PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DAIEDUE

DAIEDUE

DAIEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6107 p:/ClRC/Date0ue.indd-p.1

 

ALTERNATIVE MICROBIAL PATHWAYS OF NITROGEN REMOVAL FROM
MICHIGAN STREAMS, WETLANDS AND LAKES

By

Amy J. Burgin

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Zoology
and
Program in Ecology, Evolutionary Biology, and Behavior

2007

ABSTRACT

NOVEL MICROBIAL PATHWAYS OF NITRATE REMOVAL FROM
FRESHWATER ECOSYSTEMS

By
Amy J. Burgin

The removal of nitrogen (N) in aquatic ecosystems is of particular
interest because excessive nitrate in ground and surface waters is a growing
problem. Research on nitrate removal processes has emphasized biotic
uptake (assimilation) or respiratory denitriﬁcation by bacteria. The increasing
application of tracer techniques (e.g., stable isotopes) has yielded a growing
body of evidence for alternative microbially mediated processes of nitrate
transformation, including dissimilatory reduction of nitrate to ammonium
(DNRA), chemoautotrophic denitriﬁcation via sulfur or iron oxidation, and
anaerobic ammonium oxidation (Anammox). In Chapter 1, I review evidence
for the importance of alternative nitrate removal pathways in aquatic
ecosystems and discuss how the possible prevalence of these pathways may

alter views of N cycling and its controls.

Anaerobic microbial processes are responsible for much of the nutrient
cycling in freshwater systems. Nitrate disappearance in sediments is usually
assumed to be due to respiratory denitriﬁcation. Push-pull tracer experiments
entail adding nitrate and a conservative solute to sediment porewater,

followed by in-situ incubation with periodic subsampling. While performing

such tracer experiments to quantify rates of nitrate removal in aquatic
sediments of Michigan streams and wetlands, I found that nitrate removal
coincided with sulfate production. Push-pull experiments in a diverse set of
streams, lakes and wetlands revealed a persistent pattern of sulfate
production during nitrate removal (Chapter 2). Push-pull experiments done
with ‘5N03' also indicate the importance of DNRA to overall. nitrate removal in
these sediments.

To compare the relative importance of alternative pathways of NO3'
reduction (e.g., to NH,“ or N2), l again employed the use of stable isotopes in
conjunction with a ﬂow-through core technique (Chapter 3). Using a ﬂow-
through set up, treatment water (15N03’, 15NH4"/1"N03', or control) was
pumped over cores from six different sites. Results indicate that conversion
to N2 was the predominant nitrate loss across all six sites. I also found that
conversion into the 15NH4+ pool, indicative of DNRA, can account for a
variable fraction of the dissimilatory nitrate removal, but that anammox
accounted for very little of the overall nitrate removal.

I tested the relative importance of carbon vs. sulﬁde in regulating
DNRA using a laboratory assay (Chapter 4), by adding nitrate along with a
gradient of organic carbon (as acetate) and free sulﬁde to anoxic sediments.
I found that both carbon and sulﬁde were important in controlling nitrate
removal rates and end-products in both sites. While denitriﬁcation tended to
be the more important removal pathway in the low ambient sulﬁde site, DNRA

was of equal importance in the high ambient sulﬁde site.

DEDICATION

This dissertation is dedicated to the memory of my maternal grandfather,

Virgil Schurman
(1918-2002)

He was a farmer who only ﬁnished school through the 8th grade. Shortly before
he died, I told him what I was going to study in graduate school, and he replied, “I
always knew that nitrogen we added had to go somewhere..."

ACKNOWLEDGEMENTS

I would like to thank my advisor, Dr. Steve Hamilton, for his advice and guidance
throughout my dissertation. I would also like to thank my committee members,
Dr. Nathaniel Ostrom, Dr. Phil Robertson, Dr. Tom Schmidt, and Dr. Jay Lennon
for their feedback and help with many different stages of this process. Even
though they are not on my committee, I learned a great deal from interacting with

Kay Gross, Gary Mittelbach and Jeff Conner and other K88 and MSU faculty.

This ﬁeld and lab work would not have been possible without the help of David
Weed, Philip Riekenburg, Erin Payne, Melinda Roth, Nicole Reid, Stefanie
Whitmire, Eric Thobaben, Amanda Kurzman, Suzanne Sippel, Kathy Crowley. I
would also like to thank my newer Iabmates Jorge Celi, Leila DeSotelle, Lauren

Kinsman and Jon O’Brien for the camaraderie and feedback.

KBS is a special place to me, and I made many wonderful life-long friends while I
was here. I would especially like to thank Wendy Mahaney and Kurt Smemo for
being such great friends and for their help with science and great conversations
about how to do science. Thank you to Claire McSwiney, who was also a good
friend and role model, and willing to listen when the instruments weren’t
cooperating. Many other people also made KBS a great place to live and work,
including Katie Wharton, Lindsey Walters, Brook and Emily Wilke, Emily Grman,

Stuart Grandy, Jay Sobel, Charlotte Reemts, Jarad Mellard, and Todd Robinson.

KBS is such and easy place to live and work because of the help of numerous
support staff. I’d like to thank Nina Consollati, Sally Shaw, Alice Gillespie, John

Gorentz, and Melissa Yost for their assistance.

I have been fortunate to work with a large number of non-KBS people during my
graduate studies as well. I gained a great deal from interacting with our LINX
collogues, in particular the Notre Dame students (Jake, Laura, Clay and Denise).
Pat Mulholland provided a review of Chapter 1 before it was submitted for
publication, and his comments improved the manuscript (and chapter). I owe a
great deal to Wayne Gardner and Mark McCarthy at UTMSI for hosting my visit
there and training me on the ﬂow-through core system used in chapter 3 (and

running all of the gas samples!).

I owe special thanks to my husband and colleague, Terry Loecke, for his
encouragement and support (and ﬁeld assistance, and reading drafts of
chapters). My family, Jay and Cindy Burgin and my siblings Jerrod and Sarah,
have been wonderfully supportive of my many years of school, and didn’t ask too

often when l was going to ﬁnish.

This research was funded by grants from the National Science Foundation
(DEB). I obtained additional funds through the LTER network, MSU Zoology
Dept, MSU Ecology, Evolutionary Biology and Behavior Program, Environmental

Science and Policy Program, and Sigma Xi.

vi

TABLE OF CONTENTS

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

LIST OF FIGURES ................................................................................ xi

CHAPTER 1: HAVE WE OVEREMPHASIZED THE ROLE OF
DENITRIFICATION IN AQUATIC ECOSYSTEMS? A REVIEW
OF NITRATE REMOVAL PATHWAYS ........................................................ 1

CHAPTER 2: THE PREVALENCE OF SIMULTANEOUS NITRATE
REMOVAL AND SULFATE PRODUCTION IN STREAM, LAKE

AND WETLAND SEDIMENTS ................................................................ 31
INTRODUCTION ........................................................................ 31
METHODS ................................................................................ 36
RESULTS ................................................................................. 43
DISCUSSION ............................................................................. 61
REFERENCES ........................................................................... 69

CHAPTER 3: THE RELATIVE IMPORTANCE OF
DISSIMILATORY NITRATE REDUCTION TO AMMONIUM
(DNRA), DENITRIFICATION AND ANAEROBIC AMMOINUM

OXIDATION (ANAMMOX) IN FRESHWATER SEDIMENTS .......................... 72
INTRODUCTION ........................................................................ 72
METHODS ................................................................................. 78
RESULTS ................................................................................. 81

' DISCUSSION ............................................................................. 97
REFERENCES ......................................................................... 110

vii

CHAPTER 4: THE IMPORTANCE OF CARBON AND
SULFIDE AS CONTROLS ON NITRATE REMOVAL AND

NITRATE REDUCTION END-PRODUCTS .............................................. 115
INTRODUCTION ....................................................................... 115
METHODS ............................................................................... 118
RESULTS ................................................................................ 122
DISCUSSION ........................................................................... 158
CONCLUSIONS ........................................................................ 169
REFERENCES ......................................................................... 170

CHAPTER 5:

CONCLUSIONS ................................................................................. 173

viii

LIST OF TABLES

CHAPTER 2

Table 1: Physico-chemical characteristics of the sites in this study.
Abbreviations: DO= dissolved oxygen; Cond. = conductivity; SW= surface
water; PW= porewater; BDL= below detection limits. (*) denotes SItes that were
returned tom 2005 for the 15N03 push- pull experiments ............................... 49

Table 2. Stepwise multiple regression model (a: 0.01) results to predict the best
indicators of N03 removal rate constants and SO42 production: N03 removal
ratios (SP: NR) ...................................................................................... 60

CHAPTER 3

Table 1: Physical and chemical characteristics of the sites at the time the cores
were collected. Abbreviations: BDL = below detection limits, SW = surface
water........ .......................................................................................................... 83

Table 2: N03 processing rates and percent of overall removal of each process
by sites. Values are means of all measurements (n=9) with :1 SE. DEN =
denitriﬁcation, Anammox = anaerobic ammonium oxidation, DNRA =
dissimilatory nitrate removal to ammonium ................................................ 95

Table 3: A comparison of the relative importance of denitriﬁcation to DNRA in
studies where both have been simultaneously measured. Abbreviations: (FW)
freshwater, (ABT) acetylene block technique, (Den) denitrif' cation (DNRA)
dissimilatory nitrate reduction to ammonium. Units from the different studies are
as follows: ( )1 pmoles m 21'1'hr ,(*) umol L N“ ,(*) nmol cm 3"hr ,(#) ng N g
sediment hr ,( ) ng N g dry wt hr'. Journal abbreviations: (MEPS) Marine
Ecology Progress Series, (L&O) Limnology and Oceanography, and (FEMS-ME)
FEMS Microbiology Ecology ........................................................... 106-107

Table 4: A comparison of the relative importance of denitriﬁcation to anammox
in studies where both have been simultaneously measured. Abbreviations: (FW)
freshwater, (lPT) isotope pairing technique, (Den) denitriﬁcatio1n. Units from the
different studies are as follows: ( ) pmoles m 2‘hr' ,(‘t )umol L hr'1' ,( ) nmol N2
ml wet sed hr’1;(# ) nM hr Journal abbreviations: (AEM) Applied and

Environmental Microbiology, (EM) Environmental Microbiology, and Geochimica
et Cosmochimica (GCA) ................................................................ 108-109

CHAPTER 4

Table 1. Analysis of variance for individual response variables (NO3' and H28
removal; NHH, N2, N20, N02, 8042' ﬂux) by factor (H28 and DC) from LP (high
ambient H2S). w2 is the ﬁt of a factor to the ANOVA model, also called
magnitude of effect or effect size ...................................................... 135-136

Table 2. Analysis of variance for individual response variables (H28 removal;
NHf, N2, N20, 8042‘ ﬂux) by factor (H28 and DC) from WP (low ambient H28).
002 is the ﬁt of a factor to the ANOVA model, also called the magnitude of effect.
N02“ was not produced during WP experiments, so it is not a response variable
in these analyses. Also, because NO3’ was gone in all treatments at the end of
the experiment, NO3‘ removal is not considered as a response variable.
Abbreviations: $8 = sum of squares, DF = degrees of freedom, MSE = mean
square error, VC = variance component. .. ....................................................... 137

Table 3. Independent contrasts within the analysis of variance (ANOVA) to
compare treatments across time in LP. The estimated ﬂux (Estimate) is in
pmoles, and indicates the difference between the two treatments. A positive ﬂux
denotes that the effect of the ﬁrst factor listed (e.g, H28 in H2S vs. NO3‘ only) on
that ﬂux was greater than the second factor. NS = not signiﬁcant ........... 154-155

Table 4. Independent contrasts within the analysis of variance (ANOVA) to
compare treatments across time in LP. The estimated ﬂux (Estimate) is in
pmoles, and indicates the difference between the two treatments. A positive ﬂux
denotes that the effect of the ﬁrst factor listed (e.g, H28 in H2S vs. N03“ only) on
that ﬂux was greater than the second factor. NS = not signiﬁcant ........... 156-157

LIST OF FIGURES
CHAPTER 1

Panel 1: An introduction to heterotrophic energy production. Heterotrophic
respiration of organic matter can be either aerobic (using oxygen) or anaerobic
(no oxygen). Both forms of respiration are oxidation-reduction reactions in which
simple carbon sources are combined with electron (e') acceptors to yield oxidized
carbon (CO2), reduced products (H2O in the case of aerobic respiration), and
energy. The process of respiratory denitrification we describe in this review is a
form of anaerobic respiration in which nitrate serves as the alternate e' acceptor.
Various substances can act as electron acceptors in anaerobic respiration and
depending on the electron acceptor and its ultimate product, variable amounts of
energy are produced. Some common electron acceptors are listed in the order
from highest to lowest efﬁciency of energy yield; those microbes performing the
more efﬁcient reactions tend to outcompete others for labile organic

matter ................................................................................................... 5

Figure 1: A conceptual diagram of the nitrate removal pathways discussed in this
article. This is not meant to represent an exhaustive list of microbial
transformations, but rather to illustrate the different possible pathways and fates
of nitrate removal. Blue arrows denote autotrophic pathways, while purple
arrows denote heterotrophic pathways ....................................................... 8

Figure 2: DNRA estimates across a variety of aquatic ecosystems. The bars
represent the ranges of DNRA as a percent of the total dissimilatory nitrate
removal found in a given study site, with the balance presumably due to
denitriﬁcation. Purple bars designate marine and brackish ecosystems, blue bars
designate freshwaters. The North River site is hatched because it was
alternatively freshwater dominated and oligohaline. Many of these studies were
originally compiled by Megonigal et al. (2004) ............................................ 17

Figure 3: Anammox estimates across a variety of aquatic ecosystems. The bars
represent the ranges of total N2 production that can be attributed to Anammox in
a given study site. Purple bars designate marine and brackish ecosystems, blue
bars designate freshwaters. The Thames River Estuary is hatched because the
study spanned a range of freshwater and marine-inﬂuenced sites .................. 18

Figure 4: Hypothesized controls on predominant dissimilatory pathways of
nitrate removal. This ﬂow chart summarizes the conditions under which we
would expect a particular nitrate removal pathway to be important. C inputs refer
to labile organic carbon available to microbes. Sulﬁdic refers to the presence of
signiﬁcant amounts of either free sulﬁde (H28 or 82'), elemental 8 (8°), or metal-
bound sulﬁdes such as FeS, all of which tend to be abundant in sediment
environments with moderate to high sulfate in overlying water and high labile C

xi

inputs to support microbial sulfate reduction. Of these 8 forms, only free sulﬁde
inhibit denitrification and thus promotes DNRA. C:N ratios refer to the ratio of
labile organic carbon to nitrate. Denitrif. = denitrification; DNRA = Dissimilatory
Nitrate Reduction to Ammonium; Anammox = Anaerobic Ammonium
Oxidation.................................. .......................................................... 22

CHAPTER 2

Figure 1: Simultaneous nitrate removal and sulfate production In two different
wetlands. The panels on the left are the ratios of the reactants (NO3 and 8042 )
to the conservative tracer (Br") In two push- pull experiments. The panels on the
right2 are a comparison of observed and expected concentration of N03 and
3042' in a site with high (top) and low (bottom) sulfate production. The top
panels are data from Prairieville Creek Fen and the bottom panels are from
Windmill Pond (see Table 1 for more site information). These two examples
represent the highest and lowest rates of SO42 production relative to N03
removal that I observed” .................................................................................... 38

Figure 2: The effect of temperature on nitrate removal and sulfate production
(both in pmoles). Open symbols denote the nylon mesh injectors, closed
symbols denote the stainless steel injectors that were more commonly used and
employed in all of the ﬁeld experiments......................................................45

Figure 3: Simultaneous nitrate removal and sulfate production in surface water
over sediments after nitrate addition to the surface water (conducted In the lab In
a bucket). The top panel shows the ratios of the reactants (N03 and 8042 )to the
conservative tracer (Br.) In the top panel N03 :Br became negative because the
nitrate concentration fell below what was originally present (~0. 6 mg/L) before the
experimental N03 / Br addition. The bottom panel IS a comparison of observed
and expected concentrations of N03 and 8042 .......................................... 46

Figure 4: Nitrate removal rate constants by ecosystem. “n” refers to the number
of individual experiments (typically 3 per site), not to the number of sites of that
particular ecosystem. Boxes encompass the upper and lower quartiles, while the
line indicates the median for the dataset. Asterisks are mild outliers and open
circles are extreme outliers ..................................................................... 51

Figure 5: Ratios of total 8042 ' production. total NO3' removal (SP: NR) by
ecosystem. “n ” refers to the number of individual injectors (typically 3 per site).

A value of 1 indicates that all of the nitrate removal can be explained by the
sulfate production. Values greater than one indicate that sulfate production cana
account for more nitrate removal than was measured ................................... 52

xii

Figure 6: The fraction of measured nitrate removal that can be explained by
sulfate production using equations 1-3 (grey boxes) and equation 4-6 (black
boxes). “n” refers to the number of individual injectors (typically 3 per site), not to
the number of sites of that particular ecosystem ................................................ 54

Figure 7: Pre and post treatment NH4+ concentrations (top) and H28
concentrations (bottom) for all injectors at all sites in both 2004 and 2005.
Closed symbols denote injectors that received NO3' additions and open symbols
denote control injectors that only received a Br' addition. Error bars are standard
errors of the mean.. ........................................................................................... 55

Figure 8: Percentages of NO3' removal attributed to DNRA as measured using
15NO3'additions in combination with push-pull techniques. “n” refers to the
number of individual injectors (typically 3 per site) ...................................... 58

Figure 9: The relationship between measured 15NH4+ production (x-axis) and the
change in NH4" concentration in the ”N03 injectors (same injectors, top panel)
and the “N03 injectors (different injectors within the same site, bottom

panel) ................................................................................................ 59

CHAPTER 3

Figure 1: Average sediment oxygen demand (SOD) in the control cores from
each site over the three day incubation time (n=6). Error bars represent 1 SE. of
the mean. The ecosystem types are colored as: streams (black), wetlands (dark
grey), and lakes (light grey). Sites with the same letter are not statistically
different from each other (p=0.05) ............................................................ 85

Figure 2: 15NO3‘ ﬂux rates from each site over the three day incubation time
(n=9). Error bars represent 1 SE. of the mean. The ecosystem types are
colored as: streams (black), wetlands (dark grey), and lakes (light grey). Sites
with the same letter are not statistically different from each other (p=0.05) ......... 86

Figure 3: Denitriﬁcation rates measured as average 15N-N2 ﬂux (29N2+ 30N2) from
the ‘5NO3' amended cores from each site over the three day incubation time
(n=9). Error bars represent 1 SE. of the mean. The ecosystem types are
colored as: streams (black), wetlands (dark grey), and lakes (light grey). Sites
with the same letter are not statistically different from each other (p=0.05) ....... 87

Figure 4: Anammox as control corrected average 29N2 ﬂux from the 15NH4+

amended cores from each site over the three day incubation time (n=9). Error
bars represent1 SE. of the mean. The ecosystem types are Colored as:

xiii

streams (black), wetlands (dark grey), and lakes (light grey). Sites with the same
letter are not statistically different from each other (p=0.05) .......................... 90

Figure 5: Average NH4+ flux from the control cores (n=6; solid) and .15NO3’
amended cores (n=9; hatched) from each site over the three day incubation time.
Error bars represent 1 SE. of the mean. The ecosystem types are colored as:
streams (black), wetlands (dark grey), and lakes (light grey). (*) indicates that
the control and treatment cores had statistically different (p<0.05) NH4+ ﬂuxes..91

Figure 6: DNRA rates measured as 15NH4+ production in the 15NO3' amended
cores from each site over the three day incubation time (n=9). Error bars
represent 1 SE. of the mean. The ecosystem types are colored as: streams
(black), wetlands (dark grey), and lakes (light grey) .................................... 92

Figure 7: The relative importance of dissimilatory nitrate removal pathways to
overall nitrate removal in sediments from streams (black), wetlands (dark grey)
and lakes (light grey) ............................................................................. 93

Figure 8: The relative importance of dissimilatory nitrate removal pathways as a
percentage of the overall nitrate removal rate from each of the six sites. The
hatched area represented the amount of “unknown” NO3' removal, i.e., the N03
was removed but cannot be accounted for by the N end-products measured....94

Figure 9: Sulfate ﬂux in the ”N03 amended cores from each site and sampling
time over the three day incubation time (n=3). Error bars represent 1 SE. of the
mean. The ecosystem types are colored as: streams (black), wetlands (dark
grey), and lakes (light grey) .................................................................... 96

CHAPTER 4

Figure 1: Nitrate removal across a gradient of added sulﬁde for Loosestrife Fen
(top; high ambient H28) and Windmill Pond (bottom; low ambient H28). In these
assays, all treatments began with 14.3 umoles of added nitrate and were
sampled after 24 hours; data shown here are expressed as the total quantity of
nitrate that was removed over the 24 hours .............................................. 125

Figure 2: NH4+ ﬂuxes across the H28 and OC gradients in sediments from
Loosestrife Fen and Windmill Pond. Positive ﬂuxes indicate net production.
Note the differences in the y-axis ranges .................................................. 126

Figure 3: N2 production for Loosestrife Fen (top) and Windmill Pond (bottom)

across a gradient of OC and H28. Note the differences in the y-axis
ranges ............................................................................................. 127

xiv

Figure 4: N20 production (pmoles) for Loosestrife Fen (top)... and Windmill Pond
(bottom) across a gradient of OC and H28... ...128

Figure 5: N02 ﬂux for Loosestrife Fen across a gradient of OC and H28 ...... 129

Figure 6: H28 ﬂux for Loosestrife Fen (top) and Windmill Pond (bottom) across a
gradient of OC and H28. Negative ﬂuxes indicate net removal ..................... 133

Figure 7: 8042' ﬂux for Loosestrife Fen (top) and Windmill Pond (bottom) across
a gradient of OC and H28 ..................................................................... 134

Figure 8: Fractions of each N end-product from the gradient assays. These are
means across the H28 only gradient (i. e., they do not include treatments with
added OC). Total N ﬂux Is based on the N03 removal that was observed. Also
shown for comparison are the 8042 and H28 ﬂuxes. Quantities added: LP low
H28 (0. 5 pmoles), LP high H28 (5. 5 pmoles), WP low H28 (1.5 pmoles), and WP
high H28 (15.5 pmoles) ......... . ...................................... . ........................ 138

Figure 9: Fractions of each N end-product from the gradient assays. These are
means across OC treatments (i.e., they do not include H28 in the averages).
Total N ﬂux is based on the N03 removal that was observed. Also shown for
comparison are the 8042‘ and H28 ﬂuxes. Quantities added: LP low 00 (2.3
umoles), LP high OC (46.7 pmoles), WP low OC (2.3 pmoles), and WP high OC
(46.7 pmoles). Positive ﬂuxes indicate net production ................................ 139

Figure 10: Fractions of each N end-product from the gradient assays. These
are means across all treatments (i. e., all levels of OC in a given H28 treatment
were included). Total N ﬂux' Is based on the N03 removal that was observed.
Also shown for comparison are the 8042 and H28 ﬂuxes. Quantities added: LP
low H28 (0. 5 pmoles), LP high H28 (5.5 umoles), WP low H28 (1.5 pmoles), and
WP high H28 (15.5 umoles) .................................................................. 140

Figure 11: Nitrate removal over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 umoles of NO3‘ were added to each treatment at the start of
the experiment ................................................................................... 145

Figure 12: Ammonium ﬂux over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 umoles of NO3' was added to each treatment at the start of the
experiment ........................................................................................ 146

Figure 13: N2 production over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 pmoles of NO3' was added to each treatment at the start of the
experiment. Controls received no NO3', and all vials were maintained underwater
to ensure that atmospheric N2 and 02 did not leak into the vials .................... 148

XV

Figure 14: N20 production over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 pmoles of NO3‘ was added to each treatment at the start of the
experiment. Controls received no NO3', and all vials were maintained underwater
to ensure atmospheric N2 and 02 did not leak into the vials .......................... 149

Figure 15: N02 production over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 pmoles of NO3' was added to each treatment at the start of the
experiment ........................................................................................ 150

Figure 16: 8042' ﬂux over time in both Loosestrife Fen (LP) and Windmill Pond
(WP). Controls received no NO3‘, H28 or DC, but were maintained to evaluate
how 8042 changed in the absence of these factors over time ...................... 152

Figure 17: H28 ﬂux over time in both Loosestrife Fen (LP) and Windmill Pond

(WP). Controls received no NO3', H28 or DC, but were maintained to evaluate
how H28 changed in the absence of these factors over time ........................ 153

xvi

CHAPTER 1

As published in Frontiers in Ecology and the Environment:

Burgin, A.J. and SK. Hamilton. 2007. Have we overemphasized the role of
denitriﬁcation in aquatic ecosystems? A review of nitrate removal pathways.
Frontiers in Ecology and the Environment 5 (2): 88-96.

ABSTRACT:

The removal of nitrogen (N) in aquatic ecosystems is of particular interest
because excessive nitrate in ground water and surface water is a growing
problem. Enhanced loading of nitrate degrades water quality and is linked to
eutrophication and harmful algal blooms, especially in coastal marine waters.
Research on nitrate removal processes has emphasized plant or microbial
uptake (assimilation) or respiratory denitriﬁcation by bacteria. The increasing
application of stable isotopes and other tracer techniques to study nitrate removal
has yielded a growing body of evidence for alternative microbially mediated
processes 'of nitrate transformation, including dissimilatory reduction of nitrate to
ammonium (DNRA), chemoautotrophic denitriﬁcation via sulfur or iron oxidation,
and anaerobic ammonium oxidation (Anammox), as well as abiotic N removal
processes. Here we review evidence for the importance of alternative nitrate
removal pathways in aquatic ecosystems and discuss how the possible
prevalence of these pathways may alter views of N cycling and its controls.
These alternative pathways are particularly signiﬁcant for the management of
excess N in the environment in cases where they transform nitrate to ammonium,

a biologically available and less mobile N form, rather than to dinitrogen gas.

In a nutshell:

0 Increasing nitrogen loading causes eutrophication of aquatic ecosystems

and degrades water quality for human use.

. Most of the nitrogen added to landscapes is removed during transit to the
ocean, and this removal has been attributed largely to denitriﬁcation, with
a lesser proportion to assimilation and accumulation in ground water.

0 New research has pointed to the importance of alternative microbial
pathways of nitrate removal.

. The possible prevalence of these pathways has critical implications for

managing excess N in aquatic ecosystems.

Nitrogen in aguatic ecosystems

Excessive nitrogen (N) concentrations, often largely in the form of nitrate
(NO3'), present a water-quality problem of growing concern. Nitrate
concentrations in ground water and in rivers in developed areas of the world
have risen dramatically following the growing use of synthetic N fertilizers and
cultivation of N-ﬁxing crops (Turner and Rabalais 2003). Increasing N export
from landscapes to coastal waters has been implicated in coastal eutrophication,
creating hypoxic zones (such as in the Gulf of Mexico; (Rabalais et al. 2001) and
harmful algal blooms (Paerl et al. 2002). There is still some debate over whether
or not N alone is always the main driver of these problems (Dodds 2006), but
there is no question that the increases in N loading represent a major
perturbation of streams, rivers, estuaries and coastal marine waters.

Although N loading to coastal zones has increased, regional watershed
mass-balance studies indicate that most of the anthropogenic N that enters

watersheds is removed before reaching the oceans (Howarth et al. 1996,

Alexander et al. 2000). Surface and groundwater ﬂow through landscapes often
enters riparian wetlands and headwater streams, which can efficiently remove
nitrogen (Peterson et al. 2001, Zedler 2003). Thus, key interfaces along
landscape ﬂow paths control nitrate export to downstream surface waters, such
as large rivers and lakes, and ultimately to estuaries and marine ecosystems. In
this article, we discuss the multiple possible fates for this removed nitrate, which
include some grossly underestimated and understudied microbial pathways,
many of which have recently been gaining attention among the scientiﬁc
community. The importance and even possible prevalence of these pathways
have profound implications for the management of aquatic ecosystems to

promote nitrate removal.

 

 

Box 1: An introduction to heterotrophic energy production. Heterotrophic
respiration of organic matter can be either aerobic (using oxygen) or anaerobic
(no oxygen). Both forms of respiration are oxidation-reduction reactions in which
simple carbon sources are combined with electron (e') acceptors to yield oxidize
carbon (CO2), reduced products (H2O in the case of aerobic respiration), and
energy. The process of respiratory denitriﬁcation we describe in this review is a
form of anaerobic respiration in which nitrate serves as the alternate e' acceptor.
Various substances can act as electron acceptors in anaerobic respiration and
depending on the electron acceptor and its ultimate product, variable amounts of
energy are produced. Some common electron acceptors are listed in the order
from highest to lowest efﬁciency of energy yield; those microbes performing the
more efﬁcient reactions tend to outcompete others for labile organic matter.

Aerobic Respiration Anaerobic Respiration
Organic C C02 Organic C C02

   

  
 
 

+ ENERGY + ENERGY

‘.f-f '~ 9' acceptor Reduced
02 ”20 (other than 02) product

Electron acceptors: 02 > N03 > Fe3+ > 8042'

 

 

 

 

 

 

Where does the nitrate go?

Up to 75% of the N added to a landscape can be removed before reaching
marine ecosystems (Howarth et al. 1996). The various transformations and
eventual fate of this N as it is carried along hydrologic ﬂow paths is a problem
that has interested scientiﬁc and management communities alike. The current
consensus is that the disappearance of N is due largely to biological
transformations, since increased storage (e.g., in groundwater or biomass
accrual) cannot explain most of the “missing N” (Howarth et al. 1996). Biological
removal of nitrate from water passing through or over sediments is often
assumed to be due to either assimilation into algal or microbial biomass,
producing organic N that may be remineralized later, or respiratory denitrification
by bacteria, producing gaseous N2.

In respiratory denitriﬁcation, nitrate acts as the terminal electron acceptor
for the oxidation of organic matter under anaerobic conditions; in aquatic
sediments most the nitrate is usually converted to N2 with a variable but small
fraction escaping as nitrous oxide (N20) (Figure 1). Because N2 is unavailable to
most organisms, respiratory denitriﬁcation is considered a permanent removal of
N from the ecosystem. Denitriﬁcation rates have been estimated in soils,
wetlands, and surface waters, but estimates vary greatly within and among
environments, as well as between different measurement techniques.
Nevertheless, denitriﬁcation is thought to remove substantial fractions of the total
nitrate loads to lakes, rivers, and coastal estuaries (Seitzinger 1988, Comwell et

al. 1999). However, while nitrate disappearance in soils and aquatic sediments

is usually assumed to be largely due to denitriﬁcation, estimates of denitriﬁcation
based on direct assays (e.g., acetylene block techniques) often account for less
than half of the total nitrate disappearance (e.g., see tables in Seitzinger 1988).
This discrepancy between local denitriﬁcation estimates and the large losses of
nitrate at the landscape scale remains difﬁcult to reconcile. One possible
explanation is that we have not yet designed adequate methods to extrapolate
from site-speciﬁc rates to entire ecosystems (Comwell et al. 1999). An
alternative explanation is that much of the nitrate removal can be attributed to
processes other than respiratory denitriﬁcation or assimilation. New research
has pointed to the importance of processes that remove nitrate in freshwater
ecosystems, including dissimilatory nitrate reduction to ammonium (Tiedje 1988),
anaerobic ammonium oxidation (Jetten et al. 1998, Jetten 2001 ), denitriﬁcation
coupled to sulﬁde oxidation (Dannenberg et al. 1992, Fossing et al. 1995, Brunet
and Garcia-Gil 1996, Otte et al. 1999), and reduction of nitrate coupled to abiotic
or biotically mediated oxidation of iron (Davidson et al. 2003, Weber et al. 2006).
Here we review mounting scientiﬁc evidence for the importance of these
alternative nitrate removal pathways, and we propose that nitrate removal in

aquatic ecosystems may entail much more than denitrification and assimilation.

Respiratory iN2N20
Denitriﬁcation ’

    
 

 

:NH.+: 44
Fermentative 4 Blomass j
DNRA Assimilation

Sulfur-driven

Iron-driven ------------- . .
‘ nItrate reduction

denitriﬁcation j NOZ'I

 

 

 

     

 

 

+
8042 N2 or NH4
.-4 --éETmOX 4---.
i No ‘2 hw - 5.2-!

Figure 1: A conceptual diagram of the nitrate removal pathways discussed in
this article. This is not meant to represent an exhaustive list of microbial
transformations, but rather to illustrate the different possible pathways and fates
of nitrate removal. Blue arrows denote autotrophic pathways, while purple
arrows denote heterotrophic pathways.

Alternatives to resmtory drenitrriﬁcation

Respiratory denitriﬁcation is surely an important nitrate removal pathway,
but we will not discuss it in further detail due to the numerous reviews on the
process (Knowles 1982, Tiedje et al. 1982, Seitzinger 1988, Comwell et al.
1999). Our focus is not meant to lead the reader to the conclusion that these
alternative pathways are generally more important than denitriﬁcation, but to
point out that there are several processes that could rival denitriﬁcation in
signiﬁcance but have been much less studied until now. While there is some
evidence for each of these pathways, much more research is needed, particularly
in freshwater ecosystems, to ascertain their importance relative to respiratory

denitriﬁcation in whole-ecosystem nitrate removal.

Dissimilatory nitrate reduction to ammonium (DNRA)

The existence of DNRA has been widely recognized for at least the past
25 years, though its potential signiﬁcance as a nitrate removal pathway on an
ecosystem scale has generated increased interest within the past decade. This
microbially mediated pathway involves the dissimilatory transformation of nitrate
to ammonium (NH4+), in contrast to assimilatory processes that incorporate the N
into cellular constituents. Compared to nitrate, the resultant ammonium is a
more biologically available, and less mobile, form of inorganic N (Figure 1). Little
is known about the eventual fate of the nitrate that is converted to ammonium via

DNRA, but it is possible that under appropriate conditions, the ammonium is

converted back to nitrate via nitriﬁcation. The resultant ammonium may also be
assimilated into plant or microbial biomass.

There are two recognized DNRA pathways, one involving fermentation
and the other linked to sulfur oxidation. Early work on DNRA suggested that it
was mainly carried out by fermentative bacteria (Tiedje 1988), though in recent
years the existence of DNRA coupled to sulfur cycling has been documented in
marine and freshwater ecosystems (Brettar and Rheinheimer 1991, Brunet and
Garcia-Gil 1996). It is unknown if the two DNRA pathways are mutually
exclusive.

Fermentative DNRA couples electron ﬂow from organic matter via
fermentation reactions to the reduction of nitrate (Tiedje 1988, Megonigal et al.
2004). Many microbes perform fermentative DNRA, including species of
Clostridia, Desulfovibn'o, Vibrio, and Pseudomonas; these organisms can also
i carry out fermentation without using nitrate (Tiedje 1988). Although the
conditions promoting fermentative DNRA and respiratory denitriﬁcation are
similar (anoxia, available nitrate and organic substrates), fermentative DNRA is
thought to be favored in nitrate-limited, labile-carbon rich environments while
respiratory denitriﬁcation would be favored under carbon-limited conditions
(Kelso et al. 1997, Silver et al. 2001). Tiedje (1988) argued that high labile
carbon availability would favor organisms that used electron acceptors most
efﬁciently; DNRA transfers eight electrons per mole of nitrate reduced, whereas
denitriﬁcation only transfers ﬁve. Some studies have supported Tiedje’s

hypothesis that DNRA is more important in high carbon, low nitrate systems,

10

including Bonin (1996) and Nijburg et al. (1997). The oxidation state of the
sediments may also be important. For example, Matheson et al. (2002)
hypothesized that microzones of oxygen leakage from roots of emergent plants
in wetland sediments may favor the facultatively aerobic denitriﬁers over the
obligately anaerobic fermentative bacteria. Much more work is needed to
understand where and when DNRA is prevalent in ecosystems before we can
fully understand what factors govern its importance relative to other nitrate
removal processes.

A very different form of DNRA is chemolithoautotrophic and couples the
reduction of nitrate to the oxidation of reduced sulfur forms, including free sulﬁde
(H28 and 82') and elemental sulfur (S) (Brunet and Garcia-Gil 1996, Otte et al.
1999). The nitrate may be reduced either to ammonium, making it a form of
DNRA, or to N2 in a form of denitriﬁcation, although not all species can do both
(Zopﬁ et al. 2001). In this pathway, the predominant fate of the reduced nitrate
may be determined by the ambient concentration of free sulﬁde, which is known
to inhibit the ﬁnal two reduction steps in the denitriﬁcation sequence. Sulﬁde
inhibition of these terminal steps may drive the reduction to ammonium rather
than to nitrous oxide and N2. Brunet and Garcia-Gil (1996) studied the effects of
various sulfur forms as potential electron donors, and found that only free sulﬁde
yielded ammonium and nitrous oxide, lending support to the idea that the
enzymes that support respiratory denitriﬁcation may be inhibited by the presence
of sulfide. 0n the other hand, metal-bound sulﬁdes (e.g., FeS), which are often

abundant constituents of freshwater sediments (Holmer and Storkholm 2001),

11

also can be oxidized by these bacteria, but these compounds may not inhibit
denitriﬁcation (Brunet and Garcia-Gil 1996). A similar process that couples the
reduction of nitrate to the oxidation of methane was recently discovered in
freshwaters (Raghoebarsing et al. 2006), though it is not yet clear if this process
is important to whole-ecosystem nitrate removal.

The ability of bacteria to couple the reduction of nitrate to the oxidation of
sulfur has now been established in a number of taxa with diverse metabolic
characteristics (Dannenberg et al. 1992, Bonin 1996, Philippot and Hojberg
1999) including members of the genera Thiobacillus, Thiomicrospora, and
Thioploca (Timmertenhoor 1981, Jorgensen 1982, Kelly and Wood 2000).
Bacteria with this capability include the “big bacteria” (e.g., Thioploca) that are
able to store nitrate, sulfur, or calcite in vacuoles (Schulz and Jorgensen 2001).
This storage capability, in conjunction with their gliding motility, allows them to
take advantage of steep biogeochemical gradients, for example by taking up
nitrate from overlying oxic water and utilizing it to oxidize sulfur in sulﬁde-rich
anoxic porewaters (Schulz and Jorgensen 2001).

The biogeochemical importance of nitrate use by sulfur-oxidizing bacteria
was ﬁrst widely recognized in marine sediments, but we are beginning to
discover its importance in freshwater ecosystems. For example, mush of the
nitrate uptake in a groundwater aquifer was ascribed to Thiobacillus denitriﬁcans
(Bottcher et al. 1990), and Thioploca occurs not only in marine sediments, but
also in freshwater ecosystems including lakes Erie, Baikal, and Biwa (Megonigal

et al. 2004). Furthermore, species of Beggiatoa, a genus of sulfur oxidizers

12

common in freshwaters, also appear to be capable of using nitrate to oxidize

sulfur (Kamp et al. 2006).

Nitrate reduction coupled to iron oxidation

The reduction of nitrate coupled to iron (Fe) cycling is thought to take
place through both biotic and abiotic pathways (Weber et al. 2006, Davidson et
al. 2003). In Figure 1, we depict one example of an abiotic pathway in which
nitrate is converted to nitrite (N02') by ferrous iron (Fe2+; this could also be done
by reduced manganese, Mn2+), followed by rapid reaction of the nitrite to N2.
Postma et al. (1991) concluded that this reaction would only remove a signiﬁcant
proportion of nitrate from groundwater in areas with low nitrate inputs. Another
abiotic reaction has been proposed in which nitrate is reduced to nitrite by
reaction with Fe or Mn and the nitrite binds with organic substances to produce
DON (Davidson et al. 2003); evidence for this reaction was discovered recently in

forest soils (Dail et al. 2001), but it is not known to occur in aquatic ecosystems.

Alternatively, microbes can mediate nitrate reduction coupled to iron
oxidation in aquatic ecosystems (Weber et al. 2006). This biotic reduction occurs
at relatively low temperatures and circumneutral pH (Weber et al. 2001), and
thus it may be more likely to occur in surface waters than the equivalent abiotic
reaction. Microbes that can perform this process have been isolated from a
diverse array of aquatic sediments (Straub and Buchholz-Cleven 1998). The
majority of the work in this area has focused on describing the microbes capable

of the reaction, and we could not ﬁnd an estimate of the potential importance of

13

the reaction as an ecosystem-level process compared to other N removal
processes. The controls on the process remain poorly understood, though it may
be important in areas of high reduced iron and a limited supply of organic C

(Weber et al. 2001).

Anaerobic Ammonium Oxidation

Anaerobic ammonium oxidation (known as Anammox) is a
chemolithoautotrophic process by which ammonium is combined with nitrite
under anaerobic conditions, producing N2. The nitrite is derived from the
reduction of nitrate, possibly by denitrifying bacteria, and therefore Anammox
contributes to permanent nitrate removal. The process was discovered in a
wastewater treatment system in the 1990’s, and since its discovery, studies have
shown it occurs in anoxic wastewater, oxygen-depleted zones of the ocean,
temperate shelf sediments, sea ice, and cold Arctic shelf sediments (Jetten et al.
1998, Rysgaard and Glud 2004, Rysgaard et al. 2004), and recently it has been
reported in one freshwater ecosystem - Lake Tanganyika (Schubert et al. 2006).

Scientists still know relatively little about the bacteria that carry out
Anammox, and no pure cultures exist (Strous et al. 2006). This may be because
Anammox is performed by slow-growing organisms (doubling time is
approximately 11 days; Jetten et al. 1999), an idea further upheld by evidence
that the process has a low thermal optimum (12°C compared to 24°C for
denitriﬁcation; Jetten 2001). Those Anammox bacteria that have been identiﬁed

belong to the Planctomycetes, a group that has evolved internal

14

compartmentalization (similar to eukaryotes) and a specialized structure called
an anammoxosome, which may protect the cell from toxic Anammox
intermediates such as hydrazine (Jetten et al. 2003, Strous et al. 2006).
Anammox occurs in anoxic waters where there are suitable concentrations
of both nitrate and ammonium, and the process is inhibited by many simple
organic compounds including pyruvate, ethanol, and glucose (Jetten et al. 1999).
Thus, Anammox may be most important in ecosystems with limited labile carbon,
or that have an excess of nitrogen relative to carbon inputs. This may include
signiﬁcant parts of the open-ocean and continental shelves (Dalsgaard et al.
2005). A recent synthesis of Anammox studies suggests that in marine
ecosystems, water depth is important in regulating the relative importance of
Anammox to total nitrate removal, with Anammox producing up to 2/3 of the N2 in
areas over 20 m deep. Although Anammox seems to be less important to overall
nitrate removal in shallower marine and estuarine waters (<1 m), many of these
areas have higher absolute rates of Anammox (Dalsgaard et al. 2005). While
little is known about Anammox in freshwaters, based on what is known about the
process in marine ecosystems, one might expect that it would be more important
in very deep, large oligotrophic lakes. The only study to date on freshwaters was
in Lake Tanganyika where Shubert and others (2006) found that 7-13% of the N2

production came from Anammox.

15

How important are these pathways in aguatic ecosystem N cvclingz

This is a particularly difﬁcult question to answer at the present time
because many of the pathways we described are just beginning to be studied in
detail. In this section, we provide evidence for the importance of alternative
pathways in marine and freshwater ecosystems (Figures 2 and 3). We also
describe the conditions under which we might expect a particular pathway to be
important. Figure 4 is a ﬂow chart based on what we know about controls of
each pathway; its purpose is to synthesize the work we have summarized to this
point, and to generate testable hypotheses about when and where certain nitrate
removal processes are likely to be important.

The relative availability of labile carbon, reduced sulfur, and reduced iron
are proposed to be key determinants of nitrate removal pathways. Anammox
and respiratory denitriﬁcation have been shown to be important nitrate removal
pathways in areas of relatively low labile carbon; at this time, sulfur, and
particularly free sulfide, is not known to affect Anammox. However, because of
its effect on key enzymes in the denitriﬁcation sequence, we believe that free
sulﬁde may be a key variable in determining nitrate removal processes in
relatively high carbon environments, which includes many freshwater and near-
coastal ecosystems. When there is sulﬁde in close proximity to oxic waters, as
for example in surﬁcial sediments of many shallow waters, we hypothesize that
nitrate removal coupled to microbially mediated sulfur oxidation may be

important; in anoxic settings with relatively low sulﬁde, we expect that respiratory

16

100

 

 

 

 

90
80 -
70 ‘ P
a 60 7 a
g 50 . I
o\° 40 '7
30 . L
20 - In
10 l I l D __ D
0 - . - - .
6‘ 0‘ 0‘ 6‘ 6‘ v *0 ,o‘ 0‘ «0‘ A 6* b
0&‘90 396‘?) 8&0 3969 0409 @069 6‘29 ‘QQSX 6‘69 00‘? Q02? 9,06 {£555
09 9 9° to (9‘ {33° {‘99 06 ‘90 a” Q§P \02’ 6°
\(x 06 6* Q6 ®® 0C}. $6}. 6 .40 06 4Q~ fb\
«872 029 02°) 029 .oo‘ ‘6 ‘2 V3629 by '52 6°22 QBQ
343 (4.. Q at: 33‘ 330‘ .335 309 3° 0 <5»
0&2 0‘22 3° 0 060 ‘2 4“ 4‘20" 922’ V020 6:20
00 00 0 @‘b 9&1!
\s

Figure 2: DNRA estimates across a variety of aquatic ecosystems. The bars
represent the ranges of DNRA as a percent of the total dissimilatory nitrate
removal found in a given study site, with the balance presumably due to
denitriﬁcation. Closed bars designate marine and brackish ecosystems, open
bars designate freshwaters. The North River site is hatched because it was
alternatively freshwater dominated and oligohaline. Many of these studies were
originally compiled by Megonigal et al. (2004).

17

%ofN2from Anammox
ANw-ho'lmﬂ
OOOOOOO

-
a
‘9 -
_ o
e
G. _
—
—
\\‘

0
Q52? 02“ (2% 0&9 2° 2’6 $2
0 Yso (‘2’ <3 202’ «90 Q2 o2
\0 . \\ 9 e o g ~t~
00 06 @Q 0% *9 &Q) \rfb
«(o (\2’ .30 0 .34. ‘0 .06
\ Q) C) 0‘ \ &\ I\
C90 c. g 62‘ ’b
0‘ V‘ o <2 2’
obi 032 022‘ 0°20
03% gs 0
e“ 0

Figure 3: Anammox estimates across a variety of aquatic ecosystems. The bars
represent the ranges of total N2 production that can be attributed to Anammox in
a given study site. Closed bars designate marine and brackish ecosystems,
open bars designate freshwaters. The Thames River Estuary is hatched
because the study spanned a range of freshwater and marine-inﬂuenced sites.

18

denitriﬁcation and perhaps fermentative DNRA could be more important (Figure
4).

DNRA has been measured in a few studies of whole-system nitrate
removal (Bonin 1996, Rysgaard et al. 1996, Silver et al. 2001, Tobias et al. 2001,
Welsh et al. 2001, An and Gardner 2002), although none of these studies
determined if the apparent DNRA was chemolithoautotrophic or fermentative.
Figure 2 summarizes data from the literature to show that DNRA is potentially as
important as respiratory denitriﬁcation in diverse environments. Most work on
this pathway has been done in marine ecosystems, including marine and
estuarine sediments, brackish marsh sediments, and mangroves, where DNRA
can account for a very wide range (0-100%) of the total nitrate removal (Figure 2,
purple). Evidence for DNRA has been found in freshwaters as well, including
river sediments, rice paddies, riparian wetlands and aquifers (Figure 2, blue).
DNRA may be relatively more important in marine than freshwater ecosystems,
but this is a tenuous conclusion because of the small number of studies of DNRA
in freshwaters (Figure 2). Evidence for DNRA has also been found in certain
soils, where it can account for a large fraction (up to 75%) of total nitrate removal
(Silver et al. 2001). The observation that DNRA can be important in soils, which
are not thoroughly anoxic like aquatic sediments, highlights how little we
understand about the process and suggests that DNRA may occur in many other
environments that have yet to be investigated.

Research on Anammox in marine ecosystems was synthesized by

Dalsgaard et al. (2005), who concluded that Anammox contributes half or more

19

of the N2 production in coastal shelves and the deep sea (Figure 3), and possibly
is responsible for 1/3 to 2/3 of global oceanic nitrate removal. The role of
Anammox in freshwater nitrogen cycling remains speculative since only one
study in a natural freshwater ecosystem has been published (Schubert et al.
2006). Anammox would be expected to occur where nitrate and ammonium
coexist, which could perhaps include interfaces between surface water and
sediment pore water. However, due to inhibition by simple organic carbon
compounds, Anammox may be limited to areas that are relatively low in labile
carbon, which may not often be the case for near-surface freshwater sediments
that support high biological productivity (Figure 4).

How is it that scientists may have overlooked these pathways for so many
years? We believe this is due in large part to methodological limitations. The
importance of these pathways has recently been appreciated through the use of
stable isotope and molecular microbial methods. Prior to the widespread use of
stable isotopes, the favored method to measure denitriﬁcation was the acetylene
block technique (ABT) (Tiedje 1988). The ABT typically entails creation of a
sediment slurry, de-oxygenation with an inert gas, addition of acetylene to block
the transformation of nitrous oxide, N20, to N2, and measurement of the rate of
N20 production over time to indicate the rate of denitriﬁcation. This method
suffers from a number of problems for trying to detect alternative nitrate removal
processes, including the removal of free sulﬁde by sparging, disruption of the
steep sediment redox gradients that may favor certain organisms and reactions,

and the incorrect assumption that all of the nitrous oxide produced is from

20

denitrification (Welsh et al. 2001, Senga et al. 2006). The widespread use of the
ABT, as well as other less sensitive techniques, may have led to an
overestimation of the importance of denitriﬁcation, and an underestimation of
other nitrate removal pathways.

Conclusions and implications for management

Immense amounts of effort have been expended to study respiratory
denitriﬁcation and management decisions are being made based on that body of
knowledge. The possible importance -- or even prevalence — of alternative
nitrate removal pathways has profound implications for our management of
aquatic ecosystems to reduce nitrate loads. Nitrate is the most mobile N form, so
removal of nitrate by any of the processes described above is important to
downstream water quality, but permanent removal by denitriﬁcation is most
desirable.

Removal by other pathways can result in transformation of the nitrate to
something other than dinitrogen gas (N2). Nitrate removal via Anammox still
creates dinitrogen gas as an end-product, but removes both a nitrate and an
ammonium ion in the process. In contrast, the conversion of nitrate to
ammonium, as in DNRA, creates an even more bioavailable N form, and one that
tends to be less mobile in soils and sediments. This converted ammonium can
also be transformed back to nitrate via nitriﬁcation. Additionally, if S-oxidizers
prove to take up much of the nitrate, then N cycling is closely linked to sulﬁde

availability, which is turn is linked to Sulfate reduction. In freshwaters sulfate

21

 

,8 oxidizers:

 

    
 
 
    
 

 

   

+
S oxidizers H28 DNRA ’0 NH4
dominate
Sulfidic
F93. 30 S oxidizers:
High C >Denitrif.
inputs to N2
Low C:N RGSpir.
Denitrif.
to N2
Not sulfidic
Ferment.
High C:N DNRA .
Respir.

Denitrif. to

0w C:N N022, the"
/ Anammox
Low Fe
I'ow C / Respir.
mpms Denitrif.
High C:N to N2
\_>Fe oxidation:

High Fe Denitrif. to N2

Figure 4: Hypothesized controls on predominant dissimilatory pathways of
nitrate removal. This ﬂow chart summarizes the conditions under which we
would expect a particular nitrate removal pathway to be important. C inputs refer
to labile organic carbon available to microbes. Sulﬁdic refers to the presence of
signiﬁcant amounts of either free sulﬁde (H28 or 82'), elemental 8 (8°), or metal-
bound sulﬁdes such as FeS, all of which tend to be abundant in sediment
environments with moderate to high sulfate in overlying water and high labile C
inputs to support microbial sulfate reduction. Of these 8 forms, only free sulfide
inhibit denitriﬁcation and thus promotes DNRA. C:N ratios refer to the ratio of
labile organic carbon to nitrate. Denitrif. = denitriﬁcation; DNRA = Dissimilatory
Nitrate Reduction to Ammonium; Anammox = Anaerobic Ammonium Oxidation.

22

reduction may be controlled by sulfate inputs, and sulfate is a ubiquitous pollutant
in industrialized and agricultural regions (Schlesinger 1997). If excess sulfate
loading to freshwaters actually enhances nitrate removal, then the controls on
nitrate removal in landscapes subject to S and N pollution become more complex
than previously thought.

Ecologists and managers should accept that nitrate disappearance is no
longer synonymous with denitriﬁcation, and that there are many other pathways
that potentially remove nitrate. Much more research needs to be done on these
alternative nitrate removal pathways across a diversity of aquatic ecosystems.
Most of what we know about them is based on research done in marine
ecosystems, and thus our understanding of what controls these processes in

freshwater ecosystems subject to elevated nitrate inputs remains incomplete.

Acknowledgements:

We would like to thank the following people for their help and advice: Jay
Lennon, Terrance Loecke, Patrick Mulholland, Nathaniel Ostrom, G. Philip
Robertson, Thomas Schmidt, Stefanie Whitmire and Nancy Grimm’s N reading
group at ASU. The comments of two anonymous reviewers also improved this
manuscript. This work was supported by US. National Science Foundation
grants DEB-0111410, 0508704, 0423627, and 0516076. This is contribution

#1275 to the WK. Kellogg Biological Station.

23

References:

Alexander, R., R. Smith, and G. Schwarz. 2000. Effect of stream channel size on
the delivery of nitrogen to the Gulf of Mexico. Nature 403:758-761.

An, S. M., and W. S. Gardner. 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). Marine Ecology-Progress
Series 237:41-50.

Bonin, P. 1996. Anaerobic nitrate reduction to ammonium in two strains isolated

from costal marine sediment: A dissimilatory pathway. FEMS Microbiology
Ecology 19:27-38. *

Bottcher, J., 0. Strebel, S. Voerkelius, and H. L. Schmidt. 1990. Using Isotope
Fractionation of Nitrate Nitrogen and Nitrate Oxygen for Evaluation of

Microbial Denitriﬁcation in a Sandy Aquifer. Journal of Hydrology 114:413—
424.

Brettar, l., and G. Rheinheimer. 1991. Denitriﬁcation in the Central Baltic -
evidence for H28-oxidation as motor of denitriﬁcation at the oxic-anoxic
interface. Marine Ecology-Progress Series 77:157-169.

Brunet, R. C., and L. J. Garcia-Gil. 1996. Sulﬁde—induced dissimilatory nitrate
reduction to ammonia in anaerobic freshwater sediments. FEMS
Microbiol09y Ecology 21 :1 31-138.

Comwell, J. C., W. M. Kemp, and T. M. Kana. 1999. Denitriﬁcation in coastal
ecosystems: methods, environmental controls, and ecosystem level
controls, a review. Aquatic Ecology 33:41-54.

Dail, D. B., E. A. Davidson, and J. Chorover. 2001. Rapid abiotic transformation
of nitrate in an acid forest soil. Biogeochemistry 54:131-146.

Dalsgaard, T., B. Thamdrup, and D. E. Canﬁeld. 2005. Anaerobic ammonium

oxidation (anammox) in the marine environment. Research in Microbiology
156:457-464.

Dannenberg, 8., M. Kroder, W. Dilling, and H. Cypionka. 1992. Oxidation of H2,
organic compounds and inorganic sulfur compounds coupled to reduction

of 02 or nitrate by sulfate-reducing bacteria. Archives of Microbiology
1 58:93-99.

24

Davidson, E. A., J. Chorover. and D. B. Dail. 2003. A mechanism of abiotic
immobilization of nitrate in forest ecosystems: the ferrous wheel
hypothesis. Global Change Biology 92228-236.

Dodds, W. K. 2006. Nutrients and the "dead zone": the link between nutrient
ratios and dissolved oxygen in the northern Gulf of Mexico. Frontiers in
Ecology and the Environment 4:211-217.

Fossing, H., V. A. Gallardo, B. B. Jorgensen, M. Huttel, L. P. Nielsen, H. Schulz,
D. E. Canfield, S. Forster, R. N. Glud, J. K. Gundersen, J. Kuver, N. B.
Ramsing, A. Teske, B. Thamdrup, and O. Ulloa. 1995. Concentration and
transport of nitrate by the mat-forming sulfur bacterium Thioploca. Nature
374:713-715.

Holmer, M., and P. Storkholm. 2001. Sulphate reduction and sulphur cycling in
lake sediments: a review. Freshwater Biology 46:431-451.

Howarth, R. W., G. Billen, D. Swaney, A. Townsend, N. Jaworski, K. Lajtha, J. A.
Downing, R. Elmgren, N. Caraco, T. Jordan, F. Berendse, J. Freney, V.
Kudeyarov, P. Murdoch, and Z. L. Zhu. 1996. Regional nitrogen budgets
and riverine N&P ﬂuxes for the drainages to the North Atlantic Ocean:
Natural and human inﬂuences. Biogeochemistry 35:75-139.

Jetten, M. S. M. 2001. New pathways for ammonia conversion in soil and aquatic
systems. Plant and Soil 230:9-19.

Jetten, M. 8. M., M. Strous, K. T. van de Pas-Schoonen, J. Schalk, U. G. J. M.
van Dongen, A. A. van de Graaf, 8. Logemann, G. Muyzer, M. C. M. van
Loosdrecht, and J. G. Kuenen. 1998. The anaerobic oxidation of
ammonium. FEMS Microbiology Reviews 22:421-437.

Jorgensen, B. 1982. Ecology of the bacteria of the sulphur cycle with special
reference to anoxic-oxic interface environments. Philosophical Transitions
of the Royal Society of London 298:543-561.

Kamp, A., P. Stief, and H.N. Schulz-Vogt. 2006. Anaerobic sulﬁde oxidation with
nitrate by a freshwater Beggiatoa enrichment culture. Applied and
Environmental Microbiology 72:4755—4760. 2

Kelly, D. P., and A. P. Wood. 2000. Conﬁrmation of Thiobacillus denitriﬁcans as a
species of the genus Thiobacillus, in the beta-subclass of the
Proteobacteria, with strain NCIMB 9548 as the type strain. lntemational
Journal of Systematic and Evolutionary Microbiology 50:547-550.

Kelso, B. H. L., R. V. Smith, R. J. Laughlin, and S. D. Lennox. 1997. Dissimilatory

nitrate reduction in anaerobic sediments leading to river nitrite
accumulation. Applied and Environmental Microbiology 63:4679-4685.

25

Knowles, R. 1982. Denitriﬁcation. Microbiological Reviews 46:43-70.

Matheson, F. E., M. L. Nguyen, A. B. Cooper, T. P. Burt, and D. C. Bull. 2002.
Fate of 15N-nitrate in unplanted, planted and harvested riparian wetland
soil microcosms. Ecological Engineering 19:249-264.

Nijburg, J. W., M. J. L. Coolen, S. Gerards, P. J. A. K. Gunnewiek, and H. J.
Laanbroek. 1997. Effects of nitrate availability and the presence of
Glyceria maxima the composition and activity of the dissimilatory nitrate-

reducing bacterial community. Applied and Environmental Microbiology
63:931-937.

Otte, 8., J. G. Kuenen, L. P. Nielsen, H. W. Paerl, J. Zopﬁ, H. N. Schulz, A.
Teske, B. Strotmann, V. A. Gallardo, and B. B. Jorgensen. 1999. Nitrogen,
carbon, and sulfur metabolism in natural Thioploca samples. Applied and
Environmental Microbiology 65:3148-3157.

Paerl, H. W., R. L. Dennis, and D. R. Whitall. 2002. Atmospheric deposition of
nitrogen: Implications for nutrient over-enrichment of coastal waters.
Estuaries 25:677-693.

Peterson, B. J., W. M. Wollheim, P. J. Mulholland, J. R. Webster, J. L. Meyer, J.
L. Tank, E. Marti, W. B. Bowden, H. M. Valett, A. E. Hershey, W. H.
McDowell, W. K. Dodds, S. K. Hamilton, 8. Gregory, and D. D. Morrall.
2001. Control of nitrogen export from watersheds by headwater streams.
Science 292286-90.

Philippot, L., and O. Hojberg. 1999. Dissimilatory nitrate reductases in bacteria.
Biochimica et Biophysica Acta-Gene Structure and Expression 1446:1-23.

Postma, D., C. Boesen, H. Kristiansen, and F. Larsen. 1991. Nitrate reduction in
an unconﬁned sandy aquifer - water chemistry, reduction processes, and
geochemical modeling. Water Resources Research 27:2027-2045.

Rabalais, N. N., R. E. Turner, and W. J. Wiseman. 2001. Hypoxia in the Gulf of
Mexico. Journal of Environmental Quality 30:320-329.

Raghoebarsing, A. A., A. Pol, K. T. van de Pas-Schoonen, A. J. P. Smolders, K.
F. Ettwig, W. l. C. Rijpstra, S. Schouten, J. 8. S. Damste, H. J. M. 0p den
Camp, M. S. M. Jetten, and M. Strous. 2006. A microbial consortium
couples anaerobic methane oxidation to denitriﬁcation. Nature 4402918-
921.

Rysgaard, 8., and R. N. Glud. 2004. Anaerobic N2 production in Arctic sea ice.
Limnology and Oceanography 49:86-94.

26

Rysgaard, 8., R. N. Glud, N. Risgaard-Petersen, and T. Dalsgaard. 2004.
Denitriﬁcation and anammox activity in Arctic marine sediments.
Limnology and Oceanography 49:1493-1502.

Rysgaard, 8., N. RisgaardPetersen, and N. P. Sloth. 1996. Nitriﬁcation,
denitriﬁcation, and nitrate ammoniﬁcation in sediments of two coastal
lagoons in Southern France. Hydrobiologia 329:133-141.

Schlesinger, W. H. 1997. Biogeochemistry: an analysis of global change, 2nd
edition. Academic Press, San Diego.

Schubert, C. J., E. Durisch-Kaiser, B. Wehril, B. Thamdrup, P. Lam, M.M.M.
Kuypers. 2006. Anaerobic ammonium oxidation in a tropical freshwater
system (Lake Tanganyika). Environmental Microbiology 8:1857-1863.

Schulz, H. N., and B. B. Jorgensen. 2001. Big bacteria. Annual Review of
Microbiology 55:105-137.

Seitzinger, S. 1988. Denitriﬁcation in freshwater and coastal marine ecosystems:
ecological and geochemical significance. Limnology and Oceanography
33:702-724.

Senga, Y., K. Mochida, R. Fukumori, N. Okamoto, and Y. Seike. 2006. N20
accumulation in estuarine and coastal sediments: The inﬂuence of H28 on

dissimilatory nitrate reduction. Estuarine Coastal and Shelf Science
67:231-238.

Silver, W. L., D. J. Herman, and M. K. Firestone. 2001. Dissimilatory nitrate
reduction to ammonium in upland tropical forest soils. Ecology 8222410-
2416.

Straub, K. L., and B. E. E. Buchholz-Cleven. 1998. Enumeration and detection of
anaerobic ferrous iron-oxidizing, nitrate-reducing bacteria from diverse
European sediments. Applied and Environmental Microbiology 64:4846-
4856.

Tiedje, J. M. 1988. Ecology of denitriﬁcation and dissimilatory nitrate reduction to
ammonium. Pages 179-244 in A. J. B. Zehnder, editor. Biology of
Anaerobic Microorganisms. Wiley & Sons, New York.

Tiedje, J. M., A. J. Sexstone, D. D. Myrold, and J. A. Robinson. 1982.

Denitriﬁcation - ecological niches, competition and survival. Antonie Van
Leeuwenhoek Journal of Microbiology 48:569-583.

27

Timmer—ten-hoor, A. 1981. Cell Yield and Bioenergetics of Thiomicrospira-
Denitriﬁcans Compared with Thiobacillus-Denitriﬁcans. Antonie Van
Leeuwenhoek Journal of Microbiology 47:231-243.

Tobias, C. R., S. A. Macko, l. C. Anderson, E. A. Canuel, and J. W. Harvey.
2001. Tracking the fate of a high concentration groundwater nitrate plume
through a fringing marsh: A combined groundwater tracer and in situ
isotope enrichment study. Limnology and Oceanography 46:1977-1989.

Turner, R. E., and N. N. Rabalais. 2003. Linking landscape and water quality in
the Mississippi river basin for 200 years. Bioscience 53:563-572.

Weber, K. A., F. W. Picardal, and E. E. Roden. 2001. Microbially catalyzed
nitrate-dependent oxidation of biogenic solid-phase Fe(ll) compounds.
Environmental Science 8 Technology 35:1644-1650.

Weber, K. A., M. M. Urrutia, P. F. Churchill, R. K. Kukkadapu, and E. E. Roden.
2006. Anaerobic redox cycling of iron by freshwater sediment
microorganisms. Environmental Microbiology 8: 1 00-1 13.

Welsh, D. T., G. Castadelli, M. Bartoli, 0. Poll, M. Careri, R. de Wit, and P.
Viaroli. 2001. Denitriﬁcation in an intertidal seagrass meadow, a
comparison of 15N isotope and acetylene-block techniques: dissimilatory

nitrate reduction to ammonia as a source of N20? Marine Biology
139:1029-1036.

Zedler, J. B. 2003. Wetlands at your service: reducing impacts of agriculture at
the watershed scale. Frontiers in Ecology and the Environment 1:65-72.

Zopﬁ, J., T. Kjaer, L. P. Nielsen, and B. B. Jorgensen. 2001. Ecology of
Thioploca spp.: Nitrate and sulfur storage in relation to chemical
microgradients and inﬂuence of Thioploca spp. on the sedimentary
nitrogen cycle. Applied and Environmental Microbiology 67:5530-5537.

Due to space limitations, we would like to include this supplementary information
online:

References from Figure 2 (in order of pairs):
Goeyens, L., R. T. P. Devries, J. F. Bakker, and W. Helder. 1987. An experiment
on the relative importance of denitriﬁcation, nitrate reduction and

ammoniﬁcation in coastal marine sediment. Netherlands Journal of Sea
Research 21 :171-175.

Christensen, P. B., S. Rysgaard, N. P. Sloth, T. Dalsgaard, and S. Schwaerter.
2000. Sediment mineralization, nutrient ﬂuxes, denitriﬁcation and

28

dissimilatory nitrate reduction to ammonium in an estuarine ﬁord with sea
cage trout farms. Aquatic Microbial Ecology 21:73-84.

Koike, I., and A. Hattori. 1978. Denitrification and ammonia formation in
anaerobic coastal sediments. Applied and Environmental Microbiology
35:278-282.

Pelegri, S. P., L. P. Nielsen, and T. H. Blackburn. 1994. Denitriﬁcation in
estuarine sediment stimulated by the irrigation activity of the amphipod
Corophium volutator. Marine Ecology-Progress Series 105:285-290.

Rivera-Monroy, V. H., R. R. Twilley, R. G. Boustany, J. W. Day, F. Veraherrera,
and M. D. Ramirez. 1995. Direct denitriﬁcation in mangrove sediments in
Terrninos Lagoon, Mexico. Marine Ecology-Progress Series 126:97-109.

Tobias, 0. R., l. C. Anderson, E. A. Canuel, and S. A. Macko. 2001. Nitrogen
cycling through a fringing marsh-aquifer ecotone. Marine Ecology-
Progress Series 210225-39.

Tobias, C. R., S. A. Macko, l. C. Anderson, E. A. Canuel, and J. W. Harvey.
2001. Tracking the fate of a high concentration groundwater nitrate plume
through a fringing marsh: A combined groundwater tracer and in situ
isotope enrichment study. Limnology and Oceanography 46:1977-1989.

Bowden, W.B. 1986. Nitriﬁcation, nitrate reduction, and nitrogen immobilization
in a tidal freshwater marsh sediment. Ecology 67: 88-99.

Kelso, B. H. L., R. V. Smith, R. J. Laughlin, and S. D. Lennox. 1997. Dissimilatory
nitrate reduction in anaerobic sediments leading to river nitrite
accumulation. Applied and Environmental Microbiology 63:4679-4685.

Bengtsson, G., and H. Annadotter. 1989. Nitrate reduction in a groundwater
microcosm determined by 15N gas chromatography-mass spectrometry.
Applied and Environmental Microbiology 55:2861-2870.

Buresh, R. J., and W. H. Patrick. 1978. Nitrate reduction to ammonium in
anaerobic soil. Soil Science Society of America Journal 422913

Yin, S. X., D. Chen, L. M. Chen, and R. Edis. 2002. Dissimilatory nitrate
reduction to ammonium and responsible microorganisms in two Chinese
and Australian paddy soils. Soil Biology 8 Biochemistry 34:1131-1137.-
918. ~

Matheson, F. E., M. L. Nguyen, A. B. Cooper, T. P. Burt, and D. 0. Bull. 2002.

Fate of 15N-nitrate in unplanted, planted and harvested riparian wetland
soil microcosms. Ecological Engineering 19:249-264.

29

References from Figure 3 (in order of bars):

Dalsgaard, T., D. E. Canfield, J. Petersen, B. Thamdrup, and J. Acuna-Gonzalez.
2003. N2 production by the anammox reaction in the anoxic water column
of Golfo Dulce, Costa Rica. Nature 422:606-608.

Rysgaard, 8., and R. N. Glud. 2004. Anaerobic N2 production in Arctic sea ice.
Limnology and Oceanography 49:86-94.

Rysgaard, 8., R. N. Glud, N. Risgaard-Petersen, and T. Dalsgaard. 2004.
Denitriﬁcation and anammox activity in Arctic marine sediments.
Limnology and Oceanography 49:1493-1502.

Engstrom P, Dalsgaard T, Hulth S, and Aller, RC. 2005. Anaerobic ammonium
oxidation by nitrite (anammox): Implications for N2 production in coastal
marine sediments. Geochimica et Cosmochimica Acta 69 (8): 2057-
2065.

Thamdrup, B. and T. Dalsgaard. 2002. Production of N2 through anaerobic
ammonium oxidation coupled to nitrate reduction in marine sediments.
Applied and Environmental Microbiology 68: 1312-18.

Risgaard-Petersen N, Nielsen LP, Rysgaard S, Dalsgaard T, and Meyer RL.
2003. Application of the isotope pairing technique in sediments where
anammox and denitriﬁcation coexist. Limnology and Oceanography-
Methods 1: 63-73.

Trimmer, M., J. C. Nicholls, and B. Deﬂandre. 2003. Anaerobic ammonium
oxidation measured in sediments along the Thames estuary, United
Kingdom. Applied and Environmental Microbiology 69:6447-6454.

Schubert, C. J., E. Durisch-Kaiser, B. Wehril, B. Thamdrup, P. Lam, M.M.M.

Kuypers. 2006. Anaerobic ammonium oxidation in a tropical freshwater
system (Lake Tanganyika). Environmental Microbiology 8:1857-1863.

30

CHAPTER 2

THE PREVALENCE OF SIMULTANEOUS NITRATE REMOVAL AND SULFATE
PRODUCTION IN STREAM, WETLAND AND LAKE SEDIMENTS
Introduction:

Increases in the intensity and extent of agricultural land use have led to
dramatic alterations of the global N cycle, contributing to a doubling of the annual
input rate of bio-available, fixed N (Vitousek et al. 1997). N pollution originating
from agriculture and other anthropogenic activities has dramatically increased N
loading to aquatic ecosystems (Carpenter et al. 1998, Bernot and Dodds 2005).
Approximately 75% of the N loading toterrestrial landscapes cannot be
acc0unted for in river exports to the ocean and most of this evidently disappears
in transit (Seitzinger et al. 2006). These massive changes in N cycling have
stimulated interest in understanding how the dominant inorganic N form, nitrate
(NO3'), moves through landscapes, and what controls the fraction that is removed
in transit through aquatic ecosystems.

Nitrate removal has long been acknowledged to be driven by the
heterotrophic microbial metabolism of organic carbon, with N03“ serving either as
an electron donor for respiratory denitriﬁcation, or in a fermentation process
wherein nitrate is reduced to ammonium in dissimilatory nitrate reduction to
ammonium (DNRA) (Tiedje 1988). More recently, a few studies have reported
evidence for apparent linkages between sulfur (S) and N cycling in freshwater
ecosystems. With experimental additions of NO3' and sulfate (8042') to wetland

sediments in southwest Michigan, Whitmire and Hamilton (2005) found that in

31

approximately half of their 16 experiments, 8042' production occurred during the
period of NO3' removal. NO3' additions on the Wisconsin River ﬂoodplain also
resulted in an approximately equimolar amount of 8042' production (Forshay
2003). Evidence of this phenomenon has also been found in the Netherlands
where researchers observed that in an aquifer containing pyrite, 8042'
concentrations increased while inﬁltrating NO3' was removed (Lucassen et al.
2002). S-driven NO3‘ removal has also been exploited as a water treatment
method for removing NO3’ in cores packed with elemental sulfur granules
(Spares 2002).

The metabolic ﬂexibility of S-oxidizing bacteria has been increasingly
appreciated in recent years. Thiobacillus denitriﬁcans couples the oxidation of
inorganic sulfur compounds to the reduction of NO3' and is common in freshwater
and marine sediments (Kelly 1999, Haaijer et al. 2006). Thioploca spp. were ﬁrst
described from Lake Constance in the early 19005 and were later found to occur
in several other lakes in northern Germany (Jorgensen and Gallardo 1999).
However, their unique inﬂuence on biogeochemical cycling was appreciated only
after Thioploca was found in vast benthic mats off the coast of Chile (Gallardo
1977). Species of Thioploca have been intensively studied since the discovery of
the marine species, in large part because their distinctive metabolism couples
NO3' reduction to sulﬁde (H28) oxidation (Jorgensen and Gallardo 1999).
Additionally, they possess gliding motility that allows them to migrate upward to
oxic overlying water of higher NO3' concentration, and store both NO3' and

elemental S intracellularly (Jorgensen and Gallardo 1999). Thiomargarita, known

32

for its enormous size (750 um), also utilizes and stores NO3' and 8° intracellularly
(Schulz et al. 1999, Schulz and Jorgensen 2001). Beggiatoa spp., close relatives
of Thioploca and Thiomargarita, and are also gliding, ﬁlamentous bacteria that
can store 8° and oxidize it with NO3' (Kamp et al. 2006). These bacteria have
been known to occur in marine ecosystems for some time (Sweerts et al. 1990,
McHatton et al. 1996), and have recently been isolated from freshwater
ecosystems (Kamp et al. 2006).

This study was inspired by the observations of Whitmire and Hamilton
(2005), who examined anaerobic microbial processes including NO3' removal in a
variety of wetlands. Tracer injections were conducted using a modiﬁed “push-
pull” method to quantify rates of NO3' and 8042' uptake (see “Methods”). The
injection solution in that work was local groundwater with ambient concentrations
of N03' (14 mg ML) and 8042’ (53 mg/L) to simulate groundwater inputs to ﬁve .
different wetland sediments. In 9 of her 16 experiments, they observed marked
production of 8042' relative to the concentration of the conservative tracer (Br‘),
but only during the period when NO3' was being consumed. Figure 1 shows
examples of this phenomenon in a wetland in which the 8042' production
(triangles) was quite pronounced (top left panel) and a wetland in which the
response was less prominent (bottom left panel). In all cases, 8042' removal
(triangles) commenced soon after the available NO3‘ (circles) was removed,
presumably by 8042' reduction (Figure 1).

Nitrate removal can be stoichiometrically compared to 8042' production by

expressing the push-pull experiment data as the difference between observed

33

and expected concentrations (Figure 1, right panels). Expected concentrations
are calculated as the product of the observed Br‘ concentration and the ratio of
N03‘ or 8042' to Br' in the injection solution, after background correction
(Whitmire 2003, Whitmire and Hamilton 2005). Figure 1 illustrates two examples
from the dataset collected for this experiment. The example in the top set of
panels is from a site where high rates of sulfate production occurred. The
example from the bottom panels is from a site where relatively little 8042'
production occurred.

The variation observed in Whitmire and Hamilton’s work led me to question
if this pattern was indeed biological, and if so, how widespread it was across
aquatic ecosystems. As part of this study, I performed several tests to show that
the occurrence was enzymatically catalyzed (see Results). Once assured that
the process is biological, l was faced with discerning what reactions may be
driving the 8042' production and N03' removal, and in particular, I wanted to
know what happens to the nitrate that is reduced? Assuming the reaction
proceeds via denitriﬁcation to dinitrogen gas (N2), the initial oxidation step may

have the following stoichiometry (Fossing et al. 1995):
5H82+2NO32+7H+95S°+N2+6H20 (1)
The resultant elemental 8 may be stored in the cells before later being oxidized

to 8042'. Further oxidation to 8042' could occur by the following reaction

(Fossing et al. 1995):

58°+6NO3‘+2H2O-)58042'+3N2+4H+ (2)

If these two reactions occurred sequentially, the molar ratio of NO3' consumed to

8042' produced would be 8:5 (=1.6) as in this combined reaction:

5HS'+8NO3'+3H"->58042'+4N2+4H2O (3)

However, the nitrate may not be completely reduced to N2, but may also be
converted to NH4+ in a form of DNRA. The stoichiometric formula for the
conversion of N03' to NH4+, assuming direct transformation of sulﬁde to
elemental S (as in equation 1), is (Sayama et al. 2005):

4H2S+NO3'+2H*—) 4S°+NH4i+3H2O (4)

The resultant 8 may be stored, or further oxidized to 8042' via:

4 8° + 3 NO3'+ 7 H2O —> 3 NH4++4 $0..” + 2 H” (5)
Reactions 4 and 5 may occur sequentially producing ratio of one mole of nitrate
consumed to every one mole of sulfate produced, as in:

4 H28 + 4 N03‘ + 4 H20 —> 4 NH4+ + 4 3042' (6)
The stoichiometry of these two reactions can be used to estimate the magnitude
of the sulfate production that could be to nitrate removal.

The alternative reduction products (N2 vs. NH4+) have different
implications for ecosystem N cycling. Nitrate reduced to N2 via denitriﬁcation is
permanently removed, whereas NO3' transformed to NH4+ via DNRA is retained
within the ecosystem. Few studies have examined the role of DNRA in
freshwater systems, and the eventual fate of NH4+ produced by DNRA is
uncertain (Burgin and Hamilton 2007). Freshwater wetlands are low in 8
compared to marine systems, but often contain enough to support signiﬁcant

bacterial transformations (Lovley and Klug 1983).

35

The objectives of this study were to: 1) confirm that the 8042' production
observed by Whitmire and Hamilton (2005) was a biological phenomenon; 2)
ascertain how widespread the phenomenon is across diverse freshwater
ecosystems; 3) explore which potential pathways and reactions could be
responsible; and, 4) seek a predictive understanding of what types of aquatic

ecosystems would exhibit signiﬁcant coupled NO3'-8 cycling.

Methods:
Site selection

Experiments were conducted in nine streams, nine wetlands and three
lakes (Table 1). For the 2004 study, sites were selected to encompass a range
of free H28 concentrations In sediment porewaters (Whitmire 2003; Table 1). In
2005, sites were selected in part based on data from the larger 2004 survey, and
in part based on sites from the LINX ll experiments (streams only; Mulholland et

al., in review).

Push-pull methodology
The push-pull method has been used to estimate biogeochemical
processing-rates in aquifers (Istok et al. 1997, McGuire et al. 2002), lake
sediments (Luthy et al. 2000), and riparian wetlands (Addy et al. 2002). In my
"work, push-pull experiments were done using a near-surface push-pull technique
similar to the method developed by Whitmire and Hamilton (2005), modiﬁed
slightly to minimize dead volume in the injector system. The experiments were

conducted in situ at 5-10-cm depth in the sediments, at three points per

36

treatment (described below) in each site. Sites are described in Whitmire (2003),
Whitmire and Hamilton (2005), (Wetzel 2001) and Chapters 3 and 4 of this
dissertation.

Push-pull experiments entail removing a sample of porewater and amending
it with one or more reactive solutes (e.g., NO3') as well as a conservative solute
tracer (Br‘), reinjecting the porewater into the sediments, and withdrawing
samples over time to quantify the rates of uptake. The ratio of the reactant to the
conservative tracer concentration corrects for dilution and dispersion; use of the
conservative tracer together with the biologically active solute allows for
calculation of the net production or consumption of the reactant (see Whitmire
and Hamilton 2005 for more details on the calculations). The amendment
solution is 02-free and minimal in volume. Generally, 27 mL of porewater were
removed from the sediments, amended with 3 mL of anoxic Br' (control) or NO3' +
Br‘ (treatment) solution (ﬁnal concentrations were 10 mg L'1 NO3‘ -N and 3 mg L'1
Br'), and reinjected into the sediments. The injection line was then ﬂushed with 1
mL of porewater and 1 mL of sample was withdrawn and ﬁltered (0.45 pm Millex
syringe ﬁlters) for analysis by ion chromotrography (N03, 8042', Br'). In lake and
wetland sediments, samples were taken every 10 minutes for the ﬁrst hour and
then every 30 minutes for the next 2-3 hours. lncubations generally lasted 4
hours. Due to the increased hydrologic movement in streams, samples were

taken every ﬁve minutes for the ﬁrst 20 minutes and every 10 minutes thereafter.

37

NOB-N :Br (circles)

1.4
1.2
1.0

NOe-N :Br (circles)

0.2

0.0 '

 

0.8 §
0.6 .
0.4 4

SO42":Br (triangles)

0 Q9 ’09 ’89

TIme (min)

8042':Br (triangles)

 

0 6° <19 '39

Time (min)

Observed-Expected (uM)

Observed-Expected(pM)

150

100

50'

-100

-150 4,
.200 ~
-250 -

 

TIme (min)

i + N03 l
l {—23— 804%
i

l

 

0 30 60 90120150180240

TIme (min)

Figure 1: Simultaneous nitrate removal and sulfate production In two different
wetlands. The panels on the left are the ratios of the reactants (N03 and 8042 )
to the conservative tracer (Br) in two push- -pull experiments. The panels on the
right are 2a comparison of observed and expected concentration (eq. 8) of N03
and 8042 in a site with high (top) and low (bottom) sulfate production. The top
panels are data from Prairieville Creek Fen and the bottom panels are from
Windmill Pond (see Table 1 for more site information). These two examples
represent the highest and lowest rates of 8042 production relative to N03
removal that I observed.

38

In 2004, I performed push-pull experiments in a diverse set of wetlands
(n=9), streams (n=9) and lakes (n=3), with site details in Table 1. In 2005, I
returned to each of three wetlands, lakes and streams and repeated the push-
pull experiments in the same locations using both “N03 for the push-pull rate
measurements and ‘5NO3' to discern the dominant end products (sites marked
with asterisks in Table 1). Due to analytical constraints, we could not remove
small sub-samples of the ‘5NO3' addition; therefore, those injectors were sampled
only once at the end of the incubation period. The treatment addition was thus
undisturbed for the length of the experiment, and the whole volume (40 mL) was
removed at the end of the experiment and the 15N2 extracted using the static
headspace equilibration (Hamilton and Ostrom 2007) to quantify denitriﬁcation
and the remaining water was ﬁltered for 15NH4+ analysis to quantify DNRA. The
615N-NH4“ was measured using the ammonia diffusion procedure (Holmes et al.
1998), in which M90 is added to elevate the pH and the NH3 diffuses into an
acidiﬁed glass-ﬁber filter sealed within a Teﬂon packet suspended above the
surface of the sample. These samples on ﬁlters were analyzed for 5‘5N either in
the Stable Isotope Biogeochemistry Laboratory operated by Nathaniel and Peggy
Ostrom at MSU or at the Marine Biological Laboratory’s facility in Woods Hole,
MA. These 15N tracer data were compared to the N03' removal rates calculated
from the ”N03 addition at the same site.

NO3' removal rate constants were calculated based on N03' and Br‘

concentrations, and were modeled as ﬁrst order rate reactions using the following

39

exponential function, which relates the concentration of the reactant (N03) to the
tracer (Br‘) at a given point in time:

Creaciantﬂ) = Ciracerﬂ) * 9"“ (7)
The N03 removal rate constant (k) is the slope of a regression line fit to a plot of
(In)(Creactant(t)/Ctracer(t)) versus time (C is concentration). Linear regressions for

the calculation of nitrate removal were always signiﬁcant (i.e., p < 0.05).

Stoichiometric calculations

In addition to calculating NO3' removal rate constants from the push-pull
experiments, I also calculated 8042' production. Brieﬂy, 8042' production was
calculated as the difference between the measured 8042' concentration
(observed) and what would be expected based on the concentration of the

conservative tracer (Br'):

[804214.44 = [Brlobgd * (Isof'I/[Brnnleciale (8)
where [804219.434 is the expected concentration if 'no loss or gain of 8042' had
taken place, [Br]obs'd is the Br concentration observed at a given sampling time,
and ([8042'1/[Br‘j)33344434e is the ratio of 8042‘ to Br‘ in the injectate that was added
at the beginning of the experiment. This expected 8042' concentration was
subtracted from the measured concentrations, and the difference calculated as
8042‘ production. The same calculations can also be done for N03' removal,

substituting N03' for 8042' in equation 5. All concentrations are molar.

The 8042' production and N03' removal can then be related to the

stoichiometry of various possible reactions (equations 14) to infer which

40

processes may be occurring in these experiments. The molar ratio of total 8042'
production to total N03‘ removal over the duration of the experiment is hereafter
referred to as SP:NR. The “overall ratio” (mean of the ratios taken at each time
point) over the duration of the N03' removal period may be a better indicator of
reaction stoichiometry because of the potential temporary storage of NO3' by 8

oxidizing bacteria.

Testing for experimental artifacts

In studying this phenomenon, I ﬁrst thought that perhaps the apparent
sulfate production could be an experimental artifact of the push-pull
methodology. I therefore performed three experimental artifact tests: 1) a
temperature test to verify enzymatic catalysis and thereby rule out the possibility
of a strictly abiotic chemical reaction; 2) a test using an alternative injector
material to ensure that the phenomenon was not caused by a reaction with the
stainless steel injectors; and 3) additions of tracers to water overlying sediments
to see if the phenomenon occurred in the absence of the injectors and push-pull
experiments.

For the temperature test, I collected soil cores from a nearby wetland
(Turkey Marsh, a site that was not used in the in situ push-pull experiments
described above) and placed the cores into 1 quart canning jars along with ~100
mL overlying water, allowing one day in the dark for stabilization. To each jar, I
added a regular push-pull injector made with a stainless steel mesh tip.
Triplicate jars were incubated at 6, 22, or 50°C. For the 22°C treatment, I had

three additional jars with injectors made from a Nylon mesh material, to test the

41

second possible artifact described above. Push-pull experiments were
conducted in these jars, using the same methods described for the ﬁeld
experiments. Additionally, for the third experiment, I collected sediment from the
same site to half-fill a 20-L bucket, leaving approximately 2.5 cm of overlying
water, to which I added NO3' and Br'. This overlying water was periodically
sampled over a longer time period than the push-pull experiments because I
expected reactions to occur more slowly, limited by sediment-water diffusive
exchanges. Water samples for all experiments were analyzed as described

below.

Porewater chemistry analysis

In both control (Br) and treatment (NO3' + Br) injections, porewaters were
sampled for NH4+ and H28 concentrations prior to application of the treatment
(“pre”) and at the end of the experiment (“post”). H28 was analyzed by the
methylene-blue method (Golterrnan and Clymo 1969). NH4+ was measured
colorimetrically using the phenylhypochlorite technique (Aminot et al. 1997). NO-
3', Br’, and 8042’ were measured using membrane-suppression ion
chromatography (Dionex 4200 with an AS14A anion column). At each site a
sample of surface water was collected (ﬁltered to 0.45 pm with a Gelman
membrane ﬁlter) for comprehensive hydrochemical analysis (major solutes,

nutrients).

42

Statistical analysis

All statistical analyses were conducted using Systat version 9.0 software.
I used a non-parametric KruskaI-Wallis analysis of variance to test the
differences in the distributions of NO3' removal rate constants and SP:NR ratios
across ecosystem types. A parametric analysis of variance (ANOVA) was used
to compare differences in porewater chemistry between treatment and control
injectors. Stepwise multiple regression (MR; 0 = 0.10) was used to determine
which environmental variables (Table 1) best explained the variation in N03' rate
constants and SP:NR ratios. Variables were square root transformed when
necessary to improve normality. MR was only performed on the 2004 data to
avoid including the same sites from two years; MR was not performed on the
2005 sites alone because there were only 9 sites total, which was not enough

power for the analysis.

Results:

Experimental Artifact Tests
After observing 8042' production in ﬁeld studies, I attempted to rule out other

possible explanations (i.e., to ensure that the production was not an experimental

artifact). To rule out a chemical reaction, I performed experiments in microcosms

(jars) at various temperatures. Both 8042' production and N03' removal revealed

an intermediate thermal optimum, indicative of biological mediation via enzymatic

activity (below). I found similar results using injectors made of either stainless-

43

steel screens (closed symbols) or nylon filters (open symbols), indicating that an
interaction with the materials was not causing the phenomenon.

To ensure that the tracer ions did not somehow affect sediment ion
exchange equilibria (desorbing 8042') I injected only Br' at various
concentrations, and found no 8042' production (the NaBr comprised most of the
total ions added). My experiments were conducted in sediments free of plant
roots, ruling out NO3‘ uptake by vascular plants. Finally, to ensure that the
injection was not creating an artiﬁcial juxtaposition of reduced and oxidized
substances, I added N03' to the water overlying sediments in a wetland ﬁeld
enclosure and in a bucket of organic sediment in the lab. Water-column NO3'

additions yielded the same results, albeit over a longer time scale (Figure 3).

To summarize the large amount of variation within one ecosystem type, I
have placed all of the ﬁeld sites into an aquatic ecosystem category (stream, 8;
lake, L; wetland, W) and have created box plots (e.g., Fig. 4) that enc0mpass all
of the individual injectors frOm each site (i.e., the box plots are based on the
entire set of experiments, rather than site averages (experiments were conducted
at three points within each site). This gives the most complete illustration of the

variation within and among ecosystems.

44

8042' production (umoles)

N03' removal (umoles)

(130 -

(125 4

(120 -

CI15 -

(110 -

(105 4

C100 -

-OJ

 

 

412 -

413 -

-0.4 ~

-05 _

-06 -

417 -

 

 

-na-

10

I l l

20 30 40

Temperature (C)

Figure 2: The effect of temperature on nitrate removal and sulfate production
(both in pmoles). Open symbols denote the nylon mesh injectors, closed
symbols denote the statinless steel injectors that were more commonly used and
employed in all of the ﬁeld experiments.

45

O N03:Br
A A A SO4:Br
1 \ A 3
A
on A A
8 2 "e
0 A
0 o
-05 AAA . ’ ' ' 1
A
A
-1 » ' 0
0 20 40 60 80 100 120 140
75
A
A A
I A A A A
A 25 A
2 I AAAsz
3 £ .
u
9 l
8 -25l
ﬁt
LIJ .30
I ~
8 '75 o
2
8
8 . ‘ O
f 0 O
.s A 804! ‘0 O .
-175
0 20 40 60 80 100 120 140
Time (hours)

Figure 3: Simultaneous nitrate removal and sulfate production in surface water
over sediments after nitrate addition to the surface water (conducted in the lab in
a bucket). The top panel shows the ratios of the reactants (NO3‘ and 8042') to the
conservative tracer (Br’). In the top panel NO3‘:Br' became negative because the
nitrate concentration fell below what was originally present (~06 mg/L) before the
experimental NO3‘I Br‘ addition. The bottom panel is a comparison of observed
and expected concentrations of NO3' and 8042'.

46

The N03 removal rate constant is the fraction of the N03‘ concentration
that was removed per unit time (min‘1). For example, a rate of -0.01 min‘1 means
that 1% of the nitrate that is present (concentration dependent) is removed every
minute. NO3' removal rate constants were highly variable both among
ecosystem types and within a given ecosystem (Figure 4). Lakes tended to have
higher removal rates than did streams and wetlands (KruskaI-Wallis ANOVA
5.639, df = 2; p = 0.06). Stream NO3‘ removal rate constants ranged from -0.006
to 00545 min", wetlands ranged from -00031 to -0.0752 min", and lakes

ranged from -0.062 to -0.0842 min“.

To compare the relative amount of 8042' production in relation to NO3'
removal across sites and aquatic ecosystems, I used a ratio of the umoles of
8042’ produced (calculated from the observed 8042' concentration and the
expected concentration based on the Br‘ concentration; ﬁgures 1 and 3) to the
umoles of NO3' removed (also as in Figures 1 and 3). This generates a unitless
ratio of sulfate productionznitrate removal (abbreviated “SP:NR”); numbers near 1
reﬂect a situation wherein nearly all of the N03' removal can be accounted for by

the 8042' production.

Concurrent 8042' production and N03' removal occurred in all freshwater
ecosystem types (Figure 5). Streams had the greatest range of ratios of total
8042' production to total NO3' removal (SP:NR 002-24). In 6 stream
experiments (injectors, in 2 sites), 8042' production could account for much more
of the N03' removed than actually occurred; this also occurred in 4 wetland

injectors seen in wetlands (comprising 2 sites). Wetland SP:NR ranged from

47

0005-1 .1 and lake SP:NR from 0.03-1.1. Generally, 8042' production accounted
for 25-50% of NO3' removal (estimated from the inter-quartile ranges in Figure 4),
and the fraction of removal attributable to 8042’ production was higher in

wetlands and streams than in lakes (KW ANOVA 9.394, df=2; p=0.009).

By applying the stoichiometric model described above (equations 1-4) to
the 90 experiments (considering each experiment separately), I compared how
N03“ removal (yielding either NH4+ or N2) relates to the measured 8042'
production in different aquatic ecosystems under the two alternative reaction
stoichiometries (DNRA vs. denitriﬁcation). The amount of 8042' produced
explained a signiﬁcant fraction of the N03' removed in all types of aquatic
ecosystems (Figure 6). Lakes have the highest S-dependent N removal,
followed by streams and wetlands (medians, Figure 6), though there is
considerable variation within a given ecosystem type. The 8:5 ratio of NO3‘
consumed to 8042' produced (grey boxes, Figure 6) accounts for a greater
fraction of the overall NO3' removal. When considering the possibility of NH4+ as
the end product, as would be the case for a 1:1 ratio of NO3' consumed to NH4“
produced (black boxes, Figure 6), a slightly smaller, but still signiﬁcant proportion

of the N03‘ removal could be explained by the 8042' production.

48

Table 1: Physico-chemical characteristics of the sites in this study.
Abbreviations: D0 = dissolved oxygen; Cond. = conductivity; SW = surface
water; PW = porewater; BDL = below detection limits. (*) denotes sites that were
returned to in 2005 for the "N03 push-pull experiments.

49

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

S 3 Name N... No .0. E. 8.3 Em m. E 9.3 .553.
em Nu «at 3 mt mm... 8m. m. E emu 9.3 38923
wow «.9. 38 em Sm 5... Km 32 New 9.3 E9998:
8 ma 3% mm 58 Se 5 ea 3: 2395 can. 4......5.
F 34. a»: no ..om. SN «3 me «.8 225.5 4.95
9: we «.3. no mam 8s 2 ed New 8395 .8 XS
..2 3 a: 3 ..om ems 2 ma 4.8 239... 2 x4...
3 No 4.0% to 8.5 03 5 02 0.8 .552. 44$.
RF me am No m... :5 mm E new “5.6.... mm:
8m New. 38 a... No 3.5 one .93 EN 233.. ..e....aeaoe..
.48 3. mm to we 88 mm o... :N 2.4.6.... 53.85
59 3: «wow 5:0. E Ex 9:. 3 EN 9.30.... .....EEE
m2 3. ©me mm 39, w: m; E: E: E35 3.6 9.30.
5 QN 38 5 mi 3... ...mm 3: 3: E35 9.2.5
2 to A. 5 E. 95 8.5 NE 2. NS E35 ..Eeq
5 no St No 4. R 9: 08 E 13 E35 3.89.4
ewe we :5 ed 3 5... men 3. gm E35 E823.
3. 5.0 .58 me I: R... N3 3 0.? E35 «imam.
5 ed a. a. I «.8 Ex 8... 3 0.3 E35 933.4
8 no 32 E New 83 8.. to 4.8 E35 .95
to; no Nome a... 0.9. mm... own 32 33 E35 .5.er
E: 5.... .2: E: E: m: ._ 9: 6.
44:2 mu... .465 ...:z .62 :e .260 on 9:3
.54. .54. .5 .sm .5 .5 >5 .5 .5 E9285 9.5

 

 

 

 

 

 

 

 

 

 

 

50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00 __ T 4
TE -o.01e —
‘E’ -0.02 — —
g ,
.7. -043 - l 4
C
8 -0 04 — —-
.. - l
‘3 005- —~
t; J.
g -0.06— —
E -0.07— ~
.2 -0.08— —
Z 0
-0.09 ' L 1
Lakes Streams Wetlands
n=18 n=36 n=36

Figure 4: Nitrate removal rate constants by ecosystem. “n" refers to the number
of individual experiments (typically 3 per site), not to the number of sites of that
particular ecosystem. Boxes encompass the upper and lower quartiles, while the
line indicates the median for the dataset. Asterisks are mild outliers and open
circles are extreme outliers.

51

 

x
{-3

 

 

8042' production : NO3' removal

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 .I. .i.

Lakes Streams Wetlands
n=18 n=36 n=36

Figure 5: Ratios of total 8042' production : total NO3' removal (SP:NR) by
ecosystem. “n” refers to the number of individual injectors (typically 3 per site).
A value of 1 indicates that all of the nitrate removal can be explained by the
sulfate production. Values greater than one indicate that sulfate production can
account for more nitrate removal than was measured.

52

The pore waters of each injector were sampled for NH4+ and H28 prior to
application of the treatment (NO3' + Br“ or the Br'-only “control") and at the end of
the experiment. Figure 7 illustrates the pre and post treatment NH4+ (top) and
H28 (bottom) concentrations in all experiments. Generally there was more NH4”
production and H28 depletion in NO3'-amended sediments (solid) than in the
control injections (open). This pattern reﬂects what would be expected if DNRA
was coupled with 8 oxidation, there is no signiﬁcant difference between the
“post” control and treatment injectors for either the NH4+ concentrations (ANOVA
F1,77=1.8; p =0.19) or H28 concentrations (ANOVA F1,67=0.44; p=0.51).

Additional evidence for the importance of DNRA as a NO3' removal
pathway in aquatic ecosystems comes from the 2005 experiments in which I
added 15NO3' to the sediments. Wetlands had the greatest variation in the
percent of NO3' removal that could be attributed to DNRA, ranging from 5-110%.
Streams and lakes had comparable amounts of NO3' removal attributable to

DNRA, ranging from 3.5—37% and 0.542%, respectively (Figure 8).

53

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

120 I I l

100— .. . a -
g 80F- a
O
E
9 -
-.., 50- —*
o
E l
g 40- 4] _
B ,
..5 . -
0.. 20— g l —

O lLI III 1. l

 

Lakes Streams Wetlands
n=18 n=36 n=36

Figure 6: The fraction of measured nitrate removal that can be explained by
sulfate production using equations 1-3 (grey boxes) and equation 4-6 (black
boxes). “n” refers to the number of individual injectors (typically 3 per site), not to
the number of sites of that particular ecosystem.

54

10

 

+ N03"
T —0— Control

 

 

 

 

H28 concentration (uM)

 

 

 

Pre Post
160

 

+ NO3'

150 2 22 —0— Control

 

 

140 -

130 -

120 a

110 -

NH4? concentration (uM)

 

100 -

 

 

90‘

 

 

 

80 I r
Pre Post

Figure 7: Pre and post treatment NH4" concentrations (bottom) and H28
concentrations (top) for all injectors at all sites in both 2004 and 2005. Closed
symbols denote injectors that received N03' additions and open symbols denote

control injectors that only received a Br' addition. Error bars are standard errors
of the mean.

55

As part of the 2005 experiments, I added both “N03 and 15NO3' in
separate injectors within the same site. In both treatment types I measured the
pre and post treatment NH4+ concentration. The top panel of Figure 9 shows a
positive albeit weak relationship between the pmoles of 15NH4+ produced
(measured from the 15NO3' treatment) and the change in NH4+ concentration
(post—pre NH4+) in the same injectors (F145 = 4.2; p=0.056). The bottom panel
compares the 15NH4+ produced to the change in NH4+ concentration in the "N03
injectors within the same site (i.e., same site, but different injector). There is no
relationship between these two variables (F1,13=0.76; p=0.39).

Surface and porewater chemical characteristics were measured at each
site where a push-pull experiment was conducted (Table 1). These data were
used in a stepwise multiple regression model to examine what site characteristics
best predicted both NO3' rate constants and SP:NR across sites. NO3' rate
constants and SP:NR ratios were square-root transformed to meet the
requirement for a normal distribution.

NO3' rate constants were best predicted by surface water (8W) NO3‘ and
NH4” concentrations and porewater (PW) H28 concentrations (Table 2). Sites
with higher surface water NO3' had higher N03‘ removal rate constants.
Additionally, sites with lower surface water NH4+ and lower H28 had higher NO3‘
removal rate constants. These three variables explained 58% of the variation in
N03‘ removal rate constants across sites.

SP:NR ratios, indicative of the relative importance of 8042' production in

N03‘ removal, were best predicted by a combination of surface water 8042' and

56

porewater NH4+ concentrations (Table 2). Sites with higher SP:NR ratios had
higher surface water 8042' and porewater NH4+ concentrations, whereas sites
with greater SP:NR ratios had lower porewater NH4+ concentrations. The
combination of these two variables explained 44% of the variation in SP:NR

ratios across sites.

57

120 I I I

 

A
O
O
I
————4
I

 

00
O
I
I

 

 

 

% of N removal by DNRA
CD
0
l
l

 

 

 

 

 

 

 

 

 

 

 

 

 

40— r T —
20* ‘
O I T .L

 

Lakes Streams Wetlands
n=9 n=9 n=9

Figure 8: Percentages of NO3' removal attributed to DNRA as measured using
15N03'additions in combination with push-pull techniques. “n” refers to the
number of individual injectors (typically 3 per site).

58

900
300 y =196.13x+ 63.617

R2 = 0.136
700

600
500
400
300
200
100

p=0.056

A Nl-I4+ concentration (uM)

 

 

 

0 0.2 0.4 0.6 0.8 1
15NI-I4+ produced (umoles)

900 . w .,,. w , 4 ..
800
700
600
500 °
400
300
200
100 o

0 es . e 9

0 0.2 0.4 0.6 0.8 1

A Nl-I4+ concentration (uM)
.0
O

 

 

15NH4+ produced (umoles)

Figure 9: The relationship between measured 1"NH4+ production (x-axis) and the
change in NH4+ concentration in the ”N03 injectors (same injectors, top panel)
and the ”N03 injectors (different injectors within the same site, bottom panel).

59

Table 2: Stepwise multiple regression model (a = 0.01) results to predict the best

indicators of NO3' removal rate constants and 8042' production : NO3' removal
ratios (SP:NR).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Effect Coefficient Std Error t statistic p
NO3' rate constants model R2 = 0.582
Model 0.199 0.018 11.11 0.000
constant
SW NO3' 0.013 0.006 2.27 0.039
SW NH4+ -0.003 0.001 -3.26 0.006
PW H28 0021 0.006 -3.44 0.004
8042' production : NO3‘ removal (SP:NR) model R2 = 0.440
Model 0.309 0.119 2.59 0.018
constant
SW 8042' 0.006 0.003 2.43 0.025
PW NH4‘ -0.001 0.000 -2.06 0.055

 

60

Discussion:
Nitrate removal across aquatic ecosystems

Nitrate removal is often considered to be a beneﬁcial service provided by
aquatic ecosystems (Zedler 2003), particularly in agricultural landscapes such as
those in SW Michigan. This study is consistent with that generalization,
quantifying how quickly N03“ can be removed in sediments of many different
types of aquatic ecosystems (Figure 4). Additionally, all ecosystem types
exhibited a similar amount of variation in N03' removal rates, indicating that
perhaps there is nothing distinctive about NO3' removal in sediments of streams,
lakes, or wetlands. Nitrate removal may be driven by the transport of N03' (or
lack thereof) into the sediment pore waters. This premise, however, would
predict that there should be higher rates of NO3' removal in streams, due to their
greater turbulence and hydrologic connectivity to pore waters. I did not ﬁnd
evidence for higher N03“ removal rates in streams as compared to wetland or

lake sediments.

Evidence of linkage between NO3‘ removal and 8042’ production

This study revealed that NO3' removal and concurrent 8042' production is
a biologically mediated process that is found across diverse freshwater
ecosystems in southwest Michigan. Evidence to support the assertion that the
process is biological can be found in the intermediate temperature optimum seen
for both NO3' removal and 8042’ production (Figure 2). If the reaction was not
enzymatically mediated, one would expect increasing rates of 8042‘ production

with increasing temperatures. Evidence that this is not merely an experimental

61

artifact of the push-pull experimental method can be found from the N03‘
additions to a sediment mesocosm (Figure 3). In this experiment, I found that the
same NO3-induced 8042' production occurred, albeit over a longer time-scale.
Further evidence of the common occurrence of nitrate—driven sulfate
production can be found both in the large number of sites in which it was
observed and in the substantial fraction of NO3’ removal that could be attributed
to 8042' production (Figure 5). Sulfate production commonly explained 25-50%
of the N03' removal across aquatic ecosystems (based on the interquartile range
of SP:NR ratios). Lakes and streams generally had higher amounts of 8042‘
production relative to NO3' removal than did wetlands (Figure 5), though there
was a great deal of variation within each ecosystem type. Wetland porewaters
tended to have higher H28 concentrations (the hypothesized electron donor for
N03‘ reduction). This relative abundance of electron donors may have caused
H28 to be oxidized only to elemental sulfur (as in equations 1 and 4) rather than
all the way to 8042' (as in equations 3 and 6), which may have led to the overall
decrease in the importance of 8042' production relative to NO3' removal (SP:NR).
A ﬁnal line of evidence for the importance of NO3' reduction coupled to
H28 oxidation comes from examining the concentrations of NH4“ (a potential
product) and H28 (the hypothesized reactant) before and after the N03' additions
(Figure 7). Across all of the individual experiments from both years, NH4“
concentration increased after adding NO3' compared to the control, while the H28
concentrations simultaneously decreased. While the differences illustrated in

Figure 7 are not statistically signiﬁcant, they are consistent with the general

62

pattern I would expect to see if N-S coupling was present. This is a further line of
evidence that simultaneous N03” and H28 removal coincide with 8042' and NH4”
production, all of which indicate the presence and importance of a pathway
linking S and N cycling in these freshwater sediments. This study adds to the
observations of others (Lucassen et al. 2002, Spares 2002, Forshay 2003,
Whitmire 2003, Whitmire and Hamilton 2005) that there may be important but

relatively unexplored linkages between the N and 8 cycles.

What kinds of sites have NO3‘ linked 3042' production?

Multiple linear regression models suggested that of the surface and
porewater chemistry variables measured, surface water 8042' and porewater
NH4+ were the best predictors of a given site’s SP:NR, which in turn indicates the
amount of 8042' production relative to NO3' removal. Speciﬁcally, sites with
higher SP:NR ratios had a higher surface water 8042' concentrations and lower
porewater NH4+ concentrations (Table 2). The positive relationship between
SP:NR and surface water 8042‘ makes intuitive sense since sites with high
surface water 8042' may be sites with higher rates of 8 cycling, and in particular
higher 8042‘ reduction. H28 is the product of 8042' reduction, so both would be
necessary to support populations of S oxidizers. However, the negative
relationship between SP:NR and porewater NH4+ may be because sites with
higher 8042' reduction potential and more porewater NH4+ are more thoroughly
and consistently anoxic, and thus may have less 8 oxidation potential.

The same approach was used to predict which variables would be good

indicators of a site’s NO3' removal rate. The model showed a positive

63

relationship between NO3' removal rates and surface water NO3‘ concentrations
(Table 2), suggesting that sites with higher NO3‘ also have higher removal rates.
It also showed a negative relationship between NO3' removal rates and surface
water NH4+ and porewater H28 concentrations. Higher NO3' removal rates at
lower porewater H28 concentrations would make sense if indeed H28 was the
reactant to drive N-S coupled cycling. On the other hand, it may reﬂect the site’s
antecedent conditions of low redox state, corresponding with low NO3'
availability. The relationship between higher NO3' removal rates and lower
surface water NH4+ is less obvious, but may be linked through high rates of
nitriﬁcation, which would decrease the NH4" pool, particularly in the more
oxygenated surface water, while simultaneously producing N03, The higher NO-
3‘ production (via nitriﬁcation) could effectively prime the N03' reducing

communities and be reﬂected in the higher reduction rates.

NO3‘ removal end-products

In this study, I use two methods to estimate nitrate removal and its end-
products: a stoichiometric approach (eq. 1-6) and 15N tracer methods.
Stoichiometric methods inherently rely on a mass balance approach, but relate
the ﬂuxes of N to another element (Groffman et al. 2006), sulfur in this case.
Rates of DNRA can be directly measured using stable isotopes to track the ﬂow
of ”N from N03' to NH4+.

Sulfate production can account for a variable but signiﬁcant fraction of
overall NO3' removal when I apply the stoichiometric model outlined in equations

1-3 (for S coupled denitriﬁcation; Figure 6 grey boxes) and 4-6 (for S-coupled

64

DNRA; Figure 6 black boxes). In general, the sulfate production explained
between 25-40% of nitrate removal in lakes, 15-25% of removal in streams, and
10-15% of removal in wetlands (medians, Figure 6). There was, however, a
great deal of variation in both streams and lakes, and a smaller degree of
variation in the amount of sulfate production that could account for nitrate
removal in wetlands.

To elucidate these pathways with greater clarity than our stoichiometric
model can provide, I performed push-pull experiments with 15NO3' in the same
stream, lake and wetland sites that were sampled in 2004 (Figure 8). Wetlands
had the greatest range of nitrate removal attributable to DNRA, whereas streams
and lakes had comparable ranges of DNRA. Tiedje (1988) hypothesized that
fermentative DNRA should occur in the most biologically productive sites where
sediments were most highly reducing, which could explain why wetlands had
higher DNRA than either lakes or streams (though the pattern was not
statistically signiﬁcant), which generally don’t have as reduced conditions as
wetlands.

The median amount of NO3' removal predicted for the reduction to NH4+
(equations 4-6; Figure 6 black boxes) agrees reasonably well with the DNRA
measured via 15N methods for lakes and streams (about 20-30% in both cases).
However, there is a large discrepancy between the amount of nitrate removal to
NH4+ as predicted by the 8042' production (Figure 6, black boxes) and the DNRA
measured via 15N methods in wetlands. The 15N approach estimated that

roughly half of the N03' was converted to NH4+ in wetlands (Figure 8) whereas

65

the sulfate production only estimated 10% of the overall nitrate removal was to
NH4*. This suggests that the stoichiometric model explained above (Figures 4-6)
and represented in Figure 6 may drastically underestimate the amount of N03'
being lost to DNRA, particularly in wetlands. On the other hand, the
stoichiometric model may overestimate the amount of NO3' being lost to DNRA in
some streams and wetlands (see outliers in Figure 6). However, while there is a
great deal of variation, it is important to note that the median values of measured
DNRA (Figure 8) and the median values of the stoichiometric model (Figure 6,
black bars) are very similar, explaining ~20% of the nitrate removal. In this
regard the two methods for estimating the importance of N-S coupled cycling (via

either measured sulfate production or measured DNRA) agree well.

Measuring DNRA using ”N03 vs. ’5NO3' additions

In this study I attempted to measure DNRA by quantifying the increase in
porewater NH4+ concentrations after NO3' injection (Figure 7), and by using stable
isotope enrichment experiments (Figure 8). I can then compare the DNRA
measured via the stable isotope enrichments to the NH4+ increase in the same
injector that the ”N03 was added to (top panel, Figure 9), and also to the
increase of NH4+ in injectors that were in the same site, but at a slightly different
location (Figure 9, bottom panel). There is a signiﬁcant positive relationship in
the ﬁrst comparison, suggesting that within the same site, injectors with high
DNRA also displayed increases in NH4+ concentration. However, injectors that
were also placed in a given site, but received 14N03", did not have a signiﬁcant

positive relationship with the measured DNRA in that site. The results of both

66

comparisons (Figures 7 and 9) emphasize how difﬁcult it is to measure DNRA by
quantifying changes in NH4+ concentration alone. This is in large part due to the

high degree of variation in NH4+ production both within a given site and between

sites of a speciﬁc ecosystem.

I also measured 15N2 production (indicative of denitriﬁcation) as part of the
"N03 experiments, which would have allowed me to directly compare it to
DNRA. However, we concluded that since I was adding 99% ”N03 in the
experiments, in many cases large amounts of 30N2 may have been produced.
The 30N2 was produced over 29N2 because there is little ambient nitrate (“N03")
in the pore water for the 15NO3’ that was added to be mixed. Isotope ratio mass
spectrometers (IRMS) are often only tuned to measure 28 and 29 N2 because
the atmospheric amounts of 30N2 are very low, and that was the case when
samples from these experiments were analyzed. Thus, I was not able to get

accurate estimates of denitriﬁcation from the 15N2 measurements.

Implications for aquatic ecosystem N cycling

The existence and relative importance of denitriﬁcation versus DNRA has
profound implications for N cycling in aquatic ecosystems (Burgin and Hamilton
2007). Whereas nitrate that is converted to N2 is permanently removed from
biological availability, nitrate that is converted to NH4” becomes more biologically
available to many plants and bacteria, and is more likely to be retained within the
ecosystem. DNRA is a relatively understudied pathway compared to
denitriﬁcation, and it is not clear what happens to the resultant NH4*. In well-

oxygenated ecosystems, such as streams, the NH4+ may be re-nitriﬁed (via

67

nitriﬁcation), or it could be stored temporarily as sorbed ions in the sediments or
as organic N assimilated into biomass.

Nitrate is the most soluble N form, so removal of NO3' by either of these
processes is important to water quality; however, permanent removal (to N2 gas)
by denitriﬁcation is most desirable. If S-oxidizers are taking up and transforming
much of the N03‘ in surface or ground waters, then NO3' removal is closely linked
to S cycling, and speciﬁcally to 8042' inputs. Sulfate is a ubiquitous pollutant in
industrialized regions and atmospheric deposition of 8042' is greatly enhanced
over pre-industrial times (Schlesinger 1997). If excess 8042’ in freshwaters
actually enhances NO3‘ removal, by stimulating H28 formation through 8042‘
reduction, then the controls on N processing in landscapes subject to S and N

pollution become more complex than previously thought.

68

References:

Addy, K., D. Q. Kellogg, A. J. Gold, P. M. Groffman, G. Ferendo, and C. Sawyer.
2002. In situ push-pull method to determine ground water denitriﬁcation in
riparian zones. Journal of Environmental Quality 31:1017-1024.

Aminot, A., D. S. Kirkwood, and R. Kerouel. 1997. Determination of ammonia in
seawater by the indophenol-blue method: Evaluation of the ICES NUTS
l/C 5 questionnaire. Marine Chemistry 56:59-75.

Bernot, M. J., and W. K. Dodds. 2005. Nitrogen retention, removal, and
saturation in Iotic ecosystems. Ecosystems 82442-453.

Burgin, A. J., and S. K. Hamilton. 2007. Have we overemphasized the role of
denitriﬁcation in aquatic ecosystems? A review of nitrate removal
pathways. Frontiers in Ecology and the Environment 5:89-96.

Carpenter, S. R., N. F. Caraco, D. L. Correll, R. W. Howarth, A. N. Sharpley, and
V. H. Smith. 1998. Nonpoint pollution of surface waters with phosphorus
and nitrogen. Ecological Applications 8:559-568.

Forshay, K. J. 2003. Nitrogen dynamics in ﬂoodplain water bodies following _
inundation on the Wisconsin River ﬂoodplain. MS. Thesis. University of
Wisconsin, Madison, WI.

Fossing, H., V. A. Gallardo, B. B. Jorgensen, M. Huttel, L. P. Nielsen, H. Schulz,
D. E. Canﬁeld, S. Forster, R. N. Glud, J. K. Gundersen, J. Kuver, N. B.
Ramsing, A. Teske, B. Thamdrup, and O. Ulloa. 1995. Concentration and
transport of nitrate by the mat-forming sulfur bacterium Thioploca. Nature
374:713-715.

Gallardo, V. A. 1977. Large benthic microbial communities in sulﬁde biota under
Peru Chile subsurface countercurrent. Nature 268:331-332.

Golterman, H.L. and R8. Clymo. 1969. Methods for chemical analysis of
freshwaters. Blackwell, Oxford.

Groffman, P. M., M. A. Altabet, J. K. Bohlke, K. Butterbach-Bahl, M. B. David, M.
K. Firestone, A. E. Giblin, T. M. Kana, L. P. Nielsen, and M. A. Voytek.
2006. Methods for measuring denitriﬁcation: Diverse approaches to a
difﬁcult problem. Ecological Applications 16:2091-2122.

Haaijer, S. C. M., M. E. W. Van der Welle, M. C. Schmid, L. P. M. Lamers, M. S.
M. Jetten, and H. J. M. Op den Camp. 2006. Evidence for the involvement

69

of betaproteobacterial Thiobacilli in the nitrate-dependent oxidation of iron
sulfide minerals. FEMS Microbiology Ecology 58:439-448.

Holmes, R. M., J. W. McClelland, D. M. Sigman, B. Fry, and B. J. Peterson.
1998. Measuring N-15NH4+ in marine, estuarine and fresh waters: An
adaptation of the ammonia diffusion method for samples with low
ammonium concentrations. Marine Chemistry 60:235-243.

lstok, J. D., M. D. Humphrey, M. H. Schroth, M. R. Hyman, and K. T. OReiIly.
1997. Single-well, "push-pull" test for in situ determination of microbial
activities. Ground Water 35:619-631.

Jorgensen, B. B., and V. A. Gallardo. 1999. Thioploca spp: ﬁlamentous sulfur
bacteria with nitrate vacuoles. FEMS Microbiology Ecology 28:301-313.

Kamp, A., P. Stief, and H. N. Schulz-Vogt. 2006. Anaerobic sulﬁde oxidation with
nitrate by a freshwater Beggiatoa enrichment culture. Applied and
Environmental Microbiology 72:4755—4760.

Kelly, D. P. 1999. Thermodynamic aspects of energy conservation by
chemolithotrophic sulfur bacteria in relation to the sulfur oxidation
pathways. Archives of Microbiology 171 :219—229.

Lovley, D. R., and M. J. Klug. 1983. Sulfate reducers can out-compete
methanogens at freshwater sulfate concentrations. Applied and
Environmental Microbiology 45: 1 87-1 92.

Lucassen, E. C. H. E. T., A. J. P. Smolders, and J. G. M. Roelofs. 2002. Potential
sensitivity of mires to drought, acidiﬁcation and mobilisation of heavy
metals: the sediment S/(Ca+Mg) ratio as diagnostic tool. Environmental
Pollution 120:635-646.

Luthy, L., M. Fritz, and R. Bachofen. 2000. In situ determination of sulﬁde
turnover rates in a meromictic alpine lake. Applied and Environmental
Microbiology 66:712-71 7.

McGuire, J. T., D. T. Long, M. J. Klug, S. K. Haack, and D. W. Hyndman. 2002.
Evaluating behavior of oxygen, nitrate, and sulfate during recharge and
quantifying reduction rates in a contaminated aquifer. Environmental
Science & Technology 36:2693-2700.

McHatton, S. C., J. P. Barry, H. W. Jannasch, and D. C. Nelson. 1996. High

nitrate concentrations in vacuolate, autotrophic marine Beggiatoa spp.
Applied and Environmental Microbiology 62:954-958.

70

 

Sayama, M., N. Risgaard-Petersen, L. P. Nielsen, H. Fossing, and P. B.
Christensen. 2005. Impact of bacterial NO3' transport on sediment
biogeochemistry. Applied and Environmental Microbiology 71 :7575—7577.

Schlesinger, W. H. 1997. Biogeochemistry: an analysis of global change, 2nd
edition. Academic Press, San Diego.

Schulz, H. N., T. Brinkhoff, T. G. Ferdelman, M. H. Marine, A. Teske, and B. B.
Jorgensen. 1999. Dense populations of a giant sulfur bacterium in
Namibian shelf sediments. Science 284:493-495.

Schulz, H. N., and B. B. Jorgensen. 2001. Big bacteria. Annual Review of
Microbiology 55: 1 05-1 37.

Soares, M. l. M. 2002. Denitriﬁcation of groundwater with elemental sulfur. Water
Research 36:1392-1395.

Sweerts, J. P. R. A., D. Debeer, L. P. Nielsen, H. Verdouw, J. C. Vandenheuvel,
Y. Cohen, and T. E. Cappenberg. 1990. Denitriﬁcation by sulfur oxidizing
Beggiatoa Spp mats on freshwater sediments. Nature 344:762-763.

Tiedje, J. M. 1988. Ecology of denitrification and dissimilatory nitrate reduction to
ammonium. Pages 179-244 in A. J. B. Zehnder, editor. Biology of
Anaerobic Microorganisms. Wiley & Sons, New York.

Vitousek, P. M., J. D. Aber, R. W. Howarth, G. E. Likens, P. A. Matson, D. W.
Schindler, W. H. Schlesinger, and D. G. Tilman. 1997. Human alteration of
the global nitrogen cycle: Sources and consequences. Ecological
Applications 7:737-750.

Wetzel, R. G. 2001. Limnology: Lake and River Ecosystems, 3 edition. Academic
Press, San Diego.

Whitmire, S. L. 2003. Anaerobic biogeochemical functions of Michigan wetlands
and the inﬂuence of water source. Ph.D. Dissertation. Michigan State
University, East Lansing, MI.

Whitmire, S. L., and S. K. Hamilton. 2005. Rapid removal of nitrate and sulfate in
freshwater wetland sediments. Journal of Environmental Quality 3422062-
2071.

Zedler, J. B. 2003. Wetlands at your service: reducing impacts of agriculture at
the watershed scale. Frontiers in Ecology and the Environment 1:65-72.

71

CHAPTER 3

THE RELATIVE IMPORTANCE OF DENITRIFICATION, DISSIMILATORY
NITRATE REDUCTION TO AMMONIUM (DNRA), AND ANAEROBIC
AMMONIUM OXIDATION (ANAMMOX) TO NITRATE REMOVAL IN

FRESHWATER SEDIMENTS

Introduction:

Excessive nitrogen (N) concentrations, often in the form of nitrate (NO3‘),
present a water-quality problem of growing concern. Increasing problems with
eutrophication of coastal marine waters are linked to the export of N from
terrestrial landscapes. Surface and groundwater ﬂow within the landscape often
passes through wetlands and headwater streams, where much of the N03
present in the water can be removed (Peterson et al. 2001, Zedler 2003). Thus,
key interfaces along landscape ﬂow paths control N export to downstream
surface waters, such as large rivers and lakes, and ultimately to estuaries. The
removal of NO3’ by wetlands and streams is of particular interest in agricultural
landscapes where N export by rivers has been implicated in creating hypoxic
zones (such as in the Gulf of Mexico) and harmful algal blooms (Rabalais et al.

2001, Paerl et al. 2002).

As NO3' rich water moves through a landscape, many different processes
can remove the nitrate, reducing the amount of loading to downstream
ecosystems. These pathways include respiratory denitriﬁcation, dissimilatory
nitrate reduction to ammonium (DNRA), and anaerobic ammonium oxidation

(anammox; indirectly through the reduction of N03' to NO2'). Respiratory

72

denitriﬁcation has long been thought to be the predominant microbial pathway for
nitrate removal in freshwater sediments. Respiratory denitriﬁcation isa
microbially mediated transformation that reduces nitrate coupled to the oxidation
of organic carbon. The N03' is sequentially reduced to nitrite (NO2'), nitrous
oxide (N20) and di-nitrogen gas (N2). Many factors inﬂuence rates of respiratory
denitriﬁcation in surface waters, including oxygen, NO3' and carbon availability,
light, and the presence of plants (Knowles 1982, Goltennan 2004).

There have been many studies of denitriﬁcation in diverse ecosystems,
and attempts have been made to uncover broad patterns across ecosystem
types through meta-analyses (Pina-Ochoa and Alvarez-Cobelas 2006) and
compilations and analyses of published rate measurements (Seitzinger1988,
Comwell et al. 1999). Denitriﬁcation rates have been assessed in soils,
wetlands, and surface waters, but estimates vary greatly within and among
environments, as well as between different measurement techniques. Wetlands
can be particularly efﬁcient NO3' sinks and support high rates of denitriﬁcation
(Peterjohn and Correll 1984, Hedin et al. 1998, Tobias et al. 2001, Whitmire and
Hamilton 2005). Headwater streams are also efﬁcient N processing sites
(Peterson et al. 2001, Mulholland et al. in prep). However, in a large study of 72
streams acroSs the US, denitriﬁcation accounted for a relatively small proportion
(median of 16%) of the overall nitrate removal (Mulholland et al. in prep).
Seitzinger (1988) and Cornwell et al. (1999) concluded that denitriﬁcation
removes highly variable but signiﬁcant fractions of the total N loading to lakes,

rivers, and coastal estuaries. While denitriﬁcation can often account for a

73

signiﬁcant fraction of the overall nitrate removal, it rarely accounts for the entire
amount. The balance of this removal is often ascribed to biological assimilation
(Burgin and Hamilton 2007).

An alternative pathway of nitrate removal that has received relatively little
study is dissimilatory nitrate reduction to ammonium (DNRA). This pathway
involves the transformation of N03' to ammonium (NH4+) either by fermentative
bacteria utilizing carbon substrates (Tiedje 1988) or by chemolithoautotrophic
bacteria that can utilize N03“ to oxidize reduced sulfur compounds, such as
sulﬁde (H28), producing sulfate (Fossing et al. 1995, Brunet and Garcia-gil 1996).
Tiedje (1988) suggested that fermentative DNRA would be most important in
highly reducing environments that maintain anoxic conditions for long time
periods. Additionally, DNRA is thought to be favored in NO3'-limited, labile-

. carbon rich environments while respiratory denitriﬁcation would be favored under
carbon-limited conditions (Tiedje, 1988). Sulfur-driven DNRA, on the other hand,
may be controlled by the availability of reduced sulfur compounds, including H28
and 8°, to use as electron donors. Furthermore, H28 may be a key driver in
determining the relative importance of denitriﬁcation and DNRA by denaturing the
ﬁnal denitrification reductases, thus shunting the pathway over to DNRA (Brunet
and Garcia-gil 1996).

DNRA has mostly been studied in wetlands and marine-inﬂuenced
ecosystems (Bowden 1986, Tobias et al. 2001, Gardner et al. 2006), and in
freshwater Lake Vilar (Brunet and Garcia-gil 1996). In Ringfield Marsh, a

brackish coastal marsh, DNRA accounted for 7-70% of the overall N03'

74

reduction; the relative importance of DNRA and denitrification varied between
seasons depending on groundwater inputs (Tobias et al. 2001). Bonin (1996)
found that DNRA accounted for 80% of the total NO3‘ consumption in sediments
of the Mediterranean coast of France, even though bacterial biomass estimates
suggested that ammonium-producers were 100-fold less abundant than
respiratory denitriﬁers. An and Gardner (2002) demonstrated the relative
importance of DNRA in many Texas estuaries; in a few of the systems, DNRA
was more important than denitriﬁcation to overall NO3' removal. In Lake Vilar
(Spain), Brunet and Garcia-Gil (1996) found that adding N03' to the H28-rich,
anoxic hypolimnion resulted in H28 removal and NH4+ production. In a study of a
freshwater tidal marsh in MA, Bowden (1986) found that DNRA accounted for
less than 10% of the overall NO3' reduction. DNRA has apparently not yet been
investigated in freshwater streams.

Anaerobic oxidation of ammonium (anammox) is the process by which
NH4+ (electron donor) is oxidized by NO2‘ (electron acceptor) under anaerobic
conditions, producing N2. Anammox was discovered in sludge reactors, and has
since been shown to occur in anoxic wastewater, temperate shelf sediments, sea
ice, and more recently in cold Arctic shelf sediments (Jetten et al. 1998,
Megonigal et al. 2004, Rysgaard and Glud 2004, Rysgaard et al. 2004), where it
has been estimated to account for a wide range (1-65%) of total N2 production
(Dalsgaard et al. 2005). In marine sediments the source of the N02' for
anammox is hypothesized to be incomplete denitrification of NO3'. It is apparent

that anammox occurs in areas where there are suitable concentrations of both

75

NO3' and NH4", but no 02, and is inhibited by many simple organic compounds
including pyruvate, ethanol, and glucose (Jetten et al. 1998).

There are very few published studies of anammox in freshwater
ecosystems. The exception to this is Schubert et al.’s (2006) study of anammox
activity in Lake Tanganyika, the second largest freshwater lake in the world.
Schubert’s study found that up to 13% of the N2 produced could be attributed to
anammox activity. Because Lake Tanganyika is a unique freshwater system,
generalizations about the role of anammox in freshwater N cycling remains
speculative (Megonigal et al. 2004). However, anammox would be expected to
occur where sufﬁcient NO3' and NH4+ co-occur, which could include certain

interfaces between surface and ground waters.

In this study, I examined the relative importance of denitriﬁcation, DNRA
and anammox to overall nitrate removal rates in sediments taken from two
streams, lakes and wetlands in southwestern Michigan. While denitriﬁcation has
been extensively studied in these types of systems, very few studies have
addressed whether DNRA and anammox are important to overall nitrate removal
in freshwater ecosystems. Additionally, no published studies quantify the relative
contribution of all three processes simultaneously. I also examined whether the
relative contribution of these processes to overall nitrate removal can be

predicted by ecosystem properties (e.g., ecosystem types).

76

Sites:

Two wetlands, lakes and streams were chosen as sites for this
experiment. Sites were selected to represent the biogeochemical diversity of the
local landscape, based largely on previous work in the Hamilton Lab. The two
wetland sites were Loosestrife Fen (also called Loosestrife Pond, LP) and Turkey
Marsh (TM). Loosestrife Fen is a small (0.4 ha) fen created from sediment
inﬁlling behind a small earthen dam located in the WK. Kellogg Experimental
Forest. It is dominated by Chara sp. and has a few centimeters of surface water
year-round due in part to the continuous groundwater inputs from a spring that
drives surface ﬂow across the wetland. Turkey Marsh is a 3.1-ha isolated,
depressional wetland located near the Kellogg Biological Station. The wetland is
both precipitation- and groundwater-fed, and its surface water levels ﬂuctuate

more than those in Loosestrife Fen.

The two lake sites were Wintergreen Lake (VVGL) and Lawrence Lake
(LAW). Wintergreen Lake is approximately 15 ha in area with a maximum depth
of 6.3 m and a mean depth of 3.5 m. The lake is hyper-eutrophic due in part to
its location within the Kellogg Bird Sanctuary. Lawrence Lake is an oligotrophic,
hard-water lake 5 ha in area with a maximum depth of 12.6 m. For further

background see (Wetzel 2001).

The two stream sites were Arcadia Creek (ARC) and Bellingham Drain
(BEL); both streams were also part of the Michigan Lotic lntersite Nitrogen
eXperiment (LINX2) whole-stream 15NO3' additions in 2005. Arcadia Creek is a

small stream with a largely urban catchment located in the city of Kalamazoo, MI.

77

Groundwater maintains its baseﬂow, but many urban drains also empty into it
making it a typically “ﬂashy” (e.g., highly variable discharge) urban stream. In
contrast, Bellingham Drain was excavated for agricultural land drainage and its
catchment is largely covered by row-crop production (corn and soy). It is a

tributary of the Gun River, which in turn empties into the Kalamazoo River.

Methods:

A ﬂow-through core incubation method was used to determine the relative
importance of DNRA, denitriﬁcation and Anammox to total nitrate removal in
sediments (An and Gardner 2002). Eight 7.6-cm ID. by 30-cm long undisturbed
sediment cores were collected from each of the six locations at each site, along
with approximately 100-L of surface water. Cores were collected by hand where
the water was shallow, or in the lakes from a boat using a coring device ﬁtted
with a one-way rubber valve. The wetland cores were collected on 16 July and
the experiments were run 17-19 July 2006. The lake cores were collected on 13
August and the experiments were run from 14-16 August 2006. The stream
cores were collected on 29 Sept and the experiments run from 30 Sept -2 Oct
2006.

Once the cores were collected and returned to the lab some of the
overlying water was drained off, leaving approximately 2.5-3 cm of the overlying
water on the core. The core was ﬁtted with a tight-ﬁtting, O-ring plug that
contained Teﬂon inﬂow and outﬂow lines embedded into the plug. The inﬂow
lines were connected to a peristaltic pump that transferred the treatment water

from 20-L reservoirs to the cores at a rate of 1.2 ml min". The treatments were

78

site surface water 1) without any 15N (control - 2 cores per site); 2) with added
1°NO3' to examine the relative importance of DNRA and denitrification (increased
concentration by ~0.66 mgL"; 3 cores per site); and 3) with added 15NH4+ to test
for the presence of anammox (increased the concentration by ~0.33 mg L"; 3
cores per site). If sites had little ambient NO3' in the surface water (TM, LP,
WGL; Table 1), enough NO3' was added to the NH4" treatment to increase the
N03' concentration to 0.2 mg N L“1 to provide adequate concentrations for
anammox to occur. The treatment reservoirs (inputs) were aerated to maintain
oxic conditions. The cores were kept in the dark at ambient (lab) temperature.
The water from the outﬂow lines was collected in 1-L bottles that were repeatedly
ﬁlled and emptied throughout the experiment.

The treatments were begun on the same day as the cores were collected.
After one day of incubation to allow the cores to equilibrate, duplicate gas and
water samples were collected for three successive days. Gas samples were
collected directly from the outﬂow lines into long, narrow tubes ﬁtted with glass
stoppers. These samples were then shipped overnight to the Gardner lab
(University of Texas Marine Science Institute, Port Aransas, TX) to be analyzed
by membrane inlet mass spectrometry (MIMS) for dissolved Ar, 02, 2°N2, 29N2,
and °°N2. Samples for nutrients were taken at approximately the same time as
the gas samples, ﬁltered with a syringe ﬁlter (0.45 pm), and analyzed for NH4"
concentration using the phenylhypochlorite method (Aminot et al. 1997) and for
anions and cations (including NO3‘ and 8042') on a Dionex membrane-

suppression Ion Chromatograph. Samples were collected for.1°NH4+ and were

79

analyzed by a modified diffusion method (Holmes et al. 1998). In addition to the
later method, water samples were collected and filtered (0.22 pm) for ”NH4+
analysis by an HPLC technique (Gardner et al. 1991, Gardner et al. 1996).

All ﬂux measurements were corrected for any background ﬂuxes (e.g.,
14N03 removal in sites with high ambient N03 concentrations) and converted to
tracer ”N based on the 15N atom ratio [AR; ”N/(”N+”N)]. Fluxes are generally
reported as activity of the ”N component of the N03 pool, with the exception
being the overall NH4“ ﬂuxes. Nitrate removal rates were calculated as the
difference between ”N03 concentrations in the outﬂow and the inﬂow on a
surface area basis. I assumed that the ”N AR in the outﬂow was the same as
the AR in the inﬂow water, and calculated the ”N03 removal rate as the
difference in the ”N component of the N03 pool between the inﬂow and outﬂow.
Denitriﬁcation rates were calculated as the sum of 15N in the forms of 29N2 and
30N2 that were produced in the presence of ”NO3‘. DNRA rates were calculated
as the ”NH4" produced in the presence of ”N03. Anammox rates were

calculated as the ”N in 29N2 produced in the presence of ”NH4" and ”N03.

Statistical Analysis

Samples were analyzed for changes over time (3 day incubation time)
using a standard analysis of variance (ANOVA) with time as a covariate. For
most response variables there was no signiﬁcant effect of time, so the
measurements from all three days were pooled into a site average. Further

comparisons were made with ANOVAs (8A8 PROC GLM) using Tukey’s

80

comparisons to perform a posteriori comparisons of individual sites or ecosystem

types (e.g., streams, wetlands, lakes).

Results: >

Physical and chemical characteristics

Experiments were conducted in the mid to late summer months, and thus
surface water temperatures were generally high, ranging from 18.6-25.4 °C
(Table 1). All sites also had high dissolved oxygen (DO) in overlying water at the
time of core collection. Nitrate concentrations were highest in the two stream
sites. Nitrate was measurable in both of the lakes, but was below detection limits
(~0.01 mg N L") in the two wetland sites. Ammonium concentrations were
generally 5 50 pg N L" with the exception of Loosestrife Fen (wetland), where
the NH4+ concentration was more than triple of that at most sites (Table 1).
Surface water sulfate concentrations were lowest in Turkey Marsh (wetland) at
1.9 mg L" and highest in Bellingham Drain (stream) at 123.9 mg L"; most

however were between 10-20 mg L".

Sediment Oxygen Demand

The sediment oxygen demand (SOD) indicates the rate at which the
sediments are removing 02 from the water ﬂowing over the sediment cores. It
reﬂects all of the 02 consuming processes that occur in the sediment-surface

water interface, including decomposition, chemical oxidation (e.g. Fe or H28),

81

and microbial processes such as nitriﬁcation. All sites consumed 02 (Figure 1).
There were no statistically significant changes in SOD rates over time, so all
measurements from the three days were combined for statistical analysis (n=6
per site). Arcadia Creek and Loosestrife Fen were two sites with especially high
SOD (Figure 1). Although there were signiﬁcant differences in SOD between
particular sites, there were no differences in SOD between ecosystem types, i.e.,
streams vs. lakes vs. wetlands (df=2, 51; F = 0.15; p=0.86). Adding N03 to the
overlying water did not signiﬁcantly change the SOD relative to controls (df=1,

88; F = 1.37; p=0.2446).

Nitrogen ﬂuxes

Measured nitrogen ﬂux and transformation rates are illustrated in Figures
2-6 and summarized in Figure 7 and Table 2. Fluxes are in pmoles rn'2 h" from
2-3 replicate cores per treatment. Positive values indicate net increase, or
production, whereas negative values indicate net removal or loss. All ﬂuxes are
background corrected to account for any activity in the control cores, as well as
corrected for the 15N atom ratio (AR) so that only the ﬂux of the added ”N was

measured and compared to other ﬂuxes.

82

:99; oomtnm u >>w .me= 82093 26.8 n ..om ”mco=m_>o.ﬁn<
698:8 9m; «.28 05 we: 05 “m «cum 9.: ho 8358820 .8520 new Reign 2. 83m...

 

 

 

 

 

 

 

 

 

 

 

2: 5o «.3 a? «.8 3V 9.3 529352,
:8 8.0 :8 m. : mum 2 9.3 8:225...
3 ..8 tom w .3 . 0.8 Sc sens. >92:
2: ..om 38 our EN Sc .8". £38.83
9va mm: 02 mm 3: E 5.2a 523:8
S: 86 mé em new E :85 «682
NV... 95 c... z 95 r... 2 9: amaze
ﬂow .62 ...:z oo o. mum
26 3m 3m 3m .95:

 

 

 

 

83

Nitrate removal rates varied considerably among sites, ranging from 114-
961 pmoles N03 rn'2 hr" (Figure 2). There were no significant changes in rates
over time, so all measurements were combined for analysis (n=9 per site).
Streams had higher rates of N03 removal than either wetlands (Tukey’s p=0.023)
or lakes (Tukey’s p=0.010), though this pattern is driven in large part by the high
N03 removal rates in Arcadia Creek, which were greater than in any other site
(Figure 2). One wetland (Loosestrife Fen) also had relatively high removal rates.
Generally sites with high SOD (Figure 1) also had high N03 removal rates.

Denitriﬁcation rates were also variable among sites ranging from 49—361
umoles ”N2 generated rn'2 hr" (Figure 3). There were no statistically signiﬁcant
changes in denitriﬁcation rates over time, so all measurements from the three
days were combined for statistical analysis (n=9 per site). Streams had much
higher rates of denitriﬁcation compared to wetland (T ukey’s p<0.0001) or lake
sites (Tukey’s p<0.0001). Wetlands and lakes had relatively similar rates of
denitrification ranging from ~50-100 pmoles 15N-N2 rn'2 hr" and were not
statistically different from each other (Tukey’s p=0.6763). These general patterns

were the same when individual sites were compared (Figure 3).

84

 

 

 

 

 

 

 

0 _ .......

-500 -
,f -1000 -
E a
(3
w -1500 -
CD
6 l a.b
E
3
o -2000 -
O
a)

-2500 -

'3000 I I r T I l

(66* ‘9'6‘ 000 6%“ 9‘66 0‘66
«‘0 ,6? (ix 6» 0»
¢\ “9 5x6 *e‘l 30° (0°
‘09 ‘09 9,0 «0‘ \s‘ 69
P‘ 96“ \loo \9 gm"

Figure 1: Average sediment oxygen demand (SOD) in the control cores from
each site over the three day incubation time (n=6). Error bars represent 1 SE. of
the mean. The ecosystem types are colored as: streams (black), wetlands (dark
grey), and lakes (light grey). Sites with the same letter are not statistically
different from each other (p=0.05).

85

 

 

 

 

-200 -

 

 

 

b.c
-400 -

-600 e

-800 d

-1000 -

 

”N03 removal rate (umoles N03- N m'2 hr")

 

 

 

' 1 2 0 0 I 1 I I l V
0‘066 0‘6\(\ Q 0.06 avg“ yay‘e V660
\e (0 '60 x; 00 a“
6 “0 '6 *0 9° (0
to") («9 9,05 «0‘ qt‘ 663

Figure 2:'”NO3 ﬂux rates from each site over the three day incubation time
(n=9). Error bars represent 1 SE. of the mean. The ecosystem types are
colored as: streams (black), wetlands (dark grey), and lakes (light grey). Sites
with the same letter are not statistically different from each other (p=0.05).

86

 

400

300 -

200 ~

100 ‘

 

”N2 ﬂux (29N2 + 30N2) (umoles m'2 hr“)

 

 

 

 

 

 

 

 

 

Figure 3: Denitriﬁcation rates measured as average 15N-N2 ﬂux (29N2+ 30N2) from
the ”N03 amended cores from each site over the three day incubation time
(n=9). Error bars represent 1 SE. of the mean. The ecosystem types are
colored as: streams (black), wetlands (dark grey), and lakes (light grey). Sites
with the same letter are not statistically different from each other (p=0.05).

87

Anammox rates were calculated as the amount of 29N2 produced in the
treatment containing ”NH4" and ”N03. Rates were typically low, ranging from
1-27 umoles 29N2 m'2 hr" (Figure 4). There were nostatistically significant
changes in anammox rates over time, so all measurements from the three days
were combined for statistical analysis (n=9 per site). Streams had higher
anammox rates than either wetlands (Tukey’s p=0.0032) or lakes (Tukey’s
p=0.0011), though this is in part driven by the very large anammox rates at
Bellingham Drain (Figure 4). Lake sediments had small amounts of anammox
(~2.5 umoles 29N2 rn'2 hr"), and wetlands had the lowest rates of anammox
activity though the two groups were not statistically different (Tukey’s p=0.75).

Ammonium production was measurable in all six sites in both the control
(solid bars, Figure 5) and 15N03 treatment cores (hatched bars, Figure 5).
However, the ﬂux from the ”N03 -amended cores was sometimes not different
than the ﬂux occurring in the same site’s control cores (e.g., Bellingham Drain,
Turkey Marsh and Lawrence Lake; p>0.05). Ammonium production was
stimulated by adding N03 to the overlying water compared to the control in
Arcadia Creek (df = 1, 13; F = 8.52; p=0.012), Loosestrife Fen (df=1,13; F=8.35;
p=0.013), and Wintergreen Lake (df=1,13; F=7.22; p=0.019) (Figure 5).

Generally, there were no statistically signiﬁcant changes in NH...+ ﬂux rates
over time, so all measurements from the three days were combined for statistical
analysis (n=9 per site for treatment, and n=6 for controls). The exception to this
is Arcadia Creek, which produced a pulse of high (mostly 14N) NH4+ in all three

”N03 treatment cores on the ﬁrst day, but then the NH4+ decreased considerably

88

on days 2 and 3. Though there was a significant effect of time at this site, the
data from all three days were combined for graphical and statistical analysis to
maintain consistency between sites. This NH4+ pulse is the driver of the high
variation in NH4+ﬂux at Arcadia Creek (Figure 5).

Rates of DNRA, measured as the ”NH4“ produced in the presence of
”N03 in the overlying water, was measurable at all six sites, ranging from 7-72
pmoles ”NH4+ produced rn'2 hr" (Figure 6). Though there were no differences in
DNRA rates among ecosystem types, there was considerable variation among
different sites (Figure 6). The sites with high DNRA (Arcadia, Loosestrife and
Wintergreen) were also sites that had a measurable increase in NH4+ production
(Figure 5) and the sites with the highest SOD (Figure 1).

The relative importance of these three nitrate removal pathways to overall
15NO3 removal is illustrated by site in Figure 7 and broken down as percentages
of the overall ”N03 removal in Figure 8 and Table 2. Denitriﬁcation is the
dominant removal process in most sites; however, with the exception of
Bellingham Drain, denitriﬁcation only accounted for 20-40% of overall N03
removal (Table 2). The amount of removal that DNRA could account for was of a
slightly smaller range at 2-18% of overall N03 removal. Anammox accounted for
the smallest fraction of N03 removal (0-10°/o). ln 5 of the 6 sites, a considerable
proportion (~50°/o) of the N03ﬂux that could be explained by the ”N budgets
and rates I measured (Figure 8). ln Bellingham Drain, I could account for slightly
more (~7°/o) of the overall nitrate removal than was measured by individual

process rates.

89

 

4O

30-

20%

10*

29N2 ﬂux (umoles rn‘2 hr“)

 

 

 

 

 

 

 

Figure 4: Anammox as control corrected average 29N2 ﬂux from the ”NH4+
amended cores from each site over the three day incubation time (n=9). Error
bars represent 1 SE. of the mean. The ecosystem types are colored as:
streams (black), wetlands (dark grey), and lakes (light grey). Sites with the same
letter are not statistically different from each other (p=0.05).

90

1800

 

1600 -

1400 -

 

1200 -

1000 -

800 -

600 -

NH4+ ﬂux (umoles m‘2 hr")

400 ‘

 

 

 

 

 

Figure 5: Average NH4+ ﬂux from the control cores (n=6; solid) and ”N03
amended cores (n=9; hatched) from each site over the three day incubation time.
Error bars represent 1 SE. of the mean. The ecosystem types are colored as:
streams (black), wetlands (dark grey), and lakes (light grey). (*) indicates that
the control and treatment cores had statistically different (p<0.05) NH4” ﬂuxes.

91

 

 

 

 

 

 

DNRA as 15NH.+ production (umoles ”NH3-N m-2 hr")

 

Figure 6: DNRA rates measured as 15NH4" production in the 15N03 amended
cores from each site over the three day incubation time (n=9). Error bars
represent 1 SE. of the mean. The ecosystem types are colored as: streams
(black), wetlands (dark grey), and lakes (light grey).

92

1000

 

 

- Denitriﬁcation
:3 Anammox
— DNRA

\‘
s

800 ~

 

 

600 ‘

 

 

400 ~

200 a

N03 removal (umoles N03- N m“2 hr")

 

 

 

 

0 -
0 \0 6 5‘0 0
‘6 (“0‘6 -\er<\ “0‘ 6V3”~ (0’66
600‘” 9 05”“ «‘6‘! “$er ‘9‘06
(>8 90““ 00% \’ \34 “\e

Figure 7: The relative importance of dissimilatory nitrate removal pathways to
overall nitrate removal in sediments from streams (black), wetlands (dark grey)
and lakes (light grey).

93

 

 

120

 

 

 

 

 

100 —
‘2“

o 80 —
E
93
but)

2 60 -
E
.9

‘5 40 —
S"

20 —

0 _

 

6\ a”
N09 90”“?

Figure 8: The relative importance of dissimilatory nitrate removal pathways as a
percentage of the overall nitrate removal rate from each of the six sites. The
hatched area represented the amount of "unknown" N03 removal, i.e., the N03
was removed but cannot be accounted for by the N end-products measured.

94

.EaSEEm 2 .9080. 90.0.: E90._E.mm.0 u 5.20
60:00.08 EacoEEm 0.00.0000 u onEmc< .cozmoctgcmu u zwo .w.w ....” 5.2. 8a.... mEmEmSmmmE ..0 h.0
mcmoE 0.0 00:.0> 09.0 >0 00000.0 :000 .0 _m>0E0. __0.0>0 .0 €00.00 0:0 020. 0500000.... .62 ”N 0.00...

 

 

 

 

 

 

 

 

5 0.0 a E a me « .... 3.3
0% 0.: 3 0.8 a om. 0.0 8 «mm. 52925.5
l 3 0.0 a ed a we « .... 8.3
0+0 2: 3 :0 a N 3 me 3:. 8:233
- 3 to a we a we a ..s. can...
9mm 2: 0.0 ode « S Z we 3 e- 9:3
we so a : a we a .55“.
New wt 0.0 0.2 a we 3 8 EV. 8.588..
00 E « mm a R. a .0. £20
:8 3 no «.00 a 5 EN mew mew. 520:...8
0.0 3 a E a mm a .0. 0.8.0
one we no QB a 00 m... am So. «.082
.0582 ...: N...: 1:. N....
025.0000 _0>oEo._ .0508. .62 2.; N.E 00.0.»... F...: N.... 00.05:
.m>0E0._ .62 *0 .x. .62 u.0 .x. “_0 e\.. 00.08.. 8:. 00.0.5. 08:
.62 .83 5.20 onE~=< ...oo $.20 .52.... .50 .622 2.0

 

 

 

 

 

 

 

 

 

 

95

 

 

 

 

 

 

 

 

 

1400
+ Arcadia Creek
1200 - —o— Bellingham Drain
- —-—-v—- Loosestife Pond
e 1000 ~
E -A. Turkey Marsh
‘7’ 800 - I Lawrence Lake
E
“a j 0 Wintergreen Lake
Vt
O 600
(I)
m 400 -
2
0
E 200 ~
3
-.-..- .-..-. -- . S"
g o « i --
NV '200 -
O I: §
‘0 -400 -
-600 t t r r r
20 3o 40 50 60 7o 80

Time (hours)

Figure 9: Sulfate ﬂux in the 15N03 amended cores from each site and sampling
time over the three day incubation time (n=3). Error bars represent 1 SE. of the
mean. The ecosystem types are colored as: streams (black), wetlands (dark
grey), and lakes (light grey).

96

Sulfate ﬂuxes in two of the six sites changed signiﬁcantly over the time
course of the experiments (Arcadia Creek and Wintergreen Lake). Thus, rather
than representing sulfate ﬂuxes as a grand mean of all measurements over the
experimental time course, they are displayed as daily means by site (Figure 9).
These are not compared to the controls because the lack of N03 in the control
cores would have favored sulfate reduction, whereas the N03 added to all of the
treatments would have suppressed sulfate reduction. Sulfate ﬂuxes did not
consistently change in the same direction over time; for example, in Arcadia
Creek 3042' ﬂux decreased over time, whereas in Wintergreen Lake the SO42“

ﬂux increased over time.

Discussion:

Are there differences in N03 processing among aquatic ecosystem types?

The two stream sites had higher rates of N03 removal, denitriﬁcation and
anammox than did the wetland or lake sites (ﬁgures 2, 3, 4). This could be in
part due to their higher ambient N03 load compared to the wetlands and lakes
and thus their higher background N03 removal rates, as illustrated by the
relatively high rates in the control cores (data not shown). Loosestrife Fen (W),
however, had no background N03, but had a N03 removal rate nearly as high
as the stream sites (Figure 2). The denitriﬁcation and anammox rates at that
site, however, were relatively low compared to the other lake and wetland

sediments (Figures 3 and 4). Therefore, high ambient N03 is not necessarily a

97

predictor of higher N03 removal rates, but may be related to higher rates of
denitrification and anammox.

The rates of N03 removal are similar to those measured in other aquatic
ecosystems, both freshwater and marine. These N03 removal rates fall into the
same range as reported by (Kellman 2004) for streams in a similarly human-
dominated landscape (273-1041 pmoles rn'2 hr"). Furthermore, Arcadia Creek
and Bellingham Drain were part of the Lotic lntersite Nitrogen eXperiment 2
(LINX2), which conducted whole stream ”N03 enrichments. The results of that
study showed that N03 removal rates were 120 umoles rn'2 hr" for Arcadia
Creek and 43 umoles rn'2 hr" for Bellingham Drain. The values I report for these
same streams are much higher. Due to logistical constraints, I could only sample
one area of a given ecosystem, and l tended to sample areas that were not
sandy. Thus, the discrepancy in rates may be due to differences of scales or
increased spatial heterogeneity. Sediments collected for my experiments may
represent a relatively “hot” patch within the larger stream matrix, which would
have been more equally represented with the whole-stream ”N03 experiments.

Denitriﬁcation rates in lakes vary considerably, with reported values
ranging from 0.4 - 383 pmoles N rn'2 hr" (Seitzinger 1988). Some of this
variation is driven by differences in nitrogen loads, oxygen levels, and primary
productivity in the waters (Seitzinger 1988). Denitriﬁcation rates in the two lakes
studied here ranged from 50-100 umoles rn'2 hr", with approximately twice as
much denitriﬁcation in the highly eutrophic Wintergreen Lake than in oligotrophic

Lawrence Lake. These denitriﬁcation rates are similar to those found in the

98

sediments from a reservoir draining an agriculturally dominated landscape, which
ranged from 62-225 pmoles rn‘2 hr‘1 (David et al. 2006).

Wetlands have been particularly well-studied with regards to denitrification
because of their oft cited ability to provide the valuable ecosystem service of
nutrient processing and retention, including N03 removal (Zedler 2003).
Wetlands have exceptional variation in denitriﬁcation rates both within a given
system (Poe et al. 2003) and across time within the same site (Smith et al. 2000).
Studies of wetlands receiving agricultural drainage have found denitriﬁcation
rates ranging from 50 — 650 pmoles rn‘2 hr" in a North Carolina constructed
wetland (Poe et al. 2003), 21 - 1414 umoles m'2 hr" in a California constructed
wetland (Smith et al. 2000), and 15 -- 25 umoles m’2 hr" in a minerotrophic fen in
Denmark (Hoffmann et al. 2000). The two sites used in this study are neither
constructed nor restored ecosystems, which may in part explain why the
denitriﬁcation rates were relatively low (50-80 umoles m'2 hr") and relatively
similar between the two sites, as was found by Hoffmann et al. (2000).

The rates of denitriﬁcation measured in this study also generally agree
with those found in meta-analyses across aquatic ecosystems. Pii’la-Ochoa and
Alvarez-Cobelas (2006) reported a mean denitriﬁcation rate of ~350 pmoles m”2
hr" in lakes and ~170 umoles rn‘2 hr" in rivers, though overall, rivers had the
highest rates of denitriﬁcation accompanied by the most variation. While these
rates were of the same order of magnitude, I found that streams had much higher
rates of denitriﬁcation than did lakes (Figure 3). These ﬁndings agree better with

those reported in Seitzinger (1988), who also found that streams generally had

99

higher denitriﬁcation rates (0-345 umoles rn‘2 hr") than lakes (10-171 pmoles m‘2
hr"). The LINX2 studies reported a denitrification rate of 2.38 umoles rn'2 hr" for
Bellingham Drain and 7.8 pmoles rn‘2 hr'1 for Arcadia Creek. The wetland
denitriﬁcation rates reported here also agree approximately with rates reported
for wetlands of intermediate and high disturbance (5—135 umoles N m'2 hr")
(Seitzinger 1994).

It is more difficult to compare my estimated DNRA rates to other studies in
similar ecosystems because DNRA is not measured as frequently as
denitrification. An additional complication is that DNRA is reported as both rates
and as percentages of nitrate removal. The rates of DNRA measured here are in
the same range and slightly higher (13 — 99 umoles m'2 hr"; Table 3) compared
to those measured in the estuaries studied by Gardner et al. (2006) (3 — 50
umoles rn'2 hr"; Table 3). However, the rates I measured were much lower than
those measured in the marine-inﬂuenced Ringﬁeld Marsh (1370 - 4230 umol rn'2
hr") (Tobias et al. 2001) and the rates measured in estuarine sediments (800 -

50,000 umol m'2 hr") (Koike and Sorensen 1988).

Most studies of anammox have been done in marine ecosystems. The
only study performed in freshwaters was by Schubert et al. (2006) in Lake
Tanganyika. Schubert et al. measured rates of anammox in the anoxic water
column (>100 m depth) of up to 10 nM N2 hr", which are much lower than the
absolute rates of anammox measured in this study. Dalsgaard et al. (2005) also
found higher anammox rates but lower relative importance of anammox in overall

N03 removal when comparing near-surface anammox measurements to deep

100

ocean measurements. Although the Dalsgaard et al. (2005) synthesis was
reported results in a per volume unit (nmol cm'3 hr") and my rates were
measured on an areal basis (umoles m'2 hr"), if we assume that the active depth
of sediment in the cores is 2 cm, then the rates of annamox in these sediments
translate to 10-50 nmoles cm’3 hr" (300 nmoles cm"3 hr" in Bellingham), which
are considerably higher than those reported from various studies in the review by
Dalsgaard et al. (2005), which tended to range from 1-10 nmoles cm'3 hr".
Dalsgaard et al. (2005) suggested that water depth was a key driver in the
relative importance of anammox compared to denitrification, wherein at deeper
depths with lower mineralization (and C) anammox bacteria can compete
effectively with denitrifers. My sites seem to also ﬁt this pattern for near-surface
environments, with relatively little N03 consumed by Anammox.

These rates may be artificially high due to an enrichment or fertilization
effect created by increasing the concentration of N03 in the overlying water.
Adding N03 to the cores, even when some background N03 was present,
increased N03 removal rates in some sites (e.g., in Arcadia Creek; data not
shown) but not all. This could explain some of the disparity in the rates of N03
removal and denitriﬁcation found in the LlNX2 experiments compared to my
results. The increase in N03 availability in these experiments, however, was well
within the range of what these ecosystems might normally encounter from either
agricultural run-off or high nitrate groundwater. Thus, the rates measured here

should be well within realistic rates for these ecosystems.

101

Hypotheses to explain the missing tracer 15N

These experiments did not fully quantify all of the N end products, as
indicated by the substantial fraction of N03 removal that could not be accounted
for by measuring the end-products accounted for in this study. I was able to 3
account for roughly half in most cases, which is comparable to other studies
where one or more nitrate removal processes have been measured (Seitzinger
1988). The unaccountable ”N from N03 could be explained by a number of
possibilities: 1) underestimated DNRA rates (discussed below) due to NH4+
sorption or exchange with bound NH4“ pools; 2) microbial storage of N03, as has
been found in some strains of nitrate-reducing bacteria (Kamp et al. 2006); and
3) other processes not measured, including assimilative uptake for biomass
incorporation, although microbial use of N03 seems unlikely given available NH4+
and for the same reason N ﬁxation would seem unlikely in the sediment
environment.

Denitriﬁcation and anammox result in dissolved gaseous nitrogen (N2),
which can readily be quantiﬁed with careful sample collection and handling
because it behaves conservatively in pore waters. On the other hand, NH4“ in
sediment pore waters is known to be in equilibrium with the ion exchange
complex. Exchange of dissolved NH4“ with a large sorbed reservoir would result
in an underestimation of the tracer ”N in NH4+ and thus an underestimate of
DNRA rates because I sampled water ﬂowing over the sediments. The

importance of this sorption remains to be investigated.

102

What is the relative importance of various N removal processes?

Differences in measurement methods and rate units make it difficult to
compare processes across different ecosystems and studies. To examine the
relative importance of the various processes in this study compared to a number
of other published reports, I have taken a ratio of the rates of denitriﬁcation to
either DNRA (Table 3) or anammox (Table 4). The actual rates reported in each
study are reported in the tables, along with their respective units (as footnotes),
and the ratio is a unitless number wherein a value less than one indicates that
the alternative process is more important than denitriﬁcation, and greater than
one indicates that denitriﬁcation is more important.

DNRA accounted for a signiﬁcant proportion of nitrate removal in the six
sites used in this study, although denitriﬁcation was the dominant nitrate removal
process (Table 3). The exception to this was in streams where denitriﬁcation
was much more important than DNRA. The denitriﬁcationzDNRA ratios found in
these sites are similar to other sites, both marine and freshwater, where both
processes have been measured. This is particularly interesting because DNRA
has been shown to be an important component of the nitrogen cycle in marine
systems, but has received less study in freshwater systems. However, data in
Table 3 indicate that DNRA can be quite important relative to denitriﬁcation in
many freshwater sites (Bowden 1986, Storey et al. 2004, McCarthy et al. 2007a,
McCarthy et al. 2007b). This comparison demonstrates that the relative
importance of the two pathways is similar in both freshwater and marine

ecosystems. Also, in many cases these DNRA estimates are based on isotope

103

tracer studies that potentially suffer from the same problem of sorption in the
sediments that was discussed above, and thus may underestimate true rates.
Anammox tended to be much less important to N03 processing than
denitriﬁcation in the six sites used in this study. The relative importance of the
two processes has received a great deal of study in marine and oceanic
ecosystems, but little is known about anammox in freshwaters. Thus, there are
fewer available estimates of denitriﬁcationzanammox ratios (Table 4), and only
one freshwater study for comparison (Schubert 2006). Generally, the relative
importance of anammox to denitriﬁcation is much lower in these sites than has
been measured in other sites (Trimmer et al. 2003, Engstrom et al. 2005), but is
similar in range to ratios from other freshwater sites (Trimmer et al. 2003, site 6;
McCarthy and Gardner unpublished data). The exception to this is Bellingham
Drain, where anammox accounted for approximately 10% as much N03 removal
as did denitriﬁcation, making the denitriﬁcationzanammox ratio more similar to

that of the other studies.

Conclusions

In this study, I measured rates of nitrate removal from six freshwater
ecosystems and partitioned the nitrate removal end-products to N2, indicating
denitriﬁcation activity or NH4" indicating DNRA. Additionally, by using the isotope
pairing method, I estimated rates of anammox, which could also be indirectly
reliant on nitrate reduction. Denitriﬁcation was an important pathway of nitrate

removal in all six ecosystems. However, in certain sites, DNRA could be as

104

important as denitrification to overall nitrate removal. Anammox was not an
important pathway of nitrate removal in any of the six sites. The rates of
denitriﬁcation and DNRA were comparable to those estimated from other
freshwater and marine sites where both processes have been measured.
Anammox rates are more difﬁcult to compare to other studies due to differing
methodologies, but the anammoxzdenitriﬁcation ratio is much higher in these
sites than as been measured from other (mostly marine) ecosystems, indicating
that anammox is relatively unimportant to nitrate removal in near-surface

freshwater sediments.

105

080.com. 80.0.0925. wEmh. 3.5.05.9: 000 5000500000 000 30.050... .00... .0000m 00050.0. >mo.00m. 0000.2
ﬁlms: ”0000030000 .0E00.. ...... ts b0 :0 z 00 C ...... P..00E.000 m z 00 0.... ...... 0E0 .0E0 C ...... .... .0E: A».
...... N.... 00.00.... € 0.50:3 00 0.0 00.030 00.0.0.0 003 E00 3.03 .E=.00EE0 0. 0000000. 30...: 030......000 20.20.
00.000.00.000 .000. 030.0003 0.00.0 000.3000 Cm$ 00.0.5000... 9i. .000...0.>0.00< .00.:000E >_0:00003.:E.0
0000 0>00 £00 0.002. 00.030 0. $20 0. 00.300.003.000 .0 00000000.. 02.0.0. 00. .0 000000E00 < .0 0.00...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00..
0.0 ...m .....0 .0.0.>> 0030m. 00.00 22 meow ..0 .0 000.00 0.005. 00:00..
0.0..
mdm 0: +0. 5 .0EE3m 0030M. 00.00 22 meow ..0 .0 000.00 00.0.. 00.00w
. 0.3 >00
0.0 ..md 1N3 .0.0.>> 0030M 00.00 20.. meow ..0 .0 000.00 00.000305. .00m.
0.3 >00
«.0. .m. 5 ..mdo .0EEzw 0030.“. 00.00 20. ooom ..0 .0 000.00 00.20 030.00
0.00 050.2
0.0 Rd N. E .0EE0m .0030m. 00.00 20. meow ..0 .0 000.00 .020. 000032
00.0.2 . mums.
of. be (0.: :0... 000.005 z...F w ...m.< room ..0 .0 00.000 00.05. 0.0.50.0.
00.05. mums.
06 (TN .03 000% 000.005 20. w .50. room ..0 .0 00.000 00.0.2 0.0.0605.
00.0.. 00.0..
0.? .00 .me .0EE3m 0.00033. 00.00 20. .030 0.0. 00009000.;
0:0. ZN emv .0EE3m 00.0.. 00.00 20. 030 0.0. 00.0.. 0000030..
...0 0.03 0.0 .0EEsm 00.0.). >>u. 00.00 20e .030 0.0. 00.05. >903...
..._. 0N0 ..om .0EE3m 00h. >>m 00.00 20. >030 0.0. 00“. 03300000..
0.05
0.00 .R ER .0EE0m E000m 00.00 20. >030 0.0. E0000...0m
«.0 .00 tom .0EEzm E000m 00.00 2%.. 030 0.00 0.00.0 0.000..<
<mzo 000.. 030.. 0%:
.000 <mzo .000 00000w E0u0>000m 0050.... 000000001 35

 

 

106

0030..000 um 0.00...

 

 

 

 

 

 

 

 

 

000.00
000.0200: 0.0.E0000000.m .0020.
«.0 ..MN ..Mb 402 00000.”. 00.00 22 voow ..0 .0 >0.0.m 00000
000.00
002030300 0.0.0.00000005 0020.
0.. km. .00.. $2 00000.00. 00.00 20F .38 ..0 .0 0.05 00000
000 .020.
00.0.. . 0520.00.00. 00.00
0.. .000 .80 :0. 000950 00.8 2.2 0800 ..0 .0 200022 50.0.. 00.0.
0..0 00.0.. 0.0.2 00.00
00.0. 0.00.0.00.0>I 50.0... 00.0..
how .0 .o...‘ ..0“. 0.000..3m. 00.00 2.2 038 ..0 .0 >00000.2
..0000_0.. 00.0000m. 0...m 00.0..
I 000..0>> 000.0. .005 ... 0.00.0
0. .8 Bow .0rEE3m 0030M >>n. 00.00 20. 0.88 ..0 .0 30000.2 00.003 0.0
..0000.0.. 00.0000”. 00m 00.0..
.. 0.30.2 000.0. .005 .0 000.0
..0 .00. .000 0.05:0 00300 .5. 00.8 22. 0800 ._0 .0 2.00022 00:62. 0.0
00.0.2 20F 00.0.2
0... .0... ..0 S2 :00... 2,. .0 E< .6280 000. 000260 .020 0.32
<mzo 0.0.. 0.0. 0m».
"000 (120 .00n 000000 E0.0>000w 000.05. 02.0.0.0... 0....0

 

 

 

 

 

 

 

 

 

107

.209 mo_EEooEmoo “m «EEEoomo

ucm $965922 .mEmEcoLScm 22m: (620592.). EcmEcESCm Ucm um=aa< 22w<v .mcozmSanm 6:52,

.1... 5.: A v.1; wumm “m2, _E Nz BE: A .715 -._ .95 C .c N.E 3.9:: € .mao__£ mm 2m mmﬁam EEmtE 9: E0:
«:5 .cozmocEEmu Ewe wagccoﬂ 35mg mmoﬁﬁ Pa: 553:3: 9?: .mco=m_>m._nn< .uwSmmmE zwaomcmgsﬁm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

cmwn m>mc £02 9ch 863m 5 onEmcm 9 cozmoEzEmv yo mocmtoag 9:29 9: B cowtquoo < .v gawk
E <00 A 5V 58%
QN "we luv ES. 580 ESE ZE meow ._m 6 Eobmmcm .xmtmmmxm
2.3
kn: <00 Econ .2825
m6 Hod H3 >32 £300 356 22 moom ._m 6 Eobwmcw .xwtwmmxm
vém Ba Ba LmEEzm 9:3 880 23 >25 95 9.3 599853
9mm 1.; 3v _mEE:m 0x3 @900 25 33m $5 «in... 82226..
new F: 3v 82:25 £292 >>m $50 29 >33 mg 5222 >wx5h
mém RN Em LmEEsm cm“. >>n_ $50 29 >95 25 cw“. 85880..
_‘.o_‘ ZNN WEN .mEEzm Emmbw 988 2B 33m m_£ ESQ Emcm5=wm
Now Be tom LwEEzm Emmbm 8.8 23 >035 £5 x020 «58.6.
on. ,
E24. 32 3a.. on? _
V :50 onEmc< .cwo cammww Emum>moom .3522 ooceeom 95 L.

 

108

Emacscoo .v 2ka

 

9% 85233::

cse swam .223

 

 

 

 

 

 

 

 

 

Qua 5.3 %$ .355 9.3 £50 23 .828 new 3:822 25 8.3
98 85:93:: as 58m .228
2: BE EN .235 $3 850 22 5:23 95 3:80.). 25 9.3
2m E. 2:
nm a: ““8 22, 33 E. 2,: 88 ..m a 59:65 micamcﬂ 9.3
b”: E: 6 SE baamm
3: ho .mm .52 $35 ram 22 88 ._m 6 .5555 32m moEaE
E. .52 a 23
92 .2 .09 32 53m. him 22 meow ._m 6 5255 2E .99. 8&9;
E <00 Emv ES
9: Had “or 5:93 .5500 555 22 moow .5 5 Eobmmcm .ucsom 2.2m. mac;
E <00 83 Eva
adv ”we Hm? 5:93 5500 ESE 29 moom ._m 5 Eobmmcw .ccsom 25.2 9.0..
on 2a.. 5&3
-Ew:< x9: 38 E89?
"coo .Em:< ..50 commwm .08. 30555. 505551 95 L

 

 

 

 

 

 

 

 

 

109

References:

Aminot, A., D. S. Kirkwood, and R. Kerouel. 1997. Determination of ammonia in
seawater by the indophenol-blue method: Evaluation of the ICES NUTS
l/C 5 questionnaire. Marine Chemistry 56:59-75.

An, S. M., and W. S. Gardner. 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). Marine Ecology-Progress
Series 237141-50.

Bonin, P. 1996. Anaerobic nitrate reduction to ammonium in two strains isolated
from costal marine sediment: A dissimilatory pathway. FEMS Microbiology
Ecology 19:27-38.

Bowden, W. B. 1986. Nitrification, nitrate reduction, and nitrogen immobilization
in a tidal freshwater marsh sediment. Ecology 67:88-99.

Brunet, R. C., and L. J. Garcia-gil. 1996. Sulﬁde-induced dissimilatory nitrate
reduction to ammonia in anaerobic freshwater sediments. FEMS
Microbiology Ecology 21:131-138.

Burgin, A. J., and S. K. Hamilton. 2007. Have we overemphasized the role of
denitriﬁcation in aquatic ecosystems? A review of nitrate removal
pathways. Frontiers in Ecology and the Environment 5:89-96.

Comwell, J. C., W. M. Kemp, and T. M. Kana. 1999. Denitriﬁcation in coastal
ecosystems: methods, environmental controls, and ecosystem level
controls, a review. Aquatic Ecology 33:41-54.

Dalsgaard, T., B. Thamdrup, and D. E. Canﬁeld. 2005. Anaerobic ammonium
oxidation (anammox) in the marine environment. Research in Microbiology
156:457-464.

David, M. B., L. G. Wall, T. V. Royer, and J. L. Tank. 2006. Denitriﬁcation and the
nitrogen budget of a reservoir in an agricultural landscape. Ecological
Applications 16:2177-2190.

Engstrom, P., T. Dalsgaard, S. Hulth, and R. C. Aller. 2005. Anaerobic
ammonium oxidation by nitrite (anammox): Implications for N2 production

in coastal marine sediments. Geochimica et Cosmochimica Acta 69:2057-
2065.

110

Fossing, H., V. A. Gallardo, B. B. Jorgensen, M. Huttel, L. P. Nielsen, H. Schulz,
D. E. Canfield, S. Forster, R. N. Glud, J. K. Gundersen, J. Kuver, N. B.
Ramsing, A. Teske, B. Thamdrup, and O. Ulloa. 1995. Concentration and
transport of nitrate by the mat-forming sulfur bacterium Thioploca. Nature
374:713-715.

Gardner, W. S., L. R. Herche, P. A. St. John, and S. P. Seitzin er. 1991. High-
Performance Liquid Chromatographic determination of NH4:[”NH4 +

15NH4] ion ratios in seawater for isotope dilution experiments. Analytical
Chemistry 63: 1 838-1 843.

Gardner, W. S., M. J. McCarthy, S. M. An, D. Sobolev, K. S. Sell, and D. Brock.
2006. Nitrogen ﬁxation and dissimilatory nitrate reduction to ammonium

(DNRA) support nitrogen dynamics in Texas estuaries. Limnology and
Oceanography 51 :558-568.

Gardner, W. 8., P. A. StJohn, C. Evans, and J. F. Cavaletto. 1996. HPLC
retention-time-shift determination of nitrogen isotope ratios in enriched
water. American Laboratory 28:017.

Golterman, H. L. 2004. The chemistry of phosphate and nitrogen compounds in
sediments. Kluwer Academic Publishers, London.

Hedin, L. 0., J. C. von Fischer, N. E. Ostrom, B. P. Kennedy, M. G. Brown, and
G. P. Robertson. 1998. Thermodynamic constraints on nitrogen

transformations and other biogeochemical processes at soil-stream
interfaces. Ecology 79:684-703.

Hoffmann, C. C., S. Rysgaard, and P. Berg. 2000. Denitriﬁcation rates predicted
by nitrogen-15 labeled nitrate microcosm studies, in situ measurements,
and modeling. Journal of Environmental Quality 29:2020-2028.

Holmes, R. M., J. W. McClelland, D. M. Sigman, B. Fry, and B. J. Peterson.
1998. Measuring N-15NH4+ in marine, estuarine and fresh waters: An
adaptation of the ammonia diffusion method for samples with low
ammonium concentrations. Marine Chemistry 60:235-243.

Jetten, M. S. M., M. Strous, K. T. van de Pas-Schoonen, J. Schalk, U. G. J. M.
van Dongen, A. A. van de Graaf, S. Logemann, G. Muyzer, M. C. M. van
Loosdrecht, and J. G. Kuenen. 1998. The anaerobic oxidation of
ammonium. Ferns Microbiology Reviews 22:421-437.

Kamp, A., P. Stief, and H. N. Schulz-Vogt. 2006. Anaerobic sulﬁde oxidation with

nitrate by a freshwater Beggiatoa enrichment culture. Applied and
Environmental Microbiology 72:4755—4760.

111

Kellman, L. 2004. Nitrate removal in a first-order stream: reconciling laboratory
and ﬁeld measurements. Biogeochemistry 71:89-105.

Knowles, R. 1982. Denitriﬁcation. Microbiological Reviews 46:43-70.

Koike, l., and J. Sorensen. 1988. Nitrate reduction and denitriﬁcation in marine
sediments. Pages 251-270 in T. H. Blackburn and J. Sorensen, editors.
Nitrogen cycling in the coastal marine environments. John Wiley & Sons,
London.

McCarthy, M. J., W. S. Gardner, P. J. Lavrentyev, K. M. Moats, F. J. Jochem,
and D. M. Klarer. 2007a. Effects of hydrological ﬂow regime on sediment-
water interface and water column nitrogen dynamics in a Great Lakes
coastal wetland (Old Woman Creek, Lake Erie). Journal of Great Lakes
Research 33:219-231.

McCarthy, M. J., P. J. Lavrentyev, L. Yang, L. Zhang, Y. Chen, B. Qin, and W. S.
Gardner. 2007b. Nitrogen dynamics and microbial food web structure
during a summer cyanobacterial bloom in a subtropical, shallow, well-
mixed, eutrophic lake (Lake Taihu, China). Hydrobiologia 581:195-207.

Megonigal, J., M. Hines, and P. Visscher. 2004. Anaerobic Metabolism: Linkages
to Trace Gases and Aerobic Processes. Pages 317-424 in W. H.
Schlesinger, editor. Biogeochemistry. Elsevier-Pergamon, Oxford, UK.

Paerl, H. W., R. L. Dennis, and D. R. Whitall. 2002. Atmospheric deposition of
nitrogen: Implications for nutrient over-enrichment of coastal waters.
Estuaries 25:677-693.

Peterjohn, W. T., and D. L. Correll. 1984. Nutrient dynamics in an agricultural
watershed - observations on the role of a riparian forest. Ecology 65:1466-
1475.

Peterson, B. J., W. M. Wollheim, P. J. Mulholland, J. R. Webster, J. L. Meyer, J.
L. Tank, E. Marti, W. B. Bowden, H. M. Valett, A. E. Hershey, W. H.
McDowell, W. K. Dodds, S. K. Hamilton, S. Gregory, and D. D. Morrall.
2001. Control of nitrogen export from watersheds by headwater streams.
Science 292:86-90.

Pina-Ochoa, E., and M. Alvarez-Cobelas. 2006. Denitriﬁcation in aquatic
environments: A cross-system analysis. Biogeochemistry 81:111-130.

Poe, A. C., M. F. Pichler, S. P. Thompson, and H. W. Paerl. 2003. Denitriﬁcation

in a constructed wetland receiving agricultural runoff. Wetlands 23:817-
826.

112

Rabalais, N. N., R. E. Turner, and W. J. Wiseman. 2001. Hypoxia in the Gulf of
Mexico. Journal of Environmental Quality 30:320-329.

Rysgaard, S., and R. N. Glud. 2004. Anaerobic N2 production in Arctic sea ice.
Limnology and Oceanography 49286-94.

Rysgaard, S., R. N. Glud, N. Risgaard-Petersen, and T. Dalsgaard. 2004.
Denitriﬁcation and anammox activity in Arctic marine sediments.
Limnology and Oceanography 49:1493-1502.

Schubert, C. J., E. Durisch-Kaiser, B. Wehrli, B. Thamdrup, P. Lam, and M. M.
M. Kuypers. 2006. Anaerobic ammonium oxidation in a tropical freshwater
system (Lake Tanganyika). Environmental Microbiology 8:1857-1863.

Seitzinger, S. 1988. Denitriﬁcation in freshwater and coastal marine ecosystems:
ecological and geochemical signiﬁcance. Limnology and Oceanography
332702-724.

Seitzinger, S. P. 1994. Linkages between organic matter mineralization and
denitriﬁcation in 8 riparian wetlands. Biogeochemistry 25:19-39.

Smith, L. K., J. J. Sartoris, J. S. Thullen, and D. C. Andersen. 2000. Investigation
of denitriﬁcation rates in an ammonia-dominated constructed wastewater-
treatment wetland. Wetlands 20:684-696.

Storey, R. G., D. D. Williams, and R. R. Fulthorpe. 2004. Nitrogen processing in
the hyporheic zone of a pastoral stream. Biogeochemistry 69:285-313.

Tiedje, J. M. 1988. Ecology of denitriﬁcation and dissimilatory nitrate reduction to
ammonium. Pages 179-244 in A. J. B. Zehnder, editor. Biology of
Anaerobic Microorganisms. Wiley & Sons, New York.

Tobias, C. R., I. C. Anderson, E. A. Canuel, and S. A. Macko. 2001. Nitrogen
cycling through a fringing marsh-aquifer ecotone. Marine Ecology-
Progress Series 210:25-39.

Trimmer, M., J. C. Nicholls, and B. Deﬂandre. 2003. Anaerobic ammonium
oxidation measured in sediments along the Thames estuary, United
Kingdom. Applied and Environmental Microbiology 69:6447-6454.

Wetzel, R. G. 2001. Limnology: Lake and River Ecosystems, 3 edition. Academic
Press, San Diego.

Whitmire, S. L., and S. K. Hamilton. 2005. Rapid removal of nitrate and sulfate in

freshwater wetland sediments. Journal of Environmental Quality 3422062-
2071.

113

Zedler, J. B. 2003. Wetlands at your service: reducing impacts of agriculture at
the watershed scale. Frontiers in Ecology and the Environment 1265-72.

114

CHAPTER 4

ORGANIC CARBON AND SULFIDE AS CONTROLS ON NITRATE REMOVAL
AND NITRATE REDUCTION END-PRODUCTS IN FRESHWATER SEDIMENTS
Introduction:

Nitrogen (N), as a fundamental constituent of biomass, is critically
important to ecosystem productivity because it is often a growth-limiting nutrient.
Human activities have approximately doubled the amount of reactive nitrogen
globally (Vitousek et al. 1997). This anthropogenic perturbation has led to
increased loading of nitrate (N03) to many surface- and ground-water
ecosystems. As nitrate-rich water passes through or over freshwater sediments,
the concentration of nitrate typically decreases. The prevailing scientiﬁc belief is
that this removal is largely due to either assimilation into microbial, algal or plant
biomass, or to di-nitrogen (N2) via bacterial respiratory denitriﬁcation. While
denitriﬁcation has been intensively examined, few studies have investigated
other known processes that could directly compete with denitriﬁcation and may
well be as important for overall nitrate removal (Megonigal et al. 2004, Burgin
and Hamilton 2007).

We are only starting to understand the complexity of microbial N cycling,
especially with regard to factors that control the relative importance of multiple
potential pathways. With regard to N03' utilization by bacteria, multiple pathways
are known to simultaneously occur, including respiratory denitriﬁcation and
dissimilatory nitrate reduction to ammonium (Burgin and Hamilton 2007).

Availability of free sulﬁde (hereafter referred to has H28), NOg', and organic

115

carbon (OC) have all been put forth as potential controls on one or both of these
processes (Tiedje et al. 1982, Tiedje 1988, Brunet and Garcia-gil 1996).

In respiratory denitrification, nitrate is used in the terminal oxidation of
organic matter under anaerobic conditions; most of the nitrate is transformed to
N2, a biologically unavailable form of N. The enzymatic sequence in nitrate
reduction is well studied (Paul and Clark 1996); these enzymes can be inhibited
by 02 and low pH. Incomplete reduction can result in the accumulation of
intermediates including nitrite (N02) and nitrous oxide (N20). Therefore,
favorable conditions for respiratory denitriﬁcation include anoxia, N05, and labile
OC.

In contrast to denitriﬁcation, DNRA is a relatively understudied N03'
removal pathway. Two forms of DNRA are known to occur: fermentative DNRA,
thought to occur under conditions of high labile carbon availability (Tiedje et al.
1988), and sulfur-driven DNRA (Brunet and Garcia-gil 1996, Otte et al. 1999),
thought to occur wherever nitrate and reduced sulfur compounds coincide.
Fermentative DNRA couples electron ﬂow from organic matter via fermentation
reactions to the reduction of N03' (Tiedje 1988, Megonigal et al. 2004). Tiedje
(1988) suggested that heterotrophic DNRA would be most important in highly
reducing environments that maintain anoxic conditions for long time periods.
Although the conditions promoting fermentative DNRA and respiratory
denitriﬁcation are similar (anoxia, available N03; and labile organic matter),
fermentative DNRA is thought to be favored in NOg'-Iimited, labile-carbon rich

environments while respiratory denitriﬁcation would be favored under relatively

116

carbon-limited conditions (Kelso et al. 1997, Silver et al. 2001). A few studies
have supported this hypothesis (Nijburg et al. 1997, Bonin et al. 1998,
Christensen et al. 2000), but none to our knowledge has directly tested it by
manipulating C:N ratios.

Recent work in marine and freshwater systems has demonstrated that
certain S-oxidizing bacteria can use N03' to oxidize H28 and elemental S to
sulfate (8042') (Fossing et al. 1995, Brunet and Garcia-gil 1996). These S-
oxidizing bacteria can reduce nitrate to either N2 or ammonium (NH4")
(Dannenberg et al. 1992, Brunet and Garcia-gil 1996, Otte et al. 1999). The
predominant fate of the reduced N (NH4+ vs. N2) may be determined by the
ambient concentration of H28, which is known to inhibit denitriﬁcation (Brunet and
Garcia-Gil 1996). High ambient H28 can inhibit the ﬁnal two reduction steps of
the denitriﬁcation sequence, in which case the sulﬁde can be oxidized to
elemental S or 8042' with a simultaneous reduction of N05 to NH4+. On the
other hand, metal-bound sulﬁdes such as FeS also can be oxidized by these
bacteria, but do not show the enzymatic inhibition of denitriﬁcation (Brunet and
Garcia-Gil 1996), and these often are abundant constituents of freshwater
sediments (Holmer and Storkholm 2001). Therefore, the importance of sulfur-
driven nitrate reduction in a given site may be regulated by the availability of
electron donors (H28, Thiosulfate, elemental S) as well as the ambient

concentration of H28, which may inhibit key denitriﬁcation enzymes.

In this study, I investigated the relative importance of labile OC and H28

as controls on N03' removal and its end-products (indicative of dominant

117

pathways) in a laboratory setting using anoxic wetland sediments. I selected
sediments from a high ambient H28 site and a low ambient H28 site, assuming
that they would have different microbial communities which might differentially
utilize the N05. My goals were: 1) to examine the relative importance of OC and
H28 in regulating N2 and NH4+ production from NOa', and 2) to investigate the
time course of N transformations after additions of OC and H28. 1 have no a
priori reason to believe either 00 or H28 will be a more important regulator of
nitrate removal and the end-products; the two are not necessarily mutually
exclusive controls on nitrate processing. If H28 controls DNRA, I would expect to
see increasing NH4+ production with increasing added H28. Additionally, if H28
inhibits denitriﬁcation enzymes, I would expect to see a decrease in N2
production with increasing H28. If OC is an important regulator of DNRA, I would

expect to see increasing NH4+— production with increasing OC.

Methods:
Site selection:

To compare N03' transformations by bacterial communities charactersitic
of different levels of ambient H28 and organic carbon availability, sediment
samples were collected from a high H28 site (Loosestrife Fen, also called
Loosestrife Pond; LP) and a low H28 site (Windmill Pond; WP). Loosestrife Fen
is a small groundwater-fed fen at the Experimental Forest of the Kellogg
Biological Station (KBS). Groundwater from a spring enters the site and drives
ﬂow across the wetland, which has a residence time of ~24-48 hours. This site

tends to have high sediment porewater HZS concentrations; near-surface

118

porewater often contains >150 uM H28 (Whitmire 2003). Windmill Pond (WP) is
a shallow pond along and connected to Gull Lake at KBS, which in combination
with groundwater inputs provides a source of water for the site. This site has an
organic layer of sediment overlying a sandy bottom layer. Windmill Pond
sediments have sand colored black by iron sulﬁdes, but concentrations of free
sulﬁde tend to be low in the near surface sediment porewater (<10 uM H28:
Burgin, unpublished data). Neither site had detectable reduced iron (Fey) in the

porewaters.

Assay procedure:

Two types of assays were conducted using sediments from these sites. In
the ﬁrst set, which will be referred to as the “gradient” assays, l subjected the
sediments to levels of labile organic carbon (OC) and free sulﬁde (H28) in a full
factorial experimental design, which also included controls wherein N03" but not
OC nor H28 were added (NOg’-only treatment). On day 1 and day 3 of the
experiment, sediments were collected from each site, brought back to the lab and
gently mixed, then allowed to sit for at least 24 hours to return to anaerobic
conditions. To conduct the assays, 10 mL of sediments and 20 mL of site
surface water were placed into a 40 mL vial and capped with a silicon septum.
Five replicate vials of each treatment were sparged with He for 20 min. After,
sparging, the given treatment assignment of labile OC [as sodium acetate, NaAc;
high = 10 mg C/L (46.7 pmoles), med = 5 mg C/L (23.3 pmoles), Iow = 1 mg C/L

(2.3 pmoles)] and sulfide [as NaZS; high = ~200 pM (7-15 umoles), med = ~100

119

uM (2.5-5 pmoles), low = ~20 uM (0.5-1 umoles)] or a combination of the two
were added to the anoxic sediments. The sulfide solution was made and added
to vials in a glove bag purged with high-purity He immediately prior to the
experiment. All OC and N03' solutions were also prepared fresh for each
experiment. All sediments except for controls received a N03‘ amendment of
14.3 umoles to yield a ﬁnal concentration of 10 mg N/L or 0.7 mM. Controls
without any added C, H28 or N03' were also prepared to ensure that air leakage
was not a signiﬁcant source of 02 and N2. Sediment vials were then incubated
for approximately 24 hours with the caps and septa underwater to minimize air
contamination.

After 24 hr of incubation, vials were destructively analyzed as follows: 3
mL of He-sparged water were injected into the bottom of each vial to displace 3
mL of vial headspace into a Shimadzu gas chromatograph containing a Porapak-
Q® packed column and a thermal conductivity detector for Oz+Ar and N2 ,
quantiﬁcation. Tests showed that N2 and Oz contamination via air entry from this
procedure was insigniﬁcant. Another 4 mL of water was used to displace more
headspace into a He-ﬁlled exetainer (Labco®) for N20, CH4 and C02
quantiﬁcation via gas chromatography (electron capture, ﬂame ionization, and
infrared gas detectors). The vial cap was then removed and a 10-mL water
sample was immediately ﬁltered (0.45 pm membrane) for H28 analysis by the
colorimetric method of Golterman and Clymo (1969). Another 5 mL was ﬁltered
(0.45 pm) for analysis of NH4“, NOa', S042) and N02' via membrane-suppression

ion chromatography (Dionex). A ﬁnal 3 mL subsample was ﬁltered (Millex sterile

120

0.22 pm) and frozen for future acetate analysis. All other samples (e.g., gas and
water chemistry) were analyzed as soon as they were generated.

The same technique was used to examine the time course of N03’
transformations in sediment samples from the same two sites; these will be
called “time-course assays”. Due to logistical constraints for these experiments,
only one level of OC (as acetate; 23.3 pmoles for a ﬁnal concentration of 415
pM) and of H28 (~1.5 pmoles for a ﬁnal concentration of 55 uM) were added to
the sediments in a full-factorial design. Fifteen replicates of each of the five
treatments (control; N03“ only; H28 + NOg'; OC + N05; 00 + H28 + NOg‘) were
started at approximately the same time, and were destructively harvested at time
points chosen based on previous experiments. LP was sampled 12, 24 and 48
hrs after the start of the experiment; for WP sampling took place at 6, 12 and 24
hrs because its sediments took up N03’ more rapidly.

All data were corrected for dilution and background concentrations; graphs
indicate net ﬂuxes wherein a positive value denotes production and a negative
value is removal. The gradient assays were analyzed by two-way analysis of
variance (ANOVA) with levels of carbon and sulﬁde as ﬁxed factors. Because
factors and interactions were often signiﬁcant, the magnitude of the effect (1.02;
effect size) of each factor (treatment) was also calculated. The effect size is
based on estimating the variance in a response variable that can be explained by
the factor, and then relates that fraction of the variance to the total variance in a
response variable (Graham and Edwards 2001). The effect size is not directly

dependent on the sample size, and it does not necessarily covary with statistical

121

signiﬁcance (p values) (Graham and Edwards 2001). Effect sizes are calculated
as described by Graham and Edwards (2001) for a two-way ANOVA of ﬁxed
factors. The variance of each factor is related to the overall variance, and the
resulting ﬁgure is termed the “variance component” (Tables 1 and 2). The
variance component for a factor is then related to the overall variance (factors +
interactions + error), and multiplied by 100 to reﬂect the percent of variance
attributable to a factor, or the magnitude of that factor’s effect ((02) (Tables 1 and
2).

The time-course assays were analyzed by a two-way ANOVA at each
individual time point (3 time points per experiment). Contrast statements were
used to compare the effects of various treatments of interest. Statistical analysis
was performed in SAS using PROC MIXED. When appropriate, I used the
LSMEANS/PDIFF procedure to make a priori pairwise comparisons. In both

experiments, ﬁve replicates of each treatment were run.

Results:
OC and H28 Gradient Assays

Sediments from both sites removed a large proportion of the added 14.3
pmoles of nitrate within 24 hrs (Figure 1). LP sediments removed ~66% of the
added nitrate during the 24 hr incubation, whereas the WP sediments removed
nearly all added N031 In the LP sediments, stimulation of N03' removal by
added H28 was dependent on OC availability, as indicated by the signiﬁcant

interaction between the treatments (Table 1). At low to intermediate levels of

122

H28, N03' removal was stimulated by the presence of 00, but at high H28 levels,
OC inhibited N03' removal (Figure 1). The H28 treatment and the interaction of
H28 and OC had a larger magnitude of effect on N03' removal than did 00 alone
(Table 1). In the WP gradient experiments, I could not test the effects of OC nor
H28 on overall N03' removal because nearly all of the added nitrate was gone by
the end of the 24-hour incubation period, precluding estimation of the removal
rate.

OC and H28 additions caused NH4+ production in the'presence of added
NO3' in Sediments from the two sites (Figure 2). In LP, while there was an
increasein NH4+ production with addition of either H28 or 00 (~0.5-1 umole), the
greatest NH4+ accumulation occurred when the two were added together (up to
3.5 pmoles). H28 and the interaction of OC and H28 had much larger effect on
NH4+ ﬂux than did OC alone (Table 1). WP sediments showed net increases of
0-6 pmoles of NH4"; the largest change occurred by adding the smallest amount
of H28 in combination with OC, with the most NH4* production occurring at the
lowest levels of OC and H23 combined (Figure 2). In contrast to LP, no change
in NH4+ ﬂux was seen across the H2S- or OC-only gradients. Again, in WP as in
LP, H23 had a larger magnitude of effect than did OC, though the difference
between the two effect sizes was much smaller than was the case in LP.

N2 was produced in the sediments from both sites, but at rates up to 3
times greater in WP than in LP (Figure 3). In LP, both OC and H28 signiﬁcantly
increased N2 production with no interaction between the treatments,and H23 had

a greater magnitude of effect on N2 production than did OC (Table 1). Adding

123

H28 to 0.5 and 2.5 umoles increased N2 production across all OC additions, but
the increase did not extend to the 5.5 umoles H28 addition. In the WP
sediments, N2 production was dependent on the interaction of H28 and OC
(Table 2). However, both OC and H28 had similar magnitudes of effect on N2.
Maximal N2 production in the WP sediments occurred in the low OC (zero
ambient sulﬁde) and 0 pmoles OC and 0 umoles H2S (NO3' only) treatments;
adding OC and H28 had an inhibitory affect on N2 production (Figure 3).

N20 production accounted for a small faction of the nitrate removal in the
LP sediments (Figure 4). In LP, the addition of both OC and H28 signiﬁcantly
decreased N20 production, though there was also a signiﬁcant interaction
between these main effects (SAS). Even smaller amounts of N20 were produced
in the WP sediments. Adding the highest amount of H28 increased N20

production, whereas smaller amounts of H28 did not affect N20 production.

124

 

14‘ LP

N03 removed (umoles)

 

 

—O— 0 umoles C
....... 0... 23 umoles C
23.3 umoles C
46.7 umoles C

 

 

 

‘aath‘
"“43
O\. ‘

 

r I I ~

3 4 5 6

H28 added (umoles)

 

N03 removed (umoles)

 

 

 

 

 

——o— O umoles C

....... 0...... 23 umoles C
---v--- 23.3 umoles C
—..-A-—-- 46.7 umoles C

 

 

# l ' T

10 12 14 16 18

H28 added (umoles)

Figure 1: Nitrate removal across a gradient of added sulﬁde for Loosestrife Fen
(top; high ambient H28) and Windmill Pond (bottom; low H28). All treatments
began with 14.3 umoles of added nitrate and were sampled after 24 hours.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

+ O umoles C
5 ....... 0 ...... 2.3 umoles C
_-+-- 23.3 umolesC
LP ----A-—-- 46.7 umoles C
a 4 ‘
2
O
’5'
r; 3‘
8
D
'O
9 2 .
Q.
+
v
I
z 1 2
o . . ' ' ' .
o 1 2 3 4 5 6
H28 added (umoles)
+ 0 umoles C
....... 0...... 23 umoles C
8 2 WP
23.3 umoles C
A 46.7 umolesC
3 5 ‘
o
E
3 4 ‘ .......... ......
'0 ~11
a, ‘~~
0
.3 2 l
9
o. i
+<r o 2
I
Z
-2 -
'4 I l l r l ' T I '

 

 

0 2 4 6 8 10 12 14 16 18

H28 added (umoles)

Figure 2: NH4+ ﬂuxes across the H28 and 0C gradients in sediments from

Loosestrife Fen and Windmill Pond. Positive ﬂuxes indicate net production.
Note the differences in the y-axis ranges.

126

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—O— 0 umoles C
5 ....... O ...... 23 umoles C
. LP ---V"-- 23.3 umoles C
—.--A---- 46.7 umoles C
A 4 ‘
(I)
2
O
E
3 3 *
C
.Q
‘6
§ 2 l
a
2'
1 1
0 r f ' I ‘ I
o 1 2 3 4 5 6
H28 added (umoles)
—O— 0 umoles C
12 wp ....... 0 ...... 2.3 umoles C
—--v--- 23.3 umoles C
- —---A-—-- 46.7 umoles C
10 .
1;; "
_g_>
O
E 8 ‘
3
C
.9 6 «
‘6
3
8
a 4 *
N
z
2 ..
0 r r r I ' ' l '

 

O 2 4 6 8 10 12 14 16 18

H28 added (umoles)

Figure 3: N2 production for Loosestrife Fen (top) and Windmill Pond (bottom)
across a gradient of OC and H28. Note the differences in the y-axis ranges.

127

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.010
+ 0 umoles C
LP --O-- 2.3 umoles C
A 0.008 2 -V- 23.3 umoles C
3 —A-— 46.7 umoles C
o .
g 0.006 -
C
.9
g 0.004 .
'o
8
O. \z -----
C; 0.002 - ----l"-‘iz"~m=.5,:
2 ‘§
0.000 -
0 1 2 3 4 5 6
H.,S added (umoles)
0.025
—O— OumolesC
A ....... O ...... 23 umoles C WP
3 0.020 - ---v--- 23.3 umolesC
E: —-o-A-—-- 46.7 umoles C
3 0.015 .
C
.9
‘6
'8 0.010 1
9
a
‘2. 0.005 -
2
0.000 .

 

 

 

0 2 4 6 8 1O 12 14 16 18
H28 added (umoles)

Figure 4: N20 production (umoles) for Loosestrife Fen (top) and Windmill Pond
(bottom) across a gradient of OC and H28.

128

 

3.5

 

 

 

 

 

 

 

 

+ 0 umoles C
3 0 _ LP --O-- 2.3 umoles C
' i -v- 23.3 umoles C
a \\ -A- 46.7 umoles C
33 2.5 «
O
E
3
c 2.0 2
.9
‘3
'8 1.5 -
'a
o" 1.0 -
z
0.5 -
0.0 T f I I 1 *T
O 1 2 _ 3 4 5
H28 added (umoles)

Figure 5: N02‘ ﬂux for Loosestrife Fen across a gradient of OC and H28.

129

Conversion to NO2‘ accounted for approximately 25% of the overall N03'
removal in the LP sediments (Figure 5). The response of N02' to the OC
gradient was dependent on the level of the H28 treatment, as indicated by the
interaction between the two factors (Table 1). The effect size of H28 on N02'
production was 10 times greater than the effect size of OC. and 6 times greater
than the effect size for the interaction factor (Table 1). Adding small amounts of
H28 increased NO2' production (compare 0 and 0.5 umoles H28 on Figure 5);
however, adding larger amounts of H28 decreased NO2' production. There was
no net NO2' production in the WP sediments, possibly because of the difference
in processing times between the two sediments; transient production of NO2‘ may
not have been detected with the more rapid N03‘ disappearance in WP
sediments.

ln sediments from both sites, H2S showed net increases in the controls
(i.e., zero added H28 and DC in Figure 6), whereas in treatments where H28 was
added most of it disappeared by the end of the incubation (Figure 6). ln LP, all of
the added H2S was removed in the 0.5 and 2.5 pmole H28 treatments; however,
in the 5.5 pmole H28 treatment, the addition of OC inhibited further H2S removal.
In WP, a different trend is seen wherein adding OC did not stimulate H2S
removal, but may have inhibited it. In the medium (7.5 umoles H28) and high
(15.5 umoles H28) treatments, the zero added 0C treatment removed nearly all
of the H28 that was added. In the OC treatments, however, H2S was still

removed, but not to the degree that occurred in the zero OC treatments. In both

130

LP and WP, H2S had a much larger magnitude of effect on H2S ﬂux than did OC
(Tables 1 and 2).

8042' concentrations in the assays reﬂect the balance between production
by oxidation of reduced S, potentially coupled to N03‘ reduction, and
consumption by 8042’ reduction. Both OC and H28 additions signiﬁcantly
affected 8042' concentrations in LP sediments; however, there was a signiﬁcant
interaction between treatments (Table 1). In LP, increased additions of OC
progressively inhibited 8042' production. When 00 was not present (Figure 7,
solid line, top panel), 8042' concentrations signiﬁcantly increased with increasing
H28. Additionally, in the absence of added H28, adding additional OC
progressively decreased 8042' concentrations. In WP sediments, a similar
general pattern is seen, wherein the addition of H28 led to an increase in 8042'
when CO was not added; when CO was added, 8042‘ was consumed rather than
produced. Both OC and H28 signiﬁcantly affected 8042’ concentrations, though
there is a signiﬁcant interaction between the treatments (Table 2). This
interaction may also be caused by a similar effect of OC inhibition of 8042'
production as seen in the LP sediments. When OC was not present (Figure 7,
solid line, bottom panel), 8042' concentration signiﬁcantly increased in the 0
umoles added H2S treatment as compared to the 0.5, 2.75 and 5.5 umoles H2S
treatments (painNise comparisons df = 3, 58; t = -2.51, -2.08, -3.41; p=0.01, 0.04,
0.00 respectively). In both LP and WP, OC had a much larger magnitude of

effect than H28 on 8042' production (Tables 1 and 2).

131

I can account for a majority of the nitrate that was added in most
treatments in the LP and WP gradient assays (Figures 8, 9, 10), and the recovery
varies by treatment type. Figures 8-10 represent different ways to account for
the N removal. Figure 8 illustrates the various ﬂuxes across the low and high
H2S + NO3' treatments in both sites, which omits the possibility of the H28
interacting with DC to mask patterns. Figure 9 is the same as 8, but is averaged
across the OC treatments, with no H28 in the averages. Figure 10 averages all
OC treatments (low, medium, high) at a given H28 level, and therefore
incorporates any interactions that may occur between H28 and OC. Figure 10
generally has less N in the “unknown” category, which means that by adding in
factors I can account for more of the added N031 Conversely, in the H28 or DC
only treatments (Figures 8 and9) I did not recover the same amount of the

added N as in the treatments where both H23 and OC were added.

132

 

LP —-0— 0 umoles C
O 4 Q ....... 0 ...... 2.3 umoles C
. ——-v--- 23.3 umoles C
—---As—-- 46.7 umoles C

 

 

H28 removed (umoles)
r'o

 

 

 

 

 

   
 
  

 

 

 

 

 

-3 -

.4 u

'5 ' I I I I I

0 1 2 3 4 5 6
H28 added (umoles)
5

,0? —O—— 0 umoles C
% WP ....... 0 ...... 2.3 umoles C
e 0 - \QA. --+-— 23.3 umoles c
iv . 46.7 umoles C
E -5 I
9
8
.6 -10 *
‘8
D
'O
o
a -15 -
a)
N
I

'20 F I I I I I I I I

0 2 4 6 8 10 12 14 16 18
H28 added (umoles)

Figure 6: H28 ﬂux for Loosestrife Fen (top) and Windmill Pond (bottom) across a
gradient of 0C and H28. Negative ﬂuxes indicate net removal.

133

 

   
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3’: + 0 umoles C LP
6' 8 - ....... 0 ...... 2.3 umo|es C
E 6 _ —--v--— 23.3 umoles C
L; —---A-— 46.7 umoles C
> 4 _
O
E
9.’ 2 -
‘5
c: 0 4
g 2. ............ Q ................................... §
3 '2 " \'- ................
2 9 --------- 4 .....
'5. .4 -I éM-A .._,,_” ~~~~::.:i
“s -6 2 §..
In
-8 r I I I I I
H28 added (umoles)
10 _ —O— 0 umolesC
7,? ....... O ...... 23 umOIES C WP
% 8 _ ---v--- 23.3 umolesC
g —-.-A-—-- 46.7 umolesC
Z 6 -
m
5
E 4 ‘
2
8 2 ~
C
.9
*5 0 *
3
8
a '2 ‘
6 4 ‘
w
-6 I I T I I ' ' . I I
0 2 4 6 8 10 12 14 16 18
H28 added (umoles)

Figure 7: 8042' ﬂux for Loosestrife Fen (top) and Windmill Pond (bottom) across
a gradient of OC and H28.

134

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Qwo mmrd mmvd vm mend Lotm
md Sod mnmd : mid m Name To
Qmw Bod 80.0 v. 5 m2: m vwvé w 03¢ N2
02 mwod Sod No momd m mend 0
New movd movd we m.mw 5.5
mg: Sod 30.0 EN 036 a ﬁn mb
0..: word oood Qo nmwd m 0w m 5% .....IZ
0; wood wood N6 ONE? m Nm 0
NN® omoé omoé no 53% Lotm
0.3 mmmd mood Wm wmmd m moodm 90 RES?
0? nomd oood Qm mmmd m $0.5m w .62
Qm $0.0 $0.0 QN mNmN. m mnmﬁ o
.95 E252»,

“cocanoo Susan Co meg—um

N3 352:? w:_m> a 0:2 u. cues. 8295 we Sam .88". omconmmm

 

 

 

.oN_m 80% Lo Soto v6
ouEEmmE 8:8 02m _ouoE <>Oz< 05 9 .298 m Co E 9: 9 N3 .AmNI EoBEm :95 n3 E0: 60 new web 56$ 3
OS: 4.40m ..NOZ .Ouz .mz . ....Iz ._m>oEo._ w”: 9.6 -oozv wmﬁmtg 3:039 .3929: Lou, oocmtg Co m_w>_mc< ._. 038.

135

3:8 ”F 28¢

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

for wood wood mo Nomoo .6th
Qt who; ooo.o mo? nomow o 03.3 90

m.m mid ooo.o to mwoé m mower m 5% -Nvom
moo Sod ooo.o moor Nondor m maﬁmom 0

mo omoo omoo vo Nvoé 5cm

ow ooo.o ooo.o vow 5nd 0 oNoN mb 63E?
ﬁmo QKN ooo.o voowm n34; m SYNNN. m mNI
No ooo.o ooo.o Boo ommd m PEN o
VNN nmoo “moo vo ooo.o Lotm
w. : ooo.o ooo.o on ommo m mmoN 9o
mom vao ooo.o YNN moﬁv m 803 m 5% -NOZ
Qm mroo ooo.o ﬁw oowo m 2%; 0
NS oooooooo vonooooo 5 Fvwoooo .otm

mg mooooooo Foooov o6 Fooooooo o vmoooo o mb

fm ovoooooo Foooov o. R Emooooo m mmoooo o w xst 0N2
ow omoooooo moooo mo Soooooo m wmooooo o

.23 £0.93:
Emcanoo 9.30m ho men—Em
~ 3 35:? o:_w> a 0:2 .... S35. 3230 .8 Sam Baum". wmcoamwm

 

 

136

Table 2. Analysis of variance for individual response variables (H28 removal;
NH4“, N2, N20, 8042' ﬂux) by factor (H28 and DC) from WP (low ambient H28).
w2 is the fit of a factor to the ANOVA model, also called the magnitude of effect.
N02‘ was not produced during WP experiments, so it is not a response variable
in these analyses. Also, because N03‘ was gone in all treatments at the end of
the experiment, N03' removal is not considered as a response variable.
Abbreviations: SS = sum of squares, DF = degrees of freedom, MSE = mean
square error, VC = variance component.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Response Factor SS DF MSE F ratio p VC (.02
value
C 19.2 3 6.4 4.9 0.004 0.19 9.7
NH4+ ﬂux S 27.5 3 9.2 6.9 0.000 0.30 15.0
C*S 25.0 9 2.8 2.1 0.041 0.17 8.4
Error 81.4 64 1.3 1.31 66.9
C 14.9 3 5.0 13.3 0.000 0.18 21.4
N; ﬂux S 26.0 3 8.7 33.5 0.000 0.32 28.2
C*S 8.3 9 0.9 3.6 0.001 0.09 9.1
Error 15.5 64 0.2 0.25 31.3
C 1.2 x 3 4 x 3.4 0.024 2 x 3.1
NZOﬂux 104 10‘5 10'5
S 2.9 x 3 9.9 x 8.3 0.001 1 x 1.6
10 4 1o 4 1o '5
C*S 3.3 x 9 4 x 3.1 0.004 6 x 7.4
10 4 1o '5 1o '5
Error 6.9 x 59 6.7 x 6.7 x 87.8
10 4 1o 4 1o 4
C 107.8 3 35.9 56.4 0.000 1.3 . 4.2
H28 S 2168.1 3 722.7 1133.1 0.000 27.0 86.2
removal C*S 195.0 9 21.6 33.9 0.000 2.3 7.5
Error 40.8 64 0.6 0.6 2.0
C 182.7 3 60.9 51.7 0.000 2.2 44.7
so} ﬂux s 87.1 3 29.1 24.7 0.000 1.0 20.9
C*S 54.7 9 6.1 5.2 0.000 0.5 11.0
Error 75.3 64 1.2 1.1 ~ 23.5

 

 

 

 

 

 

 

 

 

 

137

 

LE— Low HZS_A LP - High H28

i
LNZ

: (:1 NH4+
I

E N03:
‘ CID unknown

 

 

 

 

 

 

 

umoles produced or consumed

 

 

 

 

 

15 , WP - Low H28 WP — High st

 

 

 

[:1 NH4+

umoles produced or removed

 

 

 

 

 

c712 unknown

Figure 8: Fractions of each N end-product from the gradient assays. These are
means across the H28 only gradient (i.e., they do not include treatments with
added 0C). Total N ﬂux is based on the N03’ removal that was observed. Also
shown for comparison are the SO42“ and H28 ﬂuxes. Quantities added: LP low
H28 (0.5 umoles) , LP high H28 (5.5 umoles), WP low H28 (1.5 umoles), and WP
high H28 (15.5 umoles).

138

LP low OC LP high CO

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

15 ~ . i-——~ -—~-«
8 r", M — N2
E {I [2 NH 4'
3 4
g 10 4 M :3 N02
LB) COZIBLE: N03-
‘O . < CZIJ unknown
5 .
g L...
“o
o
5. 80.:2
g 0 ‘ N st
§
-5 l
WP low OC WP high CO
— N2 i //
[:1 NH4+ 4 /
4
= N03- .
unknown

 

 

 

8042'

so.” i ’ N m E

 

 

umoles produced or consumed

 

 

 

 

Figure 9: Fractions of each N end-product from the gradient assays. These are
means across 00 treatments (i.e., they do not include H28 in the averages).
Total N ﬂux is based on the N03‘ removal that was observed. Also shown for
comparison are the SO42” and H28 ﬂuxes. Quantities added: LP low CC (2.3
umoles), LP high CO (46.7 umoles), WP low CC (2.3 umoles), and WP high 0C
(46.7 umoles). Positive ﬂuxes indicate net production.

139

LP highﬂgs ,

—N2

15'

 

(:1NH+
10‘ 4

 

NQ3'

 

 

 

c2121 unknown

0 _ I 8022' H28

N .~.g

Hr 4* *- H
[I
K [I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

umoles produced or consumed
01

 

 

LP low H28

 

 

 

 

 

 

 

 

 

-5 -
WP high st
— N2

15 ‘ . :1 NH4+

10 q — N02-

N03-
(22:2: unknown

 

 

 

 

 

 

WP low H28

 

 

 

 

 

 

 

N 80.4 st

umoles produced or removed
0

 

 

 

 

 

Figure 10: Fractions of each N end-product from the gradient assays. These
are means across all treatments (i.e., all levels of DC in a given H28 treatment
were included). Total N ﬂux is based on the N03' removal that was observed.
Also shown for comparison are the 8042' and H28 ﬂuxes. Quantities added: LP
low H28 (0.5 umoles), LP high HZS (5.5 umoles), WP low H28 (1.5 umoles), and

WP high H28 (15.5 umoles).

140

In sediments from both sites, NO3' was converted to both N2 and NH4+,
though under different circumstances. For example, increasing amounts of NH4+
production occurred in LP with increasing levels of H28 only (no 0C interaction;
Figure 8, top two panels). However, in WP, NH4+ production only occurred when
H28 was added together with OC. For an illustration of this, compare the bottom
two panels of Figures 8 and 10. In Figure 8, across the H28-only gradient (no
DC in averages), no NH4+ production occurred. However, Figure 10 shows that
adding high levels of H28 together with DC produced NH4”. Adding more QC
resulted in an increase in NH4+ production in LP (Figure 9, top panels), though
the increase was not as great as the increase in NH4+ between the low and high
H28 treatments. Adding DC in WP, however, resulted in a decrease in NH4+
production (Figure 9, bottom panels). Therefore, conversion to NH] was a
signiﬁcant N03' sink in both sites but was evidently subject to different controls.

WP consistently had more of the added N03' converted to N2 compared to
LP (Figures 8-10). Adding more H28 decreased the N2 produced in both sites
(Figures 8 and 10). Adding more H28 also decreased N02' production, which
was only generated in significant amounts in the LP sediments (Figure 8).
Adding more QC increased NOz' production in LP (Figure 9).

Both LP and WP also had evidence of H28 consumption and 8042'
production, but again, the expression of these responses was dependent on H28
and OC interactions. In LP H28 + N0; treatments, all of the added HZS was
consumed. However, adding QC tended to decrease HZS removal (compare top

right panels of Figures 8 and 10). The interactions between OC and H28 are

141

also apparent in the 8042' ﬂux patterns. In LP, adding more HZS resulted in an
increase in 8042' (Figure 8, top two panels). However, this response
disappeared when H28 was added together with DC (Figure 10, top two panels)
or in the 0C only gradient (Figure 9). This pattern of OC inhibiting 8042'
production was also seen in WP. WP sediments produced 8042’ in both the low
and high H28 + NO3' treatments (Figure 8, bottom panels), but this production

was reduced when OC was added to the treatments (Figure 10, bottom panels).

Time-course assays

The time-course assays were designed to examine the dynamics and
transient changes in N03' transformations as added N03' is removed. As in the
gradient assays, N03‘ was removed much faster from WP than from LP (Figure
11; note difference in x-axis time scales). In WP, ~75% of the added N03‘ was
gone after 12 h; in LP, ~75% was gone in 24 h. In LP, the HZS + N03' treatment
lagged behind the other treatments in N03’ removal early in the experiment, but
nearly all of the added N03' was gone by the last sampling time. The treatments
wherein either 0C or H28 were added had more N03' removed by the ﬁnal
sampling time than did the N03' only treatment. In LP, N03' removal in the 00
treatment was greater than in the N03‘ only treatment only at the ﬁnal time point,
and LP sediments tended to have higher amounts of N03‘ removed than when
NO3' was added alone (Table 3). H28 signiﬁcantly decreased N03' removal early

in the experiment, but by the last sampling point, the H28 additions signiﬁcantly

142

increased N03” removal (Table 3). The 00 treatment consistently produced
more N03' removal in LP compared to the H28 treatment.

Adding 0C or H2S, together or separately, stimulated N03' removal in the
WP sediments as well; these treatments consistently removed nitrate faster than
did the N03' only treatment (Figure 11). Both the OC and H28 treatments
increased N03‘ removal compared to the N03‘ only treatments at most time
points, and were never signiﬁcantly different from each other in their effects on
N03' removal (Table 4).

The OC and H28 treatments had different effects on NH4+ ﬂuxes in the two
sediments (Figure 12). In LP, neither the H28 nor OC treatments differed
signiﬁcantly from the N03' treatment in their NH4+ production (Table 3). They did,
however, generally differ signiﬁcantly from each other because the H28 treatment
stimulated NH4+ production, whereas the OC treatment stimulated NH4+ removal.
Additionally, the H28 treatment steadily increased in NH4+ production over time,
whereas the OC treatment did not show any production of NH4” until the 48-hr
sampling time. The results from LP in these assays are similar in the magnitude
of NH4" production compared to the LP gradient experiments (Figure 2). In the
WP timed assays, NH4+ was not produced, as was seen in the WP gradient
assays (Figure 2), but was removed or did not change. Addition of H28 generally
produced more NH4+ than did N03' alone, whereas the addition of OC stimulated
NH4+ removal compared to N03‘ only (Table 4).

N2 was produced in nearly all treatments in both LP and WP (Figure 13).

In the LP sediments, addition of H28 signiﬁcantly stimulated N2 production

143

compared to the N03' only treatment (Table 3). The OC treatment stimulated N2
production, which was significantly greater than that in the N03‘ only treatment
only at the 2nd sampling point. Generally, the H2Streatment produced more N2
than did the 0C treatment across sampling times (Table 3). In WP, the H23
treatment initially had greater N2 production, but had the lowest N2 production by
the last sampling time (Table 4). In WP the addition of OC and H28 signiﬁcantly
stimulated N2 at the second sampling time, but then signiﬁcantly decreased it
compared to N03' at the ﬁnal sampling point, making it difﬁcult to discern which
had the greater effect. However, the OC treatment generally stimulated N2
production to a greater degree than the H28 treatment (Table 4).

N20 production was low in both LP and WP (Figure 14), and as in the
gradient assays, accounted for a very small fraction of the overall N03' removal.
LP produced more N20 than WP (as was also seen in the gradient assays;
Figure 4). Neither H2S nor 00 signiﬁcantly affected N20 production compared to
the N03' only treatment (Table 3). Treatments with added H2S typically had
greater N20 production in LP sediments (Table 3). WP had the opposite trend,
wherein N20 was produced early on in the experiment, and more production
occurred when treatments included H2S. OC never had a signiﬁcant affect on

N20 production in WP (Table 4).

144

 

1: Control
16 1 CI) N03 only LP
C 4' N03

- H28 + N03

_C+H28+NO3

 

124

 

 

ix}
{’5
g
.05.
5f
33‘
f6

N03' removed (umoles)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Time (hrs)

 

[2:] Control
[:3 N03 Only
m C + NO3 WP
14 - - H28 + No3
- c+ st + N03

16

l

 

124

 

10-

11'! unmet-125.232.1952”: itﬁ‘ﬁﬁﬁl‘fﬁﬁ'l . .-

NO3‘ removed (umoles)

 

 

WWQKEQSEWMW‘W“ 4W ‘

 

 

 

héMlﬁéﬁﬁRt‘KéﬂKAﬁ-‘Tm :3

 

 

 

 

 

 

0 7 I V I r
5 1O 15 20

Time(hrs)

 

 

-24

Figure 11: Nitrate removal over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 umoles of N03' were added to each treatment at the start of
the experiment.

145

2.0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:3 Control LP

g 4 [:1 N03 only

a 1'5 1:: C + N03

5 - H28 + N03

T; 1.02-C+H28+N03 '

o l.
E

2

5 0.5 -

C

.9

‘6

T1. T

5

:5r -0.5 4

z

'1.0 T I r l l
0 10 20 30 40 50
Time (hrs)
1.0

g 0.5 2 ‘—

‘s’
13’ 00 T~ T F 1

> ~;

g .

5
.5 '1'0 d 1'
'8 .

9 -1.5 - f I?

Q _ (1"
Iv ._.
z -2.0 - i

 

 

 

'2.5 T I I l I
0 5 10 15 20 25 30
Time (hrs)

Figure 12: Ammonium ﬂux over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 umoles of N03; was added to each treatment at the start of the
experiment.

146

NO2' was produced in both LP and WP during the time series assays
(Figure 15), in contrast to the'gradient assays where only LP showed production
(Figure 5). LP produced a much greater amount of NO2' throughout the
experiment, whereas NO2' production in WP was two orders of magnitude lower.
In the ﬁrst sampling at LP, the N03'-only treatment produced more NO2’ than the
0C or H2S treatments, though the effect was not statistically signiﬁcant. In the
ﬁnal two time points at LP, the H28 treatment signiﬁcantly stimulated NO2'
production compared to N03' only and compared to the OC treatment (Table 3).
ln WP, both H23 and OC stimulated NO2' production, particularly early in the
experiment where both treatments had greater NO2' production than N03‘ alone.
NO2' production declined over time. At the ﬁnal sampling point, neither treatment
was statistically different from the N03' treatment, though the H28 treatment had

signiﬁcantly higher NO2' than the 0C treatment (Table 4).

147

 

2.5
:3 Control LP

I: N03 only

2.0 . [:3 C + N03

— H28 + N03
- C + H28 + N03

 

 

1.5 -

1.0 .

N2 ﬂux (umoles)

0.5 -

 

 

 

 

 

 

 

 

0.0
40 50

 

ﬁmemml

 

3.5
:1 Control

[:1 N03 only

C + N03 WP
- H23 + N03 1
- C + H28 + N03

3.0 -

2.5 -'

 

 

2.0 “ 11
5

1.5 4

N2 ﬂux (umoles)
H

1.0-

0.5 4

 

 

 

 

 

 

 

 

   

 

 

 

 

 

0.0 n. 55" r
0 5 10 15 20 25 30

Time (hrs)
Figure 13: N2 production over time in both Loosestrife Fen (LP) and Windmill

Pond (WP). 14.3 umoles of N03’ was added to each treatment at the start of the
experiment. Controls received no N033 and all vials were maintained undenNater

to ensure that atmospheric N2 and 02 did not leak into the vials.

148

 

0.06

:1 Control
I: N03 only

[:13 C+NO3
I: HZS+NO3
0.04 4 - C+ H28+ N03

0.05 4

 

 

0.03 -

N20 ﬂux (umoles)

0.02 -

 

 

:0 a j m

l I I I

 

 

 

 

 

 

 

 

 

0 10 20 30 40 50
Time (hrs)
0.008
[:I control
I:l N03 Only
C + N03
0.006 - - H28+ N03
13‘ - C + H28 + N03
6
E
EL
x 0.004 ~
3
:
ON
2
0.002 2
0.000 I . .
0 5 10 15 20 30

Time (hrs)

Figure 14: N20 production over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 umoles of N03' was added to each treatment at the start of the ‘
experiment. Controls received no N03', and all vials were maintained underwater
to ensure atmospheric N2 and 02 did not leak into the vials.

149

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.0
LP - Control
[:1 N03 only
C + N03
1.5 - — H28 + N03
:00? - C + H28 + N03
'5
E
3
x 1.0 -
3
c
o" I
z
0.5 -
0.0 o .
0 30 40 ~ 50
0025 Time (hrs)
WP I: Control
[:1 N03 Only
0.020 ~ C + N03
' - H28 + N03
3 g — c + H2s + N03
0 q
E 0.015 %
3 ii
>5 1
a: :4.
' N 0.010 " 3’
0.005 - g
2::
3
5
0.000 . .
0 5 15 20 25 30
Time (hrs)

Figure 15: N02' production over time in both Loosestrife Fen (LP) and Windmill
Pond (WP). 14.3 umoles of N03“ was added to each treatment at the start of the

experiment.

150

8042‘ ﬂux was also affected differently by the various treatments in
sediments from both sites (Figure 16). Both OC and H28 signiﬁcantly affected
SO42“ ﬂux at the second two sampling points, though H2S increased 8042' and
QC decreased 8042‘, as was also seen in the gradient experiments (Table 3). In
WP, neither H2S nor 0C had a consistent effect on SO42’ ﬂux (Table 4). At 24
hrs, however, it is clear that treatments with H2S added had slightly higher ﬁnal
8042' concentrations than the N03' only or DC treatments (Table 4). However, I
do not see the clear increase in 8042' that was seen in LP. These trends are
consistent with the results from the gradient assays (Figure 7). In both LP and
WP, the controls (no QC, N03' or H2S) removed 8042' at the ﬁrst sampling point
because all of the other treatments had N03' added, which would have inhibited
SO47" reduction (Figure 16). The decreased 8042' concentration at the ﬁrst
sampling point is due to 8042‘ reduction that has already occurred by that time.

This sulfate reduction is also evident in the H28 production in the same
controls (Figure 17). The N03' added to the other treatments suppressed this
reaction until the added N03' was largely removed. In LP, nearly all of the added
H2S was gone by the ﬁrst sampling point. By the 2"d sampling point some 8042'
reduction may have been occurring in some of the treatments. It is clear that
across all treatments in both sites, 8042' reduction had commenced by the last
sampling point leading to an increase in H2S by the end of the experiment,
particularly in the 0C and N03' only treatments. This pattern of a rapid decrease
in the added H2S, followed by an increase after N03' is depleted is different than

the steady increase in H2S concentration seen in the controls.

151

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8 E: Control
[:1 N03 only LP

75 6 1:: C + N03

30’ - H28 + N03

5 4‘ —c+H2s+No3

a

>

E

92 2 ‘

3

.5 0

*6

3

E -2 «

Q.

o' -

a) -4 g» .3

'6 I ﬁ I I I
0 1o 20 30 4o 50
4 Time (hrs)

7:?

d) _
B 2 WP
E l I
a T

° 1

E

2

a -2 1 4 , Fl

C l
.9 ‘
‘3 I l
'o '4 ‘

2

Q
NOV '6 _ i
(I)

'8 I I I I I
0 5 10 15 20 25 30
Time (hrs)

Figure 16: 8042‘ ﬂux over time in both Loosestrife Fen (LP) and Windmill Pond
(WP). Controls received no N033 H2S or DC, but were maintained to evaluate
how 8042' changed in the absence of these factors over time.

152

 

[:1 Control
[:1 N03 only
[:1 C + N03
- H28 + N03
2 . — C+H28+NO31E

 

 

 

 

 

, l- . 1

[I

 

 

H28 production or consumption (umoles)

 

 

 

 

‘2 I I T I I
0 10 20 30 40 50
Time (hrs)
4
3 a
a

 

 

 

 

 

 

HZS ﬂux (umoles)

 

 

 

 

0 5 10 15 20 25 30
Time (hrs)

Figure 17: H2S ﬂux over time in both Loosestrife Fen (LP) and Windmill Pond
(WP). Controls received no N03’, H2S or DC, but were maintained to evaluate
how H28 changed in the absence of these factors over time.

153

Table 3. Independent contrasts within the analysis of variance (ANOVA) to
compare treatments across time in LP. The estimated ﬂux (Estimate) is in
umoles, and indicates the difference between the two treatments. A positive ﬂux
denotes that the effect of the ﬁrst factor listed (e.g, H28 in H2S vs. N03' only) on
that ﬂux was greater than the second factor. NS = not signiﬁcant.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Response Factor Hour F120 P Estimate
(umoles)

H2S vs. N03' only 12 14.9 0.0010 -2.43

H2S vs. N03' only 24 3.51 NS -0.89

H2S vs. N03' only 48 21.5 0.0020 1.20

0C vs. NO3' only 12 0.98 NS 0.62

N03‘ OC vs. NO3' only 24 1.85 NS 0.65

removal OC vs. N03' only 48 28.2 <0.0001 1.38

H2S vs. CC 12 23.49 <0.0001 -3.05

H2S vs. DC 24 10.44 0.0042 -1.55

H28 vs. DC 48 0.45 NS -0.17

H2S vs. N03‘ only 12 0.26 NS -0.08

H2S vs. N03' only 24 2.57 NS 0.53

__H2S vs. N03' only 48 2.46 NS 0.48

0C vs. N03’ only 12 0.11 NS -0.05

NH4+ ﬂux 00 vs. N03' only 24 2.24 NS -o.49

OC vs. N03' only 48 2.65 NS -0.49

H2S vs. DC 12 0.03 NS -0.03

H2S vs. DC 24 9.60 0.0057 1.03

H28 vs. DC 48 10.23 0.0047 0.98

H2S vs. N02; only 12 9.14 0.0073 0.24

H2S vs. NO3' only 24 6.50 0.0202 0.17

H2S vs. N03' only 48 5.69 0.0282 0.59

OC vs. N03“ only 12 8.26 0.0101 -0.23

N2 ﬂUX OC vs. N02; only 24 16.04 0.0008 0.26

0C vs. N03' only 48 0.02 NS -0.03

H2S vs. DC 12 34.78 <0.0001 0.47

H2S vs. DC 24 2.12 NS -0.10

H2S vs. DC 48 6.37 0.0212 0.62

 

154

 

Table 3: cont’d.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Response Factor Hour F120 P Estimate
(umoles)

4128.13. NO3' only .. 12 4.04 NS -0.40

._ H2S vs. NOg‘onIJy_ﬁL______ﬁ2_4_ 25.76 <0.0001 0.84

H2S vs. NO3-OILIY 48 7.37 0.0142 0.43

0C vs. N02; only 12 3.13 NS -0.35

N02 ﬂUX 0C vs. N02; only 24 0.12 NS -0.06

OC vs. NOg' only 48 0.40 NS 0.11

H2S vs. DC 12 0.06 NS -0.05

H2S vs. DC 24 29.32 <0.0001 0.89

H28 vs. DC 48 3.71 NS 0.32

H2S vs. NOg' only 12 0.02 NS -0.09

H2S vs. N02; only 24 13.38 0.0016 3.32

H2S vs. No; only 48 7.34 0.0144 3.43

0C vs. N03' only 12 3.97 NS -1.29

8042‘ flux oc vs. N03' only 24 14.30 0.0012 -3.43

0C vs. N03' only 48 5.00 0.0382 -3.00

H2S vs. DC 12 3.42 NS 1.19

H2S vs. DC 24 55.33 <0.0001 6.74

H28 vs. DC 48 22.95 0.0001 6.43

 

155

 

Table 4. Independent contrasts within the analysis of variance (ANOVA) to
compare treatments across time in LP. The estimated ﬂux (Estimate) is in
umoles, and indicates the difference between the two treatments. A positive ﬂux
denotes that the effect of the ﬁrst factor listed (e.g, H28 in H2S vs. NO3'only) on
that ﬂux was greater than the second factor. NS = not signiﬁcant.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Response Factor Hour F120 P Estimate
(umoles)
__H_2S_vs__.-N03‘ only 6 25.45 <0.0001 5.66
H2S vs. N03' only 12 .. ﬂ 1.21 NS 1.48
H2S vs. N03' only 24 6.45 0.0195 0.22
0C vs. N03'onlL 6 12.79 0.0019 4.01
NOa' 0C vs. N03' only 12 4.46 0.0474 2.85
removal 00 vs. No; only 24 6.45 0.0195 0.22
H2S vs. DC 6 2.16 NS 1.65
H2S vs. DC 12 1.03 NS -1.37
HES vs. DC 24 0.00 NS 0.00
H2S vs. N03' only 6 2.78 NS 0.75
H2S vs. N03" only 12 4.44 0.0480 0.68
H2S vs. N03‘ only 24 7.54 0.0125 0.98
0C vs. N03' only 6 9.09 0.0068 -1.35
NH4+ ﬂux 00 vs. Nog' only 12 0.01 Ns -0.02
0C vs. NO3‘ only 24 0.75 NS -0.31
H2S vs. OC 6 21.93 0.0001 2.10
H28 vs. DC 12 4.76 0.0413 0.71
H28 vs. DC 24 13.07 0.0017 1.29
H2S vs. N03' only 6 1.94 NS 0.29
H2S vs. N03' only 12 5.57 0.0305 0.35
H2S vs. N03' only 24 59.1 <0.0001 -1.62
OC vs. N02; only 6 0.29 NS 011
N2 ﬂUX OC vs. N03' only 12 24.27 0.0001 0.73
0C vs. N03' only 24 7.84 0.0118 -0.51
H2S vs. QC 6 3.72 NS 0.41
H2S vs. CC 12 6.59 0.0200 -0.38
H28 vs. DC 24 27.69 <0.0001 -1.11
H2S vs. N03' only 6 37.49 <0.0001 0.0059
H2S vs. N03‘ only 12 8.64 0.0081 0.0004
H2S vs. N03‘ only 24 22.18 0.0002 0.0010
OC vs. N03' only 6 0.04 NS 0.0002
N20 ﬂUX OC vs. NO3' only 12 0.16 NS 0.0001
OC vs. NOg'only 24 0.32 NS -0.0001
H2S vs. DC 6 39.36 <0.0001 0.0057
H2S vs. CC 12 6.45 0.0195 0.0004
_H?LS vs. DC 24 27.82 <0.0001 0.0011

 

156

 

Table 4 cont’d.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Response Factor Hour F120 P estimate
(umoles)
H2S vs. NO3‘ only 6* _ 22.18 0.0001 0.009
H2S vs. NO3' only 12 0.20 NS 0.001
H2S vs. NOg' only 24 _ 1.48 NS 0.002
OC vs. N05 only 6 11.42 0.0030 0.007
N02 ﬂUX OC vs. N03' only 12 4.96 0.0375 0.004
QC vs. N03‘ only 24 1.80 NS -0.002
H2S vs. DC 6 1.77 NS 0.003
H2S vs. DC 12 3.18 NS -0.003 _
H28 vs. DC 24 6.54 0.0188 0.004
H2S vs. N03' only 6 3.09 NS -1.17
H2S vs. N02; only 12 0.41 NS 0.63
H2S vs. N03' only 24 24.30 <0.0001 2.31
0C vs. N03' only 6 0.67 NS -0.54
8042' ﬂux oc vs. N03' only 12 1.42 Ns -1.17
0C vs. N03' only 24 3.95 NS -0.93
H2S vs. DC 6 0.89 NS -0.62
H2S vs. DC 12 3.36 NS 1.80
H28 vs. DC 24 47.85 <0.0001 3.24

 

157

 

Discussion:
NO3‘ processing in high and low H2S sites

The different N03' removal rates and end-products imply that different
microbially mediated reactions are responsible for the N03' processing in
sediments from these two sites. This may be due to the biogeochemistry of the
sediments and particularly to differences in ambient H2S concentration, or it may
be due to other environmental factors. While H2S can be an electron donor via
chemolithotrophic oxidation, it is also acutely toxic to microbes (Wang and
Chapman 1999, Senga et al. 2006). The high 1H2S site (LP) may show evidence
of this toxic effect in its slower N03‘ removal rates compared to the low H28 site
(WP; Figure 1). However, the time-course experiments showed that all of the
added N02; will be removed given sufﬁcient time (Figure 11).

The N03’ removal at' both sites resulted in different N end-products,
presumably due to different reduction pathways, possibly including both DNRA
and denitriﬁcation. I hypothesize that DNRA explains the NH4+ production in both
sites, despite the fact that the responses to the OC and H28 treatments were
different between sites. In LP NH4" production increased across the H28
gradient when OC was absent (black line, Figure 2); adding 0C stimulated more
NH4+ production, though not as consistently as in the H28 gradient. There is
further support forthe idea that H2S is directly inﬂuencing NH4" production from
the timed assays (Figure 12). In this case, treatments with added H2S (as well as
the N03' only treatments) produced more NH4“ production than in the CC + NO3‘

treatment. In WP, there was no apparent increase in NH4+ across gradients of

158

either H2S or DC alone; instead, I found substantial NH4" production only when
0C and H28 were added together (Figure 2). This perhaps argues that the NH4+
is being generated from two different processes, and is therefore inﬂuenced in a
different manner by both H28 and OC. The timed assays give a different view of
the NH4+ ﬂux in WP. Here, we do not see any apparent net NH4+ production, but
rather see NH4+ uptake, even in the CC + H2S + NO3‘ interaction treatment. In
the gradient assays from both sites, the magnitude of the H28 effect was larger
than the OC effect (Tables 1 and 2). Similarly, in the timed experiments, H2S
either caused more NH4“ production than OC. as in LP (Table 3), or caused less
NH4“ removal over time compared to either N03’ only or DC treatments, as in WP
(Table 4). These ﬁndings argue that NH4+ ﬂux in both of these sites is directly
linked to H2S oxidation, possibly through DNRA.

An alternative explanation for the increased NH4+ ﬂuxes is that increasing
OC and N03' availability increased denitrification, generating more OM
breakdown and leading to greater NH4” remineralization and increased NH4+
ﬂuxes. I cannot exclude the possibility that the NH4+ produced is from DNRA or
increased OM remineralization, though earlier (Chapter 3) I measured substantial
DNRA, including in the sites studied herein, when measured using 15N methods.
Follow-up work using 15N tracer methods in combination with the assay
technique would help elucidate the mechanisms.

N2 production in the WP gradient experiments was generally 3 times
greater in LP, potentially indicating that denitriﬁcation was a more important N03‘

removal process at WP (Figure 3). In the gradient experiments from both sites,

159

 

H28 and OC had similar magnitudes of effect on N2 production, indicating that
both are of similar importance to N2 generation (Tables 1 and 2). CC, however,
affected the two sites differently. In LP, adding OC increased N2 production over
the H28 treatments, whereas 0C additions in WP decreased N2 production
compared to the H28 only gradient (black line, Figure 3). And in WP the addition
of H28 further inﬂuenced N2 ﬂux by decreasing N2 production over the H28
gradient (Figure 3); the same inhibition of 0C and H28 on N2 production was also
seen at the ﬁnal sampling point (24 hr) on the WP timed experiment, wherein the
N03' only treatment produced the highest N2, followed by the two treatments with
DC added, and ﬁnally by the H28 + NO3' treatment. This OC/H2S inhibition
contrasts with LP, wherein adding intermediate amounts of H28, particularly in
combination with DC, stimulated N2 production (Figure 13). This same effect is
seen in the LP timed experiment (Figure 13), where the H28 treatments
consistently produced signiﬁcantly more N2 than the N03' only treatments, and
generally produced more N2 than the OC treatment (Table 3). For unknown
reasons, the N2 production rate in the WP timed assays was about half of what it
was in the gradient assays.

N20 consistently accounted for only a very minor fraction of the overall
nitrate removal and transformation (<1%). lts production, however, was
inﬂuenced by both time and the gradients of OC and H28. In LP, adding any H2S
decreased the amount of N20 produced (Figure 4). This inhibition was partially
relieved by adding CO in combination with H2S, but the N20 levels did not get as

high as when OC was added alone. In the WP gradient experiments, N20

160

production is only stimulated at the highest levels of H28 and OC. although there
is a high degree of variability (Figure 4). The same is true for the WP timed
assay N20 ﬂux, wherein the highest N20 ﬂux was observed in treatments with
H2S together with DC (Figure 14).

Both sites had large differences in N02‘ production. LP consistently had
high amounts of NO2' produced (>1 umole), whereas NO2' production in WP was
far lower (~ 0.01 umoles), and not detectable at the end of the 24 hr incubation of
the gradient experiment. In LP, small amounts of H28 stimulated NO2'
production, and adding OC increased the production as well (Figure 5), though
H2S clearly had a larger magnitude of the effect compared to OC (Table 1). A
similar pattern was seen in the LP timed assays, wherein the H2S only treatment
had generally had the highest N02' production compared to the N03' only
treatments (Figure 15, Table 3). Though both OC and H28 affected NO2'
production in WP, there were no discernible patterns to the effects (Table 4).

Thus, it is clear that both DNRA and denitriﬁcation were important N03'
removal processes in both sites, but under different conditions. Denitriﬁcation
was consistently the more important N03' removal pathway in WP, whereas in LP
both denitriﬁcation and DNRA were equally important. Indeed, at high H2S
levels, DNRA (conversion to NHX) accounted for more N03' removal than
denitriﬁcation (conversion to N2). LP consistently showed measurable DNRA
across the H23 and OC gradients as well as when both H28 and OC were added.
DNRA in WP, on the other hand, only appeared to be important when H28 and

OC were added together.

161

Relative effects of OC and H23 on N03' reduction end-products

The complex interactions between OC and H28 make it difﬁcult to discern
their relative importance as controls on N03‘ removal end-products based on
signiﬁcance of the main effects alone. However, by partitioning the variation that
can be attributed to a given factor and calculating the magnitude of a factor’s
effect I can gain more insight into the relative importance of 0C and H28 in N
cycling. While it is clear that both OC and H28 are inﬂuencing NH4“ production
(Figures 2 and 12), H2S consistently had a larger magnitude of effect in both
sites (Tables 1 and 2). Additionally, the H28 treatment consistently produced
signiﬁcantly more NH4+ than did the OC treatment in both sites (Tables 3 and 4).
Both OC and H28 affected N2 production equally, based both on their magnitudes
of effects in the gradient assays (Tables 1 and 2); however, in the LP timed
assays, the H28 treatment produced more N2 than the QC treatment, but in WP,
the H28 treatment produced less N2 than the 0C treatment (Tables 3 and 4).
While DC has long been known to affect N2 production by stimulating
denitriﬁcation, the inﬂuence of H28 on N2 production has not been previously
noted, particularly in freshwater ecosystems. From these experiments, I can
infer that both H23 and OC contribute to N03' removal and to the ultimate end-
product of the N03' reduction.

Few other studies have simultaneously examined the effects of both OC
and H28 on N cycling. Published studies to date have been largely carried out in
batch reactors that are very different from natural ecosystems. In one of these

examples, Reyes-Avila et al. (2004) examined the effects of acetate and H28 on

162

denitrification. They found that denitriﬁcation rates were highest with acetate as
the sole electron donor. H2S as a sole electron donor resulted in lower
denitriﬁcation rates by an order of magnitude, but the combination of the two
resulted in intermediate denitriﬁcation rates (Reyes-Avila et al. 2004). Perhaps
one of the greatest questions from my study arises from noting that upon adding
a large amount of labile OC, we did not see more of an effect, as was the case in
Reyes-Avila et al. (2004). While I added very large amounts of QC by ecological
standards (1.5 mM or 10 mg/L), Reyes-Avila added 12,400 mg/L acetate, over
100-fold more than this study, which may in part explain why he saw a more
dramatic difference between the acetate additions. However, Cardoso et al.
(2006) added 500 (M acetate to slurries, and that only supported small amounts
of denitrification in sediments isolated from reactors that had been conditioned
with S compounds. Kelso et al. (1999) compared the effect of different OC
sources (including acetate and glucose) on N03“ reduction and the relative
importance of DNRA and denitrification. She found that the form of C matters to
the relative importance of the two pathways, and to overall N03' removal (Kelso
et al. 1999). In that study, acetate inhibited N03' removal relative to the N031
only control treatments, and glucose favored NH4+ as an end-product more so
than the other C sources (Kelso et al. 1999). Thus, the reaction of a particular
sediment to acetate may in large part depend on the microbial community that is
adapted to those conditions.

A few more studies have examined the effects of H28 concentrations on

N03' reduction without also accounting for the effects of a carbon source. Senga

163

 

et al. 2006 isolated two strains of N20 producers from brackish sediments and
subjected them to a range of sulﬁde concentrations (0->1500 pM). They found
that the amount of NH4+ produced increased over the H28 gradient, and the
greatest NH4" production occurred at 313 uM H2S, after which increasing H2S
concentrations led to little or no N03' reduction (Senga et al. 2006). This is
similar to what was found in the H28 only gradient in LP, wherein adding
increasing amounts of H28 resulted in increasing NH4+ production. However, I
cannot compare the effects of H28 on N2 production because Senga et al. 2006
did not directly measure N2 (it was estimated as the difference between the N
added and N accounted for) to make any inferences about the effect of H28 on
N2 ﬂux. Cardoso et al. 2006 did, however, examine the effects of H28 on N2
production. They found that increasing H2S from 250 to 1000 (M decreased
denitriﬁcation by 21-fold (Cardoso et al. 2006). The low H2S concentration in this
study, however, was higher than the highest H2S concentration in my gradient

assays, which makes it difﬁcult to compare the two studies.

Carbon and the Tiedje DNRA Hypothesis

It has been hypothesized that DNRA, which at one time was thought to be
predominantly fermentative, would be most important in highly reducing
environments that maintain anoxic conditions for long time periods (Tiedje et al.
1982, Tiedje 1988). Tiedje et al. (1982) also hypothesized that DNRA would be
favored in N03'-limited, labile-carbon rich environments while respiratory

denitriﬁcation would be favored under carbon-limited conditions. This hypothesis

164

has been supported by some studies examining DNRA across sites of differing
organic matter availability (Bonin 1996, Christensen et al. 2000), but has not
been tested in a manipulative fashion as in this study. If this hypothesis were
true, then we would expect that increasing levels of OC would lead to increasing
amounts of DNRA. There is some evidence for this in the LP sediments, but
none from the WP sediments (Figure 2). While adding OC in the WP sediments
stimulated DNRA, it only occurred when the QC was added together with H2S.
Across the OC gradient there was no increase in NH.“ production (all symbols at
zero H2S added in bottom panel of Figure 2). Also, adding increasing amounts of
DC with H28 in WP, generally decreased the amount of NH4+ produced. OC
clearly does inﬂuence DNRA in WP, but DNRA in this site is driven by the
interaction between OC and H28, not by either one alone. Additional support for
the assertion that DNRA is inﬂuenced by both OC and H28 comes from
examining the magnitude of the effects of both factors in both sites. H2S is
clearly as important (in the case of WP, Table 2) if not more important (as in LP,

1 Table 1) than DC in determining NH4” production (DNRA).

Evidence of S-linked N03' removal

Evidence for linkages between sulfur and nitrogen cycling has been found
in marine ecosystems (Fossing et al. 1995) where sulfate exists in much higher
concentrations. Work by Fossing et al. (1995) showed that large mats of
Thioploca off the coast of Chile were able to take up nitrate and store it in

vacuoles, and use it to oxidize sulﬁde in a chemolithoautotrophic reaction.

165

Brunet and Garcia-Gil ( 1996) invoked a similar explanation for patterns they saw
in a freshwater lake in Spain, wherein NH4+ production coincided with NO3' and
H28 depletion. Upon additions of the various S species as potential electron
donors in the presence of N03', they found that H2S produced only NH4+ and
N20, lending support to the idea that denitriﬁcation is inhibited by sulﬁde. Brunet
and Garcia-Gil (1996) further hypothesized that the presence of sulﬁde was the
main factor determining whether N03' was reduced to N2 or NH4”. The
experiments I report here provide additional evidence that the presence of H28
inﬂuences N03' reduction, particularly in freshwater ecosystems.

In sediments from both wetlands, adding H2S resulted in an increase in
8042' in both the gradient and timed experiments (Figures 7 and 16). This
production was much more pronounced in LP, where adding increasing amounts
of H28 led to increasing amounts of 8042’ in nearly a 1:1 molar ratio. WP, in
contrast, had SO47" production stimulated by slight amounts of H28, but the trend
did not continue to increase over the H28 gradient. In both sites and both
experiments, adding QC inhibited 8042' production (Figures 7 and 15), and had a
larger magnitude of effect on 8042' ﬂux than H28 in part due to this inhibitive
effect (Tables 1 and 2). This is especially clear in the LP gradient experiment
wherein adding increasing amounts of OC increasingly inhibited 8042‘
production, and actually stimulated 8042' reduction (removal). A similar effect
was also noted by Reyes-Avila et al. (2004), who also found that adding acetate
in combination with H2S forced the incomplete oxidation of H28 to S0 rather than

to 8042-.

166

Adding NO3' in combination with H2S stimulated H2S removal, and this
H2S removal often occurred by the ﬁrst sampling time in the timed assays (Figure
17). In LP, over half of the added H2S was gone within 12 hours, and in WP, the
added H2S disappeared by 12 hours. Between 12 and 24 hours, N03' was
exhausted and 3042' reduction began, which increased the H28 in the vial slightly
by the end of the experiment (Figure 17). There was no change in the H28 ﬂux in
either the N03‘ only or DC treatments. In the controls (no added QC, N03" or
H2S), reduction of 8042‘ from the added surface water caused an increase in H2S
concentration over the time course. Adding increasing amounts of H28 in the
gradient assays also led to increasing removal of H28 in both LP and WP. In LP,
nearly all of the added H2S was removed along the H28 gradient (Figure 6).
However, at the highest level of H28 addition, adding OC increasingly inhibited
H2S removal. This effect was not seen in WP, where QC stimulated more H2S
removal compared to the H28 gradient (black line, Figure 6). In both sites,
treatments along the 0C gradient (no H2S added) resulted in an increase in H2S,
presumably from 3042‘ reduction.

The simultaneous 8042' production and H28 removal in the presence of
N03; strongly suggests a link between the two cycles, but what happens to the
N02]? Some have hypothesized that the N03’ is reduced to NH4+ (Brunet and
Garcia-gil 1996, Otte et al. 1999, Sayama 2001), while others have determined
that the conversion can be to N2 (Sweerts et al. 1990). The stoichiometry of
these two reactions is different and could provide insight into which is occurring

in these experiments. If the N03' is converted to NH4”, there should be 1 mole of

167

NO3' and H28 consumed per mole of 8042' and NH] produced (Sayama et al.
2005). However, if N2 is produced, 8 moles of N02; consumption results in 5
moles of 8042’ production (Fossing et al. 1995). These stoichiometric equations
are described in more detail in Chapter 2.

If we examine, for example, the high-H2S-only treatment from LP, we see
that 4 umoles of N03“ and 5 umoles of H28 were consumed and approximately 7
umoles of 8042‘ were produced (Figure 8, top right). While this doesn’t match
perfectly with either of the two reactions discussed above, it more closely reﬂects
what we would expect if the reduction to NH4+ was occurring. LP had further
evidence of this reaction in the increasing NH4+ production along the H28
gradient (Figure 2 and Figure 8). In this case, adding ~5 umoles H2S resulted in
2-3 umole increase in NH4+ produced. While one reaction may dominate, it is
also likely that the two are not mutually exclusive and both are occurring to some
degree, perhaps shifting in response to changing H2S concentrations over the
course of the experiment. This lack of exclusivity between the two pathways is
even more likely in natural near-surface sediment environments which vary
greatly in H2S concentrations over both space and time. We have evidence for
both pathways (H2S coupled to both NH4+ and N2 production) occurring because
both end-products are affected by H28 in LP. In the LP gradient experiment,
addition of medium amounts of H28 resulted in increased N2 production (Figure
3); in the timed experiment, addition of medium amounts of H28 resulted in the

most N2 production at the end of the experiment (Figure 13).

168

Conclusions

Denitriﬁcation was the more important N03' reduction pathway in the low ambient
H2S site (WP), whereas DNRA and denitriﬁcation were of equal importance in the
high ambient H2S site (LP). DNRA, however, occurred in both sites, but under
different conditions and was particularly stimulated by adding either CC or H2S.
This study also provides evidence for the inﬂuence of H28 on N2 production,
presumably from S-Iinked denitriﬁcation, which is something that has not been
experimentally demonstrated in freshwaters to my knowledge. This work
highlights that there are other pathways of NO3' removal besides carbon-driven
denitriﬁcation in freshwater ecosystems. Thus the S and N cycles may be linked
in ways previously not appreciated. Since both N and S are heterogeneous
components of our landscape, it is safe to assume that not all wetlands process
N03‘ the same way and different pathways may be more or less important under
differing conditions. Greater efforts need to be made to understand what controls
N processing in addition to OC and 02, which have mainly been studied in the

context of denitriﬁcation.

169

 

References:

Bonin, P. 1996. Anaerobic nitrate reduction to ammonium in two strains isolated
from costal marine sediment: A dissimilatory pathway. FEMS Microbiology
Ecology 19:27-38.

Bonin, P., P. Omnes, and A. Chalamet. 1998. Simultaneous occurrence of
denitriﬁcation and nitrate ammoniﬁcation in sediments of the French
Mediterranean Coast. Hydrobiologia 389: 1 69-1 82.

Brunet, R. C., and L. J. Garcia-gil. 1996. Sulﬁde-induced diSsirnilatory nitrate
reduction to ammonia in anaerobic freshwater sediments. FEMS
Microbiology Ecology 21:131-138.

Burgin, A. J., and S. K. Hamilton. 2007. Have we overemphasized the role of
denitriﬁcation in aquatic ecosystems? A review of nitrate removal
pathways. Frontiers in Ecology and the Environment 5:89-96.

Cardoso, R. B., R. Sierra-Alvarez, P. RoWlette, E. R. Flores, J. Gomez, and J. A.
Field. 2006. Sulfide oxidation under chemolithoautotrophic denitrifying
conditions. Biotechnology and Bioengineering 95:1148-1157.

Christensen, P. B., S. Rysgaard, N. P. Sloth, T. Dalsgaard, and S. Schwaerter.
2000. Sediment mineralization, nutrient ﬂuxes, denitriﬁcation and
dissimilatory nitrate reduction to ammonium in an estuarine ﬁord with sea
cage trout farms. Aquatic Microbial Ecology 21:73-84.

Dannenberg, S., M. Kroder, W. Dilling, and H. Cypionka. 1992. Oxidation of H2,
organic compounds and inorganic sulfur compounds coupled to reduction

of O2 or nitrate by sulfate-reducing bacteria. Archives of Microbiology
158:93-99.

Fossing, H., V. A. Gallardo, B. B. Jorgensen, M. Huttel, L. P. Nielsen, H. Schulz,
D. E. Canﬁeld, S. Forster, R. N. Glud, J. K. Gundersen, J. Kuver, N. B.
Ramsing, A. Teske, B. Thamdrup, and O. Ulloa. 1995. Concentration and
transport of nitrate by the mat-forming sulfur bacterium Thioploca. Nature
374:713-715.

Golterman, H.L. and RS. Clymo. 1969. Methods for chemical analysis of
freshwaters. Blackwell, Oxford.

Graham, M. H., and M. S. Edwards. 2001. Statistical signiﬁcance versus ﬁt:
estimating the importance of individual factors in ecological analysis of
variance. Oikos 93:505-513.

170

Holmer, M., and P. Storkholm. 2001. Sulphate reduction and sulphur cycling in
lake sediments: a review. Freshwater Biology 46:431-451.

Kelso, B. H. L., R. V. Smith, and R. J. Laughlin. 1999. Effects of carbon
substrates on nitrite accumulation in freshwater sediments. Applied and
Environmental Microbiology 65:61 -66.

Kelso, B. H. L., R. V. Smith, R. J. Laughlin, and S. D. Lennox. 1997. Dissimilatory
nitrate reduction in anaerobic sediments leading to river nitrite
accumulation. Applied and Environmental Microbiology 63:4679-4685.

Megonigal, J., M. Hines, and P. Visscher. 2004. Anaerobic Metabolism: Linkages
to Trace Gases and Aerobic Processes. Pages 317-424 in W. H.
Schlesinger, editor. Biogeochemistry. EIsevier—Pergamon, Oxford, UK.

Nijburg, J. W., M. J. L. Coolen, S. Gerards, P. J. A. K. Gunnewiek, and H. J.
Laanbroek. 1997. Effects of nitrate availability and the presence of
Glyceria maxima the composition and activity of the dissimilatory nitrate-

reducing bacterial community. Applied and Environmental Microbiology
63:931-937.

Otte, S., J. G. Kuenen, L. P. Nielsen, H. W. Paerl, J. Zopﬁ, H. N. Schulz, A.
Teske, B. Strotmann, V. A. Gallardo, and B. B. Jorgensen. 1999. Nitrogen,
carbon, and sulfur metabolism in natural Thioploca samples. Applied and
Environmental Microbiology 65:3148—3157.

Paul, E. A., and F. E. Clark. 1996. Soil Microbiology and Biochemistry, 2nd
edition. Academic Press, San Diego.

Reyes-Avila, J. S., E. Razo-Flores, and J. Gomez. 2004. Simultaneous biological
removal of nitrogen, carbon and sulfur by denitriﬁcation. Water Research
38:3313-3321.

Sayama, M. 2001. Presence of nitrate-accumulating sulfur bacteria and their
inﬂuence on nitrogen cycling in a shallow coastal marine sediment.
Applied and Environmental Microbiology 67:3481-3487.

Sayama, M., N. Risgaard-Petersen, L. P. Nielsen, H. Fossing, and P. B.
Christensen. 2005. Impact of bacterial NO3- transport on sediment
biogeochemistry. Applied and Environmental Microbiology 71 :7575-7577.

Senga, Y., K. Mochida, R. Fukumori, N. Okamoto, and Y. Seike. 2006. N20
accumulation in estuarine and coastal sediments: The inﬂuence of H28 on
dissimilatory nitrate reduction. Estuarine Coastal and Shelf Science
67:231-238.

171

Silver, W. L., D. J. Herman, and M. K. Firestone. 2001. Dissimilatory nitrate

reduction to ammonium in upland tropical forest soils. Ecology 8222410-
2416.

Sweerts, J. P. R. A., D. Debeer, L. P. Nielsen, H. Verdouw, J. C. Vandenheuvel,
Y. Cohen, and T. E. Cappenberg. 1990. Denitriﬁcation by sulfur oxidizing
Beggiatoa spp mats on freshwater sediments. Nature 344:762-763.

Tiedje, J. M. 1988. Ecology of denitrification and dissimilatory nitrate reduction to
ammonium. Pages 179-244 in A. J. B. Zehnder, editor. Biology of
Anaerobic Microorganisms. Wiley & Sons, New York.

Tiedje, J. M., A. J. Sexstone, D. D. Myrold, and J. A. Robinson. 1982.
Denitriﬁcation - Ecological Niches, Competition and Survival. Antonie Van
Leeuwenhoek Journal of Microbiology 482569-583.

Vitousek, P. M., J. D. Aber, R. W. Howarth, G. E. Likens, P. A. Matson, D. W.
Schindler, W. H. Schlesinger, and D. G. Tilman. 1997. Human alteration of

the global nitrogen cycle: Sources and consequences. Ecological
Applications 7:737-750.

Wang, F. Y., and P. M. Chapman. 1999. Biological implications of sulﬁde in
sediment - A review focusing on sediment toxicity. Environmental
Toxicology and Chemistry 18:2526-2532.

Whitmire, S. L. 2003. Anaerobic biogeochemical functions of Michigan wetlands
and the inﬂuence of water source. Ph.D. Dissertation. Michigan State
University, East Lansing, MI.

172

CHAPTER 5

CONCLUSIONS

Nitrate is potentially processed by many different pathways in freshwater
ecosystems, including lakes, streams and wetlands. While some of these
pathways, such as respiratory denitriﬁcation, have been well-studied, others such
as dissimilatory nitrate reduction to ammonium (DNRA) and anaerobic
ammonium oxidation (anammox) have received relatively little scientiﬁc attention
in freshwaters. Furthermore, while the linkages between sulfur and nitrogen
cycling have been acknowledged and examined in marine-inﬂuenced
ecosystems, these linkages have not been well studied in freshwaters, in part
because of the perception that relatively low sulfate concentrations in freshwaters
render S cycling unimportant in overall ecosystem function. The dissimilatory
transformation of nitrate by chemolithoautotrophs, including anammox as well as
use of nitrate as an oxidizing agent by sulfur- and iron-oxidizing bacteria, are so
little known in freshwaters that thay can be considered novel pathways of nitrate
removal.

I found evidence of linkages between the sulfur and nitrogen cycles in
sediments from a diverse set of freshwater streams, lakes and wetlands based
on the push-pull tracer experiments (Chapter 2). I was able to show that the
amount of sulfate production relative to nitrate removal varied both within and
between different freshwater environments. Based on stoichiometric calculations

the sulfate that was produced could explain a substantial fraction of the overall

173

nitrate removal. Furthermore, by using stable isotopes (“SN-labeled NOa'), I
found that dissimilatory nitrate reduction to ammonium (DNRA) could account for
a large fraction of the overall nitrate removal, particularly in wetlands.

The push-pull experiments in Chapter 2 helped to convince me that the
phenomenon of sulfate production was indeed biologically driven and potentially
linked to N cycling, but I needed to conduct further experiments because I could
not directly compare multiple nitrate removal pathways, including denitriﬁcation
and DNRA. To discern the end-products of N03“ reduction (e.g., NH4“ or N2), I
again employed the use of stable isotopes, particularly of 15NO3‘ and 15NH4“, to
elucidate the relative importance of multiple pathways operating simultaneously
(Chapter 3). This is the ﬁrst study, to my knowledge, that simultaneously
estimates three pathways of nitrate removal. Additionally, it is one of a small
handful of studies that have examined the importance of anammox in freshwater
ecosystems. While I did not ﬁnd evidence of substantial rates of anammox in
any of my sites, I did discover that DNRA can be an important nitrate removal
pathway, and in some cases, could rival denitriﬁcation as a nitrate sink.

Once I understood that DNRA in particular was potentially important in
many of these sites, I sought to understand what factors control the relative
importance of denitriﬁcation vs. DNRA in freshwater sediments (Chapter 4). Two
forms of DNRA are known to occur: fermentative DNRA, thought to occur under
conditions of high labile carbon availability, and sulfur-driven DNRA, controlled by
H28 and other reduced sulfur compounds. Therefore, I wanted to test the effects

of carbon vs. sulﬁde in controlling both denitriﬁcation and DNRA, along the lines

174

of the hypotheses outlined in Chapter 1. I did this using an assay technique I
developed that could simultaneously manipulate carbon and sulﬁde gradients, as
well as estimate different end-products of nitrate reduction, and I conducted .
these assays in both a high and low ambient sulﬁde site. I found that both
carbon and sulﬁde were important in controlling nitrate removal rates and end-
products in both sites. While denitriﬁcation tended to be the more important
removal pathway in the low ambient sulﬁde site, DNRA was of equal importance
in the high ambient sulﬁde site. DNRA occurred in both sites, but under different
conditions and was particularly stimulated by adding both OC and H28. This
work also provided evidence for the inﬂuence of H28 on N2 production,
presumably from S-Iinked denitriﬁcation, which is something that has not been
experimentally demonstrated in freshwaters to my knowledge. I also saw sulfate
production increase over an increasing sulﬁde addition gradient, which further
suggests that the nitrogen and sulfur cycles are intricately linked, as I found in
Chapter 2.

While this study (Chapter 4) has given new insight to nitrate reduction and
processing, it also raises new questions regarding our understanding of N cycling
in freshwater ecosystems. In particular, I have become interested in
investigating whether the ambient sulﬁde concentration of a site changes how
nitrogen is cycled in that ecosystem. Stream sediments, for example, tendto be
lower in sulﬁde compared to wetland or lake sediments. Do high sulﬁde
ecosystems inherently cycle nitrogen, particularly nitrate, differently than low

sulﬁde ecosystems? We are currently performing the same assay used in

175

Chapter 4 on more sites, aiming for a total of 4-5 low and high ambient sulﬁde
sites, which will help me answer the previous question. Additionally, because we
know that sulﬁde and nitrate inputs are variable over time, there may be
differences in the relative importance of various nitrate removal pathways over
seasonal scales. This would be an interesting question to address using either
an assay approach, or 15N methods, or a combination of both.

These ﬁndings have signiﬁcant implications for our current understanding
of N cycling in freshwater ecosystems. The possible importance — or even
prevalence — of alternative nitrate removal pathways (DNRA, anammox) has
profound implications for our management of aquatic ecosystems to reduce
nitrate loads. Nitrate is the most mobile N form, so removal of nitrate by any of
the processes is important to downstream water quality, but permanent removal
by denitriﬁcation is most desirable.

Removal by other pathways can yield N2 in an alternative form of
denitriﬁcation, or they may result in transformation of the nitrate to something
other than dinitrogen gas (N2). Nitrate removal via Anammox still creates
dinitrogen gas as an end-product, but removes both a nitrate and an ammonium
ion in the process. In contrast, the conversion of nitrate to ammonium, as in
DNRA, creates an even more bioavailable N form, and one that tends to be less
mobile in soils and sediments. This converted ammonium can also be
transformed back to nitrate via nitriﬁcation. Additionally, if S-oxidizers prove to
take up much of the nitrate, then N cycling is closely linked to sulﬁde availability,

which is turn is linked to sulfate reduction. In freshwaters sulfate reduction may

176

 

be controlled by sulfate inputs, and sulfate is a ubiquitous pollutant in
industrialized and agricultural regions. If excess sulfate loading to freshwaters
actually enhances nitrate removal, then the controls on nitrate removal in
landscapes subject to S and N pollution become more complex than previously
thought. Much more research needs to be done on these alternative nitrate
removal pathways across a diversity of aquatic ecosystems. Most of what we
know about them is based on research done in marine ecosystems, and thus our
understanding of what controls these processes in freshwater ecosystems

subject to elevated nitrate inputs remains incomplete.

177