You are on page 1of 8

Chemosphere 285 (2021) 131322

Contents lists available at ScienceDirect

Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere

The effects of Fe(III) and Fe(II) on anammox process and the


Fe–N metabolism
Yao Chen, Fangxu Jia *, Yingjie Liu, Wanrou Yu, Weiwei Cai, Xiaofan Zhang, Haodong He,
Hong Yao
Beijing Key Laboratory of Aqueous Typical Pollutants Control and Water Quality Safeguard, Department of Municipal and Environmental Engineering, School of Civil
Engineering, Beijing Jiaotong University, Beijing, 100044, PR China

A R T I C L E I N F O A B S T R A C T

Handling Editor: Dr A Adalberto Noyola This study aims to compare the effects of different Fe stress on anammox (anaerobic ammonium oxidation)
process, therefore seven identical reactors were operated under different Fe(II)/Fe(III) concentrations. After 38
Keywords: days of operation, the anammox activity was highest (10.49 ± 0.41 mg-TN/(g-VSS⋅h)) under conditions of 5 mg/
Anammox L-Fe(II), while under 30 mg/L-Fe(III) displayed severe inhibition. The results showed that continuous addition of
Ferric
30 mg/L-Fe(III) would damage the composition of EPS (extracellular polymeric substances) and make anammox
Ferous
bacteria more sensitive to environmental stress. While high Fe(II) concentrations could result in precipitates
Fe(II)-dependent nitrate reduction
Nitrogen removal encasing granular sludge, affecting substrate utilization. Moreover, the results of ΔNO-3-N/ΔNH+ 4 -N indicated
that Fe(II)-dependent nitrate reduction was induced in reactors added with Fe(II). OM27_clade and nor­
ank_f__Burkholderiaceae might be candidates for this process according to the correlation of genera and functional
genes (based on the PICRUSt 2 functional prediction). Overall, this research is expected to provide new ideas to
the effects of Fe(II)/Fe(III) on anammox and to the practical application of coupled system based on anammox in
wastewater treatment.

1. Introduction Niftrik et al., 2008) which may relate to iron respiration. Therefore, the
addition of Fe might be a potential method to enhance the activity of
Anaerobic ammonium oxidation (anammox) is a promising nitrogen AnAOB. To date, previous studies showed that the addition of Fe could
removal processing for treating high ammonia wastewater. Compared to stimulate anammox bacteria. For example, the start-up period reduced
the conventional nitrification denitrification process, it is more cost- by 20 days (Bi et al., 2014) and the specific anammox growth rate
effective and environmentally friendly as it can reduce oxygen de­ enhanced by 45.8% (Liu and Ni, 2015) with the addition of 5 mg/L-Fe
mand and the emission of CO2 and N2O with no organic carbon con­ (II). The activity of AnAOB increased five times with the addition of
sumption and lower sludge yields (Op Den Camp et al., 2006; Kuenen, 3.68 mg/L-Fe(III) (Chen et al., 2014). Moreover, the promotion of
2008). However, anammox bacteria (AnAOB) are quite sensitive to the zero-valent iron (Ren et al., 2015, 2016; Zhang et al., 2017), Fe3O4 (Gao
surrounding environment and have a long generation cycle of 11–20 et al., 2014), and Fe electrode (Zhang et al., 2012; Xie et al., 2020) on
days (Jetten et al., 2009). Therefore, the recovery of anammox activity is anammox have also reported, while reports on the inhibition of anam­
time-consumed once inhibited by interfering substances contained mox by Fe are relatively few. Therefore, concentrations of 5 mg/L
within wastewater (such as heavy metal, antibiotics, etc. (Pulicharla (relatively optimal), 15 mg/L and 30 mg/L (relatively high) were chosen
et al., 2015; Chen et al., 2016). Consequently, increasing the activity of to investigate the facilitation and suppression of Fe(II)/Fe(III) on
anammox and accelerating the recovery of reactors remain important anammox. More importantly, the majority of continuous experiments
challenges. have been conducted in a single reactor, in which the Fe concentration
Iron can promote the synthesis of heme c and Fe–S proteins in was increased sequentially in a stepwise manner, rather than compared
AnAOB which are essential for their metabolism (Ferousi et al., 2017). directly in several parallel (independent) reactors. A sequential experi­
Besides, AnAOB contain a large number of dense iron particles (van mental scheme makes it difficult to determine the actual effective

* Corresponding author.
E-mail address: jiafx@bjtu.edu.cn (F. Jia).

https://doi.org/10.1016/j.chemosphere.2021.131322
Received 20 April 2021; Received in revised form 14 June 2021; Accepted 20 June 2021
Available online 25 June 2021
0045-6535/© 2021 Elsevier Ltd. All rights reserved.
Y. Chen et al. Chemosphere 285 (2021) 131322

concentration of Fe as it can easily precipitate and be absorbed into the NO−2 -N were in form of NH4Cl and NaNO2, respectively. The composi­
anammox sludge. Therefore, comparative experiments are required to tion of synthetic wastewater referred to previous publication (Bi et al.,
help understand and reduce the cumulative effects of Fe. 2014), but no trace element solution I was added to avoid extra Fe(II)
On the other hand, Fe is a potential energy source that can be used as addition. Each reactor was fed a different Fe(III)/Fe(II) concentration
either the electron donor or acceptor with its different valence states. (sequestrated by EDTA) (Fig. 1). Influent dissolved oxygen (DO) con­
The redox reactions of Fe could be carried out by microbes or abiotically. centration was controlled to be under 0.1 mg/L and the pH was adjusted
Moreover, Fe plays a significant role in nitrogen removal and nitrogen to 7.0 ± 0.2.
cycling (Melton et al., 2014), such as anaerobic ammonium oxidation Operation strategies are summarized in Table 1 and were the same
coupled with ferric reduction (Feammox), Nitrate-dependent Fe(II) for each reactor. After reaching the stable state, nitrogen loading rate
oxidation (NDFO), and Fe(II)-dependent dissimilatory nitrate reduction (NLR) was raised by increasing the influent nitrogen concentration and/
to ammonium (Fe(II)-dependent DNRA). Feammox coupled with or the flow of influent. Daily water samples were taken to determine the
anammox contributes to nitrogen removal in wetlands and paddy soils concentrations of nitrogen compounds. At the end of each phase, the
(Yang et al., 2012; Ding et al., 2014, 2020). Meanwhile, Fe mixed liquor was withdrawn from the reactors for subsequent batch
(II)-dependent nitrate reduction (NDFO and DNRA) could reduce ni­ tests. Before that, the sludges were washed three times with oxygen-free
trate with Fe(II) as the electron donor, providing a new pathway for total water to remove residual NH+ 4 -N and NO2 -N. At which point the specific

nitrogen removal (Han et al., 2020; Li et al., 2020a). Generally, there is a anammox activity (SAA), EPS contents, SS and VSS of sludges were
coupling of anammox, nitrification, DNRA, Feammox, and NDFO in determined.
majority anammox reactors (Shu et al., 2016). However, these Fe–N
metabolism pathways are often neglected, even though they are likely to 2.2. Determination of SAA
be present within reactors due to the continuous addition of Fe. Thus, it
is necessary to take these reactions into consideration when researching Batch tests were performed in serum flasks (250 mL). Approximately
the effects of Fe on anammox. Furthermore, how to establish a coupled 150 mL mixture was added into each serum flask. The initial concen­
system to achieve high treatment efficiency also requires continued trations of NH+ 4 -N and NO2 -N were 30 and 40 mg/L, respectively, and

study and could potentially be beneficial to the engineering application pH was adjusted to 7.0 ± 0.2. After flushing with 99.99% nitrogen gas,
of anammox. the serum flasks were sealed with silicon/Teflon gaskets and poly­
Consequently, the main objectives of this study were to (1) compare propylene caps, and then incubated on an orbital shaker (180 rpm, 35 ±
the facilitation and suppression of different Fe(II)/Fe(III) concentrations 1 ◦ C) in the dark. Experiments were conducted in triplicate. Water
(5, 15 and 30 mg/L) on anammox process; (2) explore the reactions samples were collected periodically over 8 h by a syringe to determine
which might be induced into systems under different Fe stress. There­ the concentrations of NH+ 4 -N, NO2 -N, and NO3 -N.
− −

fore, seven identical reactors were operated and nitrogen removal per­
formance, EPS, and community structure of reactors were investigated. 2.3. EPS extraction and determination
Meanwhile, the correlation of genera and genes (results of PICRUSt 2
functional prediction) were also assessed to determine the potential EPS was extracted by the ultrasound method according to Jia et al.
functional bacteria of reactions related to iron metabolism. (2017): 40 mL mixed liquors were taken from each reactor and washed
three times using PBS buffer and the supernatant was discarded after
2. Materials and methods sedimentation. Next the granules were crushed (mortar and pestle) and
resuspended in PBS for sonication (20 kHz and 5.59 W/mL for 1 min);
2.1. Experimental set-up and operation strategy Finally, the suspension was centrifuged (20 min, 20000 g), and super­
natant were collected after filtering by 0.45 μm polytetrafluoroethylene
Seven identical UASB, made of polymethyl methacrylate and a filters. The content of proteins, humic acid, and polysaccharides was
working volume of 0.42 L (diameter of 3.0 cm and height of 60.0 cm), measured according to Badireddy et al. (2010). DNA content was
were utilized for experiments. A thermostatic water-box was equipped measured by the diphenylamine method (De Mey et al., 2006).
to maintain a fixed temperature of 35 ± 1 ◦ C. The schematic diagram
and reactor designations are shown in Fig. 1. Anammox granular sludge,
2.4. Illumina MiSeq sequencing
refrigerated (4 ◦ C) for over one year, was inoculated into the reactors
with an initial biomass concentration of 8.0–10.0 g-VSS/L. NH+ 4 -N and Anammox biomass collected from the seven reactors (triplicate
samples for each reactor) on day 38 was explored by Illumina MiSeq
sequencing to determine the microbial community structure. Samples
were sent to Majorbio Company (Shanghai, China). The universal
primer pair 338 F (5′ -ACTCCTACGGGAGGCAGCAG-3′ ) and 806 R (5′ -
GGACTACHVGGGTWTCTAAT-3′ ) was utilized for amplification and
sequencing of the 16 S rRNA gene V3 ~ V4 region. All data analysis were
finished by I-sanger Cloud Platform (www.i-sanger.com). Sequencing
reads were clustered into operational taxonomic units (OTUs) at an
identity threshold of 0.97. The functional genes were established using
PICRUSt2 function prediction. 31,152 reads were obtained from each
sample after subsampling and the coverage was greater than 99.4%,
indicating that the diversity of the microbial community could be well

Table 1
Experimental conditions of continuous-flow experiments.
Phase NH+
4 -N (mg/L) NO−2 -N (mg/L) HRT (h) Duration time (d)

1 28.55 ± 2.00 37.68 ± 1.73 2.7 1–10


2 54.64 ± 0.98 83.44 ± 3.51 2.7 11–24
Fig. 1. The schematic diagram and the influent conditions (valence and con­
3 114.96 ± 3.55 147.78 ± 4.38 1.8 25–38
centration of iron) of the reactors.

2
Y. Chen et al. Chemosphere 285 (2021) 131322

represented by the sequences. After cultivation, the 16 S rRNA gene the nitrogen removal rate (NRR) of the seven reactors had stabilized at
community diversity of RII-30 was significantly lower than other re­ 0.48–0.54 kg-N/(m3⋅d) and the NH+ 4 -N and NO2 -N removal efficiencies

actors according to the Shannon and Simpson index (Table S1). In were above 96.0% and 96.3% (not shown in the figure), respectively.
general, the community richness and diversity of RII-5 ~ RII-30 were Although the nitrogen removal performances of the seven reactors were
lower than that of RIII-5 ~ RIII-30. Additionally, all original sequencing similar, RIII-30 had the highest SAA, at 4.21 ± 0.54 mg-TN/(g-VSS⋅h),
data have been submitted to the National Center for Biotechnology In­ while RII-30 was the lowest (2.35 ± 0.21 mg-TN/(g-VSS⋅h)) (Fig. 2(d)).
formation (NCBI) repository with BioProject PRJNA693360. During phase 2, the effluent NH+ 4 -N and NO2 -N concentrations dropped

below 1 mg/L within 2 days (3 days for RIII-30) even the influent ni­
2.5. Other analytical analysis trogen concentrations were doubled. The maximum NH+ 4 -N concentra­
tion in effluent of RIII-30 was 22.0 mg/L (NO−2 -N was 30.1 mg/L) after
Water samples were analyzed immediately after filtrating with a increasing the NLR, while all other reactors were below 10.4 mg/L
0.45 μm polyether-sulfone filter. The NH+ 4 -N, NO2 -N, and NO3 -N con­
− − (NO−2 -N was below 13.0 mg/L). Together this indicated that inhibition
centrations and the VSS and SS of sludge were determined according to by high Fe(III) concentration (30 mg/L) emerged gradually. Meanwhile,
standard methods (Eaton, 2005). The pH was determined by a WTW the SAA of RIII-30 (4.20 ± 0.37 mg-TN/(g-VSS⋅h)) did not increase as
Multi340i pH meter (WTW, Germany). The significance of the results compared to phase 1 (Fig. 2(d)). The SAA of RII-5, RII-15 and RII-30
and linear fittings were performed with OriginPro 2020 (OriginLab, (5.45 ± 0.29, 5.50 ± 0.14 and 4.71 ± 0.19 mg-TN/(g-VSS⋅h), respec­
Northampton, MA, USA) and p < 0.05 represented statistically signifi­ tively) surpassed that of their Fe(III) counterparts, RIII-5, RIII-15 and
cant value. The Pearson Indices (r) and Significant Correlation (p) of the RIII-30 (4.30 ± 0.51, 4.11 ± 0.66 and 4.20 ± 0.37 mg-TN/(g-VSS⋅h),
correlations between microbial community structure and functional respectively). This showed that the activity of AnAOB in the presence of
genes under Fe(II) or Fe(III) stress were calculated by R (v 3.6.3) using Fe(II) was higher than that with Fe(III) after 24 days of cultivation. But
the RStudio (v1.2.5033, using psych, reshape 2 packages) environments. only the SAA of RII-5 and RII-15 were higher than the control (R0, 5.27
± 0.52 mg-TN/(g-VSS⋅h)). Notably, the SAA of RII-30 increased
3. Results dramatically during phase 2 as compared to phase 1.
During phase 3, the influent NH+ 4 -N and NO2 -N concentrations were

3.1. Effects of Fe(III) and Fe(II) on AnAOB activity approximately 114.96 ± 3.55 mg/L and 147.78 ± 4.38 mg/L while the
HRT was halved, therefore the NLR was tripled as compared to phase 2.
The variation of nitrogen removal performance of the seven reactors At this point, the differences among reactors in nitrogen removal per­
after 38 days of operation is showed in Fig. 2(a ~ c). During the phase 1, formance began to emerge. The effluent NH+ 4 -N and NO2 -N concentra­

the influent NH+4 -N and NO2 -N concentrations were 28.54 ± 2.00 mg/L
− tion of most reactors were stable at ~0 mg/L after a period of adaption,
and 37.68 ± 1.73 mg/L, respectively. NLR was in the range of 0.61–0.72 however, those of RIII-30 remained above 95.0 mg/L and 117.1 mg/L,
kg-N/(m3⋅d). The maximum NH+ 4 -N concentration in effluents of the
respectively, indicating that anammox activity within RIII-30 was
seven reactors was at 23.0–25.0 mg/L (NO−2 -N was 21.0–25.8 mg/L) on dramatically inhibited. To further elucidate the differences in perfor­
day 1 and stabilized at 0 mg/L after 3–4 days. At the end of this period, mance of each reactor after increasing the NLR, linear fitting was

Fig. 2. Operation performance of the seven reactors


with the continuous addition of different Fe (Fe(III)/
Fe(II)) concentrations: The variation of (a) ammonia
(b) nitrite concentrations. (c) NLR and NRR
throughout the experiment. The variation of (d) SAA
and (e) VSS/SS ratios at the end of each phase. (f) The
linear fitting of each reactor during the adaptation
period in phase 3. “k” is the slope of the fitting curve,
representing the adaptive rate of reactors. Asterisks
indicate a significant correlation. *P < 0.05, **P <
0.01.

3
Y. Chen et al. Chemosphere 285 (2021) 131322

performed, and the adaptive rate (slope of the fitting curve, k) of ni­
trogen load impact was calculated during the adaption period (Fig. 2(f)).
The adaptive rate of RII-5 and R0 were the highest (0.98 and 0.86 kg-N/
(m3⋅d2)) followed by that of RII-15 and RII-30 (0.67 and 0.49 kg-N/
(m3⋅d2)), with RIII-5 and RIII-15 having the lowest rates (0.32 and 0.27
kg N/(m3⋅d2)). However, the SAA of RII-15 and RII-30 (8.40 ± 0.13 and
7.88 ± 0.30 mg-TN/(g-VSS⋅h), respectively) were lower than that of
RIII-5 and RIII-15 (8.66 ± 0.30 and 8.52 ± 0.23 mg-TN/(g-VSS⋅h),
respectively). These contrary results were caused by the accidental
washing out of partial sludge in RIII-5 and RIII-15 during phase 3. The
SAA of RII-5 (10.49 ± 0.41 mg-TN/(g-VSS⋅h)) was the highest and the
SAA of all reactors, except that of RIII-30, increased dramatically.
Meanwhile, the VSS/SS ratios of reactors were also calculated and are
showed in Fig. 2(e). The ratios of RII-5, RII-15 and RII-30 were generally
lower than that of others and shows a downward trend (from 0.84 ±
0.03 to 0.63 ± 0.01). In total, the ratios of RII-15, RII-30 and RIII-30
decreased dramatically with the consistent addition of corresponding
Fe(II)/Fe(III) concentrations (from 0.83 ± 0.00 to 0.75 ± 0.01, from
0.82 ± 0.00 to 0.63 ± 0.01 and from 0.87 ± 0.01 to 0.81 ± 0.03,
respectively).

3.2. Difference of stoichiometric ratio in reactors under different Fe stress

To discern which reactions occurred in each reactor, the molar ratios


ΔNO−2 -N/ΔNH+ 4 -N and ΔNO3 -N/ΔNH4 -N during the stable state were
− +

calculated for each phase (Fig. 3). In all three reactors amended with Fe
(III) (RIII-5 ~ RIII-30), the ΔNO−2 -N/ΔNH+ 4 -N ratios were generally
higher than the theoretical value (1.32) of the anammox reaction,
especially for phases 2 (1.32–1.47) and 3 (1.34–2.91) (Fig. 3(a)).
However, there was no obvious correlation between ΔNO−2 -N/ΔNH+ 4 -N
and Fe(III)/Fe(II) concentration. On the contrary, the ΔNO−3 -N/ΔNH+ 4 -N
ratios increased with increasing Fe(III) concentration in each phase
(Fig. 3(b)), from 0.24 ± 0.04 to 0.28 ± 0.06, from 0.27 ± 0.02 to 0.31 ±
0.03 (r = 0.534, P < 0.01) and from 0.24 ± 0.03 to 1.05 ± 0.49 (r = Fig. 3. The (a) ΔNO−2 -N/ΔNH+ 4 -N and (b) ΔNO3 -N/ΔNH4 -N of reactors in
− +

0.718, P < 0.01), respectively. Notably, the ΔNO−2 -N/ΔNH+ 4 -N and phases 1, 2, and 3. The black solid triangle represents the mean value of data.
ΔNO−3 -N/ΔNH+ 4 -N ratios of RIII-30 deviated highly from the theoretical The black solid pentacle above and below the box represents the maximum and
value during phase 3 as its performance was severely damaged. In minimum values of the data, respectively. The upper and lower edges of the box
general, these results demonstrated that continuous addition of Fe(III) represent the 25th and 75th percentile of the data. The heatmaps inside the blue
would induce the excessive transformation of NO−2 -N and NO−3 -N. box represent the correlation between ΔNO−2 -N/ΔNH+ 4 -N (in (a)) and ΔNO3 -N/

In the reactors which had been amended with Fe(II) (RII-5 ~ RII-30), ΔNH+ 4 -N (in (b)) with Fe(III) concentrations. Meanwhile, the heatmaps inside
the ΔNO−2 -N/ΔNH+ the red box represent the correlation between ΔNO−2 -N/ΔNH+ 4 -N (in (a)) and
4 -N ratios (1.29–1.44) were also generally higher
ΔNO3 -N/ΔNH+

4 -N (in (b)) with Fe(II) concentrations. Asterisks indicate a sig­
than the theoretical values, while the ΔNO−3 -N/ΔNH+ 4 -N ratios declined
nificant correlation. *P < 0.05, **P < 0.01. (For interpretation of the references
with increasing Fe(II) concentrations in each phase, from 0.25 ± 0.07 to
to colour in this figure legend, the reader is referred to the Web version of
0.21 ± 0.04 (r = − 0.501, P < 0.05), from 0.28 ± 0.02 to 0.23 ± 0.02 (r this article.)
= − 0.569, P < 0.01) and from 0.27 ± 0.03 to 0.19 ± 0.02 (r = − 0.614, P
< 0.01), respectively. The ΔNO−2 -N/ΔNH+ 4 -N of RII-5 was close to the
theoretical value (0.26) of anammox and the reactors with relatively
high Fe(II) concentrations were lower. In general, the ΔNO−2 -N/ΔNH+ 4 -N
molar ratios of the seven reactors were all higher than the theoretical
value throughout the entire experiment, and the ΔNO−3 -N/ΔNH+ 4 -N ratio
was heavily dependent on the valence and concentration of iron ions. In
total, the results indicated that some reactions other than anammox
were induced by Fe addition and the difference was quite significant
between the reactors to which Fe(III) and Fe(II), respectively, had been
added.

3.3. The variation of EPS contents

EPS plays an essential role in the formation of multilevel porous


structures on the cell surface for self-protection and to promote micro­ Fig. 4. The concentrations of EPS contents and the PN/PS ratio in the seven
bial aggregation (Henriques and Love, 2007). Fe(III)/Fe(II) could either reactors at the end of each phase.
affect anammox activity directly, or via the EPS. Therefore, the variation
of EPS content for each reactor was analyzed (Fig. 4). Except for reactor than that of the other reactors (268.8–300.8 mg/(g-VSS)), and the
RIII-30, the total EPS contents and PN/PS (proteins/polysaccharide PN/PS ratio displayed a similar trend. Notably, the PN/PS ratio of
ratio) ratios of the sludges increased with time, especially in phase 3. RIII-30 in phase 3 (0.73 ± 0.04) was lower than that in phase 2 (0.84 ±
The total EPS content of RIII-30 (268.2 ± 11.9 mg/(g-VSS)) was lower

4
Y. Chen et al. Chemosphere 285 (2021) 131322

0.01), indicating that excess Fe(III) addition had a greater effect on other reactions in the systems, therefore it was necessary to analyze
protein content. The EPS content of reactors with high Fe addition other potential functional bacteria, as well as their correlation with Fe
concentrations (265.51 ± 27.8, 268.4 ± 21.4, 221.18 ± 29.9 and 221.5 and N related functional genes (Fig. 5(b)). The significance of difference
± 25.9 mg/(g-VSS) in RIII-15, RIII-30, RII-15, and RII-30, respectively) among the 21 samples was also analyzed. After 38 days of cultivation
were higher than those with low concentrations (204.3 ± 32.5 and with continuous Fe addition, the community structure of RIII-30 was
189.4 ± 22.3 mg/(g-VSS) in RIII-5 and RII-5, respectively) in phase 1. It significantly different from the others (P < 0.01) as reactor had been
should also be noted that reactors amended with Fe(III) were higher severely destroyed. The abundance of NOB, including Nitrospira and
than those amended with Fe(II) at the same concentration. However, the Candidatus Nitrotoga, was higher in reactors which had been amended
increased rates of RII-5 (5.2 mg/(g-VSS⋅d)) and RIII-5 (3.4 with Fe(III) (P < 0.05). The abundance of denitrifying bacteria,
mg/(g-VSS⋅d)) were higher than others (data not shown). Therefore, the including Comamonas and Denitratisoma, displayed a similar trend.
EPS content decreased with increasing Fe(III) (from 300.0 ± 18.2 to Meanwhile, the abundance of Thermomonas and Simplicispira, which
268.9 ± 12.0 mg/(g-VSS)) and Fe(II) (from 335.0 ± 19.7 to 270.2 ± 26.1 were respectively related to Fe–Mn (Hou et al., 2020) and As(III) (Cui
mg/(g-VSS)) concentrations in phase 3. Moreover, only the total EPS et al., 2018) oxidation, was higher in the reactors with high Fe(III)/Fe
contents of RIII-5, RIII-15, and RII-5 were higher than the control, which (II) concentrations (P < 0.01 and P < 0.01, respectively). Members of the
was in accordance with the SAA results of anammox. genus Paludibaculum have been reported to be capable of fermentation
and dissimilatory Fe(III) reduction (Kulichevskaya et al., 2014), but its
abundance was reduced as Fe(III)/Fe(II) concentrations increased. The
3.4. Microbial community evolution genus Ignavibacterium, reported as potential functional Feammox bac­
teria (Takao et al., 2010), increased with reactor operation time, how­
A total of 11 abundant phyla (abundance ≥ 0.01%) were detected in ever its proportion was always less than 0.1%. The correlations between
all samples (Fig. S2). The most dominant phylum was Chloroflexi which genera and functional genes were quite different within the different Fe
accounted for 56.0–69.0% of sequences, followed by Planctomycetes valences circumstances (as shown in Fig. 5 (b)). When Fe(III) was less
(11.8–28.6%), Proteobacteria (4.8–12.6%), and Bacteroidetes (2.6–4.8%). than 15 mg/L (RIII-5 ~ RIII-15), Ca. Brocadia was closely and positively
Candidatus Brocadia was the representative genus for Planctomycetes in correlated with feoB gene which is speculated to be the only iron
this study, given that its abundance variation was highly in accordance transporter gene in anammox genomes (Ferousi et al., 2017). However,
with that of Planctomycetes as a group (r = 0.999, P < 0.01). As shown in there was no significant correlation between Ca. Brocadia and feoB
Fig. 5(a), the abundance of Ca. Brocadia in RIII-30 (11.6 ± 4.6%) was under Fe(II) stress (RII-5 ~ RII-30). These results indicated that the feoB
significantly lower than that of other reactors (18.3–28.4%) (P < 0.05). gene was inactive in Fe(II)-rich environments. NOB, denitrifying bac­
The relative abundance of Ca. Brocadia roughly decreased with teria, SWB02 and Ignavibacterium showed significant positive correla­
increasing concentration of Fe(II) or Fe(III), with the highest Ca. Bro­ tions with fieF (ferrous-iron efflux pump) and nitrate reduction genes
cadia abundance (28.4 ± 2.6%) observed with 5 mg/L-Fe(II). Addi­ (nirA, nrfA and/or nirS) only under Fe(III) exposure. OM27_clade and
tionally, the abundance of Ca. Brocadia in RIII-5 and RIII-15 (25.6 ± 2.2 norank_f__Burkholderiaceae showed significant positive correlation with
and 22.8 ± 5.8%, respectively) were higher than that of the control (R0) nirS gene, but they were positive correlated with feoA, ftnA and/or fur
(18.4 ± 1.5%). While the abundance of other two reactors (RII-15 and only under Fe(II) stress.
RII-30, 19.7 ± 3.4 and 18.2 ± 5.8%, respectively) were similar to that of
R0. These results were supported the SAA results.
As mentioned previously, the addition of Fe(III)/(II) could induce

Fig. 5. Analysis of the microbial community struc­


ture (a) at genus level (“others” represents all classi­
fied taxa that were < 1%) and (b) the correlation
between genera and functional genes under Fe(III) or
Fe(II) stress (bubble matrix). The heatmap shows the
relative abundance of functional bacteria at genus
level. The diameter of bubble represents the value of
correlation coefficient r. Orange and blue bubbles
represent positive and negative correlations, respec­
tively. The RIII-30 was excluded since its community
structure was quite different compared with others
and was severely inhibited. The data with r < 0.5 or p
> 0.05 were filtered out. “△” and “☆” represents the
significance of difference among seven samples
(heatmap) and the significance of the correlation
(bubble). △, P < 0.05; ☆, P < 0.01. (For interpre­
tation of the references to colour in this figure legend,
the reader is referred to the Web version of this
article.)

5
Y. Chen et al. Chemosphere 285 (2021) 131322

4. Discussion single reactor where the influent Fe(II) concentration was gradually
increased (from 2 to 50 mg/L and from 0 to 378.6 mg/L, respectively),
4.1. Effects of Fe(III) and Fe(II) on anammox thus it was hard to determine the true effective Fe(II) concentrations in
the reactors. In total, the mechanisms of influence (especially inhibition)
Our results showed that the seven reactors displayed similar per­ of Fe(II) and Fe(III) still remain elusive, therefore further studies and
formance during the initial period (Fig. 1), which might be due to the more advanced technologies are needed to ascertain the effect of Fe on
high biomass concentrations (8.0–10.0 g-VSS/L) and low substrate anammox.
concentrations. The difference emerged only after increasing the NLR.
The activity of anammox could be enhanced slightly by the continuous 4.2. The coupling system induced by Fe(III) and Fe(II)
addition of relatively low concentrations of Fe(III) (<15 mg/L), but
would be severely inhibited by high concentration (30 mg/L). The ni­ In this study, the extra consumption of NO−2 -N was observed
trogen removal of RIII-30 was almost undetectable during phase 3 throughout the experiment and the ΔNO-3-N/ΔNH+ 4 -N also deviated
(Fig. 2) and the relative abundance of AnAOB was greatly reduced from the theoretical value (Fig. 3). It could be presumed that other re­
(Fig. 5). This meant the amount of Fe(III) addition could have been actions, besides anammox, were present in the reactors. In reactors
beyond the iron adsorption capacity of sludge and inhibiting the activity amended with Fe(III) (RIII-5, RIII-15, and RIII-30), two genera of NOB
and growth of AnAOB. However, Li et al. (2014) and Yuan and Wang were observed and the abundance of NOB was higher than that with Fe
(2019) concluded that there was no obvious inhibition by excess Fe(III) (II). One possible explanation was that the existence of Fe(II) could
(5–65 mg/L and 10–40 mg/L, respectively) during batch tests. They consume the residual DO (Yan et al., 2019) (<0.1 mg/L) within a reactor
believed inhibition by high Fe(III) was avoided due to the precipitation to the point that the growth of NOB was inhibited. Feammox, as a newly
of Fe(III) with OH− produced during the anammox reaction. This pre­ discovered reaction, can oxidize ammonia to NO−x -N and N2 using Fe(III)
cipitation phenomenon was also observed during the present study, with as the electron acceptor (Park et al., 2009). It may provide nitrite for
the majority of sediments accumulating at the bottom of the reactors. anammox without aeration (if well controlled), which can replace par­
That was why the VSS/SS ratio of RIII-30 was slightly decreased and tial nitrification and save on energy costs. Therefore, the coupling of
those of RIII-5 and RIII-15 were relatively unchanged. But the total EPS Feammox and anammox in wastewater treatment might have unex­
contents, especially the protein contents, of RIII-30 were obviously pected benefits. However, Feammox did not exist, or was not dominant
lower than of other reactors in phase 3 (Fig. 4), the same decrease yet, in these reactors. This could have been partly due to the short in­
tendency of protein content was also observed with long-term exposure cubation time (the cultivation period of Feammox is generally long,
to 50 mg/L-Fe(III) according to Zhang et al. (2021). Given these results, more than 6 mouths), and partly to substrate competition (like NH+ 4 -N)
it is reasonable to speculate that the continuous addition of high con­ between Feammox and anammox.
centration of Fe(III) could damage the composition of EPS (especially In reactors amended with Fe(II) (RII-5, RII-15, and RII-30), the
the protein component) and make AnAOB more sensitive to the envi­ ΔNO-3-N/ΔNH+ 4 -N ratio was lower than the theoretical value and
ronmental inhibitory factors (such as nitrite), rather than inhibit the decreased with increasing Fe(II) concentration. This extra consumption
anammox directly. Therefore, the activity of RIII-30 was inhibited of NO−3 -N has been reported in other studies as well (Bi et al., 2014; Ren
severely when the NLR was tripled in phase 3. et al., 2015; Shu et al., 2016). However, denitrifying bacteria in these
As mentioned above, the reactors with the addition of Fe(II) were reactors were no more abundant than in the Fe(III) reactors. Therefore
generally superior to that with Fe(III) and the optimal concentration of denitrification might not be the main reason, but rather NDFO or Fe
Fe(II) is 5 mg/L, which was in agreement with previous studies (Qiao (II)-dependent DNRA instead. These reactions could utilize Fe(II) as
et al., 2013; Mak et al., 2019). This phenomenon can be explained in two electron donor to reduce NO−3 -N (produced by anammox process) to
aspects according to previous studies. Firstly, most of these cellular irons nitrite and ammonia, and further support their central metabolism
are divalent and the Fe(II) can be utilized directly while Fe(III) needs to (anammox) in a coupled system. Therefore, it could improve nitrogen
be reduced to Fe(II) first (Ferousi et al., 2017). Secondly, the presence of removal performance of reactors and even achieve zero nitrogen emis­
Fe(II) helps maintain an anaerobic environment for anammox. Besides, sions. Similar coupled systems were created by other researcher groups
The anammox activity of RII-30 was slightly inhibited during phase 1 through the addition of Fe (Bi et al., 2019, 2020; Li et al., 2020a) or
and promoted during phases 2 and 3 (compared to R0), revealing that NO−3 -N (Yang et al., 2020) into the original reactor. Besides, there were a
inhibition of 30 mg/L-Fe(II) was recoverable in the prophase cultiva­ few candidates, even we could not definitively link any specific func­
tion. The same recoverable inhibition was also observed by Ding et al. tional microorganisms to these metabolisms. As mentioned above (sec­
(2021), the inhibition by 56 mg/L-Fe(II) could recover in 24 h without tion 3.3), the denitrifiers, Thermomonas and Simplicispira, might have the
changing any conditions. At the end of experiment, the SAA and rela­ ability of reduce nitrate in conjunction with Fe(II) oxidation since they
tively abundance of AnAOB in RII-15 and RII-30 were close to the were related to the metal oxidizing in environment. Two genera,
control. However, the VSS/SS ratios of RII-15 and RII-30 were greatly OM27_clade and norank_f__Burkholderiaceae, might also be candidates for
decreased (Fig. 3(b)), especially that of RII-30. Compared to Fe(III), the nitrate reduction with Fe(II) oxidation since they showed close, positive
deposition rate of Fe(II) was relatively slower such that the precipitate correlation with feoA, ftnA, and/or fur only under Fe(II) stress. Mean­
was more likely to be absorbed into the EPS and even accumulate on the while the strain Candidatus Brocadia sinica was also reported to be
surface of the granular sludge, which might affect substrate utilization capable of oxidizing Fe(II) with the reduction of nitrate (Oshiki et al.,
and mass transfer (Wang et al., 2017). This absorption could be the main 2013).
cause of inhibition by Fe(II) on anammox. Therefore, it was reasonable Overall, this coupled system could be more beneficial for the appli­
to presume that the activity of RII-15 and RII-30 might be inhibited cation of anammox process since nitrate is generally present at much
during subsequent reactor operation (>38 d) since the amount of pre­ higher concentration than nitrite in wastewater and Fe is a very cheap
cipitate accumulating in the sludges would be considerable and possibly and widely available element. Wastewater rich in Fe or nitrate and
even encapsulate the sludge particles completely. In comparison, Mishra sludge produced by Fenton reaction could be utilized as substrates for
et al. (2020) found that inhibition of anammox by 30 mg/L-Fe(II) was the coupled system to turn waste into treasure. Moreover, various iron-
unrecoverable, while Li et al. (2020b) found that anammox activity was based or iron-modified fillers could be used in anammox system to form
not inhibited severely until influent Fe(II) concentration reached 378.6 a coupled system. However, the morphology and redox of Fe are
mg/L. These differences might be due to the different reactor types complicated in the reactor environment, and it was hard to determine
(UASB and UBF, respectively), sludge concentrations, operation condi­ the actual effective concentration of Fe. Therefore, the mechanism by
tions, etc. Additionally, these two experiments were both conducted in a which Fe affects these functional bacteria requires continued study.

6
Y. Chen et al. Chemosphere 285 (2021) 131322

Moreover, the best dosing method and dosing concentration need to be estimation in metabolic modeling of microorganisms. Anal. Biochem. 353 (2),
198–203.
developed to build a coupled system that does not inhibit anammox
Ding, B., Qin, Y., Luo, W., Li, Z., 2020. Spatial and seasonal distributions of Feammox
activity. from ecosystem habitats in the Wanshan region of the Taihu watershed, China.
Chemosphere 239, 124742.
Ding, J., Seow, W., Zhou, J., Zeng, R.J., Zhou, Y.J., 2021. Effects of Fe(II) on anammox
5. Conclusion community activity and physiologic response. Front. Environ. Sci. Eng. 15 (1).
Ding, L.-J., An, X.-L., Li, S., Zhang, G.-L., Zhu, Y.-G., 2014. Nitrogen loss through
In this work, the effects of three Fe(III) and Fe(II) concentrations (5, anaerobic ammonium oxidation coupled to iron reduction from paddy soils in a
chronosequence. Environ. Sci. Techno. 48 (18), 10641–10647.
15, and 30 mg/L) on anammox were investigated. It was found that 5
Eaton, AndrewD., 2005. Standard Methods for the Examination of Water & Wastewater.
mg/L-Fe(II) was optimal for anammox with the SAA of 10.49 ± 0.41 mg- American Public Health Association, Washington, DC, USA.
TN/(g-VSS⋅h), while the continuous addition of 30 mg/L-Fe(III) would Ferousi, C., Lindhoud, S., Baymann, F., Kartal, B., Jetten, M.S., Reimann, J., 2017. Iron
assimilation and utilization in anaerobic ammonium oxidizing bacteria. Curr. Opin.
inhibit AnAOB severely. The results showed that long-term addition of
Chem. Biol. 37, 129–136.
high Fe(III) concentration damaged the composition of EPS, while that Gao, F., Zhang, H., Yang, F., Li, H., Zhang, R., 2014. The effects of zero-valent iron (ZVI)
of high Fe(II) concentration affected the mass-transfer of anammox and ferroferric oxide (Fe3O4) on anammox activity and granulation in anaerobic
granular. Besides, lower ΔNO-3-N/ΔNH+ 4 -N ratios was observed in re­
continuously stirred tank reactors (CSTR). Process Biochem. 49 (11), 1970–1978.
Han, X., Peng, S., Zhang, L., Lu, P., Zhang, D., 2020. The Co-occurrence of DNRA and
actors with Fe(II) addition indicated that the process of Fe(II)-dependent Anammox during the anaerobic degradation of benzene under denitrification.
nitrate reduction was induced in reactors. And the genera, OM27_clade Chemosphere 247, 125968.
and norank_f__Burkholderiaceae, might be the candidate organisms. This Henriques, I.D.S., Love, N.G., 2007. The role of extracellular polymeric substances in the
toxicity response of activated sludge bacteria to chemical toxins. Water Res. 41 (18),
reaction could reduce the nitrate produced by anammox and improve 4177–4185.
the TNRR of reactors. Hou, D., Zhang, P., Wei, D., Zhang, J., Yan, B., Cao, L., Zhou, Y., Luo, L., 2020.
Simultaneous removal of iron and manganese from acid mine drainage by
acclimated bacteria. J. Hazard Mater. 396, 122631.
Declaration of competing interest Jetten, M.S.M., Niftrik, L.V., Strous, M., Kartal, B., Keltjens, J.T., Op Den Camp, H.J.M.,
2009. Biochemistry and molecular biology of anammox bacteria. Crit. Rev. Biochem.
Mol. Biol. 44 (2–3), 65–84.
The authors declare that they have no known competing financial Jia, F.X., Yang, Q., Liu, X.H., Li, X.Y., Li, B.K., Zhang, L., Peng, Y.Z., 2017. Stratification
interests or personal relationships that could have appeared to influence of extracellular polymeric substances (EPS) for aggregated anammox
microorganisms. Environ. Sci. Techno. 51 (6), 3260–3268.
the work reported in this paper.
Kuenen, J.G., 2008. Anammox bacteria: from discovery to application. Nat. Rev.
Microbiol. 6 (4), 320–326.
Acknowledgements Kulichevskaya, I.S., Suzina, N.E., Rijpstra, W.I.C., Damsté, J.S.S., Dedysh, S.N., 2014.
Paludibaculum fermentans gen. nov., sp. nov., a facultative anaerobe capable of
dissimilatory iron reduction from subdivision 3 of the Acidobacteria. Int. J. Syst.
This work was supported by Beijing Outstanding Young Scientist Evol. Microbiol. 64 (Pt_8), 2857–2864.
Project (grant numbers: C19H100010) and National Natural Science Li, X., Huang, Y., Wu, C., Wang, M.-k., Yuan, Y., 2014. Effect of Fe2+ and Fe3+ on the
activity of ANAMMOX. Environ. Sci. 35 (11), 4224–4229.
Foundation of China (grant numbers: 51908029).
Li, Z., Peng, Y., Gao, H., 2020a. Enhanced long-term advanced denitrogenation from
nitrate wastewater by anammox consortia: dissimilatory nitrate reduction to
Appendix A. Supplementary data ammonium (DNRA) coupling with anammox in an upflow biofilter reactor equipped
with EDTA-2Na/Fe(II) ratio and pH control. Bioresour. Technol. 305.
Li, Z., Peng, Y., Gao, H., 2020b. A novel strategy for accelerating the recovery of a Fe(II)-
Supplementary data to this article can be found online at https://doi. inhibited anammox reactor by intermittent addition of betaine: performance,
org/10.1016/j.chemosphere.2021.131322. kinetics and statistical analysis. Chemosphere 251, 126362.
Liu, Y., Ni, B.-J., 2015. Appropriate Fe (II) addition significantly enhances anaerobic
ammonium oxidation (anammox) activity through improving the bacterial growth
Author contribution rate. Sci. Rep. 5 (1).
Mak, C.Y., Lin, J.G., Chen, W.H., Ng, C.A., Bashir, M.J.K., 2019. The short- and long-term
inhibitory effects of Fe (II) on anaerobic ammonium oxidizing (anammox) process.
Yao Chen: Conceptualization, Formal analysis, Investigation, Visu­ Water Sci. Technol. 79 (10), 1860–1867.
alization, Writing – original draft; Fangxu Jia: Conceptualization, Melton, E.D., Swanner, E.D., Behrens, S., Schmidt, C., Kappler, A., 2014. The interplay of
Writing – review & editing, Supervision, Resources, Funding acquisition; microbially mediated and abiotic reactions in the biogeochemical Fe cycle. Nat. Rev.
Microbiol. 12 (12), 797–808.
Yingjie Liu: Formal analysis, Investigation; Wanrou Yu: Investigation; Mishra, P., Burman, I., Sinha, A., 2020. Performance Enhancement and Optimization of
Weiwei Cai: Writing – review & editing; Xiaofan Zhang: Validation; the Anammox Process with the Addition of Iron. Environmental Technology.
Haodong He: Visualization; Hong Yao: Resources, Funding acquisition, Op Den Camp, H.J.M., Kartal, B., Guven, D., Van Niftrik, L.A.M.P., Haaijer, S.C.M., Van
Der Star, W.R.L., Van De Pas-Schoonen, K.T., Cabezas, A., Ying, Z., Schmid, M.C.,
Writing – review & editing
Kuypers, M.M.M., Van De Vossenberg, J., Harhangi, H.R., Picioreanu, C., Van
Loosdrecht, M.C.M., Kuenen, J.G., Strous, M., Jetten, M.S.M., 2006. Global impact
References and application of the anaerobic ammonium-oxidizing (anammox) bacteria.
Biochem. Soc. Trans. 34 (1), 174–178.
Oshiki, M., Ishii, S., Yoshida, K., Fujii, N., Ishiguro, M., Satoh, H., Okabe, S., 2013.
Badireddy, A.R., Chellam, S., Gassman, P.L., Engelhard, M.H., Lea, A.S., Rosso, K.M.,
Nitrate-dependent ferrous iron oxidation by anaerobic ammonium oxidation
2010. Role of extracellular polymeric substances in bioflocculation of activated
(Anammox) bacteria. Appl. Environ. Microbiol. 79 (13), 4087.
sludge microorganisms under glucose-controlled conditions. Water Res. 44 (15),
Park, W., Nam, Y.-K., Lee, M.-J., Kim, T.-H., 2009. Anaerobic ammonia-oxidation
4505–4516.
coupled with Fe reduction by an anaerobic culture from a piggery wastewater
Bi, Z., Huang, Y., Zhang, W., Song, G., 2020. Impacts of chosen parameters on Fe-
acclimated to NH4+/Fe3+ Medium. Biotechnol. Bioproc. Eng. 14 (5), 680–685.
dependent nitrate reduction in anammox consortia: performance and bioactivity.
Pulicharla, R., Das, R.K., Brar, S.K., Drogui, P., Sarma, S.J., Verma, M., Surampalli, R.Y.,
Water 12 (5), 1379.
Valero, J.R., 2015. Toxicity of chlortetracycline and its metal complexes to model
Bi, Z., Qiao, S., Zhou, J., Tang, X., Zhang, J., 2014. Fast start-up of Anammox process
microorganisms in wastewater sludge. Sci. Total Environ. 532, 669–675.
with appropriate ferrous iron concentration. Bioresour. Technol. 170, 506–512.
Qiao, S., Bi, Z., Zhou, J., Cheng, Y., Zhang, J., 2013. Long term effects of divalent ferrous
Bi, Z., Zhang, W., Song, G., Huang, Y., 2019. Iron-dependent nitrate reduction by
ion on the activity of anammox biomass. Bioresour. Technol. 142, 490–497.
anammox consortia in continuous-flow reactors: a novel prospective scheme for
Ren, L.-F., Lv, L., Zhang, J., Gao, B., Ni, S.-Q., Yang, N., Zhou, Q., Liu, X., 2016. Novel
autotrophic nitrogen removal. Ence Total Environ. 692.
zero-valent iron-assembled reactor for strengthening anammox performance under
Chen, H., Chen, Q.Q., Jiang, X.Y., Hu, H.Y., Shi, M.L., Jin, R.C., 2016. Insight into the
low temperature. Appl. Microbiol. Biotechnol. 100, 8711–8720.
short- and long-term effects of Cu(II) on denitrifying biogranules. J. Hazard Mater.
Ren, L.-F., Ni, S.-Q., Liu, C., Liang, S., Zhang, B., Kong, Q., Guo, N., 2015. Effect of zero-
304, 448–456.
valent iron on the start-up performance of anaerobic ammonium oxidation
Chen, H., Yu, J.J., Jia, X.Y., Jin, R.C., 2014. Enhancement of anammox performance by
(anammox) process. Environ. Sci. Pollut. Res. 22, 2925–2934.
Cu(II), Ni(II) and Fe(III) supplementation. Chemosphere 117, 610–616.
Shu, D., He, Y., Yue, H., Yang, S., 2016. Effects of Fe(II) on microbial communities,
Cui, J., Du, J., Tian, H., Chan, T., Jing, C., 2018. Rethinking anaerobic As(III) oxidation
nitrogen transformation pathways and iron cycling in the anammox process:
in filters: effect of indigenous nitrate respirers. Chemosphere 196, 223–230.
kinetics, quantitative molecular mechanism and metagenomic analysis. RSC Adv. 6
De Mey, M., Lequeux, G., Maertens, J., De Maeseneire, S., Soetaert, W., Vandamme, E.,
(72), 68005–68016.
2006. Comparison of DNA and RNA quantification methods suitable for parameter

7
Y. Chen et al. Chemosphere 285 (2021) 131322

Takao, I., Koji, M., Yoshihito, U., Tatsunori, N., Shigeaki, H., Ken-Ichiro, S., 2010. Yang, W.H., Weber, K.A., Silver, W.L., 2012. Nitrogen loss from soil through anaerobic
Ignavibacterium album gen. nov., sp. nov., a moderately thermophilic anaerobic ammonium oxidation coupled to iron reduction. Nat. Geosci. 5 (8), 538–541.
bacterium isolated from microbial mats at a terrestrial hot spring and proposal of Yang, Y., Xiao, C., Lu, J., Zhang, Y., 2020. Fe(III)/Fe(II) forwarding a new anammox-like
Ignavibacteria classis nov., for a novel lineage at the periphery of green sulfur process to remove high-concentration ammonium using nitrate as terminal electron
bacteria. Int. J. Syst. Evol. Microbiol. 60 (6), 1376–1382. acceptor. Water Res. 172, 115528.
van Niftrik, L., Geerts, W.J.C., van Donselaar, E.G., Humbel, B.M., Yakushevska, A., Yuan, M., Wang, D., 2019. Effect of metal ions on nitrogen removal efficiency in
Verkleij, A.J., Jetten, M.S.M., Strous, M., 2008. Combined structural and chemical anammox sludge. Environ. Pollut. Control 41 (5), 515–525.
analysis of the anammoxosome: a membrane-bounded intracytoplasmic Zhang, J., Zhang, Y., Li, Y., Zhang, L., Qiao, S., Yang, F., Quan, X., 2012. Enhancement of
compartment in anammox bacteria. J. Struct. Biol. 161 (3), 401–410. nitrogen removal in a novel anammox reactor packed with Fe electrode. Bioresour.
Wang, R., Yang, C., Zhang, M., Xu, S.-Y., Dai, C.-L., Liang, L.-Y., Zhao, H.-P., Zheng, P., Technol. 114, 102–108.
2017. Chemoautotrophic denitrification based on ferrous iron oxidation: reactor Zhang, S., Zhang, L., Yao, H., Rong, H., Li, S., 2021. Responses of anammox process to
performance and sludge characteristics. Chem. Eng. J. 313, 693–701. elevated Fe(III) stress: reactor performance, microbial community and functional
Xie, F., Ma, X., Zhao, B., Cui, Y., Zhang, X., Yue, X., 2020. Promoting the nitrogen genes. J. Hazard Mater. 414, 125051.
removal of anammox process by Fe-C micro-electrolysis. Bioresour. Technol. 297, Zhang, Z.-Z., Xu, J.-J., Shi, Z.-J., Bai, Y.-H., Cheng, Y.-F., Hu, H.-Y., Jin, R.-C., 2017.
122429. Unraveling the impact of nanoscale zero-valent iron on the nitrogen removal
Yan, Y., Wang, Y., Wang, W., Zhou, S., Wang, J., Guo, J., 2019. Comparison of short-term performance and microbial community of anammox sludge. Bioresour. Technol.
dosing ferrous ion and nanoscale zero-valent iron for rapid recovery of anammox (243), 883–892.
activity from dissolved oxygen inhibition. Water Res. 153, 284–294.

You might also like