversity of active bacterial

taxa. Nat Commun 7: 12881

Kearns P, Bulseco-McKim A, Hoyt H, Angell J, Bowen J (2018) Nutrient enrichment alters salt

marsh fungal communities and promotes putative fungal denitrifiers. Microb Ecol In

press.

Kirwan ML, Megonigal JP (2013) Tidal wetland stability in the face of human impacts and sea-

level rise. Nature 504:53–60

Koop-jakobsen K, Giblin AE (2009) New approach for measuring denitrification in the

rhizosphere of vegetated marsh sediments. Limnol Oceanogr Meth 7:626–637

Koop-Jakobsen K, Giblin AE (2009) Anammox in tidal marsh sediments: The role of salinity,

nitrogen loading, and marsh vegetation. Estuar Coasts 32:238–245

Koop-Jakobsen K, Giblin AE (2010) The effect of increased nitrate loading on nitrate reduction

via denitrification and DNRA in salt marsh sediments. Limnol Oceanogr 55:789–802

Lambers H, Chapin III S, Pons T (1998) Plant Physiological Ecology. Springer, Berlin Germany

Langley J, Mozdzer TJ, Shepard KA, Hagerty SB, Patrick Megonigal J (2013) Tidal marsh plant

responses to elevated CO2, nitrogen fertilization, and sea level rise. Glob Chang Biol

19:1495–1503

Levin SA, Mooney HA, Field C (1989) The dependence of plant root:shoot ratios on internal

nitrogen concentration. Ann Bot 64:71–75

Macreadie PI, Nielsen DA, Kelleway JJ, Atwood TB, Seymour JR, Petrou K, Connolly RM,

Thomson ACG, Trevathan-Tackett SM, Ralph PJ (2017) Can we manage coastal

ecosystems to sequester more blue carbon? Front Ecol Environ 15:206–213

McLatchey G, Reddy K (1998) Regulation of organic matter decomposition and nutrient

release in a wetland soil. J Environ Qual 27:1268–1274

28

Mcleod E, Chmura GL, Bouillon S, Salm R, Björk M, Duarte CM, Lovelock CE, Schlesinger

WH, Silliman BR (2011) A blueprint for blue carbon: toward an improved understanding

of the role of vegetated coastal habitats in sequestering CO2. Front Ecol Environ 9:552–

560

Mendelssohn I, Morris J (2000) Ecophysiological controls on the growth of Spartina alterniflora.

In: Weinstein N, Kreeger D (eds) Concepts and controversies in tidal marsh ecology.

Kluwer, Dordrecht, p 59–80

Morris JT (1991) Effects of nitrogen loading on wetland ecosystems with particular reference to

atmospheric deposition. Annu Rev Ecol Syst 22:257–279

Morris JT, Sundareshwar PV, Nietch CT, Kjerfve B, Cahoon DR (2002) Responses of coastal

wetlands to rising sea level. Ecology 83:2869–2877

Mozdzer TJ, Kirwan M, McGlathery KJ, Zieman JC (2011) Nitrogen uptake by the shoots of

smooth cordgrass Spartina alterniflora. Mar Ecol Prog Ser 433:43–52

Nelleman C, Corcoran E, Duarte C, Valdes L, Young C De, Foncesa L, Grimsditch G (2009)

Blue carbon. A rapid response assessment. United Nations Environment Programme.

Birkelant: GRID-Arendal

Nixon S (1995) Coastal marine eutrophication: A definition, social causes, and future concerns.

Ophelia 41:199–219

Pallud C, Cappellen P Van (2006) Kinetics of microbial sulfate reduction in estuarine sediments.

Geochim Cosmochim Acta 70:1148–1162

Pallud C, Meile C, Laverman a. M, Abell J, Cappellen P Van (2007) The use of flow-through

sediment reactors in biogeochemical kinetics: Methodology and examples of

applications. Mar Chem 106:256–271

Rabalais NN, Turner RE, Dı RJ, Justic D (2009) Global change and eutrophication of coastal

waters. ICES J Mar Sci 66:1528–1537

Reddy K, Patrick Jr. W (1975) Effect of alternate aerobic and anaerobic conditions on redox

potential, organic matter decomposition, and nitrogen loss in a flooded soil. Soil Biol

Biochem 7:87–94

Redfield AC (1972) Development of a New England salt marsh. Ecol Monogr 42:201–

237

Ryther JH, Dunstan WM (1971) Nitrogen, phosphorus, and eutrophication in the coastal marine

environment. Science 171:1008–1013

29

Seitzinger SP (1988) Denitrification in freshwater and coastal marine ecosystems: Ecological

and geochemical significance. Limnol Oceanogr 33:702–724

Stumm W, Morgan J (1996) Aquatic Chemistry: An Introduction to Emphasizing Chemical

Equilibria in Natural Waters, 2nd edition. Wiley Interscience, New York, NY

Teal JM (1962) Energy flow in the salt marsh ecosystem of Georgia. Ecology 43:614–624

Teal J, Kanwisher J (1961) Gas exchange in a Georgia salt marsh. Limnol Oceanogr 6:388–399

Valiela I, Cole ML (2002) Comparative evidence that salt marshes and mangroves may protect

seagrass meadows from land-derived nitrogen loads. Ecosystems 1:92–102

Valiela I, Teal JM (1974) Nutrient limitation in salt marsh vegetation. In: Mold RJ, Queen WH

(eds) Ecology of Halophytes. Elsevier, College Park, MD

Valiela I, Teal J, Sass W (1975) Production and dynamics of salt marsh vegetation and the

effects of experimental treatment with sewage sludge. Biomass, production and species

composition. J Appl Ecol 12:973–981

Valiela I, Teal JM, Persson NY (1976) Production and dynamics of experimentally enriched salt

marsh vegetation: Belowground biomass. Limnol Oceanogr 21:245-252

Warren RS, Fell PE, Rozsa R, Brawley AH, Orsted AC, Olson ET, Swamy V, Niering WA

(2002) Salt marsh restoration in Connecticut: 20 years of science and management.

Restor Ecol 10:497–513

Wigand C, Brennan P, Stolt M, Holt M, Ryba S (2009) Soil respiration rates in coastal

marshes subjest to increasing watershed nitrogen loads in Southern New England, USA.

Wetlands 29:952–963

Zak DR, Blackwood CB, Waldrop MP (2006) A molecular dawn for biogeochemistry. Trends

Ecol Evol 21:288–95

30

Chapter 1: Nitrate addition stimulates microbial decomposition of organic matter in salt

marsh sediments

In collaboration with: Anne E. Giblin, Jane Tucker, Anna E. Murphy, Jonathan Sanderman, and

Kenly Hiller-Bittrolff

Abstract

Salt marshes store carbon at rates that are more than an order of magnitude greater than

their terrestrial counterparts, helping to mitigate negative consequences of climate change. As

nitrogen loading to coastal waters continues to rise, primarily in the form of nitrate, it is unclear

what effect it will have on carbon storage capacity of these highly productive systems. This

uncertainty is largely driven by the dual role nitrate can play in biological processes, where it can

serve as either a nutrient that stimulates primary production or a thermodynamically favorable

electron acceptor fueling heterotrophic metabolism. Here, I used a controlled flow through

reactor experiment to test the role of nitrate as an electron acceptor, and its effect on organic

matter decomposition and the associated microbial community in salt marsh sediments. I

observed a significant increase in organic matter decomposition in response to nitrate and found

that this pattern persisted even at sediment depths typically considered to be less labile. Nitrate

addition significantly altered the microbial community and decreased alpha diversity, selecting

for taxa belonging to groups known to reduce nitrate and oxidize more complex forms of organic

matter. Fourier Transform-Infrared Spectroscopy data further supported these results, suggesting

that nitrate facilitated decomposition of complex organic matter compounds into more labile

forms. Taken together, these results suggest the existence of organic matter pools that only

become accessible with nitrate and would otherwise remain stable. The existence of such pools

could have important implications for carbon storage, since greater decomposition rates may

result in less overall burial of organic-rich sediment. Given the extent of nitrogen loading along

31

our coastlines, it is imperative that we better understand the resilience of salt marsh systems to

nutrient enrichment, especially if we hope to rely on salt marshes, and other blue carbon systems,

for long-term carbon storage.

Introduction

Carbon dioxide (CO2) concentrations continue to rise as a result of fossil fuel burning and

land-use changes, thereby contributing to increases in global temperature, ocean acidification,

and sea level rise. While a number of mitigation strategies have been proposed, recent emphasis

has been placed on sequestering CO2 in blue carbon habitats (Dargusch & Thomas 2012), which

include salt marshes, mangroves, and seagrass meadows (Nelleman et al. 2009, Mcleod et al.

2011). Salt marshes are particularly efficient at storing carbon due to high levels of primary

production, the ability to trap organic rich sediments (Chmura et al. 2003), and low rates of

microbial decomposition due to largely anaerobic conditions below the first few millimeters of

the surface (Reddy & Patrick Jr. 1975). They can bury carbon at a rate more than an order of

magnitude greater than that of their terrestrial counterparts, over time scales of thousands of

years (Duarte et al. 2005, Mcleod et al. 2011). As such, they have become a major focus of

coastal restoration projects (Warren et al. 2002, Macreadie et al. 2017).

Salt marshes face several anthropogenically-driven threats that can diminish, and

potentially reverse, their capacity to store carbon. Here, I focus on the role of coastal nitrogen

(N) inputs, which continue to increase in many systems due to fertilizer production, agricultural

and urban runoff, enriched groundwater, and atmospheric deposition (Galloway et al. 2017).

While salt marshes can remove some of this anthropogenic N before entry into the coastal ocean,

either by assimilation into plant biomass (Valiela & Teal 1974) or conversion to gaseous

32

products (NO, N2, N2O) via denitrification or anammox (Kaplan et al. 1979, Hopkinson & Giblin

2008, Koop-Jakobsen & Giblin 2009), it is unclear how much N loading salt marshes can

withstand without having negative implications for carbon storage. In general, salt marshes are

more resilient to N loading when compared to other coastal systems because of their ability to

efficiently remove N (Valiela & Cole 2002). There is considerable evidence that nutrient

enrichment stimulates aboveground primary production (Kaplan et al. 1979, Morris et al. 2002,

Vivanco et al. 2015), which facilitates sediment trapping and marsh accretion (Morris et al.

2002) and augments the carbon sink potential by adding biomass. Other studies have also

observed increased belowground production in response to elevated N (Pastore et al. 2017). In

some systems, however, responses to N enrichment diminished carbon storage capacity,

including lost root biomass, increased belowground microbial respiration, and changes in species

composition, all of which can result in lower sediment stability and potential marsh collapse

(Deegan et al. 2012, Langley et al. 2013). Due to these complexities, the exact response of the

marsh carbon storage capacity to increased N loading remains unclear.

One plausible explanation for conflicting observations among marsh fertilization

experiments may be the form of N that is applied. Many studies cover small spatial scales and

apply N in its reduced form, ammonium (NH4+) or urea; although some use a mix of oxidized

and reduced forms, ammonium nitrate (NH4NO3). In contrast, much of the N delivered to the

coastal zone occurs in its oxidized form, nitrate (NO3-) (Galloway et al. 2008). In addition to

supporting primary production through assimilation by marsh vegetation, benthic microalgae,

and phytoplankton, NO3- can also serve as an energetically favorable electron acceptor to fuel

microbial oxidation of organic matter (OM) through various anaerobic respiration processes,

including denitrification (Kaplan et al. 1979, Hamersley & Howes 2005) and dissimilatory

33

nitrate reduction to ammonium (DNRA; Rysgaard et al. 1996, An & Garnder 2002, Giblin et al.

2013). Sulfate (SO42-) is another important electron acceptor in salt marsh sediments, accounting

for up to 70-90% of total sediment respiration (Howarth 1984, Howarth & Teal, 1979) due to its

virtually unlimited supply from incoming seawater. However, these two electron acceptors are

different thermodynamically in that reducing NO3- releases more free energy (ΔG°H2 = -420 kJ)

than reducing SO42- (ΔG°H2 = -98.9 kJ) (Canfield et al. 2005). Increased NO3

- availability, which

is typically limiting in coastal systems (Ryther & Dunstan 1971), may therefore affect the

microorganisms using these resources, and consequently alter the ecosystem functions they

mediate.

The mechanisms by which this change in function could occur include: 1) a shift in total

microbial community structure to an alternative state better fit for a high NO3- environment

through change in electron acceptor availability 2) alteration of metabolic capacity of the

existing microbial community to N-cycling metabolisms due to high physiological plasticity, or

3) some combination of the two (Meyer et al. 2004, Allison & Martiny 2008, Shade et al. 2012).

Considering the fundamental role microbes play in carbon decomposition, and more indirectly,

long-term carbon storage (Benner et al. 1984, Falkowski et al. 2008), it is essential that we tease

apart which of these mechanisms control microbial and ecosystem response to NO3- addition.

Regardless of the mechanism, prior studies in salt marsh systems suggest functional responses to

NO3- do occur (Koop-Jakobsen & Giblin 2010, Deegan et al. 2012). When compared to SO4

2-

reducers, NO3- reducers, as well as other microbes adapted to high N environments (Treseder et

al. 2011) may oxidize more complex forms of OM (Achtnich et al.1995), potentially resulting in

decomposition of OM that would have otherwise remained stable.

34

To better quantify the role of marshes in long-term carbon storage it is critical to

understand how these systems respond to increasing NO3- concentrations. In this study, I

investigate whether NO3- addition increases decomposition of salt marsh OM. To explicitly

address this question, I implemented a controlled flow through reactor (FTR) experiment, where

I exposed salt marsh sediments to elevated levels of NO3-. I hypothesized that the addition of

NO3- would stimulate the decomposition of OM when compared to unamended sediments, and

that these experiments would reveal the presence of a “NO3- accessible” pool of OM that

microbes could only oxidize in the presence of this more favorable electron acceptor. I also

examined whether depth and age of OM would play a role in the salt marsh sediment response to

NO3- addition. Specifically, I hypothesized that there would be little difference in decomposition

between the NO3- and unamended treatments in shallow sediments, since the OM there would be

recently deposited and relatively labile, making it accessible for both SO42- and NO3

- reduction.

Further, I hypothesized that there would be an overall reduction in decomposition in deeper

sediments, where OM lability decreases and becomes less amenable to microbial oxidation, but

that there would be a greater stimulation of decomposition at depth in the NO3- treatment

compared to the unamended sediments. Lastly, I hypothesized that these changes in metabolic

function would result from a shift in the microbial community towards taxa better adapted to use

NO3- in metabolic functions, such as denitrification and DNRA.

2. Materials and methods

2.1 Sample collection

I assessed the effect of NO3- on the decomposition of sediment OM of varying ages by

collecting samples along a depth gradient from salt marsh sediments located in West Creek, part

35

of a marsh complex located in Plum Island Sound, MA (42.759 N, 70.891 W). West Creek is a

relatively pristine reference site monitored as part of a long-term nutrient enrichment experiment

called the TIDE project (Deegan et al. 2007). I collected three replicate cores (5 cm diameter and

30 cm deep) from the tall ecotype of Spartina alterniflora, a habitat that floods daily and is

underwater approximately 35% of the time (Deegan et al. 2007). I sectioned each core into

shallow (0-5 cm), mid (10-15 cm), and deep (20-25 cm) sediments and homogenized sections

under anoxic conditions. I chose these depths to include OM of varying quality, ranging from

relatively newly deposited OM (shallow), to older OM found both within (mid) and beyond

(deep) the rooting zone (Valiela & Teal 1974, Valiela et al. 1976). Based on accretion rates taken

from nearby sites, I estimate that these sediments range from 50 to 100 years in age (Forbrich et

al. 2018). Before proceeding, I removed any root material visible to the naked eye from the

homogenized cores, standardized across all cores by total time searching. I then split each

sectioned depth into a plus-NO3- and an unamended treatment (filtered seawater). This resulted in

three replicates for each treatment at each depth.

2.2 Flow through reactors and experimental design

The flow-through reactor experimental system (Fig. S1) is a modified version of the

system described in Pallud et al. (2007) and Pallud & Van Cappellen (2006). In contrast to

whole-core batch incubations or sediment slurries, flow-through reactors provide biogeochemical

rate measurements at steady-state conditions and prevent dissolved metabolic byproducts from

accumulating in the system. Each flow-through reactor has a volume of 31.81 cm3 and consists of

two Plexiglas® caps that are radially scored for uniform flow. I confirmed unilateral,

homogenous flow in each reactor using the conservative tracer, bromide, in breakthrough

36

experiments (see supplemental methods for details and supplemental Table S1 and Fig. S2 for

flow property results).

Under anoxic conditions I loaded each reactor with homogenized sediment, and randomly

assigned each reactor a treatment, plus-NO3- (+NO3

- in 0.2 µm filtered seawater) or unamended

(0.2 µm filtered seawater only, representing natural salt marsh conditions). To prepare the two

treatment reservoirs, I filtered (0.2 µm) water collected from Woods Hole, MA, sparged each

with N2 gas for approximately 20 minutes until they reached anoxic conditions, and spiked the

NO3- reservoir with 500 µmol L-1 additional K15NO3

- (Cambridge Isotope Laboratories, Andover,

MA). 500 µmol L-1 is high when compared to natural conditions, ranging from approximately 2-

5 times higher than porewater concentrations found in nutrient enriched marshes (e.g. Negrin et

al. 2011, Peng et al. 2016). However, my goal was to compare sediment OM decomposition

under NO3- enriched and unamended conditions rather than to compare to field rates. Therefore,

it was critical that NO3- be available to microbes through the entire thickness of sediment in each

reactor of the enriched treatment. I initially added 350 µmol L-1 for the first 25 days, but found

that all the NO3- was being consumed. At this point, I increased the concentration top 500 µmol

L-1.

Half of the reactors received the plus-NO3- treatment and half received the unamended

treatment, both at a targeted flow rate of approximately 0.08 mL min-1 (see Table S1 for

measured flow rate) using peristaltic pumps rigged with 0.89 mm (inner-diameter) MasterFlex

FDA viton tubing (Cole Parmer, IL, USA). I then carried out a 92-day experiment under anoxic

conditions in a glove bag flushed with nitrogen. Once the FTRs reached steady state at the 10-

day mark, I collected samples from both the reservoirs and the effluent throughout the

experiment to measure changes in biogeochemical parameters and to monitor flow rate. To

37

assess changes in OM composition and microbial community structure, I homogenized and

aliquoted bulk sediment from the start of the experiment (pre) and from sediment in each reactor

at the end of the experiment. I dried bulk sediments overnight at 65ºC before freezing at -20°C,

and immediately flash froze additional aliquots of sediments in liquid nitrogen for nucleic acid

extraction and stored them at -80°C until further analysis.

2.3 Biogeochemical and OM analyses

I collected water samples approximately every 10 days from both the plus-NO3- and the

unamended reservoir along with all reactor outflows to measure biogeochemical processes

resulting from microbial activity. Samples for DIC, sulfide and gases were collected in glass

tubes placed in-line in the outflow with no head space. To assess total microbial respiration, I

measured dissolved inorganic carbon (DIC; CO2 + HCO3 + CO32-) on an Apollo SciTech AS-C3

DIC analyzer (Newark, DE) following methods in Dickson & Goyet (1994). I measured nitrate +

nitrite (NO3- + NO2

-) via chemiluminescence on a Teledyne T200 NOx analyzer (Teledyne API,

San Diego, CA) following methods outlined in Cox (1980), and measured ammonium (NH4+)

and sulfide colorimetrically on a Shimadzu 1601 spectrophotometer (Kyoto, Japan) following

protocols from Solorzano (1969) and Gilboa-Garber (1971), respectively. To calculate

production and consumption rates of each analyte (DIC, 3-, NH4

+, and sulfide) over time, I

calculated the difference in concentration between the inflow (reservoir) and the outflow

(effluent), corrected for flow rate in L hr-1, and divided by reactor volume (31.81 cm-3) for each

sampling point. Because I was not able to measure changes in SO42- due to high seawater

concentrations and proportionally minor changes resulting from experimental conditions, I

determined that sulfate reduction was occurring through the production of sulfide (HS-) and

38

calculated total sulfate reduction rates (SRR) by taking the sum of HS- produced and total S

storage measured at the end of the experiment (described below). I also calculated the DIC:NH4+

ratio to draw general inferences about OM pools being decomposed based on C:N stoichiometry.

To assess geochemical changes in OM, I dried samples at 65°C and fumed samples with

12N HCl before performing elemental composition analysis (percent carbon and nitrogen) on a

Perkin Elmer 2400 Series Elemental Analyzer (Perkin Elmer, Billerica, MA) using acetanilide as

a standard. I dried additional samples at 105°C overnight to obtain water content and used these

data to calculate bulk density of each reactor assuming a volume of 31.81 cm3. Lastly, I obtained

percent sulfur (%S) by combusting dried samples at 1350°C and measuring sulfur dioxide (SO2)

production on a LECO S635 S analyzer (LECO Corporation, Saint Joseph, MI).

To further characterize changes in OM as a result of NO3- addition, I used Fourier-

Transform-Infrared Spectroscopy (FT-IR), a technique that provides rapid, detailed information

about the relative abundance of chemical functional groups. To prepare samples for FT-IR

analysis, I finely ground sediment dried at 40°C for 48 hours. I ran each sample on a Bruker

Vertex 70 Fourier Transform Infrared Spectrometer (Bruker Optics Inc., Billerica, MA) outfitted

with a Pike AutoDiff diffuse reflectance Accessory (Pike Technologies, Madison, WI) and

obtained data as pseudo-absorbance (log[1/reflectance]) in diffuse reflectance mode. I collected

data at a 2 cm-1 resolution with 60 co-added scans per spectrum at the mid-IR range, from 4000-

400 cm-1, using a mirror for background correction. Resulting raw spectra were transformed

using a calculated two-point linear tangential baseline using Unscrambler X (Camo Software,

version 10.1, Woodbridge, NJ) and then assigned peaks according to Margenot et al. (2015) and

Parikh et al. (2014).

39

2.4 Nucleic acid extraction, amplification, and amplicon sequencing

I extracted genomic DNA from approximately 0.25 g wet sediment using the MoBio®

PowerSoil DNA Isolation Kit (MoBio Technologies, CA, USA) following manufacturer’s

instructions, and eluted the DNA into a 35 µL final volume. I amplified in triplicate the V4

region of the 16S rRNA gene using the general bacterial primer-pair 515F (5’-

GTGCCAGCMGCCGCGGTAA-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’)

(Caporaso et al. 2011) with Illumina adaptors (Caporaso et al. 2012) and individual 12-bp GoLay

barcodes on the reverse primer, using the following reaction: 10 µl 5-Prime Hot Master Mix

(Quanta Bio, Beverly, MA), 0.25 µl of 20 M forward and reverse primers, 13.5 µL DEPC-

treated water, and 1 µl of DNA template. PCR cycling conditions follow those outlined by the

Earth Microbiome Project (Caporaso et al. 2011). Although these primers are biased against the

SAR11 group (Apprill et al. 2015), these bacterioplankton are aerobic (Giovannoni 2017), and

should not play a large role in the microbial community associated with my anoxic experimental

conditions. Prior to sequencing, I gel-purified the pooled PCR product using a Qiagen®

QIAquick gel purification kit (Qiagen, Valencia, CA) and quantified the resulting purified

product using a Qubit® 3.0 fluorometer (Life Technologies, Thermo Fisher Scientific, Waltham,

MA). After pooling to equimolar concentrations, I performed sequencing on the Illumina MiSeq

(Illumina, San Diego, CA) platform using a 300-cycle kit and V2 chemistry. All reads are

deposited in the NCBI Sequence Read Archive under accession number TBD.

I quantified the abundance of 16S rRNA gene copies by performing quantitative PCR

(qPCR) in triplicate using 357F and 519R primers (Turner et al. 1999) and the following 20 µl

reaction: 10 µl Brilliant III Ultrafast SYBR Green qPCR Master Mix (Agilent Technologies), 0.5

µl of each primer (20 µM), 8 µl of DEPC-treated water, and 1 µl DNA template. I ran qPCR on

40

an Aria Mx Real Time PCR instrument (Agilent Technologies) using a program optimized for

16S rRNA gene targets, which consisted of an initial denaturation step at 95ºC for 3 minutes, and

40 cycles of 95ºC for 5 seconds and 60ºC for 10 seconds. I acquired data at the end of each cycle

and conducted a melt curve to confirm the size of the target gene amplification product. I ran

standard curves with each sample batch using stock from purified 16S rRNA gene product that I

quantified on a tape station (Agilent Technologies), resulting in an R2 > 0.95 and qPCR

efficiency >90%.

2.5 Statistical analyses

To investigate changes in DIC production over time, I performed a linear regression on

each core using time as the explanatory variable. I integrated between sampling points to

calculate the cumulative flux across the length of the experiment and tested for significant

differences in DIC, NH4+ production, sulfur storage, and total SO4

2- reduction, as a function of

treatment, depth, and their interaction using a two-way ANOVA. For both NO3- consumption

and sulfide production, both of which were only detectable in one of the two treatments, I

assessed differences among depths using a one-way ANOVA. To account for differences in bulk

carbon supply on DIC production, I also calculated total carbon loss by taking the proportion of

carbon released as DIC divided by the total mass of carbon per reactor using sediment

characteristic data (e.g. water content and %C).

To assess changes in %C, %N, bulk density, and %S throughout the experiment, I

calculated the difference between initial and final sediments per core and compared these relative

changes among treatment and depth using a two-way ANOVA. I performed principal coordinate

analysis (PCoA) across the entirety of the FT-IR spectra and used a PERMANOVA with 999

41

permutations to test for significant differences by treatment and depth using Manhattan distance

to construct the resemblance matrix from the FT-IR data. To further visualize trends in these

data, I also plotted the Pearson’s correlation coefficients against wavenumber to determine which

spectral bands best explained the distribution of sample scores in the PCoA based on functional

group assignments in Table 1. Lastly, I calculated a relative recalcitrance index according to the

following equation:

Eq. 1 Index II = 2924 + 2850 + 1650 + 1470 + 1405 + 920 + 840

3400 + 1270 + 1110 + 1080

where each value represents a wavenumber (Table 1) corresponding to either a carbon

(numerator) or oxygen-bonded (denominator) functional group. Higher Index II values are

typically associated with greater OM recalcitrance (Ding et al. 2002, Veum et al. 2014). I used a

two-way ANOVA to compare Index II values to infer relative recalcitrance as a function of

treatment and depth.

To investigate bacterial community composition, I analyzed sequence data in QIIME 2

(version 2017.12; Caporaso et al. 2010, QIIME 2 Development Team). I demultiplexed a total of

1,521,493 16S rRNA gene sequences across all samples and inferred amplicon sequence variants

(ASVs) using the DADA2 plugin (Callahan et al. 2016) with a maxEE of 2 and the consensus

chimera removal method. Quality filtering resulted in an average of 37,381 (± 6,693) sequences

per sample. I then assigned taxonomy with the Greengenes 16S rRNA sequence database

(version 13-8; McDonald et al. 2012) and removed ASVs occurring only once (singletons) and

any sequences matching chloroplasts and mitochondria. After aligning sequences using MAFFT

v7 (Katoh & Standley 2013), I performed beta diversity analysis with weighted UniFrac

(Lozupone et al. 2011) on ASV tables normalized to 22,999 sequences (which was my lowest

42

sequencing depth), and tested for significant differences among treatments and depth using

PERMANOVA with 999 permutations. To examine within sample diversity, I calculated a

Shannon diversity index with these normalized data and tested for differences across treatment

and depth using a two-way ANOVA. I ran a random forest model from the randomForest R

package (v4.6-12; Liaw & Wiener 2002) using 10,001 trees on a filtered feature table containing

ASVs present at least 100 times (186 ASVs total) to identify taxa most important in classifying

between plus-NO3- and unamended treatments, and confirmed model results by examining the

out-of-bag error rate (a method that uses bootstrap aggregation to assess performance without a

training set) and leave-one-out cross-validation with 999 permutations in the caret R package

(v6.0-73; Kuhn 2016). Lastly, I compared 16S rRNA gene abundance using a two-way ANOVA

with depth and treatment as fixed effects. I conducted all statistical analyses in R (R Core Team

2012) unless otherwise stated and used an alpha of 0.05 for all significance testing.

3. Results

3.1 Biogeochemical rates

Across all depths, the addition of NO3- resulted in higher DIC production rates (microbial

respiration; Fig. 1) and total cumulative production (Fig. 2) compared to the unamended

treatment, both over time and at the end of the experiment. In both treatments, total DIC

production decreased with depth (Fig. 2), with shallow sediments exhibiting significantly greater

microbial respiration than mid and deep sediments. While DIC production rates decreased over

the duration of the experiment in the shallow, unamended sediments (linear regression; p=0.002,

F(1,18) = 12.87, R2 = 0.38), no such pattern existed in the plus-NO3- treatment.

43

I measured NO3- consumption and sulfide production in reactor effluent (plus total S in

sediment) to assess nitrate reduction rates (NRR) and sulfate reduction rates (SRR), respectively.

Background NO3- concentrations in the incoming seawater were consistently low (0.6-1.2 µM).

Although all of this nitrate was removed throughout the experiment in the unamended treatment

the low initial NO3- resulted in negligible NRR. In the plus-NO3

- treatments, NRR ranged from

14.6-87.2 µmol cm-3, accounting for 86.7% ± 0.05, 98.4% ± 0.08, and 101.9% ± 0.09 of DIC

production in shallow, mid, and deep sediments respectively. There was not a significant

difference in total NRR with depth (Fig. 3), although the shallow sediments demonstrated

elevated NRR in a manner comparable to DIC production (Fig. 2).

Sulfide production occurred at all depths in the unamended treatment and was

significantly higher in shallow compared to deep sediments (Fig. 4B), but was undetectable in

the plus-NO3- treatment. However, changes in S storage indicated that SO4

- reduction was

occurring in all treatments (Fig. 4C), although increases were smaller in the plus-NO3- treatment

than in the unamended sediments (Fig. 4D). In the unamended treatment, SO4- reduction

accounted for 76.8% ± 1.4, 64.2% ± 9.3, and 59% ± 30.8 of total DIC production in shallow,

mid, and deep sediments respectively. When combined with NRR in the plus-NO3- treatment, I

could explain 97.7% ± 5.8, 110.1% ± 15.1, and 107.6% ± 8.9 of total DIC production. There was

a significant interaction between treatment and depth on sulfate reduction, with the unamended

treatment exhibiting more sulfate reduction than the plus-NO3- treatment in shallow sediments,

and unamended shallow sediments exhibiting more sulfate reduction than the unamended deep

sediments (Fig. 4D).

While there was no significant difference in NH4+ production between treatments,

shallow sediments produced significantly more NH4+ when compared to mid and deep sediments

44

in both the plus-NO3- and unamended treatments (Fig. 5). In addition, the DIC:NH4

+ ratio was

significantly higher in the plus-NO3- treatment, while the unamended treatment remained

consistently low across all depths (Fig. 6) and was similar to the C:N value of sediments from the

start of the experiment (13.66 ± 0.69).

3.2 Organic matter

I compared change in C, N, molar C:N, and S in the plus-NO3- and unamended sediments

versus sediments collected prior to the experiment (Table 2). There was no significant difference

in pre- versus post-experiment carbon or molar C:N between treatments; however, there was a

change in N by depth and total S by depth and treatment, with significantly greater S

concentrations in the unamended treatment and lower concentrations in the deep sediments (Fig.

4C).

The proportion of carbon lost as DIC throughout the experiment ranged only from 0.76 to

3.47% of the total carbon in each reactor, so it is not surprising that I did not detect significant

changes in most bulk sediment properties between treatments. To observe more precise changes

in OM, I applied FT-IR spectroscopy and explored relative shifts in chemical functional groups

related to decomposition processes. A principal coordinates analysis (PCoA; Fig. 7A) of the

whole FT-IR spectra using Manhattan distances indicated separation by depth along the first

coordinate axis (explaining 77.6% of the variance) and treatment along the second coordinate

axis (explaining 22.2% of the variance), both of which were significant according to

PERMANOVA analysis (Fig. 7A). Pairwise comparisons of mean Manhattan distances further

indicated that each depth was significantly different from the rest (shallow-mid, p=0.006,

t=3.029; mid-deep, p=0.009, t=2.72; shallow-deep, p=0.001, t=5.94), but when examining

treatment, only pre- and unamended sediments were significantly different from each other

45

(p=0.019, t=2.20). I next plotted the Pearson’s correlation coefficient against wavenumber,

which accounted for 99.8% of total variance in the PCoA. This allowed me to identify three

functional groups that exhibited the most influence on observed separation along the primary and

secondary axes (Fig. 7B): Three groups that were important in differentiating among treatments

included the lignin-like compounds at 840 and 1650 cm-1 (Artz et al., 2008), aliphatic carbon at

1470 cm-2, and polysaccharides at 1080 cm-1.To further explore whether a change in carbon

quality occurred during the incubation, I calculated Index II (Eq. 1), where higher values imply

greater recalcitrance. In both the plus-NO3- and unamended treatments, Index II was significantly

greater than in initial sediments when compared to after the incubation, although there was no

difference between the two treatments (Fig. 7C), nor was there a difference by depth.

3.3 Microbial community composition and abundance in response to nitrate

There was no difference in 16S rRNA gene abundance by treatment (p=0.385) or depth

(p=0.233) according to a two-way ANOVA (Fig. S3). A principal coordinates analysis

constructed from Weighted UniFrac similarities revealed a significant effect of both treatment

and depth on microbial community composition (Fig. 8A). There was a clear separation in

community similarity along the primary axis (43.20% of the variance explained) due to NO3-

addition, and a separation driven primarily by differences between shallow and mid/deep

sediments (Fig. 8A) along the secondary axis (20.82% of the variance explained). To determine

the effect of NO3- addition on alpha diversity, I calculated the Shannon Index and found a

significant effect of both treatment and depth, but not the interaction of the two factors. Across

all depths, alpha diversity was significantly lower in the plus-NO3- treatment when compared to

the unamended treatment (Fig. 8B).

46

A random forest model, using 10,000 trees and 186 predictor variables derived from the

most abundant ASVs, correctly classified microbial communities as belonging to either the plus-

NO3- or unamended treatment 100% of the time with a 0% out-of-bag error rate. Leave-one-out

cross-validation confirmed model performance, with a Cohen’s kappa statistic of 100%, which

compares observed accuracy to expected accuracy due to random chance. The top 30 ASVs most

important in discriminating between treatments accounted for 45.2% of total sequences and

included taxa from Phyla Bacteroidetes, Proteobacteria, Chlorobi, Caldithrix, Chloroflexi,

Planctomycetes, Acidobacteria, Gemmatimonadetes, Verrucomicrobia, and candidate group

WWE1 (Table S2; Fig. 9). Out of these 30 ASVs, classes from Flavobacteria,

Gammaproteobacteria, Alphaproteobacteria, and Ignavibacteria were more abundant in the plus-

NO3- treatment, while the unamended treatment was much more diverse, including classes from

Deltaproteobacteria, Bacteroidia, Caldithrix, Anaerolineae, Cloacamonae, BPC102, Gemm-2,

Phycisphaerae, Epsilonproteobacteria, Alphaproteobacteria, Verrucomicrobiae, and

Betaproteobacteria. I also tested the random forest model without excluding rare taxa (but still

removing singletons) to see if these rare ASVs would have a disproportionate influence on the

dataset. This also resulted in 100% classification rate and 0% out-of-bag error, but only

accounted for an additional 1.9% of all sequences (ASVs 31-41 listed in Table S1).

4. Discussion

4.1 DIC production rates decreased with depth

This study used a controlled FTR experiment to test the effect of OM quality and NO3-

addition on microbial respiration. I found that DIC production decreased as a function of depth,

with shallow sediments exhibiting significantly greater microbial respiration rates than mid and

47

deep sediments (Fig. 2). This pattern was particularly evident in the unamended treatment, where

DIC production also decreased throughout the duration of the experiment in the shallow

sediments (Fig. 1A). Decreasing DIC production with depth likely resulted from changes in OM

lability that also occurred with depth. Microbes preferentially degrade the most labile plant and

microalgae derived OM in surface sediments; the less labile components that remain accumulate

over time, are more recalcitrant, and ultimately become buried under newly deposited organic

matter and sediment (Cowie & Hedges 1994, Wakeham et al. 1997). While microbes can still

degrade these less labile organic compounds at depth, it occurs at a much slower rate (Westrich

& Berner 1984), resulting in decreased decomposition rates. Initial bulk sediment carbon and

molar C:N in this experiment did not differ significantly among the different depths (Table 1);

although the FT-IR spectra indicated a significant difference in functional groups (Fig. 7A) at

different depths driven primarily by polysaccharide depletion (Fig. 7B), which suggests

decreasing lability in deeper sediments.

Decreasing DIC production with depth can also be explained, in part, due to decreasing

availability of thermodynamically favorable electron acceptors in anoxic marsh sediments.

Energetically-favorable electron acceptors, such as NO3-, are preferentially reduced at the surface

and are therefore depleted in deeper sediments (Canfield et al. 2005). While SO42- is rarely

limiting in most marsh systems due to its high concentration in seawater and delivery via

incoming tides (Jørgensen 1977, Howarth & Teal 1979), SO42- reduction is a much less

energetically favorable metabolic pathway, releasing less free energy per mole of carbon

oxidized compared to NO3- reduction. Since there is less energy available to degrade OM that has

accumulated with time, rates of decomposition generally decrease with depth (Canfield et al.

2005, Arndt et al. 2013); though the presence of roots and bioturbation can alter this pattern

48

(Aller & Aller 1998, Kostka et al. 2002, Canfield & Farquhar 2009). This is consistent with the

decrease in DIC production (Fig. 2) and sulfide production (Fig. 4B) I observed in the deeper

unamended sediments. Further, DIC production decreased as a function of time in the surface

unamended treatment (Fig. 1), suggesting that after first oxidizing the more labile OM

compounds, only more recalcitrant, less available OM remained, leading to decreased

decomposition rates. This result corroborates other studies that find a strong relationship between

OM degradability and SRR, with decreasing OM lability resulting in lower decomposition rates

regardless of SO42- availability (Westrich & Berner 1984, Canfield 1989).

4.2 Evidence for a nitrate accessible pool of OM

The addition of NO3- resulted in significantly greater DIC production across all depths,

most notably in deeper sediments, where OM is older and less labile. While NO3- is a

thermodynamically favorable terminal electron acceptor that fuels high rates of denitrification

and DNRA in salt marshes, it is typically coupled with nitrification at oxic interfaces or rooting

zones (Hamersley & Howes 2005; Howes et al. 1981), and hence limited at depth where it

cannot be internally regenerated. By experimentally adding NO3- here, similar to what might

occur in coastal environments under high N loading, I thereby increased NO3- availability. In

doing so, I increased rates of NO3- reduction (Fig. 3) and OM oxidation (Fig. 1-2), and

consequently increased rates of decomposition. These results suggest the existence of a “NO3--

accessible” OM pool and emphasize that the definition of “recalcitrant” can differ depending on

both OM lability and electron acceptor availability. OM that is considered recalcitrant under

SO42--only conditions may no longer be stable under NO3

- availability.

High NO3- conditions may stimulate the microbial community to break down these

otherwise stable, less labile OM compounds by providing more energy for metabolic processes.

49

Higher DIC:NH4+ ratios in the plus-NO3

- treatment provide support for this claim. In general,

more DIC relative to NH4+ production indicates that microbes are using OM with higher C:N

ratios (Canfield et al. 2005). In the plus-NO3- treatment, particularly at depth, the DIC:NH4

+ ratio

was much higher, suggesting that microbial communities may be accessing a different OM pool

compared to unamended sediments, which remained consistently low and very similar to the

average sediment C:N ratio from the start of the experiment (Fig. 5). It is noteworthy that the

ratio of DIC:NH4+ in the plus-NO3

- and unamended treatment were similar in shallow sediments,

where OM is more labile and appears to be accessible to both NO3- and SO4

2- reducers. As this

ratio diverges between treatments with depth, it provides further evidence for the existence of

this separate “NO3--accessible” OM pool that microbes can access once NO3

- limitation is

released. There are other processes by which this increasing pattern in DIC:NH4+ can emerge,

including differences in microbial biomass and N uptake or anammox (Thamdrup et al. 2002,

Dalsgaard et al. 2005, Schmid et al. 2007). My NH4+ data seem to suggest, however, that there

were no significant differences in uptake or regeneration between treatments, given that

cumulative NH4+

production across the entire experiment was the same (Fig. 5). Further, I have

stable isotope 29N2 production data (Bulseco-McKim et al. In Prep) showing that anammox was

negligible in this experiment, agreeing with other studies conducted in salt marsh sediments

(Koop-Jakobsen & Giblin 2009). I therefore conclude that this increase in DIC:NH4+ ratio may

be explained, at least in part, by the oxidation of a higher C/N pool of OM at depth in the plus-

NO3- treatment.

FT-IR spectral data also suggest that microbes in the plus-NO3- treatment were accessing

a different pool of OM than microbes in the unamended treatment. A PCoA of the whole FT-IR

spectra from 4000-400 cm-1 indicated a significant difference in the OM chemistry among depths

50

and between the unamended and pre sediments, but not the plus-NO3- treatment (Fig. 7A). This

result suggests that decomposition not only caused a shift in the OM signature when compared to

pre-treatment sediments, but also, that the plus-NO3- and unamended OM composition shifted in

different ways. In addition, higher Index II values in both the plus-NO3- and unamended

treatments compared to the pre-treatment sediments show that microbial OM oxidation resulted

in more recalcitrant OM (Fig. 7C), a result that I could not detect in the bulk sediment properties

(Table 2).

Remarkably, these data also suggest that after incubation, the OM from the unamended

treatments was more recalcitrant than the OM from the plus-NO3- treatment (Fig. 7C), even

though the amount of C mineralized was less, which supports the pattern observed in the PCoA

of OM composition (Fig. 7A). One possible explanation for this finding is that NO3- addition

might facilitate decomposition of more complex OM (e.g. large cyclic compounds such as

cellulose), either through fermentation or hydrolysis, which would result in more labile, low-

molecular-weight substrates (Beauchamp et al. 1989). Rather than a strict predictable sequence

following thermodynamic theory, which asserts that electron acceptors with higher redox

potential are exclusively reduced first (Zehnder & Stumm 1988), these results suggest that NO3-

supports co-metabolism by providing more labile OM compounds for competing microbial

functional groups (Achtnich et al. 1995). Further, the fact that the pre- and plus-NO3- treatments

were not significantly different from each other suggests that there is less selective utilization of

OM in response to NO3-. In the unamended treatment, SO4

2- reducers may only have access to a

limited supply of low-molecular-weight substrates (Canfield et al. 2005), therefore creating a

more recalcitrant OM pool over time (Fig. 1, 3) and a significant shift in overall chemical

composition (Fig. 7A). With the addition of NO3-, however, these data suggest that microbes are

51

accessing a wider range of compounds; thus, despite greater decomposition rates (Fig. 2), there

was less of a shift in the overall chemical composition of the remaining OM (Fig. 7A). Similar

results have been observed in both terrestrial and oceanic studies, with N addition resulting in the

selection for microbes that responded to the N supply and that could decompose recalcitrant

carbon compounds more efficiently (Campbell et al. 2010, Treseder et al. 2011, Allison et al.

2013). Another possible explanation for a more labile signature in the plus-NO3- treatment is a

greater supply of extracellular DNA from greater microbial biomass (e.g. Dell’Anno & Danavaro

2005), however since NH4+ production rates and 16S rRNA gene abundance were both similar

between treatments (Fig. 5), this is likely not the case.

Rather than acting as an electron acceptor, another consequence of NO3- addition could

be the release of the microbial community from nutrient-limitation, which may also result in

increased DIC production rates due to higher growth rates. However, most anaerobic microbes

are not nutrient-limited because their growth-per-unit substrate-intake is much lower than with

aerobic respiration (Canfield et al. 2005), which is why typical anoxic porewater nutrient

concentrations are higher than those found in oxic sediments. If NO3- addition were affecting

growth rates via alleviation of N limitation in the plus- NO3- treatment, I would also expect to see

a spike in microbial biomass. qPCR of the 16S rRNA gene (supplemental Fig. S3) suggests this

is not occurring, with no significant differences between the plus-NO3- and unamended

treatments. However, I was only able to sample sediments for qPCR at the beginning and end of

the experiment, so I cannot comment on any microbial population growth dynamics that may

have occurred throughout the experiment. Lastly, the majority of DIC produced in the plus-NO3-

treatment can be accounted for by NRR (86.7% ± 0.05, 98.3% ± 0.08, and 101.9% ± 0.09 for

52

shallow, mid, and deep sediments, respectively), suggesting that most of the consumed NO3- can

be attributed to dissimilatory processes.

4.3. NO3- addition effects on microbial community structure

I hypothesized that the end result of NO3- addition would be 1) to fundamentally alter the

resident microbial community through a change in the competitive landscape or 2) to alter the

function of the existing community through metabolic plasticity of the microbes present (Allison

& Martiny 2008), with both scenarios resulting in shifts in the dominant metabolic pathways.

Through 16S rRNA gene sequencing, I found evidence for a combination of the two. While I

observed a core microbiome that existed in both the plus-NO3- and unamended treatment (Fig.

S4), including microbial taxa typically present in these particular salt marsh sediments (e.g.

Kearns et al. 2016), I also found a significant shift in microbial community structure (Fig. 8A)

and decreases in alpha diversity (Fig. 8B) in response to NO3-. This suggests that NO3

- addition

selects for taxa that are more competitive in a high N environment, and that this community is

fundamentally different from both pre- and unamended sediments.

Through random forest classification analysis, I identified 30 ASVs most important in

correctly classifying between plus-NO3- and unamended treatments (Fig. 9; Table S2). Out of

these 30 ASVs, ~70% were from the class Gammaproteobacteria, a widely diverse group of

gram-negative bacteria that increase in abundance as a result of fertilization (e.g. Campbell et al.

2010). Many of these ASVs were putatively assigned to orders known to reduce nitrate

(Kiloniellales; Wiese et al. 2009) oxidize sulfur/sulfide (Thiotrichales, Chromatiales; Garrity et

al. 2005, Thomas et al. 2014), ferment OM (Ignavibacteriales, Rhodospirallales; (Biebl &

Pfening 1981, Iino et al. 2010), and degrade high-molecular-weight (HMW) compounds

(Flavobacteriales, Thiotrichales, Alteromonadales). Some members of these groups can also use

53

long-chain alkanes (Fernández-Gómez et al. 2013, Guibert et al. 2016) and are stimulated in the

presence of HMW dissolved organic matter (McCarren et al. 2010, Mahmoudi et al. 2015).

These shifts in the community provide evidence for selection of taxa more adept at using nitrate

or oxidizing more complex OM. In contrast, ASVs more abundant in the unamended treatment

included orders that are ubiquitous in soil and mangrove sediments (Verrucomicrobiae,

Caldithrixales; Miroshnichenko et al. 2010, Freitas et al. 2012), that can reduce sulfate

(Desulfobacterales, Desulfarculales; Bahr et al. 2005), and that exhibit properties associated with

iron metabolism (Campylobacterales, Rhizobiales (Eppinger et al. 2004, Reese et al. 2013).

While I cannot make definitive statements regarding the exact function associated with these

taxa, identifying the taxa most responsive to NO3- addition is a step forward in understanding the

mechanistic response of microbial communities to nutrient enrichment.

4.4. Assumptions and limitations of FTR experiments

I chose a high concentration of NO3- (500 µM NO3

- ) to assure non-limiting

concentrations at a reasonable flow rate (Pallud et al. 2007). I designed this experiment

specifically to assess the potential of NO3- to mobilize carbon pools that were not being oxidized

by SO42- reduction, rather than to simulate realistic environmental conditions. In the

environment, NO3- will almost always be limiting except in the most eutrophic conditions or in

situations of continuous replacement; therefore, I cannot extrapolate the rates of decomposition

observed in this experiment to field conditions. However, I can conclude from my data that

adding NO3- may stimulate decomposition of older, more recalcitrant sediment OM. There exist

scenarios where NRR may dominate anaerobic respiration, such as freshwater wetlands or

wastewater treatment plants, but it is much less likely to occur naturally in salt marshes where

NO3- is typically limiting and there is unlimited supply of sulfate from tidal water. This study

54

suggests that, by adding NO3-, NRR does not necessarily become the dominant process, but

instead allows for the decomposition of more recalcitrant sediment that could not be mobilized

under conditions of sulfate alone.

Further, the use of FTRs eliminates the complexity involved with plant-microbe

feedbacks and competition for NO3- by benthic microalgae and phytoplankton. While these

interactions are important, the aim of this experiment was to directly assess microbial processes.

I also assumed that in this experiment, SO42- was the only electron acceptor being supplied in the

unamended treatment aside from the very small background concentration of NO3- (0.6-1.2 µM)

in the seawater I used. I do not believe that this affected the treatment differences. Since

background SO42- concentrations are so high in seawater (~28 mM), I was not able to detect

small changes at the µM level that occurred in the FTRs and had to instead infer SRR from rates

of sulfide production and changes in sediment S concentrations. These changes are likely due to

pyrite or FeS formation; although I cannot rule out the production of organic sulfur (Luther et al.

1986). Although I did not monitor the influent oxygen concentrations, I conducted the entire

experiment in an anoxic glove chamber, so oxygen should not have been present for either oxic

respiration or nitrification. In both treatments, it is possible that iron and manganese oxides were

available as electron acceptors, especially in the shallow sediments, which could have

contributed to DIC production; however, the nitrate and DIC balance suggests this was not

important in the plus-NO3- treatment. The balance was not as close in the unamended treatment

but SRR was still the dominant process. Finally, since this experiment only lasted ~90 days, I

cannot determine how large the NO3--accessible OM pool is, whether NO3

- reducers are solely

responsible for the stimulation, or if they also stimulate SO42- reducers through co-metabolism.

55

Extrapolating to the ecosystem-level from small-scale laboratory experiments is

challenging; but these FTRs are specifically designed to isolate meaningful parameters and to

allow for the extraction of kinetic rate measurements of specific microbial processes, which can

then be used to inform predictive models designed for unraveling sediment biogeochemistry

across various spatial and temporal scales (see Algar & Vallino 2014, Vallino 2011).

4.5. Implications of N-loading on salt marsh carbon storage capacity

My results show that NO3- addition stimulates DIC production and consequently,

decomposition of OM in salt marsh sediments. I observed this response even in deep sediments,

where we traditionally assume OM to be fairly recalcitrant to microbial degradation. I

hypothesize that by adding NO3- and providing a more energetically favorable electron acceptor

to the system, I am shifting the microbial community towards taxa better suited for a high NO3-

environment, and consequently changing the accessible OM pool from one that is stable and

recalcitrant to SO42- reducers, to one that is bioavailable under high NO3

- conditions. These

results suggest that comparable additions of NO3- to salt marshes could also enhance OM

decomposition in situ.

These results could have important implications for salt marsh carbon storage potential.

The effect of adding NO3- that I demonstrate here, would depend on the specific hydrology of the

marsh system. If NO3- -rich flooding waters penetrate into deep sediments, it could accelerate

decomposition of stored carbon. Not only could this decrease carbon storage potential, it could

also result in decreased belowground marsh stability (e.g. Deegan et al. 2012) and lead to greater

CO2 production. Additionally, marsh systems currently experiencing high NO3- conditions may

store less OM over time, leading to less overall carbon storage; although the OM that is buried

may be more recalcitrant, since a larger portion will already be oxidized. What this means for

56

carbon storage potential of marshes at a larger scale is unclear, since NO3- can also stimulate OM

production by acting as a nutrient, with such production offsetting respiration. Total marsh

carbon storage capacity depends heavily on the balance between these two processes. Lastly, by

stimulating N-cycling processes, we may also increase the potential for nitrous oxide (N2O)

production, a greenhouse gas with 263 times the global warming potential of CO2 (Neubauer &

Megonigal 2015), as a result of incomplete denitrification; although I did not measure N2O

fluxes in this study. Considering the degree of eutrophication in US estuaries (Bricker et al.

2008), and how NO3- addition alters processes that control OM, it is important to incorporate our

understanding of these processes when assessing the resilience of salt marsh systems to changing

climate and increasing anthropogenic pressures. This is especially critical if we hope to rely on

salt marshes for long-term carbon storage.

Acknowledgements

I would like to thank Joseph Vallino at Marine Biological Laboratory for his invaluable

contribution to the design of the flow through reactor system. I also thank researchers of the

TIDE project (NSF OCE0924287, OCE0923689, DEB0213767, DEB1354494, and OCE

1353140) for maintenance of the long-term nutrient enrichment experiment, as well as

researchers of the Plum Island Ecosystems LTER (NSF OCE 0423565, 1058747, 1637630). I

would also like to acknowledge Sam Kelsey, Khang Tran, Michael Greenwood, and members of

the Bowen lab for their assistance in the field and laboratory, as well as Inke Forbrich, Nat

Weston, and Gary Banta for their thoughtful comments on this research. This work was funded

by an NSF CAREER Award to JLB (DEB1350491) and a Woods Hole Oceanographic Sea

Grant award to AEG and JJV (Project No. NA140AR4170074 Project R/M-65s). Additional

57

support was provided by a Ford Foundation pre-doctoral fellowship award to ABM. The views

expressed here are those of the authors and do not necessarily reflect the views of NOAA or any

of its sub-agencies.

58

References

Achtnich C, Bak F, Conrad R (1995) Competition for electron donors among nitrate reducers,

ferric iron reducers, sulfate reducers, and methanogens in anoxic paddy soil. Biol Fertil soils

19:65–72

Algar CK, Vallino JJ (2014) Predicting microbial nitrate reduction pathways in coastal

sediments. Aquat Microb Ecol 71:223–238

Aller RC, Aller JY (1998) The effect of biogenic irrigation intensity and solute exchange on

diagenetic reaction rates in marine sediments. J Mar Res 56:905–936

Allison SD, Lu Y, Weihe C, Goulden ML, Adam C, Treseder KK, Martiny JBH, Allison SD, Lu

Y, Weihe C, Goulden ML, Martiny AC, Treseder KK, Martiny JBH (2013) Microbial

abundance and composition influence litter decomposition response to environmental

change. Ecology 94:714–725

Allison SD, Martiny JBH (2008) Resistance, resilience, and redundancy in microbial

communities. Proc Natl Acad Sci USA 105:11512–11519




Apprill A, Mcnally S, Parsons R, Weber L (2015) Minor revision to V4 region SSU rRNA 806R

gene primer greatly increases detection of SAR11 bacterioplankton. Aquat Microb Ecol

75:129–137

Arndt S, Jørgensen BB, LaRowe DE, Middelburg JJ, Pancost RD, Regnier P (2013) Quantifying

the degradation of organic matter in marine sediments: A review and synthesis. Earth-

Science Rev 123:53–86

Bahr M, Crump BC, Klepac-Ceraj V, Teske A, Sogin ML, Hobbie JE (2005) Molecular

characterization of sulfate-reducing bacteria in a New England salt marsh. Environ

Microbiol 7:1175–85

Beauchamp E, Trevors J, JW P (1989) Carbon sources for bacterial denitrification. Adv Soil Sci

10:113–142




Biebl H, Pfening N (1981) Isolation of members of the family Rhodospirillaceae. In: Starr M,

Stolp H, Truper H, Balows A, Schlegel H (eds) The Prokaryotes. Springer Berlin

Heidelberg, Berlin, p 267–268

59

Bricker SB, Longstaff B, Dennison W, Jones A, Boicourt K, Wicks C, Woerner J (2008) Effects

of nutrient enrichment in the nation’s estuaries: A decade of change. Harmful Algae 8:21–

32

Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP (2016) DADA2:

High-resolution sample inference from Illumina amplicon data. Nat Methods 13:581–583

Campbell BJ, Polson SW, Hanson TE, Mack MC, Schuur EAG (2010) The effect of nutrient

deposition on bacterial communities in Arctic tundra soil. Environ Microbiol 12:1842–1854

Canfield D (1989) Sulfate reduction and oxic respiration in marine sediments: implications for

organic carbon preservation in euxinic environments. Deep Sea Res Part A Oceanogr Res

Pap 36:121–138

Canfield DE, Farquhar J (2009) Animal evolution , bioturbation , and the sulfate concentration of

the oceans. Proc Natl Acad Sci USA 106:8123-8127


Press, Boston, MA

Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Peña

AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST, Knights D, Koenig JE, Ley RE,

Lozupone CA, Mcdonald D, Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ,

Walters WA, Widmann J, Yatsunenko T, Zaneveld J, Knight R (2010) QIIME allows

analysis of high- throughput community sequencing data. Nat Meth 7:335–336

Caporaso JG, Lauber CL, Walters W a, Berg-Lyons D, Huntley J, Fierer N, Owens SM, Betley J,

Fraser L, Bauer M, Gormley N, Gilbert JA, Smith G, Knight R (2012) Ultra-high-

throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms.

ISME J 6:1621–1624

Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Lozupone CA, Turnbaugh PJ, Fierer N,

Knight R (2011) Global patterns of 16S rRNA diversity at a depth of millions of sequences

per sample. Proc Natl Acad Sci USA 108:4516-4522

Chmura GL, Anisfeld SC, Cahoon DR, Lynch JC (2003) Global carbon sequestration in tidal,

saline wetland soils. Global Biogeochem Cy 17:1-12

Cox RD (1980). Determination of nitrate and nitrite at the parts per billion level by

chemiluminescence. Anal Chem 52:332-335

Cowie GL, Hedges JI (1994) Biochemical indicators of diagenetic alteration in natural organic

matter mixtures. Nature 369:304-307

Dalsgaard T, Thamdrup B, Canfield DE (2005) Anaerobic ammonium oxidation (anammox) in

60

the marine environment. Res Microbiol 156:457–64

Dargusch P, Thomas S (2012) A critical role for carbon offsets. Nat Clim Chang 2:470


C (2007) Susceptibility of salt marshes to nutrient enrichment and predation removal. Ecol

Appl 17:42–63



Dell’Anno A, Danavaro R (2005) Extracellular DNA plays a key role in deep-sea ecosystem

functioning. Science 309:2179

Dickson A, Goyet C (1994) Handbook of methods for the analysis of the various parameters of

the carbon dioxide system in sea water (Version 2). DOE (US Department of Energy)

ORNL/CDIAC-74. Carbon Dioxide Information and Analysis Center, Oak Ridge, TN

Ding G, Novak JM, Amarasiriwardena D, Hunt PG, Xing B (2002) Soil organic matter

characteristics as affected by tillage management. Soil Sci Soc Am J 66:421-429

Duarte CM, Middelburg JJ, Caraco N, Major NC (2005) Major role of marine vegetation on the

oceanic carbon cycle. Biogeosciences 1:659-679

Eppinger M, Baar C, Raddatz G, Huson DH, Schuster SC (2004) Comparative analysis of four

campylobacterales. Nat Rev Microbiol 2:872–885



Fernández-Gómez B, Richter M, Schüler M, Pinhassi J, Acinas SG, González JM, Pedrós-Alió C

(2013) Ecology of marine bacteroidetes: A comparative genomics approach. ISME J

7:1026–1037

Forbrich I, Giblin AE, Hopkinson CS (2018) Constraining marsh carbon budgets using long-term

C burial and contemporary atmospheric CO2 fluxes. J Geophys Res Biogeosci 123: 867-

878

Freitas S, Hatosy S, Fuhrman JA, Huse SM, Mark Welch DB, Sogin ML, Martiny AC (2012)

Global distribution and diversity of marine Verrucomicrobia. ISME J 6:1499–1505





61

questions, and potential solutions. Science 320:889–892

Garrity G, Bell J, Lilburn T (2005) Thiotrichales ord. nov. In: Brenner D (ed) Bergey’s Manual

of Systematic Bacteriology, 2nd edition. Boston, MA

Giblin A, Tobias C, Song B, Weston N, Banta G, Rivera-Monroy V (2013) The Importance of



Gilboa-Garber (1971) Direct spectrophotometric determination of inorganic sulfide in biological

materials and in other complex mixtures. Anal Biochem 43:129–133

Giovannoni SJ (2017) SAR11 Bacteria: The Most Abundant Plankton in the Oceans. Ann Rev

Mar Sci 9:231–255

Guibert LM, Loviso CL, Borglin S, Jansson JK, Dionisi HM, Lozada M (2016) Diverse bacterial

groups contribute to the alkane degradation potential of chronically polluted Subantarctic

coastal sediments. Microb Ecol 71:100–112



Hopkinson CS, Giblin AE (2008) Nitrogen dynamics of coastal salt marshes. In: Capone D,

Bronk D, Mulholland M, Carpenter E (eds) Nitrogen in the Marine Environment, 2nd

edition. Academic Press, Burlington, MA, p 991–1036

Howarth RW (1984) The Ecological significance of sulfur in the energy dynamics of salt marsh

and coastal marine sediments. Biogeochemistry 1:5–27


24:999–1013

Howes BL, Dacey JWH, Teal JM (1985) Annual carbon mineralization and belowground

production of Spartina alterniflora in a New England salt marsh. Ecology 66:595–605

Howes BL, Howarth RW, Teal JM (1981) Oxidation-reduction potentials in a salt marsh: Spatial

patterns and interactions with primary production. Limnol Oceanogr 26:350-360

Iino T, Mori K, Uchino Y, Nakagawa T, Harayama S, Suzuki KI (2010) Ignavibacterium album

gen. nov., sp. nov., a moderately thermophilic anaerobic bacterium isolated from microbial

mats at a terrestrial hot spring and proposal of Ignavibacteria classis nov., for a novel

lineage at the periphery of green sulfur bacteria. Int J Syst Evol Microbiol 60:1376–1382

Imhoff J (2005) Bergey’s Manual of Systematic Bacteriology. In: Brenner D, Krieg N, Staley J,

Garrity G (eds) Bergey’s Manual of Systematic Bacteriology, 2nd edition. New York, NY

62

Jorgensen BB (1977) The sulfur cycle of a coastal marine sediment (Limfjorden, Denmark).



24:726–734

Katoh K, Standley DM (2013) MAFFT multiple sequence alignment software version 7:

Improvements in performance and usability. Mol Biol Evol 30:772–780

Kearns PJ, Angell JH, Howard E, Deegan LA, Stanley RH, Bowen JL (2016) Nutrient

enrichment induces high rates of dormancy and decreases diversity of active bacterial taxa.

Nat Commun 7:12881

Koop-Jakobsen K, Giblin AE (2009) Anammox in tidal marsh sediments: The role of salinity,

nitrogen loading, and marsh vegetation. Estuar Coasts 32:238–245



Kostka JE, Gribsholt B, Petrie E, Dalton D, Skelton H, Kristensen E (2002) The rates and

pathways of carbon oxidation in bioturbated saltmarsh sediments. Limnol Oceanogr

47:230–240

Kuhn M (2015) A Short introduction to the caret package. R Found Stat Comput:1–10



19:1495–1503

Liaw A, Wiener M (2002) Classification and Regression by randomForest. R news 2:18–22

Lozupone C, Lladser ME, Knights D, Stombaugh J, Knight R (2011) UniFrac: An effective

distance metric for microbial community comparison. ISME J 5:169–172

Luther G, Church T, Scudlark J, Cosman M (1986) Inorganic and Ooganic sulfur cycling in salt-

marsh pore waters. Science 232:746–749




Mahmoudi N, Robeson MS, Castro HF, Fortney JL, Techtmann SM, Joyner DC, Paradis CJ,

Pfiffner SM, Hazen TC (2015) Microbial community composition and diversity in Caspian

Sea sediments. FEMS Microbiol Ecol 91:1–11

Margenot AJ, Calderón FJ, Bowles TM, Parikh SJ, Jackson LE (2015) Soil organic matter

63

functional group composition in relation to organic carbon, nitrogen, and phosphorus

fractions in organically managed tomato fields. Soil Sci Soc Am J 79:772–782

McCarren J, Becker JW, Repeta DJ, Shi Y, Young CR, Malmstrom RR, Chisholm SW, DeLong

EF (2010) Microbial community transcriptomes reveal microbes and metabolic pathways

associated with dissolved organic matter turnover in the sea. Proc Natl Acad Sci USA

107:16420–16427

McDonald D, Price MN, Goodrich J, Nawrocki EP, Desantis TZ, Probst A, Andersen GL,

Knight R, Hugenholtz P (2012) An improved Greengenes taxonomy with explicit ranks for

ecological and evolutionary analyses of bacteria and archaea. ISME J 6:610–618


WH, Silliman BR (2011) A blueprint for blue carbon: toward an improved understanding of

the role of vegetated coastal habitats in sequestering CO2. Front Ecol Environ 9:552–560

Meyer AF, Lipson DA, Martín AP, Schadt CW, Schmidt SK (2004) Molecular and metabolic

characterization of cold-tolerant alpine soil Pseudomonas sensu stricto. Appl Environ


Miroshnichenko ML, Kolganova T V., Spring S, Chernyh N, Bonch-Osmolovskaya EA (2010)

Caldithrix palaeochoryensis sp. nov., a thermophilic, anaerobic, chemo-organotrophic

bacterium from a geothermally heated sediment, and emended description of the genus

Caldithrix. Int J Syst Evol Microbiol 60:2120–2123

Morris JT, Sundareshwar PV, Nietch CT, Kjerfve B, Cahoon DR (2002) Responses of coastal

wetlands to rising sea level. Ecology 83:2869–2877

Negrin V, Spetter C, Asteasuain R, Perillo G, Marcovecchio J (2011) Influence of flooding and

vegetation on carbon, nitrogen, and phosphorus dynamics in the pore awter of a Spartina

alterniflora salt marsh. J Environ Sci 23:212–221

Nelleman C, Corcoran E, Duarte C, Valdes L, Young C De, Foncesa L, Grimsditch G (2009)

Blue carbon. A rapid response assessment. United Nations Environment Programme.

Birkelant: GRID-Arendal

Neubauer SC, Megonigal JP (2015) Moving beyond global warming potentials to quantify the

climatic role of ecosystems. Ecosystems 18:1000–1013




sediment reactors in biogeochemical kinetics: Methodology and examples of applications.

Mar Chem 106:256–271

Parikh SJ, Goyne KW, Margenot AJ, Mukome FND, Calderon FJ (2014) Soil chemical insights

64

provided through vibrational spectroscopy. Adv Agron 126:1-148

Pastore MA, Megonigal JP, Langley JA (2017) Elevated CO2 and nitrogen addition accelerate

net carbon gain in a brackish marsh. Biogeochemistry 133:73–87

Peng X, Ji Q, Angell JH, Kearns PJ, Yang HJ, Bowen JL, Ward BB (2016) Long term

fertilization alters the relative importance of nitrate reduction pathways in salt marsh

sediments. J Geophys Res Biogeosci 121: 2082-2095

QIIME 2 Development Team (2018) QIIME 2. https://qiime2.org/

Reddy K, Patrick Jr. W (1975) Effect of alternate aerobic and anaerobic conditions on redox

potential, organic matter decomposition, and nitrogen loss in a flooded soil. Soil Biol

Biochem 7:87–94

Reese BK, Witmer AD, Moller S, Morse JW, Mills HJ (2013) Molecular assays advance

understanding of sulfate reduction despite cryptic cycles. Biogeochemistry 118:307–319

Rysgaard S, Risgaard-Peterson N, Sloth N (1996) Nitrification, denitrification, and nitrate

ammonification in sediments of two coastal lagoons in Southern France. Hydrobiologia

117:133–141



Schmid MC, Risgaard-Petersen N, Vossenberg J Van De, Kuypers MMM, Lavik G, Petersen J,

Hulth S, Thamdrup B, Canfield D, Dalsgaard T, Rysgaard S, Sejr MK, Strous M, Op Den

Camp HJM, Jetten MSM (2007) Anaerobic ammonium-oxidizing bacteria in marine

environments: Widespread occurrence but low diversity. Environ Microbiol 9:1476–1484

Shade A, Peter H, Allison SD, Baho DL, Berga M, Bürgmann H, Huber DH, Langenheder S,

Lennon JT, Martiny JBH, Matulich KL, Schmidt TM, Handelsman J (2012) Fundamentals

of microbial community resistance and resilience. Front Microbiol 3:1–19

Solórzano L (1968) Determination of ammonia in natural waters by the phenolhypochlorite

method. Limnol Oceanogr 14:799-801

Thamdrup B, Dalsgaard T (2002) Production of N2 through anaerobic ammonium oxidation

coupled to nitrate reduction in marine sediments. Appl Environ Microbiol 68:1312–1318

Thomas F, Giblin AE, Cardon ZG, Sievert SM (2014) Rhizosphere heterogeneity shapes

abundance and activity of sulfur-oxidizing bacteria in vegetated salt marsh sediments. Front

Microbiol 5:309

Treseder KK, Kivlin SN, Hawkes C V. (2011) Evolutionary trade-offs among decomposers

determine responses to nitrogen enrichment. Ecol Lett 14:933–938

65

Turner S, Pryer KM, Miao VPW, Palmer JD (1999) Investigating deep phylogenetic

relationships among cyanobacteria and plastids by small subunit rRNA sequence

analysis. J Eukaryotic Micro 46:327-338

Valiela I, Cole ML (2002) Comparative evidence that salt marshes and mangroves may protect

seagrass meadows from land-derived nitrogen loads. Ecosystems 5:92–102

Valiela I, Teal JM (1974) Nutrient limitation in salt marsh vegetation. In: Reimold RJ, Queen

WH (eds) Ecology of Halophytes. Academic Press, New York, NY, p 547–563

Valiela I, Teal JM, Persson NY (1976) Production and dynamics of experimentally enriched salt

marsh vegetation: Belowground biomass. Limnol Oceanogr 21:245-252

Vallino JJ (2011) Differences and implications in biogeochemistry from maximizing entropy

production locally versus globally. Earth Syst Dyn 2:69–85

Veum KS, Goyne KW, Kremer RJ, Miles RJ, Sudduth KA (2014) Biological indicators of soil

quality and soil organic matter characteristics in an agricultural management continuum.

Biogeochemistry 117:81–99

Vivanco L, Irvine IC, Martiny JBH, Vivanco L, Irvine I, Martiny JBH (2018) Nonlinear

responses in salt marsh functioning to increased nitrogen addition. Ecology 96:936–947

Wakeham SG, Lee C, Hedges JI, Hernes PJ, Peterson ML (1997) Molecular indicators of

diagenetic status in marine organic matter. Geochim Cosmochim Acta 61:5363-5369


(2002) Salt marsh restoration in Connecticut: 20 years of science and management. Restor

Ecol 10:497–513

Westrich JT, Berner R a. (1984) The role of sedimentary organic matter in bacterial sulfate

reduction: The G model tested. Limnol Oceanogr 29:236–249

Wiese J, Thiel V, Gärtner A, Schmaljohann R, Imhoff JF (2009) Kiloniella laminariae gen. nov.,

sp. nov., an alphaproteobacterium from the marine macroalga Laminaria saccharina. Int J

Syst Evol Microbiol 59:350–356

Zehnder A, Stumm W (1988) Geochemistry and biogeochemistry of anaerobic habitats. In:

Zehnder A (ed) Biology of Anaerobic Microorganisms. Wiley, p 1–38

66

Tables

Table 1. Functional group assignments based on Parikh et al. (2014) and modified from

Margenot et al. (2015) to evaluate FT-IR spectra using Index II metric. ν = stretching vibration;

νas = asymmetric stretching vibration; νs = symmetric stretching vibration; δ = bending vibration.

Band (cm-1) Assignment

3400 ν(N-H), ν(O-H)

2924 aliphatic νas(C-H)

2850 aliphatic νs(C-H)

1650 aromatic ν(C = C)

1470 aliphatic δ(C-H)


1270 phenol νas(C-O), carboxylic acid ν(C-O)

1110 polysaccharide νs(C-O)


920 aromatic δ(C-H)

840 aromatic δ(C-H), less substituted

68

Fig. 1. Average (±SE) dissolved inorganic carbon (DIC) production over time (days) across three

depths that correspond to different ages of marsh organic matter (panels A-C; n = 3).

69

Fig. 2. Average (± SE) cumulative dissolved inorganic carbon (DIC) production in µmol cm-3 for

nitrate and unamended treatments at each depth. Boxes represent 25% to 75% quartiles. The

solid black line is the median value, and the whiskers are upper and lower extremes. Black dots

represent values for each individual reactor. A Two-way ANOVA indicates a significant effect

of treatment (p<0.001, F1,14=21.73) and depth (p<0.001, F2,14=48.33) on total DIC production,

but there was no significant interaction between the two. Letters represent statistically different

DIC production by depth from a Tukey’s HSD test corrected for multiple comparisons test and

asterisks indicate a significant difference between treatments.

70

Fig. 3. (A) Average (±SE) nitrate reduction rates over time (days) and (B) total nitrate reduction

at each depth in the nitrate amended treatment (nitrate was below detection in the unamended

sediments). One-way ANOVA indicated no significant difference in nitrate reduction as a

function of depth.

71

Fig. 4. (A) Average (±SE) sulfide production rates over time (days) and (B) total sulfide

production. One-way ANOVA indicated shallow sediments exhibited significantly greater

sulfide production than mid and deep sediments (p=0.0124, F2,6=9.96). (C) A two-way ANOVA

indicated that total sulfur storage was greater in the unamended treatment across all depths (as

indicated by an asterisk; p=0.0243, F1,14=7.637) and lowest in the deep sediments (p=0.0369,

F2,14=4.214). (D) There was a significant interaction between treatment and depth on total sulfate

reduction (sulfide + sulfur production) (p=0.029, F2,12=4.806). Letters represent statistically

different sulfate reduction from a Tukey’s HSD test corrected for multiple comparisons.

72

Fig. 5. Average (±SE) ammonium production rates over time (days) in the nitrate (A) and

unamended (C) treatments, and total ammonium production in µmol cm-3 across depth for nitrate

(B) and unamended (D) treatments. While there was no effect of treatment, a one-way ANOVA

indicated a significant effect of depth (nitrate: p<0.001, F2,6=37.47; unamended: p=0.005,

F2,6=14.09), as indicated by a Tukey’s HSD test corrected for multiple comparisons.

73

Fig. 6. The ratio of DIC to ammonium production calculated per core was greater in the plus-

NO3- treatment when compared to unamended sediments, while depth was insignificant,

according to a two-way ANOVA (p=0.007, F1,14=10.11). The dotted line indicates the average

C:N ratio of sediments from this experiment (13.66 ± 0.69).

74

Fig. 7. (A) Principal coordinates analysis (PCoA) of Fourier Transform-Infrared Spectra (FT-IR)

indicates significant differences by treatment (PERMANOVA; p=0.017, F2,18 = 3.314) and depth

(p=0.001 F2,18 = 16.598). (B) Pearson’s correlation coefficients plotted against wavenumber

representing regions most discriminating across two axes shown in A. Dotted lines indicate

functional group assignments listed in Table 1, with 840-920 and 1650 cm-1= aromatic carbon

and lignin-type signatures, 1080 cm-1 = polysaccharides, and 1470 and 2850-2924 cm-1 =

aliphatic carbon. (C) A two-way ANOVA of Index II values (eq. 1) indicated a significant effect

of treatment but not depth, with a higher recalcitrance index in plus-NO3- and unamended

treatments when compared to initial sediments.

75

Fig. 8. (A) Principal coordinates analysis constructed based on weighted UniFrac for pre-

experiment (green), nitrate (yellow), and unamended (blue) sediments. Shape indicates sample

depth: shallow (circle), mid (triangle), and deep (square). Results from a PERMANOVA indicate

significant differences in community composition by treatment (p=0.001, F(2,22)=11.1095) and

by depth (p=0.006, F(2,22)=3.0287). (B) Shannon diversity index. A two-way ANOVA revealed a

significant effect of both treatment (p<0.001, F(2,22)=71.207) and depth (p=0.044, F(2,22)=3.613),

as indicated by a Tukey’s HSD test corrected for multiple comparisons, but no effect of the

interaction between the two.

76

Fig. 9. Heatmap showing relative abundance of top 30 ASVs (45.2% of sequences) most

important in correctly discriminating between plus-NO3- (top 9 rows) and unamended treatments

(bottom 9 rows) according to a random forest classification model. Lighter colors indicate less

abundant taxa, while darker colors indicate more abundant taxa. Colored circles represent the

taxonomic class of each ASV. Additional taxonomic information can be found in Supplemental

Table S2.

77

Supplemental Methods

I used bromide (Br-) as a conservative tracer to confirm ideal flow conditions and

constrain transport parameters by measuring outflow concentrations as a function of time (Pallud

et al., 2007; Roychoudhury et al., 1998). I added 2 mM sodium bromide stock (NaBr-; Acros

organics) to each reservoir to increase the final concentration of Br- by 2 mM above the Br-

concentration in the reservoir seawater. I began sampling the effluent at approximately 2-hour

intervals for a 30-hour period. To analyze Br- concentrations in seawater, I followed methods

outlined in Presley (1971). Briefly, to a 500 µL sample, I added, in sequence, 5 mL phenol-red

buffer (500 mL sodium acetate buffer: 30 g sodium acetate (Fisher Scientific) + 7 mL glacial

acetic acid in 1000 mL DI water; 25 mL phenol red reagent; 0.08 g phenol sulfonephthalein

(Sigma-Aldrich) in 10 mL 0.1 N sodium hydroxide (Fisher Scientific) diluted in 500 mL DI

water), 1 ml 0.005 N N-chlorotosylamide (Sigma-Aldrich) for 30 seconds, and 2.5 mL 0.05 N

sodium thiosulfate (Sigma-Aldrich) for 5 seconds. I then immediately read absorbance at 595 nm

on a Shimadzu 1601 spectrophotometer.

To carefully measure flow conditions in the reactors, I examined the time at which the

initial bromide concentration was detected in the outflow (Fig. S2). I determined the linear flow

velocity (µ) using the following equation from Roychoudhury et al. (1998):

Eq. 1: μ = 𝑄

𝜑𝐴

Where Q is the measured flow rate in units of volume of solution per unit of time, A is the cross-

sectional area of the flow through reactor, and φ is the sediment porosity. I also calculated

residence time in the reactor with the following equation:

78

Eq. 2: 𝜏 = 𝜑𝑉

𝑄

where the function τ is the water residence time, and V is the reactor volume (31.81 cm3). I then

plotted the ratio between initial and final bromide concentrations over time (Fig. S2) and divided

the total number of hours until breakthrough by the residence time to attain the number of

porewater volume replacements required before reaching steady state (Roychoudhury et al.

1998). I performed all calculations in R (R Core Team).

References


sediment reactors in biogeochemical kinetics: Methodology and examples of applications.

Mar Chem 106:256–271



Presley BJ (1971) Techniques for analyzing interstitial water samples. Part 1: Determination of

selected minor and major inorganic constituents. Initial Report. Institute of Geophysics &

Planetary Physics 869:1749-1755

Roychoudhury AN, Viollier E, Cappellen P Van (1998) A plug flow-through reactor for studying

biogeochemical reactions in undisturbed aquatic sediments. Appl Geochemistry 13:269–280

79

Supplemental Tables

Table S1. Flow property information, including average flow rate (±SD), linear flow velocity as

calculated by Eq. 1, porosity, and residence time as calculated by eq. 2.

Sample Flow rate

(mL min-1)

Linear flow velocity

(mL hr-1 cm-2) Porosity

Residence

time (hr)

Plus-NO3- Treatment

Shallow

Core 1 0.067 (0.016) 0.422 0.60 4.74

Core 2 0.067 (0.016) 0.422 0.64 4.73

Core 3 0.072 (0.017) 0.412 0.64 4.85

Mid

Core 1 0.076 (0.015) 0.437 0.66 4.57

Core 2 0.068 (0.001) 0.414 0.62 4.83

Core 3 0.071 (0.012) 0.387 0.69 5.17

Deep

Core 1 0.077 (0.010) 0.448 0.64 4.46

Core 2 0.067 (0.010) 0.398 0.63 5.03

Core 3 0.069 (0.022) 0.400 0.65 5.00

Unamended Treatment

Shallow

Core 1 0.080 (0.021) 0.474 0.64 4.22

Core 2 0.080 (0.025) 0.462 0.66 4.32

Core 3 0.081 (0.025) 0.453 0.67 4.42

Mid

Core 1 0.082 (0.012) 0.451 0.69 4.43

Core 2 0.077 (0.013) 0.440 0.66 4.54

Core 3 0.079 (0.014) 0.453 0.66 4.42

Deep

Core 1 0.081 (0.013) 0.450 0.68 4.44

Core 2 0.066 (0.024) 0.397 0.62 5.04

Core 3 0.070 (0.037) 0.400 0.66 5.00

80

Table S2. Taxonomic information of 30 ASVs present at least 100 times that were most

important in discriminating between treatments in order of mean decreasing accuracy according

to Random Forest. I also reran the model including all taxa (excluding singletons) and obtained

the same classification result. Discriminating taxa that were rare and not included in the initial

analysis are listed as ASV 31-41 in the table and account for an additional 1.9% of sequences.

ASV Phylum Class Order Family

1 Bacteroidetes Flavobacteria Flavobacteriales NA

2 Proteobacteria Gammaproteobacteria Chromatiales NA

3 Proteobacteria Gammaproteobacteria Alteromonadales Colwelliaceae

4 Proteobacteria Gammaproteobacteria NA NA

5 Proteobacteria Alphaproteobacteria Rhodospirillales Rhodospirillaceae

6 Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobacteraceae

7 Bacteroidetes Bacteroidia Bacteroidales NA

8 WWE1 Cloacamonae Cloacamonales MSBL8

9 Proteobacteria Gammaproteobacteria Alteromonadales Alteromonadaceae

10 Caldithrix Caldithrixae Caldithrixales Caldithrixaceae

11 Proteobacteria Alphaproteobacteria Kiloniellales Kiloniellaceae

12 Proteobacteria Gammaproteobacteria Oceanospirillales Oceanospirillaceae

13 Proteobacteria Gammaproteobacteria Oceanospirillales NA

14 Chloroflexi Anaerolineae SBR1031 A4b

15 Planctomycetes Phycisphaerae MSBL9 NA

16 Chloroflexi Anaerolineae OPB11 NA

17 Proteobacteria Deltaproteobacteria Desulfarculales Desulfarculaceae

18 Proteobacteria Epsilonproteobacteria Campylobacterales NA

19 Acidobacteria BPC102 MVS-40 NA

20 Chloroflexic Anaerolineae SBR1031 NA

21 Proteobacteria Deltaproteobacteria IndB3-24 NA

22 Proteobacteria Gammaproteobacteria Thiotrichales NA

23 Gemmatimonadetes Gemm-2 NA NA

24 Chloroflexi Anaerolineae Anaerolineales Anaerolinaceae

25 Proteobacteria Alphaproteobacteria Rhizobiales NA

26 Proteobacteria Gammaproteobacteria Thiotrichales Thiotrichaceae

27 Verrucomicrobia Verrucomicrobiae Verrucomicrobiales Verrucomicrobiaceae

28 Chlorobi Ignavibacteria Ignavibacteriales lheB3-7

29 Proteobacteria Betaproteobacteria NA NA

30 Proteobacteria Gammaproteobacteria NA NA

31 Chloroflexi Anaerolineae O4D2Z37 NA

32 Chlorobi NA NA NA

33 Gemmatimonadetes Gemm-2 NA NA

34 WWE1 Cloacamonae Cloacamonales MSBL8

35 KSB3 NA NA NA

36 OP8 OP81 HMMVPog-54 NA

37 Bacteroidetes Flavobacteriia Flavobacteriales Cryomorphaceae

81

38 Chlorobi Ignavibacteria Ignavibacteriales IheB3-7

39 Acidobacteria BPC102 MVS-40 NA

40 Spirochaetes Leptospirae Leptocpirales Sediment-4

41 WWE1 Cloacamonae Cloacamonales SHA-116

82

Supplemental Figures

Fig. S1. Flow through reactor schematic modeled after Pallud et al. (2006; 2007). (A) There are

three individual reactors, each with independent in- and outflows, per block. (B) Each reactor is

radially scored to promote regular, unilateral flow, and encased with a 0.45 µM filter on either

side. (C) These reactors are held together by two Plexiglas® plates and secured with 8 screws

positioned along plate edges. All indicated measurements are in centimeters.

83

Fig. S2. Bromide breakthrough curve with the ratio of [Br-]initial/[Br-]final on the y-axis for plus-

NO3- (n = 9) and unamended (n = 9) treatments confirmed uniform and regular flow in each

reactor. Dotted line indicates a ratio of 1, where the initial and final bromide concentration are

equal, indicating breakthrough. Average breakthrough time was approximately 10 hours at a

targeted flow rate of 0.08 mL min-1 with 2.14 ± 0.13 pore volume replacements required before

reaching steady state.

84

Fig. S3. A boxplot of 16S rRNA gene abundance showed no difference by treatment (p=0.385)

or site (p=0.233) according to a two-way ANOVA.

85

Fig. S4. Stacked bar graph showing relative abundance of 20 bacterial orders present in all

samples, contributing to 45.94% of total sequences.

86

Chapter 2: Chronic exposure to nutrient enrichment lessens the effect of additional nitrate

on organic matter decomposition despite changes to microbial community structure and

activity

In collaboration with: Anna E. Murphy, Anne E. Giblin, Jane Tucker, and Jonathan Sanderman

Abstract

Salt marshes can sequester carbon at rates that are an order of magnitude greater than

terrestrial forests, but this ecosystem service is under threat from nutrient enrichment, which can

stimulate decomposition of organic matter. Despite efforts to mitigate nitrogen loading, salt

marshes continue to experience chronic enrichment, the consequences of which remain unclear.

To investigate the effect of chronic nutrient exposure on salt marsh organic matter

decomposition, I collected sediments from three sites across a range of prior nitrate exposure: a

relatively pristine marsh, a marsh enriched at ~70 µmol L-1 for 13 years, and a marsh enriched

between 100-1000 µmol L-1 for 40 years. The most chronically enriched site contained more

recalcitrant organic matter and a diverse assemblage of microbial taxa associated with the

nitrogen cycle. I also performed a controlled enrichment experiment to test whether these site

differences influenced the functional response to additional nitrate exposure. I found significant

changes to microbial community structure and activity, with the chronically enriched site having

lower rates of microbial respiration and nitrate reduction. These results suggest that long term

nutrient enrichment could lead to less carbon storage overall, but the portion of organic matter

that is buried is less vulnerable to decomposition in response to further nitrate addition. Thus, it

is important to consider the extent of nutrient enrichment when developing strategies to protect

and restore salt marshes for their carbon storage potential.

Introduction

87

The extent to which salt marshes store carbon (C), one of the many valuable ecosystem

services provided by these habitats, depends largely on the balance between inputs from primary

production and losses from organic matter (OM) decomposition, with the latter ultimately

determining how much C is buried over long time-scales (Burdige 2007). While several studies

document decomposition rates in response to shifting environmental conditions, such as

temperature (Conant et al. 2011, Kirschbaum 1995, Lehmann & Kleber 2015) and sea level rise

(Kirwan et al. 2013), there is uncertainty about the effect that nitrogen (N) has on the balance

between primary production and decomposition. Over the last few centuries, coastal habitats

have experienced increased N loading due to runoff from sewage disposal, urban and agricultural

activities, and atmospheric N deposition (Galloway et al. 2017). Tremendous efforts at

mitigating N loading to coastal waters have improved conditions in some regions (Zedler 2000,

Warren et al. 2002), however, worldwide, many salt marshes continue to experience chronic

nutrient enrichment (Bricker et al. 2008, Deegan et al. 2012). The implications of this nutrient

enrichment for C storage in salt marshes remain unclear, with some studies documenting an

increase in primary production (Valiela et al. 1975, Morris et al. 1991, Langley et al. 2013,

Morris et al. 2013), and others indicating decreased belowground biomass (Langley et al. 2009,

Watson et al. 2014), increased respiration of OM (Wigand et al. 2009, Watson et al. 2014), and

decreased sediment stability (Deegan et al. 2012, Mueller et al. 2018).

The primary form of N enrichment in coastal waters is in the oxidized form of nitrate

(NO3-) (Galloway et al. 2008), which is typically the nutrient that limits primary production

(Nixon et al. 1986, Ryther & Dunstan 1971). Increased N loading partially alleviates this

limitation and promotes primary production; however, in addition to fueling primary production,

NO3- can also serve as an electron acceptor in heterotrophic microbial metabolisms. Sulfate

88

reduction (SO42-) is typically the dominant metabolic process in salt marsh sediments due to

widespread anoxia and a large supply of ions from seawater (Howarth & Teal 1979, Howarth

1984). NO3- reduction, however, is a more thermodynamically favorable process than SO4

2-

reduction, releasing more free energy per mole of C oxidized (Canfield et al. 2005). NO3-

addition could therefore increase the rate of OM decomposition by increasing microbial

respiration. Further, because of the additional energy yielded by NO3- respiration, there is a

“NO3--accessible” pool of OM that becomes accessible only under high NO3

- availability

(Bulseco-McKim et al. In review). What remains unclear, however, is whether a legacy of NO3-

enrichment can diminish that NO3--accessible pool of OM, thereby fundamentally altering marsh

C storage.

If NO3- addition does stimulate OM decomposition during the burial process, then

sediments from chronically NO3- enriched sites may store less C than sites without chronic NO3

-

enrichment. Additionally, C that is buried in the accreting marsh may be less labile because

microbes will have already oxidized part, or all, of this NO3- accessible OM pool. The result

would be the accumulation of OM with lower substrate quality and more complex chemical

composition, resulting in sediments that are more resistant to future decomposition (Cowie &

Hedges 1994, Middelburg 1989, Wakeham et al. 1997). In a meta-analysis conducted across 900

terrestrial litter decomposition studies, chronically enriched sites exhibited decreased stimulation

of litter decay in response to nutrient addition, due to altered leaf chemistry and shifts in the

decomposer community (Knorr et al. 2005). Different OM fractions follow an asymptotic pattern

in rates of decomposition under N loading, with early stages exhibiting faster decomposition

rates than late stages (Hobbie et al. 2012). Thus, it is clear that OM quality plays an important

89

role in determining the effect of N on decomposition rates in terrestrial systems. How these

processes play out in vegetated marine sediments, however, remains unclear.

Increasing N concentrations could also alter the microbial community, causing a shift in

OM lability, and thereby also affecting the amount of OM that is decomposed. Large supplies of

exogenous N can select for N-adapted microbes that can produce more extracellular enzymes

and oxidize complex, recalcitrant OM (Treseder et al. 2011, Bulseco-McKim et al. In review).

Nutrient enrichment may also influence the metabolic strategy of the microbial community, with

increased N supply resulting in shifts in the relative proportion of dissimilatory nitrate reduction

to ammonium (DNRA) and denitrification (DNF) (Koop-Jakobsen & Giblin 2010). The shift

between these two pathways can be important as DNF results in N loss while DNRA conserves

N in the ecosystem (Burgin & Hamilton 2007). High NO3- conditions appear to favor DNF

(Tiedje 1988, Giblin et al. 2013), and NO3- limitation and labile OM appear to favor DNRA

(Burgin & Hamilton 2007, Algar & Vallino 2014, Hardison et al. 2015), though it is unclear how

nutrient enrichment will alter these controls or to what extent. Nutrient enrichment can also alter

the community structure of denitrifying microbes (Bowen et al. 2013, Angell et al. 2018, Graves

et al. 2016) and can increase both the diversity and abundance of putative fungal denitrifiers

(Kearns et al. 2018), potentially translating to more OM oxidation when compared to systems

without exogenous sources of NO3-.

We lack an understanding of whether these changes to the microbial community and

associated metabolisms persist through time. Legacy effects can persist in soil microbial

communities (Evans & Wallenstein 2012, Bernhard et al. 2015, Giauque & Hawkes 2016) and

affect enzyme activity (Averill et al. 2016) for several years after changes to the environment

(Cuddington 2011). It is possible, however, that microbial communities can remain resilient

90

against environmental changes through increased dormancy, a bet-hedging strategy where

microbes enter an inactive state under unfavorable conditions (Lennon & Jones 2011). During a

long-term nutrient enrichment experiment (Deegan et al. 2007), increased N had no effect on the

total salt marsh sediment bacterial community (Bowen et al. 2011) but significantly altered the

active bacterial community and decreased active diversity by inducing dormancy (Kearns et al.

2016). This suggests that salt marsh microbial communities possess a genetic reservoir of traits

that can respond to future changes in the environment. Understanding how the microbial

community changes in response to increased N supply, whether or not chronic nutrient

enrichment leaves behind a legacy of effects, and how this legacy translates to OM

decomposition, is critical to a better understanding of long term C storage in salt marsh systems.

My objective was to examine salt marsh sediments from sites exposed to a range of NO3-

concentrations, varying both in extent (below detection to up to 1000 µmol L-1) and duration (no

exposure to 40-years of chronic enrichment), and to characterize OM quality and the associated

microbial community. I hypothesized that sediments from sites with greater NO3- enrichment

would exhibit lower OM quality and harbor microbes better adapted to a high N environment. To

examine if these site differences influenced functional response to further nutrient enrichment, I

then performed a controlled flow through experiment, where I added 500 µmol L-1 NO3- and

measured metabolic response and changes to the associated microbial community relative to a

seawater control. I hypothesized that 1) NO3- addition would stimulate sediment microbial

respiration due to the presence of a more energetically favorable electron acceptor, 2) this

stimulation would be less pronounced in chronically enriched sediments due to a less labile OM

pool, and 3) NO3- addition would alter the microbial community to a lesser extent in chronically

91

enriched sites due to the presence of taxa already conditioned to survive in a high N

environment.

Methods

Sample collection

I collected sediment cores (5 cm diameter, 30 cm deep) from the tall ecotype of S.

alterniflora in three salt marshes varying in time and intensity of prior NO3- exposure. My sites

(Fig. 1) included (1) West Creek (reference), a relatively pristine reference marsh, (2) Sweeney

Creek (13-year enriched), which was experimentally enriched with 70 µM NO3- dissolved into

flooding tide waters for 13 years as part of the TIDE project (Deegan et al. 2007, 2012), and (3)

Greenwood Creek (40-year enriched), which has received UV-sterilized wastewater effluent

from a nearby wastewater treatment plant for 40 years, with NO3- concentrations reaching as

high as 1000 µM (Graves et al. 2016). To focus on legacy effects that occur within deeper

sediments where C storage occurs (Chmura et al. 2003), I sectioned and homogenized four cores

from each site at 20-25 cm depth.

Organic matter analyses

To compare OM characteristics among sites of varying NO3- legacy, I subsampled

sediments from each core under anoxic conditions, immediately flash froze them in liquid

nitrogen for nucleic acid extraction, and stored the sediments at -80ºC until further analysis. I

froze another subsample of sediment at -20ºC for OM characterization and set aside the

remaining sediment in anoxic jars for the decomposition experiment, described below. I

performed elemental composition analysis (%C and %N) using a Costech Elemental Analyzer

4010 (Costech Analytical Technologies, Valencia, CA) on the initial sediments that were dried at

92

65ºC and fumed with 12N hydrochloric acid. I used the same dried samples to measure %S by

combusting them at 1350ºC and measuring sulfur dioxide production on a LECO S635 S

Analyzer (LECO corporation, Saint Joseph, MI). I then dried additional samples at 105ºC to

obtain water content and used these values to calculate bulk density. I performed a one-way

ANOVA with core replicate as a random effect to compare resulting values for %OM, %C, %N,

molar C:N, and %S across sites.

I also used Fourier Transform-Infrared Spectroscopy (FT-IR) to more precisely examine

differences in OM across sites. This spectroscopic technique provides detailed information about

the relative abundance of chemical functional groups important to decomposition, thereby

allowing for inferences regarding certain OM properties such as recalcitrance. To prepare

samples for FT-IR, I finely ground sediments that were dried at 40ºC for 48 hours. I ran each

sample on a Bruker Vertex 70 FT-IR (Bruker Optics Inc., Billerica, MA) with a Pike AutoDiff

diffuse reflectance accessory (Pike Technologies, Madison, WI) and obtained data as pseudo-

absorbance (log[1/reflectance]) in diffuse reflectance mode. I collected scans at the mid-IR range

(4000-400 cm-1), at a 2 cm-1 resolution, with 60 co-added scans per spectrum, and used a mirror

for background correction and potassium bromide (KBr) and Harvard Forest soils as standards.

To baseline correct the data, I transformed each raw spectrum using a calculated two-point linear

tangential baseline in Unscrambler X (Camo Software, version 10.1, Woodbridge, NJ) and

assigned peaks (Table S1) according to Parikh et al. (2014) and Margenot et al. (2015). I also

calculated a relative recalcitrance index, routinely used to describe soils and sediments (Ding et

al. 2002, Veum et al. 2014), using the following equation:

Eq. 1 Index II = 2924 + 2850 + 1650 + 1470 + 1405 + 920 + 840

3400 + 1270 + 1110 + 1080

93

where each value represents a wavenumber (Table S1) corresponding to either a C (numerator)

or oxygen-bonded (denominator) functional group. Higher Index II values are typically

associated with greater levels of sediment OM recalcitrance (Ding et al. 2002; Veum et al. 2014).

I compared Index II values across sites using a one-way ANOVA.

Nucleic acid extraction, amplification, and sequencing

To assess differences in microbial community structure among my sites, I extracted

genomic DNA from approximately 0.25 g sediment using the MoBio® PowerSoil DNA

Isolation Kit (MoBio Technologies, CA, USA) following manufacturer’s instructions, and eluted

the DNA into a 35 µL final volume. To extract RNA, I used a method modified from Mettel et

al. (2010) according to Kearns et al. (2016). I added 700 µL PBL buffer (5 mM tris-

hydrochloride [pH 5.0], 5 mM ethylenediaminetetraacetic acid disodium salt, 0.1% [wt/vol]

sodium dodecyl sulfate, and 6% [vol/vol] water-saturated phenol), to approximately 0.5 g

sediment, and 0.5 g of 0.17 mm glass beads. After vortexing at maximum speed for 10 minutes, I

spun the samples at 20,000 x g for 30 seconds and transferred the supernatant to a new tube. To

resuspend the remaining sediment and glass beads, I added 700 µL TPM buffer (50 mM tris-

hydrochloride [pH 5.0], 1.7% [wt/vol] polyvinylpyrrolidone, 20 mM magnesium chloride), and

vortexed at maximum speed for an additional 10 minutes. I spun the sediment at 20,000 x g for

an additional 30 seconds, and pooled the supernatant with the supernatant from the previous step.

To each sample, I added an equal volume of phenol:chloroform:isoamyl alcohol (25:24:1 v/v/v),

mixed by vortexing at maximum speed for 30 seconds, and spun at 20,000 x g for 30 seconds. I

then transferred the aqueous layer to a new tube and precipitated nucleic acids using 0.7x

volumes of 100% isopropanol and 0.1x volumes of sodium acetate [pH 5.7]. After spinning at

20,000 x g for 30 minutes, I discarded the supernatant, and washed the resulting pellet using 70%

94

ethanol. I loaded the washed RNA onto an Illustra Autoseq G-50 Spin Column (GE Healthcare),

and spun at 650 x g for 7 seconds, and eluted three times with 200 µL 1.5 M NaCl (pH 5.5). I

precipitated the flow through with 0.7x volumes of 100% isopropanol and 0.1x volumes of

sodium acetate (pH 5.7), spun at 20,000 x g, and resuspended the resulting pellet in 50 µL di-

ethyl pyrocarbonate (DEPC) treated water. I checked for DNA contamination in the RNA using

general bacterial primers 515F and 806R (Bates et al. 2011), and removed contamination using

DNase I (New England BioLabs, Ipswich, MA). Lastly, I reverse transcribed 2 µL RNA to

cDNA using random hexamer primers and an Invitrogen Superscript III cDNA synthesis kit for

RT-PCR (Life Technologies, Carlsbad, CA, USA).

After confirming the presence of DNA using SYBR Safe (Thermo Fisher Scientific,

Waltham, MA), I amplified in triplicate the V4 region of the reverse transcribed 16S rRNA and

the 16S rRNA gene using the general bacterial primer-pair 515F (Bates et al. 2011; 5’-

GTGCCAGCMGCCGCGGTAA-3’) and 806R (5’-GACTACHVGGGTWTCTAAT-3’) with

Illumina adaptors (Caporaso et al. 2012) and individual 12-bp GoLay barcodes on the reverse

primer, using the following reaction: 10 µl 5-Prime Hot Master Mix (Quanta Bio, Beverly, MA),

0.25 µl of 20 µM forward and reverse primers, 13.5 µL DEPC-treated water, and 1 µl of DNA of

cDNA template. I gel purified PCR products using a Qiagen® QIAquick gel purification kit

(Qiagen, Valencia, CA) and quantified the purified product using a Qubit® 3.0 fluorometer (Life

Technologies, Thermo Fisher Scientific, Waltham, MA). After pooling to equimolar amounts, I

performed sequencing on an Illumina MiSeq (Illumina, San Diego, CA) platform using the

paired-end 250 bp 500 cycle kit with V2 chemistry.

I analyzed sequence data in QIIME 2 (version 2018.2) and demultiplexed a total of

2,871,972 sequences from the 16S rRNA gene and its gene product, 16S rRNA. Using the

95

DADA2 plugin (Callahan et al. 2016), I inferred amplicon sequence variants (ASVs) with a

maxEE of 2 and the consensus chimera removal method. I then assigned taxonomy against the

Greengenes 16S rRNA sequence database (version 13-8; McDonald et al. 2012) and removed

any sequences matching chloroplasts and mitochondria in addition to ASVs occurring only once

(singletons). Following quality filtering, I had a total of 2,094,824 sequences with an average of

29,095 sequences per sample that I aligned using MAFFT v.7 (Katoh & Standley 2013).

I examined differences in the microbial community from each site by conducting a

canonical correspondence analysis (CCA; ter Braak & Verdonschot 1995) on ASV tables

normalized to 16,426 sequences (my lowest sequencing depth for the 16S rRNA gene) and tested

for significance among sites using the ‘anova.cca’ function with 9999 permutations and 100

steps in the VEGAN package (v4.6-12; Okansen et al. 2017). I then identified environmental

factors driving community structure using the envfit vector fitting function in the VEGAN

package after screening for multicollinearity by excluding any factors with a variance inflation

factor (VIF) value exceeding 5. To identify which taxa differed most among sites, I performed a

differential abundance analysis in QIIME 2 using the ANCOM (analysis of composition of

microbiomes) plugin (Mandal et al. 2015), which compares relative abundance between groups

by calculating the Aitchison’s log-ratio of the relative abundance of a single taxon against that of

the remaining taxa (Aitchison 1986) and has stringent controls against false discovery (Weiss et

al. 2017). I applied ANCOM on a genus-level table filtered with ASVs that appeared at least 100

times in the dataset and used pseudo-count values to make all counts non-zero.

To quantify the abundance of 16S rRNA gene copies, I performed quantitative PCR

(qPCR) in triplicate using 357F and 519R primers (Turner et al. 1999) on an Aria Mx Real Time

PCR instrument (Agilent Technologies). Each 20 µl reaction consisted of 10 µl of Brilliant III

96

Ultrafast SYBR Green qPCR Master Mix (Agilent Technologies), 0.5 µl of each primer (20

µM), 8 µl of DEPC-treated water, and 1 µl DNA template. I then used the qPCR program

optimized for 16S rRNA gene targets, which consisted of an initial denaturation step at 95ºC for

3 minutes followed by 40 cycles of 95ºC for 5 seconds and 60ºC for 10 seconds, with data

acquisition at the end of each cycle. A melt curve was conducted to confirm the purity of the

target gene amplification product. Standards were prepared from purified 16S rRNA gene

amplicons, which were quantified and assessed for size with a tape station (Agilent

Technologies). Standard curves, run with each batch of samples, resulted in R2 >0.95 and qPCR

efficiency >90%. I compared abundance of 16S rRNA gene copies across sites using a one-way

ANOVA.

Decomposition experiment

I next conducted a decomposition flow through reactor (FTR) experiment to assess how

sediments from each site described above would respond to further NO3- addition. My FTR

experimental system (Bulseco-McKim et al. In review) is a modified version of the system

described in Pallud et al. (2006, 2007). In contrast to whole-core incubations or sediment

slurries, using an FTR system provides biogeochemical rate measurements under steady-state

conditions, prevents the accumulation of dissolved metabolic byproducts, and allows for the

isolation of microbial activity from other environmental conditions that may obscure

measurements in the field (e.g. influence of plant communities or tidal flux). Each FTR consists

of two polyvinyl chloride radially-scored caps that ensure uniform flow and that are sealed with

O-rings to prevent leakage. To assess flow characteristics in the reactors, which were 31.81 cm3

in volume, I performed breakthrough experiments using bromide. Flow property characteristics

can be found in Fig. S1 and Table S2.

97

Under anoxic conditions, I loaded the FTRs with homogenized sediment collected from

20-25 cm at each site and randomly assigned reactors a treatment, either plus-NO3- (+NO3

- in 0.2

µm filtered seawater) or unamended (0.2 µm filtered seawater only, representing natural marsh

conditions). To prepare the two treatment reservoirs, I filtered water collected from Woods Hole,

MA (0.2 µm pore size), sparged the reservoirs with N2 gas until they reached anoxic conditions,

and spiked the plus-NO3- reservoir with 500 µmol L-1 additional K15NO3

- (Cambridge Isotope

Laboratories, Andover, MA). Half of the reactors from each site received the plus-NO3-

treatment and half received the unamended treatment, both at a targeted flow rate of 0.08 mL

min-1 under continuously anoxic conditions. To determine nitrate reduction pathways in the plus-

NO3-treatment, I followed the fate of the 15N tracer into various end products (28, 29, 30N2). I

collected samples approximately every 10 days from both the reservoir and effluent throughout

the 100-day experiment once the FTRs reached steady state, which took approximately 10 days.

To assess changes that occurred in the bulk sediment as a result of experimental conditions, I

homogenized sediment from each FTR at the end of the experiment. I immediately sub-sampled

sediments for nucleic acid extraction and OM analysis following methods described above.

Biogeochemical analyses

I collected water samples from both the plus-NO3- and unamended treatment effluent, as

well as each reservoir, to quantify biogeochemical processes resulting from microbial activity in

the decomposition experiment. I measured dissolved inorganic C (DIC; CO2 + HCO3 + CO32-) as

an indicator of total microbial respiration on an Apollo SciTech AS-C3 DIC analyzer (Newark,

DE) and NO3- consumption on a Teledyne T200 NOx analyzer (Teledyne API, San Diego, CA)

using chemoluminescent methods (Cox 1980). I measured ammonium (NH4+) and sulfide (HS-)

colorimetrically on a Shimadzu 1601 spectrophotometer (Kyoto, Japan) following protocols

98

from Solorzano (1969) and Gilboa-Garber (1971), respectively. To quantify analyte consumption

or production rate over time, I followed eq. 2 (Pallud et al. 2006):

Eq. 2 R =(Cout−Cin)Q

V

where R is the consumption or production rate of interest, Cout and Cin are the effluent and

reservoir analyte concentrations, respectively, Q is the measured flow rate in L hour-1, and V is

the FTR volume (31.81 cm-3). I then calculated a cumulative flux for each analyte by integrating

between each measured point throughout the experiment. Since background SO42- concentrations

are typically high in seawater (~28 mmol L-1), I determined SO42- reduction rates (SRR) by

calculating the sum of the total production of hydrogen sulfide (HS-) and total sulfur (S) at the

end of the experiment.

To determine the relative contribution of each NO3- reduction pathway, I made dissolved

gas measurements of N2 on a membrane inlet mass spectrometer (Kana et al. 1994) connected to

an inline furnace set to 500ºC and copper column to remove oxygen interference (Eyre et al.

2002, Lunstrum & Aoki 2016). I monitored the production of 29N2 and 30N2 from added 15NO3-

tracer as a measure of denitrification (DNF) and calculated rates using the following equation

from Nielson et al. (1992):

Eq. 3 D15 = p29+2p30

Where D15 is denitrification from 15NO3- and p29 and p30 represent production of 29N2 and 30N2,

respectively. Because I only added NO3- in the form of 15NO3

-, and ambient concentrations of

14NO3- were largely below detection, I did not calculate D14. I also considered production of

14NO3- from nitrification, which is a largely aerobic process (Herbert 1999), as negligible since

this experiment was conducted under strictly anoxic conditions. To measure DNRA, I bubbled

water samples with helium for 10 minutes to remove any N2 and converted 15NH4+ produced

99

from DNRA to 29N2 and 30N2 using sodium hypobromite following the OX/MIMS method (Yin

et al. 2014). I calculated DNRA rates as:

Eq. 4 DNRA15 = p29+2p30

Where DNRA15 is DNRA from 15NH4+ and p29 and p30 represent production of 29N2 and 30N2,

respectively. These measurements represent denitrification and DNRA occurring in the FTRs

that result from adding 500 µmol L-1 15NO3-. I did not make these measurements in the

unamended treatment where I did not add any 15NO3-. Isotope tracer measurements were

completed at 9, 12, and 13-weeks after the start of the experiment, with no DNRA measurements

available on week 12 due to sample limitation.

Statistical analyses for decomposition experiment

To examine differences in DIC production across treatments and site, I performed a

repeated measures ANOVA. I compared cumulative fluxes of DIC production, and NO3- and

SO42- reduction (HS- production + S storage), across treatment and site using a two-way

ANOVA and calculated an effect size to measure the magnitude of response to NO3-. To further

explore the influence of initial bulk C supply on DIC production, I calculated total C loss by

taking the proportion of C released as DIC divided by the total mass of C per reactor using

sediment bulk density and %C. I performed a mass balance between DIC production and

NRR/SRR, calculated the proportion of NRR explained by DNF and DNRA in the plus-NO3-

treatment, and performed a one-way ANOVA comparing DNF and DNRA across sites. I also

calculated a DNF:DNRA ratio to assess the contribution of DNRA relative to DNF and

compared this ratio across sites using a one-way ANOVA.

To examine changes in OM characteristics (%C, %N, and %S) that occurred as a result of

the decomposition experiment, I compared the plus-NO3- and unamended treatments within site

100

using a student’s t-test. I performed a principal coordinate analysis (PCoA) across the FT-IR

spectra (4000-400 cm-1) and tested for differences among treatment and site using a

PERMANOVA with 999 permutations on a resemblance matrix constructed using Manhattan

distance. To better understand which spectral bands and associated functional groups were

responsible for observed patterns in the PCoA, I plotted the Pearson’s correlation coefficients

against wavenumber, with peaks exhibiting the greatest absolute change influencing the PCoA

most. Lastly, I performed a linear regression between DIC production and Index II values across

all treatments and sites.

To analyze the response of the microbial community to NO3- addition in the

decomposition experiment, I performed beta diversity analysis on a 16S rRNA gene ASV table

normalized to 16,427 sequences per sample, which was my lowest sampling depth, and tested for

significant differences among treatments and site using PERMANOVA with 999 permutations in

the vegan package (Oksanen et al. 2017). To further assess the effect of NO3- across sites, I

compared the weighted UniFrac dissimilarity values between plus-NO3- and unamended

treatments and tested for significance using a one-way ANOVA. I then ran a random forest

model with 10,000 trees with the randomForest R package (v4.6-12; Liaw & Wiener 2002) using

ASVs that occurred at least 100 times, to identify which taxa were most important in

discriminating between the plus-NO3- and unamended treatment within each site. This allowed

us to identify which taxa were most significantly associated with each treatment, and whether

those distinguishing taxa differed among sites. To do this, I calculated the difference in relative

abundance between the plus-NO3- and unamended treatments on a per core basis on the top ten

most important ASVs to examine how these taxa changed in response to the NO3- addition.

101

Lastly, because prior studies showed that microbes in salt marsh and soil habitats exhibit

high rates of dormancy (Kearns et al. 2017) and abundant relic DNA (Carini et al. 2016),

particularly in response to nutrient enrichment (Kearns et al. 2016), I assessed the effect of NO3-

on microbial activity at each site. To accomplish this, I calculated a 16S rRNA:16S rRNA gene +

1 ratio on ASVs that appeared at least 100 times across all samples and defined any ratio >1 as

active at the ASV level (Jones & Lennon 2010). I then identified taxa that changed most in

activity by comparing the activity ratio in the plus-NO3- and unamended treatments within

replicate cores that started with the same initial microbial community using permutation tests

against a null distribution. There are caveats associated with the use of a 16S rRNA:16S rRNA

gene ratios, such as variations in rRNA production, growth, and gene copy (Blazewicz et al.

2013; Steven et al. 2017; Papp et al. 2018); however, by comparing the putative activity ratio

between treatments only within a given taxon, these biases are minimized.

Results

Initial site characterization

There were no significant differences (Table 1) in %OM, %C, or molar C:N of initial

sediments across sites (Fig. 1); however, %N was different (p=0.003, F2,8=21.91) with higher

values at both the 13-year and 40-year enriched site. %S was greater in the reference site when

compared to the 40-year enriched site (p=0.010, F2,9 = 10.98) and Index II was higher at the 40-

year enriched site when compared to both the reference and 13-year enriched sites (Fig. 2A;

p<0.001, F2,9=20.99). These differences were also evident in the baseline corrected spectra (Fig.

2B), where the 40-year enriched sediments demonstrated polysaccharide depletion (1080 and

1110 cm-1) and enrichment of aromatic compounds (1650 cm-1). 16S rRNA gene abundance

102

varied among sites (p=0.026, F2,8=5.943) with a significantly greater copy number at the

reference site when compared to 40-year enriched site (Table 1). A PERMANOVA revealed a

significant effect of site on microbial community structure (Fig. 3A; p=0.001, pseudo-

F2,9=1.991), with %S and Index II playing a significant role (p < 0.05) at the reference and 40-

year enriched site, respectively. ANCOM analysis (Fig. 3B) showed that the top 20 ASVs most

differentially abundant among sites accounted for 13.57, 10.26, and 7.57% of the total number of

sequences for the reference, 13-year, and 40-year enriched sites, respectively. The orders

Desulfobacterales, Desulfarculales, and Bacteroidales were similar between the reference and

13-year enriched sites, but the 40-year enriched site was quite different, exhibiting greater

overall diversity. Nitrosomonadales were only present in the 40-year enriched site, and orders

Saprospirales, Cytophagales, and an unclassified Acidobacteria were present only in the enriched

sites and absent in the reference site.

Total microbial respiration

During the decomposition experiment, DIC production rate in the unamended treatment

averaged 21.67 ± 2.55, 19.41 ± 0.91, and 14.71 ± 2.53 µmol cm-3 hr-1 for the reference, 13-year,

and 40-year enriched sites, respectively (Fig. 4). NO3- addition resulted in significantly higher

DIC production rates when compared to unamended sediments across all sites, averaging 26.48 ±

5.51, 24.57 ± 3.88, and 18.57 ± 3.70 µmol cm-3 hr-1 for the same sites, respectively. Cumulative

DIC was higher in the plus NO3- sediments across all sites (Fig. 5; p=0.002, F1,20=12.246), and

higher overall at the reference and 13-year enriched sites when compared to the 40-year enriched

site (Fig. 5; p<0.001, F(2,14)=48.33) regardless of treatment. There was no significant interaction

between treatment and site, however, the effect size (Cohen’s d) in response to NO3- was 1.65,

1.90, and 0.78 for the reference, 13-year, and 40-year enriched sites, respectively, corresponding

103

to ~74, 79, and 43% non-overlap between each treatment distribution. The proportion of C lost

as DIC was 1.41% ± 0.10, 1.08% ± 0.04, and 0.72% ± 0.06 in the plus-NO3- treatment, and

1.01% ± 0.11, 0.89% ± 0.12, and 0.58% ± 0.01 for the unamended treatment, for the reference,

13-year enriched, and 40-year enriched sites, respectively. There were no significant trends by

treatment or site when examining NO3- reduction rate (p=0.39; Fig. 6A) and cumulative nitrate

reduction (p=0.149; Fig. 6B) or sulfide production rate (p=0.40; Fig. 6C) and cumulative sulfate

reduction (p=0.359; Fig. 6D).

When examining the ratio of DIC production:NO3- consumption in the plus-NO3

-

treatment, average per site was 0.61 ± 0.10, 0.55 ± 0.17, and 0.56 ± 0.12 for the reference, 13-

year, and 40-year enriched sites, respectively, which is considerably less than the ratio that

would be predicted by DNF (1.25) or heterotrophic DNRA (2) stoichiometry (Canfield et al.

2005, Giblin et al. 2013). When including any SO4- reduction that occurred in the plus-NO3

-

treatment, this ratio was even lower (0.56 ± 0.04, 0.55 ± 0.08, 0.53 ± 0.06), indicating that not all

DIC could be accounted for as a result of NO3- and SO4

- reduction. In contrast, the DIC

production:SO4- reduction ratio in the unamended treatment was much closer to, but exceeded,

the stoichiometrically predicted value of 2 (2.53 ± 1.31, 3.71 ± 1.36, and 6.08 ± 7.61 for

reference, 13-year, and 40-year enriched sites, respectively).

The sum of DNF and DNRA accounted for ~44, 40, and 61% of total NRR on week 9

and ~60, 76, and 44% of NRR on week 13 at the reference, 13-year, and 40-year enriched sites,

respectively (Table 2). Overall DNF rates were significantly greater at the 13-year enriched site

when compared to the 40-year enriched site (p=0.010, F2,26=5.531), and DNRA rates were

greatest at the reference site (p=0.002, F2,19=9.174), with significant differences between the 13-

year and 40-year enriched sites, and 40-year and reference sites (but not the 13-year and

104

reference sites) (Table 2). The DNF:DNRA ratio was 3.01 ± 0.68 (Reference), 3.03 ± 0.52 (13-

year enriched), and 11.49 ± 3.63 (40-year enriched), indicating that DNF was always the

dominant nitrate reduction process. The 40-year enriched site, however, exhibited less than half

the amount of DNRA relative to DNF when compared to the reference and 13-year enriched site

(p=0.008, F2,15=6.764).

Organic Matter Characteristics

A student’s t-test indicated a significant difference in %C between plus-NO3- and

unamended sediments in the reference marsh (p=0.002, t=-5.219, df=6; Table 3), with all other

bulk sediment properties (%OM, %N, and %S) showing no change as a result of nutrient

enrichment or site. This was not surprising considering only 0.5-1.5% of C was lost as DIC

through respiration. A PCoA of the whole FT-IR spectra allowed for a more refined examination

of OM characteristics (Fig. 7A), demonstrating a clear separation by site along the primary axis

(explaining 84.5% of the variation). There was no effect of treatment (p=0.687), however,

PERMANOVA indicated a significant effect of site (p=0.001, pseudo-F2,27=48.779), with these

differences largely resulting from decreases in polysaccharides (C-O band at 1080 cm-1) and

increases in aromatic (C=C band at 1650 cm-1), aliphatic (asymmetric and symmetric stretching

vibration C-H bands at 2924 cm-1 and 2850 cm-1, respectively), and amide C (N-H and C=N

bands at 1575 cm-1; Fig. 7B). A linear regression showed that DIC production significantly

decreased with increasing recalcitrance as indicated by greater Index II values (Fig. 7C; p<0.001,

F1,10=26.42, R2=0.698).

Microbial community in response to NO3-

A PCoA (Fig. 8A) constructed from weighted UniFrac of the microbial communities

sampled at the end of the decomposition experiment indicated that microbial communities

105

differed according to site (PERMANOVA: p=0.001, F=4.11). NO3- addition significantly altered

microbial community structure (p=0.001, pseudo-F=5.041) by shifting the similarities along the

primary axis, however the difference between treatments across sites was not significant (Fig.

8B; p=0.213). Random forest modeling correctly discriminated between the plus-NO3- and

unamended treatments 62.5, 87.5, and 87.5% of the time for the reference, 13-year enriched, and

40-year enriched sites, respectively (kappa-statistic = 25, 75, and 75%). This allowed me to

identify the top ten ASVs most important in correctly classifying between treatments (Fig. 9;

Table S3, which accounted for 7.19, 24.09, and 8.18% of the dataset. Orders Desulfarculales

(~3.4%) and Myxococcales (~0.63%) increased most in relative abundance at the reference site,

Thiotrichales (~2.5%) increased most at the 13-year enriched site, and Chromatiales increased

most at both the 13-year (~9.89%) and 40-year enriched (0.01%) sites, in response to NO3-

enrichment. Desulfobacterales decreased in response to NO3- at both the 13-year and 40-year

enriched sites, demonstrating a ~6.9 and ~10.6% decrease in relative abundance.

According to permutation tests comparing the 16S rRNA:16S rRNA gene ratios between

plus-NO3- and unamended treatments, there were 17, 12, and 15 taxa out of 1895 ASVs that

exhibited a significant change in activity at the reference, 13-year enriched, and 40-year enriched

sites, respectively, when compared to a null distribution (Fig. 10; Table S4). Orders

Thiotrichales, Chromatiales, Rhodothermales, Kiloniellales, Oceanospirillales, and

Alteromonadales were consistently more active (greater 16S rRNA:16S rRNA gene ratio) under

high NO3- conditions, while Syntrophobacterales, Desulfobacterales, Desulfarculales,

Campylobacterales, Clostridiales, Desulfovibrionales, and Candidate Phylum GN15 were more

active in the unamended treatment. Some orders, such as Myxococcales, were active in both

treatments.

106

Discussion

Nutrient-enriched sites contain less labile OM and N-adapted microbes

There were no significant differences in bulk %OM, %C, %N, and C:N ratio (Table 1),

however, sediments from the 40-year enriched site demonstrated significantly higher Index II

values compared to the 13-year enriched and reference sites (Fig. 2A). This metric, which is

commonly used to evaluate soil quality (Ding et al. 2002, Veum et al. 2014), increases with

greater recalcitrance due to the accumulation of C relative to O-containing functional groups

(Chefetz et al. 1998; Ding et al. 2002; Veum et al. 2014; Margenot et al. 2015). One explanation

for this pattern is the presence of a NO3--accessible pool of OM that is reactive only under

conditions of high NO3- (e.g. Bulseco-McKim et al. In review). Under this scenario, sites

experiencing chronic nutrient enrichment would demonstrate greater rates of decomposition

during initial deposition, thereby resulting in the accumulation of less labile OM over time. I

observed this pattern in the 40-year enriched site, despite the fact that it contained greater overall

%C. This pattern is further supported by the FT-IR spectra (Fig. 2B), where the depletion of

polysaccharides and enrichment of more aromatic forms of C in the 40-year enriched site suggest

that microbes are oxidizing more labile OM and leaving behind compounds that are more

chemically complex, such as lignin.

Another explanation for differences in OM quality may be due to changes to the

microbial community that result from shifting environmental conditions. In this study, I observed

significant site-specific differences in both microbial abundance and community structure (Table

1; Fig. 3A). In reference sediments, %S was a significant driver of community structure,

suggesting that in the absence of NO3-, S-metabolism is the dominant metabolic strategy in

107

reference sediments, as is typical in salt marshes (Howarth & Teal 1979, Howarth 1984). Out of

the top 20 ASVs that were most differentially abundant among sites, groups belonging to orders

known to reduce SO42- (Klepac-Ceraj et al. 2004, Bahr et al. 2005), including Desulfobacterales,

Desulfarculales, and other unclassified Deltaproteobacteria, were most abundant at the reference

site. The relative abundance of these orders was lower at both the 13-year and 40-year enriched

sites, suggesting that taxa associated with SO42- are not as dominant at the chronically enriched

sediments in this study when compared to reference sediments.

The divergence in the microbial community could also be due to salinity differences

(Table 1), as increases in taxonomic groups associated with SO42- reduction tend to occur at sites

with greater salinity (Howarth & Teal 1979). However, the most significant salinity-induced

changes to microbial communities seem to occur at ~5 ppt (Weston et al. 2010) when

metabolisms shift from methanogenesis to SO4- reduction. Salinity was lower at the 40-year

enriched site, but was also more variable, when compared to the reference and 13-year enriched

sites. Regardless of this difference, the same taxa associated with SO4- reduction decreased in

both the 13-year and 40-year enriched site, suggesting that salinity was less important than

nutrient enrichment in driving this pattern (Fig. 3B). Further, the 13-year enriched site had a

higher salinity than the reference site (Table 1) so the abundant ASVs associated with SO4-

reduction were highest at the site with intermediate salinity, further indicating that nutrient

enrichment, not salinity, was the major driver of microbial community structure across sites.

These shifts have important implications for OM quality, as SO42- reducers can typically only

decompose low-molecular weight compounds, leaving behind fractions that may only be

accessible to microbes under high NO3- conditions (Canfield 1989, Westrich & Berner 1984).

108

Index II was a significant driver of microbial community structure at the 40-year enriched

site (Fig. 3A), where total microbial abundance was an order of magnitude lower than the 13-

year enriched and reference sites (Table 1). This is likely due to the energy required to oxidize

more recalcitrant OM, or because competition over complex forms of OM results in lower

overall microbial abundance (Table 1). I identified several taxa that were more abundant in the

enriched sites and that belong to groups known to oxidize complex forms of OM (Fig. 3B). Taxa

from the group Anaerolineae contained the broadest range of carbohydrate hydrolytic genes in a

metagenomic study that reconstructed 82 bacterial genomes from estuarine sediment, suggesting

an important role in degrading complex C compounds (Baker et al. 2015). In addition, groups

from the Phylum Acidobacteria, also observed in this study, were widely distributed in soils

where they metabolized more refractory forms of OM (Hartman et al. 2008), providing further

evidence that microbes found in the enriched sites can oxidize less labile forms of OM that

resulted from greater decomposition rates during initial OM burial.

Several taxa, including those from the orders Cytophagales, Rhodocyclales, Chlorobi,

and Nitrosomonadales, were significantly more abundant in both enriched sites, suggesting they

might be important in N-cycling. Metagenomic sequencing of marine sediments revealed that

Cytophagales harbored high abundances of the nosZ gene, a key gene in the final step of

denitrification (Rasigraf et al. 2017). This is consistent with other metagenomic studies in salt

marsh sediments that observed elevated abundance of nosZ and atypical nosZ (Graves et al.

2016) and other genes associated with denitrification within the Cytophagales group (Didi-

Andreote et al. 2016). Both Rhodocyclales and Chlorobi include taxa with the capacity for

DNRA (Saito et al. 2008, Rasigraf et al. 2016) and were found in high concentrations in sites

receiving wastewater effluent (Lu & Lu 2014). Nitrosomonadales was also detected in degraded

109

wetland systems (Wu et al. 2013) either acting as chemolithoautotrophic or mixotrophic

ammonia oxidizers (Garrity et al. 2005). The significant increase in these groups in the enriched

sites of this study suggest that these taxa are better suited to take advantage of high N supply

and, through their N removal capacity, may enhance ecosystem resiliency by limiting

susceptibility to additional nutrient disturbance.

The nitrate accessible pool of OM is smaller at nutrient enriched sites

I hypothesized that if chronically enriched sites contained taxa more adapted to high N

conditions and less labile OM, then these communities would not respond as strongly to further

NO3- addition since buried OM that is available for oxidation would be more recalcitrant.

Although the addition of NO3- resulted in significantly greater DIC production compared to the

unamended treatment at all of my sites (Fig. 4, 5); the effect size was lowest at the 40-year

enriched site, supporting my initial hypothesis. Lower microbial respiration rates suggest that the

NO3--accessible pool of OM is smaller at the 40-year enriched site, presumably because some of

it was oxidized during burial. In the reference site, where more of the NO3--accessible OM pool

was still intact, there was greater decomposition of these deep sediments. A significant shift in

%C resulting from NO3- addition (Table 3), a higher decomposition rate (Fig. 4, 5) when

compared to the enriched sites, and the FT-IR spectral data all support the idea that chronic

enrichment may have already chemically altered the OM. These data suggest that the legacy of

nutrient enrichment may be more important in shaping OM quality than acute exposure and that

these long-term shifts in OM influence the response to additional NO3- exposure by increasing

recalcitrance (Fig. 7C).

When examining the mass balance between DIC:NO3- and SO4

2- reduction in the plus-

NO3- treatment, the ratios were considerably lower than the predicted stoichiometry for both

110

DNF and heterotrophic DNRA. This indicates that either 1) NO3- is being consumed by

assimilation or other processes that I would not detect through DIC production or 2)

chemoautotrophic processes are fixing DIC via their metabolism, thus drawing down the DIC

pool. Although a small amount of assimilation is likely occurring, I saw no significant

differences in cell abundance between the plus- NO3- and unamended treatments at the end of

the experiment, suggesting assimilation into greater cell biomass is unlikely to account for the

discrepancy in the ratios (Table 3). Further, anaerobic microbes are seldom nutrient-limited

because their growth-per-unit substrate-intake is much lower than for aerobic microbes (Canfield

et al. 2005). Chemoautotrophic processes, on the other hand, could include NO3- reduction

coupled to reduced S and iron (Burgin & Hamilton 2007, Giblin et al. 2013), both of which are

abundant in these sediments. Both processes fix DIC and would result in lower DIC production

relative to NO3- consumption than expected by stoichiometry.

Relative contributions of DNF and DNRA to nitrate reduction

The competing dissimilatory NO3- reduction pathways, DNF and DNRA, showed patterns

consistent with what would be predicted based on resource availability (Algar & Vallino 2014). I

found that the 13-year enriched marsh had significantly higher DNF rates than either the

reference or 40-year enriched sites (Table 2). DNF is favored over DNRA when the ratio of NO3-

availability to C lability is high because it provides more free energy per mole of C oxidized than

DNRA (Tiedje 1988, Giblin et al. 2013). Koop-Jakobsen & Giblin (2010) also found that

nutrient enrichment increased DNF by 16-fold in creek sediment from this same site when

compared to the reference site, although DNRA also increased by 10-fold. The more recalcitrant

nature of the sediments in the 40-year enriched marsh may explain why DNF rates were lower at

this site in spite of high nutrient enrichment (Fig. 2A). DNRA was significantly higher at the

111

reference and 13-year enriched marshes when compared to the 40-year enriched marsh (Table 2).

DNRA tends to dominate under conditions of high labile OM and NO3- limitation, since this

process can efficiently transfer 8 electrons per mole of NO3- reduced as opposed to the 5

electrons transferred in DNF (Tiedje 1988, Burgin & Hamilton 2007). What is most interesting is

the DNF:DNRA ratio was highest at the 40-year marsh (Table 2), suggesting that a combination

of greater recalcitrance (Fig. 2A) and higher NO3- concentrations favored DNF relative to DNRA

(Algar & Vallino 2014). Identifying the dominant NO3- reduction pathway is important because

each pathway plays a considerably different role in both N and OM cycling. While DNF

efficiently removes bioavailable N from the system, DNRA retains it, making it available for

autotrophic use (Tiedje 1988, Giblin et al. 2013). Furthermore, both DNF and DNRA can be

either heterotrophic or chemoautotrophic, with the latter augmenting C stores in the sediment.

Therefore, understanding the controls over NO3- reduction pathways is critical for both N

management and restoration applications.

NO3- driven shifts in microbial community structure and activity

Initial sediments from the 13-year and 40-year enriched sites contained taxa associated

with N-cycling, so I hypothesized that further NO3- addition would not alter the microbial

community structure compared to the reference site. Instead, I observed a significant shift in

microbial community structure in response to short-term NO3- addition at all sites. Regardless of

treatment, samples clustered by site, with reference and 13-year enriched site demonstrating

more similar community structure than the 40-year enriched site. This large-scale, site-specific

selection is commonly observed in microbial communities (Martiny et al. 2006, Hanson et al.

2012). More interestingly, the microbial community structure shifted in response to NO3- across

all sites, with no difference among sites in the intensity of this shift, as evidenced by equivalent

112

dissimilarity values between plus-NO3- and unamended treatments (Fig. 8B). This contradicts my

hypothesis that communities from the chronically enriched site would exhibit a decreased

response and suggests that regardless of the initial community structure, additional NO3- in this

experiment shifted the microbial community.

Random forest analysis allowed me to identify ASVs that were most important in

correctly classifying between treatments (Fig. 9). Across all sites, taxa belonging to groups

known to oxidize sulfur (Thiotrichales, Chromatiales; Garrity et al. 2005, Imhoff et al. 2005,

Thomas et al. 2014), carry out steps in denitrification (Myxococcales, Oceanospirillales), and

degrade hydrocarbons and other chemically recalcitrant substrates (Clostridiales, Marine

Crenarchaeota Group-B10; Biddle et al. 2006, Goldfarb et al. 2011, Mason et al. 2012), were all

important in distinguishing the plus-NO3- treatment. In contrast, groups known to reduce sulfate

(Desulfobacterales; Bahr et al. 2005), degrade aromatic compounds (Burkholderiales; Pérez-

Pantoja et al. 2012), and that are commonly found in deep, anoxic sediments

(Dehalococcoidales, Thermoplasmata-CCA47; Oni et al. 2015), were important in distinguishing

the unamended sediments.

The taxa important in discriminating between treatments were determined via analysis of

the 16S rRNA gene, which includes cells that are intact, recently dead, and dormant (Nielsen et

al. 2007). Salt marsh sediments can have a proportion of cells that are inactive that is as high as

90%, particularly in response to nutrient enrichment (Kearns et al. 2016). This limits my ability

to assess changes to the portion of the microbial community that is actively carrying out critical

ecosystem functions. Thus, I examined 16S rRNA and identified taxa whose activity changed

most between the plus-NO3- and unamended treatments. Consistent with my random forest

analysis of the total microbial community, in the plus-NO3- treatment I observed increases in

113

activity in orders known to reduce NO3- (Kiloniellales, Myxococcales; Wiese et al. 2009),

oxidize sulfur (Chromatiales, Thiotrichales; Garrity et al. 2005, Imhoff et al. 2005, Thomas et al.

2014), and degrade high-molecular-weight (HMW) compounds (Acidimicrobiales,

Flavobacteriales, Thiotrichales; Guibert et al. 2016, Hartman et al. 2009, Mahmoudi et al. 2015,

McCarren et al. 2010). Increases in S oxidizer activity in response to NO3- is particularly

interesting, as these groups of bacteria can use NO3- to carry out both DNF and DNRA

autotrophically. This fixation of C could, in part, explain the stoichiometric discrepancy in the

DIC:NO3- ratios I observed in my decomposition experiment (Fig. 4-6). Further, only the

reference and 13-year enriched sites demonstrated increased activity of sulfur oxidizers, whereas

the 40-year enriched site included more groups known to degrade HMW OM, corresponding

well with the significantly lower DNF and DNRA rates at the 40-year enriched site (Table 2),

where OM was considerably more recalcitrant (Table 1: Fig. 2A).

Conclusions

I showed that reference and 13-year enriched marshes exhibited similar OM

characteristics and microbial community structure, and that the 40-year enriched marsh

contained more recalcitrant OM and a unique microbial community. After exposing these

sediments to additional NO3- in a controlled FTR experiment, the reference site exhibited the

greatest rates of DIC production compared to the unamended control. In contrast, sediments from

the 40-year enriched site demonstrated a lower response to NO3-, despite significant changes to

both microbial community structure and activity. Taken together, this work suggests that long

term nutrient enrichment may lead to less overall C storage; however, the fraction of OM that

does become buried may be more stable when compared to less eutrophic systems. These results

114

highlight the need to consider the effects of chronic nutrient enrichment when determining C

storage potential in salt marsh systems

Acknowledgements

I would like to thank Joseph Vallino at Marine Biological Laboratory for his thoughtful

contribution to this experiment and Tom Goodkind at the University of Massachusetts Boston for

flow through reactor design and construction. I also thank researchers of the TIDE project (NSF

OCE0924287, OCE0923689, DEB0213767, DEB1354494, and OCE 1353140) for maintenance

of the long-term nutrient enrichment experiment and researchers at the Plum Island Ecosystems

LTER (NSF OCE 0423565, 1058747, 1637630). I would also like to thank Sam Kelsey at the

Marine Biological Laboratory and Alan Stebbins at University of Massachusetts Boston

Environmental Analytical Facility (NSF 09-42371 and DBI:MRI-RI2 to Robyn Hannigan and

Alan Christian) for assistance in the laboratory. This work was funded by an NSF CAREER

Award to JLB (DEB1350491) and a Woods Hole Oceanographic Sea Grant award to AEG and

JJV (Project No. NA140AR4170074 Project R/M-65s). Additional support was provided by a

Ford Foundation pre-doctoral fellowship award to ABM. All sequence data from this study is

available in the Sequence Read Archive under accession number TBD. The views expressed here

are those of the authors and do not necessarily reflect the views of NOAA or any of is

subagencies.

115

References

Aitchison J (1986) The statistical analysis of compositional data. J Roy Stat Soc B Met 44:139-

177



Angell J, Peng X, Ju Q, Craick I, Jayakumar A, Kearns PJ, Ward BB, Bowen JL (2018)

Community composition of nitrous oxide-related genes in salt marsh sediments exposed

to nitrogen. Front Micro 9:170

Averill C, Waring BG, Hawkes CV (2016) Historical precipitation predictably alters the shape

and magnitude of microbial functional response to soil moisture. Glob Change Biol

22:1957-1964




Baker BJ, Lazar CS, Teske AP, Dick GJ (2015) Genomic resolution of linkages in carbon,

nitrogen, and sulfur cycling among widespread estuary sediment bacteria. Microbiome

3:14

Bates ST, Berg-Lyons D, Caporaso JG, Walters WA, Knight R, Fierer N (2011) Examining the

global distribution of dominant archaeal populations in soil. ISME J 5:908–917

Bernhard AE, Dwyer C, Idrizi A, Bender G, Zwick R (2015) Long-term impacts of disturbance

on nitrogen-cycling bacteria in a New England salt marsh. Front Microbiol 6:46

Biddle JF, Lipp JS, Lever MA, Lloyd KG, Sorensen KB, Anderson R, Fredricks HF, Elvert M,

Kelly TJ, Schrag DP, Sogin ML, Brenchley JE, Teske A, House CH, Hinrichs K-U

(2006) Heterotrophic Archaea dominate sedimentary subsurface ecosystems off Peru.

Proc Natl Acad Sci USA 103:3846–3851

Blazewicz SJ, Barnard RL, Daly RA, Firestone MK (2013) Evaluating rRNA as an indicator of

microbial activity in environmental communities: limitations and uses. ISME J 7:2061-

2068

Bowen JL, Byrnes JEK, Weisman D, Colaneri C (2013) Functional gene pyrosequencing and

network analysis: An approach to examine the response of denitrifying bacteria to

increased nitrogen supply in salt marsh sediments. Front Microbiol 4:1–12

Bowen JL, Ward BB, Morrison HG, Hobbie JE, Valiela I, Deegan LA, Sogin M (2011)

Microbial community composition in sediments resists perturbation by nutrient

enrichment. ISME J 5:1540-1548

116

Bricker SB, Longstaff B, Dennison W, Jones a., Boicourt K, Wicks C, Woerner J (2008) Effects

of nutrient enrichment in the nation’s estuaries: A decade of change. Harmful Algae

8:21–32

Bulseco-McKim A, Giblin AE, Tucker J, Murphy AE, Sanderman J, Hiller-Bittrolf K, Bowen JL

(In Review) Nitrate addition stimulates microbial decomposition of organic matter in salt

marsh sediments. Glob Change Biol

Burdige DJ (2007) Preservation of organic matter in marine sediments: Controls, mechanisms,

and an imbalance in sediment organic carbon budgets? Chem Rev 107:467–485

Burgin AJ, Hamilton SK (2007) Have we overemphasized the role of denitrification in aquatic

ecosystems? A review of nitrate removal pathways. Front Ecol Environ 5:89–96



Canfield D (1989) Sulfate reduction and oxic respiration in marine sediments: implications for

organic carbon preservation in euxinic environments. Deep Sea Res Part A Oceanogr Res

Pap 36:121–138


Press, Boston, MA

Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Owens SM, Betley J,

Fraser L, Bauer M, Gormley N, Gilbert J a, Smith G, Knight R (2012) Ultra-high-


ISME J 6:1621–1624

Carini P, Marsden PJ, Leff JW, Morgan EE, Strickland MS, Fierer N (2016) Relic DNA is

abundant in soil and obscures estimates of soil microbial diversity. Nat Microbiol 2:1–6

Chefetz B, Adani R, Genevini P, Tambone F, Hadar Y, Chen Y (1998) Humic acid

transofrmation during composting of municipal solid waste. J Environ Qual 27:794-800


saline wetland soils. Glob Biogeochem Cycles 17:1-12

Conant RT, Ryan MG, Ågren GI, Birge HE, Davidson EA, Eliasson PE, Evans SE, Frey SD,

Giardina CP, Hopkins FM, Hyvönen R, Kirschbaum MUF, Lavallee JM, Leifeld J,

Parton WJ, Megan Steinweg J, Wallenstein MD, Martin Wetterstedt JÅ, Bradford MA

(2011) Temperature and soil organic matter decomposition rates - synthesis of current

knowledge and a way forward. Glob Chang Biol 17:3392–3404


matter mixtures. Nature 369:304

117



Cuddington K (2012) Legacy effects : The persistent impact of ecological interactions. Biol

Theory 6:203–210






Dini-Andreote F, Brossi MJ de L, Elsas JD van, Salles JF (2016) Reconstructing the genetic

potential of the microbially-mediated nitrogen cycle in a salt marsh ecosystem. Front

Microbiol 7:1–13

Ding G, Novak JM, Amarasiriwardena D, Hung PG, Xing B (2002) Soil organic matter

characteristics as affected by tillage management (2002) Soil Sci Soc Am 66:421-429

Evans SE, Wallenstein MD (2018) Soil microbial community response to drying and rewetting

stress: does historical precipitation regime matter? Biogeochemistry 109:101–116

Eyre BD, Rysgaard S, Dalsgaard T, Christensen PB (2002) Comparison of isotope pairing and

N2:Ar methods for measuring sediment denitrification – assumption, modifications, and

implications. Estuaries 25:1077-1087





questions, and potential solutions. Science 320:889-892

Garrity G, Bell J, Lilburn T (2005) Thiotrichales ord. nov. In: Brenner D (ed) Bergey’s Manual

of Systematic Bacteriology, 2nd edition. Boston, MA

Giauque H, Hawkes C V (2016) Historical and current climate drive spatial and temporal

patterns in fungal endophyte diversity. Fungal Ecol 20:108–114




Gilboa-Garber N (1971). Direct spectrophotometric determination of inorganic sulfide in

118

biological materials and in other complex mixtures. Anal Biochem 43:129–133

Global Administrative Areas (2015) GADM database of Global Administrative Areas, version

2.8. www.gadm.org

Goldfarb KC, Karaoz U, Hanson CA, Santee CA, Bradford MA, Treseder KK, Wallenstein MD,

Brodie EL (2011) Differential growth responses of soil bacterial taxa to carbon substrates

of varying chemical recalcitrance. Front Microbiol 2:1–10

Graves CJ, Makrides EJ, Schmidt VT, Giblin AE, Cardon ZG, Rand DM (2016) Functional

responses of salt marsh microbial communities to long-term nutrient enrichment. Appl


Guibert LM, Loviso CL, Borglin S, Jansson JK, Dionisi HM, Lozada M (2016) Diverse bacterial

groups contribute to the alkane degradation potential of chronically polluted Subantarctic

coastal sediments. Microb Ecol 71:100–112

Hanson CA, Fuhrman JA, Horner-Devine MC, Martiny JBH (2012) Beyond biogeographic

patterns: Processes shaping the microbial landscape. Nat Rev Microbiol 10:497–506

Hardison AK, Algar CK, Giblin AE, Rich JJ (2015) Influence of organic carbon and nitrate

loading on partitioning between dissimilatory nitrate reduction to ammonium (DNRA)

and N2 production. Geochim Cosmochim Acta 164:146–160

Hartman WH, Richardson CJ, Vilgalys R, Bruland GL (2008) Environmental and anthropogenic

controls over bacterial communities in wetland soils. Proc Natl Acad Sci USA

105:17842–17847

Herbert RA (1999) Nitrogen cycling in coastal marine ecosystems. FEMS Microbio Rev 23:563-

590

Hijmans RJ, van Etten J (2012) Raster: Geographic analysis and modeling with raster data. R

package version 2.0-12

Hobbie SE, Eddy WC, Buyarski CR, Adair EC, Ogdahl ML, Weisenhorn P, Monographs SE,

August N, Hobbie SE, Eddy WC, Buyarski CR, Adair EC, Ogdahl ML (2016) Response

of decomposing litter and its microbial community to multiple forms of nitrogen

enrichment. Ecol Monogr 82:389–405

Howarth RW (1984) The ecological significance of sulfur in the energy dynamics of salt marsh

and coastal marine sediments. Biogeochemistry 1:5-27

Howarth RW, Teal JM (1979) Sulfate reduction in a New England salt marsh. Limnol

Oceanogr 26:350-360

Imhoff J (2005) Bergey’s Manual of Systematic Bacteriology. In: Brenner D, Krieg N, Staley J,

119

Garrity G (eds) Bergey’s Manual of Systematic Bacteriology, 2nd edition. New York,

NY

Jones SE, Lennon JT (2010) Dormancy contributes to the maintenance of microbial diversity.

Proc Natl Acad Sci USA 107:5881-5886

Kana TM, Darkangelo C, Hunt MD, Oldham JB, Bennett GE, Cornwell JC (1994) Membrane

inlet mass spectrometer for rapid high-precision determination of N2, O2, and Ar in

environmental water samples. Anal Chem 66:4166-4170



Kearns PJ, Angell JH, Howard EM, Deegan LA, Stanley RHR, Bowen KL (2016) Nutrient

enrichment induces dormancy and decreases diversity of active bacteria in salt marsh

sediments. Nat Commun 7:12881

Kearns P, Bulseco-McKim A, Hoyt H, Angell J, Bowen J (2018) Nutrient enrichment alters salt

marsh fungal communities and promotes putative fungal denitrifiers. Microb Ecol. In

press.

Kirschbaum M (1995) The temperature dependence of soil organic matter decomposition, and

the effect of global warming on soil organic C storage. Soil Biol Biochem 27:753-760

Kirwan ML, Langley JA, Guntenspergen GR, Megonigal JP (2013) The impact of sea-level rise

on organic matter decay rates in Chesapeake Bay brackish tidal marshes. Biogeosciences

10:1869–1876

Klepac-Ceraj V, Bahr M, Crump BC, Teske AP, Hobbie JE, Polz MF (2004) High overall

diversity and dominance of microdiverse relationship in salt marsh sulphate-reducing

bacteria. Environ Microbiol 6:686–698

Knorr M, Frey SD, Curits PS (2005) Nitrogen additions and litter decomposition: A meta-

analysis. Ecology 86:3252-3257


via denitrification and DNRA in salt marsh sediments. Limnol Oceaogr 55:789-802



19:1495–1503

Lehmann J, Kleber M (2015) The contentious nature of soil organic matter. Nature 528:60–68

Lennon JT, Jones SE (2011) Microbial seed banks: The ecological and evolutionary implications

of dormancy. Nat Rev Microbiol 9:119–130

120

Liaw A, Wiener M (2002) Classification and regression by randomForest. R news 2:18–22



Lu XM, Lu PZ (2014) Characterization of bacterial communities in sediments receiving various

wastewater effluents with high-throughput sequencing analysis. Microb Ecol 67:612–

623

Lundstrum A, Aoki LR (2016) Oxygen interference with membrane inlet mass spectrometry may

overestimate denitrification rates calculated with isotope pairing technique. Limnol

Oceanogr: Meth 14:425-431

Mahmoudi N, Robeson MS, Castro HF, Fortney JL, Techtmann SM, Joyner DC, Paradis CJ,

Pfiffner SM, Hazen TC (2015) Microbial community composition and diversity in

Caspian Sea sediments. FEMS Microbiol Ecol 91:1–11

Mandal S, Van Treuren W, White RA, Eggesbø M, Knight R, Peddada SD (2015) Analysis of

composition of microbiomes: A novel method for studying microbial composition.

Microb Ecol Health D 26:27663

Margenot AJ, Calderon FJ, Bowles TM, Parikh SJ, Jackson LE (2015) Soil organic matter

functional group composition in relation to organic carbon, nitrogen, and phosphorus

fractions in organically managed tomato fields. Soil Sci Soc Am J 79:772-782

Marin-Spiotta E, Gruley KE, Crawford J, Atkinson EE, Miesel JR, Greene S, Cardona-Correa C,

Spencer RGM (2014) Paradigm shifts in soil organic matter research affect

interpretations of aquatic carbon cycling: transcending disciplinary and ecosystem

boundaries. Biogeochemistry 117:279-297

Martiny JBH, Bohannan BJM, Brown JH, Colwell RK, Fuhrman JA, Green JL, Horner-Devine

MC, Kane M, Krumins JA, Kuske CR, Morin PJ, Naeem S, Øvreås L, Reysenbach AL,

Smith VH, Staley JT (2006) Microbial biogeography: Putting microorganisms on the

map. Nat Rev Microbiol 4:102–112

Mason OU, Hazen TC, Borglin S, Chain PS, Dubinsky EA, Fortney JL, Han J, Holman HY,

Hultman J, Lamendella R, Mackelprang R, Malfatti S, Tom LM, Tringe SG, Woyke T,

Zhou J, Rubin EM, Jansson JK (2012) Metagenome, metatranscriptome and single-cell

sequencing reveal microbial response to Deepwater Horizon oil spill. ISME J 6:1715–

1727

McCarren J, Becker JW, Repeta DJ, Shi Y, Young CR, Malmstrom RR, Chisholm SW, DeLong

EF (2010) Microbial community transcriptomes reveal microbes and metabolic pathways

associated with dissolved organic matter turnover in the sea. Proc Natl Acad Sci USA

107:16420–16427

121


Knight R, Hugenholtz P (2012) An improved Greengenes taxonomy with explicit ranks

for ecological and evolutionary analyses of bacteria and archaea. ISME J 6:610–618

Mettel C, Kim Y, Shrestha PM, Liesack W (2010) Extraction of mRNA from soil. Appl Environ


Middelburg JJ (1989) A simple rate model for organic matter decomposition in marine

sediments. Geochim Cosmochim Acta1 53:1577–1581

Morris JT (1991) Effects of nitrogen loading on wetland ecosystems with particular reference to

atmospheric deposition. Annu Rev Ecol Syst 22:257–279

Morris JT, Sundberg K, Hopkinson CS (2013) Salt marsh primary production and its responses

to relative sea level and nutrients in estuaries at Plum Island, Massachusetts, and North

Inlet, South Carolina, USA. Oceanography 26:78–84

Mueller P, Schile-Beers LM, Mozdzer TJ, Chmura GL, Dinter T, Kuzyakov Y, Groot AV De,

Esselink P, Smit C, D’Alpaos A, Ibanez C, Lazarus M, Neumeier U, Johnson BJ,

Baldwin AH, Yarwood SA, Montemayor DI, Yang Z, Wu J, Jensen K, Nolte S (2018)

Global change effects on decomposition processes in tidal wetlands : implications from a

global survey using standardized litter. Biogeosciences 15:3189–3202

Nielsen KM, Johnsen PJ, Bensasson D, Daffonchio D (2007) Release and persistence of

extracellular DNA in the environment. Environ Biosaf Res 6:37–53

Nixon SW, Oviatt CA, Frithsen J, Sullivan B (1986) Nutrients and the productivity of estuarine

and coastal marine ecosystems. J Limnol Soc S Afr 12:43-71

Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O'hara B,

Simpson GL, Solymos P, Stevens MHH, Szoecs E, Wagner H (2017) Vegan: Community

ecology package version 2.4-3

Oni OE, Schmidt F, Miyatake T, Kasten S, Witt M, Hinrichs KU, Friedrich MW (2015)

Microbial communities and organic matter composition in surface and subsurface

sediments of the Helgoland mud area, North Sea. Front Microbiol 6:1–16




sediment reactors in biogeochemical kinetics: Methodology and examples of

applications. Mar Chem 106:256–271

Papp K, Hungate BA, Schwartz E (2018) Microbial rRNA synthesis and growth compared

122

through quantitative stable isotope probing with H218O. Appl Environ Microbiol 84:1-11

Parikh SJ, Goyne KW, Margenot AJ, Mukome FND, Calderon FJ (2014) Soil chemical insights

provided through vibrational spectroscopy. Adv Agron 126:1-148

Peng X, Ji Q, Angell JH, Kearns PJ, Yang HJ, Bowen JL, Ward BB (2016) Long-term

fertilization alters the relative importance of nitrate reduction pathways in salt marsh

sediments. J Geophys Res Biogeochem 121:2082-2095

Pérez-Pantoja D, Donoso R, Agulló L, Córdova M, Seeger M, Pieper DH, González B (2012)

Genomic analysis of the potential for aromatic compounds biodegradation in

Burkholderiales. Environ Microbiol 14:1091–1117

Rasigraf O, Schmitt J, Jetten MSM, Lüke C (2017) Metagenomic potential for and diversity of

N-cycle driving microorganisms in the Bothnian Sea sediment. Microbiology 6:1–13

Reese BK, Witmer AD, Moller S, Morse JW, Mills HJ (2013) Molecular assays advance

understanding of sulfate reduction despite cryptic cycles. Biogeochemistry 118:307–319



Saito T, Ishii S, Otsuka S, Nishiyama M, Senoo K (2008) Identification of novel

betaproteobacteria in a succinate-assimilating population in denitrifying rice paddy soil

by using stable isotope probing. Microbes Environ 23:192–200

Steven B, Hesse C, Soghigian J, Gallegos-Graves LV, Dunbar J (2017) Simulated rRNA/DNA

ratios show potential to misclassify active populations as dormant. Appl Environ

Microbiol 83:1-11

Solórzano L (1968) Determination of ammonia in natural waters by the phenolhypochlorite

method. Limnol Oceanogr 14:799-801

ter Braak CJF, Verdonschot PFM (1995) Canonical correspondence analysis and related

multivariate methods in aquatic ecology. Aquat Sci 57:255–289

Thomas F, Giblin AE, Cardon ZG, Sievert SM (2014) Rhizosphere heterogeneity shapes

abundance and activity of sulfur-oxidizing bacteria in vegetated salt marsh sediments.

Front Microbiol 5:309

Tiedje J (1988) Ecology of the denitrification and dissimilatory nitrate reduction to ammonium.

In: Zehnder A (ed) Biology of Anaerobic Microorganisms. Wiley and Sons, New York.

NY, p 179–244

Treseder KK, Kivlin SN, Hawkes CV (2011) Evolutionary trade-offs among decomposers

determine responses to nitrogen enrichment. Ecol Lett 14:933-938

123




Valiela I, Teal J, Sass W (1975) Production and dynamics of salt marsh vegetation and the

effects of experimental treatment with sewage sludge. Biomass, production and species

composition. Br Ecol Soc 12:973–981

Veum KS, Goyne KW, Kremer RJ, Miles RJ, Sudduth KA (2014) Biological indicators of soil

quality and soil organic matter characteristics in an agricultural management continuum.

Biogeochemistry 117:81-99


diagenetic status in marine organic matter. Geochim Cosmochim Acta 61:5363–5369




Watson EB, Oczkowski AJ, Wigand C, Hanson AR (2014) Nutrient enrichment and precipitation

changes do not enhance resiliency of salt marshes to sea level rise in the Northeastern

US. Clim Change 125:501–509

Weiss S, Xu ZZ, Peddada S, Amir A, Bittinger K, Gonzalez A, Lozupone C, Zaneveld JR,

Vázquez-Baeza Y, Birmingham A, Hyde ER, Knight R (2017) Normalization and

microbial differential abundance strategies depend upon data characteristics. Microbiome

5:1–18

Weston NB, Vile MA, Neubauer SC, Velinsky DJ (2010) Accelerated microbial organic matter

mineralization following salt-water intrusion into tidal freshwater marsh soils.

Biogeochemistry 102:135–151

Westrich JT, Berner RA. (1984) The role of sedimentary organic matter in bacterial sulfate


Wigand C, Brennan P, Stolt M, Holt M, Ryba S (2009) Soil respiration rates in coastal

marshes subject to increasing watershed nitrogen loads in southern New England, USA.

Wetlands 29:952–963

Wilson CA, Hughes ZJ, FitzGerald DM, Hopkinson CS, Valentine V, Kolker AS (2014)

Saltmarsh pool and tidal creek morphodynamics: Dynamic equilibrium of northern

latitude saltmarshes? Geomorphology 213:99–115

Wu L, Hui L, Wang X, Li J, Yu J, Zhao J (2013) Abundance and composition of ammonia-

oxidizing bacteria in degraded wetland. Soil Res 51:554–560

124

Yin GL, Hou L, Liu M, Liu Z, Gardner WS (2014) A novel membrane inlet mass sepctrometer

method to measure 15NH4+ for isotope-enrichment experiments in aquatic ecosystems.

Environ Sci Technol 48:9555-9562

Zedler JB, Kercher S (2005) Wetland resources: Status, Trends, Ecosystem Services, and

Restorability. Annu Rev Environ Resour 30:39–74

125

Tables

Tab

le 1

. A

ver

age

(± S

EM

) ch

arac

teri

stic

s of

each

stu

dy s

ite

pri

or

to s

tart

of

the

dec

om

posi

tion e

xper

imen

t.

Dif

fere

nt

lett

ers

ind

icat

e si

gnif

ican

ce (

α =

0.0

5)

acco

rdin

g t

o a

Tukey’s

HS

D t

est.

126

Tab

le 2

. A

ver

age

± S

E r

ates

(nm

ol

cm-3

hr-1

) fo

r den

itri

fica

tion (

DN

F)

and d

issi

mil

atory

nit

rate

red

uct

ion t

o

amm

on

ium

(D

NR

A)

po

tenti

al m

easu

red a

s 2

9+

30N

2 p

roduct

ion f

rom

15N

O3

- addit

ion a

long w

ith t

he

DN

F:D

NR

A

rati

o, n

itra

te r

edu

ctio

n r

ates

(N

RR

) in

nm

ol

cm-3

hr-1

, an

d r

ange

of

pro

port

ion o

f N

RR

acc

ounte

d f

or

by D

NF

+

DN

RA

. D

iffe

rent

lett

ers

ind

icat

e si

gnif

ican

ce (

α =

0.0

5)

by s

ite

acco

rdin

g t

o a

Tu

key’s

HS

D t

est.

N

A =

no d

ata.

127

Tab

le 3

. A

ver

age

(± S

EM

) bulk

den

sity

, (%

) org

anic

mat

ter,

% c

arbon,

% n

itro

gen

, m

ola

r C

:N, %

sulf

ur

(N=

4),

and 1

6S

rR

NA

gen

e co

pie

s p

er g

ram

wet

wei

ght

mea

sure

d a

t th

e en

d o

f th

e dec

om

po

siti

on e

xper

imen

t. D

iffe

rent

lett

ers

indic

ate

signif

ican

ce (

α =

0.0

5)

acco

rdin

g t

o a

Tukey’s

HS

D.

128

Figures

Fig. 1. Location of my study sites in northeastern Massachusetts, USA: West Creek (Reference;

42.759 N, 70.891 W), Sweeney Creek (13-year enriched; 42.722 N, 70.847 W), and Greenwood

Creek (40-year enriched; 42.690 N, 70.819 W). Maps were generated by downloading data from

the Database of Global Administrate Areas (GADM; Global Administrative Areas) using the

raster package in R (Hijmans & Jacob van Etten 2012).

129

Fig. 2. (A) Boxplot of Index II, a measure of recalcitrance, across sites. Boxes represent 25% to

75% quartiles. The solid black line is the median value, and the whiskers are upper and lower

extremes. Black dots represent values for each individual reactor (n = 4), and asterisk indicates

significant difference between groups. (B) Baseline corrected mean mid-IR spectra ± SE of each

site before the start of the experiment.

130

Fig. 3. (A) Constrained correspondence analysis and (B) top 20 most differentially abundant

ASVs among sites from the field experiment according to ANCOM analysis.

131

Fig. 4. Average (±SE) dissolved inorganic carbon (DIC) production over time (days) across three

sites that correspond to different levels and durations of chronic nitrate exposure (panels A-C; n

= 4).

132

Fig. 5. Cumulative DIC production. Boxes represent 25% to 75% quartiles. The solid black line

is the median value, and the whiskers are upper and lower extremes. Black dots represent values

for each individual reactor (n = 4). Letters represent statistically different DIC production across

sites from a Tukey’s HSD test corrected for multiple comparisons and asterisks indicate a

significant difference between treatments.

133

Fig. 6. (A) Average (± SE) nitrate reduction rate over time and (B) total nitrate reduction across

sites. (C) Average (± SE) sulfide production rate over time and (D) total sulfate reduction across

sites.

134

Fig. 7 (A) principal coordinates analysis of Mid-IR spectra; (B) Pearson’s correlation

coefficients plotted against wavenumber representing regions most discriminating between the

two axes shown in A; and (C) Total DIC production per site as a function of the recalcitrance

index (Index II). Dotted lines in (B) indicate functional group assignments as follows: 840-920

and 1650 cm-1= aromatic C and lignin-type signatures, 1080 cm-1 = polysaccharides, and 2850-

2924 cm-1 = aliphatic C.

135

Fig. 8. (A) PCoA based on weighted UniFrac similarity after the experiment according to site

(color) and treatment (shape). (B) Average pairwise weighted UniFrac dissimilarities between

plus-NO3- and unamended treatment.

136

Fig. 9. Stacked bar plots of the top ten ASVs most important in discriminating between plus-

NO3- and unamended treatments aggregated at the order level from (A) reference (B) 13-year

enriched and (C) 40-year enriched sites as a result of random forest analysis. Orders marked with

an asterisk indicate importance in at least two sites. Inlaid box in (C) represents relative

abundance of remaining 9 taxa after order Desulfobacterales is removed from analysis. See Table

S3 for additional taxonomic information.

137

Fig. 10. ASV’s whose activity ratio was significantly different (p<0.05) between plus-NO3- and

unamended treatments per site aggregated at the order level. Darker colors indicate more active

taxa, with white boxes represent inactive taxa (ratio < 1). See Table S4 for additional taxonomic

information.

138

Supplemental Tables

Table S1. Functional group assignments based on Parikh et al. (2014) and modified from

Margenot et al. (2015) to evaluate FT-IR spectra using Index II metric. ν = stretching vibration;

νas = asymmetric stretching vibration; νs = symmetric stretching vibration; δ = bending vibration.

Band (cm-1) Assignment

3400 ν(N-H), ν(O-H)

2924 aliphatic νas(C-H)

2850 aliphatic νs(C-H)

1650 aromatic ν(C = C)



1270 phenol νas(C-O), carboxylic acid ν(C-O)



920 aromatic δ(C-H)

840 aromatic δ(C-H), less substituted

139

Table S2. Flow property information, including average flow rate (±SD), porosity, and linear

flow velocity.

Sample Flow Rate (mL

min-1) Porosity

Linear Flow Velocity

(mL hr-1 cm-2)

Plus-NO3-

Reference

Core 1 0.090 (0.007) 0.44 0.701

Core 2 0.091 (0.001) 0.43 0.742

Core 3 0.074 (0.003) 0.44 0.513

Core 4 0.083 (0.002) 0.48 0.591

13-year enriched

Core 1 0.088 (0.004) 0.46 0.677

Core 2 0.091 (0.001) 0.48 0.682

Core 3 0.089 (0.013) 0.51 0.477

Core 4 0.080 (0.004) 0.43 0.770

40-year enriched

Core 1 0.073 (0.001) 0.59 0.458

Core 2 0.082 (0.006) 0.39 0.698

Core 3 0.086 (0.003) 0.40 0.683

Core 4 0.082 (0.005) 0.62 0.484

Unamended Treatment

Reference

Core 1 0.082 (0.015) 0.42 0.803

Core 2 0.085 (0.011) 0.41 0.834

Core 3 0.060 (0.002) 0.40 0.693

Core 4 0.076 (0.012) 0.45 0.692

13-year enriched

Core 1 0.082 (0.002) 0.42 0.792

Core 2 0.087 (0.003) 0.42 0.827

Core 3 0.065 (0.004) 0.50 0.679

Core 4 0.087 (0.009) 0.41 0.729

40-year enriched

Core 1 0.071 (0.003) 0.57 0.484

Core 2 0.072 (0.001) 0.36 0.855

Core 3 0.073 (0.002) 0.41 0.797

Core 4 0.079 (0.005) 0.60 0.517

140

Table S3. Top ten ASVs most important in discriminating between plus-NO3- and unamended

treatments at each site, listed in decreasing order of importance. Asterisks indicate ASVs shared

between at least two sites and bolded indicate taxa that increased in relative abundance with

NO3- addition.

Site Kingdom Phylum Class Order

Δ Rel

Abun.

Ref. Archaea Euryarchaeota Thermoplasmata CCA47* 0.11%

Bacteria Proteobacteria Deltaproteobacteria Myxococcales 0.63%

Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales -0.01%

Bacteria Proteobacteria Deltaproteobacteria Desulfarculales 3.42%

Bacteria Chloroflexi Anaerolineae envOPS12 -0.79%

Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales -0.44%

Bacteria Proteobacteria Deltaproteobacteria Myxococcales 0.04%

Archaea Crenarchaeota MCG B10 0.20%

Bacteria Tenericutes Mollicutes unclassified 0.26%

Bacteria SAR406 AB16 Arctic96B-7 0.11%

13-year Bacteria Proteobacteria Gammaproteobacteria Oceanospirillales 0.34%

enriched Bacteria WS3 unclassified unclassified -0.16%

Bacteria Proteobacteria Gammaproteobacteria Thiotrichales 2.51%

Bacteria Chloroflexi Anaerolineae SBR1031 -0.04%

Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales* -6.96%

Bacteria Plancotymycetes unclassified unclassified -0.02%

Bacteria Proteobacteria Deltaproteobacteria unclassified -0.01%

Bacteria Proteobacteria Gammaproteobacteria Chromatiales 9.89%

Bacteria Chloroflexi Dehaloccoidetes Dehalococcoidales -0.05%

Archaea Euryarchaeota Thermoplasmata CCA47* -0.06%

40-year Bacteria Clamydiae Chlamydiia Chlamydiales -0.04%

enriched Bacteria TM6 unclassified unclassified -0.01%

Bacteria Firmicutes Clostridia Clostridiales 0.03%

Bacteria Proteobacteria Betaproteobacteria Burkholderiales -0.02%

Bacteria Plancotymycetes Agg27 unclassified 0.02%

Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales*

-

10.62

%

Bacteria Chloroflexi Anaerolineae DRC31 0.02%

Bacteria Bacteroidetes unclassified unclassified -0.13%

Bacteria FCPU426 unclassified unclassified -0.08%

Bacteria Proteobacteria Gammaproteobacteria Chromatiales 0.01%

141

Table S4. Taxa whose activity level (± SE), as assessed by 16S rRNA:16S rRNA gene ratios

averaged across each core, significantly differed (p<0.05) between plus-NO3- and unamended

treatments at the end of the decomposition experiment on a per site basis. Bolded text indicates

greater ratios under that experimental condition.

Site Phylum Class Order Nitrate Unamen. p-value

Ref. Proteobacteria Gammaproteobacteria Thiotrichales

24.16

(18.13) 0 (0) 0.025

Proteobacteria Alphaproteobacteria NA

14.12

(12.9) 0 (0)

0.033

Proteobacteria

Gammaproteobacteria Thiotrichales

11.86

(19.4) 0 (0) 0.022

Proteobacteria

Betaproteobacteria NA

1.05

(0.70) 0 (0) 0.025

Proteobacteria

Gammaproteobacteria Alteromonadales

1.04

(0.96) 0 (0) 0.023

Proteobacteria

Gammaproteobacteria Chromatiales

1.00

(0.38) 0 (0) 0.031

Proteobacteria

Gammaproteobacteria Chromatiales

0.88

(0.13)

5.97

(8.70) 0.019

Chloroflexi

Anaerolineae GCA004

0.69

(0.25)

1.87

(0.90) 0.032

Proteobacteria

Betaproteobacteria NA

0.49

(0.40)

1.59

(0.39) 0.032

Proteobacteria Deltaproteobacteria Desulfobacterales

0.40

(0.53)

9.28

(14.48) 0.038

Proteobacteria

Deltaproteobacteria Desulfovibrionales

0.29

(0.23)

1.15

(0.91) 0.03

Proteobacteria

Deltaproteobacteria Myxococcales

0.19

(0.26)

13.45

(7.93) 0.028

Firmicutes

Clostridia Clostridiales

0.11

(0.14)

5.47

(3.36) 0.025

Proteobacteria Alphaproteobacteria NA

0.11

(0.06)

2.59

(4.28) 0.035

Proteobacteria Deltaproteobacteria Myxococcales 0 (0)

13.73

(25.51) 0.045

Poribacteria

NA NA 0 (0)

22.56

(21.04) 0.019

Proteobacteria

Deltaproteobacteria NA 0 (0)

6.80

(10.15) 0.022

13-year Proteobacteria Alphaproteobacteria Kiloniellales

8.34

(4.24) 0 (0)

0.031

Proteobacteria Deltaproteobacteria DTB120

7.65

(13.57) 0 (0)

0.021

Planctomycetes Planctomycetia Pirellulales

56.00

(20.07)

13.00

(15.10) 0.036

Proteobacteria Gammaproteobacteria Thiotrichales

38.93

(76.72) 0 (0)

0.027

Proteobacteria Gammaproteobacteria Chromatiales

175.10

(349.27) 0 (0)

0.03

Proteobacteria Gammaproteobacteria Oceanospirillales

17.48

(24.30) 0 (0)

0.032

142

Proteobacteria Deltaproteobacteria Desulfobacterales

39.25

(46.13)

107.25

(11.59) 0.029

GN04 GN15 NA

0.24

(0.29)

8.59

(14.94) 0.034

Proteobacteria Deltaproteobacteria Desulfarculales

0.22

(0.45)

23.09

(26.58) 0.03

GN04 GN15 NA

0.18

(0.21)

9.15

(17.23) 0.033

Proteobacteria Betaproteobacteria SBla14

0.10

(0.12)

4.70

(8.86) 0.028

Proteobacteria Epsilonproteobacteria Campylobacterales

0.05

(0.10)

13.44

(26.38) 0.032

40-year Bacteroidetes Rhodothermi Rhodothermales

9.10

(10.36) 0 (0)

0.025

Bacteroidetes Rhodothermi Rhodothermales

3.31

(0.66) 0 (0)

0.021

Actinobacteria Acidimicrobiia Acidimicrobiales

2.52

(4.33) 0 (0)

0.036

Proteobacteria Deltaproteobacteria Myxococcales

15.94

(14.59) 0 (0)

0.021

Proteobacteria Alphaproteobacteria Kiloniellales

11.61

(19.59) 0 (0)

0.032

Proteobacteria Deltaproteobacteria Desulfuromonadales

1.58

(0.37)

0.59

(0.22) 0.021

Proteobacteria Zetaproteobacteria Mariprofundales

1.52

(0.41) 0 (0)

0.029

Bacteroidetes Flavobacteriia Flavobacteriales

1.08

(0.66) 0 (0)

0.03

Proteobacteria Deltaproteobacteria Syntrophobacterales

9.04

(9.79)

56.25

(28.04) 0.023

Firmicutes Clostridia Clostridiales

0.82

(0.64)

9.22

(10.55) 0.023

Proteobacteria Alphaproteobacteria Rhizobiales

0.32

(0.41)

42.94

(48.13) 0.022

Proteobacteria Epsilonproteobacteria Campylobacterales 0 (0)

5.11

(8.59) 0.024

Proteobacteria Deltaproteobacteria Myxococcales 0 (0)

29.5

(18.63) 0.019

Proteobacteria Deltaproteobacteria Desulfobacterales 0 (0)

3.28

(3.83) 0.028

Proteobacteria Deltaproteobacteria Desulfobacterales 0 (0)

6.93

(5.02) 0.029

143


Fig. S1. Bromide breakthrough curve with the ratio of [Br-]initial/[Br-]final on the y-axis to confirm

uniform and regular flow in each reactor. Dotted line indicates a ratio of 1, where the initial and

final bromide concentration are equal, indicating breakthrough.

144

Fig. S2. Average ± SE rates of ammonium production for (A) plus-NO3

- and (B) unamended

treatment throughout experiment.

145

Chapter 3: Stochastic processes shape microbial communities in deep salt marsh sediments

In collaboration with: Thomas J. Mozdzer and Donald C. Barber

Abstract:

Salt marshes can sequester carbon at rates that are an order of magnitude greater than

terrestrial counterparts due to slow rates of decomposition. Microbes mediate this critical

ecosystem service, yet we know virtually nothing about their distribution and interaction with

buried organic matter in deep salt marsh sediments. Further, there is evidence that nutrient

enrichment stimulates organic matter decomposition and alters surface sediment microbial

communities, though it is unclear if this pattern holds in deeper salt marsh sediments, where long

term carbon storage occurs. To address these critical knowledge gaps, I collected three-meter-

deep cores spanning ~3000 years of sediment accumulation as part of a long-term nutrient

enrichment experiment at the Plum Island LTER. I characterized sediment organic matter in

parallel with high throughput sequencing of the 16S rRNA gene/16S rRNA to assess microbial

community diversity, abundance, structure, and assembly along a depth gradient between an

experimentally nutrient-enriched marsh and its paired reference marsh. I found that both

microbial diversity and gene abundance decreased with depth, with diverging patterns between

the 16S rRNA gene and 16S rRNA in deeper sediments that suggest significant rates of

microbial inactivity at depth. Depth and associated changes in organic matter explained a large

portion of microbial community structure in shallower sediments, with patterns driven by shifts

in rare taxa. However, in deeper sediments beyond the rooting zone, changes to the community

could no longer be attributed to parameters I measured, likely due to a transition from

deterministic to stochastic assembly at depth. The only detectable differences between the

reference and enriched marshes occurred in deeper sediments, suggesting that these differences

146

resulted from stochastic processes rather than experimental nutrient enrichment. Overall, this

work highlights the stability of salt marsh sediments, and provides novel information on the

microbes mediating carbon cycling in these critical ecosystems.

Introduction:

Salt marshes, and other vegetated coastal habitats (e.g. seagrasses and mangroves),

contribute to ~50% of carbon storage in marine sediments, despite occupying only a small

portion of coastal area (Duarte et al. 2013, Najjar et al. 2018). These “blue carbon” systems

efficiently store carbon due to high aboveground productivity (Mendelssohn & Morris 2002),

and because rates of decomposition are inhibited by anoxic, water-logged soils. As such, they

bury organic matter (OM) over millennial time scales without becoming saturated (Zedler &

Kercher 2005, Mcleod et al. 2011). Microbes largely mediate the amount of carbon decomposed

and/or buried in these deep sediments (Benner at al. 1984, Sutton-Grier et al. 2011), and while

considerable work has gone into understanding microbial distribution and activity in deep

subseafloor marine sediments (Inagaki et al. 2015, Oni et al. 2015, Starnawski et al. 2016, Walsh

et al. 2016, Petro et al. 2017, Marshall et al. 2018), lake sediments (Vueillemin et al. 2018), and

terrestrial soils (Fierer et al. 2003, Hartmann et al. 2009, Eilers et al. 2012, Carini et al. 2016),

very little is known regarding factors that control microbial communities in deep salt marsh

sediments. Given the increased prevalence of salt marsh restoration to promote carbon storage

and other beneficial ecosystem services (Warren et al. 2002, Macreadie et al. 2017, Narayan et

al. 2017), it is critical that we better understand what controls the assembly, structure, and

function of microbial communities in deep salt marsh sediments and examine how their

distribution changes under shifting environmental conditions.

147

There are many factors that can control the vertical distribution of subsurface salt marsh

sediment microbial communities. OM quality and quantity, since it serves as an electron donor

fueling heterotrophic microbial metabolisms, may select for specific microbes based on the

complexity of its chemical composition (Walsh et al. 2016). At the surface, easily degradable

OM is preferentially oxidized at higher rates due to the lower energetic demand it imparts on the

decomposer community (Cowie & Hedges 1994, Wakeham et al. 1997), with net rates of

heterotrophic respiration decreasing with depth (Middelburg 1989, Walsh et al. 2016). The less

labile forms of OM left behind accumulate and eventually become buried (Hedges et al. 2000).

As a result, total microbial biomass, diversity, and decomposition rates tend to decrease with

increasing sediment depth because microbes still need to meet their energetic demands but with

more recalcitrant OM (Burdige 2007, Middelburg 1989, Parkes et al. 1994, Westrich & Berner

1984). While some microbial groups contain special adaptations for complex OM oxidation (e.g.

Chloroflexi, candidate divisions JS1, OP9, and archaeal members of the Miscellaneous

Crenarchaeota Group (MCG) and Marine Benthic Group B (MBG; Biddle et al. 2006, Inagaki et

al. 2006, Teske et al. 2008, Kubo et al. 2012), competition for what limited resources remain

typically leads to smaller and less diverse microbial communities.

Another factor that controls microbial patterns in the subsurface environment is the

availability of electron acceptors that fuel heterotrophic metabolisms and the geochemical

zonation that occurs as a consequence. Microbes preferentially reduce electrons that yield the

most energy first (greater Gibb’s free energy; ΔGº), leading to a predictable sequence of

metabolic processes that starts with oxygen reduction, and proceeds through manganese, nitrate,

iron, and sulfate reduction, as well as methanogenesis (Froelich et al. 1979, Thamdrup et al.

1994, Canfield et al. 2005). In salt marshes, oxygen is rapidly oxidized within the top few

148

millimeters of the surface (Teal & Kanwisher 1961) unless there are physical modifications to

sediments resulting from either bioturbation (Aller 1994, Aller & Aller 1998, Kristensen &

Holmer 2001) or active diffusion of gaseous oxygen by macrophytic roots (Lee et al. 1999,

Holmer et al. 2002). As a result, in salt marsh sediments, microbial metabolism quickly switches

to anaerobic respiration, which is generally less efficient at oxidizing complex OM (Reddy &

Patrick 1975).

The anaerobic decomposition of OM is important for salt marsh carbon storage and it

also plays a role in structuring the microbial community, either by selecting for taxa that can best

use available electron acceptors, or through shifting the functional capacity of facultative

microbes. For example, sulfate is the dominant electron acceptor in salt marsh sediments due to

its very high concentration in seawater. Sulfate fuels the metabolic processes of sulfate-reducing

microbes and can support up to 70-90% of total sediment respiration (Howarth & Teal 1979,

Howarth 1984). However, nitrate can be another important electron acceptor in salt marshes,

because after oxygen, it is the most energetically favorable electron acceptor and it fuels

ecologically important processes such as denitrification (Kaplan et al. 1979, Valiela & Teal

1979, Seitzinger 1988, Sousa et al. 2012) and dissimilatory nitrate reduction to ammonium

(DNRA; Koop-Jakobsen & Giblin 2010, Giblin et al. 2013). Although nitrate is typically limiting

in coastal waters (Ryther & Dunstan 1979), nutrient enrichment from fertilizer production,

agricultural and urban runoff, enriched groundwater, and atmospheric deposition may increase

its availability to sediment microbes (Galloway et al. 2017). Increased supplies of nitrate can

stimulate respiration of belowground OM by providing additional reducing capacity to the

system (Bulseco-McKim et al. In review), which can lead to lower sediment stability and

potential marsh collapse (Darby & Turner 2008, Deegan et al. 2012, Mueller et al. 2018).

149

Previous studies, however, focused on decomposition in surface sediments, but the effect of

nitrate addition on deeper sediments is unclear.

Microbial communities that are responsible for decomposition in both surface and deep

marsh sediments can be assembled by both deterministic and stochastic processes (Stegen et al.

2012, Mittelbach & Schemske 2015, Petro et al. 2017). Deterministic processes result when

abiotic and biotic factors facilitate environmental filtering of the taxa present (Fierer & Jackson

2006, Nemergut et al. 2013). Stochastic processes, in contrast, result in microbial distribution

patterns indistinguishable from random chance. These processes include drift (random

birth/death events and unpredictable disturbances), speciation/diversification, and dispersal

events (Vellend et al. 2010, Petro et al. 2017). While both deterministic and stochastic processes

can act simultaneously on microbial community structure (Dumbrell et al. 2010, Stegen et al.

2012), their relative importance is unclear. Some studies have found deterministic processes to

dominate in deep sediments where limited resources select for specific taxa (Stegen et al. 2013),

while others have observed increasing stochasticity with sediment depth (Chu et al. 2016,

Tripathi et al. 2017). Disentangling the relative contribution of these processes is critical to

understanding the mechanism of microbial community assembly in marsh sediments and

ultimately, the role that community plays in long-term marsh carbon storage.

There are a number of challenges to linking shifts in subsurface sediment microbial

communities to changes in environmental parameters, including the presence of relic DNA

(Carini et al. 2016) and the proportion of inactive or dormant cells (Lennon & Jones 2011).

Extracellular nucleic acids from dead microbes may persist in soils for extended periods of time

(Levy-Booth et al. 2007), contributing to the estimate that total soil DNA can be up to 80% relic

DNA (Carini et al. 2016, Lennon et al. 2018). Consequently, a large portion of sequence data

150

from sediments may inaccurately represent the microbial community, since much of what is

present is not actively contributing to ecosystem processes. The extent to which relic DNA

skews diversity estimates remains unclear (Lennon et al. 2018). Prior work in salt marshes

documented high rates of inactivity, likely a result of a combination of relic DNA and microbial

dormancy (Kearns et al. 2016). Given the resource limitation in deep salt marsh sediments,

microbial communities likely also exhibit high rates of inactivity, thus obscuring our ability to

document how changes in microbial community structure feedback on ecosystem function.

Analysis of transcripts of the small subunit of prokaryotic ribosomes (16S rRNA) can partially

address this issue due to their correlation with protein synthesis (Kerkhof & Kemp 1999).

Although there are several caveats involved with this approach (Blazewicz et al. 2013, Steven et

al. 2017, Papp et al. 2018), it is essential we disentangle how these active members of the

microbial community, which are directly responsible for critical ecosystem processes, vary in

deep marsh sediments.

I examined OM characteristics and performed high throughput sequencing of the 16S

rRNA gene (total community) and its product, 16S rRNA (potentially active community), on six

deep salt marsh cores collected from two marshes in the Plum Island Ecosystem LTER. My

objectives were to 1) examine differences in OM characteristics with depth and between sites 2)

characterize both the total and potentially active community at these two sites, 3) determine the

assembly mechanisms of the microbial community in deep marsh sediments, and 4) examine

whether these mechanisms differ as a result of experimental nutrient enrichment. I hypothesized

that OM characteristics would vary along a depth gradient and that they would be different

between the two sites as a result of nutrient enrichment. Further, I hypothesized that the

microbial community, particularly the active portion, would vary concurrently with these

151

changes in OM composition. I also hypothesized that the overall microbial community diversity,

biomass, and activity would decrease with depth as there would be more competition for limited

resources in deeper sediments, resulting in deterministic selection for adapted taxonomic groups.

Methods:

Study sites and sample collection

I collected samples at salt marshes associated with the TIDE project, a long-term nutrient

enrichment experiment at the Plum Island Ecosystem LTER (Deegan et al. 2007) in

Northeastern, Massachusetts, USA (42.759 N, 70.891 W) (Fig. 1). In the Spartina patens habitat,

I collected three sediment cores each from the Sweeney Creek salt marsh (“enriched”), which

has experimentally received nitrogen in the form of 70 µM nitrate for 13 years and from its

paired reference site, West Creek (“reference”) using a Russian peat corer until reaching the

point-of-refusal. Core depths were 90, 100, and 240 cm, and 80, 150, and 280 cm for enriched

and reference marshes, respectively. I sectioned each core at 2 cm intervals and from each

section, I subsampled sediment for OM analyses, which I stored at -20°C. I also subsampled

sediment for nucleic acid extraction by homogenizing each 2 cm interval in a 50 mL falcon tube,

flash freezing the sediments in liquid nitrogen, and transferring them to a -80°C freezer for

storage until further analysis.

Organic matter analyses

From each homogenized 2 cm section, I subsampled sediment to determine bulk OM

characteristics. I calculated bulk density by dividing the sediment mass by the total volume of

each section and I determined %OM through loss-on-ignition (LOI) by drying sediment in

crucibles for 24 hours at 105ºC followed by 4 hours at 550ºC in a muffle furnace. From each 10

152

cm interval I dried sediment at 60°C for 48 hours, fumed finely ground sediment with 12N

hydrochloric acid (HCl) for three days, and performed elemental analysis (%C and %N) on a

Perkin Elmer 2400 Series II CHN Analyzer (Perkin Elmer, Waltham, MA) using aspartic acid

(Thermo Scientific, Camrbidge, UK) as a standard after correcting the mass for loss of

carbonate. To prepare samples for carbon isotope analysis (δ13C) of bulk organic matter, I freeze-

dried sediments in a VirTisTM benchtop freeze dryer (SP Scientific, Warminster, PA) and then

finely ground the sediments using a RetschTM ball mill (Verder Scientific, Newtown, PA). I ran

samples at Bryn Mawr College using cavity ring-down (CRDS) laser spectroscopy following

methods outlined in Balslev-Clausen (2013) on a PicarroTM G2201-i CRDS (Picarro Inc, Santa

Clara, CA) with combustion carried out at 980ºC on a Costech ECS 4010 elemental analyzer

(Costech Inc, Valencia, CA) using N2 as the carrier gas and USGS40 (glutamic acid) as a

standard. I report isotope values in δ notation according to the following equation:

Eq. 1 δX = [(Rsample/Rstandard) − 1] × 1000

standardized to Vienna Pee Dee Belemnite (VPDB) with reproducibility of δ13C values <0.2‰

based on repeat analysis of USGS40.

Age dating

I collected four rhizome samples from the two deepest cores at each site (Core 3 at the

reference site and Core 1 at the enriched site) for radiocarbon (14C) analysis to produce an age

date model for each site. All radiocarbon analyses were conducted at the National Ocean

Sciences Accelerator Mass Spectrometry (NOSAMS) facility located at the Woods Hole

Oceanographic Institution (Woods Hole, MA). The data were modeled using the Clam package

in R (Classical Age-Depth Modelling of Cores from Deposits; Blauuw 2010).

Nucleic acid extraction

153

I extracted genomic DNA from approximately 0.25 g sediment using the MoBio®


instructions, and the DNA was eluted into a 35 µL final volume. To extract RNA, I used a

method modified from Mettel et al. (2010), which is specifically designed to handle samples with

high humic acid content. To approximately 0.5 g of sediment, I added 700 µL PBL buffer (5 mM

tris-hydrochloride [pH 5.0], 5 mM ethylenediaminetetraacetic acid disodium salt, 0.1% [wt/vol]

sodium dodecyl sulfate, and 6% [vol/vol] water-saturated phenol), followed by 0.5 g of 0.17 mm

glass beads. After vortexing at maximum speed for 10 minutes, I spun the samples at 20,000 x g

for 30 seconds, and transferred the supernatant to a new tube. I then resuspended the remaining

sediment and glass beads with 700 µL TPM buffer (50 mM tris-hydrochloride [pH 5.0], 1.7%

[wt/vol] polyvinylpyrrolidone, 20 mM magnesium chloride) and vortexed the suspension at

maximum speed for an additional 10 minutes. I spun the sediment at 20,000 x g for 30 seconds

and pooled the supernatant with the supernatant from the previous step. To each tube, I added an

equal volume of phenol:chloroform:isoamyl alcohol (25:24:1 v/v/v), mixed by vortexing at

maximum speed for 30 seconds, and spun at 20,000 x g for 30 seconds. I then transferred the

aqueous layer to a new tube and precipated nucleic acids using 0.7 volumes of 100% isoproponal

and 0.1 volumes sodium acetate [pH 5.7]. After spinning at 20,000 x g for 30 minutes, I

discarded the supernatant, and washed the resulting pellet using 70% ethanol. I loaded the

washed RNA onto an Illustra Autoseq G-50 Spin Column (GE Healthcare) prepared with 500 µl

Q-Sepharose (GE Healthcare; Marlborough, MA), spun the column at 650 x g for 7 seconds, and

eluted from the column three times with 200 µL 1.5 M NaCl (pH 5.5). To precipitate the flow

through, I added 0.7 volumes of 100% isopropanol and 0.1 volumes sodium acetate (pH 5.7),

spun at 20,000 x g, and resuspended the resulting pellet in 50 µL di-ethyl pyrocarbonate-treated

154

(DEPC) water. I checked for DNA contamination in the RNA using general bacterial primers

515F and 806R (Bates et al. 2011) and removed contamination using DNase I (New England

BioLabs, Ipswich, MA). Lastly, I reverse transcribed 2 µL of RNA to cDNA using random

hexamer primers and an Invitrogen Superscript III cDNA synthesis kit for RT-PCR (Life

Technologies, Carlsbad, CA, USA).

Illumina library preparation and sequencing

After confirming the presence of DNA using SYBR Safe (Thermo Fisher Scientific,

Waltham, MA), I amplified DNA and cDNA in triplicate using 12 bp Golay barcoded primers

that target the V4 region of the 16S rRNA gene and contain adaptors for Illumina sequencing

(Caporaso et al. 2012). The PCR reaction was as follows: 12.5 µl Phusion High-Fidelity PCR

Master Mix with HF buffer (Thermo Fisher Scientific, Waltham, MA), 0.25 µl of 0.20 µM 515F

(5’-GTGCCAGCMGCCGCGGTAA-3’) forward primer, 0.25 µl of 0.20 µM 806R (5’-

GGACTAC HVGGGTWTCTAAT-3’) reverse primer, 11 µL of DEPC-treated water, and 1 µl of

DNA or cDNA template. I acknowledge the systematic bias of these primers against the SAR11

clade (Apprill et al. 2015), however, salt marsh sediments typically become anoxic within 2-3

mm of the surface, so these strictly aerobic bacterioplankton should not play a large role in the

microbial community associated with this study (Giovannoni 2017). I gel purified PCR products

using the Qiagen® QIAquick gel purification kit (Qiagen, Valencia, CA) and quantified the

purified product using a Qubit® 3.0 fluorometer (Life Technologies, Thermo Fisher Scientific,

Waltham, MA). After pooling to equimolar amounts, I sequenced the nucleic acids using the

Illumina MiSeq (Illumina, San Diego, CA) platform with paired-end 250 bp V2 chemistry.

Quantitative PCR

155

From each 20 cm interval, I performed quantitative PCR (qPCR) in triplicate using 357F

and 519R primers (Turner et al. 1999) to quantify the abundance of the 16S rRNA gene and 16S

rRNA, on an Aria Mx Real Time PCR instrument (Agilent Technologies). I used the following

reaction: 10 µL of Brilliant III Ultrafast SYBR Green qPCR Master Mix (Agilent Technologies),

0.5 µL of 20 µM of each forward and reverse primer, 8 µL of DEPC-treated water, and 1 µL of

nucleic acid template (either DNA or cDNA). I ran a 16S rRNA-targeted program, consisting of

an initial denaturation step at 95ºC for 3 minutes followed by 40 cycles of 95 ºC for 5 seconds

and 60 ºC for 10 seconds. I acquired data at the end of each cycle and conducted a melt curve to

confirm that the target gene was amplified. Standards were prepared from purified 16S rRNA

gene product, which I quantified and assessed for size with a tape station (Agilent Technologies).

I ran standard curves with each batch of samples, resulting in R2 > 0.95 and qPCR efficiency >

95%. To calculate gene copies per gram wet weight (g ww-1), I multiplied my results by the mass

of sediment extracted and divided that value by the total elution volume during the

extraction/reverse transcription process.

Nucleic acid sequence processing and analysis

I joined paired-end reads using fast-q join (Aronesty et al. 2011) with default parameters

and performed downstream analysis in QIIME2 (version 2018.2; Caporaso et al. 2010, QIIME 2

Development Team). After demultiplexing and quality filtering following parameters

recommended by Bokulich et al. (2013), I inferred amplicon sequence variants (ASVs) using

DADA2 (Callahan et al. 2016) with a maxEE of 2 and the consensus chimera removal method.

To assign taxonomy I used the Naïve Bayes classifier q2-feature-classifier plugin, trained on the

Greengenes 99% OTUs database (version 13-8; McDonald et al. 2012), and filtered out all

sequences matching chloroplasts and mitochondria, as well as sequences occurring only once

156

(singletons). I was left with a total of 5,732,572 sequences for final analysis; 3,629,037

sequences for the 16S rRNA gene and 2,103,535 sequences for 16S rRNA (RNA) that I aligned

using MAFFT v.7 (Katoh & Standley 2013).

Statistical Analyses

To characterize OM across depth and site, I applied linear mixed effects models fit by

maximum likelihood, with core as a random effect, using the lme4 package in R (Bates et al.

2015). I included a null model, a model fit by site or depth only, and an additive and

multiplicative model between these two parameters. I determined the most parsimonious model

by comparing the Akaike’s Information Criterion (AIC) values and performed model selection

using the bbmle package in R (Bolker et al. 2014), which ranks each model according to its

Akaike weight (wi). This metric represents the likelihood of the model normalized by the sum of

all models with values closest to 1 representing the best fit (Johnson & Omland 2004). I

compared models by calculating the difference in AIC (ΔAIC) (Richards 2005).

I normalized the amplicon sequence variant (ASV) tables generated from DADA2 to the

lowest sequencing depth and computed the Shannon diversity index using 10,000 restarts in

QIIME 2 on the 16S rRNA gene and 16S rRNA separately to examine the within-sample

diversity. To test the role of depth on Shannon diversity, I performed an analysis of covariance

(ANCOVA) including site as a fixed effect. I also used a linear mixed effects model fit by

maximum likelihood, with core as a random effect, to investigate the effect of OM characteristics

on Shannon diversity, again examining wi and ΔAIC to assess model appropriateness. To

examine patterns in gene copy number along a depth gradient, I fit a linear model to the 16S

rRNA gene data, and I fit an exponential model to the 16S rRNA data, and compared slopes

between sites using ANCOVA.

157

I then calculated beta diversity to explore between sample diversity using weighted

UniFrac in QIIME 2 on 16S rRNA gene and 16S rRNA ASV tables normalized to the lowest

sequencing depth (Lozupone et al. 2005). I tested for differences in community structure as a

function of depth and site with a PERMANOVA using 999 permutations and repeated the

analysis on tables separated into two depth categories: 0-50 cm (shallow) and 60+ cm (deep). To

better visualize changes in microbial community structure, I plotted the first PCoA axis as a

function of depth and I calculated average dissimilarity between sites, also as a function of depth,

for the top 50 cm. To explore which environmental parameters, other than depth, were

responsible for explaining variation in the microbial community structure, I used a linear mixed

effects model to test the effect of depth, %C, %N, and C:N on PCoA axis 1 for both the shallow

and deep regions of the core, with separate analyses per site in the deeper sediments. Lastly, I

analyzed the correlation between all environmental parameters and microbial community

structure with Mantel tests (Mantel 1967) on Euclidean distance using 999 permutations in the

VEGAN package in R (Okansen et al. 2017) on the full dataset and separately based on the two

depth categories.

To gain a better understanding of the dominant taxa in both shallow and deep sediments,

I performed the following steps separately for each depth category. First, I examined the top 100

most abundant ASVs for the total and active community in the top 50 cm. Next, I performed a

random forest regression against depth with 10,000 trees in the randomForest R package (v4.6-

12; Liaw & Wiener 2002) using ASVs that accounted for at least 0.5% of the dataset as predictor

variables, and estimated model performance using leave-one-out cross-validation in the Caret R

package (v6.0-71; Kuhn et al. 2016). From this model, I was able to identify which taxa were

158

most important in explaining the variance observed within each depth increment by separately

ranking them in decreasing order of % mean squared error explained.

Finally, I was interested in determining how patterns of microbial community assembly

changed with depth. One way to infer the relative influence of deterministic versus stochastic

processes is to examine phylogenetic community composition in relation to environmental

conditions (Fine & Kembel 2011, Stegen et al. 2012). Assuming that more closely related taxa

are more ecologically similar (Andersson et al. 2010, Phillipot et al. 2010), then a community

whose phylogenetic composition does not significantly differ by chance (i.e. phylogenetically

overdispersed) is considered stochastic. In contrast, if deterministic processes dominate, then

observed taxa should be more closely related than expected by chance (i.e. phylogenetically

constrained). To test this, I constructed a phylogenetic tree using FastTree (Price et al. 2010) in

QIIME 2 and calculated the standardized effect size of the mean pairwise distance (SESMPD) with

999 runs and null model = ‘taxa.labels’ using the Picante package in R (Kembel et al. 2010).

This metric is analogous to -1 times the net relatedness index (-NRI), which quantifies the

phylogenetic distance from the root to terminal leaves (Webb et al. 2008). Positive values (> 0)

indicate phylogenetic overdispersion (greater phylogenetic distance than expected according to a

null model) and negative values (< 0) indicate phylogenetic clustering (smaller phylogenetic

distances than expected according to a null model). All statistical analyses were conducted in R

(R Core Team) unless otherwise stated.

Results

Organic matter characteristics and age date modeling

159

After log-transforming bulk density data, a linear mixed effects model showed that depth

best explained patterns for both the reference and enriched site (wi=0.43; Fig. 2A, G; Table 1,

S1). At 250 cm, bulk density more than doubled to 2.03 g cm-3, which coincided with a decrease

in LOI from 19% to 3.1% at the same depth (Fig. 2B, H; Table 1). Besides this section at around

250 cm, LOI was variable down core, and the interactive effect between depth and site best

explained this variation (wi=1.0; Table S1). Values of %N and %C also demonstrated variable

patterns down core, ranging from 0.2 to 1.4% and 0.1 to 1.3% N and 2.2 to 19.2 and 1.9 to

22.6% C at the reference and enriched sites, respectively. There was a slight peak in %C and %N

at ~40 cm in all cores, and a decrease between ~150-180 cm at the enriched marsh; but these fell

within the range of values at each site (Fig. 2C, D, I, J). An additive effect of depth and site best

explained both %C (wi=0.59) and %N (wi=0.58; Table S1), where the reference site had lower

%C by an average of 2.6% ± 0.89 and lower %N by an average 0.095% ± 0.046 when compared

to the enriched site. Site best explained patterns in C:N, which was fairly consistent down core

(wi=0.56; Table 1, S1), with the enriched site exhibiting a C:N that was, on average, 2.25 ± 2.2

higher than the reference marsh (Fig. 2E, K; Table 1). At both sites, δ13C values indicated a C4

plant carbon source except in the fertilizer marsh at 170 cm depth and from 240+ cm where a

more negative δ13C value suggests a transition to a C3 plant source (Fig. 2F, L). Radiocarbon

dating indicated that the deepest cores at the reference and enriched sites were 2290 and 3260

years, respectively (Table 2). Classic age-depth modeling indicated that corresponding depths at

each site were comparable in age by ± 200 years.

Alpha diversity and gene abundance

Shannon diversity of the total community (16S rRNA gene) decreased with depth linearly

by approximately 30% (Fig. 3A; p < 0.001, F1,94=77.91, R2 = 0.44) with no difference between

160

reference and enriched sites (p=0.894). Shannon diversity of the inferred active community (16S

rRNA) showed a similar, though less pronounced pattern, with values decreasing with depth

(Fig. 3B; p<0.001, F1,99=36.7, R2 = 0.26) and no difference between sites (p=0.261). Linear

mixed effects modeling revealed that depth was the best predictor of observed changes in

Shannon diversity for both the 16S rRNA gene (wi=0.61) and 16S rRNA (wi=0.62; Table S2).

Abundance of the 16S rRNA gene and its product, 16S rRNA, also exhibited decreasing

patterns with depth. Average 16S rRNA gene copies (per g ww-1) decreased linearly with depth

while remaining within the 108-109 range (Fig. 3C). ANCOVA results indicated that, in addition

to the effect of depth, there was also a difference in slopes between sites (p=0.006, F1,50=8.163),

with a smaller rate of change at the reference site (p<0.001, F1,22 = 8.326, R2 = 0.24, y=-0.0021x

+ 9.4) when compared to the enriched site (p<0.001, F1,27=13.92, R2=0.32, y=-0.003x + 9.3).

Abundance of 16S rRNA, on the other hand, declined by an order of magnitude in the top 10 cm

following an exponential decay (p<0.001, F1,54=22.45, R2 = 0.28) and reached values 3-4 orders

of magnitude lower in the deepest sediments when compared to the surface with equivalent

slopes between sites (Fig. 3D; p=0.126).

Microbial community structure

A principal coordinate analysis (PCoA) constructed from Weighted UniFrac of the total

microbial community (via analysis of the 16S rRNA gene) revealed extensive changes with

depth (Fig. 4A); however, a PERMANOVA indicated there was no effect of site on microbial

community structure (p=0.22, pseudo-F=1.36, n = 98). The first PCoA axis explained 41% of the

variance in the community structure, and when plotted against depth (Fig. 4B), indicated that the

microbial community changed rapidly through the first 50 cm (red line), at which point depth no

longer explained much variation. Considering the dynamic changes occurring in the top 50 cm, I

161

examined dissimilarity values between sites at each depth interval from 0-50 cm (Fig. 4C) and

found that while the two sites were very similar at the surface, they became increasingly different

with depth.

As a result of these different patterns, I split the dataset into two depth categories, 0-50

cm (shallow) and 60+ cm (deep) and repeated the above analyses. A PERMANOVA revealed

that the community structure between sites was not significantly different in the shallow

sediments (p=0.422, pseudo-F=0.845, n=36), however, it was different in the deep sediments

(p=0.005, pseudo-F=3.818, n=62). This pattern was also true for the potentially active

community (via analysis of the16S rRNA; Supplemental Fig. S1), with a significant difference

between sites in the deep sediments (p=0.034, pseudo-F=2.438, n=65) but not in the shallow

sediments (p=0.143, pseudo-F=1.716, n=36). In the shallow sediments, a linear mixed effects

model revealed that additive models with either depth and %N (16S rRNA gene; Table S3) or

depth and %C (16S rRNA; Table S4) best explained the variation observed in the first PCoA

axis. Since there was a significant difference between sites in the deeper sediments, I performed

linear mixed models within each site separately. In the deep reference sediments, the null

intercept model best explained variation in the first PCoA axis for both the 16S rRNA gene

(Table S3) and 16S rRNA (Table S4). Similarly, the null intercept model was most appropriate

for the 16S rRNA gene at the enriched site (Table S3), however, C:N best explained variation in

the first PCoA axis in the 16S rRNA (Table S4). Finally, A Mantel test revealed a significant

correlation between microbial community structure and a matrix of parameters for the shallow

(p=0.001, r=0.35) and the deep sediments (p=0.001, r=0.35) for the 16S rRNA gene.

Since there were no differences among sites in the microbial community of the shallow

sediments, I pooled by site to examine the most abundant taxa in the 16S rRNA gene (Fig. 5A)

162

and 16S rRNA (Fig. 5B). I found that taxa exhibited tremendous variability over the top 50 cm.

Out of the top 100 most abundant ASVs in the 16S rRNA gene, which accounted for 35% of the

total number of sequences, several classes decreased in abundance with depth including

Gammaproteobacteria and Ignavibacteria, while classes Actinobacteria, Bacteroidia, Nitrospira,

Parvarchaeota, and candidate phyla AC1 and OP9 were relatively more abundant in the deeper

sediments. To further explore the taxa driving shifts in the top 50 cm, I ran a random forest

regression model using 1920 and 2057 ASVs (as predictor variables) to explain the variance in

the 16S rRNA gene and 16S rRNA, respectively. I found that they explained 63.8% and 64.5%

(root-mean-square error; RMSE = 10.28 and 9.65) of the variance observed with depth, with

unique ASVs becoming more abundant at different depths.

The model identified 35 ASVs for the 16S rRNA gene, spread across 16 taxonomic

classes, that significantly contributed to the variance explained in the regression model. These 35

ASVs accounted for only ~5.5% of sequences from the top 50 cm dataset. Of these 35 ASVs, the

orders Anaerolineales, Marinicellales, Chromatiales, and an unclassified Chloroflexi decreased

and unclassified Deltaproteobacteria, unclassified Phycisphaerae, Candidate division AC1

(B04R032), Dehalococcoidales, Bacteroidales, and the Miscellaneous Crenarchaeotal Group

(MCG) increased with depth. Nitrospirales, unclassified Thermoplasmata, and Mariprofundales

peaked at intermediate depths (Fig. 6; Supplemental table S7). The model also identified 41

unique ASVs, distributed across 16 classes, in the 16S rRNA that significantly contributed to the

variance within the top 50 cm. These sequences accounted for only ~10% of sequences from the

top 50 cm dataset. Gammaproteobacteria, an unclassified Anaerolineae, candidate phyla TG3,

and Alphaproteobacteria decreased within the first 10-20 cm, while Dehalococcoidales,

unclassified Chlorobi, unclassified Deltaproteobacteria, and Bacteroidia all increased in relative

163

abundance. Spirochaetes, Brachyspirales, Leptospirae, Ignavibacteria, and candidate phylum

GN15 peaked at intermediate depths (Fig. 7; Supplemental table S8).

The top 100 ASVs in the deep sediments (60+ cm) represented a total of 37 unique

taxonomic groups aggregated at the class level, accounting for ~69 and 60% of the total and

active community at the reference site (Fig. 8A, C), and ~67 and 55% of the total and active

community at the enriched site (Fig. 8B, D). Random forest classification of the two sites in the

deep sediments had a classification accuracy of 90.3% (kappa = 0.80) for the 16S rRNA gene

and 81.5% (kappa = 0.62) for 16S rRNA. This was in contrast to the shallow sediments, where

classification success was much lower (50 and 61% for the 16S rRNA gene and 16S rRNA,

respectively). The taxa most important in differentiating between the reference and enriched sites

included classes Anaerolineae, Bacteroidia, Brachyspirae, Caldithrixae, Dehalococcoidetes,

Parvarchaea, Phycisphaerae, S085 from phylum Chloroflexi, and Candidate phyla GN04 (clade

GN15). Bacteroidia and Dehalococcoidetes were also important in differentiating the potentially

active community, as was Deltaproteobacteria, Gammaproteobacteria, and Candidate Phyla OP8

and AC1 (clade B04R032).

At both sites, the total community exhibited clear patterns with depth. Acidomicrobiia,

Alphaproteobacteria, and Nitrospira disappeared within the first 50-100 cm, while Chloroflexi,

and Candidate phyla AC1 and OP8 did not become important until 100-150 cm (Fig. 8A, B;

Table S9, 10). Further, some groups were present throughout the depth gradient, including

Parvarchaea, Clostridia, Dehalococcoidetes, and Candidate phylum OP1. In contrast, there were

fewer taxonomic groups that comprised the top 100 ASVs in the potentially active dataset, with

Deltaproteobacteria and Chloroflexi accounting for a much larger portion compared to the total

community (Fig. 8C, D; Table S11, 12).

164

Patterns of microbial community assembly

In the 16S rRNA gene (Fig. 9A), SESMPD values were strongly negative at the surface,

indicating phylogenetic clustering, and exhibited similar patterns for both the reference and

enriched sites. Values became positive, indicating phylogenetic overdispersion, at ~30 cm, and

then varied between -8.3 to 9.7 down to the deepest depths, staying mostly positive with some

exceptions. The pattern in 16S rRNA (Fig. 9B) was less pronounced, with positive excursions

occurring more sporadically and less often throughout the depth profile.

Discussion

Organic matter profiles exhibit local-scale shifts rather than long-term patterns

OM data exhibited patterns characteristic of New England salt marshes (Morris et al.

2016, Forbrich et al. 2018), demonstrating long term carbon storage and OM stability down core.

Bulk density varied significantly with depth (Fig. 2A, G), but much of this variation occurred

deeper than 240 cm where the average at both sites approximately doubled. This transition

provides evidence for the presence of marine sediments as a result of transgression and is further

supported by a negative excursion in δ13C values, suggesting a shift from C3 to C4 plants.

Presumably, this marks the establishment of the Spartina marsh ~3000 years ago, which is

comparable to values reported by Kirwan et al. (2011). Across all other depths, average bulk

density values I measured also agree with other studies conducted within similar marsh sites in

Plum Island Sound (Morris et al. 2016, Forbrich et al. 2018). I found that %OM also varied by

depth, but that this variation by depth differed between the reference and enriched sites. This

pattern is likely driven by the higher %OM values in Core 3 at the reference site, which is

located near a mosquito ditch and a salt marsh panne (Fig. 1). Prior to ditching that occurred in

165

the 1930’s (Adamowicz & Roman 2005, Wilson et al. 2014), this sampling location was located

furthest from the natural creek bank, potentially resulting in less inorganic sediment delivery

from the tides. Further, proximity to salt marsh pannes can influence sediment profiles, resulting

in heterogeneous peat deposition strata (Wilson et al. 2014, Spivak et al. 2017). Both %C (Fig.

2D, J) and %N (Fig. 2C, I) were lower at the reference marsh compared to the enriched marsh,

but this is likely due to spatial differences resulting from inherently different characteristics by

site as opposed to consequences from nutrient enrichment.

Decreases in diversity and cell abundance reflect resource limitation

For both the total and potentially active microbial communities, I observed significant

decreases in Shannon diversity and abundance of the 16S rRNA gene and 16S rRNA with depth.

On the surface, Shannon diversity of the 16S rRNA gene was similar to high levels of diversity

typically observed in salt marshes (Kearns et al. 2016) and other terrestrial soils (Fierer et al.

2007). Significant decreases by depth suggest the presence of an ecological filter, where only a

subset of microbes present at the surface can thrive at depth, a pattern that is widely observed in

both marine (Oni et al. 2015, Walsh et al. 2016) and terrestrial subsurface environments

(Hartmann et al. 2009, Eilers et al. 2012). The availability of OM with depth and the increasing

need to oxidize more recalcitrant forms to gain energy, is a likely factor driving these patterns, as

fewer taxa are physiologically adapted to use these complex forms of OM, and those that are

must compete for resources, resulting in lower diversity. Electron acceptor availability may be

another selecting factor, driving competition among groups that use the same pool of available

electron acceptors. In sediment cores from permafrost zones in Alaska, depth, pH, electrical

conductivity, total organic carbon, total nitrogen, and methane all significantly influenced

diversity. In this study, depth best explained the variation in diversity, potentially due to the

166

relatively stable nature of the OM in my cores (Table S2). The potentially active microbial

community demonstrated a lower Shannon diversity at the surface when compared to the total

community (Fig. 3B), suggesting that a portion of the DNA was inactive. Similar to the 16S

rRNA gene, depth best predicted patterns in diversity in the active community (Table S2).

Abundance of the 16S rRNA gene decreased with depth at both sites (Fig. 3C). likely

driven by limited resources, with decreasing OM quantity and quality determining the amount of

energy available for growth. Consequences of these energetic limitations could include slower

growth rates or community turnover time (Jørgensen & Marshall 2016), fewer cells due to

energy allocation towards cell repair rather than cell division (Langerhaus et al. 2012), and/or

smaller community size due to selection for taxa that can survive under such conditions (Petro et

al. 2017), all of which would contribute to the lower microbial abundance I observed. Unlike

abundances of the 16S rRNA gene, abundance of 16S rRNA decreased most within the top 10

cm (Fig. 3D), and then followed an exponential decay, reaching values 3-4 orders of magnitude

lower at depth (Fig. 3D). This deviation between the 16S rRNA gene and 16S rRNA copy

number indicates that relic DNA and inactive cells are abundant at depth. Relic DNA can persist

in sediment for long periods of time, potentially altering our interpretation of the ecosystem scale

effects of shifts in the microbial community (Carini et al. 2016, Vuillemin et al. 2017). Lennon et

al. (2018), however, suggest that relic DNA contributes minimally to estimates of diversity due

to degradation being proportional to total abundance. Further, microbes can enter a metabolic

resting state of dormancy under unfavorable environmental conditions (Lennon & Jones 2011).

Since they are not metabolically active, these organisms are not necessarily exposed to forces of

selection, which could result in a difference between the 16S rRNA gene and 16S rRNA

abundance profiles.

167

Deterministic processes are important in shallow (0-50 cm) sediments

I found no differences in the microbial community by site (Fig. 4A), but both the total

and active community changed significantly with depth in the top 50 cm (Fig. 4B, S1). The lack

of differentiation between the reference and enriched site suggests that the environmental

conditions resulting from nutrient enrichment had no measurable influence on microbial

community structure when compared to the reference site. Alternatively, the experimental

nutrient enrichment may not have percolated deeper than the surface sediments. I collected cores

in the high marsh from Spartina patens habitat, which receives tidal inundation only 3-4 times a

month (Johnson et al. 2016). It is also unlikely, given the bulk density of these sediments (Fig.

2A, G), that nutrient enrichment could penetrate into deeper sediments. I found that variation in

the first PCoA axis was best explained by depth and %N in the 16S rRNA gene and by depth and

%C in the 16S rRNA, suggesting that the nature of OM in the top 50 cm, in part, dictated

microbial community structure. Further, Mantel tests indicated a significant relationship between

microbial community structure and a full matrix of environmental parameters, further indicating

that other parameters, in addition to depth, played a role in determining microbial community

structure in the top 50 cm. At the surface, I observed phylogenetic clustering (Fig. 9) indicating

that highly selective processes dominated control over microbial community structure (Fine &

Kembel 2011, Stegen et al. 2012). I would expect this pattern where strong plant-microbe

associations tend to dictate microbial community structure (Burke et al. 2002), such as in the

microbial communities in salt marsh surface sediments and in the rooting zone below. This could

explain the similarity among all cores from both sites in surface sediments in both the 16S rRNA

gene and 16S rRNA, because the effect of plant species and other related parameters outweighed

any other selective forces (Fig. 4, S1).

168

By applying random forest regression modeling, I was able to identify ASVs that were

most important in explaining the variance in microbial community structure by depth in the

shallow sediments. Groups from orders Gammaproteobacteria, Betaproteobacteria, and

Nitrospira in the total community (Fig. 6), and Alphaproteobacteria in the potentially active

community (Fig. 7), all decreased within the top 50 cm. Proteobacterial taxa are typically

copiotropic and found to decrease with depth as nutrient concentrations, including organic rich

carbon, decrease (Hansel et al. 2008, Will et al. 2010, Eilers et al. 2012). Nitrospira belong to

groups known to oxidize nitrite and they possess enormous metabolic flexibility (Watson et al.

1986, Daims et al. 2015). Their decreasing pattern with depth and ultimate disappearance > 40

cm suggests that they are closely associated with plant roots where microaerophilic regions of

oxygen allow for tight coupling between nitrification and denitrification (Hamersley & Howes

2005, Koop-Jakobsen & Giblin 2010), which supports the deterministic nature of the sediments

throughout the top 50 cm.

In contrast, orders Deltaproteobacteria, Dehalococcoidetes, Bacteroidia, Phycisphaerae,

and Crenarchaeota all increased within the top 50 cm, suggesting these groups survive better in

in deeper sediments. Deltaproteobacteria are known to contain groups that contribute to sulfur

cycling (Bahr et al. 2005). These bacterial groups are widely detected in salt marsh sediments

where sulfate reduction accounts for a large portion of microbial respiration (Howarth & Teal

1979, Howarth 1984), and can become especially dominant at depths where other more

energetically favorable electron acceptors, such as nitrate, are depleted. Further, groups from

Dehalococcoidia and Crenarchaeota can oxidize a wide range of OM substrates, possessing

genes capable of degrading aromatic hydrocarbons (Vigeron et al. 2014, Wasmund et al. 2014),

suggesting these groups are better selected for depths where OM is less labile. Several ASVs,

169

despite driving the patterns I observed across the top 50 cm, were < 1% in relative abundance

indicating that rare taxa had a disproportionate influence on microbial community structure and

were important in discriminating among unique communities along a depth gradient where

deterministic processes determine their success (Shade et al. 2014, Shade & Gilbert 2015).

Stochastic processes are important in deeper (60+ cm) sediments

There was no significant difference in microbial community structure between sites near

the surface (0-50 cm), however dissimilarity values increased with depth (Fig. 4C), resulting in

significant differences in deeper sediments (60+ cm). Since the difference between sites occurred

in sediments that were deeper than 50 cm and 800+ years old, it is unlikely these patterns by site

resulted from nutrient enrichment; rather, it is possible that stochastic processes, or differences

during the specific time frame measured, structured microbial communities in these deeper

sediments. This is supported by SESMPD values that shifted from largely negative to

predominantly positive at this depth (Fig. 9), indicating a transition from deterministic to

stochastic assembly (Fine & Kembel 2011, Stegen et al. 2012). As depth increased and OM

characteristics changed, communities demonstrated phylogenetic overdispersion, suggesting that

stochastic processes such as drift, dispersal, and/or diversification exerted a greater force on

controlling community dynamics. This is in contrast to my expectation that selection by

environmental filtering and competition would increase as resources and niche space (i.e. OM

and electron acceptor availability decline) became more limited, and it contrasts with findings

from other studies conducted in subsurface communities (Stegen et al. 2013).

These results suggest, instead, that deep salt marsh sediments with fairly consistent OM

properties provide a unique environment in which stochastic processes dominate. Contrary to my

hypothesis, competition over resources can result in phylogenetic overdispersion, where taxa

170

occupying a limited range of niche spaces must diversify to survive (Koeppel & Wu 2014).

Further, the influence of stochastic processes may be amplified as a result of the stable nature of

these sediments. In systems with low diversity, drift and diversification play a disproportionate

role in controlling assembly processes (Chase & Meyers 2011), and systems with lower biomass

can experience a greater response to dispersal (Hoehler & Jørgensen 2013, Holyoak et al. 2015).

I observed both of these characteristics in deep sediments, and although I cannot determine

whether diversification or dispersal drive these processes, it is clear that stochastic processes

become more important in deeper sediments. The fact that the null intercept model best

explained variation in microbial community structure for the deeper sediments (Table S3, S4),

further supports this claim. However, C:N was a better predictor of community structure for the

active community at the enriched site, potentially suggesting that OM quality imparted stronger

selection on the active portion of the community at that site.

Conclusions

I found that OM quality was consistent with depth, exhibiting properties of long term

carbon storage potential characteristic of salt marshes and other blue carbon systems. Microbial

diversity and gene abundance decreased with depth, likely due to resource limitation, but these

patterns could not be linked to any other environmental parameter measured in this study. It is

likely that C:N was not a sensitive enough metric to capture changes in OM quality, and more

high resolution techniques are required to assess OM quality and composition. I also found that

while deterministic processes dominated microbial community assembly at the surface, it gave

way to stochastic assembly at depth, perhaps due to the stable nature of deep salt marsh

sediments. This work highlights the patterns of diversity, structure, and assembly of microbial

communities in relation to OM in deep salt marsh sediments, though we still lack a fundamental

171

understanding of functional capacity of these microbes at depth. Future work should aim to apply

meta-omic techniques to gather more information about the genetic machinery present in

microbes driving carbon cycling in salt marsh sediments.

Acknowledgements

I would like to thank researchers of the TIDE project (NSF OCE0924287, OCE0923689,

DEB0213767, DEB1354494, and OCE 1353140) for maintenance of the long-term nutrient

enrichment experiment, as well as researchers of the Plum Island Ecosystems LTER (NSF OCE

0423565, 1058747, 1637630). I would also like to acknowledge Annie Murphy, Joseph Vineis,

Khang Tran, Michael Greenwood, and members of the Bowen lab for contributions in the

laboratory, as well as Anne Giblin, Jane Tucker, and Inke Forbrich for their thoughtful

comments on this research. This work was funded by an NSF CAREER Award to JLB

(DEB1350491). Additional support was provided by a Ford Foundation pre-doctoral fellowship

award to ABM.

172

References

Adamowicz SC, Roman CT (2005) New England salt marsh pools: A quantitative analysis of

geomorphic and geographic features. Wetlands 25:279–288

Aller R (1994) Bioturbation and remineralization of sedimentary organic matter; effects of redox

oscillation. Chem Geol 114:331–345

Aller RC, Aller JY (1998) The effect of biogenic irrigation intensity and solute exchange on

diagenetic reaction rates in marine sediments. J Mar Res 56:905–936

Andersson AF, Riemann L, Bertilsson S (2010) Pyrosequencing reveals contrasting seasonal

dynamics of taxa within Baltic Sea bacterioplankton communities. ISME J 4:171–181

Apprill A, Mcnally S, Parsons R, & Weber L (2015). Minor revision to V4 region SSU rRNA

806R gene primer greatly increases detection of SAR11 bacterioplankton. Aqua Mic Ecol

75:129–137

Aronesty E (2011) Command-line tools for processing biological sequencing data. Ea-utils:

FASTQ processing utilities. http://code.google.com/p/ea-utils




Balsev-Clausen D, Dahl TW, Saad N, Rosing MT (2013) Precise and accurate δ13C analysis of

rock samples using Flash Combustion-Cavity Ring Down Laser Spectroscopy. J Anal

Atom Spectrom 28:516-523

Bates ST, Berg-Lyons D, Caporaso JG, Walters WA, Knight R, Fierer N (2011) Examining the

global distribution of dominant archaeal populations in soil. ISME J 5:908–917

Bates D, Maechler M, Bolker B, Walker S (2015) Fitting linear mixed-effects models using

lme4. Journal of statistical software 67:1-48




Biddle JF, Lipp JS, Lever MA, Lloyd KG, Sorensen KB, Anderson R, Fredricks HF, Elvert M,

Kelly TJ, Schrag DP, Sogin ML, Brenchley JE, Teske A, House CH, Hinrichs K-U

(2006) Heterotrophic Archaea dominate sedimentary subsurface ecosystems off Peru.

Proc Natl Acad Sci USA 103:3846–3851

Blaauw M (2010) Methods and code for ‘classical’ age-modelling of radiocarbon sequences.

Quaternary Geochronology 5:512-518

173

Blazewicz SJ, Barnard RL, Daly RA, Firestone MK (2013) Evaluating rRNA as an indicator of

microbial activity in environmental communities: limitations and uses. ISME J 7:2061-

2068

Bokulich N a, Subramanian S, Faith JJ, Gevers D, Gordon JI, Knight R, Mills D a, Caporaso JG

(2013) Quality-filtering vastly improves diversity estimates from Illumina amplicon

sequencing. Nat Methods 10:57–59

Bolker B, RDC Team (2014) bbmle: Tools for general maximum likelihood estimation. R

package version 01.20

Bulseco-McKim A, Giblin AE, Tucker J, Murphy AE, Sanderman J, Hiller-Bittrolf K, Bowen J

(In Review) Nitrate addition stimulates microbial decomposition of organic matter in salt

marsh sediments. Glob Chang Biol

Burdige DJ (2007) Preservation of organic matter in marine sediments: Controls, mechanisms,

and an imbalance in sediment organic carbon budgets? Chem Rev 107:467–485

Burke DJ, Hahn D, Hamerlynck EP (2002) Interactions among plant species and

microorganisms in salt marsh sediments. Appl Environ Microbiol 68:1157–1164




Press, Boston, MA

Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Peña

AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST, Knights D, Koenig JE, Ley RE,

Lozupone CA, Mcdonald D, Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ,

Walters WA, Widmann J, Yatsunenko T, Zaneveld J, Knight R (2010) QIIME allows

analysis of high- throughput community sequencing data. Nat Meth 7:335–336

Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Owens SM, Betley J,

Fraser L, Bauer M, Gormley N, Gilbert J a, Smith G, Knight R (2012) Ultra-high-


ISME J 6:1621–1624

Carini P, Marsden PJ, Leff JW, Morgan EE, Strickland MS, Fierer N (2016) Relic DNA is

abundant in soil and obscures estimates of soil microbial diversity. Nat Microbiol 2:1–6

Chase JM, Myers JA (2011) Disentangling the importance of ecological niches from stochastic

processes across scales. Philosph Trans R Soc B Biol Sci 366:2351–2363


saline wetland soils. Glob Biogeochem Cycles 17:1-12

174

Chu H, Sun H, Tripathi BM, Adams JM, Huang R, Zhang Y, Shi Y (2016) Bacterial community

dissimilarity between the surface and subsurface soils equals horizontal differences over

several kilometers in the western Tibetan Plateau. Environ Microbiol 18:1523–1533


matter mixtures. Nature 369:304-307

Daims H, Lebedeva EV, Pjevac P, Han P, Herbold C, Albertsen M, Jehmlich N, Palatinszky M,

Vierheilig J, Bulaev A, Kirkegaard RH, Bergen M Von, Rattei T, Bendinger B, Nielsen

PH, Wagner M (2015) Complete nitrification by Nitrospira bacteria. Nature 528:504–509








Duarte CM, Losada IJ, Hendriks IE, Mazarrasa I, Marba N (2013) The role of coastal plant

communities for climate change mitigation and adaptation. Nat Clim Change 3:961-968

Duarte CM, Middelburg JJ, Caraco N, Major NC (2005) Major role of marine vegetation on the

oceanic carbon cycle. Biogeosciences 2:1–8

Dumbrell AJ, Nelson M, Helgason T, Dytham C, Fitter AH (2010) Relative roles of niche and

neutral processes in structuring a soil microbial community. ISME J 4:337–345

Eilers KG, Debenport S, Anderson S, Fierer N (2012) Digging deeper to find unique microbial

communities: The strong effect of depth on the structure of bacterial and archaeal

communities in soil. Soil Biol Biochem 50:58–65

Fierer N, Jackson RB (2006) The diversity and biogeography of soil bacterial communities. Proc

Natl Acad Sci USA 103:626–631

Fierer N, Breitbart M, Nulton J, Salamon P, Lozupone C, Jones R, Robeson M, Edwards RA,

Felts B, Rayhawk S, Knight R, Rohwer F, Jackson RB (2007) Metagenomic and small-

subunit rRNA analyses reveal the genetic diversity of bacteria, archaea, fungi, and

viruses in soil. Appl Environ Microbiol 73:7059–7066

Fierer N, Schimel JP, Holden PA (2003) Variations in microbial community composition

through two soil depth profiles. Soil Biol Biochem 35:167–176

175

Fine PVA, Kembel SW (2011) Phylogenetic community structure and phylogenetic turnover

across space and edaphic gradients in western Amazonian tree communities. Ecography

34:552–565



878

Froelich P, Klinkhammer G, Bender M, Luedtke N, Heath G, Cullen D, Dauphin P (1979) Early

oxidation of organic matter in pelagic sediments of the eastern equatorial Atlantic:

suboxic diagenesis. Geochim Cosmochim Acta 43:1075–1090






Giovannoni SJ (2017). SAR11 bacteria: The most abundant plankton in the oceans. Ann Rev

Mar Sci 9:231–255



Hansel CM, Fendorf S, Jardine PM, Francis CA (2008) Changes in bacterial and archaeal

community structure and functional diversity along a geochemically variable soil profile.

Appl Environ Microbiol 74:1620–1633

Hartmann M, Lee S, Hallam SJ, Mohn WW (2009) Bacterial, archaeal and eukaryal community

structures throughout soil horizons of harvested and naturally disturbed forest stands.


Hedges JI, Eglinton G, Hatcher PG, Kirchman DL, Arnosti C, Derenne S, Evershed RP, Kögel-

Knabner I, de Leeuw JW, Littke R, Michaelis W (2000) The molecularly-uncharacterized

component of nonliving organic matter in natural environments. Organic Geochem 31:945-

58

Hijmans RJ, van Etten J (2012) Raster: Geographic analysis and modeling with raster data. R

package version 2.0-12

Hoehler TM, Jørgensen BB (2013) Microbial life under extreme energy limitation. Nat Rev


Holmer M, Gribsholt B, Kristensen E (2002) Effects of sea level rise on growth of Spartina

anglica and oxygen dynamics in rhizosphere and salt marsh sediments. Mar Ecol Prog

176

Ser 225:197–204

Holyoak M, Leibold MA, Holt RD (2005) Metacommunities: Spatial Dynamics and Ecological

Communities. The University of Chicago Press, Chicago

Howarth RW (1984) The Ecological significance of sulfur in the energy dynamics of salt marsh

and coastal marine sediments. Biogeochemistry 1:5–27


24:999–1013

Inagaki F, Hinrichs K-U, Kubo Y, Bowles MW, Heuer VB Exploring deep microbial life in coal-

bearing sediment down to ~2.5 km below the ocean floor (2015) Science 349:16–21

Johnson JB, Omland KS (2004) Model selection in ecology and evolution. Trends Ecol Evol

19:101–108

Johnson DS, Warren RS, Deegan LA, Mozdzer TJ (2016) Saltmarsh plant responses to

eutrophication. Ecol Appl 26:2647–2659

Jørgensen BB, Boetius A (2007) Feast and famine - Microbial life in the deep-sea bed. Nat Rev


Jørgensen BB, Marshall IPG (2016) Slow microbial life in the seabed. Ann Rev Mar Sci

8:311–332


24:726–734



Kearns PJ, Angell JH, Howard EM, Deegan LA, Stanley RHR, Bowen JL (2016) Nutrient

enrichment induces dormancy and decreases diversity of active bacteria in salt marsh

sediments. Nat Commun 7:1–9

Kembel SW, Cowan PD, Helmus MR, Cornwell WK, Morlon H, Ackerly DD, Blombger SP,

Webb CO (2010). Picante: R tools for integrating phylogenies and ecology.

Bioinformatics 26:1463-1464

Kerkhof L, Kemp P (1999) Small ribosomal RNA content in marine Proteobacteria during non-

steady-state growth. FEMS Microbiol Ecol 30:253-260

Kirwan ML, Murray AB, Donnelly JP, Corbett DR (2011) Rapid wetland expansion during

European settlement and its implication for marsh survival under modern sediment

delivery rates. Geology 39:507–510

177

Koeppel AF, Wu M (2014) Species matter: The role of competition in the assembly of

congeneric bacteria. ISME J 8:531–540



Kristensen E, Holmer M (2001) Decomposition of plant materials in marine sediment exposed to

different electron acceptors (O2, NO3-, and SO4

2-), with emphasis on substrate origin,

degradation kinetics, and the role of bioturbation. Geochim Cosmochim Acta 65:419–433

Kubo K, Lloyd KGF, Biddle J, Amann R, Teske A, Knittel K (2012) Archaea of the

Miscellaneous Crenarchaeotal Group are abundant, diverse and widespread in marine

sediments. ISME J 6:1949-1965

Kuhn M (2015) A Short introduction to the caret package. R Found Stat Comput:1–10

Langerhuus AT, Røy H, Lever MA, Morono Y, Inagaki F, Jørgensen BB, Lomstein BA (2012)

Endospore abundance and d: L-amino acid modeling of bacterial turnover in holocene

marine sediment (Aarhus Bay). Geochim Cosmochim Acta 99:87–99

Lee RW, Kraus DW, Doeller JE (1999) Oxidation of sulfide by Spartina alterniflora roots.


Lennon JT, Jones SE (2011) Microbial seed banks: The ecological and evolutionary implications

of dormancy. Nat Rev Microbiol 9:119–130

Lennon JT, Muscarella ME, Placella SA, Lehmkuhl BK (2018) How, when, and where relic

DNA affects microbial diversity. MBio 9:1–14

Levy-Booth DJ, Campbell RG, Gulden RH, Hart MM, Powell JR, Klironomos JN, Pauls KP,

Swanton CJ, Trevors JT, Dunfield KE (2007) Cycling of extracellular DNA in the soil

environment. Soil Biol Biochem 39:2977–2991

Liaw A, Wiener M (2002) Classification and regression by randomForest. R news 2:18–22






Mantel N (1967) The detection of disease clustering and a generalized regression approach.

Cancer Res 27:209-220

Marshall IPG, Karst SM, Nielsen PH, Jørgensen BB (2018) Metagenomes from deep Baltic Sea

178

sediments reveal how past and present environmental conditions determine microbial

community composition. Mar Genomics 37:58–68


Knight R, Hugenholtz P (2012) An improved Greengenes taxonomy with explicit ranks

for ecological and evolutionary analyses of bacteria and archaea. ISME J 6:610–618


WH, Silliman BR (2011) A blueprint for blue carbon: toward an improved understanding

of the role of vegetated coastal habitats in sequestering CO2. Front Ecol Environ 9:552–

560

Mendelssohn I, Morris J (2000) Ecophysiological controls on the growth of Spartina alterniflora.

In: Weinstein N, Kreeger D (eds) Concepts and controversies in tidal marsh ecology.

Kluwer, Dordrecht, p 59–80

Mettel C, Kim Y, Shrestha PM, Liesack W (2010) Extraction of mRNA from soil. Appl Environ


Middelburg J (1989) A simple rate model for organic matter decomposition in marine sediments.


Mittelbach GG, Schemske DW (2015) Ecological and evolutionary perspectives on community

assembly. Trends Ecol Evol 30:241-247

Morris JT, Barber DC, Callaway JC, Chambers R, Hagen SC, Hopkinson CS, Johnson BJ,

Megonigal P, Neubauer SC, Troxler T, Wigand C (2016) Contributions of organic and

inorganic matter to sediment volume and accretion in tidal wetlands at steady state.

Earth’s Futur 4:110–121

Mueller P, Schile-Beers LM, Mozdzer TJ, Chmura GL, Dinter T, Kuzyakov Y, Groot AV De,

Esselink P, Smit C, D’Alpaos A, Ibanez C, Lazarus M, Neumeier U, Johnson BJ,

Baldwin AH, Yarwood SA, Montemayor DI, Yang Z, Wu J, Jensen K, Nolte S (2017)

Global change effects on decomposition processes in tidal wetlands: Implications from a

global survey using standardized litter. Biogeosciences 15:3189–3202

Najjar RG, Herrmann M, Alexander R, Boyer EW, Burdige DJ, Butman D, Cai WJ, Canuel EA,

Chen RF, Friedrichs MA, Feagin RA (2018). Carbon budget of tidal wetlands, estuaries,

and shelf waters of eastern North America. Global Biogeochem Cy 32:389-416

Narayan S, Beck MW, Wilson P, Thomas CJ, Guerrero A, Shepard CC, Reguero BG, Franco G,

Ingram JC, Trespalacios D (2017) The value of coastal wetlands for flood damage

reduction in the Northeastern USA. Sci Rep 7:1–12

Nemergut DR, Schmidt SK, Fukami T, O’Neill SP, Bilinski TM, Stanish LF, Knelman JE, Darcy

JL, Lynch RC, Wickey P, Ferrenberg S (2013) Patterns and processes of microbial

179

community assembly. Microbiol Mol Biol Rev 77:342–356




Oni OE, Schmidt F, Miyatake T, Kasten S, Witt M, Hinrichs KU, Friedrich MW (2015)

Microbial communities and organic matter composition in surface and subsurface

sediments of the Helgoland mud area, North Sea. Front Microbiol 6:1–16

Papp K, Hungate BA, Schwartz E (2018) Microbial rRNA synthesis and growth compared

through quantitative stable isotope probing with H218O. Appl Environ Microbiol 84:1-11

Parkes RJ, Cragg B, Roussel E, Webster G, Weightman A, Sass H (2014) A review of

prokaryotic populations and processes in sub-seafloor sediments, including biosphere:

Geosphere interactions. Mar Geol 352:409–425

Petro C, Starnawski P, Schramm A, Kjeldsen KU (2017) Microbial community assembly in

marine sediments. Aquat Microb Ecol 79:177–195

Philippot L, Andersson SGE, Battin TJ, Prosser JI, Schimel JP, Whitman WB, Hallin S (2010)

The ecological coherence of high bacterial taxonomic ranks. Nat Rev Microbiol 8:523–

529

Price MN, Dehal PS, Arkin AP (2010) FastTree 2 - Approximately maximum-likelihood trees

for large alignments. PLoS ONE 5:e9490

QIIME 2 Development Team (2018) QIIME 2. https://qiime2.org/

Reddy K, Patrick Jr W (1975) Effect of alternate aerobic and anaerobic conditions on redox

potential, organic matter decomposition and nitrogen loss in a flooded soil. Soil Biol

Biochem 7:87–94

Richards SA (2005) Testing ecological theory using the information-theoretic approach :

Examples and cautionary results. Ecology 86:2805–2814





Shade A, Gilbert JA (2015) Temporal patterns of rarity provide a more complete view of

microbial diversity. Trends Microbiol 23:335–340

Shade A, Jones SE, Gregory Caporaso J, Handelsman J, Knight R, Fierer N, Gilbert JA (2014)

180

Conditionally rare taxa disproportionately contribute to temporal changes in microbial

diversity. MBio 5:1–9

Sousa AI, Lillebø AI, Risgaard-Petersen N, Pardal MA, Caçador I (2012) Denitrification: an

ecosystem service provided by salt marshes. Mar Ecol Prog Ser 448:79-92

Spivak AC, Gosselin K, Howard E, Mariotti G, Forbrich I, Stanley R, Sylva SP (2017) Shallow

ponds are heterogeneous habitats within a temperate salt marsh ecosystem. J Geophys

Res Biogeosci 122:1371–1384

Starnawski P, Bataillon T, Ettema TJG, Jochum LM, Schreiber L, Chen X, Lever MA, Polz MF,

Jørgensen BB, Schramm A, Kjeldsen KU (2017) Microbial community assembly and

evolution in subseafloor sediment. Proc Natl Acad Sci USA 114:2940–2945

Stegen JC, Lin X, Fredrickson JK, Chen X, Kennedy DW, Murray CJ, Rockhold ML, Konopka

A (2013) Quantifying community assembly processes and identifying features that

impose them. ISME J 7:2069–2079

Stegen JC, Lin X, Konopka AE, Fredrickson JK (2012) Stochastic and deterministic assembly

processes in subsurface microbial communities. ISME J 6:1653–1664

Steven B, Hesse C, Soghigian J, Gallegos-Graves LV, Dunbar J (2017) Simulated rRNA/DNA

ratios show potential to misclassify active populations as dormant. Appl Environ

Microbiol 83:1-11

Sutton-Grier AE, Keller JK, Koch R, Gilmour C, Megonigal JP (2011) Electron donors and

acceptors influence anaerobic soil organic matter mineralization in tidal marshes. Soil

Biol Biochem 43:1576–1583

Teal J, Kanwisher J (1961) Gas exchange in a Georgia salt marsh. Limnol Oceanogr 6:388–399

Teske A, Sorensen KB (2008) Uncultured archaea in deep marine subsurface sediments: have we

caught them all? ISME 2:3-18

Thamdrup B, Fossing H, Jorgensen BB (1994) Manganese, iron and sulfur cycling in a coastal

marine sediment, Aarhus bay, Denmark. Geochim Cosmochim Acta 58:5115–5129

Tripathi BM, Kim M, Kim Y, Byun E, Yang JW, Ahn J, Lee YK (2018) Variations in bacterial

and archaeal communities along depth profiles of Alaskan soil cores. Sci Rep 8:1–11





diagenetic status in marine organic matter. Geochim Cosmochim Acta 61:5363–5369

181

Walsh EA, Kirkpatrick JB, Pockalny R, Sauvage J, Spivack AJ, Murray RW, Sogin ML,

D’Hondt S (2016) Relationship of bacterial richness to organic degradation rate and

sediment age in subseafloor sediment. Appl Environ Microbiol 82:4994–4999




Wasmund K, Schreiber L, Lloyd KG, Petersen DG, Schramm A, Stepanauskas R, Jørgensen BB,

Adrian L (2014) Genome sequencing of a single cell of the widely distributed marine

subsurface Dehalococcoidia, phylum Chloroflexi. ISME J 8:383–397

Watson SW, Bock E, Valois FW, Waterbury JB, Schlosser U (1986) Nitrospira marine gen. nov.

sp.:a chemolithotrophic nitrite-oxiziding bacterium. Arch Microbiol 144:1-7

Webb CO, Ackerly DD, Kembel SW (2008) Phylocom: Software for the analysis of

phylogenetic community structure and trait evolution. Bioinformatics 24:2098–2100

Westrich JT, Berner RA (1984). The role of sedimentary organic matter in bacterial sulfate


Will C, Thürmer A, Wollherr A, Nacke H, Herold N, Schrumpf M, Gutknecht J, Wubet T,

Buscot F, Daniell R (2010) Horizon-specific bacterial community composition of german

grassland soils, as revealed by pyrosequencing-based analysis of 16S rRNA genes. Appl







Vellend M (2010) Conceptual synthesis in community ecology. Q Rev Biol 85:183–206

Vigneron A, Cruaud P, Roussel EG, Pignet P, Caprais JC, Callac N, Ciobanu MC, Godfroy A,

Cragg BA, Parkes JR, Nostrand JD Van, He Z, Zhou J, Toffin L (2014) Phylogenetic and

functional diversity of microbial communities associated with subsurface sediments of

the Sonora Margin, Guaymas Basin. PLoS ONE 9:9–13

Vuillemin A, Ariztegui D, Horn F, Kallmeyer J, Orsi WD, Anselmetti F, Corbella H, Francus P,

Lücke A, Maidana NI, Ohlendorf C, Zolitschka B, Schabitz F, Wastegård S (2018)

Microbial community composition along a 50 000-year lacustrine sediment sequence.

FEMS Microbiol Ecol 94:1–14

182

Zedler JB, Kercher S (2005) Wetland resources: Status, trends, ecosystem services, and

restorability. Ann Rev Env Resources 30:39-74

183

Tables

Table 1. Average ± SE for bulk density (g cm-3), %OM, %N, %C, C:N ratio, and δ13C (‰) for

the reference and enriched marshes (n=3)

Reference Enriched

Bulk Density 0.33 (0.15) 0.35 (0.29)

%OM 23.2 (9.50) 25.9 (10.30)

%N 0.64 (0.23) 0.72 (0.24)

%C 10.64 (4.15) 12.93 (4.77)

C:N 18.86 (3.32) 21.11 (3.85)

δ13C -15.2 (1.00) -16.4 (3.17)

184

Table 2. Sediment depth intervals and age date results using radiocarbon dating.

Site Depth (cm) δ14C Age (years) Error (± years)

Reference 81-82 18.08 > Modern 1

97-98 -181.22 1540 15

129-130 -210.47 1830 15

163-164 -241.54 2160 25

213-214 -254.21 2290 20

Enriched 67-68 -89.65 690 15

113-114 -192.76 1660 15

179-180 -256.07 2310 15

253-256 -338.67 3260 20

185

Figures

Fig. 1. Location of my study sites in northeastern Massachusetts, USA: West Creek (Reference;

42.759 N, 70.891 W) and Sweeney Creek (Enriched; 42.722 N, 70.847 W). Shapes indicate core

replicates per site (squares = 1, diamonds = 2, triangles = 3). Maps were generated by

downloading data from the Database of Global Administrate Areas (GADM; Global

Administrative Areas) using the raster package in R (Hijmans & Jacob van Etten 2012).

186

Fig. 2. Organic matter characteristics along a depth gradient for cores collected from reference

(blue, top) and enriched (green, bottom) sites. Shapes indicate core replicate (squares = 1,

diamonds = 2, triangles = 3), and age date along right y-axis is calculated from 14C radiocarbon

dating. Panels A and G report bulk density values, B and H report % organic matter, C and I, and

D and J, report %N and %C respectively, E and K report C:N and F and L report the δ13C values.

187

Fig. 3. Shannon diversity decreases with depth at both sites for the 16S rRNA gene (A) and 16S

rRNA (B). Log 16S rRNA gene abundance decreases with depth linearly (C), while Log 16S

rRNA decreases with depth following an exponential pattern (D). Note the difference in scale in

the Y axis of Figures C and D.

188

Fig. 4. (A) Principal coordinate analysis (PCoA) constructed from weighted Unifrac of the total

community colored by depth, with no effect of site according to a PERMANOVA. (B) The first

PCoA axis versus depth shows the microbial community structure changes drastically with depth

in the top 50 cm, marked with the dotted red line, but then becomes relatively stable 60+cm. (C)

Boxplots of dissimilarity between sites for each depth interval between 0-50 cm.

189

Fig. 5. Bar plots showing relative abundance of top 100 ASVs aggregated at the class-level in

shallow (0-50 cm) sediments for the (A) total and (B) active microbial community combined for

both sites, accounting for an average of ~35 and 40% of the total dataset respectively. Black lines

indicate family-level distinctions within each order. Additional taxonomic information can be

found in Supplemental Table S5-6.

190

Fig. 6. Barplots showing relative abundance of the 35 ASVs most important in explaining the

variance in the total (DNA) microbial community structure along the top 50 cm depth gradient

according to a random forest regression model. Data are aggregated at the class level,

represented by faceted boxes, and colors within each box represent different orders within each

class. Black lines indicate family-level distinctions within each order. Additional taxonomic

information can be found in Supplemental Table S7.

191

Fig. 7. Barplots showing relative abundance of the 35 ASVs most important in explaining the

variance in the active (RNA) microbial community structure along a depth gradient according to

a random forest regression. Data are aggregated at the class level, represented by faceted boxes,

and colors within each box represent different orders. Black lines indicate family-level

distinctions within each order. Additional taxonomic information can be found in Supplemental

Table S8.

192

Fig. 8. Bar plots showing relative abundance of top 100 ASVs aggregated at the class-level in

deep (60+ cm) sediments for the total community at the reference (A) and enriched (B) site (~69

and 67% of total dataset) and the potential active community at the reference (C) and enriched

(D) site (~60 and 55% of total dataset). Black lines indicate family-level distinctions within each

order, and a blank bar indicates that no sequence data are available at that depth. Additional

taxonomic information can be found in Supplemental Table S9-12.

193

Fig. 9. Standardized effect size of mean pairwise distances (SESMPD), equivalent to -Net

relatedness Index, along a depth gradient for (A) the 16S rRNA gene and (B) 16S rRNA.

Negative values (<1) indicate phylogenetic clustering and positive (>1) values indicating

phylogenetic overdispersion, with the red dotted line indicating 50cm.

194

Supplemental Tables

Table S1: Model selection results for linear mixed effects models assessing the response of bulk

density, %OM, %N, %C, and C:N ratio to depth and site (fixed effects) and core replicate

(random effect). Each model includes a null intercept model, reduced models (depth or site), or a

full model (additive or multiplicative relationships between depth and site). Models that are

bolded represent best fit as indicated by an Akaike weight (wi) closest to 1.

Response Variable Model df ΔAIC Weight

Bulk density Bulk density = 1 + (1|Core) 3 234.7 <0.001

Bulk density = Site + (1|Core) 4 235.3 <0.001

Bulk density = Depth + (1|Core) 4 0.0 0.43

Bulk density = Depth + Site + (1|Core) 5 0.4 0.34

Bulk density = Depth*Site + (1|Core) 6 1.2 0.23

%OM %OM = 1 + (1|Core) 3 185.2 <0.001

%OM = Site + (1|Core) 4 184.6 <0.001

%OM = Depth + (1|Core) 4 13.3 0.001

%OM = Depth + Site + (1|Core) 5 12.8 0.002

%OM = Depth*Site + (1|Core) 6 0.0 0.9970

%C %C = 1 + (1|Core) 3 5.5 0.038

%C = Site + (1|Core) 4 3.4 0.109

%C = Depth + (1|Core) 4 4.6 0.059

%C = Depth + Site + (1|Core) 5 0.00 0.589

%C = Depth*Site + (1|Core) 6 2.1 0.204

%N %N = 1 + (1|Core) 3 8.7 0.008

%N = Site + (1|Core) 4 8.6 0.008

%N = Depth + (1|Core) 4 2.0 0.214

%N = Depth + Site + (1|Core) 5 0.0 0.581

%N = Depth*Site + (1|Core) 6 2.2 0.190

C:N Ratio Molar C:N Ratio = 1 + (1|Core) 3 3.3 0.111

Molar C:N Ratio = Site + (1|Core) 4 0.0 0.564

Molar C:N Ratio = Depth + (1|Core) 4 4.7 0.053

Molar C:N Ratio = Depth + Site + (1|Core) 5 2.0 0.205

Molar C:N Ratio = Depth*Site + (1|Core) 6 4.2 0.068

195

Table S2: Model selection results for linear mixed effects models assessing the response of

Shannon diversity to depth, %C, %N, and C:N ratio treating core as a random effect. Models that

are bolded represent best fit as indicated by an Akaike weight (wi) closest to 1.


16S rRNA gene Shannon = 1 + (1|Core) 3 56 <0.001

Shannon Diversity Shannon = Depth + (1|Core) 4 0.0 0.368

Shannon = %C + (1|Core) 4 57.4 <0.001

Shannon = %N + (1|Core) 4 54.7 <0.001

Shannon = C:N + (1|Core) 4 56 <0.001

Shannon = Depth + %C + (1|Core)* 5 1.9 0.141

Shannon = Depth + %N + (1|Core) 5 2.2 0.122

Shannon = Depth + C:N + (1|Core)* 5 0.6 0.271

Shannon = Depth + %C + %N + (1|Core) 6 3.8 0.055

Shannon = Depth + %C + %N + C:N + (1|Core) 7 4.3 0.043

16S rRNA Shannon = 1 + (1|Core) 3 19.9 <0.001

Shannon Diversity Shannon = Depth + (1|Core) 4 0.0 0.406

Shannon = %C + (1|Core) 4 22 <0.001

Shannon = %N + (1|Core) 4 21.7 <0.001

Shannon = C:N + (1|Core) 4 22 <0.001

Shannon = Depth + %C + (1|Core) 5 2.1 0.145

Shannon = Depth + %N + (1|Core)* 5 1.8 0.163

Shannon = Depth + C:N + (1|Core)* 5 1.6 0.186

Shannon = Depth + %C + %N + (1|Core) 6 4 0.056

Shannon = Depth + %C + %N + C:N + (1|Core) 7 4.4 0.044

Shannon = 1 + (1|Core) 3 19.9 <0.001

196


PCoA axis 1 to depth, %C, %N, and C:N ratio for the 16S rRNA gene treating core as a random

effect. Models that are bolded represent best fit as indicated by an Akaike weight (wi) closest to

1.


Shallow (0-50 cm) PCoA 1 = 1 + (1|Core) 3 27.4 <0.001

PCoA 1 = Depth + (1|Core) 4 0 0.315

PCoA 1 = %C + (1|Core) 4 29.3 <0.001

PCoA 1 = %N + (1|Core) 4 28.8 <0.001

PCoA 1 = C:N + (1|Core) 4 29.6 <0.001

PCoA 1 = Depth + %C + (1|Core)* 5 1.3 0.164

PCoA 1 = Depth + %N + (1|Core) 5 0 0.321

PCoA 1 = Depth + C:N + (1|Core) 5 2.7 0.084

PCoA 1 = Depth + %C + %N + (1|Core) 6 2.5 0.092

PCoA 1 = Depth + %C + %N + C:N + (1|Core) 7 5.3 0.023

Deep (60+ cm) PCoA 1 = 1 + (1|Core) 3 0 0.3611

Reference only PCoA 1 = Depth + (1|Core)* 4 1.7 0.153

PCoA 1 = %C + (1|Core) 4 2.6 0.0994

PCoA 1 = %N + (1|Core) 4 2.2 0.1176

PCoA 1 = C:N + (1|Core) 4 2.7 0.0934

PCoA 1 = Depth + %C + (1|Core) 5 4.1 0.0469

PCoA 1 = Depth + %N + (1|Core) 5 3.6 0.0589




Deep (60+ cm) PCoA 1 = 1 + (1|Core) 3 0 0.3701

Enriched only PCoA 1 = Depth + (1|Core) 4 2.6 0.103

PCoA 1 = %C + (1|Core) 4 2 0.1383

PCoA 1 = %N + (1|Core)* 4 1.7 0.1611

PCoA 1 = C:N + (1|Core) 4 2.5 0.1075

PCoA 1 = Depth + %C + (1|Core) 5 4.6 0.0365

PCoA 1 = Depth + %N + (1|Core) 5 4.3 0.0424




197


PCoA axis 1 to depth, %C, %N, and C:N ratio for 16S rRNA treating core as a random effect.

Models that are bolded represent best fit as indicated by an Akaike weight (wi) closest to 1.


Shallow (0-50 cm) PCoA 1 = 1 + (1|Core) 3 48.6 <0.001

PCoA 1 = Depth + (1|Core) 4 3.1 0.099

PCoA 1 = %C + (1|Core) 4 48.8 <0.001

PCoA 1 = %N + (1|Core) 4 49.9 <0.001

PCoA 1 = C:N + (1|Core) 4 45.5 <0.001

PCoA 1 = Depth + %C + (1|Core) 5 0 0.458

PCoA 1 = Depth + %N + (1|Core) 5 2 0.167

PCoA 1 = Depth + C:N + (1|Core) 5 3 0.102



Deep (60+ cm) PCoA 1 = 1 + (1|Core) 3 0 0.3149

Reference only PCoA 1 = Depth + (1|Core) 4 1 0.1904

PCoA 1 = %C + (1|Core) 4 2.6 0.0876

PCoA 1 = %N + (1|Core) 4 2.5 0.0924

PCoA 1 = C:N + (1|Core) 4 2.8 0.0774

PCoA 1 = Depth + %C + (1|Core) 5 2.9 0.0727

PCoA 1 = Depth + %N + (1|Core) 5 2.9 0.0743




Deep (60+ cm) PCoA 1 = 1 + (1|Core) 3 1.2 0.222

Enriched only PCoA 1 = Depth + (1|Core) 4 3.8 0.06

PCoA 1 = %C + (1|Core) 4 2.9 0.095

PCoA 1 = %N + (1|Core) 4 3.6 0.067

PCoA 1 = C:N + (1|Core) 4 0 0.397

PCoA 1 = Depth + %C + (1|Core) 5 5.7 0.023

PCoA 1 = Depth + %N + (1|Core) 5 6.4 0.016




198

Table. S5. Unique taxonomic information for top 100 ASVs from shallow (0-50cm) total

microbial community for both sites combined

Kingdom Phylum Class Order Family

Bacteria Spirochaetes Spirochaetes Spirochaetales Spirochaetaceae






Bacteria Proteobacteria Epsilonproteobacteria Campylobacterales Helicobacteraceaea

Bacteria Proteobacteria Epsilonproteobacteria Campylobacterales Helicobacteraceae

Bacteria Chlorobi Ignavibacteria Ignavibacteriales Ignavibacteriaceae




Bacteria Bacteroidetes Bacteroidia Bacteroidales na











Archaea Euryarchaeota Thermoplasmata E2 DHVEG-1

Archaea Crenarchaeota MBGB na na


Unassigned na na na na

Archaea Crenarchaeota MCG B10 na

Archaea Parvarchaeota Parvarchaea WCHD3-30 na


Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales koll13

Bacteria Caldithrix Caldithrixae Caldithrixales BA059

Bacteria Caldithrix Caldithrixae Caldithrixales BA059

Bacteria Spirochaetes Brachyspirae Brachyspirales A0-023

Bacteria Actinobacteria Actinobacteria WCHB1-81 At425_EubF1


Bacteria Chloroflexi Anaerolineae SB-34 na

Bacteria Chloroflexi Anaerolineae SB-34 na

Bacteria Chloroflexi Anaerolineae GCA004 na


199


Bacteria Chloroflexi Anaerolineae CFB-26 na

Bacteria Chloroflexi Anaerolineae CFB-26 na

Bacteria Chloroflexi Dehalococcoidetes Dehalococcoidales Dehalococcoidaceae






Bacteria AC1 na na na

Bacteria AC1 B04R032 na na


Bacteria OP9 JS1 SB-45 na


Bacteria WS2 SHA-109 na na

Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales Rhodospirillaceae

Bacteria Proteobacteria Gammaproteobacteria Thiotrichales Thiotrichaceae

Bacteria Proteobacteria Betaproteobacteria Gallionellales Gallionellaceaeb

Bacteria Proteobacteria Betaproteobacteria MND1 na

Bacteria Proteobacteria Gammaproteobacteria Chromatiales na



Bacteria Proteobacteria Deltaproteobacteria na na


Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobacteraceae

Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobacteraceaec




Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobulbaceae

Bacteria Proteobacteria na na na

Bacteria Proteobacteria Deltaproteobacteria MBNT15 na









Bacteria Proteobacteria Alphaproteobacteria Rhizobiales na

Bacteria Proteobacteria Alphaproteobacteria Rhizobiales Hyphomicrobiaceaed

Bacteria Proteobacteria Alphaproteobacteria Rhizobiales Hyphomicrobiaceaed

Bacteria Proteobacteria Alphaproteobacteria Rhizobiales Hyphomicrobiaceaee

200

Bacteria Proteobacteria Alphaproteobacteria Rhizobiales Hyphomicrobiaceae




Bacteria Proteobacteria Deltaproteobacteria Desulfarculales Desulfarculaceae







Bacteria TPD-58 na na na

Bacteria Caldithrix Caldithrixae Caldithrixales Caldithrixaceae

Bacteria Planctomycetes Phycisphaerae MSBL9 na

Bacteria Nitrospirae Nitrospira Nitrospirales Thermodesulfovibrionaceae

Bacteria na na na na aGenus Sulfurimonas bGenus Gallionella cGenus Desulfococcus dGenus Rhodoplanes eGenus Hyphomicrobium

201

Table. S6. Unique taxonomic information for top 100 ASVs from shallow (0-50cm) active

microbial community for both sites combined.

Kingdom Bacteria Class Order Family










Bacteria na na na na

Bacteria Proteobacteria Epsilonproteobacteria Campylobacterales Helicobacteraceaea

Bacteria Proteobacteria Gammaproteobacteria na na

Bacteria Proteobacteria Betaproteobacteria MND1 na



Bacteria Proteobacteria Gammaproteobacteria Chromatiales Chromatiaceae













Bacteria Proteobacteria Gammaproteobacteria Chromatiales Ectothiorhodospiraceae




Bacteria Proteobacteria Alphaproteobacteria Rhizobiales na


Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobacteraceaeb






202







Bacteria Proteobacteria Deltaproteobacteria Myxococcales na


Bacteria Proteobacteria Deltaproteobacteria Syntrophobacterales Syntrophaceaed



Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobulbaceae

Bacteria Proteobacteria Deltaproteobacteria AF420338 na
































203







Bacteria Firmicutes Clostridia Clostridiales Clostridiaceaee

Bacteria GN04 GN15 na na


Bacteria LCP-89 SAW1_B44 na na






Bacteria Chloroflexi Dehalococcoidetes GIF9 na aGenus Sulfurimonas bGenus Desulfococcus cGenus Desulfosarcina dGenus Desulfobacca eGenus Clostridium

204

Table. S7. Top ASVs (n=35) identified from a random forest regression model for the total

microbial community (DNA), ordered by decreasing importance.


Bacteria Caldithrix Caldithrixae Caldithrixales na

Bacteria Chlamydiae Chlamydiia Chlamydiales na

Bacteria Bacteroidetes Bacteroidia Bacteroidales GZKB119







Bacteria Bacteroidetes Bacteroidia Bacteroidales SB-1



Bacteria Proteobacteria Zetaproteobacteria Mariprofundales Mariprofundaceae

Bacteria GN04 na na na

Bacteria Chloroflexi Anaerolineae Anaerolineales Anaerolinaceae


Bacteria Chloroflexi S085 na na






Bacteria WS2 SHA-109 na na


Bacteria Proteobacteria Gammaproteobacteria Marinicellales Marinicellaceae



Bacteria Proteobacteria na na na








205

Table S8. Top ASVs (n=41) identified from a random forest regression model for the active

microbial community (RNA), ordered by decreasing importance.


Bacteria OD1 na na na


Bacteria OD1 ABY1 na na

Bacteria SAR406 AB16 noFP_H7 na




Archaea Euryarchaeota Methanobacteria Methanobacteriales WSA2


Archaea Euryarchaeota Thermoplasmata E2 20c-4








Bacteria GN04 MSB-5A5 na na

Bacteria Chloroflexi Anaerolineae OPB11 na

Bacteria Chloroflexi Dehalococcoidetes GIF9 na





Bacteria CD12 na na na




Bacteria OP1 MSBL6 na na



Bacteria Proteobacteria Gammaproteobacteria Marinicellales Marinicellaceae




206





Bacteria BHI80-139 na na na


207

Table S9. Unique taxonomic information for top 100 ASVs from deep sediment total microbial

community (60+ cm) for the reference marsh.









Bacteria OP8 OP8_1 HMMVPog-54 na




















Archaea Crenarchaeota MCG pGrfC26 na












208





Bacteria Chloroflexi Anaerolineae na na

Bacteria Chloroflexi Anaerolineae Anaerolineales Anaerolinaceae


Bacteria Chloroflexi Dehalococcoidetes na na
















Bacteria AC1 SHA-114 na na





















209

Bacteria Proteobacteria Gammaproteobacteria Pseudomonadales Pseudomonadaceaea














Bacteria Planctomycetes Phycisphaerae CCM11a na


Bacteria na na na na aPseudomonas veronii bGenus Desulfococcus

210

Table S10. Unique taxonomic information for top 100 ASVs from deep sediment total microbial

community (60+ cm) for the enriched marsh.

















Bacteria Chlorobi BSV26 C20 na






Archaea Crenarchaeota MCG pGrfC26 na















Bacteria OP3 na na na




211

Bacteria Chloroflexi Ellin6529 na na
























Bacteria AC1 SHA-114 na na

















Bacteria Proteobacteria Gammaproteobacteria Pseudomonadales Pseudomonadaceaea



212
















Bacteria LD1 na na na aPseudomonas veronii bGenus Desulfococcus

213

Table S11. Unique taxonomic information for top 100 ASVs from deep sediment active

microbial community (60+ cm) for the reference marsh.


Archaea Euryarchaeota ANME-1 na na
















Bacteria Planctomycetes Planctomycetia Pirellulales Pirellulaceae






Bacteria Actinobacteria Actinobacteria Actinomycetales Intrasporangiaceae

Bacteria Actinobacteria Actinobacteria Actinomycetales Micrococcaceaea

Bacteria Proteobacteria Gammaproteobacteria Pseudomonadales Pseudomonadaceae

Bacteria Proteobacteria Gammaproteobacteria Vibrionales Pseudoalteromonadaceaeb


Bacteria Proteobacteria Betaproteobacteria Burkholderiales Oxalobacteraceaec

Bacteria Proteobacteria Betaproteobacteria Burkholderiales Comamonadaceaed

Bacteria Proteobacteria Gammaproteobacteria Pseudomonadales Moraxellaceaee







Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae



Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales Erythrobacteraceaef

214

Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales Sphingomonadaceaeg

Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales Rhodospirillaceaeh

Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobacteraceaei

Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobacteraceaei





Bacteria Proteobacteria Deltaproteobacteria Myxococcales Haliangiaceae


Bacteria NKB19 na na na



Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales na

Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales na






















Bacteria Spirochaetes [Brachyspirae] [Brachyspirales] A0-023








215



Bacteria OP8 OP8_1 na na













Bacteria Chloroflexi Dehalococcoidetes GIF9 na aMicrococcocus luteus bGenus Pseudoalteromonas cGenus Ralstonia dVariovorax paradoxus eGenus Enhydrobacter fGenus Erythrobacter gGenus Spingomonas hGenus Skermanella iGenus Desulfococcus

216

Table S12. Unique taxonomic information for top 100 ASVs from deep sediment active

microbial community (60+ cm) for the enriched marsh.


Archaea Euryarchaeota Thermoplasmata E2 20c-4







Bacteria Bacteroidetes Flavobacteriia Flavobacteriales Weeksellaceaea




Bacteria Planctomycetes Planctomycetia Pirellulales Pirellulaceae










Bacteria Actinobacteria Actinobacteria Actinomycetales Micrococcaceaeb

Bacteria Proteobacteria Gammaproteobacteria Enterobacteriales Enterobacteriaceaec

Bacteria Proteobacteria Gammaproteobacteria Vibrionales Vibrionaceaed

Bacteria Proteobacteria Betaproteobacteria Burkholderiales Burkholderiaceaee

Bacteria Proteobacteria Betaproteobacteria Burkholderiales Oxalobacteraceaef

Bacteria Proteobacteria Betaproteobacteria Burkholderiales Comamonadaceae

Bacteria Proteobacteria Betaproteobacteria Burkholderiales Comamonadaceaeg

Bacteria Proteobacteria Gammaproteobacteria Pseudomonadales Moraxellaceaeh

Bacteria Proteobacteria Gammaproteobacteria Pseudomonadales Moraxellaceaei

Bacteria Proteobacteria Gammaproteobacteria Pseudomonadales Moraxellaceaej






Bacteria Proteobacteria Alphaproteobacteria Rhizobiales Methylobacteriaceaek


Bacteria Proteobacteria Alphaproteobacteria Rhizobiales Rhodobiaceael


217

Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales Erythrobacteraceae

Bacteria Proteobacteria Deltaproteobacteria Desulfobacterales Desulfobacteraceaem































Bacteria Firmicutes Clostridia Clostridiales Clostridiaceaen


Bacteria Tenericutes Mollicutes na na

Bacteria Firmicutes Bacilli Bacillales Staphylococcaceaeo









218











Bacteria Chloroflexi Dehalococcoidetes Dehalococcoidales na






aGenus Cloacibacterium bMicrococcus luteus cGenus Yersinia dGenus Vibrio eGenus Burkholderia fGenus Ralstonia gVariovorax paradoxus hGenus Enhydrobacter iAcinetobacter johnsonii jGenus Acinetobacter kGenus Methylobacterium lGenus Afifella mGenus Desulfococcus nClostridium bowmanii oGenus Staphylococcus

219


Fig. S1. (A) Principal coordinate analysis (PCoA) constructed from weighted Unifrac of the

active community (RNA) colored by depth, with no effect of site according to a PERMANOVA.

(B) The first PCoA axis versus depth shows the microbial community structure changes

drastically with depth in the top 50 cm, but then becomes relatively stable 60+cm.

220

Appendix: Nitrate reduction pathways and functional potential in response to nutrient

enrichment

In collaboration with: Joseph Vineis, Anna E. Murphy, Amanda C. Spivak, Anne E. Giblin, Jane

Tucker

Background and Objectives:

Salt marshes are efficient at removing excess nutrients from that land that could have

detrimental effects in coastal waters, such as harmful algal blooms, anoxia/hypoxia, and

increased turbidity (Anderson et al. 2002, Diaz & Rosenberg 2008). Situated at the intersection

between the land and sea, salt marshes can intercept nutrients either by assimilation into plant

biomass (Valiela & Teal 1974) or through competing nitrogen (N) cycling processes that use

nitrate (NO3-) as an electron acceptor, such as denitrification (DNF) or dissimilatory nitrate

reduction to ammonium (DNRA). Partitioning which microbial NO3- reduction processes are

occurring is critical for understanding the fate of N in coastal systems, because while DNF

results in the removal of biologically available N in the form of N2 gas (Kaplan et al. 1979,

Seitzinger 1988), DNRA retains this N in the system as NH4+ (An & Gardner 2002, Gardner et

al. 2006, Giblin et al. 2013), which can then stimulate primary production or be transformed

back to NO3- through nitrification. High NO3

- conditions appear to favor DNF (Giblin et al. 2013,

Tiedje 1988), while NO3- limitation and labile organic matter appear to favor DNRA (Algar &

Vallino 2014, Burgin & Hamilton 2007, Hardison et al. 2015), however, the relative contribution

of these controls are not well understood. Being able to distinguish controls over competition

among these microbial pathways is important, as they produce different byproducts with

important implications for water quality. Further, it is unclear how partitioning among these

processes will change with increasing nutrient enrichment, as N loading continues to occur in

221

coastal waters with largely unknown consequences for salt marsh systems (Deegan et al. 2012,

Galloway et al. 2017).

One informative way to understand controls on these important rates is to concurrently

study the microbial community responsible for these processes, along with the fluxes they

produce, under changing conditions. The microbial consortia inhabiting salt marsh sediments

mediate much of the biogeochemical transformations involved in N cycling (Falkowski et al.

2008, Jetten 2008, Kuypers et al. 2018). Despite this critical role, comparatively little is known

about the microbes that facilitate these biogeochemical pathways and how they will respond to

nutrient enrichment. Metagenomics analysis allows us to investigate the genetic diversity and

functional potential of the resident microbial community (Franzosa et al. 2015). The technique

does not rely on culturing member organisms, which is notoriously challenging, and avoids

biases involved in Polymerase Chain Reaction (PCR) amplification that may limit sequence

information we can obtain using techniques such as 16S rRNA sequencing (Zhou et al. 2015). It

also has the potential to reveal phylogenetic and genomic novelty, providing insight into

previously unknown lineages of subsurface microbes we know surprisingly little about.

I used a controlled flow through experiment to examine competing NO3- reduction

pathways and assessed shifts in functional potential in response to nutrient enrichment in salt

marsh sediments across a depth gradient varying in organic matter lability. I hypothesized that

NO3- reduction rates would be greatest in surface sediments, where organic matter is most labile,

and that the contribution of DNRA relative to denitrification would decrease as organic matter

becomes more recalcitrant. I also hypothesized that the addition of NO3- would result in a shift in

the functional potential of the microbial community when compared to an unamended control,

resulting in greater abundance of genes associated with the N cycle. Consequently, this shift in

222

the functional potential would coincide with degradation of more complex forms of organic

matter. The outcome of this work not only provides information on the controls over competing

NO3- reduction processes, but also adds to our understanding of the response of microbes to N

enrichment and its effect on carbon storage.

Methods:

Sample collection & experimental design

To test my hypotheses, I collected sediment along a depth gradient in the tall ecotype of

Spartina alterniflora at West Creek, a relatively pristine marsh complex located in Plum Island

Sound, MA (42.759 N, 70.891 W) that is monitored as part of a long-term enrichment

experiment called the TIDE project (Deegan et al. 2007). I collected three replicate cores (5 cm

diameter, 30 cm deep) and sectioned them under anoxic conditions into shallow (0-5 cm), mid

(10-15 cm), and deep (20-25 cm) depths, thus representing a range in OM quality from newly

deposited material to sediments ranging from 50-100 years in age (Wilson et al. 2014, Forbrich

et al. 2018). After homogenizing sediment under anoxic conditions and removing as much root

material as possible, I split each core section into a plus-NO3- and an unamended treatment

(filtered seawater), resulting in three replicates for each treatment and depth combination.

A more detailed description of the methodology I followed for this experiment can be

found in chapter 1 of this thesis. Briefly, I loaded flow through reactor (FTRs; modified from

Pallud et al. (2006; 2007) with homogenized sediment under anoxic conditions, and randomly

assigned each reactor a treatment, either plus-NO3- (+NO3

- in 0.2 µm filtered seawater) or

unamended (0.2 µm filtered seawater only, representing natural marsh conditions). I prepared

each reservoir with filtered water from Woods Hole, MA by sparging with N2 gas for

223

approximately 20 minutes and spiking the plus-NO3- reservoir with an additional 500 µmol L-1

K15NO3- (Cambridge Isotope Laboratories, Andover, MA). For the first 25 days of the

experiment, I only added 350 µmol L-1 15NO3-, but since NO3

- was being fully consumed, I

increased the added concentration to 500 µmol L-1 to ensure it was never limiting. Half of the

reactors (n = 9) received the plus-NO3- treatment and half (n = 9) received the unamended

treatment, both at a targeted flow rate of 0.08 mL min-1 using MasterFlex FDA viton tubing

(Cole Parmer, IL, USA). The experiment lasted for 92 days, throughout which I collected water

samples to assess microbial respiration and NO3- reduction pathways. I also collected sediment

from before (pre; n = 9) and after (post; n = 18) the experiment to assess OM characteristics and

microbial community functional potential by flash-freezing sediment and storing at -80ºC until

further analysis.

Metabolism measurements

I collected water samples from both the plus-NO3- and unamended treatment effluent, as

well as each corresponding reservoir, to assess biogeochemical processes as a result of microbial

activity in each reactor. I measured dissolved inorganic carbon (DIC; CO2 + HCO3 + CO32-) as

an indicator of total microbial respiration on an Apollo SciTech AS-C3 DIC analyzer (Newark,

DE) and NO3- consumption as an indicator of assimilation/dissimilation on a Teledyne T200

NOx analyzer (Teledyne API, San Diego, CA) using chemoluminescent methods (Cox 1980). I

also collected water samples in airtight cut-off volumetric pipettes, preserved a portion with zinc

chloride (ZnCl2) for denitrification measurements, and froze ~50 mL sample at -20ºC for DNRA

measurements.

To determine the relative contribution of each NO3- reduction pathway, I made dissolved

gas measurements of N2 on a membrane inlet mass spectrometer (Kana et al. 1994) connected to

224

an inline furnace set to 500ºC and copper column to remove oxygen interferences (Eyre et al.

2002; Lunstrum & Aoki 2016). I monitored the production of 29N2 and 30N2 from added 15NO3-

tracer as a measure of denitrification (D15) and calculated rates using the following equation from

Nielson et al. 1992:

Eq. 1 D15 = p29+2p30

where p29 and p30 represent production of 29N2 and 30N2, respectively. Because I only added

NO3- in the form of 15NO3

-, and ambient concentrations of 14NO3- were very low to below

detection (as high as 0.6-1.2 µM), I did not calculate D14. Furthermore, I ignored the production

of 14NO3- from nitrification, which is a largely aerobic process (Herbert 1999), since this

experiment was conducted under strictly anoxic conditions. To measure DNRA, I bubbled water

samples with helium for 10 minutes to remove any N2 and converted 15NH4+ produced from

DNRA to 29N2 and 30N2 using sodium hypobromite following OX/MIMS methodology as

outlined in Yin et al. (2014). I then calculated DNRA as the following:

Eq. 2 DNRA15 = p29+2p30

It is important to note that I am not attempting to calculate ambient rates nor making these

measurements in the unamended treatment, where I did not add any 15NO3-.

Once I made measurements of each analyte, I calculated either production (DIC, N2 gas)

or consumption (NO3-) as a rate following eq. 3 (Pallud et al. 2006):

Eq. 3 R =(Cout − Cin)Q

V

where R is the consumption or production rate of interest, Cout and Cin are the effluent and

reservoir analyte concentrations, respectively, Q is the measured flow rate in L hour-1, and V is

the FTR volume (31.81 cm-3). From this, I also calculated the cumulative flux by integrating

between each measured time point across the length of the experiment.

225

Lipid Extraction and analysis

I extracted and analyzed lipid biomarker compounds from the sediments after the

experiment using a modified method from Bligh & Dyer (1959). To extract total fatty lipids, I

mixed ~3 g wet sediment with a chloroform:methylene chloride:phosphate buffer

(MeOH:CHCl3:PBS) saline mixture (2:1:0.8) and heated to 80ºC in a microwave-accelerated

reaction system (MARS6). I then partitioned samples with a 1:1:0.9 ratio of MeOH:CHCl3:PBS,

removed the organic phase, and concentrated the samples under N2. To separate each sample into

fractions of neutral and glycolipids (F1/2) and phospholipids or PLFAs (F3/4; Guckert et al.

1985), I used a silica gel column and eluted with chloroform, acetone, and methanol,

respectively. I dried the PLFAs under N2 and saponified them with 0.5 M sodium hydroxide

(NaOH) at 70ºC for 4 hours following Osburn et al. (2011). I then acidified the samples with 3

mL 3N hydrochloric acid (HCl) before extracting 3x with hexane. To methylate the PLFA

extract, I added acidic methanol (95:5 methanol:HCl) and heated overnight at 70 ºC to form fatty

acid methyl esters (FAME). I analyzed FAMEs on an Agilent 7890 gas chromatograph mass

spectrometer (Agilent, Santa Clara, CA) with a flame ionization detector located at Woods Hole

Oceanographic Institution using a DB-5 column with methyl heneicosanoate (Supelco 37

Component FAME mix) as an internal standard following methods in Canuel et al. (2007) and

references therein. I designated fatty acids as A:BωC, where A is the number of carbon atoms, B

represents the number of double bonds, and C indicates the \ position of the double bond relative

to the aliphatic end of the molecule as designated by “ω” (Canuel et al. 1995). I then calculated

the µg of fatty acid per total grams of organic carbon. An overview of the compounds I analyzed

in this study can be found in Table 1.

DNA extraction, library preparation, and sequencing

226

I extracted genomic DNA from approximately 0.25 g wet sediment using the MoBio®


instructions, and eluted the DNA into a 35 µL final volume. I then confirmed DNA quality

(260/280) and concentration (ng/µL) using a NanoDrop (ThermoFisher Scientific, Waltham,

MA). To shear DNA at a 270 bp target size, I transferred ~100 ng DNA into a V2 8-microtube

strip and ran it on a Covaris ME220 focused-ultra sonicator (Covaris Inc., Woburn, MA) at 1000

cycles/burst, 20% duty factor, and 70% peak power for 88 seconds per sample.

I prepared 27 metagenomic libraries using the NuGEN Ovation Ultralow System V2

(NuGEN, San Carlos, CA), performing end repair and barcode ligation with the recommended

PCR cycling conditions (25ºC for 30 min and 70ºC for 10 min) and purification methods

(Agencourt AMPure XP; Beckman Coulter, Pasadena, CA). I then amplified the final library

under the following conditions: 72ºC for 2 min, 95ºC for 3 min, 9 cycles (98ºC for 20 sec, 65ºC

for 30 sec, 72ºC for 30 sec), and 72ºC for 1 min, and size-selected on a per-sample basis at 390

bp using a PippinPrep (Sage Science, Beverly, MA). After confirming a target insert size of 270

bp on an Agilent 4200 TapeStation (Agilent Technologies Inc, Santa Clara, CA), I quantified

each library using a KAPA library quantification kit (Roche Sequencing, Pleasanton, CA) and

performed sequencing on an Illumina NextSeq Hi-Output 2x150 Illumina flow cell (Illumina Inc,

San Diego, CA) at the Marine Biological Laboratory Keck Facility (Woods Hole, MA).

Sequence analysis and annotation

I joined paired end reads using illumina-utils with a P=0.1, which allows for 1 error in

every ten bases, and used a PHRED score of 30 to remove any regions with low quality reads,

resulting in 93% of raw reads retained after quality filtering (Table 2). I submitted merged reads

227

to the MG-RAST server (Meyer et al. 2008) and performed functional annotation using SEED

subsystems (Aziz et al. 2008) with an 80% minimum cutoff identity.

Statistical analyses

I examined rates of denitrification and DNRA over time and calculated cumulative rates

by integrating between each measured point across the length of the experiment. To compare

cumulative rates by depth as the factor, I ran a one-way ANOVA using core replicate as a

random effect for each process separately. I then divided DIC by the sum of denitrification and

DNRA and plotted values as boxplots by depth to better understand the contribution of these

processes to total microbial respiration. To assess how much each process compared to NO3-

reduction, I plotted denitrification and DNRA as stacked bar plots against NO3- consumption,

which was calculated as the difference between the reactor effluent and treatment reservoirs.

Using all subsystem annotations present in the dataset with at least 80% cutoff identify, I

constructed a non-dimensional scaling plot and tested for significance by depth and treatment

using ‘adonis’ in the Vegan package in R (Oksanen et al. 2017; R Core Team). Finally, to

examine the response of functional potential to nutrient enrichment, I constructed a heat map

showing normalized abundance of genes associated with N cycling for the unamended and plus-

NO3- treatment, which were clustered using unweighted pair group method with arithmetic mean

(UPGMA) on Euclidean distances (Michener & Sokal 1957). Lastly, I examined differences

among total fatty acids and various sub-classes (Table 1) by treatment and depth.

Results and general conclusions:

Denitrification and DNRA were higher in shallow sediments when compared to mid and deep

sediments (Fig. 1, 2A-B): Denitrification rates were similar half way throughout the experiment,

228

which suggests that NO3- addition stimulated microbial metabolism to nearly the same rates

despite differences in organic matter lability that occurs with depth. These rates then diverged

~week 10, likely as a result of decreased organic matter lability over time. DNRA, on the other

hand, exhibited stimulated rates in the shallow sediments; however, rates were lower in the mid

and deep sediments, likely because organic matter was less labile at those depths. When

examining cumulative rates, both denitrification (p=0.05, F2,6=6.48) and DNRA (p<0.001,

F2,6=27.29) were highest in the shallow sediments when compared to the mid and deep

sediments. This is likely a reflection of organic matter lability, where compounds available for

microbial oxidation tend to decrease with depth (Middelburg 1989, Cowie & Hedges 1994).

The relative contribution of DNRA to DIC production was greatest in the shallow sediments

(Fig. 2C): The stoichiometry of DIC to NO3- reduction is 1.25 and 2 for denitrification and

heterotrophic DNRA, respectively (Canfield et al. 2005, Giblin et al. 2013). By dividing DIC by

the sum of measured rates of denitrification and DNRA, I can get a better sense for the relative

contribution of these processes to total microbial respiration. I found that, across all depths, this

ratio averaged 1.42 ± 0.08 (shallow), 1.42 ± 0.15 (mid), and 1.33 ± 0.10 (deep). Statistically

equivalent ratios across depth (p=0.84) suggest that, in conjunction with lower denitrification and

DNRA rates in deeper sediments, DIC also decreased such that the ratio remained the same.

DNF and DNRA account for the majority of nitrate consumption throughout the experiment (Fig.

3): Across all depths, denitrification and DNRA rates account for the majority of NO3-

consumption, suggesting minimal contribution of assimilative processes in this experiment. This

supports the notion that NO3- can play a large role as an electron acceptor fueling heterotrophic

229

microbial metabolism, which in part explains why some marsh fertilization experiments observe

decreases in belowground biomass (Darby & Turner 2008, Deegan et al. 2012).

Functional potential differed by treatment, with NO3- enrichment resulting in significantly

different genetic composition (Fig. 4): In addition to the evidence I have that NO3- leads to

significant shifts in microbial community structure (Fig. 8 of chapter 1 and Fig. 8 of chapter 2),

particularly towards taxa associated with the N cycle (Fig. 9 of chapter 1 and Fig. 9 of chapter 2),

these data also show that community functional potential changes as well. Metagenomes from

the nitrate-enriched samples were significantly different from those in the unamended and pre

sediments (p=0.001, pseudo-F = 30.50). It is also interesting that there seems to be separation by

depth (p=0.001, pseudo-F = 8.88), with the shallow sediments from pre-incubation exhibiting

different functional composition when compared to mid and deep sediments from the pre and

unamended treatments. This suggests that experimental conditions may have resulted in

homogenization of the microbial community.

Out of those genes that were altered in response to NO3- addition, many of them that increased

are associated with N cycling (Fig. 5): A heatmap of genes associated with N cycling exhibited

clear differences in relative abundance when comparing the plus-NO3- and unamended

treatments. While genes encoding for enzymes associated with some maintenance functions,

such as ammonium transport, were similar across treatments, other genes associated with

dissimilatory N reduction, such as nitrite reductases (nir), nitric oxide reductases (nor), and

nitrous oxide reductases (nos) all increased in abundance in the plus-NO3- treatment (Kuypers et

al. 2018). This suggests that the addition of NO3- fundamentally altered the genetic make-up of

230

these microbial communities, increasing functional potential to carry out N transformations.

Future work should consider whether these changes in gene abundance correlate with

biogeochemical measurements of ecosystem function, which is a central goal in microbial

ecology.

It is also interesting to note that genetic potential is most similar by depth in the

unamended sediments, but this pattern does not hold in the plus-NO3- treatment. The addition of

NO3- appears to alter the microbial community such that organic matter lability is not as critical

in defining composition. This supports the idea that NO3- acts as an electron acceptor in

microbial metabolism, and allows for the decomposition of more complex organic matter, as

proposed in chapter 1 of my dissertation. Overall, this work provides a high resolution

characterization of the microbes responsible for N cycling, particularly in response to nutrient

enrichment.

Lower fatty acid concentrations in plus-NO3- treatment suggest enhanced microbial organic

matter degradation (Fig. 6): Fatty acids are a class of lipid biomarkers that demonstrate high

source fidelity and, depending on its source, exhibits a range of chemical reactivity (Canuel et al.

1995). By examining different sub-classes of fatty acids under various environmental conditions,

I can assess both sediment organic matter quality and decomposition in relation to its source. To

test the effect of nutrient enrichment on organic matter decomposition, I quantified phospholipid-

linked fatty acids (PFLAs), which are indicators of recently viable cells, between the plus-NO3-

and unamended treatments at the end of the FTR experiment.

I found that total fatty acid concentration was lower in the plus-NO3- treatment in shallow

and deep sediments (Fig. 6A), suggesting more overall organic matter loss at these depths when

231

compared to the unamended control. This pattern was not evident in the mid sediments. For both

the algal (PUFA and SCFA; Fig. 6G,B) and microbial-derived (MUFA and 10-Methyl C16:0; Fig.

6F, I) fatty acid compounds, concentrations were lower in the plus-NO3- treatment. A similar

pattern also occurred for the less labile compounds that were detritus-based (LCFA; Fig. 6H) and

Spartina derived (C18:2+C18:3; Fig. 6D), although this difference only existed in the deepest

sediments for the LCFAs. Overall, these results suggest that by adding NO3- and providing a

more energetically favorable electron acceptor for microbial metabolism, I am stimulating

organic matter decomposition. In particular, as a result of this stimulation, microbes may be

accessing more complex organic matter that would otherwise remain stable under ambient

nutrient conditions, resulting in decreased capacity for carbon storage in salt marsh systems.

232

References






Anderson DM, Glibert PM, Burkholder JM (2002) Harmful algal blooms and eutrophication:

nutrient sources, composition, and consequences. Estuaries 25:704–726

Aziz RK, Bartels D, Best A, DeJongh M, Disz T, Edwards RA, Formsma K, Gerdes S, Glass

EM, Kubal M, Meyer F, Olsen GJ, Olson R, Osterman AL, Overbeek RA, McNeil LK,

Paarmann D, Paczian T, Parrello B, Pusch GD, Reich C, Stevens R, Vassieva O,

Vonstein V, Wilke A, Zagnitko O (2008) The RAST Server: Rapid annotations using

subsystems technology. BMC Genomics 9:1–15

Bligh EG, Dyer WJ (1959) A rapid method of total lipid extraction and purification. Can J

Biochem Physiol 37:911–917

Burgin AJ, Hamilton SK (2007) Have we overemphasized the role of denitrification in aquatic

ecosystems? A review of nitrate removal pathways. Front Ecol Environ 5:89–96


Press, Boston, MA

Canuel E, Cloern J, Ringelberg D, Guckert JB, Rau GH (1995) Molecular and isotopic tracers

used to examine sources of organic matter and its incorporation into the food webs of San

Francisco Bay. Limnol Oceanogr 40:67–81




matter mixtures. Nature 369:304








233

Diaz RJ, Rosenberg R (2008) Spreading dead zones and consequences for marine ecosystems.

Science 321:926–929

Eyre BD, Rysgaard S, Dalsgaard T, Christensen PB (2002) Comparison of isotope pairing and

N2:Ar methods for measuring sediment denitrification – assumption, modifications, and

implications. Estuaries 25:1077-1087





878

Franzosa EA, Hsu T, Sirota-Madi A, Shafquat A, Abu-Ali G, Morgan XC, Huttenhower C

(2015) Sequencing and beyond: integrating molecular “omics” for microbial community

profiling. Nat Rev Microbiol 13:360–72

Galloway JN, Leach AM, Erisman JW, Bleeker A (2017) Nitrogen: The historical progression


Gardner WS, McCarthy MJ, An S, Sobolev D, Sell KS, Brock D (2006) Nitrogen fixation and

dissimilatory nitrate reduction to ammonium (DNRA) support nitrogen dynamics in

Texas estuaries. Limnol Oceanogr 51:558–568




Guckert JB, Antworth CP, Nichols PD, White DC (1985) Phospholipid, ester-linked fatty acid

profiles as reproducible assays for changes in prokaryotic community structure of

estuarine sediments. FEMS Microbiol Lett 31:147–158

Hardison AK, Algar CK, Giblin AE, Rich JJ (2015) Influence of organic carbon and nitrate

loading on partitioning between dissimilatory nitrate reduction to ammonium (DNRA)

and N2 production. Geochim Cosmochim Acta 164:146–160

Herbert RA (1999) Nitrogen cycling in coastal marine ecosystems. FEMS Microbio Rev 23:563-

590

Jetten MSM (2008) The microbial nitrogen cycle. Environ Microbiol 10:2903–9

Kana TM, Darkangelo C, Hunt MD, Oldham JB, Bennett GE, Cornwell JC (1994) Membrane

inlet mass spectrometer for rapid high-precision determination of N2, O2, and Ar in

environmental water samples. Anal Chem 66:4166–4170

234


24:726–734

Kuypers MMM, Marchant HK, Kartal B (2018) The microbial nitrogen-cycling network. Nat

Rev Microbiol 16:263–276

Lundstrum A, Aoki LR (2016) Oxygen interference with membrane inlet mass spectrometry may

overestimate denitrification rates calculated with isotope pairing technique. Limnol

Oceanogr: Meth 14:425-431

Meyer F, Paarmann D, D’Souza M, Etal. (2008) The metagenomics RAST server—a public

resource for the automatic phylogenetic and functional analysis of metagenomes. BMC

Bioinformatics 9:386

Michener CD, Sokal RR (1957) A quantitative approach to a problem in classification. Evolution

11:130–162

Middelburg JJ (1989) A simple rate model for organic matter decomposition in marine

sediments. Geochim Cosmochim Acta1 53:1577–1581




Osburn MR, Sessions AL, Pepe-Ranney C, Spear JR (2011) Hydrogen-isotopic variability in

fatty acids from Yellowstone National Park hot spring microbial communities. Geochim

Cosmochim Acta 75:4830–4845





Tiedje J (1988) Ecology of the denitrification and dissimilatory nitrate reduction to ammonium.

In: Zehnder A (ed) Biology of Anaerobic Microorganisms. Wiley and Sons, New York.

NY, p 179–244






235

Yin G, Hou L, Liu M, Liu Z, Gardner WS (2014) A novel membrane inlet mass spectrometer

method to measure 15NH4+ for isotope-enrichment experiments in aquatic ecosystems.

Environ Sci Technol 48:9555–9562

Zhou J, Zhili H, Yang Y, Deng Y, Tringe SG, Alvarez-Cohen L (2015) High-throughput

metagenomic technologies for complex microbial community analysis: open and closed

formats. CEUR Workshop Proc 1542:33–36

236

Tables

Table 1. Sub-class groups used in this study

Source Subclass Abbreviation Lipid Number

Algal & microbial Short Chain Fatty

Acids SCFA C12:0+C14:0

Sediment heterotrophic

bacteria

Branched Chain

Fatty Acids BCFA

Iso- and anteiso-

C13:0+C15:0+C17:0+C19:0

Sediment microbial and

bacterial OM (sulfate

reducers)

Short Chain Fatty

Acid

10-Methyl

C16:0 10-Methyl C16:0

Microbial OM Monounsaturated

Fatty Acid MUFA C16:1

Microbial OM Monounsaturated

Fatty Acid MUFA C16:1+C17:1+C18:1+C19:1

Spartina derived OM Polyunsaturated

Fatty Acid PUFA C18:2+C18:3

Labile algal OM &

epiphytic diatoms

Polyunsaturated

Fatty Acid PUFA C20:4+C20:5

Detritus derived from

plants

Long Chain Fatty

Acids LUFA C24:0+C26:0+C28:0+C30:0

237

Table 2. Number of basepairs and quality-filtered sequences per metagenome submitted to the

MG-RAST server for annotation

Sample Name Treatment Depth # Basepairs # Sequences

1SPRE_merged Pre Shallow 8.06E+09 3.35E+07

2SPRE_merged 5.74E+09 2.30E+07

3SPRE_merged 4.20E+09 1.76E+07

1MPRE_merged Mid 4.95E+09 2.00E+07

2MPRE_merged 5.51E+09 2.23E+07

3MPRE_merged 5.49E+09 2.25E+07

1DPRE_merged Deep 5.09E+09 2.05E+07

2DPRE_merged 8.47E+09 3.39E+07

3DPRE_merged 7.82E+09 3.11E+07

N1SPOST_merged Nitrate Shallow 5.43E+09 2.19E+07

N2SPOST_merged 4.32E+09 1.85E+07

N3SPOST_merged 5.47E+09 2.30E+07

N1MPOST_merged Mid 5.55E+09 2.21E+07

N2MPOST_merged 4.68E+09 1.91E+07

N3MPOST_merged 5.88E+09 2.42E+07

N2DPOST_merged Deep 5.30E+09 2.16E+07

N3DPOST_merged 7.29E+09 3.03E+07

S1SPOST_merged Unamended Shallow 5.38E+09 2.13E+07

S2SPOST_merged 5.52E+09 2.23E+07

S3SPOST_merged 4.96E+09 2.03E+07

S2MPOST_merged Mid 1.14E+10 4.64E+07

S3MPOST_merged 5.29E+09 2.19E+07

S1DPOST_merged Deep 5.73E+09 2.25E+07

S2DPOST_merged 5.32E+09 2.31E+07

S3DPOST_merged 6.16E+09 2.67E+07

238

Figures

Fig. 1. Average ± SE denitrification (DNF) and dissimilatory nitrate reduction to ammonium

(DNRA) for shallow (A,D), mid (B,E), and deep (C,F) sediments.

239

Fig. 2. (A) Cumulative denitrification (B) DNRA and (C) ratio of cumulative DIC to the sum of

cumulative DNF+DNRA, with the red line representing the stoichiometrically expected ratio of

1.25 and 2 for DNF and DNRA, respectively. Boxes represent 25% to 75% quartiles. The solid

black line is the median value, and the whiskers are upper and lower extremes. Black dots represent

values for each individual reactor (n = 3). Letters represent statistically different DIC production

across sites from a Tukey’s HSD test corrected for multiple comparisons test and asterisks indicate

a significant difference between treatments.

240

Fig. 3. Average ± SE denitrification and DNRA rates over time (weeks) for (A) shallow (B) mid

and (C) deep sediments. Average ± SE nitrate consumption rates indicated by black line.

241

Fig. 4. Non-metric multidimensional scaling plot of subsystems-level functional annotations by

treatment (color) and depth (shape).

242

Fig. 5. Heatmap showing normalized abundance of genes associated with nitrogen cycling for

the unamended and plus-NO3- treatment, which were clustered using unweighted pair group

method with arithmetic mean (UPGMA) on Euclidean distances. Shapes represent sample depth.

Lighter colors represent lower abundance while darker colors represent higher abundance.

243

Fig. 6. Boxplots colored by treatment of (A) total fatty acids and various subclasses of

phospholipid-linked fatty acid (PLFA) subclasses representing (B) short-chain fatty acids

(SCFA; C12:0+C14:0) (C) mono-unsaturated fatty acid (C16:1) (D) polyunsaturated fatty acids

(C18:2+C18:3) (E) branched chain fatty acid (BCFA; iso- and anteiso-C13:0, C15:0, C17:0, C19:0) (F)

mono-unsaturated fatty acids (MUFA) (G) Poly-unsaturated fatty acids (PUFA; C20:4+C20:5) (H)

Long-chain fatty acids (LCFA; C24:0+C26:0+C28:0+C30:0) (I) 10-methyl C16:0.

nitrate as an electron acceptor in microbial decomposition ...cj82st421/fulltext.pdf1 nitrate as an...

Documents