Influence of season and plant species on the 
abundance and diversity of sulfate reducing 
bacteria and ammonia oxidizing bacteria in 
constructed wetland microcosms
Authors: Jennifer L. Faulwetter, Mark D. Burr, Albert 
E. Parker, Otto R. Stein, & Anne K. Camper
NOTICE: The final publication is available at Springer via http://dx.doi.org/10.1007/
s00248-012-0114-y. 
Faulwetter JL, Burr MD, Parker AE, Stein OR, Camper AK, "Influence of season and plant species 
on the abundance and diversity of sulfate reducing bacteria and ammonia oxidizing bacteria in 
constructed wetland microcosms," Microbial Ecology, July 2013 65(1):111–127.
Made available through Montana State University’s ScholarWorks 
scholarworks.montana.edu 
Influence of Season and Plant Species on the Abundance 
and Diversity of Sulfate Reducing Bacteria and Ammonia 
Oxidizing Bacteria in Constructed Wetland Microcosms
Jennifer L. Faulwetter, Mark D. Burr,  Albert E. Parker, Otto R. Stein, & Anne K. 
Camper
Abstract 
Constructed wetlands offer an effective means for 
treatment of wastewater from a variety of 
sources. An un-derstanding of the microbial 
ecology controlling nitrogen, carbon and sulfur 
cycles in constructed wetlands has been 
identified as the greatest gap for optimizing 
performance of these promising treatment 
systems. It is suspected that op-erational factors 
such as plant types and hydraulic operation 
influence the subsurface wetland environment, 
especially redox, and that the observed variation 
in effluent quality is due to shifts in the 
microbial populations and/or their activ-ity. This 
study investigated the biofilm associated 
sulfate reducing bacteria and ammonia oxidizing 
bacteria (using the dsrB and amoA genes, 
respectively) by examining a variety of surfaces 
within a model wetland (gravel, thick roots, fine 
roots, effluent), and the changes in activity 
(gene abun-dance) of these functional groups 
as influenced by plant species and season. 
Molecular techniques were used includ-ing 
quantitative PCR and denaturing gradient gel 
electro-phoresis (DGGE), both with and 
without propidium monoazide (PMA) 
treatment. PMA treatment is a method for 
excluding from further analysis those cells with 
compro-mised membranes. Rigorous statistical 
analysis showed an interaction between the 
abundance of these two functional groups with 
the type of plant and season (p<0.05). The 
richness of the sulfate reducing bacterial 
community, as indicated by DGGE profiles, 
increased in planted vs. unplanted 
microcosms. For ammonia oxidizing bacteria, 
season had the greatest impact on gene 
abundance and diversity (higher in summer than 
in winter). Overall, the primary influence of 
plant presence is believed to be related to root 
oxygen loss and its effect on rhizosphere redox.
Introduction
Constructed wetlands (CW) are a technology 
utilizing engi-neered wetland systems for the 
treatment of a variety of wastewaters ranging 
from domestic sources to storm runoff. Although 
the operation of CW is relatively well 
understood, there is a lack of understanding of 
the microbial impact on performance [21]. 
Many studies have investigated micro-bially 
mediated processes in CW by focusing on net 
changes in the concentration of specific 
chemicals or waste constit-uents [78, 79, 93]. 
Microbial communities have been shown to 
effectively contribute to the removal of 
chemical pollu-tants in CW [34, 71]; however, 
microbiology remains the most underrepresented 
area of research in this field. Further-more, the 
role of biofilms and their function within these 
systems have been virtually ignored. Presently, 
these gaps in knowledge may impede the 
effective design and operation of CW [21, 81]. 
An understanding of the community struc-ture of 
microorganisms involved in biogeochemical 
process-es, their location and distribution 
within the CW, and the extent of seasonal 
population shifts are important for ad-vancing 
CW technology.
Microbial biofilm communities are found on 
every avail-able surface in CW, including roots, 
as biofilms are capable of creating stable, 
protected environments for microbial survival 
[31]. It is also possible for oxygen gradients 
to develop within the biofilm due to the activity 
of the biofilm microorganisms themselves [12, 
80]. Therefore, biofilms may contain 
microenvironments that allow for the simulta-
neous presence and growth of both anaerobic 
and aerobic
organisms. Root surfaces are very dynamic and heterogeneous,
which promotes colonization by varying populations of micro-
organisms. Plant roots support large biofilm populations pri-
marily because they are sites of oxygen release and root
exudation [6, 13, 53]. Chemical exudates released by plant
roots are generally plant species specific, and plants are known
to selectively enrich the rhizosphere with microorganisms ac-
cordingly [13]. Because of their effects on rhizosphere biofilm
populations, root oxygen loss (ROL) and root exudation may
ultimately affect effluent water quality.
Sulfate reducing bacteria (SRB) are important in CW
because sulfate is a common pollutant in a variety of waste-
water types including domestic wastewater and acid mine
drainage. SRB have a vital role in the geochemical cycling
of sulfur, are important for many biological processes and
for the generation of alkalinity in CW [41]. SRB utilize
sulfate (SO4
−2) as a terminal electron acceptor in the anaer-
obic oxidation of organic substrates [35] and are critically
important since they are the only known organisms to per-
form this function. SRB improve effluent water quality by
removing sulfate from the aqueous phase. However, precip-
itates of sulfide remain within the system and CW probably
accumulate sulfur over time. SRB are found mainly below
the soil (or water) surface where anoxic environments are
best suited for sulfate reduction. Historically, SRB have
been considered strict anaerobes, but more recent research
has shown that SRB can persist in oxic conditions and
survive extended periods of oxygen exposure [12, 18, 72].
Many studies have assumed SRB activity in CW based on
observations of high sulfate input, low sulfate output, high
sulfide concentrations, minimal sulfate uptake by CW
plants, and low redox conditions [3, 9, 35, 76, 79].
Ammonia is another common pollutant found in waste-
waters (e.g., domestic and agricultural wastewater) and can
have detrimental effects on the environment when dis-
charged. Nitrification plays a critical role in the biogeo-
chemical cycling of nitrogen and has been documented in
a variety of CW systems [17, 82, 83]. In bacteria, nitrifica-
tion occurs via two aerobic reactions: ammonia to nitrite, by
ammonia oxidizing bacteria (AOB), and nitrite to nitrate, by
nitrite oxidizing bacteria (NOB). Ammonia oxidizing ar-
chaea (AOA) have been detected in wastewater treatment
[62], natural wetlands [74] and CW [73]. The oxic condi-
tions required for nitrification can limit transformation of
ammonia in traditional wastewater treatment systems [61],
but plants in CW may be able to provide sufficient oxygen
for nitrification through their roots [10, 11, 39, 71, 96].
Previous research in our laboratory has long inferred the
presence of a robust SRB population within CW micro-
cosms [3, 9, 79]. Most recently, Taylor et al. [86] showed
that microcosms planted with Deschampsia cespitosa had
high removal efficiencies for organic carbon (COD) (nearly
100 %) and seasonally dependent removal of sulfate
(highest in summer) while unplanted microcosms and those
planted with L. cinereus had slightly lower overall COD
removal (80 %) but constant sulfate removal regardless of
season. Since constant sulfate removal was observed in the
unplanted control and L. cinereusmicrocosms, the fluctuations
in seasonal sulfate removal performance (determined by efflu-
ent water quality) were assumed to be linked to both plant
presence (vs. unplanted control) and plant species selection.
Redox data revealed very consistent values in the range of −200
to −250 mV for the unplanted control and L. cinereus micro-
cosms regardless of season. D. cespitosa microcosms also
tracked in this redox range in summer, but had redox readings
during successive winters in the 0 to +400 mV range [84, 85].
The effect ofD. cespitosa on increasing redox in the entire bulk
water during winter is attributed to ROL, and was the primary
reason it was selected for study in the current research.
Molecular methods are increasingly being used to inves-
tigate microbial diversity and activity in environmental sys-
tems such as CW, because culture-based methodologies are
severely limiting [4, 69]. An overall perspective of the
microbial diversity can be obtained by targeting the 16S
rRNA gene, but this approach has limited utility for inves-
tigating specific microbial groups. For this reason, it is
becoming more common to target specific functional genes.
Furthermore, since DNA from nonviable cells has been
shown to remain intact for a certain time in the environment
[40], DNA-based PCR technologies cannot normally distin-
guish between DNA extracted from live (potentially active)
and dead (inactive) cells. Differentiating between microbial
populations that are merely present in biofilms vs. those that
are potentially active may provide a better link between
biogeochemical processes and community structure [42,
51]. The live–dead continuum continues to be debated,
however, one criterion for a live cell is an intact cell membrane,
which can be determined experimentally as the ability to
exclude from the cell membrane-impermeable dyes [30, 56].
Recently, Nocker et al. [57, 58] introduced a method designed
to distinguish between cells with intact vs. compromised mem-
branes based on treatment with propidium monoazide (PMA)
prior to DNA extraction. PMA is capable of entering the cells
of bacteria with compromised or damaged cell membranes and
binding to the DNA. Exposure to light during processing
causes PMA to become irreversibly bound to the DNA, which
is then permanently inhibited from subsequent PCR amplifi-
cation. This process leaves only DNA from cells with intact
cell membranes for further analysis using PCR methodology
[57]. PMA treatment also excludes extracellular DNA from
subsequent amplification.
The microbial diversity found in natural environments is
an asset that can be utilized for bioremediation, but the
diversity also makes the study of microbial ecology in these
environments difficult. In CW, optimization of effluent water
quality is the main objective; therefore, bulk water sampling
would seem appropriate for understanding the community
dynamics within. However, most nutrient removal in CW very
likely takes place in root and matrix surface biofilms, yet these
environments are difficult to sample. The plant rhizosphere in
CW and other vegetated soils has been found to have a differ-
entmicrobial community compared to the bulk substratum [46,
75]. Furthermore, composite sampling methods (such as soil
cores) can underestimate actual microbial diversity because
they tend to combine regions of high and low diversity [42,
44]. Exactly how a CW should be sampled is still unknown.
The goal of this study was to determine the diversity and
activity of SRB andAOB communities within CWmicrocosms
by evaluating the relative DNA copy numbers by quantitative
PCR (qPCR) of two functional genes, dsrB (involved in sulfate
reduction) and amoA (involved in nitrification). dsrB, the
smaller fragment of the dissimilatory sulfite reductase gene
(dsrAB), was selected as it is essential for anaerobic sulfate
reduction and has been found in all dissimilatory sulfate-
reducing prokaryotes examined thus far [1, 5, 27, 51]. AOB
were targeted by examining the beta proteobacterial ammonia
monooxygenase gene (amoA) as this gene is responsible for the
initial rate limiting step in nitrification and has been used in a
variety of ecological studies investigating AOB [19, 38, 45, 67,
95]. The potentially active community was defined as consist-
ing of cells with intact cell membranes as determined by PMA
treatment. The diversity within each of these genes was also
evaluated as the number of bands (richness) in denaturing
gradient gel electrophoresis (DGGE) profiles.
CW microcosms were planted with one of two different
macrophyte species or left unplanted and maintained in a
greenhouse with temperature regulation that simulated natural
ambient conditions in a seasonally cold climate. Microcosms
were fed a synthetic wastewater that simulated post-primary
domestic wastewater effluent. Microcosms were destructively
sampled in summer and winter seasons to investigate the
effects of plant species and season on the SRB and AOB
functional communities. To investigate whether specific loca-
tions within the CW were ecologically more relevant habitats
for AOB and SRB than others, i.e., selectively enriched, sam-
ples were collected from bulk effluent water, gravel (two
depths) and roots (thick, thin, ultra-fine). Effluent water quality
analyses were also performed to assess reduction of sulfate and
removal of ammonia as well as used to correlate CW perfor-
mance (by pollutant removal) with the relative quantities of the
targeted genes.
Methods
Constructed Wetland Operation
This research was conducted in the same facilities used in
earlier studies [78, 79, 86]. CW microcosms were used to
simulate an operational subsurface flow CWand were main-
tained in a greenhouse at the Plant Growth Center at Mon-
tana State University in Bozeman, MT (46°N, 111°W). Four
replicates each of unplanted controls and monocultures of
D. cespitosa and L. cinereus were planted in model subsur-
face wetlands consisting of 15 cm diameter by 30 cm tall
polyvinyl chloride columns filled with 1–5 mm diameter
gravel. Greenhouse temperature was changed every 60 days
to mimic natural seasonal cycles. The annual temperature
sequence was 4, 8, 16, 24, 16, 8, and 4°C. Supplemental
lighting was not used. Patterns of natural light and con-
trolled temperature induced normal seasonal cycles of plant
dormancy and growth. Microcosms were fed synthetic
wastewater simulating post-primary domestic wastewater
effluent [86]. There were three 20-day batches at each
temperature. Between batches, the microcosms were com-
pletely drained. Periodic measurement of sulfate, ammonia,
and COD in the bulk water was by standard methods as
reported previously [85]. Plants were grown for a minimum
of 12 months prior to the first sampling date. All sampling
for this research was done during one winter (4°C) and the
following summer (24°C).
Plant Species Selection
The two plant species investigated in this research were
selected from a list of 19 species based upon their perfor-
mance in earlier CW microcosm experiments [84, 86]. The
objective was to compare two species at opposite ends of the
spectrum in terms of their apparent effects on nutrient re-
moval (carbon and sulfate) and root oxygen release in the
CW. Plants were selected based upon COD removal and
oxygen release because these are standard criteria for eval-
uating CW performance [71]. CW columns planted with D.
cespitosa were very effective at carbon removal, as demon-
strated by reductions in chemical oxygen demand (COD).
D. cespitosa also readily released oxygen from its roots
regardless of season [84-86]. Thus, it had the potential to
create oxic microenvironments immediately surrounding the
root in the otherwise anoxic depths of the CW. In contrast,
CW columns planted with L. cinereus provided poor COD
removal and undetectable levels of root oxygen release [84,
86]. It was also advantageous that both species belonged to
the same family (Poaceae), because potential differences in
plant physiology were minimized. Unplanted control col-
umns (containing only gravel) were also included in the
experimental design.
The plants were transplants from microcosms used in
earlier experiments and were 12 months old at the time of
the first sampling and 18 months old at the second sampling.
In addition, the gravel used in each microcosm was also
obtained from earlier experiments and was plant-species
specific, that is, only gravel from earlier D. cespitosa
microcosms was used for replanting this species, and so on.
There is no standard for such studies, and the literature
reports a wide range of plant ages. Molle et al. [52] allowed
plants to mature for 3 years to minimize plant aging effects.
In contrast, Gagnon et al. [26] sampled plant microcosms
that were less than 6 months old. Iasur-Kruh et al. [36] used
3-month-old microcosms to model a 3-year-old CW. Wang
et al. [92] used plantings that were at least 2 years old, but
noted that even with mature plants, there were some differ-
ences in plant performance between senescent plants in the
fall compared to fast growing plants in the spring.
Microcosm Destructive Sampling
CWmicrocosms were destructively sampled, in duplicate, at
the completion of the third batch (day 20) at the specified
temperature (4 °C, winter, and 24 °C, summer). Six samples
were obtained from each microcosm for the ultimate purpose
of DNA extraction. First, the column was completely drained
and the water was collected as effluent (E1). Next, the plant
and root ball were removed and separated from any residual
rootbound gravel by swirling the plant in a container of tap
water. Roots of two diameters were then aseptically excised
from the plant, thicker and older roots near the crown (~1–
2 mm diameter, R2), and thinner and younger roots near the tip
(~0.5–1 mm diameter, R3). Ultra-fine roots that had detached
from the plant during processing (R4, < ~0.5 mm diameter)
were found floating in the container of tap water and were
skimmed from the surface. Finally, gravel from the top of the
column (G5) and from the bottom (G6) were collected.
The gravel from the bottom of the column was consid-
ered anoxic and interesting for its potential to harbor anaer-
obic (or facultative) microorganisms. For that reason, it was
immediately transferred to an anaerobic chamber to protect
anaerobes against oxygen exposure. The purpose of this
intensive sampling was to determine whether certain sample
locations were selectively enriched for the target genes
(amoA or dsrB) and/or more responsive to seasonal or plant
species differences.
Duplicate samples were collected from each column lo-
cation, one for PMA treatment, the other untreated. Briefly,
the effluent sample was filtered (250 ml), and the filter was
aseptically divided in half for PMA-treated and untreated
processing. For PMA treatment, clear 1.7 ml microcentri-
fuge tubes (www.biotang.com) were used for maximum
transmittance of light (see PMA protocol below). Untreated
samples were transferred directly to DNA extraction tubes
(MO BIO PowerSoil™ DNA Isolation Kit; www.mobio.
com). Thin (0.03 g) and thick roots (0.05 g) were collected
in duplicate and aseptically transferred to 1.7 ml clear tubes
or to DNA extraction tubes. Ultra-fine roots were skimmed
into a 100 ml volume of water, filtered (41 Whatman paper
filter (www.whatman.com)), and the filter contents were
also transferred to either clear tubes or DNA extraction
tubes. To remove biofilm from the gravel surface, each of
the gravel samples collected (7.5 ml) was vortexed with
phosphate buffered saline (pH 7.2) and 10 ml sterile sand
for 60 s. For PMA treatment, 3 ml of the supernatant was
transferred in 0.5 ml aliquots to six clear 1.7-ml microcen-
trifuge tubes. For the untreated sample, six 0.5-ml aliquots
were sequentially pelleted into DNA extraction tubes.
PMA Treatment
A total volume of 0.5 ml for each sample was used for PMA
treatment. For the effluent samples and all root samples,
0.5 ml of sterile effluent filtrate was added to the clear tubes.
Since PMA treatment is based upon the integrity of the cell
membrane, sterilized filtrate was used to minimize any
osmotic shock or artificial membrane damage to the cells
prior to treatment with PMA. PMA treatment was as de-
scribed by Nocker et al. [58], except as modified below. To
each tube, 1.5 μl of PMA (20 mM stock solution prepared in
20 % dimethyl sulfoxide; Biotium, Inc., Hayward, CA) was
added and each tube was shaken vigorously. Tubes were
incubated in the dark on ice for 5 min with the exception of
bottom gravel samples which were exposed to PMA for 7 to
10 min. PMAwas then inactivated by exposing the tubes to
light for 2 min using a 650-W halogen light source (sealed
beam lamp, FCW 120 V, 3,200 K; GE Lighting, General
Electric Co., Cleveland, OH). Samples were shaken during
light exposure. Root samples were exposed to light for
4 min to minimize the effect of shadowing by root material.
Upon completion of PMA treatment, samples were trans-
ferred to MO BIO PowerBead Tubes (MO BIO Power-
Soil™ DNA Isolation Kit) for DNA extraction. Gravel
samples were processed in six 0.5-ml aliquots and after
PMA inactivation. Aliquots were combined and concentrat-
ed by sequential centrifugation for 5 min at 5,000×g into a
DNA extraction tube.
DNA Extraction
The MO BIO PowerSoil™ DNA Isolation Kit was used to
complete the DNA extraction as described in the manufac-
turer’s protocol with the exception that instead of vortexing,
PowerBead tubes were placed into the FastPrep® Instrument
(Qbiogene, Inc.) at speed 5.5 for 45 s. DNA yield was
estimated on an agarose gel with ethidium bromide staining,
serial dilutions were performed for PCR, and the DNA
preparations were stored at −20 °C.
Conventional PCR
Conventional PCR was performed to screen for the target
genes and to obtain products to be analyzed by DGGE. In
some cases, primers for both genes were combined in a
single multiplex PCR reaction, primarily to screen for the
target genes and to conserve sample DNA and PCR
reagents. All PCR reactions (20 μl) were performed using
2× GoTaq® Green Master Mix (www.promega.com). The
PCR reaction mixture consisted of 10 μl 2× GoTaq® Green
Master Mix, 0.5 μl ultrapure bovine serum albumin (BSA)
(50 mg/ml; Ambion, www.ambion.com), 2.5 μl nuclease-
free water, 1 μl each of forward and reverse primers
(12.5 μM), and 5 μl 1:10 diluted (unquantified) template
DNA. Oligonucleotide primers were synthesized by Inte-
grated DNATechnologies (www.idtdna.com) and described
in Table 1. Multiplex reactions were adjusted to contain
10 μl 2× GoTaq® Green Master Mix, 0.5 μl ultrapure
BSA (50 mg/ml; Ambion, www.ambion.com), 0.5 μl
nuclease-free water, 1 μl of each forward and reverse primer
(12.5 μM), and 5 μl 1:10 diluted (unquantified) template
DNA. All PCR amplifications were performed on an Eppen-
dorf Mastercycler® ep thermal cycler (Eppendorf North
America, www.eppendorfna.com) using the program speci-
fied in Table 2. PCR products were confirmed by agarose
gel electrophoresis and staining with ethidium bromide.
Sulfate Reducing Bacteria (for DGGE)
PCR primers DsrBF and Dsr4R (Table 1) target the β-subunit
of the dissimilatory sulfite reductase gene (dsrB, required for
sulfate reduction). Presumptive presence of the dsrB gene was
indicated on an agarose gel by a 370-bp PCR product. Primer
DsrBF was also synthesized with a 5′ 40-bp GC clamp and
was paired with primer Dsr4R for amplifying fragments to be
analyzed by DGGE. The PCR program specified by Geets et
al. [27] (Table 2) was used for amplification. Products were
analyzed by DGGE.
Multiplex PCR
PCR primers DsrBF and Dsr4R (Table 1) were combined
with primers RottF and RottR (Table 1). Presumptive pres-
ence of the genes was indicated on an agarose gel by 370-
and 491-bp PCR products. The PCR program specified by
Bahr et al. [5] (Table 2) was used for amplification. Products
were diluted and used for additional PCR amplification with
the Rott primers and analysis by DGGE.
Ammonia Oxidizing Bacteria (for DGGE, with Multiplex
Above)
PCR primers RottF and RottR (Table 1) target the ammonia
monooxygenase gene (amoA, required for ammonia oxida-
tion to nitrite). Presumptive presence of the amoA gene was
indicated on an agarose gel by a 491-bp PCR product.
Primer RottR was also synthesized with a 5′ 40-bp GC
clamp and was paired with primer RottF for amplifying
fragments to be analyzed by DGGE. Multiplex PCR product
was diluted 1:100 and 5 μl was used as template in the
reactions subsequently analyzed by DGGE. For PCR prod-
ucts to be cloned and used as qPCR standards, 5 μl 1:10
diluted (unquantified) template DNA was used for each
reaction. The PCR program specified by Bahr et al. [5]
(Table 2) was used for amplification. Reamplified multiplex
PCR products were used for DGGE analysis.
DGGE
DGGE was performed on functional PCR products (dsrB
and amoA) (with and without PMA treatment) from com-
munity DNA using a DCode™ system (www.biorad.com)
and reagents from Sigma-Aldrich (www.sigmaaldrich.com).
Table 1 Primer sequences used in this study
Primer Target Sequence (5′ to 3′) Method Reference
DsrBF dsrB gene (dsr β-subunit) CAACATCGTYCAYACCCAGGG PCRa-DGGE/qPCR [27]
Dsr4R dsrB gene (dsr β-subunit) GTGTAGCAGTTACCGCA PCRa-DGGE/qPCR [91]
RottF amoA gene GGGGTTTCTACTGGTGGT PCRa-DGGE/qPCR [67]
RottR amoA gene CCCCTCKGSAAAGCCTTCTTC PCRa-DGGE/qPCR [67]
VectF pCR®2.1-TOPO plasmid AGTGTGCTGGAATTCGCC DGGE Marker [14]
VectR with GC pCR®2.1-TOPO plasmid ATATCTGCAGAATTCGCC DGGE Marker [14]
Eub341F 16S rRNA CCTACGGGAGGCAGCAG qPCR [55]
V3 region
Eub534R 16S rRNA ATTACCGCGGCTGCTGGC qPCR [55]
V3 region
GC Clamp Attach at 5’ end of primer CGCCCGCCGCGCCCCGCGCCCG
GCCCGCCGCCCCCGCCCC
PCR-DGGE [22]
a Used in multiplex PCR screening reactions without the GC clamp
Gels had a gradient of denaturant concentrations from 40 %
at the top of the gel to 70 % at the bottom, where 100 %
denaturant is defined as 7 M urea and 40 % formamide. Gels
also contained an 8 % to 12 % polyacrylamide gradient from
top to bottom [29]. Electrophoresis was at 60 V for 16 h.
Gels were stained with Sybr®Gold (www.invitrogen.com)
and documented using a FluorChem™ 8800 fluorescence
imager (www.alphainnotech.com). Three marker lanes (gener-
ated from five pooled unidentified clones according to the
method described by Burr et al. [14], data not shown) were
included in each DGGE gel so that cross-comparison between
gels would be possible. Bands in DGGE images were identified
using GelCompar II software (Version 6.1, AppliedMaths Inc.)
and confirmed visually.
Quantitative PCR
All PCR amplifications were performed in a Rotor-Gene
3000 real-time PCR cycler (QIAGEN, formerly Corbett Life
Science, www1.quiagen.com) in a 72-well rotor using the
programs described in Table 2: 16S rRNA gene program
was modified from Agrawal and Lal [1], dsrB from Geets et
al. [27] (see qPCR modifications in Table 2), amoA from
Geets et al. [28]. Data were acquired using the FAM/Sybr
detection channel during the extension step. For each sam-
ple, Ct values, the number of cycles required for the signal to
exceed background fluorescence, were converted to initial
DNA template concentrations using a standard curve. The
standard curve was a plot of Ct values vs. standard template
concentrations. Standards and samples were prepared in
triplicate and appropriate negative controls containing no
template DNA were included to ensure that no contamina-
tion had occurred. Melt curve analysis was also performed
after thermal cycling to verify PCR products. Melt curve
analysis was performed from 65 °C to 95 °C in 0.3 °C
increments held for 5 s with an initial pre-melt hold for
90 s at the first step.
The 16S rRNA gene was quantified and used to normal-
ize the amounts of dsrB and amoA genes in each sample
location. PCR primers Eub341F and Eub534R target the
variable V3 region of the 16S rDNA gene. Quantitative
PCR reactions (25 μl) were performed using Power SYBR®
Green PCR Master Mix (www.appliedbiosystems.com).
The PCR reaction mixture consisted of 12.5 μl Power
SYBR® Green PCR Master Mix, 0.5 μl ultrapure BSA
(50 mg/ml, Ambion), 2 μl nuclease-free water, 1 μl of
forward and reverse primers (12.5 μM), and 8 μl 1:10
diluted template DNA. These primers amplified a 193-bp
fragment of the 16S rDNA gene.
qPCR data were expressed in terms of relative gene
abundance by normalizing absolute copy numbers to copy
numbers of the 16S rRNA gene from the same sample. This
allowed for comparison among samples in which different
amounts of DNAwere extracted, and seemed to be the only
way to compare qPCR results for such different sample
materials as water, gravel, and root biomass. It also allowed
a straightforward assessment of the extent to which a sample
location was enriched for either functional gene compared to
other sample locations. A limitation of this approach is that
it does not determine the absolute contribution of SRB and
AOB associated with effluent vs. gravel vs. roots. It was not
possible to relate gene abundance to cell numbers of AOB
and SRB because the intracellular copy number for each of
the genes investigated is variable. The 16S rRNA gene can
range from 1 to 15 copies per cell [23, 28, 32], and the amoA
gene from 2 to 3 copies per cell [16, 28, 59, 88]. The dsrB
gene has been reported to have a single copy per cell [43,
48]. Samples below the level of detection were distorted
when the copies of 16S rRNA genes from that same location
were also low. This occurred with only one sample (from the
ultra-fine roots in a summer D. cespitosa microcosm). PCR
primers DsrBF and Dsr4R (Table 1) were used for amplifi-
cation of SRB and RottF and RottR were used for amplifi-
cation of AOB. Quantitative PCR reactions (25 μl) were
performed using Power SYBR® Green PCR Master Mix
(www.appliedbiosystems.com). The PCR reaction mixture
consisted of 12.5 μl Power SYBR® Green PCR Master Mix,
0.5μl ultrapure BSA (50mg/ml; Ambion, www.ambion.com),
2 μl nuclease-free water, 1 μl of forward and reverse primers
(12.5 μM), and 8 μl undiluted template DNA.
Table 2 PCR programs used in this study
Application PCR Program Reference
dsrB cloning, DGGE, qPCR Initial denaturation for 4 min at 94 °C, followed by 35 cycles of: 94 °C for 60 s, 55 °C
for 60 s, and 72 °C for 60 s. The program ended with an extension step at 72 °C for
10 min.qPCR program run for 40 cycles, denaturation temperature increased to 95 °C,
the initial denaturation step increased to 10 min.
[27]
Multiplex PCR, amoA and
DGGE Marker amplification
Initial denaturation for 60 s at 94 °C, followed by 35 cycles of: 94 °C for 60 s, 54 °C for
60 s, and 72 °C for 3 min. The program ended with an extension step at 72 °C for 10 min.
[5]
16S rRNA qPCR Initial denaturation for 10 min at 95 °C, followed by 45 cycles of: 95 °C for 10 s, 60 °C
for 15 s, and 72 °C for 20 s.
Modified from [1]
amoA qPCR Initial hold at 50 °C for 2 min and denaturation for 10 min at 95 °C, followed by 40
cycles of: 95 °C for 60 s, 50 °C for 60 s, and 72 °C for 60 s.
[28]
qPCR Standard Curve Preparation
Purified (16S rRNA gene, dsrB, amoA) PCR products
obtained from DNA extracted from CW samples were
cloned into plasmid vector pCR®2.1-TOPO® using the
TOPO® TA Cloning kit (Invitrogen, www.invitrogen.com)
following the manufacturer’s protocol. Clones were sent to the
Research Technology Support Facility (RTSF) at Michigan
State University and were sequenced using the M13F primer.
Edited sequences were compared with known sequences in
the GenBank database using the Basic Local Alignment Se-
quence Tool (BLAST) (http://blast.ncbi.nlm.nih.gov/
Blast.cgi). The 16S rRNA gene sequence was 99 % identical
to the 16S rRNA gene of an uncultured beta proteobacterium
clone (accession no. FJ535165). The dsrB sequence was 84%
identical to the dsrB gene of a Desulfatiferula sp. isolate
(accession no. HE613445). The amoA sequence was 99 %
identical to the amoA gene of an uncultured bacterium from an
environmental sample (accession no. JN177542).
Standard curves were obtained with serial dilution of the
quantified standard plasmids carrying the target 16S rRNA
gene, dsrB, or amoA gene. Standard curves were generated
for each gene, in triplicate, according to the protocol de-
scribed above. The copy number of the standard plasmids
carrying the targeted genes ranged from 1.09×107 to
1.09×102 copies/μl for the 16S rRNA gene, 3.12×107 to
3.12×102 copies/μl for dsrB, and 2.53×107 to 2.53×102
copies/μl for amoA.
Statistical Analyses
DGGE
DGGE gel images were processed and normalized using the
GelCompar II software (Version 6.1, Applied Maths Inc.).
Bands in DGGE images were identified on a presence/
absence basis. Band intensities were not considered during
statistical analysis. Subsequent statistical analyses of the
presence-absence data were performed using R software
(Version 2.11.1) libraries labdsv [65] and optpart [66]
(www.r-project.org). Dissimilarity matrices were calculated
using the Sorensen method [50, 87].
Hierarchical clustering (HC) analyses from the Sorensen
dissimilarity matrices were performed on DGGE band pat-
terns profiles for each of the functional genes, and dendro-
grams were generated using the Unweighted Pair Group
Method with Arithmetic Mean (UPGMA) [8, 37]. To opti-
mize the dendrograms to the most informative number of
clusters, a stride plot was generated which shows the global
(partana ratio) and local (silhouette width) values for the
cluster analysis. The partana (partition analysis) ratio eval-
uates within-cluster similarity with among-cluster similarity
and is a tool to measure the cluster validity [2], while the
silhouette width measures the mean similarity of each sam-
ple in the cluster to the mean similarity of the next most
similar cluster [68]. The optimized dendrogram was sliced
(according to the optimized partana ratio) and chi-square
tests performed on the sliced trees to determine impact of the
environmental parameters (season, plant type, sample type) on
the clusters (<0.05 indicated a significant relationship).
Separately, principal coordinate analysis (PCO) was ap-
plied to the DGGE dissimilarity matrix. General surface
plots (i.e., logistic regression with the first two PCOs as
predictors and one of the categorical environmental varia-
bles as the response) were generated to visualize, in two
dimensions, important partitions of the data with respect to
the environmental variables (e.g., summer vs. winter,
planted vs. unplanted) for each of the functional genes (R
software library labdsv [65]). The goodness of fit of the
surface plots was reported by D2, which is analogous to R2
(for quantitative responses). Large D2 values (maximum
was 1) indicated a good fit of the logistic regression of the
two most informative PCOs to the environmental partition-
ing of the data. The most influential environmental factor
(the one with the highest D2 value) was selected and the
dataset subdivided according to this parameter (e.g., sea-
son). Subsequent analysis of these two subgroup datasets
determined the next most influential variable (e.g., plant
species or sample type). Surface analysis is capable of
comparing two categories at a time; for those with variables
with three categories, one was compared to the remaining
two as was relevant to the CW system (e.g., unplanted
(control bands) vs. planted (D. cespitosa and L. cinereus
bands considered together)). This process was repeated until
all environmental variables had been assessed and sub-
groups created and analyzed, or surface analysis no longer
revealed associations between the PCO data and the envi-
ronmental parameter applied. Results from surface analyses
were compared with HC results to determine whether sim-
ilar conclusions could be drawn using a second independent
approach.
Quantitative PCR
Quantitative PCR results were analyzed using analysis of
variance (ANOVA) and multivariate ANOVA (MANOVA)
in Minitab (version 15). All data were normalized to 16S
rRNA gene quantity [1, 19, 28] and transformed to the log
scale to satisfy the normal assumption of the statistical
models. The Benjamini–Hochberg correction was used to
maintain a false discovery rate of either 5 % or 10 % [7].
Due to the limited number of data points, MANOVA anal-
ysis could not be performed on all six sample types simul-
taneously. Effluent (E1) was analyzed with an ANOVA. The
other five responses were separated into root (R2, R3, R4)
and gravel (G5, G6) subsets and these two groups were
analyzed separately. The factors in these models were Plant,
Season, and the two-way interaction. If there was no statis-
tically significant interaction, the main effects were inter-
preted directly. That is, an overall effect due to Plant Type
could be interpreted across all of the Seasons. Otherwise, a
“follow up” MANOVA was performed with a single factor
with levels across all of the combinations of Plant and
Season in order to examine the interactions between Plant
and Season. The linear discriminate functions (LDF), which
are the directions of maximum discriminability of the group
means produced by the MANOVA [64], were calculated,
rounded to an integer value, and interpreted with regard to
our CW system. The coefficients of each LDF were stan-
dardized by multiplying by the within group standard devi-
ation. By comparing the magnitude of the standardized
coefficients, we ranked the importance or contribution of
the original gravel and root measurements to the discrimi-
nability of the group means [64]. Statements of significance
are made based on Wilks’ likelihood ratio test.
Results
The combined approach of incorporating PMA, DGGE and
qPCR with functional gene primers allowed for a comprehen-
sive analysis of membrane-intact microorganisms that had
amoA and dsrB genes. PMA removed the DNA from cells
with compromised cell membranes (and extracellular DNA);
DGGE was a measure of the diversity of the community in
each sample, and qPCR quantified gene copy number. In this
way, we attempted to determine if diversity and/or abundance
of a particular gene within the membrane-intact microbial
community varied with time and sample location.
CW performance, as determined by effluent water qual-
ity, had been the focus of previous research [86] with COD
and sulfate removal closely examined. To ensure that the
CW microcosms were performing as anticipated, water
quality was evaluated prior to destructive sampling with
methods used in the previous study [86]. For water quality
data, there was good agreement between the previous and
current research. In the current research, there was a season-
al trend in COD removal for both the unplanted control and
L. cinereus, ranging from 70 % to 80 % in winter and up to
90 % in summer. D. cespitosa COD removal was typically
highest in winter (nearly 100 %) and slightly reduced in
summer (about 90 %). Sulfate data also correlated well with
previous research for all treatments, with nearly 100 %
sulfate removal in summer and winter for L. cinereus and
the unplanted control, and for D. cespitosa nearly 100 %
sulfate removal in the summer and 50 % removal in the
winter. Ammonia was most efficiently removed by the D.
cespitosa columns (near complete removal in both summer
and winter), followed by L. cinereus (75 % removal in
summer, 60 % in winter) and the unplanted control
(~50 % removal in summer and winter).
Quantitative PCR
The efficiency of qPCR was near 0.9 and the standard
curves for all genes were linear over six orders of magnitude
(R2>0.99). For each DNA extract, copy numbers of the 16S
rRNA gene and the two functional genes, dsrB and amoA,
were calculated from their respective qPCR standard curves.
The functional gene copy numbers were then normalized to
the 16S rRNA gene copy number as ratios (copies dsrB or
amoA/copies 16S rRNA gene) [1, 19, 28]. This calculation
expressed functional gene abundance relative to the total
bacterial population, and allowed comparisons between dif-
ferent samples in which DNA yield might have varied. For
samples below the level of detection, the qPCR threshold
value was substituted (678 copies/μl for dsrB and 267
copies/μl for amoA) since the statistical analyses required
a numerical value for each sample. These values were de-
termined by averaging the y-intercepts from the qPCR runs
performed for each gene and calculating the copy number
for the threshold Ct value.
Each sample was split between PMA treated and untreat-
ed fractions prior to DNA extraction. The effects of PMA
treatment were used to evaluate 16S rRNA gene copy num-
ber when compared to untreated samples. Ct values for the
16S rRNA gene treated with PMA indicated that removal of
DNA was no more than 10 % when compared to the un-
treated samples. Furthermore, the difference between PMA
treatment vs. no treatment was only significantly different
for the ultra-fine root samples and top gravel (ANOVA,
5 %). Therefore, all subsequent qPCR data for dsrB and
amoA are reported for the PMA treated samples only.
Dissimilatory Sulfite Reductase Gene (dsrB)
ANOVA was performed for each of the six individual loca-
tions to see how season and plant species affected relative
gene abundance. The ultra-fine roots and the top gravel layers
showed seasonal variation for SRB with significantly higher
gene ratios observed in the summer (10 % significance;
Tables 3, 4 and 5). Overall, D. cespitosa had the lowest
relative dsrB abundance (for all sample locations except the
ultra-fine roots; Tables 3–5). In previous studies, D. cespitosa
microcosms also typically had the highest winter redox po-
tential; in summer redox was comparable to L. cinereus and
the unplanted control [84, 85]. The unplanted control column
and L. cinereus had similar relative dsrB abundance for all
sample locations, with the effluent and bottom gravel samples
containing the highest relative abundance of dsrB. This corre-
lates well with the similarly efficient sulfate removal rates
observed for both of these treatments.
MANOVA was performed for a more comprehensive
examination of the qPCR data for environmental responses.
Effluent data (E1) was analyzed separately by standard
ANOVA with no significant results. There were significant
plant and season interactions within the root samples
(p<0.05). For dsrB, 87 % of the variability of the group
means (for the different plant and season combinations) was
discriminated by the equation: log(R2)− log(R3)+0.5 log
(R4). This discriminant function can be used to transform
the measurements on each sample that were actually mea-
sured, R2, R3 and R4, to a new variable which yields
maximal discriminability between the group means. Al-
though the coefficients for log(R2) and log(R3) have the
same magnitude, standardization of these coefficients show
that log(R3) is the single most important root measurement
to monitor in order to ascertain mean dsrB differences
amongst plants and seasons. L. cinereus had the highest
relative dsrB abundance on its roots in the summer season,
whereas D. cespitosa had minimal seasonal variation in the
relative dsrB abundance in root samples. Similar to the root
samples, dsrB showed a plant and seasonal interaction with-
in the gravel (p<0.05) and 86 % of the variability of the
group means was discriminated by the equation: log(G5)+
2log(G6). This relationship shows the bottom (anaerobic)
gravel to be more important in explaining plant and seasonal
mean differences in relative dsrB abundance, compared to
the top gravel (more aerobic). In addition, D. cespitosa
gravel was always observed to have lower relative dsrB
abundance in the summer season compared to the unplanted
control or L. cinereus gravel locations (Tables 3, 4 and 5).
Table 3 Average (n=2) relative
gene abundances for dsrB and
amoA compared to the16S
rRNA gene, separated by plant
species: unplanted control
E1 effluent, R2 ~1–2 mm diam-
eter roots from near the plant
crown, R3 ~0.5–1 mm diameter
roots near the root tips, R4<
~0.5 mm diameter ultrafine roots
detached during plant process-
ing, G5 gravel from column top,
G6 gravel from the column
bottom
Unplanted Control
Summer Winter
E1 G5 G6 E1 G5 G6
Copies/μl
AOB 6.6E+02 2.4E+04 8.4E+02 7.0E+02 1.3E+04 1.3E+03
SRB 1.8E+04 2.7E+03 1.7E+04 2.1E+03 4.4E+03 5.2E+04
16S 5.9E+06 5.4E+05 3.1E+06 1.0E+06 1.1E+07 1.0E+07
Relative abundance
AOB/16S (%) 0.011 4.560 0.027 0.067 0.118 0.013
AOB/16S (log10) −3.95 −1.34 −3.57 −3.17 −2.93 −3.89
SRB/16S (%) 0.311 0.495 0.541 0.203 0.041 0.502
SRB/16S (log10) −2.51 −2.31 −2.27 −2.69 −3.39 −2.30
Table 4 Average (n=2) relative gene abundances for dsrB and amoA compared to the16S rRNA gene, separated by plant species: D. cespitosa
D. cespitosa
Summer Winter
E1 R2 R3 R4 G5 G6 E1 R2 R3 R4 G5 G6
Copies/μl
AOB 5.3E+02 3.5E+02 3.8E+02 2.7E+02 3.5E+03 1.2E+03 2.7E+02 3.1E+02 2.7E+02 2.7E+02 2.7E+02 3.0E+03
SRB 1.4E+04 7.5E+02 2.2E+03 6.8E+02 2.8E+03 9.4E+03 6.8E+02 6.8E+02 8.1E+03 2.4E+03 8.5E+03 3.9E+04
16S 4.9E+06 2.9E+06 1.0E+06 9.6E+04 2.9E+06 1.0E+07 2.4E+06 1.3E+06 9.7E+06 4.7E+06 1.1E+07 8.9E+06
Relative abundance
AOB/16S (%) 0.011 0.012 0.037 0.279 0.119 0.011 0.011 0.024 0.003 0.006 0.002 0.034
AOB/16S
(log10)
−3.97 −3.92 −3.43 −2.55 −2.92 −3.94 −3.96 −3.63 −4.56 −4.24 −4.61 −3.47
SRB/16S (%) 0.278 0.026 0.221 0.708 0.096 0.093 0.028 0.051 0.083 0.051 0.077 0.433
SRB/16S
(log10)
−2.56 −3.59 −2.65 −2.15 −3.02 −3.03 −3.56 −3.29 −3.08 −3.29 −3.11 −2.36
E1 effluent, R2 ~1–2 mm diameter roots from near the plant crown, R3 ~0.5–1 mm diameter roots near the root tips, R4<~0.5 mm diameter
ultrafine roots detached during plant processing, G5 gravel from column top, G6 gravel from the column bottom
Ammonia Monooxygenase Gene (amoA)
The ultra-fine roots and the top gravel layers showed sea-
sonal variation for AOB, with significantly higher gene
ratios observed in the summer (5 % significance; Tables 3,
4 and 5). In addition, both planted species contained similar
relative amoA abundance for each sample location through-
out the columns. MANOVA indicated that for the amoA gene
there were significant plant and season interactions within the
root samples (p<0.05) with 86 % of the variability of the
group means for the different plant and season combinations
was discriminated by the equation: log(R2)−2 log(R3)+0.5
log(R4). For gravel samples, only a seasonal effect was ob-
served (no plant involvement) with respect to relative gene
quantity (p<0.10), with relatively higher amoA abundance
observed in the summer and 100 % of the variability of the
group means discriminated by the equation: log(G5). This
indicated that only the gravel nearest the surface of the col-
umns (and associated with the roots of the planted columns)
had an effect on the mean amoA abundance observed. It also
implied that the bottoms of the microcosms are relatively
unimportant with respect to nitrification, which was intuitive
as the bottoms of the columns are assumed to be the most
anaerobic regions in the microcosms (as evidenced by black-
ening and odor upon destructive sampling).
DGGE Analysis
For consistency in comparing the DGGE data with the
qPCR results, only those samples that were PMA treated
and above the level of detection for qPCR analysis were
included in DGGE analysis. For each of the functional
genes, HC analysis and PCO combined with general surface
plotting were performed on the DGGE community profiles
for each of the functional genes (the 16S rRNA gene was
not analyzed by DGGE). Figure 1 shows representative
DGGE profiles for dsrB. Profiles for both PMA-treated
and untreated samples are included to demonstrate that there
was no visible difference between them. DGGE profiles
appeared to indicate differences in the SRB community
profiles from planted and unplanted microcosms as well as
from summer to winter seasons among the planted micro-
cosms (Fig. 1). Statistical analyses were performed to deter-
mine if the apparent differences were significant. Even
though no visible differences were observed in the amoA
DGGE community profiles (data not shown), statistical
analyses were performed on these profiles also.
Richness of dsrB and amoA genes was measured as the
number of bands in the DGGE profiles. For dsrB, the most
interesting observation was that of 72 total bands detected
(all dsrB DGGE profiles combined), profiles from planted
microcosms contained 71of these, while profiles from
unplanted controls contained only 62 bands. The most inter-
esting result for amoA DGGE profiles was the comparison
between summer and winter samples. There were a total of 44
bands in winter profiles and 45 in summer profiles; however,
seven or eight bands were only detected in one of the seasons,
suggesting a slight shift in the AOB community.
HC analysis of the dsrB gene revealed differences be-
tween the community profiles of planted and unplanted
microcosms (p<0.005). It also indicated that sample type
(gravel, roots, effluent) within the column affected commu-
nity structure (p00.01). PCO surface analysis also detected
a difference between the planted and unplanted SRB
Table 5 Average (n=2) relative gene abundances for dsrB and amoA compared to the16S rRNA gene, separated by plant species: L. cinereus
L. cinereus
Summer Winter
E1 R2 R3 R4 G5 G6 E1 R2 R3 R4 G5 G6
Copies/μl
AOB 2.6E+03 4.5E+03 3.8E+03 1.7E+03 4.9E+03 3.8E+03 6.7E+02 2.3E+03 4.2E+03 2.7E+02 7.2E+03 8.0E+03
SRB 1.6E+04 9.0E+03 3.9E+03 1.1E+03 1.1E+04 3.3E+04 3.1E+04 1.9E+04 3.3E+04 3.2E+03 1.9E+04 2.7E+04
16S 1.1E+07 1.6E+06 1.5E+06 2.1E+05 2.2E+06 4.9E+06 5.8E+06 7.3E+06 4.4E+06 2.4E+07 2.3E+07 1.1E+07
Relative abundance
AOB/16S (%) 0.022 0.276 0.257 0.783 0.219 0.076 0.012 0.032 0.096 0.001 0.031 0.070
AOB/16S
(log10)
−3.65 −2.56 −2.59 −2.11 −2.66 −3.12 −3.94 −3.49 −3.02 −4.95 −3.50 −3.16
SRB/16S (%) 0.143 0.548 0.268 0.539 0.471 0.664 0.528 0.263 0.750 0.014 0.084 0.239
SRB/16S
(log10)
−2.85 −2.26 −2.57 −2.27 −2.33 −2.18 −2.28 −2.58 −2.12 −3.87 −3.07 −2.62
E1 effluent, R2 ~1–2 mm diameter roots from near the plant crown, R3 ~0.5–1 mm diameter roots near the root tips, R4<~0.5 mm diameter
ultrafine roots detached during plant processing, G5 gravel from column top, G6 gravel from the column bottom
communities (D200.5799; Fig. 2). The figure provides a
visual representation of the surface plot analysis performed
on the data, with the contour lines illustrating the separation
of the data in a third dimension (analogous to a topographic
map) according to the presence or absence of plants. When
analysis was limited to data only from planted microcosms,
there was a significant seasonal effect on SRB community
structure (p<0.05, as determined by chi-square analysis; D20
0.3852, as determined by surface analysis). No seasonal effect
was observed within the control column communities (p>
0.5). HC analysis of the planted columns also showed sample
type to be important in determining the community structure
(p<0.05), with the effluent samples being the most different
from any of the other locations within the microcosm.
Analysis of the amoA community profiles by HC indicated
both a seasonal effect on AOB community structure as well as
an effect due to sample type (effluent, roots, gravel; p<0.01).
However, PCO surface analysis revealed no remarkable
effects on the community structure by season, plant species,
etc. Although we only used the PCO surface analysis to
provide a visualization of possible effects found by HC, since
the two methods do not provide consistent results, we do not
emphasize the statistically significant differences in the amoA
community profiles found by HC in this manuscript.
Overall, HC showed sample type (effluent, roots, gravel)
to select for unique communities for both amoA and dsrB.
Season also affected microbial community composition for
both genes. For dsrB, only planted columns were seasonally
affected, whereas the amoA gene indicated a seasonal effect
for both planted and unplanted microcosms.
Discussion
The research presented here investigated the effect of two
independent variables, season and plants species selection
(including unplanted controls), on the diversity and abundance
of two functional genes, amoA (nitrification) and dsrB (sulfate
reduction). Both are good model genes because PCR and
qPCR methods are well established for them [27, 28, 67, 91].
They are also molecular markers for very important biological
transformations in wastewater treatment in CW. Diversity was
.
W
Lc
2R
R
A
+
.
S
D
c
4R
R
A
-
.
S
Lc
2R
R
A
+
..
WW
DcDc
3R4R
RR
AA
--
..
SS
DcDc
3R3R
RR
AA
+-
.
W
D
c
3R
R
A
+
.......
SSSSWWS
CCCCCCC
6G6G6G6G6G6G5G
GGGGGGG
AABBAAB
+-+-+-+
....
SSSS
LcDcLcDc
5G5G5G5G
GGGG
AABA
-+--
..
SS
LcLc
5G5G
GG
AB
++
Unplanted Planted
A BFigure 1 RepresentativeDGGE profiles of SRB
communities (dsrB gene). a
SRB community profiles
comparing gravel samples from
unplanted (left) and planted
(right) microcosms. b SRB
community profiles comparing
summer and winter seasons
from rhizosphere (R2, R3, R4)
biofilm samples. Images were
created using GelCompar II
software v. 6.1. S summer,
W winter, C unplanted control,
Dc D. cespitosa, Lc L. cinereus,
A/B replicate microcosms,
+/− PMA treatment
Figure 2 Analysis of SRB biofilm community profiles from planted
and unplanted microcosms post PMA treatment. The contours indicate
values between 0 (planted) and 1 (unplanted) and offer a visualization
of the separation of DGGE results from unplanted and plantedmicrocosm
communities. Pound symbol unplanted control, asterisk L. cinereus,
ampersand D. cespitosa
measured as presence/absence of bands in DGGE profiles as
used in the study by Calheiros et al. [15]. Band intensity was
not evaluated since it has been considered a poor indicator of
the abundance of the corresponding species [25, 54]. Total
gene abundance was then quantified by qPCR.
A secondary objective was to identify sampling locations
that were (1) selectively enriched with either gene (greater
relative gene abundance or greater diversity) and/or (2) more
responsive to the independent variables relative to other sam-
pling locations. Although planktonic bacteria are not thought
to be primarily responsible for nutrient removal in CW, inves-
tigations typically focus on these easily obtained samples. To
ensure that the functional diversity, gene abundance and met-
abolic activity are more realistically assessed, the communi-
ties partitioned into microhabitats in biofilms on gravel and
varied root surfaces were investigated. It was hypothesized
that these samples might be more responsive to the variables
of season and species than bulk water and therefore more
sensitive or ecologically relevant. However, even intricate
sampling such as that performed in this research cannot sep-
arate very different microhabitats that might exist within
micrometers to millimeters of each other.
Other research groups have performed similar intensive
sampling in full scale CWand/or microcosms. Gagnon et al.
[26] combined sand and root samples for analysis of bacte-
rial density, respiration, and enzyme activity. Wang et al.
[92] sampled complete below ground biomass, but for mi-
crobial enzyme assays, not for DNA-based analysis. Iasur-
Kruh et al. [36] separated roots from gravel in a microcosm
study, but chose to analyze only the gravel samples for
DGGE analysis of the bacterial rRNA gene. Sims et al.
[74] sampled water and wetlands soils in their study of
seasonal effects on AOB. However, we are not aware of
any research in which there was an attempt to determine the
relative importance of different sampling locations (includ-
ing three different root classes) in explaining variations in
the data among samples. For example, our statistical analy-
ses indicated that root sample R3 was the most important in
explaining seasonal and plant species effects on dsrB abun-
dance. Although we have only begun to investigate how a
CW should be sampled, we have presented a method and
statistical rationale for making that determination.
PMA treatment was integrated into the sampling proto-
col to target membrane-intact cells within the CW micro-
cosms. The conclusion that the majority of bacterial cells
in a community have intact cell membranes does not
indicate the extent to which these cells are currently
metabolically active or replicating but does focus attention
on potentially active cells. Because PMA treatment re-
duced the copy number of 16S rRNA genes by only about
10 %, subsequent data analyses were performed only on
PMA treated sample results. Furthermore, ANOVA of the
PMA untreated functional genes normalized to PMA
untreated 16S rRNA genes produced results that were
very similar to those for PMA treated samples. There
are two possible interpretations of these results: (1) the
bacterial communities consisted almost entirely (~90 %)
of cells with intact membranes, or (2) that PMA treatment
was not completely effective at inhibiting PCR amplifica-
tion of DNA from cells with compromised membranes
and/or extracellular DNA. However, since the PMA meth-
od [57, 58] convincingly demonstrated the effectiveness of
PMA in excluding from PCR amplification the DNA from
cells killed by various treatments (e.g., heat, isopropanol),
the conclusion that the bacterial communities in these
microcosms consisted primarily of cells with intact mem-
branes is probably justified. This result is similar to that of
Varma et al. [89], who found little difference in qPCR
copy numbers between PMA-treated and -untreated sam-
ples from wastewater in the absence of any biocidal
treatment. However, when bacteria were heat killed, there
was at least a three-log reduction in the qPCR signal in
PMA-treated compared to untreated samples. Another pos-
sible effect on PMA efficacy is amplicon length. Luo et
al. [49] reported that PMA treatment could fail to suppress
PCR amplification from membrane-compromised cells
when the amplicons were shorter than about 190 bp. In
the current study, the amplicon lengths were 193, 370, and
491 bp for 16S rRNA, dsrB and amoA, respectively.
Using the approaches and rationales described above, the
microbial ecology of the wetlands with respect to sulfate re-
duction and ammonia oxidation were explored. The ratios of
the two functional genes to 16S rRNA gene abundance in the
CW samples were closely related to the ranges previously
reported in the literature from locations such as rice roots
[70], marine sediments [19], sludge [28], and oil field produc-
tion water [1]. Average dsrB/16S rRNA genes ranged from
0.013 % to 0.750 % and amoA/16S rRNA genes from 0.002 %
to 4.6 %. Other researchers have shown similar relative quan-
tities of SRB in the environment, generally not representing
more than 5 % of the total microbial community present [20,
70] (both compared SRB rRNA with total rRNA). Most re-
cently, Dang et al. [19] reported the ratio of amoA/16S rRNA
genes in marine sediments to be between 0.003 % and 0.07 %.
For SRB, the attached roots (R2 and R3) were most
important in explaining relative abundance, but the ultra-
fine roots (R4) could not be completely discounted without
altering the results. In general, all the root locations were
equally important, but for gravel, the bottom gravel loca-
tions were most important. SRB appeared to be a more
robust community, compared to AOB, as few samples were
below detectable levels. SRB were found in all of the
sampled microcosm locations, supporting previous evidence
that these organisms are capable of existing in both anaero-
bic and aerobic environments [12, 18, 24]. H2S was not
measured, but the odor of H2S was detected during sampling
and there was blackening of the gravel (especially the gravel
from the bottom of the column (G6)), indicative of SRB activity.
Increased SRB presence in the unplanted control and L.
cinereus indicated that these treatment conditions were more
appropriate and less variable for treating wastewater with high
levels of sulfate. Sulfate, carbon, or redox could limit sulfate
reduction. In this research, adequate sulfate and carbon were
supplied in the feed water, leaving redox as the variable most
likely to limit SRB activity and also the most site specific of
these variables. D. cespitosa had the lowest quantity of SRB
of all the CW microcosms tested (for all sample locations
except the ultra-fine roots (R4)), which correlated well with
the observed redox and sulfate data [86]. Higher quantities of
SRB within the bottom gravel (compared to top gravel) also
correlated well with the expected anaerobic conditions in
these locations, which are likely to be enhanced in the
unplanted or L. cinereus conditions (lowest redox values).
Conversely, another study (in a wetland lake system) reported
the increased presence of SRB (MPN methodology) in the
rhizosphere compared to surrounding sediments [90].
Overall, D. cespitosa had the lowest relative SRB quantity
with very little seasonal variation. This implies a stable SRB
community; however, it had been hypothesized that an in-
crease in the SRB population in summer would correspond to
increased sulfate removal. It is possible that the increased
redox within the D. cespitosa microcosms in winter is high
enough to inhibit sulfate reduction but not high enough to
affect the abundance of the SRB community present.
In addition to gene abundance, it is important to consider
the potential diversity of organisms carrying the dsrB gene. A
difference in the SRB community DGGE profile was ob-
served depending on plant presence or absence, with
unplanted control columns having unique SRB communities.
This was also observed in other wetland studies [90]. Since
water quality and qPCR data consistently matched for both
unplanted microcosms and L. cinereus planted microcosms, a
difference due to plant species was expected to separate D.
cespitosa from the other treatments. This result may suggest
that the rhizosphere of L. cinereus was leaking oxygen. Since
L. cinereus had such shallow roots, the oxygen released may
not have been sufficient to impact sulfate removal to the extent
observed for D. cespitosa. This could indicate that the SRB
community structure was dependent on plant presence or
absence and that sulfate removal efficiency did not predict
the differences in SRB community profiles. Statistical analy-
ses of qPCR showed dsrB present in higher ratios for L.
cinereus roots than for D. cespitosa roots; however DGGE
showed similar community profiles for the roots of both of
these plants. This shows that variation in relative gene abun-
dance does not necessarily indicate a difference in community
structure, but rather population density. In general, the DGGE
community structure for SRB appeared to be most greatly
influenced by the presence or absence of a plant while the
gene quantities appeared to be most greatly influenced by a
combination of plant species and seasonal variation.
For AOB, the attached roots were apparently the most
ecologically relevant samples. The fine roots (R3) were
more important for AOB compared to SRB in describing
the variability of the means. This implicates the “fine root”
region as a site of preferential colonization by AOB and an
important sampling site. The ultra-fine roots (R4) were
equally important for AOB and SRB and could not be
removed from analysis without affecting the outcome. It
was difficult to characterize the AOB community as thor-
oughly as the SRB because many of the samples were below
the level of detection by qPCR. AOB were most frequently
detected in the summer season and generally attached to root
surfaces. The top gravel was also important for explaining
AOB abundance (by both ANOVA and MANOVA), where-
as the bottom gravel was not important. These results sug-
gest top gravel as another targeted location for future AOB
investigation. Although not significant for the MANOVA
results, AOB were detected in the bottom gravel of all the
CW microcosms, supporting other research indicating their
ability to survive under anaerobic conditions [26]. It has also
been reported that AOB and anammox organisms can coex-
ist [63], although screening (by conventional PCR) for
anammox organisms in our systems did not yield any pos-
itive results (data not shown).
There was a similarity between the relative gene abundan-
ces of amoA inD. cespitosa and L. cinereusmicrocosms. This
was surprising as the overall redox and performance for D.
cespitosa was much greater than L. cinereus with respect to
ammonia removal. L. cinereus roots did not penetrate deeply
into the gravel substrate, potentially causing any oxygen re-
leased to be concentrated within the uppermost portion of the
column and supporting a higher proportion of AOB. Addi-
tionally, this may be simply explained by many samples being
below the level of detection and normalized to similar values.
Although strong seasonal effects were observed for AOB
community diversity with respect to DGGE, the quantity of
the amoA gene was generally below the level of detection
for most winter samples. This was not surprising given that
the literature has often reported a strong temperature effect
on nitrification [33, 47, 60, 77]. Although decreased quan-
tities of amoAwere observed in these winter samples, water
quality analyses indicated that nitrification was still occur-
ring (as measured by loss of ammonia). Ammonia loss could
have been due to activity of archaeal ammonia oxidizers
(AOA) or anammox organisms; however, screening of sev-
eral samples for the genes of these two groups of organisms,
by conventional PCR, did not detect any (data not shown).
Another CW research group also found a seasonal shift in
AOB DGGE profiles, but was investigating changes from
autumn to spring [95]. Sims et al. [74] generally found AOA
in equal or greater numbers compared to AOB (qPCR data)
in wetland soil and water samples. AOB were also more
sensitive to winter than AOA.
The majority of CW research has focused on the micro-
bial population associated with the gravel and effluent water
and neglects those associated with the roots. The root surface
is very dynamic and heterogeneous, making it an ideal surface
for colonization by varying populations of microorganisms
such as SRB [90] and AOB [10, 11, 71]. It was unexpected
that the ultra-fine roots (R4) ofD. cespitosa in summer would
simultaneously support a relatively large SRB and AOB com-
munity, as reported by ANOVA. As mentioned previously,
correction for samples below the level of detection resulted in
this location being misrepresented as artificially high in indi-
vidual summer samples for both genes because the
corresponding copies of 16S rRNA gene were also low.
Season also influenced the microbial community structure
with effects on the abundance of both the dsrB and amoA
genes. Our results for the root samples differed for the two
genes. It is possible that each plant species affected the mi-
crobial community abundance by uniquely altering the redox
environment within the microcosm.
Although there were differences in gene abundance, and
diversity, both AOB and SRBwere present at the same sample
locations, confirming that both aerobic and anaerobic organ-
isms can exist in close proximity in CW biofilms [80]. This
could be due to protected niches in the environment provided
by plant species. For example, one study found that high
efficiency nitrification and sulfate reduction was possible in
CW microcosms planted with Juncus effusus. It was postulat-
ed that plant presence, and thus ROL, limited sulfide accumu-
lation and toxicity within the microcosms [94]. Our sampling
methods were not capable of separating aerobic from anaero-
bic regions, since oxygen gradients, especially at root surfa-
ces, are probably on a scale of less than one mm. Because of
the predominant anaerobic condition in these columns, there is
the potential for denitrification. Although the denitrifying
populations were not directly assessed in this study, nitrite
and nitrate were monitored in the effluent. The data suggested
that denitrification was occurring in the summer in all col-
umns, while nitrite accumulated during the winter in the
control and L. cinereus columns. The increased redox in the
D. cespitosa system in winter may have contributed to this
phenomenon by limiting the competing SRBs.
The results of this study begin to illustrate the interactions
occurring within CW and the methods used herein can be
applied to a variety of environmental systems and genes for
a more in depth understanding of microbial processes. Quan-
tification of controllable parameters and their influences on
the microbial ecology will lead to better design and operation
of CW systems. Optimization of the microbial community
structure and function should be a priority for the effective
design of wastewater treatment systems [28, 81]. The research
reported here was focused on a fundamental understanding of
wetland ecology at the microbial level and the influence of
that ecology on critical chemical cycles.
Acknowledgements This study was supported by: USDA-NRI
Competitive Grants Program, Award 2004-35102-14832, the Montana
Board of Research and Commercialization Technology (MBRCT),
Grant Agreement #09-26, and the Society of Wetland Scientists
through the Student Research Grant Program. We would like to thank
Bennett Hisey and Rachel Van Kempen-Fryling for their assistance in
preparation and production of DGGE gels for this project.
References
1. Agrawal A, Lal B (2009) Rapid detection and quantification of
bisulfite reductase genes in oil field samples using real-time PCR.
FEMS Microbiol Ecol 69:301–312
2. Aho K, Roberts DW, Weaver T (2008) Using geometric and non-
geometric internal evaluators to compare eight vegetation classification
methods. J Veg Sci 19:549–562
3. Allen WC, Hook PB, Biederman JA, Stein OR (2002) Temperature
and wetland plant species effects on wastewater treatment and root
zone oxidation. J Enviro Qual 31:1010–1016
4. Amann RI, Ludwig W, Schleifer KH (1995) Phylogenetic
identification and in-situ detection of individual microbial-cells
without cultivation. Microbiol Rev 59:143–169
5. Bahr M, Crump BC, Klepac-Ceraj V, Teske A, Sogin ML, Hobbie
JE (2005) Molecular characterization of sulfate-reducing bacteria
in a New England salt marsh. Environ Microbiol 7:1175–1185
6. Bais HP, Weir TL, Perry LG, Gilroy S, Vivanco JM (2006) The
role of root exudates in rhizosphere interations with plants and
other organisms. Ann Rev Plant Biol 57:233–266
7. Benjamini Y, Hochberg Y (1995) Controlling the false discovery
rate — a practical and powerful approach to multiple testing. J R
Stat Soc Ser B-Methodol 57:289–300
8. Boon N, De Windt W, Verstraete W, Top EM (2002) Evaluation of
nested PCR-DGGE (denaturing gradient gel electrophoresis) with
group-specific 16S rRNA primers for the analysis of bacterial
communities from different wastewater treatment plants. FEMS
Microbiol Ecol 39:101–112
9. Borden D, Stein O, Hook P (2001) Seasonal effects of supplemental
organic carbon on sulfate reduction and zinc sulfide precipitation in
constructed wetland microcosms. Intern Ecol Eng Soc Meet Lincoln
University, New Zealand, In, pp 296–300
10. Brix H (1994) Functions of macrophytes in constructed wetlands.
Water Sci Technol 29:71–78
11. Brix H (1997) Do macrophytes play a role in constructed treatment
wetlands? Water Sci Technol 35:11–17
12. Brune A, Frenzel P, Cypionka H (2000) Life at the oxic–anoxic
interface: microbial activities and adaptations. FEMS Microbiol
Rev 24:691–710
13. Burgmann H, Meier S, Bunge M, Widmer F, Zeyer J (2005)
Effects of model root exudates on structure and activity of a soil
diazotroph community. Environ Microbiol 7:1711–1724
14. Burr MD, Clark SJ, Spear CR, Camper AK (2006) Denaturing
gradient gel electrophoresis can rapidly display the bacterial diversity
contained in 16S rDNA clone libraries. Microb Ecol 51:479–486
15. Calheiros CSC, Teixeira A, Pires C, Franco AF, Duque AF, Crsipim
LFC,Moura SC, Castro PML (2010) Bacterial dynamics in horizontal
flow constructed wetlands with different plants for high salinity
industrial wastewater polishing. Water Res 44:5032–5038
16. Chain P, Lamerdin J, Larimer F, Regala W, Lao V, Land M et al
(2003) Complete genome sequence of the ammonia-oxidizing
bacterium and obligate chemolithoautotroph Nitrosomonas europaea.
J Bacteriol 185:2759–2773
17. Cooper P, Griffin P (1999) A review of the design and performance
of vertical-flow and hybrid reed bed treatment systems. Water Sci
Technol 40:1–9
18. Cypionka H (2000) Oxygen respiration by Desulfovibrio species.
Annu Rev Microbiol 54:827–848
19. Dang HY, Li J, Chen RP, Wang L, Guo LZ, Zhang ZN, Klotz MG
(2010) Diversity, abundance, and spatial distribution of sediment
ammonia-oxidizing betaproteobacteria in response to environmental
gradients and coastal eutrophication in Jiaozhou Bay, China. Appl
Environ Microbiol 76:4691–4702
20. Devereux R, Hines ME, Stahl DA (1996) S cycling: characterization
of natural communities of sulfate-reducing bacteria by 16S rRNA
sequence comparisons. Microb Ecol 32:283–292
21. Faulwetter JL, Gagnon V, Sundberg C, Chazarenc F, Burr MD,
Brisson J et al (2009) Microbial processes influencing performance
of treatment wetlands: a review. Ecol Eng 35:987–1004
22. Ferris MJ, Muyzer G, Ward DM (1996) Denaturing gradient gel
electrophoresis profiles of 16S rRNA-defined populations inhabiting
a hot spring microbial mat community. Appl Environ Microbiol
62:340–346
23. Fogel G, Collins C, Li J, Brunk C (1999) Prokaryotic genome size
and SSU rDNA copy number: estimation of microbial relative
abundance from a mixed population. Microb Ecol 38:93–113
24. Fortin D, Goulet R, Roy M (2000) Seasonal cycling of Fe and S in
a constructed wetland: the role of sulfate-reducing bacteria.
Geomicrobiol J 17:221–235
25. Fromin N, Hamelin J, Tarnawski S, Roesti D, Jourdain-Miserez K,
Forestier N et al (2002) Statistical analysis of denaturing gel
electrophoresis (DGE) fingerprinting patterns. Environ Microbiol
4:634–643
26. Gagnon V, Chazarenc F, Comeau Y, Brisson J (2007) Influence of
macrophyte species on microbial density and activity in constructed
wetlands. Water Sci Technol 56:249–254
27. Geets J, Borrernans B, Diels L, Springael D, Vangronsveld J, van
der Lelie D, Vanbroekhoven K (2006) DsrB gene-based DGGE for
community and diversity surveys of sulfate-reducing bacteria. J
Microbiol Methods 66:194–205
28. Geets J, de Cooman M, Wittebolle L, Heylen K, Vanparys B, De
Vos P et al (2007) Real-time PCR assay for the simultaneous
quantification of nitrifying and denitrifying bacteria in activated
sludge. Appl Microbiol Biotechnol 75:211–221
29. Girvan MS, Bullimore J, Pretty JN, Osborn AM, Ball AS (2003)
Soil type is the primary determinant of the composition of the total
and active bacterial communities in arable soils. Appl Environ
Microbiol 69:1800–1809
30. Gregori G, Citterio S, Ghiani A, Labra M, Sgorbati S, Brown S,
Denis M (2001) Resolution of viable and membrane-compromised
bacteria in freshwater and marine waters based on analytical flow
cytometry and nucleic acid double staining. Appl Environ Microbiol
67:4662–4670
31. Hall-Stoodley L, Costerton JW, Stoodley P (2004) Bacterial biofilms:
from the natural environment to infectious diseases. Nat RevMicrobiol
2:95–108
32. Hallin S, Jones CM, Schloter M, Philippot L (2009) Relationship
between N-cycling communities and ecosystem functioning in a
50-year-old fertilization experiment. ISME J 3:597–605
33. Hammer M, Hammer MJ (2001) Water and wastewater technology.
Prentice Hall, Upper Saddle River, NJ, USA
34. Hatano K, Frederick DJ, Moore JA (1994) Microbial ecology of
constructed wetlands used for treating pulp-mill waste-water. Water
Sci Technol 29:233–239
35. Hsu S, Maynard J (1999) The use of sulfur isotopes to monitor the
effectiveness of constructed wetlands in controlling acid mine
drainage. Environ Eng Policy 1:223–233
36. Iasur-Kruh L, Hadar Y, Milstein D, Gasith A, Minz D (2010)
Microbial population and activity in wetland microcosms con-
structed for improving treated municipal wastewater. Microb Ecol
59:700–709
37. Ibekwe AM, Papiernik SK, Gan J, Yates SR, Yang CH, Crowley
DE (2001) Impact of fumigants on soil microbial communities.
Appl Environ Microbiol 67:3245–3257
38. Ibekwe AM, Grieve CM, Lyon SR (2003) Characterization of
microbial communities and composition in constructed dairy wetland
wastewater effluent. Appl Environ Microbiol 69:5060–5069
39. Johnson K, Martin C, Moshiri G, McCrory W (1999) Performance
of a constructed wetland leachate treatment system at the Chunchula
landfill, Mobile County, Alabama. In: Mulamoottil F, McBean E,
Rovers F (eds) Constructed wetlands for the treatment of landfill
leachates. CRC Press, Boca Raton, FL, USA, pp 57–70
40. Josephson KL, Gerba CP, Pepper IL (1993) Polymerase chain
reaction detection of non-viable bacterial pathogens. Appl Environ
Microbiol 59:3513–3515
41. KalinM, FysonA,WheelerWN (2006) The chemistry of conventional
and alternative treatment systems for the neutralization of acid mine
drainage. Sci Total Environ 366:395–408
42. Kirk JL, Beaudette LA, Hart M, Moutoglis P, Khironomos JN, Lee
H, Trevors JT (2004) Methods of studying soil microbial diversity.
J Microbiol Methods 58:169–188
43. Klein M, Friedrich M, Roger A, Hugenholtz P, Fishbain S, Abicht
H et al (2001) Multiple lateral transfers of dissimilatory sulfite
reductase genes between major lineages of sulfate-reducing
prokaryotes. J Bacteriol 183:6028–6035
44. Konopka A (2009) What is microbial community ecology? ISME J
3:1223–1230
45. Kowalchuk GA, Stienstra AW, Heilig GHJ, Stephen JR,Woldendorp
JW (2000) Molecular analysis of ammonia-oxidising bacteria in soil
of successional grasslands of the Drentsche A (The Netherlands).
FEMS Microbiol Ecol 31:207–215
46. Kowalchuk GA, Buma DS, de Boer W, Klinkhamer PGL, van
Veen JA (2002) Effects of above-ground plant species composition
and diversity on the diversity of soil-borne microorganisms. Antonie
Van Leeuwenhoek Intern J Gen Mol Microbiol 81:509–520
47. Kuschk P, Wiessner A, Kappelmeyer U, Weissbrodt E, Kastner M,
Stottmeister U (2003) Annual cycle of nitrogen removal by a pilot-
scale subsurface horizontal flow in a constructed wetland under
moderate climate. Water Res 37:4236–4242
48. Leloup J, Fossing H, Kohls K, Holmkvist L, Borowski C, Jorgensen
BB (2009) Sulfate-reducing bacteria in marine sediment (Aarhus
Bay, Denmark): abundance and diversity related to geochemical
zonation. Environ Microbiol 11:1278–1291
49. Luo J-F, Lin W-T, Guo Y (2010) Method to detect only viable cells
in microbial ecology. Appl Microbiol Biotechnol 86:377–384
50. McCune B, Grace JB (2002) Analysis of ecological communities.
MjM Software Design, Gleneden Beach, OR, USA
51. Minz D, Flax J, Green S, Muyzer G, Cohen Y, Wagner M et al
(1999) Diversity of sulfate-reducing bacteria in oxic and anoxic
regions of a microbial mat characterized by comparative analysis
of dissimilatory sulfite reductase genes. Appl Environ Microbiol
65:4666–4671
52. Molle P, Prost-Boucle S, Lienard A (2008) Potential for total nitrogen
removal by combining vertical flow and horizontal flow constructed
wetlands: a full scale experiment study. Ecol Eng 34:23–29
53. Munch C, Neu T, Kuschk P, Roske I (2007) Ecological considerations
—the root surface as the definitive detail for microbial transformation
processes in constructed wetlands—a biofilm characteristic. Water Sci
Technol 56:271–276
54. Murray AE, Hollibaugh JT, Orrego C (1996) Phylogenetic compo-
sitions of bacterioplankton from two California estuaries compared
by denaturing gradient gel electrophoresis of 16S rDNA fragments.
Appl Environ Microbiol 62:2676–2680
55. Muyzer G, Dewaal EC, Uitterlinden AG (1993) Profiling of
complex microbial populations by denaturing gradient gel elec-
trophoresis analysis of polymerase chain reaction-amplified
genes coding for 16S ribosomal-RNA. Appl Environ Microbiol
59:695–700
56. Nocker A, Camper AK (2009) Novel approaches toward preferential
detection of viable cells using nucleic acid amplification techniques.
FEMS Microbiol Lett 291:137–142
57. Nocker A, Cheung CY, Camper AK (2006) Comparison of propidium
monoazidewith ethidiummonoazide for differentiation of live vs. dead
bacteria by selective removal of DNA from dead cells. J Microbiol
Methods 67:310–320
58. Nocker A, Sossa-Fernandez P, Burr MD, Camper AK (2007) Use
of propidium monoazide for live/dead distinction in microbial
ecology. Appl Environ Microbiol 73:5111–5117
59. Okano Y, Hristova KR, Leutenegger CM, Jackson LE, Deni-
son RF, Gebreyesus B et al (2004) Application of real-time
PCR to study effects of ammonium on population size of
ammonia-oxidizing bacteria in soil. Appl Environ Microbiol
70:1008–1016
60. Painter H (1986) Nitrification in the treatment of sewage and
waste-waters. In: Prosser J (ed) Nitrification. Information Printing,
Oxford, UK, pp 185–211
61. Paredes D, Kuschk P, Koser H (2007) Influence of plants and
organic matter on the nitrogen removal in laboratory-scale model
subsurface flow constructed wetlands inoculated with anaerobic
ammonium oxidizing bacteria. Eng Life Sci 7:565–576
62. Park H-D, Wells GF, Bae H, Criddle CS, Francis CA (2006)
Occurrence of ammonia-oxidizing archaea in wastewater treatment
plant bioreactors. Appl Environ Microbiol 72:5643–5647
63. Pathak B, Kazama F, Tanaka Y, Mori K, Sumino T (2007)
Quantification of anammox populations enriched in an immo-
bilized microbial consortium with low levels of ammonium
nitrogen and at low temperature. Appl Microbiol Biotechnol
76:1173–1179
64. Rencher AC (1998) Multivariate statistical inference and applications.
Wiley
65. Roberts DW (2009) labdsv: Ordination and multivariate analysis for
ecology. In: R package version 1.3-3, http://ecology.msu.montana.edu/
labdsv/R
66. Roberts DW (2010) optpart: Optimal partitioning of similarity rela-
tions. In: R package version 2.0-1, http://CRAN.R-project.org/
package00optpart
67. Rotthauwe JH, Witzel KP, Liesack W (1997) The ammonia mono-
oxygenase structural gene amoA as a functional marker: molecular
fine-scale analysis of natural ammonia-oxidizing populations.
Appl Environ Microbiol 63:4704–4712
68. Rousseeuw PJ (1987) Silhouettes — a graphical aid to the inter-
pretation and validation of cluster-analysis. J Comput Appl Math
20:53–65
69. Saleh-Lakha S, Miller M, Campbell RG, Schneider K, Elahimanesh
P, Hart MM, Trevors JT (2005) Microbial gene expression in soil:
methods, applications and challenges. J Microbiol Methods 63:1–19
70. Scheid D, Stubner S (2001) Structure and diversity of Gram-
negative sulfate-reducing bacteria on rice roots. FEMS Microbiol
Ecol 36:175–183
71. Scholz M, Lee B (2005) Constructed wetlands: a review. Intern J
Environ Stud 62:421–447
72. Sigalevich P, Baev MV, Teske A, Cohen Y (2000) Sulfate reduction
and possible aerobic metabolism of the sulfate-reducing bacterium
Desulfovibrio oxyclinae in a chemostat coculture withMarinobacter
sp strain MB under exposure to increasing oxygen concentrations.
Appl Environ Microbiol 66:5013–5018
73. Sims A, Gajaraj S, Hu Z (2012) Seasonal population changes of
ammonia-oxidizing organisms and their relationship to water quality
in a constructed wetland. Ecol Eng 40:100–107
74. Sims A, Horton J, Gajaraj S, McIntosh S, Miles RJ, Mueller R,
Reed R, Hu Z (2012) Temporal and spatial distributions of
ammonia-oxidizing archaea and bacteria and their ratio as an
indicator of oligotrophic conditions in natural wetlands. Water
Res 46:4121–4129
75. Smalla K, Wieland G, Buchner A, Zock A, Parzy J, Kaiser S et al
(2001) Bulk and rhizosphere soil bacterial communities studied by
denaturing gradient gel electrophoresis: plant-dependent enrich-
ment and seasonal shifts revealed. Appl Environ Microbiol
67:4742–4751
76. Song Y, Fitch M, Burken J, Nass L, Chilukiri S, Gale N, Ross C
(2001) Lead and zinc removal by laboratory-scale constructed
wetlands. Water Environ Res 73:37–44
77. Song ZW, Zheng ZP, Li J, Sun XF, Han XY, Wang W, Xu M (2006)
Seasonal and annual performance of a full-scale constructed wetland
system for sewage treatment in China. Ecol Eng 26:272–282
78. Stein OR, Hook PB (2005) Temperature, plants, and oxygen: how
does season affect constructed wetland performance? J Environ Sci
Health Part A Toxic/Hazard Subst Environ Eng 40:1331–1342
79. Stein OR, Borden-Stewart DJ, Hook PB, Jones WL (2007) Seasonal
influence on sulfate reduction and zinc sequestration in subsurface
treatment wetlands. Water Res 41:3440–3448
80. Stewart P, Franklin M (2008) Physiological heterogeneity in
biofilms. Nat Rev Microbiol 6:199–210
81. Stottmeister U, Wieβner KP, Kappelmeyer U, Kastner BO, Muller
RA, Moormann H (2003) Effects of plants and microorganisms in
constructed wetlands for wastewater treatment. Biotechnol Adv
22:93–117
82. Sundberg C, Stendahl JSK, Tonderski K, Lindgren PE (2007) Over-
land flow systems for treatment of landfill leachates — potential
nitrification and structure of the ammonia-oxidising bacterial commu-
nity during a growing season. Soil Biol Biochem 39:127–138
83. Sundblad K (1998) Sweden. In: Vymazal J, Brix H, Cooper PF,
Green MB, Haberl R (eds) Constructed wetlands for wastewater
treatment in Europe. Backhuys Publishers, Leiden, The Netherlands,
pp 251–259
84. Taylor CR, Stein OR, Hook PB, Zabinski CA (2008) In: Comparing
19 plant species’ seasonal effects on COD removal in model treatment
wetlands. 11th Intern Conf Wetl Syst Water Pollut Control. Indore,
India, pp 132–138
85. Taylor CR (2009) Selecting plant species to optimize wastewater
treatment in constructed wetlands. MSc thesis, Mont State Univ
86. Taylor C, Hook P, Stein O, Zabinski C (2011) Seasonal effects of
19 plant species on COD removal in subsurface treatment wetland
microcosms. Ecol Eng 37:703–710
87. Van der Gucht K, Sabbe K, De Meester L, Vloemans N, Zwart G,
Gillis M, Vyverman W (2001) Contrasting bacterioplankton com-
munity composition and seasonal dynamics in two neighbouring
hypertrophic freshwater lakes. Environ Microbiol 3:680–690
88. van der Wielen PWJJ, Voost S, van der Kooij D (2009) Ammonia-
oxidizing bacteria and archaea in groundwater treatment and drink-
ing water distribution systems. Appl Environ Microbiol 75:4687–
4695
89. Varma M, Field R, Stinson M, Rukovets B, Wymer L, Haugland R
(2009) Quantitative real-time PCR analysis of total and propidium
monoazide-resistant fecal indicator bacteria in wastewater. Water
Res 43:4790–4801
90. Vladar P, Rusznyak A, Marialigeti K, Borsodi AK (2008) Diversity
of sulfate-reducing bacteria inhabiting the rhizosphere of Phragmites
australis in Lake Velencei (Hungary) revealed by a combined
cultivation-based and molecular approach. Microb Ecol 56:64–75
91. Wagner M, Roger AJ, Flax JL, Brusseau GA, Stahl DA (1998)
Phylogeny of dissimilatory sulfite reductases supports an early
origin of sulfate respiration. J Bacteriol 180:2975–2982
92. Wang R, Baldy V, Perissol C, Korboulewsky (2012) Influence of
plants on microbial activity in a vertical-downflow wetland system
treating waste activated sludge with high organic matter concen-
trations. J Environ Mgt 95:S158–S164
93. Whitmire SL,Hamilton SK (2005) Rapid removal of nitrate and sulfate
in freshwater wetland sediments. J Environ Qual 34:2062–2071
94. Wiessner A, Kuschk P, Jechorek M, Seidel H, Kastner M (2008)
Sulphur transformation and deposition in the rhizosphere of Juncus
effusus in a laboratory-scale constructed wetland. Environ Pollut
155:125–131
95. Yin J, Jiang LY, Wen Y, Yao ZL, Zhou Q (2009) Treatment of
polluted landscape lake water and community analysis of
ammonia-oxidizing bacteria in constructed wetland. J Environ
Sci Health Part A Toxic/Hazard Subst Environ Eng 44:722–
731
96. Zhu T, Sikora FJ (1995) Ammonium and nitrate removal in vegetated
and unvegetated gravel bed microcosm wetlands. Water Sci Technol
32:219–228