-
Bioresource Technology 241 (2017) 552–562
Contents lists available at ScienceDirect
Bioresource Technology
journal homepage: www.elsevier .com/locate /bior tech
Nitrate removal, spatiotemporal communities of denitrifiers and
theimportance of their genetic potential for denitrification in
noveldenitrifying bioreactors
http://dx.doi.org/10.1016/j.biortech.2017.05.2050960-8524/� 2017
Elsevier Ltd. All rights reserved.
⇑ Corresponding author.E-mail address: [email protected] (L.
Wang).
1 Equal contributions.
Yimin Zhang a,1, Longmian Wang a,⇑,1, Wei Han b, Xu Wang c,
Zhaobing Guo d, Fuquan Peng a, Fei Yang a,Ming Kong a, Yuexiang Gao
a, Jianying Chao a, Dan Wu a, Bin Xu a, Yueming Zhu a
aNanjing Institute of Environmental Sciences, Ministry of
Environmental Protection, No. 8 Jiang Wang Miao Street, Nanjing
210042, PR Chinab Sino-Japan Friendship Center for Environmental
Protection, No. 1 Yu Hui Nan Road, Chao Yang District, Beijing
100029, PR Chinac School of Resource and Environmental Sciences,
Wuhan University, 129 Luoyu Road, Wuhan 430079, PR ChinadNanjing
University of Information Science & Technology, No. 219 Ningliu
Road, Nanjing 210044, PR China
h i g h l i g h t s
� Higher-rate NO�3 -N removal is achieved in NUA–DNBF than in
DAS–DNBF.� The potential N2O production rate was much lower in
DAS–DNBF than NUA–DNBF.� Burkholderiales, Rhodocyclales and
Rhizobiales were dominant in both substrates.� qnosZ and
Pqnir/qnosZ may serve as biological indicators for NO3-N removal
in DNBF.
� The NO�3 -N removal rate in NUA increased linearly with the
DEA.
a r t i c l e i n f o
Article history:Received 7 April 2017Received in revised form 27
May 2017Accepted 30 May 2017Available online 1 June 2017
Keywords:Denitrification enzyme activityDenitrifying
biofilterDewatered alum sludgeFunctional geneNeutralized used
acid
a b s t r a c t
Nitrate treatment performance and denitrification activity were
compared between denitrifying biolog-ical filters (DNBFs) based on
dewatered alum sludge (DAS) and neutralized used acid (NUA). The
spa-tiotemporal distribution of denitrifying genes and the genetic
potential associated with denitrificationactivity and nitrate
removal in both DNBFs were also evaluated. The removal efficiency
of NUA–DNBFincreased by 8% compared with that of DAS–DNBF, and the
former NUA–DNBF emitted higher amountof N2O. Analysis of abundance
and composition profiles showed that denitrifying gene patterns
variedmore or less in two matrices with different depths at three
sampling times. Burkholderiales,Rhodocyclales, and Rhizobiales were
the most commonly detected in both media during stable
periods.Denitrification was determined by the abundance of specific
genes or their ratios as revealed by control-ling factors. The
enhanced nitrate removal could be due to increasing qnosZ or
decreasing
Pqnir/qnosZ.
Furthermore, NUA–DNBF solely reduced nitrate by increasing the
denitrification enzyme activity.� 2017 Elsevier Ltd. All rights
reserved.
1. Introduction
The impact of agricultural production on the nitrogen (N)
cycleleads to N enrichment of surface and ground water, as well
asincreased nitrous oxide (N2O) emissions (Wang et al.,
2016a).Nitrate (NO�3 -N), which is an important component of N,
causesagricultural runoff pollution vulnerably owing to its high
watersolubility and mobility (Hua et al., 2016). High levels of
NO�3 -N
from agricultural drainage to receiving waters can pose a risk
tothe environment. Additionally, N2O is a greenhouse gas
andozone-depleting substance emitted as a result of incomplete
deni-trification that also leads to undesired effects on the
atmosphereand ecosystem (Syakila and Kroeze, 2011). Thus, it is
necessaryto remove excess NO�3 -N from agricultural fields and
control N2Oproduction simultaneously to ensure the security of
waterresources and human health.
Denitrifying bioreactors/biofilters (DNBFs) are a
promisingapproach to reducing NO�3 -N loads from agriculture runoff
dis-charged into waterways. These systems typically use media in
con-tainers to convert NO�3 -N to N gas via microbial
denitrification
http://crossmark.crossref.org/dialog/?doi=10.1016/j.biortech.2017.05.205&domain=pdfhttp://dx.doi.org/10.1016/j.biortech.2017.05.205mailto:[email protected]://dx.doi.org/10.1016/j.biortech.2017.05.205http://www.sciencedirect.com/science/journal/09608524http://www.elsevier.com/locate/biortech
-
Y. Zhang et al. / Bioresource Technology 241 (2017) 552–562
553
(Schipper et al., 2010). The medium acts as a carbon and
energysource for denitrifying microorganisms, and is the key factor
influ-encing nitrate removal and N2O emission. Neutralized used
acid(NUA) and dewatered alum sludge (DAS) originating from
indus-trial byproducts, which have demonstrated nitrate removal
effi-ciencies ranging of 50–82%, have emerged as low-cost
substratesof constructed wetlands (CWs) and laboratory columns to
reduceNO�3 -N from synthetic wastewater and surface water
(Wendlinget al., 2012; Hu et al., 2016). However, the studies
mentionedabove only focused on reducing NO�3 -N using NUA or
DAS-basedbiofilters for short-term treatment, while few studies
have investi-gated the results of long-term operation or their
usage for nitrateremoval in agricultural runoff treatment. In
support of active den-itrification in bioreactors, elevated levels
of denitrification enzymeactivity (DEA) have commonly been measured
in DNBFs (Warnekeet al., 2011), but elevated DEA does not always
mean that signifi-cant NO�3 -N removal is occurring (Schipper and
McGill, 2008).Indeed, it is currently not known if NUA and DAS in
DNBFs selectfor substrates capable of reducing NO�3 -N via
increasing DEA. Inaddition, N2O production during denitrification
is an importantissue that must be addressed when studying DNBFs. A
study ofadverse effects in different substrates used in
denitrification bedsbyWarneke et al. (2011) revealed that a
combination of maize cobsand woodchips enhanced NO�3 -N removal
while minimizing theadverse effects of N2O release, whereas Greenan
et al. (2009) mea-sured negligible N2O emission of only
0.003–0.028% of the totalNO�3 -N removed in a woodchip column
study. However, less atten-tion has been given to examining
potential N2O production inDNBFs containing NUA or DAS.
Conventional heterotrophic denitrification, which is the
domi-nant mechanism of NO�3 -N removal in DNBFs, consists of
consecu-tive reaction steps (Schipper et al., 2010). Nitrite
reductase genes(nirS and nirK), which catalyze the reduction of
nitrite to nitric-oxide, are considered to be the key functional
genes involved indenitrification. Additionally, the last step of
denitrification (thereduction of N2O to N2) is catalyzed by the
nitrous oxide reductase(nosZ) gene (Huang et al., 2013). These
denitrifying genes areresponsible for N transformations resulting
in NO�3 -N removaland N2O alleviation. Although the denitrifying
genes described todate have been used as molecular markers for
quantitative andconstituent studies of denitrifying bacteria in the
media of biofil-ters and CWs (Warneke et al., 2011; Hu et al.,
2016; Wang et al.,2016a), the spatiotemporal abundance and
distribution hetero-geneity of denitrifying genes at NUA and DAS in
DNBFs isunknown. Many studies have shown that differences in the
com-munity structure patterns and abundance of denitrifying
bacterialgenes were correlated with a variety of physical and
biogeochem-ical conditions (Ruiz-Rueda et al., 2009; Chen et al.,
2014; Liu et al.,2016; Zhang et al., 2016). However, the abundance
of denitrifyingbacteria in these NUA or DAS-based DNBFs in relation
to the efflu-ent quality, DEA and potential N2O production rate
(Pot N2O) underconstant environmental conditions are currently not
wellunderstood.
Hence, this study was conducted to compare the removal
effi-ciency (total organic carbon (TOC) and NO�3 -N removal)
usingDNBFs for synthetic agricultural runoff treatment with media
ofNUA and DAS. The primary focus was the NO�3 -N removal mecha-nism
in these bioreactors, which included exploring the DEA andPot N2O,
as well as the abundance of nirS, nirK and nosZ at differentdepths
of NUA and DAS over time, and analyzing the compositionsof these
bacteria in the systems. Furthermore, the mechanisms bywhich the
abundance of denitrification bacteria influences denitri-fying
activity (DEA and Pot N2O) and effluent concentrations, aswell as
the correlation between DEA and NO�3 -N removalefficiencies were
examined to determine the main genetic factors
influencing NO�3 -N removal and N2O emission. Therefore, the
TOCvariation, N removal, and microbial community shift presented
inthis study might provide insights into strategies that
fine-tunethe operational parameters of biological processes in
agriculturalrunoff treatment by using DNBFs.
2. Materials and methods
2.1. Properties of NUA and DAS
Table S1 summarizes the characteristics of the filter
materialsused for this study. NUA and DAS were collected from a
rutile plantand waterworks in Jiangsu Province, respectively.
Following collec-tion, these industrial byproducts were washed
using distilledwater to remove dirt and floating fine particles,
after which theywere naturally air dried, ground and sieved
(particle size 4–10 mm). Bulk density and porosity were determined
using stan-dard soil science methods (Liu, 1996). The pH of the
substratewas measured in a 10% (w/v) aqueous solution using a
digital pHmeter. The specific surface area of the NUA and DAS was
measuredby the Brunauer-Emmett-Teller method with N using a
NOVA3000e surface area and pore size analyzer (Quantachrome
Instru-ments, USA).
2.2. Experimental design and operation
The laboratory-scale experimental downward-flow DNBFs
arepresented in Fig. S1. The system consisted of three parts:
NUAand DAS reaction columns, a raw water feeding subsystem, and
abackwashing subsystem. Two identical cylinder-shaped
Plexiglascolumns (110 cm in height and 40 cm in diameter) were
designed(from bottom to top) with a supporting layer of 10 cm of
cobble-stones (diameter 4–10 cm), 90 cm of reaction layers and a 10
cmwater distribution layer. NUA and DAS were added to each
con-tainer as the media at the reaction layers after preprocessing.
EachDNBF had three sampling ports distributed at different depths
inthe two columns for collection of filter materials. Three
outletswere located at the base of the columns (30 cm, effluent
1;60 cm, effluent 2; 90 cm, effluent 3) to collect samples
fromdifferent-layer effluents.
Simulated agricultural runoff was prepared according to
themethod described by Hua et al. (2016), and glucose and KNO3
dis-solved in water were respectively used as carbon and NO�3
-Nsources. Synthetic wastewater was pumped through a water
dis-tributor to ensure that the influent was uniformly
distributed.Afterward, this wastewater infiltrated the NUA or DAS
beds andwas finally discharged from the outlets. The
characteristics of thesynthetic raw water are shown in Table S2.
The overall experimentwas composed of a start-up phase and
operation periods. TheDNBFs were inoculated with the activated
sludge during start-upphase. The start-up process included the
following three stages:(i) the first stage (days 0–20), in which
there was a 0.1 m3 m�2 d�
hydraulic loading rate. During this phase, synthetic
waterremained in the columns for 3 h after the influent reached the
totalreaction volume. (ii) The second stage (days 21–40), in which
therewas a 0.15 m3 m�2 d� hydraulic loading rate under conditions
ofhydraulic retention time (HRT) = 2 h. The raw water quality inthe
second stage was the same as that used in the first stage. (iii)The
third stage (days 41–60), in which there was identical rawwater
with a 0.2 m3 m�2 d� hydraulic loading rate under aHRT = 2 h. When
the NO�3 -N concentration of effluent 3 was below4.0 mg L�1, the
start-up of the DNBFs was deemed complete. Sub-sequently, the
operation experiments were performed under thesame operational
conditions in the third stage of start-up for150 days. During the
start-up period, backwashing was not
-
554 Y. Zhang et al. / Bioresource Technology 241 (2017)
552–562
conducted to prevent the loss of biomass. At steady state,
theDNBFs were regularly backwashed every 7 days to prevent
clog-ging of the filter bed. Additionally, the system was fed
daily, oneprocess of introducing water was performed once a day
throughoutthe period, and the temperature was maintained at 21.0 �C
± 1.5 �C.
2.3. Water quality analysis
The influent and effluent samples were collected and
analyzedevery 5 days during the first 30 days and every 10 days in
the fol-lowing 180 days. The concentrations of NHþ4 -N; NO
�3 -N; NO
�2 -N,
total N and pH were determined according to the standard
meth-ods (APHA, 2002). TOC was determined by the combustion
oxida-tion nondispersive infrared absorption method using a
TOCanalyzer (TOC-L CPH, Shimadzu, Japan). The dissolved oxygen(DO)
was measured in situ with a DO meter (YSI Model no. 550A,USA). All
samples, including the controls, were analyzed intriplicate.
2.4. Denitrification activity measurements
DEA was determined in media using an acetylene
inhibitiontechnique as previously described (Ryden and Dawson,
1982).Samples of initial NUA and DAS (N0 and D0), NUA and DAS
col-lected on day 60 from 0–30 cm (N11 and D11), 30–60 cm (N12and
D12) and 60–90 cm (N13 and D13), and NUA and DAS col-lected on day
180 from depths of 0–30 cm (N21 and D21), 30–60 cm (N22 and D22)
and 60–90 cm (N23 and D23) were obtainedfrom the columns. One type
of media was prepared by mixing 25 gof sample and 25 mL of a
solution containing 1 mM glucose, 1 mMKNO3 and 1 g L�1
chloramphenicol in a 125 mL glass bottle. Chlo-ramphenicol in a
bottle was used as a supplementary nitrificationinhibitor because
acetylene diffused slowly in viscous media. Theheadspace was
evacuated and flushed for 10 min with He, afterwhich 10 mL of
acetylene were added. The samples were shakenat 25 �C, after which
the concentration of N2O was measured inthe headspace after 30, 40,
50, and 60 min of incubation by gaschromatography as previously
described (Warneke et al., 2011).N2O increased during incubation in
the presence of acetylene andthus inhibited the nitrification and
reduction of N2O to N2, whichprovided a basis for the application
of this technique to calculateDEA indirectly by utilizing Bunsen
coefficient for N2O dissolvedin water. Pot N2O was determined by
incubating parallel sampleswithout acetylene.
2.5. Quantification of the functional genes
Total genomic DNA from microbial samples (0.25 g) collected
atdifferent times and depths was first extracted and purified using
anUltra Clean Soil DNA Isolation Kit (MO BIO Laboratories, Loker
AveWest, USA), then analyzed by 1% agarose gel electrophoresis
andstored at �20 �C until use. The 16S rRNA and main
bacteriainvolved in the denitrification processes were quantified
usingreal-time polymerase chain reaction (PCR). Primers and
thermalcycling conditions used for each reaction are summarized
inTable S3. All quantitative PCR (qPCR) reactions were performedon
an ABI PRISM7500 Real-Time PCR System (Applied Biosystems,CA, USA)
in a total volume of 25 mL containing 12.5 mL SYBR PremixEx TaqTM
(Takara), 2 mL template DNA, 1 mL of each primer(5 mmol L�1), 0.5
mL ROX Reference Dye II (50�) and 8 mL RNase-free water. All
samples were run in triplicate. Standard curves wereobtained by
serial dilution from 103 to 108 copies of linearizedplasmids
containing the respective functional genes. The R2 valuefor each
standard curve exceeded 0.99 and the amplification effi-ciencies
were 85–110%.
2.6. PCR–denaturing gradient gel electrophoresis (DGGE),
sequencingand phylogenetic analysis
DNA extracted from different filter materials was subjected
tofunctional PCR amplification targeting specific genes involved
indenitrification. The primers and PCR-DGGE reaction conditionsare
shown in Table S4. A detailed description of the PCR/DGGE set-tings
used is available in Wang et al. (2016b).
The numbered dominant bands from DGGE gels were excised,washed,
and dissolved in sterile water. The eluted DNA was sub-sequently
reamplified as templates using the primer sets shownin Table S4,
but without the GC clamp. Amplified DNA was thenpurified and
ligated with the pTG19-T PCR Product Cloning Kit(Generay, Shanghai,
China) according to the manufacturer’s pro-tocols, after which the
clones were used as templates forsequencing by Biolinker Inc.
(Shanghai, China). The retrievedsequences were then compared with
those available in the Gen-Bank database using the Basic Local
Alignment Search Tool. Theobtained sequences have been deposited in
GenBank underaccession numbers KX000671 to KX000684 for nirS,
KX000652to KX000670 for nirK and KX000685 to KX000704 for nosZ.
Phy-logenetic analysis was performed using Molecular
EvolutionaryGenetics Analysis version 4.0, and neighbor-joining
trees wereconstructed using the p-distance model with a bootstrap
of1000 replications.
2.7. Statistical analysis
The intensity of the bands identified from the DGGE images
wasmeasured using the Quantity One image analysis software
(Version4.0, BioRad, USA), after which principal component analysis
(PCA)was conducted based on similarity in relative band intensity
andposition of DGGE profiles using the CANOCO software 4.5
(Micro-computer Power, Ithaca, USA).
The data presentation and treatment was accomplished usingthe
SPSS software (Version 16.0). The removal rates of TOC andNO�3 -N
were calculated as the differences between the influentand effluent
concentrations divided by the influent concentration.Differences in
DEA and Pot N2O were tested for the effects of NUAvs. DAS
substrates at different depths at the same sampling time byone-way
ANOVA and Duncan’s multiple range test. The Pearsoncorrelations
among the abundances of 16S rRNA and denitrifyingfunction genes,
gene ratios, DEA and Pot N2O of substrates andthe concentrations of
wastewater from the effluent were deter-mined. Linear regression
mode was used to predict the relationshipbetween DEA and NO�3 -N
removal efficiencies in NUA- and DAS-based DNBF. The significance
level for all tests was 0.05.
3. Results
3.1. TOC and NO�3 -N removal performance
The average removal efficiencies of TOC by NUA- and DAS-based
DNBF during the stable stage were 60.3% and 70.1%, respec-tively
(Fig. S2a). The main form of nitrogen in effluents was NO�3 -N,and
low NHþ4 -N and NO
�2 -N concentrations of 0.01–0.5 mg L
�1
were obtained (data not shown). As shown in Table S5, theNO�3 -N
level decreased gradually from influent to effluent whileflowing
through the filter material layers. After the start-up period,the
NO�3 -N removal rates were 74.8–92.2% and 73.5–85.6% for theNUA and
DAS columns, respectively (Fig. S2b), with the averageremoval
efficiency of the NUA–DNBF being 8% higher than that ofthe
DAS–DNBF.
-
Y. Zhang et al. / Bioresource Technology 241 (2017) 552–562
555
3.2. Denitrification enzyme activity and potential N2O
production rate
The DEA of DAS and NUA increased gradually over time(Fig. 1a).
The DEA of substrate sampled in the same depth at iden-tical time
was significantly higher in NUA than DAS (P < 0.05),except
between the initial media and D21 and N21. The DEA forboth
substrates at various depths did not differ significantly onday 60
and 180 (P > 0.05), except for that between N21 and N23(P <
0.05).
When compared to the untreated media, the Pot N2O of NUAand DAS
increased significantly for both sampling times (Fig. 1b).As shown
in Fig. 1b, the average Pot N2O of DAS was reduced by135% and 74%
compared to that of NUA for day 60 and 180, respec-tively. Similar
to the DEA, the Pot N2O of DAS did not change signif-icantly with
depth, but that of N12 and N13 was significantlyenhanced relative
to N11, and the highest value was observed inN23 on day 180 (Fig.
1b).
Fig. 1. DEA (a) and Pot N2O (b) of different substrates during
operation period inboth systems. Columns with the same letter for
the identical sampling time do notdiffer significantly (ANOVA;
Duncan’s multiple range test, P < 0.05).
3.3. Copies of 16S rRNA and denitrification genes (nirS, nirK
and nosZ)
In Fig. 2a–d, low copy numbers of 16S rRNA and
denitrificationgenes were found in the initial media, and the 16S
rRNA, nirS, nirKand nosZ in filter materials in the stable stage
were present at 109–1010, 108–109, 107–108 and 106–107 copies g�1
substrate, respec-tively. For comparison, the ratios of nirS/nirK
and
Pnir/nosZ of
these substrates ranged from 0.01 to 31.4 and 83.7 to
886.7,respectively. Overall,
Pnir genes were in the range of 0.3–20.4%
of the estimated 16S rRNA copies.As shown in Fig. 2a–e, the gene
copies of nirK, nosZ and 16S
rRNA in NUA were much higher than those at the same depth inDAS
on day 60 and 180, while the nirS and
Pnir numbers varied
slightly between the two substrates. The gene copy numbers
ofnirK, nosZ and 16S rRNA increased in both media as time
pro-gressed, but nirS and
Pnir remained nearly unchanged from day
60 to 180. Moreover, the top layer of media had relatively
lowergene copy numbers of most genes than the middle and lower
sub-strate. However, nirS and
Pnir of DAS decreased slightly on day
180 throughout the depth gradient.As shown in Fig. 2f–h, there
were no major differences in any of
the gene ratios spatially in both filter materials on day 60 and
180.Moreover, the nirS/nirK in DAS and
Pnir/16s rRNA in both sub-
strates decreased dramatically from day 60 to 180, while no
obvi-ous changes in nirS/nirK in NUA and
Pnir/nosZ in either media
were observed throughout the stable sampling period.
3.4. Community composition and phylogenetic analysis of
denitrifiers
Comprehensive DGGE analysis of denitrifying genes from
allsamples showed distinct patterns for each time and
location(Figs. 3a–5a). No obvious differences in nirS community
structurewere detected between DAS on day 60 and NUA (except N23)
onday 60 and 180 (Fig. S3). Moreover, with the exception of N23,the
structures of nirS at different layers did not shift
significantlyat the same sampling time in either substrate. PCA of
the nirSDGGE profiles also indicated good separation for DAS, but
not forNUA among different stable operation periods. In the
phylogenetictree of nirS (Fig. 3b), sequences from the both
substrates were sep-arated into twomain groups. For the initial
media, band 3 belongedto Group I.a at D0 and no bands were observed
at N0. When thetwo operation times were considered as a whole,
sequences fromthe various depths of NUA and DAS did not separate
from eachother clearly, and they included all of subgroup I and
Group II. Spe-cially, Ideonella (band 5) belonged to Group I.a and
undeterminedbacteria (band 2) in Group II were only found in
D21–D23, butbands 3 and 10 clustered into Group I.a were not seen
in N23.
PCA revealed clear variations in the nirK community structurefor
both substrates from the initial time to day 60 and 180(Fig. S4).
However, no temporal or spatial pattern of significantlychanging
community structures between NUA and DAS (exceptD21, N11, N21) was
observed from 60 day to 180. Additionally,the nirK community from
N11 and N21, D21, and other substrateswere distinctly clustered
into three groups during the operationperiod, and DAS rather than
NUA at the top layer followed atime-dependent shift in nirK.
Phylogenetic analysis showed thatsequences present in Group I.a,
I.c, II.a and II.c were the major typesin all samples, whereas
Group I.b was more heterogeneous anddominated by band 5 retrieved
from D21, N11 and N21. As shownin Fig. 4b, nirK compositions in
both the initial and top-layer sub-strates (except D11) were
different from those of the other media.Specifically, only 2 bands
in D0, 3 bands in N0 and 4 bands in N11and N21 were grouped into
Group I.a, while no bands in D21belonged to Group II.b. Therefore,
nirK compositions from the ini-tial media, D21, N11 and N21 were
separated from the othermedia.
The PCA results of the DGGE banding pattern from DNBFs atNUA and
DAS (except D23) revealed differences in nosZ commu-nity structures
(Fig. S5). During the stable operation period, a tem-poral pattern
of significantly changing nosZ community structurefor DAS (except
the middle-layer medium) was seen, but notime-dependent shift was
observed for NUA. In the spatial patternof the nosZ community, no
depth-dependent variation at NUA wasfound. The structure of nosZ at
DAS shifted significantly along thedepth gradient on day 180, but
no distinct variation existed amongDAS on 60 day. In Fig. 5b, 11 of
16 and 5 of 16 bands belonged to
-
Fig. 2. Number of copies of nirS (a), nirK (b), nosZ (c),
bacterial 16S rRNA (d),P
nir (e), and ratios of gene copies of nirS/nirK (f), ratios of
gene copies ofP
nir/16S rRNA (g) andgene copies of
Pnir/nosZ (h).
556 Y. Zhang et al. / Bioresource Technology 241 (2017)
552–562
Group I.a and II.b, respectively, both of which were generally
foundin all substrates except N0. The distinct bands, 7 and 17,
whichwere clustered within Group I.a and I.b, respectively, were
foundin all NUA and D23 samples, but not in other substrates. The
nosZcompositions of D0 and D21 were different from those of
othermedia since Group II.a and I.b were not observed in DO and
D21,respectively.
3.5. Factors controlling denitrification
As shown in Table 1, the concentrations of TOC and NO3-N inthe
effluents had a significant negative and positive relationshipwith
qnosZ and
Pqnir/qnosZ, respectively, but no other correla-
tions between biological gene indicators and effluents were
found.Moreover, a significant positive correlation existed between
the
-
Group .c
Group .b
Group
Group .a
(a)
(b)
Band 2Band 3
Band 1
Band 4Band 5
Band 6
Band 7Band 8
Band 9
Band 10Band 11
Band 12Band 14
Band 13
D0 D11 D12 D13 D21 D22 D23 N23 N22 N21 N13 N12 N11 N0
Fig. 3. DGGE profiles of nirS gene fragments from different
spatial-temporal NUA and DAS (a), and neighbor-joining phylogenetic
tree of nirS sequences retrieved from thenumbered DGGE bands (b).
Arrows indicate the bands excised (1–14) for sequencing. The
sequences are 425 bp. The numbers at branch points are bootstrap
values. Scale barindicates 5 changes per 100 nucleotide
positions.
Y. Zhang et al. / Bioresource Technology 241 (2017) 552–562
557
abundance of functional genes (qnirK, qnosZ, q16S rRNA) and
DEA,and between qnirS, qnirK,
Pqnir and Pot N2O. There were also neg-
ative relationships between qnirS/qnirK and DEA, and between
Pqnir/qnosZ and DEA. As shown in Fig. S6, the NO�3 -N removal
rate
in NUA–DNBF increased linearly with the DEA of NUA, but that
inDAS–DNBF was not linearly correlated with the DEA of DAS.
-
Group .c
Group .b
Group .c
Group .b
Group .a
Group .a
(a)
(b)
D0 D11 D12 D13 D21 D22 D23 N23 N22 N21 N13 N12 N11 N0
Band 1
Band 2
Band 3
Band 4
Band 5
Band 6
Band 8
Band 7
Band 9Band 10Band 11
Band 12Band 14Band 15Band 13Band 16
Band 17Band 18
Band 19
Fig. 4. DGGE profiles of nirK gene fragments from different
spatial-temporal NUA and DAS (a), and neighbor-joining phylogenetic
tree of nirK sequences retrieved from thenumbered DGGE bands (b).
Arrows indicate the bands excised (1–19) for sequencing. The
sequences are 473 bp. The numbers at branch points are bootstrap
values. Scale barindicates 5 changes per 100 nucleotide
positions.
558 Y. Zhang et al. / Bioresource Technology 241 (2017)
552–562
4. Discussion
4.1. Effective NO�3 -N reduction
Both NUA- and DAS-based DNBFs can substantially removeNO�3 -N
from wastewater. It is well known that efficient NO
�3 -N
removal rates based on suitable porosity and high surface area
ofmedia at the bioreactor are expressed in wetlands or other
terres-
trial ecosystems (Schipper et al., 2010). These substrates
werefound to have effective porosity values and adequate specific
sur-face areas (Table S1), which is helpful to biofilm growth in
thepores of the media. Denitrifying bacteria in the attached
biofilmcan use NO�3 -N as the electron acceptor via conventional
microbialdenitrification, which in turn reduces the NO�3 -N
concentration insewage. The removal rate of NO�3 -N by NUA–DNBF was
higher thanthat by DAS–DNBF possibly because of different
fluctuations of TOC
-
Group .b
Group .b
Group .a
Group .a
D0 D11 D12 D13 D21 D22 D23 N23 N22 N21 N13 N12 N11 N0
Band 1
Band 2Band 3
Band 4
Band 5
Band 6
Band 7
Band 8Band 9
Band 10
Band 11
Band 12
Band 13
Band 17Band 14Band 15
Band 16
Band 18
Band 19Band 20
(a)
(b)
Fig. 5. DGGE profiles of nosZ gene fragments from different
spatial-temporal NUA and DAS (a), and neighbor-joining phylogenetic
tree of nosZ sequences retrieved from thenumbered DGGE bands (b).
Arrows indicate the excised bands (1–20) for sequencing. The
sequences are 453 bp. The numbers at the branch points are
bootstrap values. Scalebar indicates 5 changes per 100 nucleotide
positions.
Y. Zhang et al. / Bioresource Technology 241 (2017) 552–562
559
contents between these DNBFs. The relatively slow depletion
ofavailable carbon in NUA–DNBF but not in DAS–DNBF enhancedNO�3 -N
removal efficiencies through denitrification by usingorganic carbon
from the feed water in the NUA column. It has beenconfirmed that
high TOC release is usually coupled with highNO�3 -N removal rates
(Warneke et al., 2011).
4.2. Denitrification activity and its relationship with the
abundance oftotal and denitrifying bacterial communities
In general, denitrification activity is primarily controlled by
var-ious environmental factors, including carbon availability,
moisture,properties of filter materials, influent qualities, and
denitrifying
-
Table 1Pearson correlation coefficients (R) between
denitrification functional genes, denitrification enzyme activity,
potential N2O production rate and the concentrations of
wastewaterfrom the effluent.
Eff TOC Eff NO�3 -N Eff DO DEA Pot N2O
qnirS �0.210NS �0.240NS �0.411NS 0.237NS 0.620*qnirK �0.539NS
�0.374NS 0.063NS 0.954** 0.631*qnosZ �0.718** �0.708* �0.229NS
0.881** 0.533NSq16S rRNA �0.449NS �0.410NS 0.124NS 0.768**
0.209NSP
qnir �0.301NS �0.290NS �0.376NS 0.416NS 0.717**qnirS/qnirK
0.401NS 0.210NS �0.213NS �0.799** �0.373NSP
qnir/16SrRNA bacteria 0.271NS 0.229NS �0.236NS �0.559NS
0.002NSP
qnir/qnosZ 0.652* 0.615* 0.132NS �0.720** �0.300NS
Eff: Effluent wastewater. DEA: denitrification enzyme activity.
Pot N2O: Potential N2O production rate.*and** indicate significant
(2-tailed) effects at P < 0.05 and P < 0.01, respectively.
The bold values represent significant Pearson correlation.
560 Y. Zhang et al. / Bioresource Technology 241 (2017)
552–562
bacteria (Correa-Galeote et al., 2013). The abundance of
denitrifiercommunities is often considered the dominant factor
influencingDEA (Lyautey et al., 2013). Both DEA of DAS and NUA
showed acommon temporal pattern, with the lowest value being
observedin the initial samples and the highest on day 180. This
patternwas markedly different from that of the spatial variation,
whichshowed no significant differences in DEA among major
different-depth media. These findings were supported by the
significant pos-itive correlation between DEA and the numbers of
functional bac-teria (qnirK, qnosZ, q16S rRNA), which suggests that
the peak ofDEA occurred on day 180, when the functional bacterial
abun-dances were highest, and that similar values were present in
depthgradients in which most bacteria levels (qnirK, qnosZ, q16S
rRNA)are alike. Similar results were reported in the sediments of
CWsand biofilms of rivers (Correa-Galeote et al., 2013; Lyautey et
al.,2013). Specifically, these studies revealed that the DEA of
sedi-ments was positively affected by q16S rRNA and qnirK, and
thatthis positive correlation was observed between the DEA of
biofilmsand the qnosZ. However, previous reports also found that
qnirS butnot qnosZ, or neither qnirS nor qnirK, imposed significant
controlon DEA (Correa-Galeote et al., 2013; Lyautey et al., 2013).
Theseinverse results might account for the spatiotemporal
distributionof denitrifying genes abundances among different media.
In thisstudy, the qnirS of both media changed slightly with the
depth ofDNBFs from day 60 to 180, but qnirK and qnosZ changed
substan-tially. These situations caused distinct relationships
between DEAand denitrifying genes. Furthermore, higher DEA at NUA
wascaused by the increased abundance of q16S rRNA, nirK and qnosZin
this medium. The absence of differences in DEA between D21and N21
may partly be explained by the similar rich nutrients(especially
NO�3 -N) in the upper layers of NUA- and DAS–DNBF.Tiedje (1988)
also suggested that nitrate rich environments sup-port higher DEA.
Indeed, as the bacterial community matures onday 180, limited
oxygen diffusion to the deeper layers favors thepresence of
bacterial functional groups with low oxygen require-ments at the
bottom layers (Lyautey et al., 2005), which wasadvantageous to
elevated DEA. Thus, the discrepancies of DEAbetween N21 and N23
were partially due to a slight decrease inDO at the outlet.
Denitrification is considered to be the major source of N2Ounder
most conditions (Huang et al., 2013). The spatial-temporalpatterns
of Pot N2O in DAS most likely depended on
Pqnir. We
found a strong positive relationship betweenP
qnir and Pot N2O.The slight changes in abundance of
Pnir during operation periods
as depth increased explain the lack of obvious differences in
PotN2O from DAS, regardless of temporal and spatial variations.
Whencompared to DAS, the spatiotemporal distribution of Pot N2O
fromNUA was heterogeneous and complex. NirS/
Pqnir and nirK abun-
dances showed different spatiotemporal distributions, with
nocharacteristic time or depth dependence being observed for PotN2O
at NUA, but a positive correlation in Pot N2O value. Similarly,
the genes involved in the denitrification processes were not
dis-tributed according to the spatial patterns of Pot N2O, but a
few cor-relations were observed between denitrification variables
(PotN2O) and gene (nirS, nirK) abundances (Correa-Galeote et
al.,2013). Although in this study higher qnosZ had the ability
toreduce N2O to N2 in NUA, the Pot N2O of NUA was clearly
muchhigher than that of DAS. The marginal contribution of nosZ
toN2O reduction may have been due to
Pnir and nosZ having very
different abundance ranges (P
nir was two orders of magnitudemore abundant than nosZ), which
suggested the relatively smallqnosZ brought about a low activity of
nosZ. Šimek et al. (2004)demonstrated that a low pH increases N2O
production from deni-trification. The faintly acid property of NUA
likely caused N2O pro-duction to become prominent through
accelerated denitrification.Therefore, these findings configure a
scenario of complex relation-ships between biogeochemical
properties of the substrates, nutri-ent contents, genes and
spatiotemporal distributions of DEA andPot N2O, which were
dominated by correlations with specific geneabundances.
4.3. Structure and compositions of denitrifying bacterial
communities
DGGE analysis of the denitrifying genes from substrates showeda
quickly well-adapted denitrifier community attached to the
NUA,regardless of how space and time changed, and a complicated
den-itrifier community along the depth gradient attached to DAS
dur-ing different operation periods. With the exception of very
lowdiversity or of the absence of denitrifying genes in N0,
communitystructures of nirS, nirK and nosZ in most of the NUA
tended to bestable, suggesting that NUA directly influenced the
micro-environment conditions and provided a suitable ecological
nichefor denitrifying bacterium growth and propagation. Thus,
specificphysiochemical characteristics of NUA (such as sorptive,
high sur-face area) seemingly have significant impacts on certain
denitrify-ing bacterial community structures and compositions
relative toother environmental variables. The study has also shown
that spe-cial material geochemical properties were strongly related
to bac-terial community diversity and structure (Despland et al.,
2014).The temporal differences in DAS suggested that all nirS and
mostnosZ show a shift in the community’s structures from a
circum-stance of rapid colonization with a high rate of
reproduction to astabilized environment colonized by specific
species. These find-ings are in accordance with those of previous
studies that revealedchanges in the community structure of nirS and
nosZ over time inthe CW (Ruiz-Rueda et al., 2009). However, most
nirK structureswere time-independent in this study, which differed
from generalvariations in nirK structures that have been reported
with time insoil (Yoshida et al., 2009). This apparent preference
of microorgan-isms harboring nirS at DAS for certain environments
might rule outvariations in the structure and compositions of nirK
at the samedepth throughout the operation period. Ruiz-Rueda et al.
(2007)
-
Y. Zhang et al. / Bioresource Technology 241 (2017) 552–562
561
also observed inversely related diversity indices for nirK and
nirS.Intriguingly, most structures of denitrifying genes were
space-independent. Although the differences in denitrifying
communitystructures were usually related to the depth gradient and
differ-ences in HRT among various depths (Truu et al., 2009), the
similarHRT in the DNBFs evaluated in this study likely contributed
to thestability with experimental depth. Moreover, some individuals
(i.e.,nirS of N23) separated from the main group,
temporal-independentnosZ for DAS in middle layer, and
depth-dependent nosZ for DAS onday 180, did not comply with the
prevailing trends of denitrifyingcommunity structure variations.
These discrepancies could proba-bly be ascribed at least in part to
the roles of N contents or DO dif-fusion gradients in DNBFs. It has
been verified that nutrientssupply rates and oxygen depletion
influenced denitrifier commu-nity structures (Despland et al.,
2014; Iribar et al., 2015); however,the occurrence of these
phenomena in the present study are stilldifficult to explain.
Phylogenetic analysis showed that no unique sequences dif-fered
between the DAS- and NUA-based DNBF, and that denitrifierscomprised
a diverse microbial community. Most sequences in bothsubstrates
were closely related to Burkholderiales (Group I.a in nirS,I.b in
nirK and II in nosZ), Rhodocyclales (Group II in nirK) and
Rhi-zobiales (Groups I.a, II.a and II.c in nirK, I.a in nosZ),
which areknown to dominate the denitrification process in
wastewatertreatment systems and grasslands (Chon et al., 2010; Pan
et al.,2016). In addition to being commonly detected in all
samples,Burkholderiales, Rhodocyclales, and Rhizobiales were more
adaptedto exist in nirS-nosZ, nirS, and nirK-nosZ, respectively, as
indicatedin Figs. 3b–5b Some minor genera (i.e., Pseudomonas
containingboth nirS-harboring and nosZ-harboring denitrifiers
belonging toGammaproteobacteria in Group I.b at nirS and nosZ,
Arthrobacterbelonging to Actinobacteria in Group I.c at nirS,
Staphylococcusbelonging to Bacilli in Group II.b at nirK) were also
observed in bothDNBFs. Despite only a few sequences showing
homology to theaforementioned genera, those possessing
denitrification functionswere universally obtained from
environmental samples, such assediments, soils and activated sludge
(Faulwetter et al., 2009;Harbi et al., 2010; Mulec et al., 2015).
In contrast, the nirS and nirKcommunities were more diverse than
the nosZ communities. Fur-thermore, some of the nirK sequences in
Group I.c were not clus-tered with known denitrifying populations,
suggesting novel andunique nirK communities in these DNBFs.
4.4. Genetic factors controlling denitrification and N2O
production
Microbial denitrification might be the mechanism of NO�3
-Nremoval in the experimental DNBFs because qnosZ and
Pqnir/
qnosZ ratio were identified as the main factors limiting NO�3 -N
con-centrations in effluents. Changes in q16S rRNA or
Pqnir/16SrRNA
could not be used to infer NO�3 -N variations because most
bacteriapossess at least a few copies of 16S rRNA (Zhang et al.,
2016).Nitrite reductase genes, such as nirS and nirK, which were
rela-tively more abundant than nosZ in this study, were
responsiblefor the second step of denitrification and were
indirectly involvedin NO�3 -N transformation. The principle of
these function genesmight reveal that NO�3 -N removal was not
associated with qnirS,qnirK,
Pqnir, and qnirS/qnirK. Conversely, a previous study showed
that Eff NO�3 -N decreased linearly with theP
qnir, but not qnosZ, inCWs (Chen et al., 2014). In the present
study, although no correla-tions were detected between Eff DO and
the abundance of anyfunctional genes, the similar DO concentrations
among differenttimes at both DNBFs led to considerable distribution
variations inqnosZ rather than
Pqnir. These findings suggested that increasing
qnosZ or decreasingP
qnir/qnosZ could facilitate the removal ofNO�3 -N in DNBF. The
positive relationship between DEA and Pot
N2O and the specific abundances of functional genes conformedthe
characteristic distributions of these gene quantities, whichhave
been mentioned above. Furthermore, the results shown inTable 1
indicated that the ratios of qnirS/qnirK and
Pqnir/qnosZ
would have a negative impact on enhancement of DEA via
thechanging relative abundances of the corresponding genes.
The NO�3 -N removal rate in NUA–DNBF was significantly
posi-tively correlated with DEA, which demonstrated that DEA
couldbe used to estimate NO�3 -N removal from this system.
However,there is currently debate regarding whether or not
increasingDEA is conductive to denitrification (Schipper et al.,
2010). The dis-crepancy in DAS–DNBFs might be explained by the
concurrent Nimmobilization, dissimilatory nitrate to ammonium, or
Anammox(Burgin and Hamilton, 2007), all of which could alter NO�3
-N fates.
5. Conclusion
The combination of NUA and DAS in DNBFs may enhance
nitrateremoval and minimize the adverse effects of N2O during
agricul-tural runoff treatment. Phylogenetic analysis confirmed
that thedominance of denitrifying bacterial community at both
substrateswas the same during stable and prolonged operation.
Analysis ofgenetic factors indicated that the absolute or relative
abundanceof specific functional genes primarily contributed to
denitrification.Therefore, the qnosZ and ratio of
Pqnir/qnosZ may serve as biolog-
ical indicators for nitrate removal at DNBFs. DEA could be used
topredict nitrate removal in NUA rather than DAS.
Acknowledgements
This study was supported by the National Natural
ScienceFoundation of China (No. 51308247) and the Foundation
ResearchProject of Jiangsu Province (No. BK20161100).
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
inthe online version, at
http://dx.doi.org/10.1016/j.biortech.2017.05.205.
References
APHA, 2002. Standard Methods for the Examination of Wastewater.
AmericanPublic Health Association, American water Works Association
and WaterEnvironment Federation, Washington, DC.
Burgin, A.J., Hamilton, S.K., 2007. Have we overemphasized the
role ofdenitrification in aquatic ecosystems? A review of nitrate
removal pathways.Front. Ecol. Environ. 5, 89–96.
Chen, Y., Wen, Y., Zhou, Q., Vymazal, J., 2014. Effects of plant
biomass ondenitrifying genes in subsurface-flow constructed
wetlands. Bioresour.Technol. 157, 341–345.
Chon, K., Kim, Y., Chang, N.I., Cho, J., 2010. Evaluation
wastewater stabilizingconstructed wetland, through diversity and
abundance of the nitrite reductasegene nirS, with regard to
nitrogen control. Desalination 264, 201–205.
Correa-Galeote, D., Marco, D.E., Tortosa, G., Bru, D.,
Philippot, L., Bedmar, E.J., 2013.Spatial distribution of N-cycling
microbial communities showed complexpatters in constructed wetland
sediments. FEMS Microbiol. Ecol. 83, 340–351.
Despland, L.M., Clark, M.W., Vancov, T., Aragna, M., 2014.
Nutrient removal andmicrobial communities’ development in a young
unplanted constructedwetland using BauxsolTM pellets to treat
wastewater. Ecol. Eng. 484, 167–175.
Faulwetter, J.L., Gagnon, V., Sundberg, C., Chazarenc, F., Burr,
M.D., Brisson, J.B.,Camper, A.K., Stein, O.R., 2009. Microbial
processes influencing performance oftreatment wetlands: a review.
Ecol. Eng. 35, 987–1004.
Greenan, C.M., Moorman, T.B., Kaspar, T.C., Parkin, T.B.,
Jaynes, D.B., 2009.Denitrification in wood chip bioreactors at
different water flows. J. Environ.Qual. 38, 1664–1671.
Harbi, B., Chaieb, K., Jabeur, C., Mahdouani, K., Bakhrouf, A.,
2010. PCR detection ofnitrite reductase genes (nirK and nirS) and
use of active consortiaof constructed ternary adherent
staphylococcal cultures via mixture design fora denitrification
process. World J. Microbiol. Biotechnol. 26, 473–480.
Hu, Y., He, F., Ma, L., Zhang, Y., Wu, Z.B., 2016. Microbial
nitrogen removal pathwaysin integrated vertical-flow constructed
wetland systems. Bioresour. Technol.207, 339–345.
http://dx.doi.org/10.1016/j.biortech.2017.05.205http://dx.doi.org/10.1016/j.biortech.2017.05.205http://refhub.elsevier.com/S0960-8524(17)30880-5/h0005http://refhub.elsevier.com/S0960-8524(17)30880-5/h0005http://refhub.elsevier.com/S0960-8524(17)30880-5/h0005http://refhub.elsevier.com/S0960-8524(17)30880-5/h0010http://refhub.elsevier.com/S0960-8524(17)30880-5/h0010http://refhub.elsevier.com/S0960-8524(17)30880-5/h0010http://refhub.elsevier.com/S0960-8524(17)30880-5/h0015http://refhub.elsevier.com/S0960-8524(17)30880-5/h0015http://refhub.elsevier.com/S0960-8524(17)30880-5/h0015http://refhub.elsevier.com/S0960-8524(17)30880-5/h0020http://refhub.elsevier.com/S0960-8524(17)30880-5/h0020http://refhub.elsevier.com/S0960-8524(17)30880-5/h0020http://refhub.elsevier.com/S0960-8524(17)30880-5/h0025http://refhub.elsevier.com/S0960-8524(17)30880-5/h0025http://refhub.elsevier.com/S0960-8524(17)30880-5/h0025http://refhub.elsevier.com/S0960-8524(17)30880-5/h0030http://refhub.elsevier.com/S0960-8524(17)30880-5/h0030http://refhub.elsevier.com/S0960-8524(17)30880-5/h0030http://refhub.elsevier.com/S0960-8524(17)30880-5/h0030http://refhub.elsevier.com/S0960-8524(17)30880-5/h0030http://refhub.elsevier.com/S0960-8524(17)30880-5/h0035http://refhub.elsevier.com/S0960-8524(17)30880-5/h0035http://refhub.elsevier.com/S0960-8524(17)30880-5/h0035http://refhub.elsevier.com/S0960-8524(17)30880-5/h0040http://refhub.elsevier.com/S0960-8524(17)30880-5/h0040http://refhub.elsevier.com/S0960-8524(17)30880-5/h0040http://refhub.elsevier.com/S0960-8524(17)30880-5/h0045http://refhub.elsevier.com/S0960-8524(17)30880-5/h0045http://refhub.elsevier.com/S0960-8524(17)30880-5/h0045http://refhub.elsevier.com/S0960-8524(17)30880-5/h0045http://refhub.elsevier.com/S0960-8524(17)30880-5/h0045http://refhub.elsevier.com/S0960-8524(17)30880-5/h0045http://refhub.elsevier.com/S0960-8524(17)30880-5/h0045http://refhub.elsevier.com/S0960-8524(17)30880-5/h0050http://refhub.elsevier.com/S0960-8524(17)30880-5/h0050http://refhub.elsevier.com/S0960-8524(17)30880-5/h0050
-
562 Y. Zhang et al. / Bioresource Technology 241 (2017)
552–562
Hua, G.H., Salo, M.W., Schmit, C.G., Hay, C.H., 2016. Nitrate
and phosphate removalfrom agricultural subsurface drainage using
laboratory woodchip bioreactorsand recycled steel byproduct
filters. Water. Res. 102, 180–189.
Huang, L., Gao, X., Guo, J.S., Ma, X.X., Liu, M., 2013. A review
on the mechanism andaffecting factors of nitrous oxide emission in
constructed wetlands. Environ.Earth Sci. 68, 2171–2180.
Iribar, A., Hallin, S., Pérez, J.M.S., Enwall, K., Poulet, N.,
Garabétian, F., 2015. Potentialdenitrification rates are spatially
linked to colonization patterns of nosZgenotypes in an alluvial
wetland. Ecol. Eng. 80, 191–197.
Liu, G.S., 1996. Soil Physical and Chemical Analysis &
Description of Soil Profiles.Standards Press of China, Beijing.
Liu, X.H., Wang, H.C., Long, H., Qi, L., Fan, H.T., 2016.
Optimizing and real-timecontrol of biofilm formation, growth and
renewal in denitrifying biofilter.Bioresour. Technol. 209,
326–332.
Lyautey, E., Jackson, C.R., Cayrou, J., Rols, J.L., Garabetian,
F., 2005. Bacterialcommunity succession in natural river biofilm
assemblages. Microb. Ecol. 50,589–601.
Lyautey, E., Hallin, S., Teissier, S., Iribar, A., Compin, A.,
Philippot, L., Garabetian, F.,2013. Abundance, activity and
structure of denitrifier communities inphototrophic river biofilms
(River Garonne, France). Hydrobiologia 716, 177–187.
Mulec, J., Kristufek, V., Chronakova, A., Oarga, A., Scharfen,
J., Sestauberova, M.,2015. Microbiology of healing Mud (Fango) from
Roman thermae aquae IASAEarchaeological site (VaraA3/4dinske
Toplice, Croatia). Microb. Ecol. 69, 293–306.
Pan, H., Li, Y., Guan, X.M., Li, J.Y., Xu, X.Y., Liu, J., Zhang,
Q.C., Xu, J.M., Di, H.J., 2016.Management practices have a major
impact on nitrifier and denitrifiercommunities in a semiarid
grassland ecosystem. J. Soils Sediments 16, 896–908.
Ruiz-Rueda, O., Trias, R., Garcia-Gil, L.J., Bañeras, L., 2007.
Diversity of the nitritereductase gene nirS in the sediment of a
free-water surface constructedwetland. Int. Microbiol. 10,
253–260.
Ruiz-Rueda, O., Hallin, S., Bañeras, L., 2009. Structure and
function of denitrifyingand nitrifying bacterial communities in
relation to the plant species in aconstructed weland. FEMS
Microbiol. Ecol. 67, 308–319.
Ryden, J.C., Dawson, K.P., 1982. Evaluation of the acetylene
inhibition technique forthe measurement of denitrification in
grassland soils. J. Sci. Food Agric. 33,1197–1206.
Schipper, L.A., McGill, A., 2008. Nitrogen transformation in a
denitrification layerirrigated with dairy factory effluent. Water.
Res. 42, 2457–2464.
Schipper, L.A., Robertson, W.D., Gold, A.J., Jaynes, D.B.,
Cameron, S.C., 2010.Denitrifying bioreactors—an approach for
reducing nitrate loads to receivingwaters. Ecol. Eng. 168,
493–504.
Šimek, M., Elhottová, D., Klimeš, F., Hopkins, D.W., 2004.
Emission of N2O and CO2,denitrification measurements and soil
properties in red clover and ryegrassstands. Soil Biol. Biochem.
36, 9–21.
Syakila, A., Kroeze, C., 2011. The global nitrous oxide budget
revisited. GreenhouseGases Meas. Manage. 1, 17–26.
Tiedje, J.M., 1988. Ecology of denitrification and dissimilatory
nitrate reduction toammonium. In: Zehnder, A.J.B. (Ed.), Biology of
Anaerobic Microorganisms.Johan Wiley and Sons, New York, pp.
179–244.
Truu, M., Juhanson, J., Truu, J., 2009. Microbial biomass,
activity and communitycomposition in constructed wetlands. Sci.
Total Environ. 407, 3958–3971.
Wang, H.L., Ji, G.D., Bai, X.Y., 2016a. Distribution patterns of
nitrogen micro-cyclefunctional genes and their quantitative
coupling relationships with nitrogentransformation rates in a
biotrickling filter. Bioresour. Technol. 209, 100–107.
Wang, L.M., Wang, X., Yang, F., Kong, M., Peng, F.Q., Chao,
J.Y., Gao, Y.X., Wu, D., Zhu,Y.M., Zhang, Y.M., 2016b. Nitrogen
removal performance and ammonia- andnitrite-oxidizing bacterial
community analysis of a novel industrial waste-based biofilter.
Chem. Eng. J. 299, 156–166.
Warneke, S., Schipper, L.A., Matiasek, M.G., Scow, K.M.,
Cameron, S., Bruesewitz, D.A., McDonald, I.R., 2011. Nitrate
removal, communities of denitrifiers andadverse effects in
different carbon substrates for use in denitrification beds.Water.
Res. 45, 5463–5475.
Wendling, L.A., Douglas, G.B., Coleman, S., Yuan, Z., 2012.
Nutrient and dissolvedorganic carbon removal from water using
mining and metallurgical by-products. Water. Res. 46,
2705–2717.
Yoshida, M., Ishii, S., Otsuka, S., Senoo, K., 2009. Temporal
shifts in diversity andquantity of nirS and nirK in a rice paddy
field soil. Soil. Biol. Biochem. 41, 2044–2051.
Zhang, Y., Ji, G.D., Wang, R.J., 2016. Functional gene groups
controlling nitrogentransformation rates in a groundwater-restoring
denitrification biofilter underhydraulic retention time
constraints. Ecol. Eng. 87, 45–52.
http://refhub.elsevier.com/S0960-8524(17)30880-5/h0055http://refhub.elsevier.com/S0960-8524(17)30880-5/h0055http://refhub.elsevier.com/S0960-8524(17)30880-5/h0055http://refhub.elsevier.com/S0960-8524(17)30880-5/h0060http://refhub.elsevier.com/S0960-8524(17)30880-5/h0060http://refhub.elsevier.com/S0960-8524(17)30880-5/h0060http://refhub.elsevier.com/S0960-8524(17)30880-5/h0065http://refhub.elsevier.com/S0960-8524(17)30880-5/h0065http://refhub.elsevier.com/S0960-8524(17)30880-5/h0065http://refhub.elsevier.com/S0960-8524(17)30880-5/h0070http://refhub.elsevier.com/S0960-8524(17)30880-5/h0070http://refhub.elsevier.com/S0960-8524(17)30880-5/h0075http://refhub.elsevier.com/S0960-8524(17)30880-5/h0075http://refhub.elsevier.com/S0960-8524(17)30880-5/h0075http://refhub.elsevier.com/S0960-8524(17)30880-5/h0080http://refhub.elsevier.com/S0960-8524(17)30880-5/h0080http://refhub.elsevier.com/S0960-8524(17)30880-5/h0080http://refhub.elsevier.com/S0960-8524(17)30880-5/h0085http://refhub.elsevier.com/S0960-8524(17)30880-5/h0085http://refhub.elsevier.com/S0960-8524(17)30880-5/h0085http://refhub.elsevier.com/S0960-8524(17)30880-5/h0085http://refhub.elsevier.com/S0960-8524(17)30880-5/h0090http://refhub.elsevier.com/S0960-8524(17)30880-5/h0090http://refhub.elsevier.com/S0960-8524(17)30880-5/h0090http://refhub.elsevier.com/S0960-8524(17)30880-5/h0090http://refhub.elsevier.com/S0960-8524(17)30880-5/h0095http://refhub.elsevier.com/S0960-8524(17)30880-5/h0095http://refhub.elsevier.com/S0960-8524(17)30880-5/h0095http://refhub.elsevier.com/S0960-8524(17)30880-5/h0100http://refhub.elsevier.com/S0960-8524(17)30880-5/h0100http://refhub.elsevier.com/S0960-8524(17)30880-5/h0100http://refhub.elsevier.com/S0960-8524(17)30880-5/h0105http://refhub.elsevier.com/S0960-8524(17)30880-5/h0105http://refhub.elsevier.com/S0960-8524(17)30880-5/h0105http://refhub.elsevier.com/S0960-8524(17)30880-5/h0110http://refhub.elsevier.com/S0960-8524(17)30880-5/h0110http://refhub.elsevier.com/S0960-8524(17)30880-5/h0110http://refhub.elsevier.com/S0960-8524(17)30880-5/h9000http://refhub.elsevier.com/S0960-8524(17)30880-5/h9000http://refhub.elsevier.com/S0960-8524(17)30880-5/h0115http://refhub.elsevier.com/S0960-8524(17)30880-5/h0115http://refhub.elsevier.com/S0960-8524(17)30880-5/h0115http://refhub.elsevier.com/S0960-8524(17)30880-5/h0120http://refhub.elsevier.com/S0960-8524(17)30880-5/h0120http://refhub.elsevier.com/S0960-8524(17)30880-5/h0120http://refhub.elsevier.com/S0960-8524(17)30880-5/h0120http://refhub.elsevier.com/S0960-8524(17)30880-5/h0120http://refhub.elsevier.com/S0960-8524(17)30880-5/h0125http://refhub.elsevier.com/S0960-8524(17)30880-5/h0125http://refhub.elsevier.com/S0960-8524(17)30880-5/h0130http://refhub.elsevier.com/S0960-8524(17)30880-5/h0130http://refhub.elsevier.com/S0960-8524(17)30880-5/h0130http://refhub.elsevier.com/S0960-8524(17)30880-5/h0135http://refhub.elsevier.com/S0960-8524(17)30880-5/h0135http://refhub.elsevier.com/S0960-8524(17)30880-5/h0140http://refhub.elsevier.com/S0960-8524(17)30880-5/h0140http://refhub.elsevier.com/S0960-8524(17)30880-5/h0140http://refhub.elsevier.com/S0960-8524(17)30880-5/h0145http://refhub.elsevier.com/S0960-8524(17)30880-5/h0145http://refhub.elsevier.com/S0960-8524(17)30880-5/h0145http://refhub.elsevier.com/S0960-8524(17)30880-5/h0145http://refhub.elsevier.com/S0960-8524(17)30880-5/h0150http://refhub.elsevier.com/S0960-8524(17)30880-5/h0150http://refhub.elsevier.com/S0960-8524(17)30880-5/h0150http://refhub.elsevier.com/S0960-8524(17)30880-5/h0150http://refhub.elsevier.com/S0960-8524(17)30880-5/h0155http://refhub.elsevier.com/S0960-8524(17)30880-5/h0155http://refhub.elsevier.com/S0960-8524(17)30880-5/h0155http://refhub.elsevier.com/S0960-8524(17)30880-5/h0160http://refhub.elsevier.com/S0960-8524(17)30880-5/h0160http://refhub.elsevier.com/S0960-8524(17)30880-5/h0160http://refhub.elsevier.com/S0960-8524(17)30880-5/h0165http://refhub.elsevier.com/S0960-8524(17)30880-5/h0165http://refhub.elsevier.com/S0960-8524(17)30880-5/h0165
Nitrate removal, spatiotemporal communities of denitrifiers and
the importance of their genetic potential for denitrification in
novel denitrifying bioreactors1 Introduction2 Materials and
methods2.1 Properties of NUA and DAS2.2 Experimental design and
operation2.3 Water quality analysis2.4 Denitrification activity
measurements2.5 Quantification of the functional genes2.6
PCR–denaturing gradient gel electrophoresis (DGGE), sequencing and
phylogenetic analysis2.7 Statistical analysis
3 Results3.1 TOC and [$] {{\rm NO}}_{3}^{-} {\rm {\hyphen}N} [$]
removal performance3.2 Denitrification enzyme activity and
potential N2O production rate3.3 Copies of 16S rRNA and
denitrification genes (nirS, nirK and nosZ)3.4 Community
composition and phylogenetic analysis of denitrifiers3.5 Factors
controlling denitrification
4 Discussion4.1 Effective [$] {{\rm NO}}_{3}^{-} {\rm
{\hyphen}N} [$] reduction4.2 Denitrification activity and its
relationship with the abundance of total and denitrifying bacterial
communities4.3 Structure and compositions of denitrifying bacterial
communities4.4 Genetic factors controlling denitrification and N2O
production
5 ConclusionAcknowledgementsAppendix A Supplementary
dataReferences