Nitrogen and Oxygen Isotope Signatures of Nitrogen Compounds during Anammox in the Laboratory and a Wastewater Treatment Plant.

Isotopic fractionation factors against 15N and 18O during anammox (anaerobic ammonia oxidization by nitrite) are critical for evaluating the importance of this process in natural environments. We performed batch incubation experiments with an anammox-dominated biomass to investigate nitrogen (N) and oxygen (O) isotopic fractionation factors during anammox and also examined apparent isotope fractionation factors during anammox in an actual wastewater treatment plant. We conducted one incubation experiment with high δ18O of water to investigate the effects of water δ18O. The N isotopic fractionation factors estimated from incubation experiments and the wastewater treatment plant were similar to previous values. We also found that the N isotopic effect (15εNXR of -77.8 to -65.9‰ and 15ΔNXR of -31.3 to -30.4‰) and possibly O isotopic effect (18εNXR of -20.6‰) for anaerobic nitrite oxidation to nitrate were inverse. We applied the estimated isotopic fractionation factors to the ordinary differential equation model to clarify whether anammox induces deviations in the δ18O vs δ15N of nitrate from a linear trajectory of 1, similar to heterotrophic denitrification. Although this deviation has been attributed to nitrite oxidation, the O isotopic fractionation factor for anammox is crucial for obtaining a more detailed understanding of the mechanisms controlling this deviation. In our model, anammox induced the trajectory of the δ18O vs δ15N of nitrate during denitrification to less than one, which strongly indicates that this deviation is evidence of nitrite oxidation by anammox under denitrifying conditions.

Anammox (anaerobic ammonia oxidization by nitrite) has been intensively investigated since the discovery of its importance as a N removal process in natural ecosystems (Dalsgaard ; Kuypers ). The rates of anammox and denitrification are frequently similar (Kuypers ; Hamersley ; Lam ). The detection of anammox in ecosystems is key for further investigations on the relative (quantitative) importance of anammox and denitrification to N losses. Molecular techniques, such as qPCR (Hamasaki ) and biomarker analyses (ladderane lipids; Jaeschke ), have generally been applied to detect anammox bacteria, followed by 15N tracer experiments (Amano ) to assess anammox activities (rates) in the laboratory. Although this approach is promising, it only estimates potential anammox rates. Thus, it is crucial to develop screening techniques that estimate anammox in the field.

The naturally occurring stable isotope ratios of N (15N / 14N, expressed as δ15N) and O (18O / 16O, expressed as δ18O) are useful tracers for investigating the origins, transport, and biogeochemical processes of dissolved inorganic N (DIN), such as nitrate (NO3–), nitrite (NO2–), and ammonium (NH4+), in ecosystems (Casciotti, 2016a, 2016b; Denk ; Thuan ). Regarding the use of δ15N and δ18O to interpret the complex dynamics of DIN, it is essential to apply the isotopic fractionation factors of specific reactions of DIN production and consumption. Previous studies on heterotrophic denitrification estimated 15N (Blackmer and Bremner, 1977; Chien ; Mariotti , 1982; Bryan ; Kawanishi ; Barford ) and 18O fractionation factors (Böttcher ; Granger ; Kritee ; Frey ; Martin and Casciotti, 2016; Osaka ; Wang ). Detailed information on isotopic fractionation during denitrification has encouraged the use of the δ15N and δ18O of NO3– in investigations on the occurrence and magnitude of denitrification in many ecosystems (Mariotti ; Koba ; Ostrom ; Lehmann ; Sigman ; Houlton ; Houlton and Bai, 2009; Miyajima ; Fang ; Lennon and Houlton, 2017).

In contrast to denitrification, few studies have used the δ15N and δ18O of DIN to examine anammox (Prokopenko ; Prokopenko ; Wenk ; Dähnke and Thamdrup, 2016), and only two studies (Brunner ; Kobayashi ) have reported isotopic fractionation factors for the anammox reaction. These factors must be known in order to estimate the importance of anammox in a studied ecosystem with the δ15N and δ18O of DIN. Brunner estimated a large inverse N isotope effect (i.e., the heavier isotope, 15N, reacts faster than the lighter isotope, 14N) during NO3– production (by anaerobic nitrite oxidation) in anammox as well as a large normal (i.e., the lighter 14N reacts faster than the heavier 15N) isotope effect for ammonium oxidation, which was confirmed in a later study by Kobayashi . However, they only reported the combined 18O isotope fractionation factors and do not provide the isotope fractionation factors for the NO2– oxidation and its relevant oxygen atom incorporation from water, involved in the combined factors.

Studies on 15N and 18O fractionation factors revealed that NO3– consumption (the assimilatory and dissimilatory reduction of NO3–) generally induced a 1:1 increase in the δ18O and δ15N of NO3– (Granger ; Granger ; Karsh ; Rohde ; Osaka ). This finding prompted the use of the δ18O and δ15N of NO3– to detect NO3– consumption in the actual ecosystem as well as investigations on NO3– isotope anomalies, specifically isotopic deviations from a slope of 1 in the δ18O vs δ15N of NO3– (Δ[15, 18]; Sigman ), in order to deepen insights into NO3– dynamics (Casciotti ; Casciotti and Buchwald, 2012; Bourbonnais ; Peters ; White ). Granger and Wankel (2016) proposed that widely observed deviations in the δ18O vs δ15N of NO3– from the denitrification slope of 1 in freshwater systems (Sigman ; Granger ; Kritee ) must result from concurrent NO3– production (nitrification or anammox) in the denitrifying system that has been largely overlooked. However, they lacked information on the 18O fractionation factor for anammox, and assumed that the 18O fractionation factor during NO3– production by anammox was similar to that for aerobic nitrite oxidation to NO3– by nitrifiers (nitrification). Thus, it is essential to investigate the 15N and 18O fractionation factors during anammox not only for the better use of the δ18O and δ15N of NO3– in anammox studies, but also to obtain a more detailed understanding of 15N and 18O fractionation.

We herein report unique data on 18O fractionation factors during anammox. We calculated apparent 15N and 18O fractionation factors with data collected from a wastewater treatment plant (WWTP) at which anammox reactors were installed at the final stage of treatment (Isaka ). We also performed anaerobic laboratory incubations with an anammox-dominated biomass to obtain more information on isotopic fractionation during anammox. We conducted one incubation experiment with high δ18O of water to investigate the effects of water δ18O. We then simulated system behavior with the observed isotopic fractionation factors to establish whether deviations in the δ18O vs δ15N of NO3– from the denitrification slope of 1 may be used to detect anammox activity.

Materials and Methods

Full-scale anammox wastewater treatment plant

Influent and treated water in the full-scale anammox wastewater treatment plant (Isaka ) were sampled three times (28th April, 7th and 12th May 2015). Detailed information on water chemistry and plant performance have been reported by Isaka . The anammox plant consists of a denitrifier reactor (DN), biochemical oxygen demand (BOD) oxidation reactor (BD), nitrite-nitrification reactor (NT), and anammox reactor (ANX). We collected water samples from each reactor (Fig. S1). The wastewater introduced into this anammox plant was effluent from an ammonia plant, which was pure water containing mainly NH4+ and methanol. Average NH4+ and total organic concentrations were 658 and 37‍ ‍mg L–1, respectively (Isaka ). Solutions were sampled from these reactors (Fig. S1) to measure the concentrations and isotope ratios of DIN. Samples (10‍ ‍mL each) were immediately filtered through 0.45-μm disk filters (25CS045AN; ADVANTEC Toyo Kaisha) and then collected in plastic centrifuge tubes. Samples were frozen until further measurements.

Biomass incubation experiments

We performed batch incubations with anammox bacteria. Details on a small-scale anammox reactor with activated sludge, including start-up information, maintenance, performance, input solutions, and microbial communities of the reactor, are provided in Supplemental Information (SI Text 1.1).

In the first experiment (Experiment A), the biomass in the reactor was sampled and incubated with the media used for the reactor, while the sampled biomass was re-suspended in fresh, chemically defined media in the second (Experiment B) and third (Experiment C) experiments. Difficulties were associated with performing incubation experiments with the anammox biomass for isotopic measurements, and, thus, we employed slightly different settings and operations to facilitate constant and active anammox reactions. In each experiment, 15‍ ‍mL of the biomass suspension in media solution was filtered with filter paper (Reeve Angel, Whatman) and differences in filter weights before and after filtration were used to calculate the suspended solid (S.S.) concentration after the filter had been oven-dried (at 105°C).

Experiment A

The biofilm and incubation media solution (500‍ ‍mL in total) were sampled from the incubation membrane in the anammox reactor (SI Text 1.1). The incubation was performed anaerobically in the glovebox at room temperature (25–30°C) after the purging of media by N2 gas to remove dissolved oxygen (DO). pH, DO, and the concentrations of NH4+ and NO2– were regularly measured to confirm the anammox activity of the biofilm, and pH (8.0) was maintained by adding KH2PO4 and Na2HPO4·12H2O solution. After the addition of NaNO2, (NH4)2SO4, and NaHCO3, we started the incubation and sampled 10‍ ‍mL of media. Sampled media were filtered with a 0.20-μm syringe filter and then split into three; one for NO3– followed by the removal of NO2– (Granger and Sigman, 2009), another one for NO2– with high pH by the addition of 2M NaOH solution to prevent oxygen atom exchange between NO water (Bourbonnais ), and one for NH4+ with low pH by the addition of 4.8 M H2SO4 to prevent NH4+ from volatilizing. These subsamples were frozen (–30°C) until further analyses.

Experiment B

The granule biomass that accumulated at the bottom of the anammox reactor was sampled. Granules were rinsed anaerobically with new media (N2 purged) in the glovebox. Media consisted of NaHCO3 502‍ ‍mg L–1; MgSO4·7H2O 603‍ ‍mg L–1; CaCl2 180.5‍ ‍mg L–1; KH2PO4 169‍ ‍mg L–1; Na2HPO4·12H2O 282‍ ‍mg L–1; trace elements of solution I (containing EDTA 6.369‍ ‍g L–1; FeSO4·7H2O 9.14‍ ‍g L–1) 0.5‍ ‍mL and solution II (containing EDTA 19.106‍ ‍g L–1; ZnSO4·7H2O 0.43‍ ‍g L–1; CoCl2·6H2O 0.24‍ ‍g L–1; MnCl2·4H2O 0.99‍ ‍g L–1; CuSO4·5H2O 0.25‍ ‍g L–1; Na2MoO4·2H2O 0.22‍ ‍g L–1; NiCl2·6H2O 0.19‍ ‍g L–1; Na2SeO4·10H2O 0.21‍ ‍g L–1; H3BO3 0.014‍ ‍g L–1) 0.5‍ ‍mL. We added NaNO2 and (NH4)2SO4 to media (500‍ ‍mL) with the anammox granules and then started the incubation at room temperature (25–30°C). We monitored pH (7.9 to 8.8) and NO2– to assess the progress of anammox. Sampling was performed as described in Experiment A.

Experiment C

We incubated the biofilm collected from the incubation membrane in the anammox reactor with the same media used in Experiment B; however, the δ18O of water (δ18OH2O) was markedly higher (229‰) than that in Experiments A and B (–8‰). This “heavy” water was prepared by mixing 18O-labeled water (10% atom 18O) with Milli-Q water. During the incubation, media with biofilms were shaken at a constant temperature (30°C) and continuously purged with a gas mixture (95% Ar + 5% CO2) to maintain low DO levels. pH ranged between 7.1 and 7.5 and the monitoring and sampling scheme was identical to Experiment B

Chemical analysis

DIN concentrations in water samples from the anammox plant and incubation experiments were measured using colorimetric methods with an autoanalyzer (Quatro, BL-Tec) (Thuan ) after appropriate dilutions. In Experiment A, NH4+ concentrations were measured during the incubation by the o-phthaldialdehyde (OPA) method (Holmes ). The DO and pH of the incubation media were monitored during the incubation with a DO meter (HQ30d; Hach) and pH meter (D-71; Horiba).

δ15N and δ18O values were assessed by GC-IRMS (Sercon 20–22 with Cryoprep) (Thuan ) with the denitrifier method (Sigman ; Casciotti ) for NO3– (δ15NNO3– and δ18ONO3–, respectively) with USGS 32, 34, 35, and IAEA-2 as standards, and with the azide method (McIlvin and Altabet, 2005) for NO2– (δ15NNO2– and δ18ONO2–, respectively) with TUAT-NO2-1 to TUAT-NO2-5 (Thuan ) calibrated against N-23, N-7373, and N-10219 (Casciotti ) as the standards. Analytical precision (expressed as the standard deviation of repeatedly measured samples) was ±0.2‰ for δ15NNO2– and δ15NNO3–, and ±0.5‰ for δ18ONO2– and δ18ONO3–. The δ15N values of NH4+ (δ15NNH4+) were evaluated using GC-IRMS with the denitrifier method after the conversion of NH4+ to NO3– by persulfate oxidation (Koba ; Thuan ) with USGS 25, 26, and IAEA-N-2 as the standards. Analytical precision was ±0.5‰ for the δ15N of NH4+. Water with high δ18O (229‰; measured by GC-IRMS with the modified azide method; McIlvin and Casciotti, 2006; Thuan ) from Experiment C was used to prepare NO3– and NO2– isotope standards for NO3– and NO2– measurements in order to correct for the effects of oxygen atom incorporation during the analysis. δ15N and δ18O are expressed as (R_SampleN/R_Nitrogen)–1 and (R_SampleO/R_Oxygen)–1 where R_SampleN and R_SampleO are [15N/14N] and [18O/16O] of the sample, respectively, R_Nitrogen is [15N/14N] of atmospheric N2 and R_Oxygen is [18O/16O] of Vienna Standard Mean Ocean Water (Table S1).

Calculation of apparent isotopic fractionation factors for the anammox plant

Apparent isotopic fractionation factors regarding anammox in the anammox plant were calculated as described by Kobayashi based on steady-state, open-system isotope systematics reported by Fry (2006).

Apparent N isotope effects of the ammonium oxidation to N2, and nitrite reduction and oxidation for the anammox plant

The ammonium oxidation to N2 by NO2– has isotope fractionation defined as 15ΔAMX. The δ15N of influx NH4+ (δ15NNH4+_NT), residual NH4+ (δ15NNH4+_ANX), and the fraction of NH4+ reacting (f) in ANX reactor (Fig. S1) at a steady state are used to estimate 15ΔAMX (Fry, 2006; Kobayashi ):

15ΔAMX = (δ15NNH4+_ANX – δ15NNH4+_NT) / f --- eq. (1)

f = ([NH4+]NT – [NH4+]ANX) / [NH4+]NT

where [NH4+]NT and [NH4+]ANX are the NH4+ concentrations in NT and ANX reactors, respectively.

The δ15N of NO2–, NO3–, and N2 in ANX reactor (δ15NNO2–_ANX, δ15NNO3–_ANX, and δ15NN2_ANX, respectively) at the steady state are given as follows (Fry, 2006; Kobayashi ):

15ΔAMXNIR = δ15NNO2–_ANX – δ15NN2_ANX --- eq. (2)

15ΔNXR = δ15NNO2–_ANX – δ15NNO3–_ANX --- eq. (3)

where 15N fractionation for nitrite reduction in anammox and nitrite oxidation are 15ΔAMXNIR and 15ΔNXR, respectively.

δ15NNO2–_NT is defined as:

δ15NNO2–_NT = (1 – a – b) × δ15NN2_ANX + a × δ15NNO2–_ANX + b × δ15NNO3–_ANX --- eq. (4)

with

a = [NO2–]ANX / ([NO2 – ]NT – [NO2–]ANX)

b = ([NO3–]ANX – [NO3–]NT) / ([NO2–]NT – [NO2–]ANX)

where δ15NNO2–_NT is δ15NNO2– in NT reactor and the concentrations of NO2– and NO3– in NT and ANX reactors are [NO2–]NT, [NO2–]ANX, [NO3–]NT, and [NO3–]ANX, respectively.

The combination of eqs. (2) and (4) gives

δ15NNO2–_NT = (1 – a – b) × δ15NN2_ANX + a × δ15NNO2–_ANX + b × δ15NNO3–_ANX

= (1 – a – b) × (δ15NNO2–_ANX – 15ΔAMXNIR) + a × δ15NNO2–_ANX + b × δ15NNO3–_ANX

= δ15NNO2–_ANX – 15ΔAMXNIR – a × δ15NNO2–_ANX+ a × 15ΔAMXNIR – b × δ15NNO2–_ANX + b × 15ΔAMXNIR+ a × δ15NNO2–_ANX + b × δ15NNO3–_ANX

= b × (δ15NNO3–_ANX – δ15NNO2–_ANX) + δ15NNO2–_ANX+ 15ΔAMXNIR × (a + b – 1)

= – (

15ΔAMXNIR = [δ15NNO2–_ANX – δ15NNO2–_NT – b × 15ΔNXR] / (a + b – 1) --- eq. (5)

Apparent combined O isotope effect of nitrite oxidation for the anammox plant

To calculate 18O fractionation during nitrite oxidation to nitrate, we followed the approach described by Kobayashi to calculate combined isotope fractionation (18EAMXcombined) because of the lack of detailed information on isotopic fractionation for the nitrite oxidation and oxygen atom incorporation during nitrite oxidation. Thus, we calculated 18EAMXcombined as follows:

18EAMXcombined = 2/3 δ18ONO2–_ANX + 1/3 δ18OH2O – δ18ONO3–_ANX --- eq. (6)

where δ18ONO2–_ANX, δ18OH2O, and δ18ONO3–_ANX are the 18O ratios of NO2–, water, and NO3– in ANX reactor, respectively.

Calculation of isotopic fractionation factors for incubations and a simulation with the dynamic model (the anammox model)

We developed an ordinary differential equation model as described by Casciotti and Buchwald (2012), Granger and Wankel (2016), and He and Bao (2019). We prepared the model (the anammox model) to calculate the isotopic fractionation factors for Experiments A, B, and C. The N transformations and associated isotopic fractionation (Fig. 1) were implemented in the anammox model with Berkeley Madonna (BM) software (Macey ), with a 4th-order Runge–Kutta method for integration. We initially used the curve-fitting function in BM software (least squares fitting) to calculate the rate constant of the ammonium oxidation based on concentration data in each experiment. Isotopic fractionation factors and the exchange rate of oxygen atoms between water and NO2– were then estimated from isotopic data.

Fig. 1.

Schematic of the anammox and denitrification system. Dotted arrows indicate denitrification processes that were not included in the anammox model.

Fluxes regarding the anammox process (Fig. 1) are defined as

AMX = AMXNIR = kAMO14N × [14NH4+] --- eq. (7)

NXR = AMXNIR × (x / (1 – x)) --- eq. (8)

where AMX, NXR, and AMXNIR are the (14N) fluxes of ammonium oxidation, nitrite oxidation, and reduction by anammox (Fig. 1), kAMO14N is the rate constant for AMX, and x is a stoichiometric ratio (increase in [NO3–]/decrease in [NO2–]) (Brunner ). We omitted the two N transformation processes regarding denitrification (nitrate and nitrite reduction by denitrification, NAR, and DENNIR, respectively; Fig. 1) in the anammox model because of the small contributions of denitrifying bacteria to the total microbial community (Fig. S2) and the small contribution of denitrification of only 5–10% at most to the total N removal rate in this study (estimated by 15N tracer measurements; D. Ikeda, personal communications).

Regarding NO2–;

d/dt [14NO2–] = – NXR – AMXNIR --- eq. (9)

d/dt [15NO2–] = – (R_NitriteN × NXR / 15εNXR)– (R_NitriteN × AMXNIR /

d/dt [N16O2–] = – 2NXR – 2AMXNIR --- eq. (11)

d/dt [N16O18O–] = – (R_NitriteO × 2NXR / 18εNXR)– (R_NitriteO × 2AMXNIR / 18εAMXNIR)– N16O18O–exch_OUT + N16O18O–exch_IN

= – (R_NitriteO × 2NXR / 18εNXR)– (R_NitriteO × 2AMXNIR / 18εAMXNIR)– kexch × [N16O18O–] + kexch×R_WaterO / 18εEQ --- eq. (12)

where R_NitriteO, R_NitrateO, R_NitriteN, and R_NitrateN are the 18O/16O and 15N/14N of [NO2–] and [NO3–], respectively. R_WaterO is the [18O/16O] of H2O. 15εNXR and 15εAMXNIR are the 15N fractionation factors of NXR and AMXNIR. 18εNXR and 18εAMXNIR are the 18O fractionation factors of NXR and AMXNIR, respectively. 18εEQ, the 18O fractionation factor of the equilibration between NO2– and H2O, was set at 13‰ in the present study based on the incubation temperature and pH (Table S1; Buchwald and Casciotti, 2013). We applied kexch (rate coefficient for oxygen atom exchange), N16O18O–exch_OUT, and N16O18O–exch_IN (N16O18O– efflux and influx regarding the N16O18O– pool, respectively) as described by He and Bao (2019) to implement oxygen atom exchange rates between NO2– and H2O.

Regarding NO3–;

d/dt [14NO3–] = NXR --- eq. (13)

d/dt [15NO3–] = (R_NitriteN × NXR / 15εNXR) --- eq. (14)

d/dt [N16O3–] = 3 NXR --- eq. (15)

d/dt [N18O16O2–] = (R_NitriteO × 2 NXR / 18εNXR)+ (R_WaterO × NXR) / 18εH2ONXR) --- eq. (16)

where 18εH2ONXR (assigned as 10.0‰; Table S1; Buchwald and Casciotti, 2010; Casciotti and Buchwald, 2012) is the 18O fractionation factor for the incorporation of oxygen from H2O into NO3– during the NXR reaction (Fig. 1).

Regarding NH4+;

d/dt [14NH4+] = – AMX --- eq. (17)

d/dt [15NH4+] = – (R_AmmoniumN × AMX/ 15εAMX) --- eq. (18)

where R_AmmoniumN is the 15N / 14N of [NH4+] and 15εAMX is the N isotopic fractionation factor for NH4+ consumption by anammox (Fig. 1).

The approximate stoichiometry of the anammox process converting NO2– and NH4+ to N2 and NO3– is as follows (Brunner ):

1.3NO2– + 1NH4+ → 1N2 + 0.3NO3– + 2H2O --- eq. (19)

However, this stoichiometry between nitrite removal and nitrate production has been reported to vary (Brunner ). Thus, we estimated this stoichiometry (x) together with kAMO14N with concentration data, which provided the AMX, AMXNIR, and NXR fluxes used in the calculation above (eq. [7] and [8]; Table S1). After estimating x and kAMO14N, we estimated the kexch of oxygen atoms between H2O and NO2– (Table S1) using the curve-fitting functions for Experiments A and B. In Experiment C with high δ18OH2O, we performed another incubation without the anammox biofilm (Fig. S3) to measure kexch. At the same time, we estimated other isotopic fractionation factors (15εAMXNIR, 15εNXR, 15εAMX, 18εAMXNIR, and 18εNXR). We assigned the range from 0 to 60‰ (with 5 and 10‰ as the initial values for the curve-fitting function of BM software) to estimate isotopic fractionation factors. We considered this 60‰ range for the curve-fitting estimate to be reasonable because isotopic fractionation factors larger than 60‰ are rarely observed (Denk ). It is important to note that curve-fitting for Experiments B and C was not successfully achieved for 18εAMXNIR, resulting in extremely high or low estimated values (calculated 18εAMXNIR values were 0 and 60‰ for Experiments B and C, respectively. In addition, 18εNXR (calculated as –11.2 and –84.3‰ for Experiments B and C, respectively) and consequently 18EAMXcombined (calculated as –4.2 and –52.9‰ for Experiments B and C, respectively), were not all successfully estimated for Experiments B and C. Based on these uncertainties in parameter estimations, we did not report these calculated values for Experiments B and C; however, we speculate that these calculated parameter sets support 18εAMXNIR as normal and 18εNXR being inverse isotope fractionation, as discussed below for Experiment A. The curve-fitting function (“multiple-fit” in BM software) (Macey ) was applied with a tolerance of 1 × 10–6. BM codes for the anammox model for curve fittings with concentrations and isotopic data are provided in the Zenodo website (https://doi.org/10.5281/zenodo.3895346) and Table S2 showed the root mean square errors (RMSE) for concentrations and isotope values for the fitted model.

Simulation exercise for denitrification and anammox (the anammox-denitrification model)

We added the fluxes of denitrification (NAR and DENNIR; Fig. 3) to the anammox model in order for the anammox-denitrification model to simulate anammox and denitrification as follows:

Fig. 3.

Isotopic fractionation factors applied in the anammox-denitrification simulation model. These factors were from Experiment A or previous studies; (A) Buchwald and Casciotti, 2010; (B) Granger and Wankel, 2016; (C) Casciotti ; (D) Buchwald and Casciotti, 2013.

Regarding NO2–;

d/dt [14NO2–] = – NXR + NAR – DENNIR – AMXNIR --- eq. (20)

d/dt [15NO2–] = – (R_NitriteN × NXR / 15εNXR)+ (R_NitrateN × NAR / 1518εNAR)– (R_NitriteN × DENNIR / 15εDENNIR)– (R_NitriteN × AMXNIR / 15εAMXNIR) --- eq. (21)

d/dt [N16O2–] = – 2 NXR + 2 NAR – 2 DENNIR – 2 AMXNIR --- eq. (22)

where 1518εNAR (assigned as 15‰, Granger ; Table S1) is the N and O isotopic fractionation factor of NAR (i.e., 15εNAR = 18εNAR, Sigman ; Granger ; Granger ; Rohde ; Osaka ) and 15εDENNIR (assigned as 5‰, Granger and Wankel, 2016; Table S1) is the 15N fractionation factor of DENNIR.

In the case of no exchange of oxygen atoms between NO

d/dt [N16O18O–] = – (R_NitriteO × 2 NXR / 18εNXR)+ (R_NitrateO × 2 NAR / 1518εNAR) / 18εH2OBRNAR– (R_NitriteO × 2 DENNIR / 18εDENNIR)– (R_NitriteO × 2 AMXNIR / 18εAMXNIR) --- eq. (23a)

where 18εH2OBRNAR is the 18O fractionation factor for the “branching effect” (assigned as 25‰, Casciotti and McIlvin, 2007; Table S1) during NAR (Fig. 1).

In the case of full exchange between NO2– and H2O,

d/dt [N --- eq. (23b)

where + 18εEQ) and δ18ONO2– is always set to δ18ONO2–_EQ.

Regarding NO3–;

d/dt [14NO3–] = NXR – NAR --- eq. (24)

d/dt [15NO3–] = (R_NitriteN × NXR / 15εNXR) – (R_NitrateN × NAR / 1518εNAR) --- eq. (25)

d/dt [N18O16O2–] = 3 NXR – 3 NAR --- eq. (26)

d/dt [N18O16O2–] = (R_NitriteO × 2 NXR / 18εNXR)+ (R_WaterO × NXR / 18εH2ONXR)– (R_NitrateO × 3 NAR / 1518εNAR) --- eq. (27)

We applied the estimated isotopic fractionation factors from Experiment A (Table 2) together with the reported values for fractionation factors (Fig. 3) to simulate whether the stronger contribution of anammox to denitrification alters the slope of the δ18O vs δ15N of NO3– from the denitrification slope of 1 with or without oxygen atom exchange between H2O and NO2– in freshwater (δ18OH2O = –8‰) or seawater (0‰) environments. The BM code for the anammox and anammox-denitrification models is provided on the Zenodo website (https://doi.org/10.5281/zenodo.3895346).

Table 2.

Isotopic fractionation factors during anammox (‰)

Open system	Anammox Plant (This study)				Reported values
	20150428	20150507	20150512		Kobayashi et al., (2019)
15ΔAMXNIR	11.8	12.0	12.4		5.9~29.5
15ΔNXR	–30.4	–31.1	–31.3		–30.1~ –45.3
15ΔAMX	34.0	34.8	34.4		30.9~32.7
18EAMXcombined*	–3.8	–2.5	–3.2		–1.5~ –12.1
Closed system	Batch incubations (This study)				Reported values
	Experiment A	Experiment B	Experiment C		Brunner et al., (2013)
15εAMXNIR	13.7	21.8	15.6		16.0
15εNXR	–77.8	–65.9	–71.1		–31.1
15εAMX	32.5	25.4	19.3		23.5~29.1
18EAMXcombined*	–10.4	n.d.	n.d.		n.d.
18εAMXNIR**	3.1	n.d.	n.d.		n.d.
18εNXR**	–20.6	n.d.	n.d.		n.d.

*: 18εNXR × 2 / 3 + 18εH2ONXR / 3 (Kobayashi )

**: assuming 18εEQ = 1.013, 18εH2ONXR = 1.010 (Table S2)

Results and Discussion

Anammox plant data

Throughout the 17-day span of the three sampling times, the DIN concentrations and their isotopic signatures were stable (Table 1) for each reactor. Stoichiometries for the anammox process were calculated by changes in DIN concentrations between NT and ANX reactors (decreases in the concentrations of NO2– and NH4+ ΔNO2– / ΔNH4+, for NO2– consumption, and an increase in NO3– with a decrease in NO ranged 1.22 ~ 1.26 and ΔNO3– / ΔNH4+ was 0.14 ~ 0.15, both of which were within the range for anammox reactions (1.03 to 1.32 for ΔNO2– / ΔNH4+ and 0.14 to 0.35 for ΔNO3– / ΔNH4+) (Yao ). Isaka also reported that ΔNO2– / ΔNH4+ was 1.23 from this anammox plant, which indicates the appropriate performance of anammox in ANX reactor. The ammonium in the influent with a high concentration (44.5‍ ‍mM) and low δ15NNH4+ (–10.4‰) was gradually consumed in the reactors with normal isotopic fractionation (i.e., with increasing δ15N), resulting in a low concentration (1.9‍ ‍mM) with high δ15NNH4+ (50.2‰) at final ANX reactor (Table 1). Nitrite produced in DN, BD, and NT reactors and consumed in ANX reactor had low δ15NNO2– values (–38.8 to –21.6‰) and relatively stable δ18ONO2– values (3.3 to 4.7‰; Table 1). Nitrate was not produced before ANX reactor and δ15NNO3– (9.3‰) was higher than δ15NNO2– (–21.6‰), with no significant difference between δ18ONO3– and δ18ONO2– in ANX (Table 1). In comparisons with the isotopic data for other types of WWTP (Table 1), we found that the lower δ18ONO3– and higher δ15NNH4+ from the anammox plant was useful for tracking the fate of N derived from the anammox wastewater plant.

Table 1.

Average concentrations and isotopic compositions of DIN in the anammox plant and isotopic data from different types of WWTP

Reactor	[NH4+] (mM)	[NO2–] (mM)	[NO3–] (mM)	δ15NNH4+ (‰)	δ15NNO2– (‰)	δ18ONO2– (‰)	δ15NNO3– (‰)	δ18ONO3– (‰)
Influent	44.5 (0.1)	0 (0)	0 (0)	–10.4 (0.2)	n.d.	n.d.	n.d.	n.d.
DN	40.7 (0.2)	0.2 (0)	0 (0)	–8.4 (0.2)	–27.3 (0.4)	4.7 (0.2)	n.d.	n.d.
BD	32.8 (0.7)	7.7 (0.6)	0 (0)	0.6 (0.9)	–38.8 (0.8)	4.3 (0.2)	n.d.	n.d.
NT	18.7 (0.4)	21.2 (0.3)	0 (0)	19.2 (0.7)	–28.2 (0.4)	4.5 (0.1)	n.d.	n.d.
ANX	1.9 (0.5)	0.4 (0)	2.5 (0)	50.2 (1.4)	–21.6 (0.4)	3.3 (0.3)	9.3 (0.4)	2.7 (0.2)
WWTP type#									Reference
CAS				0 to 36			13 to 15	–1 to 0	Tumendelger et al., 2014
A2O							8.1*	–4.5*	Toyoda et al., 2011
Preliminary							11.5 (3.1)**	4.9 (4.2)**	Archana et al., 2016
Primary							14.8 (3.9)**	8.6 (3.4)**	Archana et al., 2016
CEPT							10.6 (4.9)**	–2.1 (3.6)**	Archana et al., 2016
Secondary							12.5 (4.3)**	3.8 (2.4)**	Archana et al., 2016
Tertiary							90.7 (83.9)**	87.7 (90.6)**	Archana et al., 2016

Means from three sampling times with standard errors (in parentheses) are shown. n.d.: not determined

* Data from the sampling point closest to the outlet to the river

** Means from several WWTP with standard deviations (in parentheses) are shown.

# CAS: Conventional activated sludge, A2O: Anaerobic-Anoxic-Oxic treatment, CEPT: Chemically Enhanced Primary Treatment

The calculated 15ΔAMX was large (34.0 ~ 34.8‰; Table 2), which was similar to the reported value for anammox (30.9 ~ 32.7‰; Kobayashi ) and to the isotope effect for aerobic ammonia oxidization (29.6 ± 4.9‰; Denk ). The two NO2– consumption pathways in the ANX reactor had different 15N fractionation; normal (positive), large (–30.4 ~ –31.3‰), which fell within reported values ( (Table 2). The present results confirmed an inverse 15N effect during anaerobic NO2– oxidation to NO3–, as previously reported (Brunner ; Kobayashi ) for aerobic NO2– oxidation to NO3– (Casciotti, 2009; Buchwald and Casciotti, 2010). Similarly, the small and negative apparent “combined” 18O fractionation for ammonium oxidization by NO also fell within the reported range of –1.5 to –‍12‰ ( (Table 2). The negative 18EAMXcombined values reported here and by Kobayashi during NO3– production in anammox agree with the inverse 18O fractionation for aerobic nitrite oxidation to NO3– (Casciotti, 2009; Buchwald and Casciotti, 2010).

Incubation experiments

In all experiments, [NH4+] and [NO2–] concurrently decreased as [NO3–] increased (Fig. 2a, b, and c). Averaged stoichiometries during anammox were 1.29, 1.51, and 1.48 for ΔNO2– / ΔNH4+ and 0.16, 0.17, and 0.21 for ΔNO3– / ΔNH4+ in Experiments A, B, and C, respectively (Fig. 2a, b, and c). These results were more consistent in their stoichiometry than previous findings with the same anammox bacterium (1.00 to 2.12 for ΔNO2– / ΔNH4+ and 0.10 to 0.37 for ΔNO3– / ΔNH4+; Ali ). The estimated values of x, kAMO14N, and kexch were shown in Table S1. The estimated x values (0.13 to 0.21; (0.15 to 0.48; Brunner ), and kexch values were negligible (Table S1). The anammox rates based on NH4+ consumption were 39.7, 61.7, and 12.4‍ ‍μM (g-S.S.)‍ ‍–1‍ ‍h–1 for Experiments A, B and C, respectively.

Fig. 2.

Concentrations and isotopic signatures of inorganic N in incubation experiments. The lines represent changes in the concentrations and isotopic signatures estimated by the curve-fitting of rate constants (for concentrations, upper panels) and 15N and 18O fractionation factors (for δ15N and δ18O values, middle and lower panels). The root mean square error (RMSE) for each fitting was shown in Table S1.

δ15NNH4+, δ15NNO2–, and δ15NNO3– increased as [NH4+] and [NO2–] decreased during anammox (Fig. 2d, e, and f). In contrast, δ18ONO2– and δ18ONO3– did not change in Experiment A (Fig. 2g), while δ18ONO3– increased by ~2‰ and δ18ONO2– decreased by ~3‰ in Experiment B (Fig. 2h). In Experiment C with the high δ18O of H2O (229‰), δ18ONO2– and δ18ONO3– rapidly increased (Fig. 2i). δ18ONO2– also rapidly increased in the negative control experiment without the biomass (Fig. S3). The isotope exchange between NO2– and NO3– needs to be taken into consideration (Brunner ) when the rapid and large changes in δ15N and δ18O at the beginning of the incubation are observed. Since we did not observe such a marked change in δ15N and δ18O, indicating isotope exchange (Fig. 2), we did not include isotope exchange between NO2– and NO3– in the present study.

15εAMX values were calculated as 32.5, 25.4, and 19.3‰ for Experiments A, B, and C, respectively (Table 2, Fig. 2d, e, and f). These 15εAMX values were similar to previously reported values (23.5 ~ 29.1‰) for Kuenenia stuttgartiensis in batch incubation experiments (Brunner ). 15εAMXNIR values were estimated to be 13.7, 21.8, and 15.6‰ (Table 2, Fig. 2d, e, and f), while 18εAMXNIR values were 3.1, 0 and 60.0‰ (Table 2, Fig. 2d, e, and f) for Experiments A, B and C, respectively. Although 18εAMXNIR values in Experiments B and C were not successfully measured (Table 2; see the Methods), estimated 15εAMXNIR values were consistent with the 15N values reported for NO2– reduction by Cu-NIR coded by the nirK gene (22 ± 2 and 2 ± 2‰) (Martin and Casciotti, 2016). The similarity in these values was attributed to the Cu-NIR of “Candidatus Jettenia” with the nirK gene (Hira ; Ali ), the dominant microbe in incubation experiments (Fig. S2).

In addition to normal isotopic fractionation, we estimated 15N and 18O fractionation factors during anaerobic nitrite oxidization to NO3– of –77.8‰ for 15εNXR and –20.6‰ for 18εNXR (Table 2, Fig. 2g, h, and i) in Experiment A. Although we also estimated 15N and 18O fractionation factors of –65.9 and –71.1‰ for 15εNXR and –11.2 and –84.3‰ for 18εNXR for Experiments B and C, respectively, 18εNXR values for these experiments were not precisely measured. The large inverse 15εNXR is consistent with the reported value with K. stuttgartiensis (–31.1 ± 3.9‰; Brunner ), as well as aerobic nitrite oxidation to NO3– by nitrite-oxidizing bacteria (–12.8 ± 1.5‰; Casciotti, 2009). Regarding oxygen, although only 18εNXR in Experiment A was successfully assessed, the estimated 18εNXR value was negative and close to the inverse 18O fractionation factors for aerobic nitrite oxidization to NO3– by nitrite-oxidizing bacteria (–10 to –‍1‰; Buchwald and Casciotti, 2010).

Simulation for denitrification and anammox

We developed an anammox-denitrification model with the estimated isotopic fractionation factors (from Experiment A; Table 2) and reported values (Fig. 3) to clarify whether anammox induces a deviation in δ18ONO3– vs δ15NNO3– from the denitrification slope of 1 (i.e., Δ(15, 18); defined as (δ15N – δ15Ninitial) – (18ε / 15ε)(δ18O – δ18Oinitial), where 18ε / 15ε is the ratio of isotopic fractionation for O and N during denitrification, respectively, and assigned as 1; see the inset in Fig. 4a; Sigman ). Each simulation was run until more than 25% of NO2– was consumed. In the case of denitrification in which AMX / NAR is equal to 0 (indicating no anammox), δ18ONO3– vs δ15NNO3– was set to show a slope of 1 (the dotted lines in Fig. 4a, d, c, and d). As reported in previous studies (Casciotti and Buchwald, 2012; Granger and Wankel, 2016; He and Bao, 2019), larger anammox rates (larger AMX / NAR) induced greater offsets (larger Δ[15, 18]) from the 1:1 relationship between δ18ONO3– and δ15NNO3– in all cases (Fig. 4a, b, c, and d). The effect of oxygen atom exchange was small, but obvious (Fig. 4c for freshwater and Fig. 4d for seawater) with a larger offset with the exchange. Although the present results revealed that Δ(15, 18) is a sensitive parameter for the occurrence of anammox, its usefulness diminishes with smaller 15εNXR values (–31.1‰, Table 2 and Fig. S4), indicating the sensitivity of Δ(15, 18) against 15εNXR (i.e., the stronger 15εNXR, the larger Δ [15, 18]). To elucidate the relationship between Δ(15, 18), AMX / NAR, and isotopic fractionation factors, we simulated the δ18ONO3– and δ15NNO3– trajectories along with the different AMX / NAR ratios and 15εAMXNXR (Fig. 5 with 15εNXR = –77.8‰ and Fig. S5 with –31.1‰). Similar to 15εNXR, stronger 15εAMXNXR resulted in larger Δ(15, 18); however, Δ(15, 18) depended on AMX / NAR, 15εAMXNXR, and 15εNXR (Fig. 5 and S5). Therefore, a simple comparison of Δ(15, 18) data does not permit quantitative estimations of AMX / NAR because Δ(15, 18) increases with NO2– and NO3– consumption whenever anammox is active (AMX / NAR > 0; Fig. 4 and 5) and Δ(15, 18) levels strongly depend on many parameters, including 15εAMXNXR and 15εNXR. Although more information on isotopic fractionation factors is needed for quantitative interpretations due to the sensitivity of Δ(15, 18), our simulation exercise revealed that Δ(15, 18), the offset from the 1:1 relationship between δ15N and δ18O, may be useful for detecting NXR (nitrite oxidation) in denitrifying systems in both freshwater and seawater.

Fig. 4.

Results from the anammox-denitrification model for variable ratios of anammox (AMX) and denitrification (NAR), and with or without oxygen atom exchange between water and NO2–. The simulation was run with 15εNXR = –77.8‰ (Table 2) until more than 25% of the initial NO2– pool was consumed; however, NO2– consumption in simulations with the same run times varied according to the different AMX / NAR ratios. The end point of each simulation run was not important, whereas the slope of each run was. The dotted line in each panel illustrated the denitrification slope (1:1) and the inset in Fig. 4a shows Δ(15, 18) in the δ15N and δ18O space.

Fig. 5.

Results of the anammox-denitrification model for variable ANX / NAR ratios with variable 15εAMXNIR values with the full oxygen atom exchange between water (freshwater with δ18OH2O = –8‰) and NO2–. The simulation was run with 15εNXR = –77.8‰ (Table 2) until more than 25% of the initial NO2– pool was consumed; however, NO2– consumption in simulations with the same run times varied according to the different AMX / NAR ratios. The end point of each simulation run was not important, whereas the slope of each run was crucial. The inset in Fig. 5c shows Δ(15, 18) in the δ15N and δ18O space.

Besides NXR, some chemolithoautotrophic (e.g., sulfide-dependent) denitrification with auxiliary Nap NO3– reductase may exhibit a 2:1 rather than 1:1 relationship between δ15N and δ18O, resulting in an offset from the 1:1 relationship (Frey ). Although this non-respiratory pathway (i.e., Nap NO3– reduction) is not considered to be a major environmental sink for NO3– (Granger and Wankel, 2016), and, thus, was not included in our models, it is worthwhile considering this autotrophic denitrification as a driver of the offset in a sulfide-rich environment in which anammox may be inhibited (Jensen ) and sulfide-dependent denitrification enhanced.

Conclusion

We estimated 15N and 18O fractionation factors during anammox. The inverse 15N effects for NXR (and possibly inverse O isotope effects) may induce an offset from the denitrification trajectory (1:1 relationship between δ15N and δ18O of NO3–, Δ[15, 18]). In practice, Δ(15, 18) may be evaluated with time-course samplings or short incubation studies to investigate the occurrence of anammox, similar to denitrification. This technique will be advantageous because of its potential in evaluations of the quantitative contribution in situ of anammox versus denitrification. Although the detection and quantification of functional genes in denitrification and anammox may be readily performed, difficulties are associated with detecting the in situ occurrence of denitrification and anammox. Although the isotopic fractionation factors used also need to be considered, Δ(15, 18) is a promising parameter to complement molecular data and the results from laboratory incubation experiments in the study of anammox.

Citation

Kotajima, S., Koba, K., Ikeda, D., Terada, A., Isaka, K., Nishina, K., et al. (2020) Nitrogen and Oxygen Isotope Signatures of Nitrogen Compounds during Anammox in the Laboratory and a Wastewater Treatment Plant. Microbes Environ 35: ME20031.

https://doi.org/10.1264/jsme2.ME20031

Supplementary Material