Content uploaded by Yuanyuan Liu
Author content
All content in this area was uploaded by Yuanyuan Liu on Jun 08, 2018
Content may be subject to copyright.
Nitrate bioreduction in redox-variable low permeability sediments
Sen Yan
a,b
, Yuanyuan Liu
b
,ChongxuanLiu
a,b,
⁎, Liang Shi
b
, Jianying Shang
b
, Huimei Shan
a,b
, John Zachara
b
,
Jim Fredrickson
b
, David Kennedy
b
, Charles T. Resch
b
, Christopher Thompson
b
, Sarah Fansler
b
a
China University of Geosciences, Wuhan 430074, China
b
Pacific Northwest National Laboratory, Richland, WA 99354, USA
HIGHLIGHTS
•Low permeability zones (LPZ) can
microbially remove nitrate in
groundwater.
•The rate and end product of nitrate
bioreduction vary within LPZ.
•Greenhouse gas N
2
O can be the end
product of nitrate bioreduction in LPZ.
•Organic carbon, denitrifier mass, and
gene expression are the controlling
factors.
GRAPHICAL ABSTRACT
abstractarticle info
Article history:
Received 25 July 2015
Received in revised form 21 August 2015
Accepted 21 August 2015
Available online xxxx
Editor: D. Barcelo
Keywords:
Nitrate bioreduction
Low permeability zone
Subsurface redox transitional sediments
Nitrous oxide
Organic carbon speciation
Kinetic model
Low permeability zone(LPZ) can play an important roleas a sink or secondary source in contaminant transport in
groundwater system. This study investigated the rate and end product of nitrate bioreduction in LPZ sediments.
The sediments were from the U.S. Department of Energy's Hanford Site, where nitrate is a groundwater contam-
inant as a by-product of radionuclide waste discharges.The LPZ at the Hanford site consists of two layers with an
oxidized layer on top and reduced layer below. The oxidized layer is directly in contact with the overlying
contaminated aquifer, while the reduced layer is in contact with an uncontaminated aquifer below. The experi-
mental results showed thatnitrate bioreduction rate and end-product differed significantly in the sediments. The
bioreduction rate in the oxidized sediment was significantly faster than that in the reduced one. A significant
amount of N
2
O was accumulated in the reduced sediment; while in the oxidized sediment, N
2
O was further
reduced to N
2
. RT-PCR analysis revealed that nosZ, the gene that codes for N
2
O reductase, was below detection
limit in the reduced sediment. Batch experiments and kinetic modeling were performed to provide insights
into the role of organic carbonbioavailability,biomass growth, and competition between nitrate and its reducing
products for electrons from electron donors. The results revealed that it isimportant to considersediment redox
conditions and functional genes in understanding and modeling nitrate bioreduction in subsurface sediments.
Science of the Total Environment 539 (2016) 185–195
⁎Corresponding author at: Pacific Northwest National Laboratory, PO Box 999, MSIN: K8-96, Richland, WA 99354, USA.
E-mail address: chongxuan.liu@pnnl.gov (C. Liu).
http://dx.doi.org/10.1016/j.scitotenv.2015.08.122
0048-9697/Published by Elsevier B.V.
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
The results also implied that LPZ sediments can be important sink of nitrate and a potential secondary source of
N
2
O as a nitrate bioreduction product in groundwater.
Published by Elsevier B.V.
1. Introduction
Nitrate is a major contaminant in groundwater worldwide (Rivett
et al., 2008). Nitrate attenuation in groundwater is primarily controlled
by biogeochemical processes that reduce nitrate (NO
3
−
) to gaseous ni-
trous oxide (N
2
O), dinitrogen (N
2
), or ammonium (NH
4
+
)(Zumft,
1997). In aquatic systems, nitrate bioreduction predominately occurs in
sediments (Laverman et al., 2006), removing nitrate entering rivers
(Laverman et al., 2010) and oceans (Ward, 2013). Extensive research
has been performed to understand various processes and factors control-
ling the rate, extent, and intermediate and end products of nitrate
bioreduction in environments (Rivett et al., 2008; Seitzinger et al.,
2006). The rate and extent of nitrate bioreduction are affected by dis-
solved oxygen concentration (Rassamee et al., 2011), organic carbon spe-
ciation and contents (Babbin et al., 2014; Lu and Chandran, 2010), and
denitrifier abundance and gene expression (Zhang et al., 2014). Other fac-
tors including nitrate concentration, nutrient availability, pH, and temper-
ature can also affect the rate and intermediate products of nitrate
bioreduction (Firestone et al., 1980; Pan et al., 2012; Rivett et al., 2008).
In reduced sediments, the presence of sulfide and Fe(II) species may
stimulate nitrate bioreduction via sulfide or Fe(II) oxidation (Vaclavkova
et al., 2014; Zhang et al., 2009).Arecentstudyfoundthattheendproduct
of nitrate bioreduction in oligotrophic lake sediments is largely deter-
mined by sediment redox conditions (Small et al., 2014). However, the
mechanisms and process-level understanding on the effect of redox
conditions on nitrate bioreduction rate in the subsurface and its relation-
ship with microbial functional gene are not well-understood.
Nitrate contaminant in the subsurface can be attenuated via denitri-
fication (Rivett et al., 2008), which refers to an assemblage of nitrate, n i-
trite, NO, and N
2
O respiration in the absence of oxygen (Zumft, 1997).
N
2
O can be the end product of incomplete denitrification (Zumft,
1997). Various factors have been identified that lead to N
2
O emission
as the end product of nitratebioreduction. The presence of trace oxygen,
minor free nitrous acid species, low pH, high initial nitrate, and nitrite
concentrations can individually or collectively inhibit N
2
Oreductionto
N
2
(Firestone et al., 1980; Kampschreur et al., 2009; Laverman et al.,
2010; Pan et al., 2012; Rassamee et al., 2011). On the other hand, high
organic carbon concentration, low nitrate and nitrite concentration,
and long duration of anaerobiosis favor N
2
as the end product
(Butterbach-Bahl et al., 2013; Kampschreur et al., 2009; Weymann
et al., 2010). These findings were primarily derived from studies in
wastewater treatment systems, agricultural soils, streams and rivers.
Limited studies indicated that groundwater systems are also important
sources or sinks of N
2
O(Ronen et al., 1988; Weymann et al., 2010). The
relative importance of N
2
O production in groundwater to land/surface
N
2
Oflux, the mechanisms leading to N
2
O accumulation in groundwater,
and reactive transport of N
2
O in sediment-groundwater system have
not been well studied.
Low permeability zones (LPZs) are important subsurface units that
can retard and attenuate contaminant migration in groundwater
(Arighi et al., 2005; Robertson et al., 1996; Liu and Ball, 2002). Nitrate
bioreduction in LPZ sediments is, however, an under-studied area
(Rivett et al., 2008). Limited studies indicate that LPZ sediments may
strongly affect the fate and transformation of nitrate in groundwater
(Lin et al., 2012a; Lee et al., 2012). These studies revealed that nitrate
bioreduction primarily resulted from heterotrophic activities, and the
effect of lithoautotrophic Fe(II)- and sulfide-oxidizing microorganisms
(such as E. shelobolina)wasnegligible(Percak-Dennett and Roden,
2014). A diffusion-based reactive transport model had been established
by incorporating nitrate bioreduction kinetics that provided important
insights into nitrate migration and attenuation in LPZ (Percak-Dennett
and Roden, 2014). These previous studies in the LPZ sediments, howev-
er, focused on the loss of nitrate without considering the reduction
kinetics of nitrite and N
2
O. Field monitoring at the U.S. Department of
Energy's Hanford site, however, found the accumulation of N
2
Oin
groundwater in LPZ and nearby aquifer zones, indicating that it is
important to consider the bioreduction of nitrateand its reduction prod-
ucts. In addition, the redox potential of LPZ sediments often varies as a
function of distance from nearby aquifer. Consequently, the objective
of this study is to investigate the bioreduction kinetics of nitrate and
its reduction products in LPZ sediments with different redox potentials.
Batch experiments using both the oxidized and reduced sediments
were performed to characterize the bioreduction rates of nitrate and
its reduction products, and to determine the end product of nitrate
bioreduction (e.g., N
2
O, N
2
,orNH
4
+
). Real-time polymerase chain reac-
tion (RT-PCR) analysis was performed to investigate the mechanisms
controlling N
2
OorN
2
as the end product of nitrate bioreduction. Con-
trolled experiments and kinetic modeling were performed to provide
insights into the effect of organic carbon content and speciation, micro-
bial growth, and various model parameters on nitrate bioreduction.The
results provided important insights into the effect of sediment redox
state, indigenous organic carbon, and long-term adaption of microbial
community on the rate and end product of nitrate bioreduction in LPZ
sediments as subsurface units.
2. Materials and methods
2.1. Sediments
The sediments used in this study were obtained from a LPZ at the U.S.
Hanford 300 Area Integrated Field Research Challenge (IFRC) site (http://
ifchanford.pnl.gov), where nitrate is a contaminant in the overlying aqui-
fer (Bjornstad et al., 2009). The LPZ consists of an upper layer of yellow-
brown Miocene–Pliocene-aged lacustrine fine-grained oxidized sediment
(OS) and a lower layer of dark gray to black Miocene–Pliocene-aged la-
custrine fine-grained reduced sediment (RS). The LPZ separates the over-
lying aquifer formation from the underlying Ringold formation (Lin et al.,
2012b). The OS and RS were collected aseptically from ca. 18.0–18.3 m
depth and ca. 18.6–18.9 m depth of core C-6209 recovered during drilling.
The sediment core was stored in a −80 °C freezer and thawed at 4 °C
before use for the experiments. The geochemical andmicrobial communi-
ty properties in the sediments have been characterized previously as de-
scribed elsewhere (Lee et al., 2012; Lin et al., 2012a, 2012b;
Percak-Dennett and Roden, 2014; Peretyazhko et al., 2012). Important
properties from the previous characterization and the present study are
summarized in Table 1.Briefly, the RS contains over 9 times of organic
carbon and HCl-extractable Fe(II) than the OS. The biomass and microbial
diversity in the RS are much lower than those in the OS. The biomass
numbers determined by most probable number culture method (Lin
et al., 2012a) show that denitrifiers, aerobes, and fermenters reside in
both OS and RS with higher abundance in the OS. Sulfate reducer popula-
tion is similar in the OS and RS.
2.2. Nitrate reduction experiments
2.2.1. Batch experiments
Batch experiments were performed in duplicates to determine
nitrate bioreduction rate and to identify factors controlling the rate
and end product of nitrate bioreduction in the sediments. Sediment
suspensions were prepared in 160 mL glass bottles, by mixing a
186 S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
known weight (from 10 to 40 g) of sediment with 50 mL anoxic filter-
sterilized synthetic groundwater (SGW), which mimicked Hanford
300 area groundwater chemical composition (Liu et al., 2013). The
glass bottles were sealed with butyl rubber stoppers and all prepara-
tions were made in an anoxic chamber, filled with N
2
(Innovative
Technology Inc.). A summary of theexperimental conditions is provided
in Table S1 in Supplementary Information (SI). Sediment/SGW ratio
(200, 400, or 800 g/L) and initial nitrate concentration (0.46–2 mM)
were varied to investigate the effects of sediment-associated initial
biomass and organic carbon content, and nitrate concentration on the
rate of nitrate bioreduction. In the reactors with sediment/SGW ratio
at 200 g/L, a mixed organic carbon (OC) source (0.9 mM acetate,
0.6 mM lactate, and 0.3 mM glucose) was spiked to evaluate the rate
limitation of organic carbon. Control experiments were performed in
duplicates by autoclaving the sediments at 120 °C for 30 min and for
three times during three consecutive days to examine whether nitrate
and nitrite could be reduced abiotically.
The sediment suspensions were incubated at room temperature on a
reciprocal shaker operated at 300 rpm, within the anoxic chamber
where oxygen levels were maintained below 0.1 ppm. Suspension was
sampled periodically using 3 mL syringes fit with 18G needles. Pretests
were performed to determine sampling frequency and timing. For each
suspension sample, 0.25 mL subsample was mixed with 0.25 mL
40 mM Ethylenediaminetetraacetic acid (EDTA) for Adenosine Triphos-
phate (ATP) analysis, and the remaining 2.75 mL subsample was filtered
(0.2 μm). Among the filtrate, 1.5 mL was used for analyzing NO
3
−
,NO
2
−
,
SO
4
2−
and NH
4
+
concentrations, and 0.5 mL was acidified with 5 μL
1.25 M H
2
SO
4
for organic carbon analysis in the incubations with OC
spike. At the end of kinetic experiments, the remaining suspensions
were collected, frozen in liquid nitrogen, and stored at −80 °C until
DNA extraction. In the reactors spiked with OC, N
2
O in the head space
was sampled before aqueous sampling. To preserve headspace gas pres-
sure, 1 mL of N
2
was injected into the reactor using a 1 mL gas-tight
syringe before 1 mL gas sample was taken from the headspace. The
1 mL gas sample was mixed with 7 mL N
2
in a sealed serum vial for
N
2
O analysis. The resulting dilution of the headspace gas before taking
samples was considered in calculating N
2
O concentrations. After the gas
sampling, 3 mL of N
2
was injected into the reactors before taking 3 mL
suspension samples for aqueous phase analysis as described above.
2.2.2. Sample analysis
NO
3
−
,NO
2
−
,andSO
4
2−
concentrations in aqueous samples were de-
termined using ion chromatograph (Dionex ICS-2000). The analysis
was performed using isocratic 15-min elution with 22 mM KOH eluent
at 30 °C at a flow rate of 1 mL min
−1
. A IonPac AG18 guard column and
an IonPac AS18 separate column were used with an anion suppressor.
Standards were made from Spex CertiPrep solution (1000 mg/L anion
standards, Metuchen, NJ) with dilutions calibrated from 0.19 to
120 mg/L (except for nitrite, from 0.06 to 40 mg/L). Aqueous NH
4
+
con-
centrations were colorimetrically analyzed using the Nessler reagent
(Streuli and Averell, 1970). The acetate, lactate and glucose concentra-
tions in aqueous samples were determined by HPLC (Agilent 1100)
with a 300 × 7.8 mm Aminex HPX-87 H column (Bio-Rad, Hercules,
CA), a 0.008 N H
2
SO
4
mobile phase with a flow rate of 0.6 mL/min and
variable wavelength detector (VWD) at 210 nm for acetate and lactate
and refraction index detector (RID) for glucose. The linear ranges for ac-
etate, lactate, and glucose calibrated from 0.113 mM to 0.9 mM,
0.075 mM to 0.6 mM, and 0.038 mM to 0.3 mM, respectively. We didn't
measure dissolved organic carbon in the reactors without OC spike,
since a few pretests using the same sediments found that the
dissolved organic carbon was below detection limit. ATP concentration
was measured using a luminometer following Promega's protocol
(Stanley, 1989). Total organic carbon contents in the sediment samples
were determined using Carbon analyzer (Shimadzu SSM-5000A). N
2
O
concentrations in gas samples were determined using gas chromato-
graph (SRI 8610C Greenhouse Gas Monitor) equipped with a
63
Ni
Table 1
Biogeochemical properties in the oxidized sediment (OS) and reduced sediment (RS).
Sediment C
org
Fe(II)
HCl a
S
pyritic a
Fe
total a
Denitrifiers
b
Fe reducers
b
Aerobes
b
Fermenters
b
sulfate reducers
b
OTUs
c
16S rRNA
nosZ
(%) (μmol/g) (μmol/g) (μmol/g) (cells/g) (copies/g)
OS 0.05 9.4 ± 0.4 nd 671 2.67 × 10
3
1.71 × 10
3
4.60 × 10
3
4.6 × 10
3
1.71 × 10
3
325 nd
d
43,195 ± 4170
e
112,662 ± 16,086
f
nd
d
3215 ± 775
e
34,536 ± 6,078
f
RS 0.51 85.1 ± 0.5 86.6 679 1.73 × 10
3
nd 1.73 × 10
2
1.95 × 10
2
1.73 × 10
3
50 nd
d
nd
e
1509 ± 1173
f
nd
d
nd
e
nd
f
a
The Fe(II)
HCl
and Fe
total
data were adopted from Peretyazhko et al. (2012);
b
From Lin et al. (2012a);
c
OTUs: operational taxonomic units at the 97% identity level (from Lin et al., 2012b);
d
In fresh sediments before incubation;
e
In incubated sediments without organic carbon spike;
f
In incubated sediments with organic carbon spike.
Note that the errors of nosZ and 16S rRNA data are standard deviation of three RT-PCR measurements.
187S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
electron capture detector and a flame ionization detector. A 1-mL gas
sample loop was used to inject samples onto the packed separation col-
umns, which consisted of 2 m Haysep-Dand and 1 m Shincarbon joined
with a 30 cm length of 1/8〞outside diameter (OD) copper tubing. The
N
2
carrier gas pressure was set at 20 psi, and the column temperature
was 100 °C. The calibration range for N
2
O was from 0.017 to
4.266 ppm. For each analysis, 3 mL gas was injected into the 1 mL sam-
ple loop using 5 mL gas-tight syringe, which was flushed with high-
purity N
2
gas (99.999%) twice before sampling. Blanks and standards
were run regularly between samples. Standard recoveries ranged from
96%to101%forN
2
O (mean = 99%), indicating no contamination from
atmospheric N
2
O during sampling and analysis.
The DNA in the initial and incubated sediments (from 8 to 12 g)
were extracted following an established procedure (Lin et al., 2012a;
Lin et al., 2012b). Briefly, the cells were lysed by bead-beating the sedi-
ments with 70 °C lysis buffer, sodium dodecyl sulfate, and casein solu-
tion for 10 min, and heated in 70 °C in water bath for 20 min, and
then centrifuged at 5000 gfor 10 min. The supernatant containing
DNA was collected and then mixed with a solution containing 1 part
3 M Na-acetate and 7 parts bioreagent 2-propanol (N99.5%) to precipi-
tate DNA. The mixture was centrifuged at 9000 gfor 60 min at 4 °C,and
the precipitates were collected. The DNA in the precipitates was first
extracted using phenol:chloroform:isoamyl alcohol (25:24:1), and
then twice extracted using chloroform/isoamyl alcohol (24:1) and 2-
propanol. The DNA was finally diluted in 100 μLof10mM
tris(hydroxymethyl)aminomethane. The plasmid 16S rRNA and nosZ
gene standards were prepared following an established protocol de-
scribed elsewhere (Lin et al., 2012b). All the quantitative PCR reactions
were performed with a StepOnePlus™real-time PCR system (Applied
Biosystems Inc., Foster City, CA) programmed for 40 cycles. Power
SYBR® green PCR master mix was used for all the samples assays, and
16S rRNA and nosZ gene fragments were included in each run to estab-
lish standard curves. The primers 27F and 1492R, nosZ2-F and nosZ2-R
were used to amplify 16S rRNA and nosZ gene, respectively. After
amplification, the melting curves of amplicons were analyzed to ensure
that a single homogenous product was generated. The results herein
were reported as the gene copies per gram sediment.
2.3. Mathematical interpretation of observations
The half-reactions involved in the nitrate bioreduction are provided
in Table S2 in SI. The overall reactions for the nitrate and nitrite reduc-
tion can be assembled from the half-reactions:
1. Nitrate reduction to nitrite
1
4CH2Oþ1
28 þ13
28 fe;NO−
3→NO−
2
NO−
3¼1−fe;NO−
3→NO−
2
28 C5H7O2Nþfe;NO−
3→NO−
2
2NO−
2
þ1
8HCO−
3þ5
28 fe;NO−
3→NO−
2−3
56
CO2
þ5
56 þ1
28 fe;NO−
3→NO−
2
Hþ
þ1
56 þ3
28 fe;NO−
3→NO−
2
H2O
ð1Þ
2. Nitrite reduction to dinitrogen
1
4CH2Oþ1
26 þ23
78 fe;NO−
2→N2
NO−
2¼1−fe;NO−
2→N2
26 C5H7O2Nþfe;NO−
2→N2
6N2
þ1
8HCO−
3þ5
26 fe;NO−
2→N2−7
104
CO2
þ9
104 −23
78 fe;NO−
2→N2
Hþ
þ1
104 þ11
39 fe;NO−
2→N2
H2O
ð2Þ
where f
e
is the fraction of total number of electrons generated from
organic carbon oxidation for the respiration (i.e., nitrate or nitrite
reduction). Organic carbon speciation in the sediments was unknown
and a molecular formula CH
2
O was used for modeling purpose to
balance elemental composition in the reactions. Similarly, C
5
H
7
O
2
N
was used to describe elemental composition of biomass in the overall
reactions. Reaction 1 is the overall reaction describing nitratereduction
to nitrite in coupling with organic carbon oxidation and biomass pro-
duction. Reaction 2 is the overall reaction that merges intermediate re-
actions involving N
2
O (Table S2). These overall reactions have been
used widely for mass and electron balance in modeling nitrate
bioreduction in wastewater treatment engineering and agricultural sys-
tems (Ahn, 2006; Mastrocicco et al., 2011; Matějůet al., 1992; McCarty
et al., 1969; Rittmann and McCarty, 2001; U.S. EPA, 1993). As described
and discussed in the result section, however, N
2
O, instead of N
2
was the
end product of nitrate bioreduction in the RS. Consequently, in this
study, reaction 2 is replaced with the following reaction to balance
overall mass and electron in describing nitrite reduction in the RS:
1
4CH2Oþ1
26 þ6
13 fe;NO−
2→N2O
NO−
2¼1−fe;NO−
2→N2O
26 C5H7O2Nþfe;NO−
2→N2O
4N2O
þ1
8HCO−
3þ5
26 fe;NO−
2→N2O−7
104
CO2
þ9
104 −6
13 fe;NO−
2→N2O
Hþ
þ1
104 þ19
52 fe;NO−
2→N2O
H2O
ð3Þ
A saturation-type model for heterotrophic growth systems wasused
to describe the rate-limiting roles of biomass, organic carbon, and elec-
tron acceptor concentrations during nitrate bioreduction (Rittmann and
McCarty, 2001). The model considers nitrate reduction to nitrite and ni-
trite reduction to dinitrogen in coupling with organic carbon oxidation
and biomass growth. The overall rates for nitrate and nitrite reduction
were described using a multiplicative Monod model (Rittmann and
McCarty, 2001):
rNO−
3¼qNO−
3XCoc
KNO−
3
s;oc þCoc
CNO−
3
Ks;NO−
3þCNO−
3
ð4Þ
rNO−
2¼qNO−
2
1þCNO−
3=KI
XCoc
KNO−
2
s;oc þCoc
CNO−
2
Ks;NO−
2þCNO−
2
ð5Þ
dCi
dt ¼X
j¼NO−
3;NO−
2
αj
irjð6Þ
dX
dt ¼X
j¼NO−
3;NO−
2
βjrj−bX ð7Þ
where qj(mmol/g/d) is the maximum specific rate of utilization for
electron acceptor j(j=NO
3
−
or NO
2
−
); X(g/L) is the biomass concentra-
tion; C
i
(mM) is the concentration of substrates (organic carbon, NO
3
−
and NO
2
−
); K
js,oc
(mM) is the half-maximum rate constant with respect
to organic carbon when electron acceptor jwas used; K
s,j
(mM) is the
half-maximum rate constant with respect to electron acceptor j;K
I
(mM) is the inhibition constant; α
i
j
and β
j
, which are functions of f
e
,
are the stoichiometric coefficients of substrate ior biomass Xin the
overall nitrate or nitrite reduction reactions, respectively; b(1/d) is
the endogenous decay coefficient.
In the modeling, literature values for parameters qNO−
3,qNO−
2,K
s,oc
NO
3
−
,
K
s,oc
NO
2
−
,K
s,NO
3
−
,K
s,NO
2
−
, and bwere first tried (Almeida et al., 1997; Ni
et al., 2011, 2013; Pan et al., 2012, 2013; Wang et al., 2003), and if neces-
sary fine-tuned to match experimental results. Parameters f
e,NO
3
−
→NO
2
−
,
f
e,NO
2
−
→N
2
(or f
e,NO
2
−
→N
2
O
for the RS), and K
I
were fitted to evaluate
whether it is necessary to consider biomass growth and competitive
inhibition during the nitrate and nitrite reduction in the LPZ sediments.
These parameters were estimated by simultaneously matching the
measured nitrate and nitrite concentrations in the reactors with
188 S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
variable sediment/SGW ratios and initial nitrate concentrations. The
goodness of the model fit was quantified by minimizing the sum of
squares of the differences between the simulated and measured values,
i.e. S¼∑n
i¼1ðCi−CiÞ2(Bard, 1974).
For the convenience to estimate relative standard deviation of the
estimated parameters (σ(θ
k
)/θ
k
), Eq. (6) was rewritten with respect to
fitting parameters:
dCi
dt ¼fθ;Ci;tðÞ ð8Þ
where C
i
is the concentration ofNO
3
−
or NO
2
−
and θis a vector containing
fitting parameters. The variances for the estimated parameters can be
calculated from a covariance matrix of the parameter estimates (Bard,
1974; Beck and Arnold, 1977; Liu and Zachara, 2001):
Cov θðÞ¼ JTV−1J
−1ð9Þ
where Cov(θ) is a covariance matrix of the estimated parameters.V
−1
is
the inverse of the covariance matrix of the measurement errors. Assum-
ing that measurements are independent, then Vis a diagonal matrix
with its diagonal element assumed to be proportional to the sum of
squares of the errors between the measurements and simulations
(V=SI). Superscript Tin Eq. (9) denotes the matrix transpose of sensi-
tivity coefficient matrix J:
J¼
∂S1
∂θ1
⋯∂S1
∂θL
⋮⋱⋮
∂Sn
∂θ1
⋯∂Sn
∂θL
2
6
6
6
4
3
7
7
7
5
ð10Þ
In the covariance matrix Cov(θ), the value of the diagonal element is
the square of standard deviation of θ
k
,i.e.σ
2
(θ
k
)(Bard, 1974; Beck and
Arnold, 1977). The relative standard deviation of the estimated param-
eters (σ(θ
k
)/θ
k
) will be used to discuss the uncertainty of a parameter
estimate. The confidence region volume (CRV) for a set of the estimated
parameters is proportional to |Cov(θ)|
1/2
.
3. Results and discussion
3.1. Nitrate bioreduction in untreated OS and RS
Nitratewas reduced in both untreated OS andRS sediments with rel-
atively faster rates in the OS (Fig. 1aandd).Nonitratewasreducedin
the controls using the autoclaved sediments (Fig. S1 in SI), indicating
that nitrate reduction was microbially mediated. Nitrite was an inter-
mediate product of nitrate bioreduction as its concentration first in-
creased with decreasing nitrate concentration, and then decreased
(Fig. 1b and e) when nitrate was depleted or nitrate reduction rate
slowed with time (Fig. 1a and d). Nitrite reduction was also microbially
mediated because no nitrite reduction occurred in the controls (Fig. S2).
Unlike nitrate bioreduction, the nitrite bioreduction rate was similar in
both sediments (Fig. S2). Consequently, duringthe nitrate bioreduction,
the accumulated nitrite concentration was higher in the OS than that in
the RS (Fig. b and e). The ATP concentrations had similar trends as those
for nitrite concentrations (Fig. 1c and f) with their peak concentrations
shifted toa later time. The results indicated that initial metabolic energy
production was faster than consumption, leadingto the accumulation of
ATP. When the rates of nitrate and nitrite bioreduction slowed down,
energy consumption became faster than its production.
The fasterrate of nitrate bioreduction in the OS relative to the RS was
unexpected in light of the general understanding that lower redox con-
ditions favor nitrate bioreduction (Small et al., 2014). Redox condition
can affect the nitrate bioreduction rate by influencing denitrifier popu-
lation and gene expression, and electron donor concentration and
speciation. Higher redox potential would decrease the rate of gene ex-
pression for denitrifying reductase production (Zhang et al., 2014),
and would also cause the oxidation of electron donors including re-
duced iron and sulfur species and organic carbon, thus decreasing
their capacity and bioavailability for nitrate bioreduction and lead to a
lower rate of nitrate bioreduction (Van Trump et al., 2011). The faster
nitrate bioreduction rate in the OS, despite that the RS contained over
9 times of organic carbon than the OS, was, however, qualitatively
consistent with the variation in denitrifier population in the sediments
(Table 1). A likely reason for the lower population of denitrifiers in the
RS is the long-term adaption of denitrifiers at the site where the low
permeability of the LPZ and faster nitrate bioreduction in the OS made
nitrate less available for the RS, which locates below the OS. Conse-
quently, the microbial community in the RS is adapted to low or no
nitrate condition.
The rates of nitrate bioreduction generally increased with increasing
sediment/SGW ratio in both OS and RS (Fig. 1), indicating that a higher
sediment-associated initial organic carbon content and biomass led to a
higher nitrate bioreduction rate. In the OS, nitrate and its reduction
product nitrite were reduced completely at sediment/SGW ratio of
800 g/L, but only partially at 200 g/L (Fig. 1a and b). In the RS, nitrate
was only bio-reduced partially at all sediment/SGW ratios (Fig. 1dand
e). The partial nitrate bioreduction that depended on the initial
sediment/SGW ratio could be attributed to the limitation of organic
carbon in the sediment. When organic carbon was higher in the higher
solid/SGW ratio reactor, more nitrate was reduced. Based on the
stoichiometric balance as described by reactions 1–3 and the extents
of nitrate and nitrite reduction (Fig. 1), only 4.3% and 0.22% of the
indigenous organic carbon was involved or bioavailable for nitrate
bioreduction in the OS and RS, respectively. The much lower bioavail-
able organic carbon than the total organic carbon in both sediments
suggested that most of the labile organic carbon had been consumed
in field.
3.2. Effects of organic carbon and initial nitrate concentration on nitrate
bioreduction
Spiking organic substrates significantly increased the rates of
nitrate and nitrite bioreduction in the OS (Fig. 2), confirming that or-
ganic carbon was a key rate-limiting factor. Without added exoge-
nous organic carbon, nitrate and nitrite concentrations stabilized
after 10–20 days of incubation depending on initial nitrate concen-
trations, and only 25% of nitrate was reduced at the end of experi-
ment (Fig. 2a). With the spiked organic substrates, nitrate was
completely reduced within 3 days, and nitrite was produced and
then completely reduced within 7 days (Fig. 2b). The total concen-
tration of spiked organic carbon was 0.324 mg C/g of sediment
after normalizing to the sediment mass. This value represented 65%
of total indigenous organic carbon in the OS (Table 1), indicating
that organic carbon bioavailability was an important factor influenc-
ing nitrate bioreduction. Moreover, it is interesting to observe that
nitrite in un-spiked OS reactors (Fig. 2a) accumulated to a common
asymptotic value (~ 0.2 mM) as a result of ~ 0.2 mM of nitrate
reduction, despite with different initial nitrate concentrations. The
common asymptotic accumulation of nitrite is likely due to the
incomplete nitrate bioreduction limited by the low organic carbon
content and bioavailability in the OS.
The rate-limiting effect of organic carbon was consistent with
other reports in the literature (Pulou et al., 2012; Weymann et al.,
2010). In some river and stream sediments, however, nitrate
bioreduction via denitrification was not limited by organic carbon
content (Herrman et al., 2008; Jha and Minagawa, 2013; Laverman
et al., 2010). Analysis of these different cases indicated that
10mgC/gofsedimentwastheboundarydividingwhetherorganic
carbon would limit denitrification (Herrman et al., 2008). When
organic carbon content is lower than 10 mg C/g of sediment,
189S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
denitrification would be limited by organic carbon. The total organic
carbon contents were 0.6 and 5.3 mg C/g of sediment in the OS and
RS, respectively, consistent with this classification of the rate-
limiting role of organic carbon. Furthermore, our results highlighted
that organic carbon bioavailability played even more important role
than the total organic carbon in limiting nitrate bioreduction.
Fig. 1. Experimental and modeling results of nitrate bioreduction in the OS(plots a-c)and RS (plots d–f) suspensions under variable sediment/SGW ratios (200, 400, 800 g/L) and initial
nitrate concentrations (0.46 and 2.0 mM). The figure shows nitrate reduction (plots a and d), nitrite production and reduction (plots b and e), and measured ATP and simulated biomass
(plots c and f). Symbols are experimental results and lines are the simulated results. The error bars denote the standard deviations of duplicate experiments.
Fig. 2. Nitrate reduction,and nitrite production and reduction in the OS suspensions with different initialnitrate concentrations, and with/without spiked organic carbon (OC). Sediment/
SGW ratio= 200 g/L. Plota: nitrate and nitriteconcentrationswithout OC spike(symbol ■for NO
3
−
and symbol□for NO
2
−
under initial NO
3
−
= 0.46 mM; symbol▲for NO
3
−
and symbol△
for NO
2
−
under initial NO
3
−
=2mM),Plotb: nitrate and nitrite concentrations with OC spike (symbol ■for NO
3
−
and symbol □for NO
2
−
under initial NO
3
−
= 0.46 mM; symbol●fo r NO
3
−
and symbol ○for NO
2
−
under initial NO
3
−
= 1.5 mM; symbol ▲for NO
3
−
and symbol △for NO
2
−
under initial NO
3
−
= 2 mM), plot c,dand e: spiked OC compound concentrations corre-
spondingto nitrate reduction in plot bunderdifferent initialNO
3
−
concentrations(symbol ■for ace tate, symbol ●for lactate, and symbol▲for glucose).The error bars denotethe standard
deviation of duplicate experiments.
190 S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
Among spiked organic substrates, glucose and lactate were preferen-
tially used, and then acetate (Fig. 2c–e). The sequence of organic carbon
degradation and assimilation by microorganisms is an important, but
unresolved question in understanding and predicting microbial activi-
ties in soils and sediments (Bahr et al., 2011). Free energy, molecular
structure, and redox state of organic compounds have been proposed
as major factors governing the sequence of their utilization (Bahr
et al., 2011). Glucose, lactate and acetate have different free energies
and molecular structure, and are the most commonly employed OC
sources for denitrification. The results in Fig. 2c indicated the incongru-
ent consumption of these OC sources, and were approximately consis-
tent with the free energies involved in the respiration reaction of OC
oxidation and nitrate reduction as glucose (−2686 kJ/reaction) yielded
more energy than lactate (−1260 kJ/reaction), and lactate more than
acetate (−843 kJ/reaction) (Cervantes, 2009). However, when normal-
izing the reaction free energy to each carbon in the organic compounds,
the difference in reaction free energy using these different organic
carbon species was small (420 to 447 kJ/reaction). The sequence of
organic compound degradation in Fig. 2c was also in contrast to the
molecular structure explanation as acetate has a simpler molecular
structure than lactate and glucose. It was not consistent with the
redox potential trend either because the average redox state of carbon
is the same (valence = 0) in these three organic compounds. Further
research is needed to understand factors controlling the degradation
and assimilation of organic substrates.
Spiking organic carbon also enhanced nitrate bioreduction in theRS,
but to a much lesser degree than the OS (Fig. 3). The rates of nitrate
bioreduction in the RS with organic carbon addition (Fig. 3b) were
10–20% higher than those without spiked organic carbon (Fig. 3a).
This result was in contrast to the OS where the rate of nitrate
bioreduction increased 50 times after organic carbon addition (Fig. 2).
The result, together with that increasing sediment/SGW ratio signifi-
cantly increased the rate of nitrate bioreduction (Fig. 1d and e), indicat-
ed that nitrate bioreduction was more limited by denitrifier population
and gene expression in the RS (Table 1). The sequence of organic carbon
degradation in the RSwas, however, the sameas that in the OS (Fig. 3c).
Sulfate concentration change was negligible (Fig. S3) during nitrate
bioreduction. This observation is consistent with the result of long-
term (3-year) nitrate bioreduction reported by Percak-Dennett and
Roden (2014), which demonstrated that nitrate bioreduction in the
LPZ sediment was driven by heterotrophic denitrification. The
enhanced aqueous sulfate concentration in the OS on day 1 (Fig. S3)
resulted from the release of sediment-associated sulfate, which was
confirmed by the quick leaching of sulfate into DI-water from the OS.
In addition, ammonium production was not observed (data not
shown), suggesting that dissimilatory reduction of nitrate to ammoni-
um was negligible in both sediments.
Increasing initial nitrate concentration had a minor effect on the
nitrate bioreduction in both sediments without organic carbon addition
(Figs. 2aand3a). In order to compare the nitrate bioreduction rate
between the OS and RS, a pseudo first-order rate with respect to nitrate
concentration was estimated as described below. When initial nitrate
concentration increased from 0.46 to 2 mM, the observed pseudo first-
order nitrate bioreduction rates only increased from 0.005 to
0.006 mM/d in the OS and from 0.002 to 0.003 mM/d in the RS. The re-
sult was expected given that nitrate bioreduction in the OS was mainly
limited by organic carbon, while the bioreduction in the RS was mainly
controlled by denitrifier population and gene expression as mentioned
above. When organic carbon limitation was removed by spiking organic
substrates, nitrate bioreduction rate increased with increasing nitrate
concentration in the OS (Fig. 2b). In the RS reactors spiked with organic
substrates, however, the rate of nitrate bioreduction only increased
from 0.005 to 0.007 mM/d when nitrate concentration increased from
0.46 to 2 mM, indicatingthat the initial nitrate concentration had negli-
gible impact on the nitrate bioreduction in the RS. This was consistent
with the observation that the nitrate bioreduction rate in the RS was
mainly limited by denitrifier population.
3.3. End product of nitrate bioreduction
N
2
O was an intermediate product of nitrate bioreduction in the OS as
its concentration initially increased and then decreased (Fig. 4a). N
2
O
concentration increased as nitrate and nitrite were reduced (Fig. 2).
When nitrate was depleted, N
2
O concentration decreased with time.
As an intermediate species, N
2
O concentration was affected by the
rate of N
2
O production from nitrate and nitrite reduction, and N
2
O
Fig. 3. Nitratereduction, and nitrite production and reduction in the RS suspensions with different initial nitrate concentrations, and with/without spiked organic carbon (OC). Sediment/
SGW ratio= 200 g/L. Plota: nitrate and nitriteconcentrationswithout OC spike(symbol ■for NO
3
−
and symbol□for NO
2
−
under initial NO
3
−
= 0.46 mM; symbol▲for NO
3
−
and symbol△
for NO
2
−
under initial NO
3
−
=2mM),Plotb: nitrate and nitrite concentrations with OC spike (symbol ■for NO
3
−
and symbol □for NO
2
−
under initial NO
3
−
= 0.46 mM; symbol●fo r NO
3
−
and symbol ○for NO
2
−
under initial NO
3
−
= 1.5 mM; symbol ▲for NO
3
−
and symbol △for NO
2
−
under initial NO
3
−
= 2 mM), plot c,dand e: spiked OC compound concentrations corre-
spondingto nitrate reduction in plot bunderdifferent initialNO
3
−
concentrations(symbol ■for ace tate, symbol ●for lactate, and symbol▲for glucose).The error bars denotethe standard
deviations of duplicate experiments.
191S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
reduction to N
2
. The results indicated that the N
2
Obioreductioninthe
OS was a fast process as its concentration was low and quickly
decreased to zero once nitrate and nitrite were completely reduced
(Figs. 2 and 4). A slight shift of N
2
O concentration profile to the late
time in the case of 2 mM NO
3
−
was apparently caused by the higher
initial concentration of nitrate, and the subsequent accumulation of
nitrite required a longer time to completely transform to N
2
O
(Fig. 2b). In the RS, N
2
O accumulated to a much higher concentration
than in the OS (Fig. 4b). Nitrogen mass balance calculation (Table S3)
by considering the equilibrium of N
2
O in gas and aqueous phases
indicated that almost all reduced nitrate (91–102%) was transformed
to N
2
O. The continuous, quantitative accumulation of N
2
O as a result
of nitrate and nitrite reduction indicated that N
2
O was the end product
of nitrate bioreduction in the RS.
The reduction of N
2
OtoN
2
in the OS was consistent with the quan-
titative PCR measurements that showed the abundance of nosZ gene in
the OS upon incubation (Table 1). Lin et al. (2012a) found that the deni-
trifier population in Hanford subsurface sediment was dominated by
Ochrobactrum anthropi, and its nosZ clones had 97% similarity with
Ochrobactrum anthropiYD50.2, which can tolerate up to 100 mM nitrite.
The increase and then decrease with time in N
2
O concentration (Fig. 4a)
indicated that nosZ was expressed and N
2
O reductase functional in the
OS upon incubation. In the RS, however, the quantitative production
of N
2
O from nitrate and nitrite reduction indicated that N
2
Oreductase
did not exist or its activity was inhibited. Undetectable nosZ gene in
the RS (Table 1)reflected little or no potential for N
2
O reduction, and
led to the N
2
O accumulation as the end product of nitrate bioreduction.
The long-term adaptation of the denitrifiers at the site as speculated
before for the low denitrifier population in the RS might also be
responsible for the low or non-existence of nosZ gene.
3.4. Kinetic model analysis
Various models havebeen proposed to describe nitrate bioreduction
including multiplicative Monod model (Henze et al., 2000; Rittmann
and McCarty, 2001), enzyme-based model (Hamilton et al., 2005), and
modified Monod model with electron competition (Ni et al., 2011; Ni
et al., 2013; Pan et al., 2013). All models adopt saturation-type expres-
sion with respect to electron donor and acceptor concentrations, but
with different ways to balance mass and electron for the reactions
involvedin nitrate bioreduction. The rate-limiting role of organic carbon
and microbial biomass in the sediments as described before indicated
that a multiplicative saturation-type model is needed to describe the ni-
trate and nitrite bioreduction in the LPZ sediments. The multiplicative
Monod model was modified by adding an inhibition term to account
for the electron competition between nitrate and nitrite bioreduction
(first part containing K
I
in the right hand side of Eq. (5)). This inhibition
term is required to describe the phenomenon in Fig. 1a and b, which
showed that nitrite reduction started only after nitrate reduction was
completed in the OS. The electron competition concept (Ni et al.,
2011; Ni et al., 2013; Pan et al., 2013) assumed that the electrons
generated from organic carbon oxidation can be competitively used by
different reductases involved in the nitrate bioreduction (Richardson
et al., 2009). The kinetic model as described by Eqs. (4)–(7) with param-
eters provided in Table 2 was able to simultaneously describe nitrate
and nitrite concentration changes as a function of time under variable
solid/SGW ratios (i.e. initial biomass and organic carbon concentra-
tions) and initial nitrate concentrations (Fig. 1).
Although most nitrate and nitrite concentration profiles were fitted
well by the model (Eqs. (4)–(7)), discrepancy still existed between the
model and measured results. In the OS, the simulated nitrate and nitrite
concentrations for the 0.46 mM nitrate treatment at 400 and 800 g/L
solid water ratio matched wellwith the measured values, while the sim-
ulated nitrate and nitrite concentrations for the 0.46 mM nitrate treat-
ment at 200 g/L were lower and higher than the corresponding
measured values, respectively. For the treatment with 2 mM nitrate at
200 g/L, nitrate was fitted well but the simulated nitrite concentrations
were higher than the experimental results. In the RS, thenitrate concen-
trations were simulated well for most batch experiments except for the
treatment with 0.46 mM nitrate at 400 g/L solid water ratio, for which
the simulated nitrite concentrations were either lower or close to the
measured values. It's important to note that the model was used to
simultaneously fit all the treatments using a common set of model
parameters. Fitting can be improved by adjusting parameters for each
treatment (results not shown), suggesting that the model parameters
changed with experimental conditions. These changes, however, are
within the uncertainty ranges as discussed below (Table 2). The ATP
results were poorly described by the model. In the simulation, ATP
was treated as a surrogate for the biomass of denitrifiers. Comparing
the modeling and experimental results indicated that the ATP concen-
tration changed more dynamically than the simulated biomass. Similar
phenomena were observed previously in the growth of fermentative
Fig. 4. Temporal evolution of N
2
O concentration in thereactor headspacesduring the nitratebioreduction inthe OS (a) and RS (b) suspensions undervariable initialnitrate concentrations
corresponding to those in plot b in Figs. 2 and 3. Sediment/SGW ratio = 200 g/L with OC spike. The error bars denote the standarddeviationsof duplicate experiments.
192 S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
bacteria in cultures and microbial growth in moist soils (Meyer and
Papoutsakis, 1989; Nannipieri et al., 1978), suggesting that ATP was
not a good indicator for modeling the dynamic change of biomass in
the LPZ sediments. The poor correlation of ATP content with total
biomass concentrations would be expected, because the ATP content
of the sample should somewhat reflect the product of the biomass
concentration and the cell-specific activity. Moreover, the simulated
biomass was independent of initial nitrate concentration (Fig. 1). This
can be seen in the two treatments with different initial nitrate
concentrations at the same solid/water ratio (Fig. 1c). The result is
consistent with the experimental observation that the bioavailability
of sediment-associated organic carbon limited the rate of nitrate
bioreduction in the sediments, and consequently the change in nitrate
concentration had negligible effect on biomass growth. Difference in
simulated biomass at different solid/water ratio is because of the differ-
ence in initial biomass in the reactors.
The estimated inhibition concentration of nitrate (K
I
)fornitrite
reduction was 0.0039 mM in the OS (Table 2), which was much smaller
than the initial nitrate concentration (0.46–2 mM) used in the experi-
ments. While this is an empirically fitted parameter to match the exper-
imental data, the small K
I
value suggested that the reactivity of nitrite
reductases in the OS was strongly inhibited by the presence of nitrate.
The macroscopic effect is that nitrite reduction in the OS started after
the completion of nitrate reduction. In the RS, the fitting K
I
value was
20 mM, suggesting that nitrate had little impact on nitrite reduction.
The large K
I
reflected the fact that nitrite reduction in the RS was not
inhibited by the presence of 0.46–2.0 mM nitrate (Fig. 1d and e).
These results support the assumption of electron competition between
the nitrate and nitrite reductases. When the nitrate reduction rate was
faster than nitrite reduction in the OS, electrons generated from organic
carbon oxidation were less available for nitrite reductase. On the other
hand, when nitrate reduction rate was slower than or close to nitrite
reduction in the RS, nitrite reductases was able to effectively compete
for electrons with nitrate reductases.
The values of most other model parameters were generally consis-
tent with those reported in the literature (Table 2). Some differences
exist, however. The much larger fitted values of f
e
than those reported
in the literature for nitrate and nitrite bioreduction indicated that
biomass synthesis was small in the OS (f
e
N0.9), and negligible in the
RS (f
e
= 1). The result indicated that almost all electrons and energy
generated from organic carbon oxidation were used for the respiration,
suggesting that the nitrate and nitrite concentrations can be simulated
by ignoring the growth term. Model simulations by ignoring the growth
term (i.e., setting f
e
=1inreactions 1 to 3) confirmed this conclusion
(results not shown). In the RS, a larger fitted K
s,NO
3
−
value with respect
to nitrate was used to match the slow rate of nitrate reduction
(Fig. 1d), and a smaller value of K
s,oc
NO2−
with respect to organic carbon
was used to match the relative faster reduction of nitrite (Fig. 1e).
The estimated parameters contained large uncertainties (Table 2).
This is expected because a previous theoretical analysis revealed that
the kinetic parameters in the saturation-type models are linearly corre-
lated and the relative standard deviation (i.e., σ(θ
k
)/θ
k
) of correlated pa-
rameters could be as large as 80 depending on experimental conditions
(Liu and Zachara, 2001). The relative standard deviation of all the esti-
mated parameters except for K
s,oc
NO
2
−
in this study ranged from 0.5 to 63,
within the theoretical ranges described above. The larger uncertainties
for K
s,oc
NO
2
−
value in the RS, was attributed to the larger variations in the
measured nitrate and nitrite concentrations as shown by the large
value of CRV (Table 2). Because of the correlated nature of the kinetic
parameters, their estimated values should be used in sets to minimize
simulation uncertainties.
4. Conclusions
This study found that organic carbon bioavailability and gene
expression can play critical roles in determining the rate and end
product of nitrate bioreduction in the LPZ sediments, highlighting that
a common assumption that organisms capable of denitrification are
ubiquitous in the subsurface environment has a limitation. The lower
rate of nitrate bioreduction in the reduced than that in the oxidized
LPZ sediments was also in contrast to the observations in river and
lake sediments where nitrate bioreduction is typically faster in more re-
duced sediments. Long-term adaptation of microbial community to the
biogeochemical environmentin the LPZ was the likely cause. Themicro-
bial community in thedeeper region of the LPZ has less chance to access
nitrate because nitrate would be consumed when it diffuses from the
overlyingaquifer. This study also indicated that the bioavailable organic
carbon can be much less than the total organic carbon in the LPZ sedi-
ments. Using total organic carbon for modeling would significantly
over-estimate the potential and rates of microbial activities. In addition,
modeling analysis indicated that the denitrification model in the LPZ
can be simplified by ignoring growth term under natural conditions.
Moreover, the model for predicting nitrate migration in the LPZ systems
will need to carefully consider organic carbon bioavailability and the in-
congruent carbon substrate utilization by the ba cteria consortium, func-
tional gene expression, and redox heterogeneity in the rate and end
product of nitrate bioreduction in the LPZ sediments.The LPZ, such as
an aquitard, typically separates upper and lower aquifers in subsurface
systems. When groundwater tables unequally change in the aquifers
through preferential recharge or pumping extraction in one aquifer, a
Table 2
Parameters used for simulation in the OS and RS.
Kinetic parameters OS RS Literature values
θσ(θ)/θθ σ(θ)/θ
X0 (g/L) 2.1 × 10
−6
1.4 7.2 × 10
−7
11.8 –
Bioavailable OC (%) 4.3 8.0 0.18 1.0 –
qNO−
3(mmol·g-1·d-1) 140 1.5 185 12.7 55
a
−203
b
qNO−
2(mmol·g-1·d-1) 170 5.9 85 12.0 28
a
−96
b
K
s,oc
NO
3
−
(mM) 0.1 –0.1 –0.1
b
K
s,oc
NO
2
−
(mM) 0.1 –0.001 364 0.1
b
K
s,NO
3
−
(mM) 0.0018 –0.1 14.2 0.002
b
−0.003
a
K
s,NO
2
−
(mM) 0.0041 –0.0041 –0.0041
b
K
I
(mM) 0.0039 39.5 20 63.0 –
f
e,NO
3
−
→NO
2
−
0.999 0.5 1 –0.5
c
f
e,NO
2
−
→N
2
0.94 2.1 –– –
f
e,NO
2
−
→N
2
O
––1––
b (d-1) 0.04 5.0 0 –b0.05
c
CRV –233 –4.8 × 10
5
–
a
Parameters from Almeida et al., 1997;
b
Parameters from Pan et al., 2013;
c
Parameters from Pan et al., 2013; Rittmann and McCarty, 2001.
193S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
hydraulic pressure gradient will form, which can force groundwater
flow from one aquifer to the other through the LPZ. The result in this
study indicated that nitrate carried by groundwater can be retarded
and removed through bioreduction in the LPZ. On the other hand, N
2
O
may be produced in the LPZ and carried by the groundwater into the
receiving aquifer, affecting the groundwater quality and chemical
composition. Consequently the bioreduction of nitrate and reducing
products in the LPZ needs to be carefully considered in groundwater
quality control and management.
Acknowledgments
This research is supported by the U.S. DOE, Office of Biological and
Environmental Research (BER) as part of the Subsurface Biogeochemical
Research (SBR) Program through the Pacific Northwest National
Laboratory (PNNL) SBR Science Focus Area Research Project. Part of
this research was performed in the Environmental Molecular Science
Laboratory, a user facility sponsored by the DOE's Office of BER and
located at PNNL. PNNL is operated for DOE by Battelle Memorial
Institute under contract DE-AC05-76RL01830. S. Y. would like to
acknowledge the fellowship from the China Scholarship Council. We
also thank the anonymous reviewers for their careful reading and
constructive comments.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.scitotenv.2015.08.122.
References
Ahn, Y.H., 2006. Sus tainable nitrogen elimination biotechnologies: a review. Process
Biochem. 41, 1709–1721.
Almeida, J.S., Reis, M.A.M., Carrondot, M.J.T., 1997. A unifying kinetic model of denitrifica-
tion. J. Theor. Biol. 186, 241–249.
Arighi, L., Haggerty, R., Myrold, D.D., Iverson, J., Baham, J.E., Madin, I.P., Arendt, J., 2005.
Nitrate attenuation in the Missoula flood deposits aquitard (Willamette silt) of the
Willamette Valley, Oregon. AGU fall meeting abstracts 12/2005.
Babbin, A.R., Keil, R.G., Devol, A.H., Ward, B.B., 2014. Organic matter stoichiometry, flux,
and oxygen control nitrogen loss in the ocean. Science 344, 406–408.
Bahr, M., Stams, A.J.M., De la Rosa, F., Garcia-Encina, P.A., Munoz, R., 2011. Assessing the
influence of the carbon oxidation-reduction state on organic pollutant biodegrada-
tion in algal-bacterial photobioreactors. Appl. Microbiol. Biotechnol. 90, 1527–1536.
Bard, Y., 1974. Nonlinear parameter estimation. Academic Press, New York.
Beck, J.V., Arnold, K.J., 1977. Parameter estimation in engineering and science. John Wiley
& Sons Inc., New York.
Bjornstad, B.N., Horner, J.A.,Vermeul, V.R., Lanigan, D.C.,Thorne, P.D., 2009.Borehole com-
pletion andconceptual hydrogeologicmodel for the IFRC wellfield, 300 area, Hanford
Site. Pacific Northwest National Laboratory, Richland, WA.
Butterbach-Bahl, K., Baggs, E.M., Dannenmann, M., Kiese, R., Zechmeister-Boltenstern, S.,
2013. Nitrous oxide emissions from soils: how well do we understand the processes
and their controls? Philos. Trans. R. Soc. B 368.
Cervantes, F.J., 2009. Environmental technologies to treat nitrogen pollution. IWA Pub-
lishing, London.
Firestone, M.K., Firestone, R.B., Tiedje, J.M., 1980. Nitrous-oxide from soil denitrification:
factors controlling its biological production. Science 208, 749–751.
Hamilton, R., Casasus, A., Rasche,M., Narang, A., Svoronos, S.A., Koopman, B., 2005. Struc-
tured model for denitrifier dialuxic growth. Biotechnol. Bioeng. 90, 501–508.
Henze, M., Gujer, W., Mino, T.,Van Loosedrecht, M., 2000. Activated sludge models ASM1,
ASM2, ASM2d and ASM3. IWA scientific and technical report, London, U.K. vol. 9
Herrman, K.S., Bouchard, V., Moore, R.H., 2008. Factors affecting denitrification in agricul-
tural headwater streams in northeast Ohio, USA. Hydrobiologia 598, 305–314.
Jha, P.K., Minagawa, M., 2013.Assessment of denitrification process in lowerIshikari river
system, Japan. Chemosphere 93, 1726–1733.
Kampschreur, M.J., Temmink, H., Kleerebezem, R., Jetten, M.S.M., van Loosdrecht, M.C.M.,
2009. Nitrous oxide emission during wastewate r treatment. Water Res. 43,
4093–4103.
Laverman, A.M., Van Cappellen, P., van Rotterdam-Los, D., Pallud, C., Abell, J., 2006. Poten-
tial rates and pathways of microbial nitrate reduction in coastal sediments. FEMS
Microbiol. Ecol. 58, 179–192.
Laverman, A.M., Garnier, J.A., Mounier, E.M., Roose-Amsaleg, C.L., 2010. Nitrous oxide
production kinetics during nitrate reduction in river sediments. Water Res. 44,
1753–1764.
Lee, J.H., Fredrickson, J.K., Kukkadapu, R.K., Boyanov, M.I., Kemner, K.M., Lin, X.J., Kennedy,
D.W., Bjornstad, B.N., Konopka, A.E., Moore, D.A., Resch, C.T., Phillips, J.L., 2012.
Microbial reductive transformation of phyllosilicate Fe(III) and U(VI) in fluvial sub-
surface sediments. Environ. Sci. Technol. 46, 3721–3730.
Lin, X.J., Kennedy, D., Peacock, A., McKinley, J., Resch, C.T., Fredrickson, J., Konopka, A.,
2012a. Distribution of microbial biomass and potential for anaerobic respiration in
Hanford Site 300 area subsurface sediment. Appl. Environ. Microbiol. 78, 759–767.
Lin, X.J., Kennedy, D., Fredrickson, J., Bjornstad, B., Konopka, A., 2012b. Vertical stratifica-
tion of subsurface microbial community composition across geological formations at
the Hanford Site. Environ. Microbiol. 14, 414–425.
Liu, C.X., Ball, W. P., 2002. Back diffusion of chlorinate d solvent contaminants from a
natural aquitard to a remediated aq uifer under well-controlled field conditions:
predictions and measurements. Ground Water 40, 175–184.
Liu, C.X., Shang, J.Y., Kerisit, S., Zachara, J.M., Zhu, W.H., 2013. Scale-dependent rates of
uranyl surface complexation reaction in sediments. Geochim. Cosmochim. Acta 105,
326–341.
Liu, C., Zachara, J.M., 2001. Uncertainties of Monod kinetic par ameters nonlinearly
estimated from batch experiments. Environ. Sci. Technol. 35, 133–141.
Lu, H.J., Chandran, K., 2010. Factors promoting emissions of nitrous oxide and nitric oxide
from denitrifying sequencing batch reactors operated with methanol and ethanol as
electron donors. Biotechnol. Bioeng. 106, 390–398.
Mastrocicco, M., Colombani, N., Salemi, E., Castaldelli, G., 2011. Reactive modeling of deni-
trification in soils with natural and depleted organic matter. Water Air Soil Pollut.
222, 205–215.
Matějů,V.,Čižinská , S., Krejčí, J., Janoch, T., 1992. Biological water denitrificatio n —a
review. Enzym. Microb. Technol. 14, 170–183.
McCarty, P.L., Beck, L., St. Amant, P., 1969. Biological denitrification of wastewaters by
addition of organic materials. Proceedings of the 2 4th annual Purdue Industrial
Waste Conference, pp. 1271–1285.
Meyer, C., Papoutsakis, E., 1989. Increased levels of ATP and NADH are associated with in-
creased solvent produ ction in continuous cultures of Clostri dium acetobutylicum .
Appl. Microbiol. Biotechnol. 30, 450–459.
Nannipieri, P., Johnson, R.L.,Paul, E.A., 1978. Criteria for measurement of microbial growth
and activity in soil. Soil Biol. Biochem. 10, 223–229.
Ni, B.J., Ruscalleda, M., Pellicer-Nàcher, C., Smets, B.F., 201 1. Modeling nitrous oxide
production during biological nitrogen removal via nitrification and denitrification:
extensions to the general ASM models. Environ. Sci. Technol. 45, 7768–7776.
Ni, B.J., Ye,L., Law, Y.Y., Byers, C., Yuan, Z.G., 2013.Mathematical modeling of nitrous oxide
(N
2
O) emissions from full-scale wastewater treatment plants. Environ. Sci. Technol.
47, 7795–7803.
Pan, Y.T., Ye, L., Ni, B.J., Yuan, Z.G., 2012. Effect of pH on N
2
O reduction and accumulation
during denitrification by methanol utilizing denitrifiers. Water Res. 46, 4832–4840.
Pan, Y.T., Ni, B.J., Yuan, Z.G., 2013. Modeling electron competition among nitrogen oxides
reduction and N
2
O accumulation in denitrification.Environ.Sci.Technol.47,
11083–11091.
Percak-Dennett, E.M., Roden, E.E., 2014. Geochemical and microbiological responses to
oxidant introduction into reduced subsurface sediment from the Hanford 300 Area,
Washington. Environ. Sci. Technol. 48, 9197–9204.
Peretyazhko, T.S., Zachara, J.M., Kukkadapu, R.K., Heald, S.M., Kutnyakov, I.V., Resch, C.T.,
Arey, B.W., Wang, C.M., Kovarik, L., Phillips, J.L., Moore, D.A., 2012. Pertechnetate
(TcO
4
−
) reduction by reactive ferrous iron forms in naturally anoxic, redox transition
zone sediments from the Hanford Site, USA. Geochim. Cosmochim. Acta 92, 48–66.
Pulou, J., Tournebize, J., Chaumont, C., Haury, J., Laverman, A.M., 2012. Carbon availability
limits potential denitrification in watercress farm sediment. Ecol. Eng. 49, 212–220.
Rassamee, V., Sattayatewa, C., Pagilla, K., Chandran, K., 2011. Effect of oxic and anoxic
conditionson nitrous oxide emissions from nitrification and denitrification processes.
Biotechnol. Bioeng. 108, 2036–2045.
Richardson, D., Felgate, H., Watmough, N., Thomson, A., Baggs, E.,2009. Mitigating release
of the potent greenhous e gas N
2
O from the nitrogen cycle —could enzymic regula-
tion hold the key? Trends Biotechnol. 27, 388–397.
Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology: Principles and Applica-
tions. McGraw-Hill Education, New York.
Rivett, M.O., Buss, S.R., Morgan,P., Smith, J.W.N., Bemment, C.D., 2008. Nitrate attenuation
in groundwater: a review of biogeochemical controlling processes. Water Res. 42,
4215–4232.
Robertson, W.D., Russell, B.M., Cherry, J.A., 1996. Attenuation of nitrate in aquitard
sediments of southern Ontario. J. Hydrol. 180, 267–281.
Ronen, D., Magaritz, M., Almon, E., 1988. Contaminated aquifers are a forgotten compo-
nent of the global N
2
O budget. Nature 335, 57–59.
Seitzinger, S., Harrison, J.A., Bohlke, J.K., Bouwman, A.F., Lowrance, R., Peterson, B., Tobias,
C., Van Drecht, G., 2006. Denitrification across landscapes and watersc apes: a
synthesis. Ecol. Appl. 16, 2064–2090.
Small, G.E., Cotner, J.B., Finlay, J.C., Stark, R.A., Sterner, R.W., 2014. Nitrogen transforma-
tions at the sediment–water interface across redox gradients in the Laurentian
Great Lakes. Hydrobiologia 731, 95–108.
Stanley, P.E., 1989. A review of bioluminescent ATP techniques in rapid microbiology. J
Biolumin. Chemilumin. 4, 375–380.
Streuli, C.A., Averell,P.R., 1970. Analytical Chemistry of Nitrogen and its Compounds. John
Wiley & Sons Inc., New York.
U.S. EPA, 1993. Process Design Manual: Nitrogen Control (Cincinnati, Ohio).
Vaclavkova, S., Jorgensen, C.J., Jacobsen, O.S., Aamand, J., Elberling, B., 2014. The impor-
tance of microbial iron sulfide oxidat ion for nitrate depletion in anoxic Danish
sediments. Aquat. Geochem. 20, 419–435.
Van Trump, J.I., Wrighton, K.C., Thrash, J.C., Weber, K.A., Andersen, G.L., Coates, J.D., 2011.
Humic acid-oxidizing, nitrate-reducing bacteria in agricultural soils. mBio 2.
Wang, S., Jaffe, P.R., Li, G., Wang, S.W., Rabitz, H.A., 2003. Simulating bioremediation of
uranium-contaminated aquifers; uncertainty assessment of model p arameters.
J. Contam. Hydrol. 64, 283–307.
194 S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
Ward, B.B ., 2013. How nitrogen is lost. Science 341, 352–353.
Weymann, D., Geistlinger, H., Well, R., von der Heide, C., Flessa, H., 2010. Kinetics of N
2
O
production and reduction in a nitrate-contaminated aquifer inferred from laboratory
incubation experiments. Biogeosciences 7, 1953–1972.
Zhang, J., Yu, Y., Zhu, T., Cai, Z., 2014. The mechanisms governing low denitrification ca-
pacity and high nitrogen oxide gas emissions in subtropical forest soils in China.
J. Geophys. Res. Biogeosci. 119, 1670–1683.
Zhang, Y.C., Slomp, C.P., Broers, H.P., Passier, H.F., Van Cappellen, P., 2009. Denitrification
coupled to pyrite oxidation and changes in groundwater quality in a shallow sandy
aquifer. Geochim. Cosmochim. Acta 73, 6716–6726.
Zumft, W.G., 1997. Cell biology and molecular basis of denitrification.Microbiol. Mol. Biol.
Rev. 61, 533–616.
195S. Yan et al. / Science of the Total Environment 539 (2016) 185–195
SUPPLEMENTARY INFORMATION TO:
Nitrate Reduction in Redox Transitional Sediments from a Groundwater Aquitard
Sen Yan1,2, Yuanyuan Liu1, Chongxuan Liu1*, Liang Shi1, Jianying Shang1, Huimei Shan1,2, John
Zachara1, Jim Fredrickson1, David Kennedy1, Charles T. Resch1, Christopher Thompson1, and
Sarah Fansler1
1Pacific Northwest National Laboratory, Richland, WA 99354, USA
2China University of Geosciences, Wuhan, 430074, China
* Corresponding author:
E-mail: chongxuan.liu@pnnl.gov.
Tel: 1-509- 371-6880, Fax: 1-509- 371-6354;
Pacific Northwest National Laboratory, PO Box 999, MSIN: K8-96, Richland, WA 99354, USA
Table S1. The experimental conditions for the batch experiments.
sediment
type
sediment/water
ratio
(g/L)
[NO3
-]0
(mM)
organic
carbon(OC)
addition
OC
speciation
N2O
measurement
OS
200 2.0
×a × ×
200
400 0.46
800
0.46
200 1.5 √
b
√ √
2.0
RS
200 2.0
× × ×
200
400 0.46
800
0.46
√ √ √
200 1.0
2.0
anot added or not measured, badded or measured
Table S2. Half reactions involved in the OS and RS.
Function of the Half
Reaction Half Reactionsa Electron
Equivalent
R-1: Electron donor 22 2 3
2311 9
8888 8
CH O H O CO HCO H e
1
R-2: Synthesis
utilizing NO3
- 32 572 2
1529 1 11
28 28 28 28 28
NO CO H e C H O N H O
1-
32
,eNO NO
f
R-3: Synthesis
utilizing NO2
- 22 572 2
1527 1 10
26 26 26 26 26
NO CO H e C H O N H O
1-
222
,/eNO N NO
f
R-4: Respiration
utilizing NO3
- 322
111
222
NO H e NO H O
32
,eNO NO
f
R-5a: Respiration
utilizing NO2
- in OS 222
14 12
33 63
NO H e N H O
22
,eNO N
f
R-5b: Respiration
utilizing NO2
- in RS 222
13 13
22 44
NO H e N O H O
22
,eNO NO
f
aHalf reactions were from Environmental Biotechnology: Principles and Applications (Rittmann and
McCarty 2001).
Table S3. Nitrogen mass balance calculation in the RS incubations with 200 g/L sediment and OC spike.
Initial
concentration
condition
N0
a
(μmol)
NNO3
-b
(μmol)
NNO2
-c
(μmol)
NN2O (μmol) Nloss
f
(μmol)
N
N
ratio %
gas
phased
aqueous
phasee
[NO3
-]0 = 0.46
mM 23 7.84 3.94 9.76 0.39 11.22 91
[NO3
-]0 = 1.0 mM 50 26.06 10.85 12.15 0.49 13.10 97
[NO3
-]0 = 2.0 mM 100 78.32 9.20 12.23 0.49 12.48 102
a the initial total mass of N added as nitrate, b the mass of nitrate including the nitrate remaining in the
reactor and the nitrate taken out for sampling, c the mass of nitrite including the nitrite remaining in the
reactor and the nitrite taken out for sampling, d the amount of gas phase N2O (residual N2O plus N2O
consumed during sampling for measurements), e the calculated amount of aqueous phase N2O based on
gas phase N2O and using the Henry’s law. f the mass of nitrate loss during reaction, Nloss = N0 – (NNO3
-
+NNO2
-).
Figure S1. Nitrate concentrations in the batch reactors containing the autoclaved OS and RS.
Figure S2. Temporal changes of nitrite concentrations in the OS and RS initially spiked with 2 mM NO2
-
(No nitrate was provided, sediment/SGW ratio = 200 g/L with spiked organic carbon).
Figure S3. Temporal changes of sulfate concentrations during reaction in the OS (a-c) and RS (d-f). Plots
a and d showed sulfate concentrations in the reactors with variable sediment/SGW ratio without organic
carbon spike; Plots b and e showed sulfate concentrations in the reactors with different initial nitrate
concentrations without organic carbon spike; and plots c and f showed sulfate concentrations in the
reactors with organic carbon spike.