Enhanced nitrogen removal of low carbon wastewater in denitrification bioreactors by utilizing industrial waste toward circular economy

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC-ND 4.0 https://creativecommons.org/licenses/by-nc-nd/4.0/ Enhanced nitrogen removal of low carbon wastewater in denitrification bioreactors by utilizing industrial waste toward circular economy © Elsevier Ltd. Accepted version (Final draft) Kiani, Sepideh; Kujala, Katharina; Pulkkinen, Jani; Aalto, Sanni L.; Suurnäkki, Suvi; Kiuru, Tapio; Tiirola, Marja; Kløve, Bjørn; Ronkanen, Anna-Kaisa Kiani, S., Kujala, K., Pulkkinen, J., Aalto, S. L., Suurnäkki, S., Kiuru, T., Tiirola, M., Kløve, B., & Ronkanen, A.-K. (2020). Enhanced nitrogen removal of low carbon wastewater in denitrification bioreactors by utilizing industrial waste toward circular economy. Journal of Cleaner Production, 254, Article 119973. https://doi.org/10.1016/j.jclepro.2020.119973 2020 Journal Pre-proof Enhanced nitrogen removal of low carbon wastewater in denitrification bioreactors by utilizing industrial waste toward circular economy Sepideh Kiani, Katharina Kujala, Jani Pulkkinen, Sanni L. Aalto, Suvi Suurnäkki, Tapio Kiuru, Marja Tiirola, Bjørn Kløve, Anna-Kaisa Ronkanen PII: S0959-6526(20)30020-2 DOI: https://doi.org/10.1016/j.jclepro.2020.119973 Reference: JCLP 119973 To appear in: Journal of Cleaner Production Received Date: 30 August 2019 Revised Date: 12 December 2019 Accepted Date: 2 January 2020 Please cite this article as: Kiani S, Kujala K, Pulkkinen J, Aalto SL, Suurnäkki S, Kiuru T, Tiirola M, Kløve Bjø, Ronkanen A-K, Enhanced nitrogen removal of low carbon wastewater in denitrification bioreactors by utilizing industrial waste toward circular economy, Journal of Cleaner Production (2020), doi: https:// doi.org/10.1016/j.jclepro.2020.119973. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier Ltd. Sepideh kiani: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Writing - Original Draft, Writing - Review & Editing, Visualization Anna-Kaisa Ronkanen: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Writing - Original Draft, Writing - Review & Editing, Visualization, Supervision, Project administration Björn Klöve: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Writing - Original Draft, Writing - Review & Editing, Visualization, Supervision, Project administration Katharina Kujala: Methodology, Validation, Formal analysis, Investigation, Resources, Writing - Original Draft, Writing - Review & Editing, Visualization, Supervision Jani Pulkkinen: Validation, Formal analysis, Investigation, Resources, Writing - Review & Editing, Tapio Kiuru: Investigation, Resources, Project administration Sanni L. Aalto: Methodology, Validation, Formal analysis, Investigation, Resources, Writing - Review & Editing, Visualization Suvi Suurnäkki: Methodology, Validation, Formal analysis, Investigation, Resources, Writing - Review & Editing, Visualization Marja Tiirola: Methodology, Validation, Formal analysis, Investigation, Resources, Writing - Review & Editing, Visualization, Supervision, Project administration 1 Enhanced nitrogen removal of low carbon wastewater in denitrification bioreactors by utilizing industrial waste toward circular economy Sepideh Kiani a *, Katharina Kujala a , Jani Pulkkinen b , Sanni L. Aalto c,d , Suvi Suurnäkki c , Tapio Kiuru b , Marja Tiirola c , Bjørn Kløve a and Anna-Kaisa Ronkanen a a Water, Energy and Environmental Engineering Research Unit, Faculty of Technology, P.O. Box 4300, FI90014 University of Oulu, Finland b Natural Resources Institute Finland, Survontie 9A, 40500 Jyväskylä, Finland c Department of Biological and Environmental Science, Nanoscience Center, 40014 University of Jyväskylä, Finland d Department of Environmental and Biological Sciences, University of Eastern Finland, P.O. Box 1627, 70211 Kuopio, Finland *Corresponding author: Sepideh Kiani (Email: sepide[email protected]) 1 Abstract 1 Aquaculture needs practical solutions for nutrient removal to achieve sustainable fish production. Passive 2 denitrifying bioreactors may provide an ecological, low-cost and low-maintenance approach for wastewater 3 nitrogen removal. However, innovative organic materials are needed to enhance nitrate removal from the low 4 carbon effluents in intensive recirculating aquaculture systems (RAS). In this study, we tested three 5 additional carbon sources, including biochar, dried Sphagnum sp. moss and industrial potato residues, to 6 enhance the performance of woodchip bioreactors treating the low carbon RAS wastewater. We assessed 7 nitrate (NO 3- ) removal and microbial community composition during a one-year in situ column test with real 8 aquaculture wastewater. We found no significant differences in the NO 3removal rates between the 9 woodchip-only bioreactor and bioreactors with a zone of biochar or Sphagnum sp. moss (maximum removal 10 rate 31-33 g NO 3- -N m -3 d -1 ), but potato residues increased NO 3removal rate to 38 g NO 3- -N m -3 d -1 , with 11 stable annual reduction efficiency of 93%. The readily available carbon released from potato residues 12 increased NO 3- -N removal capacity of the bioreactor even at higher inflow concentrations (>52 mg L -1 ). The 13 microbial community and its predicted functional potential in the potato residue bioreactor differed markedly 14 from those of the other bioreactors. Adding potato residues to woodchip material enabled smaller bioreactor 15 size to be used for NO 3removal. This study introduced industrial potato by-product as an alternative carbon 16 source for the woodchip denitrification process, and the encouraging results may pave the way toward 17 growth of blue bioeconomy using the RAS. 18 19 Keywords: Recirculating aquaculture system, woodchip bioreactor, carbon source, potato residues, nitrate, 20 microbial community 21 22 23 24 25 26 27 2 1 Introduction 28 Recirculating aquaculture systems (RAS) are environmentally friendly solutions that aim to achieve zero 29 waste from fish production. Although RAS have been used for more than 10 years in different countries, 30 including two largest RAS in Finland with a production capacity of over 4000 tons, nitrate (NO 3- ) removal is 31 still a critical challenge (Pulkkinen et al., 2018). Removal of NO 3is a challenge as aquaculture wastewater 32 has low carbon (C) but high nitrogen (N) concentrations. A few previous studies have examined the use of 33 denitrifying bioreactors for treating aquaculture effluent. So far, such studies have focused on RAS effluents 34 with high chemical oxygen demand (COD) (Lepine et al., 2016), added bicarbonate (HCO -3 ) to inlet water 35 (von Ahnen et al., 2016b) and diluted effluent from an outdoor fish farm with low recirculation intensity and 36 low NO 3- -N concentration (~6 mg L -1 ) (von Ahnen et al., 2018, 2016a). In contrast, treatment of highly 37 intensive indoor RAS effluents with low COD (12.9 ± 1.8 mg L -1 ) and high NO 3- -N concentration (>50 mg 38 L -1 ) has received little attention. 39 In denitrifying bioreactors, nitrogen (N) is removed by heterotrophic denitrifiers converting NO 3to nitrogen 40 gas under anoxic conditions. Under nitrate-rich conditions, this process depends on the availability of the 41 carbon source as the organic electron donor (Wang and Chu, 2016). External carbon sources, such as acetate 42 or methanol, are often supplied to the system to achieve efficient denitrification (Cherchi et al., 2009). 43 However, the cost of carbon addition is typically high (Zhang et al., 2016) and the process needs regulation 44 to prevent overor under-dosing of the liquid carbon sources (Rocher et al., 2015). Solid carbon sources can 45 provide a cost-effective alternative to the classical carbon sources mentioned above. In recent years, research 46 has focused on solid carbon sources with high quality, optimal efficiency and slow-release ability in the 47 treatment of excessively nitrate-contaminated water, particularly surface water (Beutel et al., 2016) and 48 groundwater (Zhang et al., 2012). Wood-particle products (e.g. woodchip and sawdust) have been widely 49 used, due to their ability to supply carbon to the denitrification process for 5-15 years and thus allow good 50 NO 3removal with minimum bioreactor maintenance (Schipper et al., 2010). However, the large space 51 requirement for full-scale woodchip bioreactors has prompted efforts to enhance the denitrification rate by 52 using innovative natural carbon sources (Tangsir et al., 2017). Inexpensive industrial food by-products, such 53 3 as industrial potato residue, could have high potential to be utilized in identifying bioreactor to enhance 54 nitrate removal. Potato industries can generate 20-25 % waste from peeling, trimming and cutting processes 55 (Liang and McDonald, 2014). 56 This study examined the use of a denitrifying bioreactor to treat indoor intensive RAS effluent with low 57 COD and high NO 3concentration, as part of the unique RAS research platform (see Pulkkinen et al., 2018), 58 and compared different carbon sources, including potato residue, for improving the nitrogen removal 59 performance of woodchip bioreactors. The overall aim was to evaluate the performance of denitrifying 60 bioreactors in removing NO 3from aquaculture wastewater with low COD for a period of over one year. 61 Specific objectives were to (1) study the suitability of wood-based bioreactors for treating RAS effluent, (2) 62 assess whether the NO 3removal performance of woodchip process can be enhanced by additional carbon 63 sources, (3) to assess the effect of different carbon sources on the microbial community composition in 64 different compartments of the bioreactors, and (4) to identify dominant bacteria and their functional potential 65 in the bioreactors studied. The intention was to find solutions for improving water treatment and for 66 enhancing NO 3removal in the recirculating aquaculture systems. 67 2 Material and methods 68 2.1 RAS effluent water quality 69 The study was conducted at the Laukaa fish farm of the Natural Resources Institute Finland (LUKE) in 70 central Finland, in the research platform examining RAS. The RAS design is described in detail in Pulkkinen 71 et al. (2018). In brief, effluent was obtained from a RAS consisting of a feed collector unit, swirl separator, 72 drum filter (60 µm mesh) and fixed bed bioreactor, followed by a moving bed bioreactor and a trickling 73 filter. In order to prevent any changes in water chemistry, microbiology or water temperature, all tests were 74 performed using the natural RAS effluent. The effluent is characterised by low carbon (15.3 mg L -1 on 75 average), but high N content (mean NO 3- -N content 34.7 mg L -1 ) (Table 1). Due to the efficient nitrification 76 unit before the bioreactors, NO 3is dominating N fraction. 77 Table 1. Mean inflow water quality parameters (SD = standard deviation, n = number of sample) 78 4 Water quality parameters Inflow (mean ± SD) n Total organic carbon (mg L - 1 ) 15.3 ± 2.1 5 Dissolved organic carbon (mg L - 1 ) 14 ± 1.3 5 Chemical oxygen demand (mg L - 1 ) 12.9 ± 1.8 5 Biological oxygen demand (mg L - 1 ) 3.8 ± 2.2 13 Nitrate-nitrogen (mg L - 1 ) 34.7 ± 15.6 27 Nitrite-nitrogen (mg L - 1 ) 0.1 ± 0.06 30 Ammonium-nitrogen (mg L - 1 ) 0.5 ± 0.2 30 Dissolved oxygen (mg L - 1 ) 8.1 ± 1.7 29 pH 6.9 ± 0.2 28 Oxidation - reduction potential ( Eh , mV) 178.6 ± 60.4 35 Alkalinity (mg CaCO 3 L - 1 ) 54.2 ± 18 25 Sulphate (mg L - 1 ) 10.5 ± 3.2 24 2.2 Bioreactor design 79 The performance of denitrifying bioreactors was studied in four transparent acrylic columns (0.1 m diameter 80 × 0.32 m high) with upward flow direction applying a theoretical retention time (HRT) of 48 h at controlled 81 temperature (15.5±0.8°C) (Fig. 1). In each column, the reactive media were placed on top of an inert quartz 82 gravel bed, from which they were separated by plastic netting with 2 mm pore size, to prevent clogging with 83 materials containing organic matter. A constant inflow rate of 0.6 mL min -1 was applied to each bioreactor 84 for 346 days, using a peristaltic pump. The upward flow direction and the quartz gravel layer at the base of 85 the columns prevented the development of preferential flow pathways and ensured uniform distribution of 86 flow into the columns. The columns consisted of packed-media zones (zone 1, zone 2, zone 3) containing 87 woodchips, industrial potato waste, biochar or dried Sphagnum sp. moss in the ratios shown in Fig. 1. The 88 packed-media has not been replaced during the study period. All bioreactors with additional layer contain 89 same total volume of woodchips. However, Sphagnum sp. moss was mixed with woodchips in the zone 2, 90 due to its different characteristic and small particle size distribution. It is well known that natural peat has 91 5 typically low hydraulic conductivity (e.g. Ronkanen and Kløve 2005), which could cause risks in longer 92 HRT or even clogging of the bioreactor. In order to avoid this, moss was mixed with woodchips. The 93 packed-media zones were separated from the outlet free water zone by a fixed perforated PVC plate 94 (thickness 5 mm) at a height of 4.5 cm from the top of the column. The columns were sealed at both ends to 95 provide controlled conditions. 96 The selected carbon sources had different C/N ratios, ranging from 28 to 249 (Table 2). Woodchips had the 97 highest C/N ratio, but biochar contained the highest amount of carbon. The used woodchips were obtained 98 locally from fresh birch trees (provided by the energy company Vapo Group). The average woodchip size 99 was around 3 cm × 1.5 cm × 0.4 cm and mean porosity 63%. The Sphagnum sp. moss used was common 100 mire flora provided by Vapo Group. The biochar (porosity 46%) was obtained from RPK Hiili Oy. The 101 potato material tested comprised industrial residues from POHJOLAN PERUNA Oy with a dry matter 102 content of 12% (determined after drying the material at 105°C for 24 h). 103 Prior to the experiments, solid materials (woodchips and biochar) were washed with distilled water and 104 saturated for 48 h. In order to prevent fermentation, the potato residues were kept in the freezer prior to use. 105 The frozen potato residues were defrosted at room temperature for 8 h before the test. 106 12 start, but only after stable denitrification rates are established and low nitrite concentrations are detected in 211 the outflow. 212 The inflow NH 4+ -N concentration ranged between 0.17-1.0 mg L -1 (Table 1; Fig. S1b). Low NH 4+ -N 213 production was detected in all bioreactors, with outflow concentrations of 0.8±0.5 mg L -1 , 0.9±0.5 mg L -1 , 214 0.9±0.6 mg L -1 and 3.8±3.4 mg L -1 in BR1, BR2, BR3 and BR4, respectively. Less than 2 mg L -1 of NH 4+ -N 215 was recorded in the first three weeks in BR1-BR3 (Fig. S1b). However, the bioreactor with potato residues 216 (BR4) showed relatively high NH 4+ -N, with a mean concentration of 10 mg L -1 , in the first 10 days of the 217 experiment, but it then declined to lower than 4 mg L -1 to reach the background level. The continuous 218 production of ammonium in BR4 indicates the occurrence of dissimilarity nitrate reduction to ammonium 219 (DNRA). In general, a reducing environment and high TOC/NO 3ratio (1400/15-110/16 in BR4; days 1-70) 220 can indicate the occurrence of DNRA (Kraft et al., 2014; van Rijn et al., 2006). DNRA has also been 221 observed in previous woodchip bioreactor studies (Lu et al., 2013; Zhao et al., 2018). Reducing conditions, 222 indicated by Eh values, were also seen in this study, which led the system to SO 42reduction (Fig. 4). 223 In the start-up phase, all bioreactors released DOC. The rate of release was highest in BR4, with outflow 224 concentrations of 1380 mg L -1 measured on day 6 after start-up (Table. S1). The DOC release from the other 225 bioreactors was much lower (<100 mg L -1 ; Table S1). Within 70 days after start-up, outflow DOC 226 concentration decreased to 81 mg L -1 in BR4 and to the background level (14 ± 1.3 mg DOC L -1 ) in BR1-227 BR3 (Table S1). Initial carbon content flush-out is common in bioreactors. The start-up COD concentration 228 in the outflow ranged 59-940 mg L -1 in BR1-BR4 (Table. S1) exceeding temporarily the maximum 229 concentration of 42 mg L -1 observed in Finnish rivers (Niemi and Raateland, 2007). However, start-up phase 230 of the woodchip bioreactor is short compared to estimated lifetime (5-15 years), so the potential pollution for 231 carbon is minor compared to the amount of nitrogen removed. Lepine et al. (2016) reported an 232 approximately 50-day flush-out period for a plywood bioreactor treating aquaculture effluent at HRT of 42 h. 233 Somewhat higher carbon leaching (200 mgL -1 ) has been reported for bioreactors packed with fresh 234 woodchips and a mixture of woodchips and biochar (Hassanpour et al., 2017; Hoover et al., 2016). Release 235 of high DOC concentrations to recipient water bodies from use of bioreactors as an end-of-pipe treatment 236 can adversely affect aquatic ecosystems, e.g. by causing a DO concentration reduction, light and temperature 237 13 changes (Prairie, 2008; Solomon et al., 2015), resulting in lower fish production (Stasko et al., 2012). Hence, 238 at sites governed by strict regulations or when recycling outflow to fish farms, high DOC might need to be 239 controlled. Schipper et al. (2010) identified HRT as a factor controlling the initial magnitude of DOC 240 depletion and its duration in wood-based bioreactors. However, the fact that carbon was more readily 241 released from potato residues than from the other carbon sources used in this study proves that HRT is not 242 the only controlling factor and that carbon quality also plays a key role. In the present study, there was 243 significantly lower outflow DOC concentration of 53, 68 and 81 mg L -1 in bioreactors BR1, BR2 and BR3, 244 which can be partly explained by higher nitrate loading (Hassanpour et al., 2017) and partly by the type of 245 carbon source used. Dependence of TOC leaching and variations in NO 3- -N concentration have also been 246 reported by Zhao et al. (2018). In order to control the carbon content due to leaching, it is recommended to 247 consider post-bioreactors treatment units (e.g. constructed wetland, sand filter) or recirculating the start-up 248 effluent back to the bioreactor (Schipper et al., 2010). 249 The SO 42concentrations were on average higher in the outflow than in the inflow waters of BR1 and BR2, 250 indicating leaching or production of SO 42- (Fig. 4). This resulted in cumulative leaching/production of 165 g 251 and 474 g SO 42in BR1 and BR2, respectively, for the whole study period. In contrast, SO 42were on average 252 lower in outflow than in inflow waters of BR3 and BR4 (Fig. 4), indicating SO 42reduction/removal. 253 Cumulative SO 42removal of 350 g and 546 g was observed in BR3 and BR4, respectively, for the whole 254 study period. SO 42leaching/removal increased the SO 42concentration in the outflow by up to 20% 255 compared with the cumulative inflow SO 42of 2.6 kg. Sulphate leaching/production indicated the potential of 256 internal sulphur cycling in bioreactors with incomplete N removal. BR1 and BR2 had incomplete nitrate 257 removal during the study period due to sulphide re-oxidation to sulphate by sulphur oxidizing bacteria 258 (SOB), which can use oxygen or nitrate as electron acceptor (Faulwetter et al., 2009) (Fig. S1 and Fig. 3). 259 Sulphate production was observed previously by Lepine et el. (2016) for a woodchip bioreactor with 260 incomplete N removal. However, higher nitrate removal in BR3 and BR4 combined with their reduced 261 conditions (Fig.4) favored sulphate reduction. 262 14 Fig. 4. Sulphate reduction/removal (+ values) and leaching/production (-values) in bioreactors BR1-BR4 263 over time at different redox potential values (Eh) in inflow and outflow for each bioreactor. 264 Redox potential was on average +340, +354, +312 and +181 mV in BR1, BR2, BR3 and BR4, respectively 265 (Fig. 4), indicating more oxidising conditions in BR1-BR3 and more reducing conditions in BR 4. It is well-266 known that denitrification and microbial sulphate removal cause decline in redox potential and rise in pH 267 (Jog and Parry., 2006). In BR4, for the entire study period when outlet Eh reduced from 412 to 116 mV, the 268 pH tended to increase about 2.2 pH units (from 4.6-6.82) (Fig. S 5). Similarly, in BR1-3 by decreasing the 269 outlet redox potential, the pH increased 0.89,1.65 and 1.4 pH units, respectively . 270 Inflow water pH was rather stable throughout the experiment (6.5-7.5) (Fig. 5). Outflow pH of bioreactors 271 during start-up was 6, 4.3, 5.2 and 3.8 in BR1 BR2, BR3 and BR4, respectively. It was thus lower than 272 inflow pH in the early stages of the experiment, most likely as a result of release of organic acids from the 273 packed materials (Fig. 5). All bioreactors showed lower alkalinity in outflow than in inflow during the start-274 up period (Fig. 5). After 2-5 weeks, alkalinity production was observed in all bioreactors. 275 15 3.2 Factors affecting nitrate removal in woodchip bioreactors 276 The results of one-way ANOVA showed that NO 3removal rates for whole study period did not differ 277 significantly between BR1, BR2 and BR 3 (p=0.75), while nitrate removal in BR4 was higher (Fig. 2d-2h). 278 In the first three months of the experiment, when inflow NO 3- -N concentration varied between 15 and 52 mg 279 L -1 , all bioreactors showed similar removal rates (Fig. 2). After that, the bioreactors responded differently to 280 increasing NO 3- -N inflow concentrations, e.g. the removal rate declined in BR1-BR3 but increased in BR4 281 (Fig. 2). BR4 reached its maximum removal rate of 38 g NO 3- -N m -3 d -1 at the highest NO 3- -N inflow 282 concentration (70 mg L -1 ; days 152-184), whereas BR1, BR2 and BR3 had a removal rate of 9, 13 and 12 g 283 NO 3- -N m -3 d -1 , respectively (Fig. 2). Those differences persisted until day 250, after which all reactors again 284 had similar stable removal rates of around 15 g NO 3- -N m -3 d -1 until the end of the experiment. Similarly to 285 removal rate, the NO 3removal efficiency in BR1-BR3 showed fluctuations throughout the study period (Fig. 286 2e and 2g). However, BR4 reached stable removal efficiency of 93% after a period of fluctuation at start-up 287 (Fig. 2h). 288 The wide range of NO 3removal rates (3-38 g NO 3- -N m -3 d -1 ) recorded in all bioreactors followed the NO 3- -289 N inflow concentration fluctuations. High removal rate in all bioreactors occurred when the inflow had high 290 Fig. 5 . Alkalinity production (+values) and inflow and outflow pH in bioreactors (BR1-BR4). 16 NO 3- -N concentrations. This is consistent with previous findings that inflow concentrations control removal 291 rate (e.g. Schipper et al., 2010; Addy et al., 2016). 292 In the present study, NO 3removal rate in BR4 increased significantly with increasing NO 3- -N inflow 293 concentration during the entire study period (R 2 = 0.93; removal rate = 0.6 × influent nitrate concentration - 294 1.85) (Fig. 6). This regression illustrated the actual relationship between inflow NO 3- -N concentration and 295 removal rate by excluding NO 3- -N limited events (NO 3- -N concentration <0.5 mg L -1 ) (Addy et al., 2016). 296 Likewise, bioreactors BR1-BR3 showed a similar response to NO 3- -N when days 152-212, with high NO 3- -N 297 concentration (55-70 mg L -1 ), were excluded from the data (Fig. 6). The sharply decline in NO 3- -N removal 298 during days 152-212 was caused due to exceeding the maximum denitrification capacity in those bioreactors. 299 This indicates that NO 3removal in BR1-BR3 was controlled by an independent parameter at high NO 3- -N 300 concentrations. The release rate of degradable carbon from the packed media presumably controlled NO 3301 removal in this concentration range (>52 mg L -1 ) (Schipper et al., 2010). Hence, the type of carbon source 302 used in denitrifying bioreactors can control NO 3removal, by providing more carbon availability and 303 different microbial composition (Xu et al., 2018; Tangsir et al., 2017). Observed DOC in the bioreactors 304 showed that carbon was much more readily released from potato residues than from any of the other carbon 305 sources tested (Table S1). The easily soluble carbon in potato residues resulted in rapid formation of a 306 complex microbial community structure with strong adaptive growth to the new environment (Zhao et al., 307 2018). 308 309 17 The maximum NO 3removal rates observed in this study were greater than those previously reported (22 g 310 NO 3- -N m -3 d -1 ) (David et al., 2015; Schipper et al., 2010). This could be due to a combination of optimal 311 factors: sufficient HRT (Lepine et al., 2016; Tangsir et al., 2017) as a result of distributed upward flow 312 (section 2.2) combined with high NO 3inflow concentration (Schipper et al., 2010), the organic C 313 compounds used (Gibert et al., 2008) and water temperature (Addy et al., 2016), here 15.5 ± 1 °C (mean ± 314 SD). A removal rate of >39 g NO 3m -3 d -1 reported by Lepine et al. (2016) for comparable water quality was 315 associated with high COD:NO 3ratio (0.86-1.66) in treated wastewater. This ratio can provide 42% COD 316 required for denitrification. The COD:NO 3ratio has been reported to be a significant parameter affecting 317 denitrification in bioreactors (Jafari et al., 2015). However, in the present study inflow COD provided less 318 than 8% of the C/N required for complete NO 3reduction (Narkis et al., 1979). Hence, the reported NO 3319 removal rates in this study represent the net values without a contribution from inflow COD. Enhancing 320 nitrate removal efficiency with different carbon substrates has been investigated previously (Gebert et al., 321 2008; Schipper et al., 2010; Hashemi et al., 2011). Hashemi et al., (2011) improved nitrate removal of 36% 322 in wood bioreactor to 65%, 56 % and 77 % by utilizing barley straw, rice husk and date palm leaf, 323 respectively. Gebert et al., (2008) reported softwood (branches and bark with small amounts of leaves from a 324 variety of trees) as top performing substrate in denitrification efficiency (>98%) with denitrification rate of ~ 325 Fig. 6 . Nitrate removal rate versus nitrate influent loading in BR14 for the study period of 346 days. 18 17 g NO 3- -N m -3 d -1 . However, other investigated materials such as mixture of wood chips, shredded bark and 326 topsoil, compost (obtained from the biological decomposition of organic wastes – wood trimmings, leaves, 327 rotten vegetables and food scraps) and willow woodchips identified as unsuitable carbon sources (see Gebert 328 et al., 2008). Warneke et al., (2011) reported nitrate removal of ~ 6.5, 6.2 and 3.5 g NO 3- -N m -3 d -1 for wheat 329 straw, maize and green waste materials, respectively compare to the removal rate of 1.3 g NO 3- -N m -3 d -1 in 330 soft wood (pine) bioreactor for 2-fold lower nitrate inlet concentration than used in this study. However, 331 additional potato residue to woodchip bioreactor increased 13% of nitrate removal to 38 g NO 3- -N m -3 d -1 332 which is remarkably higher than reported removal above. 333 3.3 Microbial community composition and process potential in the bioreactors 334 A total of 9261 quality-filtered sequences per library were obtained from water and solid samples from the 335 four bioreactors (Table 3). Library coverage was ≥94% in all cases, indicating that the sequencing depth was 336 sufficient. The number of observed and Chao 1-estimated OTUs was significantly lower (p<0.001) in filtered 337 water and solid material from BR4 than in corresponding samples from BR1-BR3. The Shannon diversity 338 index was also significantly lower (p<0.001) in BR4 (4.5) than in BR1-BR3. 339 The microbial community in BR4 differed strongly from the microbial community in BR1-BR3 (Figs. S 2A). 340 Smaller differences were detected between the microbial communities in BR1-BR3 and between water and 341 solid samples from all bioreactors (Figs. S2 B and C). In solid material, differences were observed between 342 microbial communities in zone 3 (i.e. top-layer woodchip) and in zone 2 in BR1, BR2 and BR4 (containing 343 biochar, Sphagnum sp. moss and potato residues, respectively) but not BR3 (containing woodchips) (Fig. 1). 344 In water, the differences were much less pronounced (Figs. S2 B and C). 345 Table 3. Prokaryotic diversity in bioreactors BR1-BR4. Numbers of sequences are taken from the original 346 OTU tables, while all other diversity indicators are based on OTU tables rarified at a depth of 4098 347 sequences. Average values for 1-2 replicates per sampling point are shown. Zone 2 and zone 3 refer to the 348 carbon source material tested and the top-layer woodchip, respectively, as indicated in Fig. 1 349 No. of No. of Coverage OTUs OTUs Shannon 19 sequences samples (%) richness (observed) richness (estimated) a BR 1: Woodchip/ Biochar Water Zone 2 8 550 2 95 441 802 4.64 Zone 3 7 310 2 95 468 761 4.68 Solid Zone 2 6 844 2 94 496 827 4.77 Zone 3 4 935 2 95 398 739 4.36 BR 2: Woodchip/ Sphagnum Water Zone 2 0 Zone 3 7 500 1 95 450 821 4.67 Solid Zone 2 7 358 1 96 383 697 4.42 Zone 3 6 711 2 96 354 674 4.2 BR 3: Woodchip/ woodchip Water Zone 2 8 198 2 95 433 749 4.53 Zone 3 8 304 2 94 480 854 4.72 Solid Zone 2 6 942 2 96 378 713 4.29 Zone 3 6 956 1 95 389 897 4.26 BR 4: (Woodchip/ potato) Water Zone 2 9 261 2 96 303 583 3.61 Zone 3 8 148 2 96 337 605 3.78 Solid Zone 2 9 256 2 97 287 505 3.67 Zone 3 8 359 2 96 296 578 3.39 a OTUs richness estimated by Chao1. 350 Only bacterial sequences (no archaeal sequences) were detected in the bioreactors. In BR1-BR3, the 351 microbial community was dominated by Proteobacteria, Bacteroidetes and Verrucomicrobia (Fig. 7). 352 Within the Proteobacteria, Betaproteobacteria were most abundant (24-40% relative abundance), followed 353 by Gammaproteobacteria (7-26%) and Alphaproteobacteria (11-28%). In BR4, the microbial community 354 was dominated by Epsilonproteobacteria (15-36%), Bacteroidetes (16-29%) and Firmicutes (17-34%) (Fig. 355 7). Amongst the most abundant genera, Uliginosibacterium (up to 11% relative abundance), Sulfurospirillum 356 (up to 29%), Prevotella (up to 19%) and Lactobacillus (up to 18%) were almost exclusively detected in BR4, 357 while Rhodobacter (up to 4%), Sphingobium (up to 4%), Rhodoferax (up to 5%), Pseudomonas (up to 13%), 358 Thermomonas (up to 6%) and Luteolibacter (up to 10%) were almost exclusively detected in BR1-BR3 (Fig. 359 20 S3). The genera Lactobacillus, Prevotella and Sulfurispirillum include known fermenters, some of which can 360 also reduce nitrate to ammonium (e.g. Kruse et al., 2018; Salvetti et al., 2012). The genera Rhodobacter, 361 Rhodoferax, Pseudomonas and Thermomonas include known denitrifiers (e.g. Finneran et al., 2003; 362 Mergaert et al., 2003). 363 Fig. 7. Composition of the microbial community based on sequence analysis of bacterial and archaeal 16S 364 rRNA genes from (A) solid material and (B) water samples from woodchip bioreactors with a zone 365 containing biochar (BR1), Sphagnum sp. moss (BR2), woodchip (BR3) and potato residues (BR4). Average 366 relative abundances of 1-2 replicates per sample are shown. Samples were taken from the top-layer 367 woodchip (zone 3) and the carbon source material (zone 2). 368 Functional profiles of the bacterial communities were predicted based on 16S rRNA gene sequences using 369 PICRUSt. It proved possible to use around 31% of all OTUs and 83% (76-90%) of all sequences for 370 functional prediction. Overall functional profiles of microbiological communities were rather similar in the 371 different bioreactors. Selected functions related to the nitrogen cycle were assessed in more detail (Fig. 8). 372 21 Functions related to denitrification (NarG, NapA, NirK, NorB, NorC, NosZ) and DNRA (NarG, NapA, 373 NrfA) were predicted, while functions specific to nitrification (AmoA, AmoB, AmoC) were not predicted. 374 The membrane-bound nitrate reductase NarG was predicted in similar relative abundance in all bioreactors, 375 while higher relative abundance of the periplasmic nitrate reductase NapA was predicted in BR4 than in 376 BR1-BR3 (Fig. 7). The denitrification-associated functions NirK, NorB, NorC and NosZ were predicted with 377 higher relative abundances for BR1-BR3 than for BR4, while the nitrite reductase NrfA (which catalyses the 378 reduction of nitrite to ammonia in DNRA) was more frequently predicted for BR4 (Fig. 8). This indicates 379 that bioreactors BR1-BR3 had higher predicted potential for denitrification, while the bioreactor with potato 380 residues (BR4) had higher predicted potential for DNRA. The nitrite reductase NirK may also be present in 381 nitrifying organisms. However, the contribution of nitrifiers such as Nitrospira sp. or Nitrobacter sp. to NirK 382 was only 0.15%. 383 Fig. 8. Relative abundance of predicted nitrogen cycle-related genes in functional profiles of (A) solid 384 material and (B) water samples from woodchip bioreactors with a zone containing biochar (BR1), Sphagnum 385 sp. moss (BR2), woodchip (BR3) and potato residues (BR4). Functional profiles were predicted based on 386 16S rRNA gene sequences using PICRUSt. Average relative abundances of 1-2 replicates per sample are 387 shown. 388 28 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. https://doi.org/10.1016/j.biortech.2008.02.012 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. https://doi.org/10.2134/jeq2005.0247 Hashemi, S.E., Heidarpour, M., Mostafazadeh-Fard, B., 2011. Nitrate removal using different carbon substrates in a laboratory model. Water Sci. Technol. 63, 2700. Hassanpour, B., Giri, S., Pluer, W.T., Steenhuis, T.S., Geohring, L.D., 2017. Seasonal performance of denitrifying bioreactors in the Northeastern United States: Field trials. J. Environ. Manage. 202, 242–253. https://doi.org/10.1016/j.jenvman.2017.06.054 Herbert, R.B., Winbjörk, H., Hellman, M., Hallin, S., 2014. Nitrogen removal and spatial distribution of denitrifier and anammox communities in a bioreactor for mine drainage treatment. Water Res. 66, 350–360. https://doi.org/10.1016/j.watres.2014.08.038 Hoover, N.L., Bhandari, A., Soupir, M.L., Moorman, T.B., 2016. Woodchip denitrification bioreactors: Impact of temperature and hydraulic retention time on nitrate removal. J. Environ. Qual. 45, 803–812. Jafari, S.J., Moussavi, G., Yaghmaeian, K., 2015. High-rate biological denitrification in the cyclic rotating-bed biological reactor: Effect of COD / NO 3 - , nitrate concentration and salinity and the phylogenetic analysis of denitrifiers. Bioresour. Technol. 197, 482–488. https://doi.org/10.1016/j.biortech.2015.08.047 Jong, T., Parry, D.L., 2006. Microbial sulfate reduction under sequentially acidic conditions in an upflow anaerobic packed bed bioreactor. Water Res. 40, 2561-2571. 29 Kraft, B., Tegetmeyer, H.E., Sharma, R., Klotz, M.G., Ferdelman, T.G., Hettich, R.L., Geelhoed, J.S., Strous, M., 2014. The environmental controls that govern the end product of bacterial nitrate respiration. Science 345, 676–679. Kroupova, H., Machova, J., Svobodova, Z., 2005. Nitrite influence on fish: a review. Vet. Med.- Praha50, 461. Kruse, S., Goris, T., Westermann, M., Adrian, L., Diekert, G., 2018. Hydrogen production by Sulfurospirillum species enables syntrophic interactions of Epsilonproteobacteria. Nat. Commun. 9, 4872. https://doi.org/10.1038/s41467-018-07342-3 Langille, M.G., Zaneveld, J., Caporaso, J.G., McDonald, D., Knights, D., Reyes, J.A., Clemente, J.C., Burkepile, D.E., Thurber, R.L.V., Knight, R., 2013. Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat. Biotechnol. 31, 814. Lepine, C., Christianson, L., Sharrer, K., Summerfelt, S., 2016. Optimizing Hydraulic Retention Times in Denitrifying Woodchip Bioreactors Treating Recirculating Aquaculture System Wastewater. J. Environ. Qual. 45, 813. https://doi.org/10.2134/jeq2015.05.0242 Liang, S., McDonald, A.G., 2014. Chemical and thermal characterization of potato peel waste and its fermentation residue as potential resources for biofuel and bioproducts production. J. Agric. Food Chem. 62, 8421–8429. Lu, W.-W., Zhang, H.-L., Shi, W.-M., 2013. Dissimilatory nitrate reduction to ammonium in an anaerobic agricultural soil as affected by glucose and free sulfide. Eur. J. Soil Biol. 58, 98– 104. Mergaert, J., Cnockaert, M.C., Swings, J., 2003. Thermomonas fusca sp. nov. and Thermomonas brevis sp. nov., two mesophilic species isolated from a denitrification reactor with poly(εcaprolactone) plastic granules as fixed bed, and emended description of the genus Thermomonas. Int. J. Syst. Evol. Microbiol. 53, 1961–1966. 30 Narkis, N., Rebhun, M., Sheindorf, C.H., 1979. Denitrification at various carbon to nitrogen ratios. Water Res. 13, 93–98. Natural Resources Institute Finland [referred 26.06.2019]. Access method: https://stat.luke.fi/en/aquaculture-2016_en Niemi, J., Raateland, A., 2007. River water quality in the Finnish Eurowaternet. Boreal Environ. Res. 12, 571–84. Nordström, A., Herbert, R.B., 2018. Determination of major biogeochemical processes in a denitrifying woodchip bioreactor for treating mine drainage. Ecol. Eng. 110, 54–66. https://doi.org/10.1016/j.ecoleng.2017.09.018 Oksanen, J., Blanchet, F.G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P.R., O’Hara, R.B., Simpson, G.L., Solymos, P., 2017. vegan: Community Ecology Package 2017. R package version 2.4–4. Parada, A.E., Needham, D.M., Fuhrman, J.A., 2016. Every base matters: assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol. 18, 1403–1414. Prairie, Y.T., 2008. Carbocentric limnology: looking back, looking forward. Can. J. Fish. Aquat. Sci. 65, 543–548. https://doi.org/10.1139/f08-011 Pulkkinen, J.T., Kiuru, T., Aalto, S.L., Koskela, J., Vielma, J., 2018. Startup and effects of relative water renewal rate on water quality and growth of rainbow trout (Oncorhynchus mykiss) in a unique RAS research platform. Aquac. Eng. 82, 38–45. https://doi.org/10.1016/j.aquaeng.2018.06.003 Robertson, W.D., 2010. Nitrate removal rates in woodchip media of varying age. Ecol. Eng. 36, 1581–1587. https://doi.org/10.1016/j.ecoleng.2010.01.008 Rocher, V., Laverman, A.M., Gasperi, J., Azimi, S., Guérin, S., Mottelet, S., Villières, T., Pauss, A., 2015. Nitrite accumulation during denitrification depends on the carbon quality and 31 quantity in wastewater treatment with biofilters. Environ. Sci. Pollut. Res. 22, 10179–10188. https://doi.org/10.1007/s11356-015-4196-1 Ronkanen, A.K. and Kløve, B., 2005. Hydraulic soil properties of peatlands treating municipal wastewater and peat harvesting runoff. Suo. 56(2), 43-56. Salvetti, E., Torriani, S., Felis, G.E., 2012. The Genus Lactobacillus: A Taxonomic Update. Probiotics Antimicrob. Proteins 4, 217–226. https://doi.org/10.1007/s12602-012-9117-8 Schipper, L.A., Robertson, W.D., Gold, A.J., Jaynes, D.B., Cameron, S.C., 2010. Denitrifying bioreactors—An approach for reducing nitrate loads to receiving waters. Ecol. Eng. 36, 1532–1543. https://doi.org/10.1016/j.ecoleng.2010.04.008 Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., 2009. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Env. Microbiol 75, 7537–7541. Solomon, C.T., Jones, S.E., Weidel, B.C., Buffam, I., Fork, M.L., Karlsson, J., Larsen, S., Lennon, J.T., Read, J.S., Sadro, S., Saros, J.E., 2015. Ecosystem Consequences of Changing Inputs of Terrestrial Dissolved Organic Matter to Lakes: Current Knowledge and Future Challenges. Ecosystems 18, 376–389. https://doi.org/10.1007/s10021-015-9848-y Stasko, A.D., Gunn, J.M., Johnston, T.A., 2012. Role of ambient light in structuring northtemperate fish communities: potential effects of increasing dissolved organic carbon concentration with a changing climate. Environ. Rev. 20, 173–190. https://doi.org/10.1139/a2012-010 Tangsir, S., Moazed, H., Naseri, A.A., Hashemi Garmdareh, S.E., Broumand-nasab, S., Bhatnagar, A., 2017. Investigation on the performance of sugarcane bagasse as a new carbon source in two hydraulic dimensions of denitrification beds. J. Clean. Prod. 140, 1176–1181. https://doi.org/10.1016/j.jclepro.2016.10.044 32 van Rijn, J., Tal, Y., Schreier, H.J., 2006. Denitrification in recirculating systems: Theory and applications. Aquac. Eng. 34, 364–376. https://doi.org/10.1016/j.aquaeng.2005.04.004 von Ahnen, M., Aalto, S.L., Suurnäkki, S., Tiirola, M., Pedersen, P.B., 2019. Salinity affects nitrate removal and microbial composition of denitrifying woodchip bioreactors treating recirculating aquaculture system effluents. Aquaculture 504, 182–189. von Ahnen, M., Pedersen, P.B., Dalsgaard, J., 2018. Performance of full-scale woodchip bioreactors treating effluents from commercial RAS. Aquac. Eng. 83, 130–137. https://doi.org/10.1016/j.aquaeng.2018.10.004 von Ahnen, M., Pedersen, P.B., Dalsgaard, J., 2016a. Start-up performance of a woodchip bioreactor operated end-of-pipe at a commercial fish farm—A case study. Aquac. Eng. 74, 96–104. https://doi.org/10.1016/j.aquaeng.2016.07.002 von Ahnen, M., Pedersen, P.B., Hoffmann, C.C., Dalsgaard, J., 2016b. Optimizing nitrate removal in woodchip beds treating aquaculture effluents. Aquaculture 458, 47–54. https://doi.org/10.1016/j.aquaculture.2016.02.029 Wang, J., Chu, L., 2016. Biological nitrate removal from water and wastewater by solid-phase denitrification process. Biotechnol. Adv. 34, 1103–1112. https://doi.org/10.1016/j.biotechadv.2016.07.001 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(17), 5463-5475. Westcott, S.L., Schloss, P.D., 2017. OptiClust, an improved method for assigning amplicon-based sequence data to operational taxonomic units. MSphere 2, e00073-17. Xu, Z., Dai, X., Chai, X., 2018. Effect of different carbon sources on denitrification performance, microbial community structure and denitrification genes. Sci. Total Environ. 634, 195-204 33 Zhang, H., Jiang, J., Li, M., Yan, F., Gong, C., Wang, Q., 2016. Biological nitrate removal using a food waste-derived carbon source in synthetic wastewater and real sewage. J. Environ. Manage. 166, 407–413. https://doi.org/10.1016/j.jenvman.2015.10.037 Zhang, J., Feng, C., Hong, S., Hao, H., Yang, Y., 2012. Behavior of solid carbon sources for biological denitrification in groundwater remediation. Water Sci. Technol. 65, 1696–1704. https://doi.org/10.2166/wst.2012.070 Zhao, J., Feng, C., Tong, S., Chen, N., Dong, S., Peng, T., Jin, S., 2018. Denitrification behavior and microbial community spatial distribution inside woodchip-based solid-phase denitrification (W-SPD) bioreactor for nitrate-contaminated water treatment. Bioresour. Technol. 249, 869–879. https://doi.org/10.1016/j.biortech.2017.11.011 - Woodchip bioreactors removed 31-38 g NO 3- -N m -3 d -1 from intensive aquaculture effluent - Additional potato residues to woodchip material increased 13 % of nitrate removal rate - The potato residue bioreactor hosted a distinctly different microbial community Declaration of interests ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: