Biological denitrification in marine aquaculture systems: A multiple electron donor microcosm study

Bioresource Technology 263 (2018) 340–349

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Biological denitrification in marine aquaculture systems: A multiple electron donor microcosm study Qiaochong Hea,b, Dongqing Zhangb, Kevan Mainc, Chuanping Fenga, Sarina J. Ergasb,

T

⁎

a

School of Water Resources and Environment, China University of Geosciences (Beijing), Beijing 100083, China Department of Civil & Environmental Engineering, University of South Florida, 74202 E. Fowler Ave, ENB 118, Tampa, FL 33620, USA c Fisheries and Aquaculture, Mote Marine Laboratory, 1600 Ken Thompson Parkway, Sarasota, FL 34236, USA b

A R T I C LE I N FO

A B S T R A C T

Keywords: Fish waste solids Mixotrophic denitrification Saline wastewater Elemental sulfur Methanol Wood chips

There is a lack of information on denitrification of saline wastewaters, such as those from marine recirculating aquaculture systems (RAS), ion exchange brines and wastewater in areas where sea water is used for toilet flushing. In this study, side-by-side microcosms were used to compare methanol, fish waste (FW), wood chips, elemental sulfur (S0) and a combination of wood chips and sulfur for saline wastewater denitrification. The highest denitrification rate was obtained with methanol (23.4 g N/(m3·d)), followed by FW (4.5 g N/(m3·d)), S0 (3.5 g N/(m3·d)), eucalyptus mulch (2.6 g N/(m3·d)), and eucalyptus mulch with sulfur (2.2 g N/(m3·d)). Significant differences were observed in denitrification rate for different wood species (pine > oak ≫ eucalyptus) due to differences in readily biodegradable organic carbon released. A pine wood-sulfur heterotrophic-autotrophic denitrification (P-WSHAD) process provided a high denitrification rate (7.2–11.9 g N/ (m3·d)), with lower alkalinity consumption and sulfate generation than sulfur alone.

1. Introduction Marine recirculating aquaculture systems (RAS) have been developed to minimize land and water use and wastewater discharges caused by rapid expansion of the aquaculture industry (Martins et al., 2010; Christianson et al., 2015). In RAS, nitrification processes, such as moving bed bioreactors (MBBR), are used to transform fish-toxic total ammonia nitrogen (TAN) and nitrite (NO2−) to nitrate (NO3−) (van Rijn et al., 2006). However, high NO3− concentrations have a chronic detrimental effect on marine cultured fish species production, and concentrations less than 75 mg NO3−-N/L are recommended for fish health (Davidson et al., 2014). The most common method to control NO3− in RAS is through water exchanges, which consume large amounts of water, and result in discharges of NO3− polluted wastewater to the environment, leading to aquatic ecosystem deterioration (Martins et al., 2010). Biological denitrification is an effective solution for NO3− removal in marine RAS (van Rijn et al., 2006; Simard et al., 2015). The most common RAS denitrification systems are based on heterotrophic metabolism, in which easily biodegradable liquid carbon sources, such as methanol or ethanol, are used as electron donors (Tsukuda et al., 2015). However, careful dosing is required as NO2− accumulates when the organic carbon supply is insufficient, while organic substrates are

⁎

Corresponding author. E-mail address: [email protected] (S.J. Ergas).

https://doi.org/10.1016/j.biortech.2018.05.018 Received 29 March 2018; Received in revised form 3 May 2018; Accepted 4 May 2018 Available online 08 May 2018 0960-8524/ © 2018 Elsevier Ltd. All rights reserved.

carried over to the effluent when provided in excess of the amount required for denitrification (Hamlin et al., 2008). Denitrification using fish waste (FW) as an internal organic carbon source offers economic and environmental benefits, owing to the concurrent reduction of NO3− and the solid waste stream (Martins et al., 2010; Suhr et al., 2014; Tsukuda et al., 2015). Klas et al. (2006) evaluated distinct phases of NO3− removal in a denitrification reactor treating RAS water with FW, and reported that only 4% of the total chemical oxygen demand (COD) in FW was readily biodegradable, while 30% was slowly biodegradable requiring ≥5 days to be utilized. Wood chips (WC) have gained attention as a biofilter media and carbon source for stormwater and domestic wastewater denitrification applications (Saliling et al., 2007; Lopez-Ponnada et al., 2017). WC media delivered long-term NO3− removal (5–15 years), while requiring minimum maintenance (Robertson, 2010). In a recent review, LopezPonnada et al. (2017) reported total nitrogen (TN) removal efficiency was higher with softwood (75.2%) compared with hardwoods (63.0%). In contrast, Cameron and Schipper (2010) reported that mean NO3− removal rates were similar for hardwood (3.3–4.4 g N/(m3·d)) and softwood (3.0–4.9 g N/(m3·d)) and there was no difference in long-term performance. Elemental sulfur (S0), which is a non-toxic by-product of petroleum refining, is a low cost electron donor for autotrophic denitrification

Bioresource Technology 263 (2018) 340–349

Q. He et al.

(Christianson et al., 2015). Advantages of sulfur oxidizing denitrification (SOD) include elimination of carry-over of organic carbon and lower excess biomass production than heterotrophic denitrification (Christianson et al., 2015). A disadvantage of SOD is alkalinity consumption (4.57 mg CaCO3/mg NO3−-N; Batchelor and Lawrence 1978), necessitating addition of a pH buffer material such as oyster shells (Sengupta et al., 2007). In addition, sulfate (SO42−) production (7.54 mg SO42− per mg NO3−-N reduced; Batchelor and Lawrence 1978) could potentially negatively affect fish health in RAS; however, we were unable to find any research this topic. A mixotrophic process combining heterotrophic and SOD is a potential strategy to limit SO42− production (Sahinkaya and Kilic, 2014; Sahinkaya et al., 2011). In addition, alkalinity generated by heterotrophic denitrification (3.57 mg CaCO3/mg NO3−-N) can compensate for alkalinity consumption by SOD (Oh et al., 2001; RodriguezGonzalez, 2017). Krayzelova et al. (2014) reported a high NO3− removal efficiency (90%) with reduced SO42− production by including scrap tire chips in an SOD column. Li et al. (2016) evaluated the denitrification performance of wood-sulfur heterotrophic-autotrophic denitrification (WSHAD) microcosms and reported a higher denitrification rate (0.055 h−1 to 0.066 h−1) than SOD alone (0.010 h−1 to 0.013 h−1). Limited studies have been conducted for WSHAD in RAS, and the impact of different wood species on denitrification in marine systems has not been investigated. Different electron donors have been investigated in freshwater aquaculture (Hamlin et al., 2008), wastewater (Saliling et al., 2007), groundwater microcosms (Fowdar et al., 2015) and drinking water (Sahinkaya et al., 2011). However, no prior study has evaluated denitrification performance using different electron donors in side-by-side trials for treatment of saline wastewater. High salinity can affect denitrification performance by preventing microorganisms from maintaining their osmotic pressure balance, giving rise to bacterial plasmolysis (Lay et al., 2010). In this study, side-by-side denitrification microcosm experiments were set up to compare the denitrification capacity of different electron donors for fully nitrified marine RAS water. Specific objectives were to investigate: i) the effect of different electron donors (methanol, WC, FW, S0, and a mix of WC and S0) on NO3− removal from marine water; ii) the influence of different wood species (pine, eucalyptus and oak) on denitrification performance; and iii) the effect of wood species (pine and eucalyptus) and alkalinity addition (oyster shells) on WSHAD performance.

Table 1 Experimental phases and materials added in each microcosm. Phase

Microcosms

Electron donor

Inoculum

Phase I

Methanol FW WC SOD

0.336 mL methanol 200 mL fish waste 10 g eucalyptus mulch 10 g elemental sulfur + 4 g crushed oyster shells 5 g wood chips + 5 g elemental sulfur

Plastic carriers IMTA Sand IMTA Sand IMTA Sand

WSHAD

IMTA Sand

Phase II

P-WC E-WC O-WC

10 g pine wood chips 10 g eucalyptus wood chips 10 g oak wood chips

IMTA Sand IMTA Sand IMTA Sand

Phase III

P-WSHAD

5 g pine wood chips + 5 g elemental sulfur 5 g pine wood chips + 5 g elemental sulfur + 2 g oyster shell 5 g eucalyptus wood chips + 5 g elemental sulfur 5 g eucalyptus wood chips + 5 g elemental sulfur + 2 g oyster shell

Pilot RAS reactor Pilot RAS reactor Pilot RAS reactor Pilot RAS reactor

PO-WHSAD E-WSHAD EO-WSHAD

SOD SOD SOD SOD

2.2. Electron donors and inoculum A summary of the materials used in each experimental phase is provided in Table 1. Methanol (> 99.9%) was purchased from Fisher Science (Fisher Science, USA). Elemental sulfur pellets (4.0–6.0 mm) were obtained from Southern Ag in Palmetto, Florida. Crushed oyster shells, an alkalinity source for SOD, were obtained from Myco Supply (Pittsburgh, Pennsylvania) and sieved to a size of 1.0–2.0 mm. FW was collected from a 0.085 m3 drop filter (Aquaculture Systems Technologies, L.L.C, New Orleans, LA) for solids removal in a marine RAS containing marine broodstock fish (Centropomus undecimalis) at Mote Aquaculture Research Park (MAP; Rhody et al., 2014). Different WC species were chosen based on their local availability in Florida (United States) and prior performance for denitrification (Lopez-Ponnada et al., 2017). Natural eucalyptus mulch (100% Florida-Grown Eucalyptus) was used in Phase I. The mulch was obtained from Scotts Company LLC (Marysville, Ohio, USA); however, no additional wood species information for this eucalyptus was available. In Phases II and III eastern white pine (P-WC; Pinus strobus; soft wood), eucalyptus (E-WC; Eucalyptus camaldulensis; hardwood) and red oak (O-WC; Quercus rubra; hardwood), were obtained from a specialty lumber supplier in Tampa, Florida. To maintain uniformity, the WCs were cut into blocks of approximately 4–6 mm (L) × 4–6 mm (W) × 2–4 mm (D). Different sources of inoculum were used to have an appropriately acclimated microbial community for the different electron donors tested (Table 1). Plastic carriers (AMBTM media, EEC, Blue Bell, PA, USA) were obtained from a methanol-fed denitrification reactor in the MAP red drum RAS described above. Sand was collected from a partially submerged denitrification filter in a marine integrated multitrophic aquaculture (IMTA) system described by Boxman et al. (2015). Inoculum used in WSHAD microcosms was biomass from a SOD reactor in a pilot-scale marine RAS set up in the USF laboratory. Regardless of the inoculum source, all microcosms were inoculated with 500 ± 21 mg VSS/L (1422 ± 27 mg TSS/L) except for un-inoculated controls.

2. Materials and methods Three experimental phases were set up in this study (Table 1 and Supplementary material): 1) Phase I was a screening study (no duplicates) to compare denitrification performance of methanol, FW, WC, SOD and WSHAD; 2) Phase II investigated the influence of wood species on denitrification performance; 3) Phase III investigated the effect of different wood species on WSHAD performance and the effect of oyster shell on the WSHAD performance. 2.1. Synthetic marine RAS water Synthetic marine RAS water was prepared by adding 15 g/L Instant Ocean Sea Salt (Instant Ocean®), 0.607 g/L sodium nitrate (NaNO3) and 0.044 g/L potassium dihydrogen phosphate (KH2PO4) to tap water. This resulted in a solution with a salinity of 15 ppt, NO3− concentration of 97.2 ± 1.8 mg NO3−-N/L and phosphorus concentration of 10.0 ± 0.5 mg PO43−-P/L, which are typical values for land-based marine RAS (Boxman et al., 2015). According to the manufacture’s information, Instant Ocean contains major, minor and trace elements, and is free of NO3− and phosphate. The COD in the synthetic marine RAS water was 9 ± 2 mg/L.

2.3. Experimental setup and operation The experiments were carried out in the Environmental Engineering Laboratory at the University of South Florida (USF), Tampa. Microcosms were set up in 1 L glass bottles containing 800 mL of synthetic RAS water. To maintain anoxic conditions, bottles were purged with nitrogen gas for 5 min to remove oxygen after the addition of all materials (excluding methanol). For the methanol microcosm, 341

Bioresource Technology 263 (2018) 340–349

Q. He et al.

shell in SOD microcosms were based on Tong et al. (2018), while the weight of WC added to the heterotrophic microcosms was equal to that of S0 in the autotrophic microcosms. In the WSHAD microcosms, dosages of WC, S0 and oyster shell were half of the dosages in independent heterotrophic and SOD microcosms. Four control microcosms were set up in Phase I: 1) no inoculum and no electron donor to investigate abiotic removal mechanisms; 2) plastic carriers to investigate denitrification due to endogenous decay of biomass on the carriers; 3) IMTA sand inoculum to investigate denitrification due to endogenous decay of biomass on the sand inoculum; 4) oyster shell and IMTA sand inoculum to investigate denitrification due to organics present in the oyster shells. After a 20-day acclimation period, microcosm studies were performed in two cycles. Considering that methanol and FW were liquid carbon sources, methanol (0.336 mL) and FW (100 mL) were added into methanol and FW microcosms, respectively, while no additional materials were added to the other microcosms. At the beginning of Cycle 2, WC of 5 g, FW of 100 mL and WC of 2.5 g were separately replenished into WC, FW and WSHAD microcosms to provide more electron donor for denitrification. In Phase II, all denitrification microcosms, including P-WC, O-WC and E-WC, were carried out in duplicate. P-WC and O-WC were performed in three cycles, while E-WC was performed in two cycles. Five

methanol was added after N2 purging. Microcosm bottles were sealed with screw caps with two holes with cemented tubes: one tube inserted below the water level for liquid sample collection and the other tube connected the bottle head space to a 1L SKC (Eighty Four, PA) gas bag containing nitrogen gas to avoid air entering the microcosms during sampling (see Supplementary information for details). Bottles were incubated under static conditions at 23.5 ± 2.1 °C, a typical value for commercial marine RAS in Florida (Boxman et al., 2015). Prior to daily sampling, bottles were shaken manually and the biomass was allowed to settle for 30 mins. When the measured concentrations of NO3−-N and NO2−-N were near the method detection limit (MDL) or NO3− removal rates decreased by less than 0.5 mg NO3−-N/(L·d)), the suspended solids in the microcosms were allowed to settle, a portion of the supernatant was drawn off and replaced with fresh synthetic marine RAS water with 200 mg NO3−-N/L to obtain an initial NO3− concentration of approximately 100 mg NO3−-N/L at the beginning of the subsequent cycle. The methanol dose used in Phase I was based on the theoretical value of 2.47 mg methanol/mg NO3−-N for heterotrophic denitrification (Sahinkaya et al., 2011). FW was added to achieve a COD to NO3−N ratio of 3 based on the stoichiometric value of 2.86 required for complete denitrification (Henze et al., 2002). Dosages of S0 and oyster

Fig. 1. Changes in NO3−, NO2−, TAN, COD and SO42− concentrations in methanol, WC, FW, SOD and WSHAD microcosms. 342

Bioresource Technology 263 (2018) 340–349

Q. He et al.

Table 2 Phase I results obtained in microcosms using different elecctron donors. Microcosms

Cycle 1

Cycle 2 NO3

Methanol* WC FW SOD WSHAD

−

removal rate

NO3− removal rate

Alkalinity (mg CaCO3/L)

pH

ΔTP (mg P/L)

TN (mg N/L)

Average denitrification rate

Operation (days)

k (d−1)

R2

Operation (days)

k (d−1)

R2

Initial

Final

Initial

Final

FinalInitial

Final

g N/(m3·d)

5 28 22 28 28

0.912 0.033 0.166 0.104 0.046

0.9917 0.8935 0.9577 0.9802 0.9216

5 47 24 39 47

– 0.037 0.170 0.071 0.035

– 0.9719 0.9211 0.9738 0.9815

482 321 612 227 79

786 576 959 154 75

7.33 7.02 8.09 7.63 7.54

7.54 7.56 7.21 7.34 7.16

−2.1 −0.3 +18.0 −2.2 −1.2

0.5 3.2 25.8 1.6 4.2

23.4 2.6 4.5 3.5 2.2

Methanol*-the results for methanoll microcosm was obtained in its fouth cycle.

2.5. Data analysis

controls were simultaneously carried out: 1) an abiotic control without inoculum or electron donor; 2) an endogenous decay control containing 500 mg VSS/L inoculum with no WC addition; 3) three inactivated controls with synthetic wastewater without inoculum, in which each type of 10 g WC were dried at 105 °C overnight prior to adding it to the corresponding bottle. In Phase III, four microcosm types were set up in duplicate and each was performed in two cycles: P-WSHAD and E-WSHAD without oyster shells and PO-WSHAD and EO-WSHAD with oyster shells.

Changes in NO3−-N, NO2−-N, TAN, COD and SO42− concentrations were plotted over time to visualize the reaction dynamics and by-product production during denitrification. Denitrification rate was calculated according to Eq. (1):

Denitrification rate (g N/(m3∙d)) = (Cj−Ci)/(j−i) NO3−-N

(1) −

where Cj is the concentration on Day j and Ci is the NO3 -N concentration on Day i (j > i) (mg NO3−-N/L). The specific NO3− removal rate was calculated by dividing the denitrification rate by the VSS concentration. A first-order kinetic model was used to describe denitrification process kinetics. The NO3− removal rate constant (k) was estimated based on Tong et al. (2018):

2.4. Analytical methods Samples of supernatant (4 mL) were collected using syringes from each bottle for water quality analysis, including TAN, NO2−-N, NO3−N, COD, and SO42−. Alkalinity, pH, S2−, total nitrogen (TN) and total phosphorus (TP) were measured at the beginning and end of each cycle. Samples were filtered through 0.45 µm membrane filters for TAN, NO2−-N, NO3−-N, COD, TN, TP and SO42− analysis. TAN and oxidized nitrogen (NOx−-N) was measured using an Ammonia Analyzer (TL2800, Timberline Instrument, USA) (method detection limit (MDL): 0.05 mg/L TAN and 0.05 mg/L NOx−-N). NO2−-N was analyzed using a combination of Standard Methods 4500 (APHA, 2012) and Strickland and Parsons (1972) (MDL: 0.01 mg/L NO2−-N). NO3−-N concentrations were calculated by subtracting the NO2−-N concentration from the NOx−-N concentration. COD was measured using HACH method 8000 (3–150 mg/L) adapted from Standard Methods 5220D (APHA, 2012) with addition of 0.5 g of HgSO4 to each vial to eliminate chloride interference (MDL, 3.1 mg/L COD). Measurement of readily biodegradable COD (rbCOD) was based on the method of Mamais et al. (1993). TN and TP were measured using HACH method 10071 (0–25 mg/L) adapted from Standard Methods 4500C (APHA, 2012) and HACH method 10127 (1.0–100.0 mg/L PO43−) adapted from Standard Methods 4500B-C (APHA, 2012), respectively. S2− and SO42− were measured using the Methylene Blue Method (Standard Method 4500D) according to HACH method 8131 (5–800 µg/L S2−) and SulfaVer 4 Method (Standard Method 4500E) according to HACH method 10248 (2–7000 mg/L SO42−), respectively. An Orion 5 Star (Thermo Scientific Inc., Beverly, Massachusetts) meter with a calibrated probe was used to measure pH. Alkalinity was measured using Standard Methods 2320B (865 Dosimat plus and 827 pH Lab, Metrohm AG, Switzerland; MDL: 20 mg/L). To determine VSS on the plastic carriers, three carriers were added to 50 mL vials containing 30 mL deionized (DI) water and vortexed for 15 min using a Vortex-genie 2 (Scientific Industries, USA). After decanting the supernatant, the process was repeated and the VSS was measured in the collected liquid. To determine VSS per g sand, 50 g of wet sand was used for VSS detection. TSS and VSS concentrations in water or sludge and sand were detected using Standard Methods 2540 (APHA, 2012).

Ce = C0exp(−k×t)

(2)

NO3−

where Ce is the concentration at time t (mg NO3−-N/L), C0 is the initial NO3− concentration (mg NO3−-N/L), and k is the NO3− removal −1

rate constant (d ). The value of k was obtained by linear regression of ln (Ce /C0 ) versus t. Tests to determine statistical differences between electron donors were carried out by comparing the critical value through one-way ANOVA (Tong et al., 2018) using Excel 2010. Comparisons were considered significantly different for p < 0.05. Origin 9.0 software was used to determine if differences between duplicates were well-modeled by a normal distribution. The results showed that the data were drawn from a normally distributed population at a significance level of 0.05. 3. Results and discussion 3.1. Phase I: Influence of electron donor on NO3− removal There was significant reduction in NO3− concentrations in all denitrification microcosms (Fig. 1a), with no observed lag phase. There were statistically significant (p < 0.001) differences in first order NO3− removal rate constants (k; d−1) between substrates (Table 2). The highest denitrification rate was observed with methanol, followed by FW, SOD, WC and WSHAD. NO3− removal rates in Cycle 2 were not significantly different than Cycle 1 except for a decrease in the SOD microcosm, which may have been due to a decrease in alkalinity (Table 2). Changes in NO3− concentrations in abiotic microcosms were negligible and little denitrification was observed in endogenous decay and oyster shell control microcosms. Differences in denitrification rates between heterotrophic substrates, methanol, FW and WC, were likely due to difference in labilities of organic matter (Fowdar et al., 2015). The highest NO3− removal rate (23.4 g N/(m3·d) and 46.8 mg NO3−-N/(g VSS·d)) was obtained in the methanol microcosm, as methanol is a directly utilizable substrate (Fernández-Nava et al., 2010) that is easily assimilated by the

343

Bioresource Technology 263 (2018) 340–349

Q. He et al.

Fig. 2. Changes in NO3−, NO2− and TAN concentrations in the microcosms using different wood species.

WC is in good agreement with Cameron and Schipper (2010), who reported that WC achieved the lowest NO3− removal rate, ranging from 3.0 to 4.9 g N/(m3·d), compared with other solid organic carbon sources (i.e., cobs, green waste and wheat straw), due to the lower lability of organic substrates released from WC. Although there was a significant difference (p < 0.05) in COD release from WC (Fig. 1d) between Cycles 1 and 2, NO3− removal rates remained constant (p > 0.05). This finding illustrates that detected COD was not a direct reason for the lower reaction rate and rbCOD measurements were made in subsequent tests. An increase in alkalinity was observed in all heterotrophic microcosms (Table 2), as heterotrophic denitrification generates hydroxyl ions (Hang et al., 2017). The alkalinity increase was 3.00 and 3.47 mg CaCO3/mg NO3−-N in the WC and FW microcosms, respectively, which is lower than the stoichiometric value of 3.57 mg CaCO3/mg NO3−-N (van Rijn et al., 2006). This may be attributed to acid release and generation from solids degradation. The rates of SOD was 3.5 g N/(m3·d) and 7.0 mg NO3−-N/(g VSS·d), which was lower than observed for methanol or FW. Autotrophs grow

acclimated methylotrophic denitrifying bacteria from the MAP biofilter. Similar methanol-based denitrification rates (43 and 158 g N/(m3·d)) have been reported for full-scale marine RAS (Hamlin et al., 2008). FW, a combination of liquid and solid carbon sources (soluble COD/total COD was 9.4% for the FW in our study), effectively removed NO3− from marine RAS water at a high rate of 4.5 g N/(m3·d) and 9.0 mg NO3−-N/(g VSS·d). The soluble fraction of the FW can directly support denitrifying bacteria, while the solid fraction can slowly release organic substrates for denitrification. Timmons et al. (2002) reported that the suspended solids in FW vary greatly in size from the cm size to the micron (µm) size and the majority of particles by weight are less than 100 µm in RAS. Fine solids (1–100 µm) are dispersed throughout the solution (Timmons et al., 2002), providing large surface area for enhanced microbial attachment and colonization. It should be noted that the hydrolysis/fermentation products of FW considerably solubilized TAN and phosphorus. The TAN concentration increased up to 23.7 mg/L (Fig. 1c), which would be acutely toxic to fish, as discussed below. The TP concentration increased by 18.0 mg P/L (Table 2). The relatively low denitrification rate (2.6 g N/(m3·d)) observed for 344

Bioresource Technology 263 (2018) 340–349

Q. He et al.

molecules that can pass through cell membrane and be metabolized within minutes (Mamais et al., 1993). Although COD concentrations in O-WC microcosms were higher than that those in P-WC microcosms, rbCOD concentrations were almost same (Fig. 3). Eucalyptus had a slow release of COD, and its rbCOD concentrations were lower than those in P-WC and O-WC microcosms. The higher NO3− removal rates observed for pine (softwood) than oak and eucalyptus (hardwoods) (Table 3) is in agreement with previous studies (Cameron and Schipper, 2010; Gibert et al., 2008; Robertson, 2010). Robertson et al. (2000) discussed the dual porosity characteristics of WC media, where denitrification occurs several millimeters into the wood particles rather than being restricted to the particle surfaces. Softwood (coarse) tends to have a higher permeability than hardwood (fine), and thus softwood is prone to capture water, enhancing contact between microorganisms and substrates (van Driel et al., 2006). Wood age also plays a significant role in denitrification rate. Warneke et al. (2011) indicated that when the WC had been used for > 2.5 years, the denitrification rate for E-WC (hardwood) was higher than that for P-WC (softwood). Cameron and Schipper (2010) also concluded that after 23 months, there were no significant differences in NO3− removal rate between hardwood (3.3–4.4 g N/(m3·d)) and softwood (3.0–4.9 g N/(m3·d)), although softwood did exhibit a higher denitrification rate in the first several months. This finding indicates that hardwoods, such as oak and eucalyptus, may be appropriate for long-term use as a carbon source due to their slow carbon release. In addition, it worth noting that NO3− removal rates in the present study were significantly lower than that reported by Saliling et al. (2007) (810 g N/(m3·d)), possibly due to the WC’s size and the use of marine water. The dimensions of the WC particles used in our study were 2–6 mm by 2–4 mm, while WCs used by Saliling et al. (2007) were approximately 8–50 mm on one side and 2–15 mm on the other. This finding is consistent with Cameron and Schipper (2010) who indicated that the removal rate for large size WC (15 mm) was significantly greater than that of small size (4 mm) WC. Water in WC microcosms was brown colored during operation; the highest intensity of brown colored water was observed in O-WC, followed by E-WC and P-WC. This was due to tannins leached from the WC (Hartz et al., 2017). Although colored water is not necessarily a disadvantage for fish performance, clear water could enhance the ability of fish to capture feed and therefore lead to enhanced growth and improved feed conversion ratios; and allows the farmer to observe fish health, behavior, and feeding activity (Davidson et al., 2016). The water clarity problems could require the use of ozone and ultraviolet light to address this problem. During Phase II, changes in NO3− concentrations in the abiotic, endogenous decay and Cycle 1 of the “inactivated” control microcosms were negligible. However, NO3− removal was observed in Cycle 2 in the “inactivated” control WC microcosms after the replacement of

at a slower rate than heterotrophs (Sahinkaya and Kilic, 2014), resulting in increased residence time requirements but lower sludge production. Simard et al. (2015) reported SOD rates between 1.71 and 1.96 g N/(m3·d) for aquarium water with 30 ppt salinity. SOD rates for saline water were lower than for freshwater RAS (710–810 g N/(m3·d); Christianson et al., 2015) and drinking water (300 g N/(m3·d) (Sahinkaya and Dursun, 2012). As discussed previously, high salinity can have a negative effect on denitrification rate (Simard et al., 2015). A decrease in NO3− removal rate were observed in Cycle 2 compared to Cycle 1 in SOD microcosms (Table 2). This may have been due decreased alkalinity, as oyster shell was not replenished in Cycle 2. Simard et al. (2015) indicated that the addition of oyster shell columns between SOD columns increased the pH (mean of 7.4) and alkalinity (mean of 134 mg CaCO3/L) of the treated water. SO42− concentrations increased over time during Phase I (Fig. 1e). SO42− generation was 7.36 and 7.65 mg SO42−/mg NO3−-N for Cycle 1 and Cycle 2, respectively. This finding is consistent with the theoretical value of 7.54 mg SO42−/mg NO3−-N (Batchelor and Lawrence 1978). Christianson et al. (2015) investigated SOD performance in a RAS and reported an average of 1.72–3.35 mg SO42−/mg NO3−-N, most likely due to partial heterotrophic denitrification (caused by FW input) in addition to SOD. Unexpectedly, Phase I denitrification rates for WSHAD were lower than for SOD alone (Table 2). Note that the eucalyptus used in the Phase I WC and WSHAD microcosms was a commercial mulch material, which may have included inhibitory substances. As discussed previously, to gain more information on wood species and to standardize the size, a different source of eucalyptus was used for Phases II and III. 3.2. Phase II: Influence of wood species The selection of wood species is a matter of great importance for enhancing denitrification performance. As shown in Fig. 2, NO3− was effectively removed in all wood species microcosms at different rates, and at a lower rate for each wood species in subsequent cycles (Table 3). No lag phase was observed for any of the microcosms, indicating that the inoculum was well acclimated to the applied conditions. The highest NO3− removal rate was obtained with P-WC, followed by O-WC and E-WC. A significant release in COD was observed for each wood species at the beginning of the experiment, which subsequently decreased over time (Fig. 3). Differences in NO3− removal rates between the three species were attributed to the difference in rbCOD:COD ratio (Fig. 3). Wood is a solid substrate primarily composed of lignocellulose, consisting of cellulose (45–55% content), hemi-cellulose (24–40%) and lignin (18–35%). Hydrolysis occurs when bacteria excrete extracellular enzymes that break down solid substrates into low molecular weight (MW) carbohydrates (Lopez-Ponnada et al., 2017). Readily biodegradable organic matter consists of simple organic Table 3 Phase II results obtained in microcosms using different wood species. Microcosms Operation (days)

NO3−-N (mg/L) Final

TON (mg/L)

ΔTP (mg/L)

pH

Final

Final-Initial

Initial

Alkalinity (mg CaCO3/L)

NO3− removal rate

Final

Initial

Final

k (d−1)

R2

Average denitrification rate g N/(m3·d)

Cycle 1

P-WC E-WC O-WC

9 39 13

BDL* 8.3(2.6)* BDL

1.7(0.7) 2.2(0.9) 0.9(0.3)

+2.4(0.2) +5.2(0.7) +5.5(1.3)

7.07(0.02) 6.82(0.03) 6.88(0.05)

7.12(0.13) 7.60(0.26) 6.76(0.19)

112(12) 112(20) 113(10)

438(23) 219(16) 473(32)

0.250 0.080 0.201

0.8766 0.9215 0.9058

16.5(1.1) 4.9(1.7) 9.5(0.6)

Cycle 2

P-WC E-WC O-WC

10 51 26

BDL 76.7(2.4) BDL

3.6(1.2) 1.3(0.9) 1.5(1.0)

−2.2(0.0) −0.3(0.2) −2.2(0.6)

7.06(0.06) 7.45(0.02) 7.02(0.05)

7.27(0.05) 7.26(0.09) 7.25(0.21)

299(36) 184(20) 360(30)

607(35) 423(21) 675(32)

0.247 0.006 0.140

0.8982 0.9436 0.9800

15.4(0.9) 1.2(1.0) 6.7(0.8)

Cycle 3

P-WC O-WC

17 30

BDL BDL

1.3(1.2) 0.5(0.5)

−3.2(0.7) −4.0(1.6)

7.28(0.25) 7.00(0.03)

7.65(0.29) 7.12(0.11)

360(25) 412(15)

720(10) 713(26)

0.154 0.06

0.9762 0.9411

11.6(0.4) 4.7(0.2)

BDL-below detection limit. Values shown in parentheses represent standard deviation. 345

Bioresource Technology 263 (2018) 340–349

Q. He et al.

Fig. 3. Changes in COD and rbCOD concentrations in the microcosms using different wood species.

synthetic RAS water. Although this was unexpected, it is likely a result of denitrifying bacteria present in WC that were not inactivated by heating at 105 °C. Once the RAS water was replaced the microcosms reached an optimal pH range of 6.5–7.5 for denitrifying bacteria (Lu et al., 2014). 3.3. Phase III: WSHAD performance with different wood species NO3− was effectively removed in all Phase III WSHAD microcosms with no observed lag period (Fig. 4). NO3− removal rate was higher in P-WSHAD than E-WSHAD and decreases in NO3− removal rates were observed in the second cycle. There were no significant differences in NO3− removal between P-WSHAD and PO-WSHAD or between EWSHAD and EO-WSHAD (p > 0.05). WSHAD resulted in a significantly higher (p < 0.05) NO3− removal rate than SOD alone (Tables 2 and 4). This may have been due to the presence of facultative chemolithrotrophic denitrifiers, such as Thiobacillus spp., as they can grow mixotrophically by using reduced sulfur compounds as energy sources and organic carbon for biosynthesis (Oh et al., 2001). SO42− generation in Cycle 1 of P-WSHAD was less than half of the theoretical value (7.54 mg SO42−/mg NO3−-N) of SOD (Fig. 5), another indicator of simultaneous autotrophic and heterotrophic denitrification. SO42− generation was lower for pine than eucalyptus (Table 4), possibly due to the more slowly biodegradable organic carbon released from E-WC than P-WC. Hang et al. (2017) reported that NO3− was primarily reduced by heterotrophic denitrification when the carbon source was adequate, and the residual NO3− would be reduced by SOD. When it comes to the effect of alkalinity addition on mixotrophic denitrification performance, although no significant differences were observed in denitrification performance with and without oyster shell (p > 0.05), there was a slight improvement in EO-WSHAD performance, possibly because of the lower bioavailability of organic carbon

Fig. 4. Changes in NO3− concentration in the microcosms using different wood species-fed WSHAD.

and higher fraction of SOD in the eucalyptus based microcosms. 3.4. Denitrification by-products Efficient control of TAN and NO2− concentrations are essential issues in relation to commercial fish production, as even low concentrations can be toxic to fish (< 3 mg TAN/L and < 1 mg NO2−/L for 346

Bioresource Technology 263 (2018) 340–349

225(16) 309(8) 87(10) 197(3) 189(27) 197(10) 173(2) 189(7) 6.97(0.02) 7.10(0.05) 6.34(0.12) 6.95(0.03)

0.9771 0.9679 0.9872 0.9847 0.114 0.115 0.058 0.070

Fig. 5. Changes in SO42− concentration in the microcosms using different wood species-fed WSHAD.

2.8(0.6) 2.5(0.5) 0.1(0.0) 0.3(0.2)

2.5(0.2) 1.3(0.6) 0.7(0.3) 0.5(0.1)

-0.8(0.5) -0.3(0.2) -0.6(0.1) -0.8(0.0)

7.64(0.02) 7.71(0.06) 7.51(0.05) 7.54(0.05)

fish health; Timmons et al., 2002). In addition, TAN and NO2− will increase oxygen demand in nitrification process and fish tanks. TAN concentrations were < 4 mg N/L (4% of total NO3− removed) in all microcosms except for high TAN generation of FW, which was discussed previously (Fig. 1c, Fig. 2). Some TAN generation may have been due to dissimilatory NO3− reduction to ammonium (DNRA), which converts NO3− to ammonium rather than removing it as N2 (He et al., 2016; Li et al., 2016). Greenan et al. (2006) added 15N-labelled NO3− to WC columns and incubations and found that DNRA accounted for less than 4% of total NO3− removed. Gibert et al. (2008) concluded that less than 10% of removed NO3− was attributable to DNRA. Environmental factors such as high C/N ratio, high temperature, high carbon loads and increasing S2− may favor DNRA over denitrification (Suhr et al., 2014). Total organic nitrogen (TON) values (Tables 3 and 4) reveal that organic nitrogen was released from FW and WC, its mineralization under anaerobic conditions likely led to the observed increase in TAN (Crab et al., 2007). Decreases in TAN were attributed to the microbial uptake for bacterial growth (Hang et al., 2017). Transient NO2− accumulation was observed in all microcosms with different profiles observed for different electron donors, as shown in Fig. 1b, Fig. 2, and Fig. 6. NO2− is an intermediate species in denitrification process and may accumulate transiently depending on the activities of NO3− and NO2− reductases (Fowdar et al., 2015). NO2− reductase has been shown to be enzymatically inhibited by high NO3− concentrations, resulting transient NO2− accumulation (Carrey et al., 2014). Different rates of NO2− accumulation with different electron donors were possibly due to differences in the dominant microbial species in the systems (Lu et al., 2014). For example, populations related to Methylophilus, Paracoccus, Methyloversatilis and Hyphomicrobium spp., have been identified in methanol-fed denitrification systems (Lu et al., 2014), while Thiobacillus and Sulfurimonas are the most commonly reported sulfur-oxidizing autotrophic denitrifiers (Zhang et al., 2015). In some bacterial strains, NO3− reduction outcompetes NO2− reduction, thus NO2− will be reduced only after most of the NO3− has been removed (Fowdar et al., 2015). In the presence of complex organic molecules found in WC (e.g. proteins, lipids and lignin) denitrification has been shown to be facilitated by fermentative bacteria that are capable of reducing NO3− to NO2− but unable to use NO2− or other reduced nitrogen oxides as electron acceptors (Fowdar et al., 2015). Lower NO2− accumulation in subsequent cycles than in the first cycle in all microcosms indicated that denitrifying bacteria acclimated to the

Values shown in parentheses represent standard deviation.

21(1) 676(89) 19(3) 37(23) 5.52(0.03) 6.15(0.05) 5.65(0.01) 5.97(0.11) P-WSHAD PO-WSHAD E-WSHAD EO-WSHAD Cycle 2

23 23 37 37

0.9937 0.9925 0.9876 0.9939 0.248 0.230 0.194 0.193 7.65(0.06) 7.61(0.11) 7.58(0.04) 7.61(0.08) +0.6(0.3) +0.4(0.2) +0.8(0.3) +0.4(0.2) 0.5(0.2) 0.1(0.1) 1.7(0.3) 2.9(0.6) 2.5(0.0) 1.9(0.2) 2.8(0.3) 3.3(0.6) 277(26) 297(2 6 0) 43(28) 74(80) 3.17(0.55) 4.77(0.22) 5.39(0.05) 6.15(0.12) P-WSHAD PO-WSHAD E-WSHAD EO-WSHAD Cycle 1

16 16 21 21

7.2(0.3) 7.5(0.2) 4.7(0.6) 5.0(1.2)

223(1) 259(30) 217(2) 240(18) 122(8) 112(12) 112(13) 124(6) 6.58(0.02) 6.91(0.05) 6.79(0.12) 7.08(0.03)

11.9(0.4) 11.7(0.3) 10.5(0.4) 10.6(0.7)

g N/(m3·d) k (d−1) Final Initial Final Final Operation (days)

mgSO42−/mgNO3−-N

Final

Final

Final-Initial

Initial

R2

Average denitrification rate NO3− removal rate TAN (mg/L) S2− (µg/L) SO42− generation Microcosms

Table 4 Phase III results obtained in WSHAD microcosms using different wood species.

TON (mg/L)

ΔTP (mg/L)

pH

Alkalinity (mg CaCO3/L)

Q. He et al.

347

Bioresource Technology 263 (2018) 340–349

Q. He et al.

addition, COD released from WC may create an oxygen demand in downstream processes if RAS systems including WCs are not designed properly; and tannins released from WC may result in brown colored water, which may require ozone or ultraviolet treatment to reduce the color of RAS water. The WSHAD process exhibited a high NO3− removal rate, while COD release, SO42− generation and the alkalinity demands were relatively low. Results from this study also provide insight into the selection of sustainable electron donors for denitrification of other saline wastewaters that are difficult to treat. For example, seawater is used for toilet flushing in many countries, such as Hong Kong, to alleviate water shortages (Huang et al., 2018), resulting in saline domestic wastewater. Ion-exchange resins are used to treat NO3− contaminated groundwater for use as drinking water. The spent resins are regenerated using high concentration NaCl solutions, resulting in generation of high-NO3− brines (Trögl et al., 2011). Industrial processes, such as seafood processing, tanneries, pulp and paper and textile dyeing, also generate saline wastewater containing NO3− (Liang et al., 2017). These wastewaters could be treated in biofilters packed with WC, sulfur or WSHAD media. 4. Conclusions Fig. 6. Changes in NO2− concentration in the microcosms using different wood species-fed WSHAD.

Low cost, readily available substrates was evaluated for saline wastewater denitrification. Fish waste resulted in a high denitrification rate and can simultaneously decrease saline organic waste discharges; however, post-treatment is needed for TAN and COD removal. Readily biodegradable COD release from wood chips significantly affects the rate of denitrification. The highest denitrification rate was obtained with pine (rbCOD:COD = 82.5%) followed by oak (46.9%) and eucalyptus (42.2%). Wood-sulfur heterotrophic-autotrophic denitrification (WSHAD) resulted in a higher denitrification rate and lower SO42− generation and alkalinity consumption than SOD alone and a lower COD release than WC alone.

conditions, and the inhibitory levels of NO3− for NO2− reductases decreased. A high TP release was observed in Phase I FW microcosms, as discussed previously. Decreases in TP were observed in all other microcosms in Phase I. In Phases II and III, decreases in TP were observed in the first cycle, with subsequent decreases in subsequent cycles most likely due to initial release of P from the wood followed by microbial uptake (Hang et al., 2017). As discussed previously, a major by-product of SOD and WSHAD was SO42−, which was generally proportional to the extent of SOD (Christianson et al., 2015; Oh et al., 2001). SO42− productivity in microcosms with oyster shell was higher than without oyster shell (Table 4). No reports were found on the effect of SO42− concentration on marine fish health; however, more research is needed in this area as a continuous increase in SO42− concentration over time may have an adverse effect. Conversely, the production of S2− can pose a toxicological threat to the cultured product. S2− generation was most likely due to the activity of sulfate-reducing bacteria or sulfur disproportionation (Sahinkaya et al., 2011). S2− production was observed in the WSHAD microcosms (Table 4). Hamlin et al. (2008) reported that hydrogen sulfide was acutely toxic to fish and led to extreme hypoxia in the water. In contrast, Yogev et al. (2017) suggested that the combination of high concentrations of TAN and hydrogen sulfide has the potential to control fish pathogens and thus reduce potential infection if water is routed to a nitrification unit after denitrification.

Acknowledgements This research work was supported by China Scholarship Council, China (CSC, NO. 201606400027). This material is based upon work supported by the National Science Foundation, United States under Grant No. 1243510 and Florida Sea Grant, United States. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or Florida Sea Grant. The authors would also like to express our gratitude to Victoria Burnett for her assistance with sampling and analysis. References APHA, 2012. Standard Methods for the Examination of Water & Wastewater. American Public Health Association (APHA), American Water Works Association (AWWA) and Water Environment Federation (WEF). Batchelor, B., Lawrence, A.W., 1978. A kinetic model for autotrophic denitrification using elemental sulfur. Water Res. 12, 1075–1084. Boxman, S.E., Kruglick, A., McCarthy, B., Brennan, N.P., Nystrom, M., Ergas, S.J., Hanson, T., Main, K.L., Trotz, M.A., 2015. Performance evaluation of a commercial land-based integrated multi-trophic aquaculture system using constructed wetlands and geotextile bags for solids treatment. Aquac. Eng. 69, 23–36. Cameron, S.C., Schipper, L.A., 2010. Nitrate removal and hydraulic performance of carbon substrates for potential use in denitrification beds. Ecol. Eng. 36, 1588–1595. Carrey, R., Oteroa, N., Vidal-Gavilana, G., Ayora, C., Soler, A., Gómez-Alday, J.J., 2014. Induced nitrate attenuation by glucose in groundwater: Flow-through experiment. Chem. Geol. 370, 19–28. Christianson, L., Lepine, C., Tsukuda, S., Saito, K., Summerfelt, S., 2015. Nitrate removal effectiveness of fluidized sulfur-based autotrophic denitrification biofilters for recirculating aquaculture systems. Aquac. Eng. 68, 10–18. Crab, R., Avnimelech, Y., Defoirdt, T., Bossier, P., Verstraete, W., 2007. Nitrogen removal techniques in aquaculture for a sustainable production. Aquaculture 270, 1–14. Davidson, J., Good, C., Welsh, C., Summerfelt, S.T., 2014. Comparing the effects of high vs. low nitrate on the health, performance, and welfare of juvenile rainbow trout Oncorhynchus mykiss within water recirculating aquaculture systems. Aquac. Eng. 59,

3.5. Implications for marine RAS and other applications The use of FW as an electron donor for marine RAS has the potential to simultaneously decrease saline organic waste discharges and control NO3− levels. However, an aerobic nitrification process would be needed for TAN removal prior to returning water to the fish tanks. In addition, fine solids, dispersed throughout the FW solution, need to be captured and degraded in the denitrification reactor; otherwise, the use of FW can result in an increase in suspended solids concentration in fish tank, increasing the cost of solids removal. S0 results in a stable denitrification at a slow rate; however, SO42− generation may be a concern. WC are a good potential electron donor and the alkalinity generated can improve nitrification performance, reducing alkalinity supplement requirements; however, wood species needs to be carefully considered. In 348

Bioresource Technology 263 (2018) 340–349

Q. He et al.

Oh, S.E., Yoo, Y.B., Young, J.C., Kim, I.S., 2001. Effect of organic on sulfur-utilizing autotrophic denitrification under mixotrophic conditions. J. Biotechnol. 92, 1–8. Rhody, N.R., Puchulutegui, C., Taggart, John B., Main, K.L., Migaud, H., 2014. Parental contribution and spawning performance in captive common snook Centropomus undecimalis broodstock. Aquaculture 144–153. Robertson, W.D., Blowes, D.W., Ptacek, C.J., Cherry, J.A., 2000. Long-term performance of in situ reactive barriers for nitrate remediation. Groundwater 38, 689–695. Robertson, W.D., 2010. Nitrate removal rates in woodchip media of varying age. Ecol Eng. 36, 1581–1587. Rodriguez-Gonzalez, L.C., 2017. Advanced Treatment Technologies for Mitigation of Nitrogen and Off-flavor Compounds in Onsite Wastewater Treatment and Recirculating Aquaculture Systems (PhD Dissertation). Department of Civil and Environmental Engineering, University of South Florida. Sahinkaya, E., Dursun, N., Kilic, A., Demirel, S., Uyanik, S., Cinar, O., 2011. Simultaneous heterotrophic and sulfur-oxidizing autotrophic denitrification process for drinking water treatment: control of sulfate production. Water Res. 45, 6661–6667. Sahinkaya, E., Dursun, N., 2012. Sulfur-oxidizing autotrophic and mixotrophic denitrification processes for drinking water treatment: elimination of excess sulfate production and alkalinity requirement. Chemosphere 89, 144–149. Sahinkaya, E., Kilic, A., 2014. Heterotrophic and elemental-sulfur-based autotrophic denitrification processes for simultaneous nitrate and Cr(VI) reduction. Water Res. 50, 278–286. Saliling, W.J.B., Westerman, P.W., Losordo, T.M., 2007. Wood chips and wheat straw as alternative biofilter media for denitrification reactors treating aquaculture and other wastewaters with high nitrate concentrations. Aquac. Eng. 37, 222–233. Sengupta, S., Ergas, S.J., Lopez-Luna, E., 2007. Investigation of solid-phase buffers for sulfur-oxidizing autotrophic denitrification. Water Environ. Res. 79, 2519–2526. Simard, M., Masson, S., Mercier, G., Benmoussa, H., Blais, J., Coudert, L., 2015. Autotrophic denitrification using elemental sulfur to remove nitrate from saline aquarium waters. J. Environ. Eng. 141, 04015037. Strickland, J.D.H., Parsons, T.R., 1972. A Practical Handbook of Seawater Analysis, second ed. Fisheries Research Board of Canada Bulletin. Suhr, K.I., Pedersen, L.F., Nielsen, J.L., 2014. End-of-pipe single-sludge denitrification in pilot-scale recirculating aquaculture systems. Aquac. Eng. 62, 28–35. Timmons, M.B., Ebeling, J.M., Wheaton, F.W., Summerfelt, S.T., Vinci, B.J., 2002. Recirculating Aquaculture Systems, second ed. NRAC Publication No. 01-002. Tong, S., Rodriguez-Gonzalez, L.C., Payne, K.A., Stocks, J.L., Feng, C., Ergas, S.J., 2018. Effect of pyrite pretreatment, particle size, dose and biomass concentration on particulate pyrite autotrophic denitrification (PPAD) of nitrified domestic wastewater. Environ. Eng. Sci. (accepted). Trögl, J., Boušková, A., Mrákota, J., Pilařová, V., Krudencová, J., Měchurová, J., Kříženecká, S., Stloukal, R., 2011. Removal of nitrates from simulated ion-exchange brines with Paracoccus denitrificans encapsulated in Lentikats Biocatalyst. Desalination 275, 82–86. Tsukuda, S., Christianson, L., Kolb, A., Saito, K., Summerfelt, S., 2015. Heterotrophic denitrification of aquaculture effluent using fluidized sand biofilters. Aquac. Eng. 64, 49–59. van Driel, P.W., Robertson, W.D., Merkley, L.C., 2006. Denitrification of agricultural drainage using wood-based reactors. T. ASABE 49, 565–573. van Rijn, J., Tal, Y., Schreier, H.J., 2006. Denitrification in recirculating systems: theory and applications. Aquac. Eng. 34, 364–376. Warneke, S., Schipper, L.A., Matiasek, M.G., Scow, K.M., Cameron, S., Bruesewitz, D.A., McDonald, I.R., 2011. Nitrate removal, communities of denitrifiers and adverse effects in different carbon substrates for use in denitrification beds. Water Res. 45, 5463–5473. Yogev, U., Sowers, K.R., Mozes, N., Gross, A., 2017. Nitrogen and carbon balance in a novel near-zero water exchange saline recirculating aquaculture system. Aquaculture 467, 118–126. Zhang, L., Zhang, C., Hu, C., Liu, H., Bai, Y., Qu, J., 2015. Sulfur-based mixotrophic denitrification corresponding to different electron donors and microbial profiling in anoxic fluidized-bed membrane bioreactors. Water Res. 85, 422–431.

30–40. Davidson, J., Barrows, F.T.P., Kenney, B., Good, C., Schroyer, K., Summerfelt, S.T., 2016. Effects of feeding a fishmeal-free versus a fishmeal-based diet on post-smolt Atlantic salmon Salmo salar performance, water quality, and waste production in recirculation aquaculture systems. Aquac. Eng. 38–51. Fernández-Nava, Y., Marañón, E., Soons, J., Castrillón, L., 2010. Denitrification of high nitrate concentration wastewater using alternative carbon sources. J. Hazard. Mater. 173, 682–688. Fowdar, H.S., Hatt, B.E., Breen, P., Cook, P.L.M., Deletic, A., 2015. Evaluation of sustainable electron donors for nitrate removal in different water media. Water Res. 85, 487–496. Gibert, O., Pomierny, S., Rowe, I., Kalin, R.M., 2008. Selection of organic substrates as potential reactive materials for use in a denitrification permeable reactive barrier (PRB). Bioresour. Technol. 99, 7587–7596. Greenan, C.M., Moorman, T.B., Kaspar, T.C., Parkin, T.B., Jaynes, D.B., 2006. Comparing carbon substrates for denitrification of subsurface drainage water. J. Environ. Qual. 35, 824–829. Hamlin, H.J., Michaels, J.T., Beaulaton, C.M., Graham, W.F., Dutt, W., Steinbach, P., Losordo, T.M., Schrader, K.K., Main, K.L., 2008. Comparing denitrification rates and carbon sources in commercial scale upflow denitrification biological filters in aquaculture. Aquac. Eng. 38, 79–92. Hang, Q., Wang, H., Chu, Z., Hou, Z., Zhou, Y., Li, C., 2017. Nitrate-rich agricultural runoff treatment by Vallisneria-sulfur based mixotrophic denitrification process. Sci. Total Environ. 587–588, 108–117. He, Q., Feng, C., Peng, T., Chen, N., Hu, Q., Hao, C., 2016. Denitrification of synthetic nitrate-contaminated groundwater combined with rice washing drainage treatment. Ecol. Eng. 95, 152–159. Henze, M., Harremoës, P., la Cour, J., Arvin, E., 2002. Wastewater Treatment: Biological and Chemical Processes, third ed. Springer-Verlag, Berlin Heidelberg, Germany. Hartz, T., Smith, R., Cahn, M., Bottoms, T., Bustamante, S.C., Tourte, L., Johnson, K., Coletti, L., 2017. Wood chip denitrification bioreactors can reduce nitrate in tile drainage. Calif. Agric. 71, 41–47. Huang, Z., Wang, Y., Jiang, L., Xu, B., Wang, Y., Zhao, H., Zhou, W., 2018. Mechanism and performance of a self-flocculating marine bacterium in saline wastewater treatment. Chem. Eng. J. 334, 732–740. Krayzelova, L., Lynn, T.J., Banihani, Q., Bartacek, J., Jenicek, P., Ergas, S.J., 2014. A tiresulfur hybrid adsorption denitrification (T-SHAD) process for decentralized wastewater treatment. Water Res. 61, 191–199. Klas, S., Mozes, N., Lahav, O., 2006. A conceptual, stoichiometry-based model for singlesludge denitrification in recirculating aquaculture systems. Aquaculture 259, 328–341. Lay, W.C., Liu, Y., Fane, A.G., 2010. Impacts of salinity on the performance of high retention membrane bioreactors for water reclamation: a review. Water Res. 44, 21–40. Li, R., Feng, C., Hu, W., Xi, B., Chen, N., Zhao, B., Liu, Y., Hao, C., Pu, J., 2016. Woodchipsulfur based heterotrophic and autotrophic denitrification (WSHAD) process for nitrate contaminated water remediation. Water Res. 89, 171–179. Liang, Y., Zhu, H., Banuelos, G., Yan, B., Shutes, B., Cheng, X., Chen, X., 2017. Removal of nutrients in saline wastewater using constructed wetlands: Plant species, influent loads and salinity levels as influencing factors. Chemosphere 187, 52–61. Lopez-Ponnada, E.V., Lynn, T.J., Peterson, M., Ergas, S.J., Mihelcic, J.R., 2017. Application of denitrifying wood chip bioreactors for management of residential nonpoint sources of nitrogen. J. Biol. Eng. 11, 1–14. Lu, H.J., Chandran, K., Stensel, D., 2014. Microbial ecology of denitrification in biological wastewater treatment. Water Res. 64, 237–254. Mamais, D., Jenkins, D., Prrr, P., 1993. A rapid physical-chemical method for the determination of readily biodegradable soluble cod in municipal wastewater. Water Res. 27, 195–197. Martins, C.I.M., Eding, E.H., Verdegem, M.C.J., Heinsbroek, L.T.N., Schneider, O., Blancheton, J.P., d'Orbcastel, E.R., Verreth, J.A.J., 2010. New developments in recirculating aquaculture systems in Europe: a perspective on environmental sustainability. Aquac. Eng. 43, 83–93.

349