The effects of different combinations of fixed and moving bed bioreactors on rainbow trout (Oncorhynchus mykiss) growth and health, water quality and nitrification in recirculating aquaculture systems

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/ The effects of different combinations of fixed and moving bed bioreactors on rainbow trout (Oncorhynchus mykiss) growth and health, water quality and nitrification in recirculating aquaculture systems © 2019 Elsevier B.V. Accepted version (Final draft) Pulkkinen, Jani T.; Eriksson-Kallio, Anna M.; Aalto, Sanni L.; Tiirola, Marja; Koskela, Juha; Kiuru, Tapio; Vielma, Jouni Pulkkinen, J. T., Eriksson-Kallio, A. M., Aalto, S. L., Tiirola, M., Koskela, J., Kiuru, T., & Vielma, J. (2019). The effects of different combinations of fixed and moving bed bioreactors on rainbow trout (Oncorhynchus mykiss) growth and health, water quality and nitrification in recirculating aquaculture systems. Aquacultural Engineering, 85, 98-105. https://doi.org/10.1016/j.aquaeng.2019.03.004 2019 The effects of different combinations of fixed and moving bed bioreactors on rainbow trout 1 (Oncorhynchus mykiss) growth and health, water quality and nitrification in recirculating aquaculture 2 systems 3 Jani T. Pulkkinena*, Anna M. Eriksson-Kalliob, Sanni L. Aaltocd, Marja Tiirolac, Juha Koskelaa, Tapio Kiurue, 4 Jouni Vielmaa 5 aNatural Resources Institute Finland, Survontie 9A, 40500 Jyväskylä, Finland 6 bFinnish Food Authority, Mustialankatu 3, 00790 Helsinki, Finland 7 cDepartment of Biological and Environmental Science, Nanoscience center, University of Jyväskylä, P.O. 8 Box 35, 40014 Jyväskylä, Finland 9 dDepartment of Environmental and Biological Sciences, University of Eastern Finland, P.O. Box 1627, 10 70211 Kuopio, Finland 11 eNatural Resources Institute Finland, Vilppulantie 415, 41340 Laukaa, Finland 12 *Corresponding author. jani.pulkki[email protected] +358 295323297 13 Highlights 14 - Organic material accumulated in the two-moving-bed systems 15 - Nitrite concentrations increased in the two-fixed-bed systems 16 - Different bioreactor designs did not affect fish health or growth 17 Abstract 18 The effect of bioreactor design on nitrification efficiency has been well studied, but less is known about 19 the overall impacts on water quality. Besides nitrification, submerged fixed bed bioreactors (FBBR) trap 20 fine solid particles, whereas moving bed bioreactors (MBBR) grind solids, possibly increasing solids and 21 particle accumulation in the system. In this experiment, the effects of different combinations of fixed 22 bed and moving bed bioreactors on water quality, solids removal, particle size distribution, fish health 23 based on histopathological changes and nitrification efficiency were studied in laboratory scale 24 recirculating aquaculture systems (RAS) with rainbow trout (Oncorhynchus mykiss). Three set-ups with 25 triplicate tanks were used: 1. two consecutive fixed bed bioreactors (FF); 2. a fixed bed bioreactor 26 followed by a moving bed bioreactor (FM) and 3. two consecutive moving bed bioreactors (MM). Fish 27 performance was not influenced by the design of the bioreactor, specific growth rate (SGR) being 28 between 1.59 and 1.64% d-1 and feed conversion ratio (FCR) between 0.95 and 0.98. Water nitrite 29 concentration was higher in the FF systems compared to FM and MM systems, whereas the average 30 total ammonia nitrogen concentration (TAN) was not influenced by the treatments. Nitrification rate, 31 which was measured in the laboratory, followed the water nitrite levels, indicating highest total 32 ammonium oxidation rates in the MM systems. UV254 absorbance and total organic carbon (TOC) 33 concentrations were higher in the groups with moving bed systems, indicating accumulation of organic 34 substances in the circulating water. The total volume of particles was higher in the MM systems as 35 compared to the FF systems. The total solids balance was similar in all the bioreactor groups, since the 36 removal of solids by the FBBR backwash was compensated by the drum filter in the FM and MM 37 systems. In general, no significant histopathological difference in gill, kidney, heart and liver tissue were 38 observed between the RAS treatment groups and the flow-through treatment. 39 Keywords: biofiltration; histopathology; particle size distribution; water quality monitoring 40 List of abbreviations 41 FBBR Fixed bed bioreactor 42 FCR Feed conversion ratio 43 FF Two consecutive fixed bed bioreactors 44 FM Fixed bed, followed by moving bed bioreactor 45 LEH Lamellar epithelial cell hyperplasia 46 MBBR Moving bed bioreactor 47 MM Two consecutive moving bed bioreactors 48 PSD Particle size distribution 49 SGR Specific growth rate 50 TAN Total ammonia nitrogen 51 TGC Thermal growth coefficient 52 TOC Total organic carbon 53 TS Total solids 54 55 1. Introduction 56 Nitrifying bioreactor operation and management is one of the most important and complex steps in 57 recirculating aquaculture systems (RAS) (Badiola et al., 2012; Svobodova et al., 2005). Typical RAS use 58 so-called fixed-film bioreactors, where biofilm is formed on artificial plastic carrier media or media 59 generated from natural substances such as sand and stones (Malone and Pfeiffer, 2006). Bacteria in the 60 media convert toxic ammonia into less toxic nitrate in a two phase nitrification process. The nitrification 61 process allows lower water usage rates, therefore decreasing the volume of effluents requiring the 62 treatment before discharged into the environment. There is a wide variety of nitrifying bioreactors used 63 in RAS, which all have particular strengths and weaknesses with no single reactor type being dominant 64 (e.g. Timmons and Ebeling, 2013). 65 The nitrification capacity of the following bioreactor types has been widely studied: moving bed 66 bioreactors (MBBR) (Kamstra et al., 2017), fixed bed bioreactors (FBBR) (Pedersen et al., 2015), fluidized-67 sand biofilters (Summerfelt, 2006), rotating biological contactor (Brazil, 2006) and trickling filters 68 (Greiner and Timmons, 1998; Lekang and Kleppe, 2000). Besides nitrification, different bioreactor types 69 can also have other impacts on water quality, depending on how they are designed and operated. 70 Trickling filters, MBBRs and RBCs are constantly interacting with air, which increases the oxygen (O2) 71 levels and reduces carbon dioxide concentration (CO2) (Timmons and Ebeling, 2013). However, there is 72 very little information on how the choice of bioreactor design can affect fish health and water quality 73 parameters. 74 The moving bed bioreactor was designed in Norway in the late 1980s (Rusten et al., 2006). The reactor 75 chamber is agitated continuously with compressed air or mechanically, the carrier media being 76 constantly moved so as to create a scrubbing effect against each other. Because of that scrubbing effect, 77 the reactor shears solid particles, leading to the accumulation of the total amount of particles in the 78 system (Fernandes et al., 2017). These types of reactors are easy to operate, because there is a low head 79 loss and no need for backwashing. In addition, the constant movement enables efficient use of the 80 whole reactor volume, and mixing with air provides oxygen for the nitrification process. Because of 81 scrubbing, surplus microbial biomass created in the biofilm detaches from the carrier media and is later 82 removed from the system either by outflow or in solids removal units (Ødegaard, 2006). 83 Fixed bed bioreactor or fixed bed biofilm reactor (FBBR) is a reactor type, where carrier media is 84 structurally fixed in the reactor chamber (Kadic and Heindel, 2014). Depending on the fixed media type, 85 the reactor can be susceptible to clogging and must be backwashed frequently (Schlegel and Koeser, 86 2007). When using small carrier media, suspended solids particles are commonly trapped in these 87 reactors (Fernandes et al., 2017). The distribution of flow into the reactor and inside the reactor is 88 important: turbulent flow can cause uneven distribution of substrate in the reactor and the total 89 effective surface area for nitrification may be diminished. Turbulent flow might also create pockets, 90 where oxygen can be depleted and hydrogen sulphide might form. 91 Since O2 is added and CO2 is removed mainly in the other compartments of RAS, the main water quality 92 difference between FBBR and MBBR is probably the fate of solid particles in the reactor. High suspended 93 solids loads have been reported to cause sub-lethal stress and damages to gill structure in some fish 94 species (Au et al., 2004; Bilotta and Brazier, 2008). Thus, the amount of solid particles may influence fish 95 health and welfare. In addition, there is a positive correlation between bacterial numbers and the 96 surface area of particles (Pedersen et al., 2017), which may indicate that MBBR accumulates more 97 bacteria in the circulating water than FBBR. 98 In this study, we compared two widely used bioreactor types: moving bed and fixed bed bioreactors. 99 The comprehensive approach was used for comparing the effects of different bioreactor setups on 100 ammonium removal rates, fish health in terms of histopathological lesions and growth parameters, 101 water quality, solids accumulation and microbial dynamics. Our hypothesis was that the accumulation of 102 solids in the circulating water causes histopathological changes and chronic stress in the fish, which 103 affect fish growth and feed efficiency. 104 2. Materials and methods 105 2.1. Experimental setup 106 The experiment was carried out in the Natural Resources Institute Finland (Luke) Laukaa fish farm using 107 an experimental RAS platform. The platform has 10 individual freshwater recirculating systems, each 108 consisting of a 500 l bottom drained plastic rearing tank (Arvo-Tec, Joroinen, Finland), feed collector 109 unit, 24 cm swirl separator (Eco-Trap Collector1, Pentair Aquatic Eco-Systems, Minneapolis, USA), drum 110 filter with 60 µm filter panels (Hydrotech HDF501, Veolia, Paris, France), 2 separate 147 l bioreactor 111 tanks (Arvo-Tec, Joroinen, Finland), trickling filter acting as a forced-ventilated cascade aeration column 112 (Bio-Blok® 200, EXPO-NET Danmark A/S, Hjørring, Denmark) and pump sump (Fig. 1). Water pH was 113 adjusted to 7.2 in pump sump with diluted sodium hydroxide using automated system (Prominent, 114 Heidelberg, Germany). Sodium bicarbonate was dosed to the inlet water source to achieve an alkalinity 115 of 50 mg l-1 (CaCO3) in the RAS replacement water. Oxygen saturation was kept above 80% in the fish 116 tanks. The system is described in more detail by Pulkkinen et al. (2018). 117 In the trial, three bioreactor setups were compared with triplicate units: Treatment 1. two consecutive 118 fixed bed bioreactors (FF); Treatment 2. fixed bed bioreactor followed by moving bed bioreactor (FM) 119 and Treatment 3. two consecutive moving bed bioreactors (MM). Two bioreactors per RAS unit were 120 used, so that all units had similar amount of bioreactors. The experiment lasted 14 weeks. In one 121 treatment group (FF), only two units existed for the second half of the experiment due to a technical 122 failure with pH in one tank in week 8 of the experiment. A separate 500 l flow-through tank was used to 123 grow fish of the same origin with same feed, serving as a flow-through treatment for fish 124 histopathological sampling. Water temperature was adjusted to 16 °C by controlling the air temperature 125 and in the flow-through group by controlling the inlet water temperature. 126 Similar plastic (PP) carrier media (RK Biolements heavy in fixed bed systems and medium in moving bed 127 systems, RK Plast A/S, Skive, Denmark), tank hydraulic retention time and make-up water flow were 128 used and measured constantly in all RAS units (Table 1). Carrier media, used in two earlier experiments, 129 was mixed four weeks before the trial started, and divided evenly between the bioreactors to ensure 130 similar bacterial seed in all the RAS units. In FF and FM units, the first bioreactor was backwashed once 131 every two weeks. In FF units the second bioreactor was backwashed once every four weeks. The FBBR 132 backwash water amount was not taken into account in the make-up water flow calculations, because it 133 increased the total water volume by less than 4%. 134 2.2. Fish and feeding 135 Three weeks before the trial started, a total of 820 one year old rainbow trout (Oncorhynchus mykiss) 136 (average weight 99 g) originating from the National JALO-selective breeding programme (Natural 137 Resources Institute Finland, Tervo, Finland) were divided into the 9 RAS units. When the trial started, 138 the fish were weighed and their biomasses were equalized. The fish were weighed twice during the 139 experiment at weeks four and eight and group weighing was used in all of the weightings. Fish were 140 fasted one day prior to and after the weighing. Feeding was carried out with a commercial feeding 141 system (T Drum 2000, Arvo-Tec, Joroinen, Finland) 10–14 times per day. Feed intake rate was constantly 142 monitored using sieve in the tank outlet and uneaten feed pellets were calculated. Feed company 143 feeding table was used for feeding rate and it was reduced by 0.1 %-unit, when uneaten feed was 144 observed. 1:1 mixture of two commercial diets was used to ensure that the results can be better 145 generalized across various commercial feeds. Diets were produced by Raisioaqua (Circuit Red 5 mm, 146 Raisio, Finland) and BioMar (Orbit 929 4.5 mm, Aarhus, Denmark). The crude protein and lipid contents 147 of the diets were 43% and 42%, and 26% and 31%, respectively. 148 The feed conversion ratio (FCR) was calculated as: FCR = F / G, where F = cumulative feed intake 149 between weightings and G = total tank biomass gain between weightings. Specific growth rate (SGR) was 150 calculated as: SGR = (ln(Wi+1 ) - ln(Wi)) / (ti+1 - ti) x 100, where W = average fish weight at given time and 151 ti+1 - ti = duration of feeding days. The thermal growth coefficient (TGC) was calculated for the whole 152 experiment according to Jobling (2003) as: TGC = ((We1/3) - (Wi1/3)) x (T x t) x 1000, where We = average 153 fish weight in the end, Wi = average fish weight at the beginning, T = average water temperature, t = 154 duration of feeding days. 155 2.3. Histopathological sampling and analysis 156 Tissue samples (gill, kidney and liver) from 5 fish per tank were collected at the start of the experiment, 157 twice during the experiment and again at the end of the experiment, at approximately one month 158 intervals. The second gill arch from the right hand side was sampled and sectioned parasagittally. 159 Kidney tissue was sampled as approximately 2 cm long sections from the distal third of the kidney and 160 sectioned transversely. Liver tissue was sampled in approximately 1×1 cm sections and sectioned 161 sagittally. The tissue samples were fixed in 10% neutral buffered formalin, embedded in paraffin, 162 sectioned at 4 μm and stained with haematoxylin and eosin (H&E) according to standard laboratory 163 practice. The sections were examined using light microscopy. 164 The histopathological changes were reported on a scale from 0–3: minimal, mild, moderate and severe 165 as described by Wolf et al. 2015. One section per tissue and per fish was examined. 166 The following parameters were studied and classified according to the severity of the lesions: 167 Gills: Lamellar epithelial cell hyperplasia (LEH): proliferation of the squamous epithelial cells lining the 168 gill surface. General diffuse proliferative branchitis: filling of interlamellar spaces by a mixed population 169 of epithelial and inflammatory cells. Focal branchitis: a local unspecific inflammatory change consisting 170 mainly of mononuclear, lymphocytic cell types involving a smaller area, usually only a few lamellae. 171 Lamellar fusion: one or more interlamellar sulci filled by proliferating pavement cells (with or without 172 increased mucous cells, chloride cells, and/or leucocytes). Lamellar adhesion: the often focal attachment 173 of adjacent lamellae with little or no evidence of cell proliferation. Lamellar thrombosis: formation of 174 blood clots inside lamellar capillaries consisting of fragmented thrombocyte nuclei and/or pink fibrinous 175 material within the distended capillaries. 176 Kidney: Tubular necrosis: Necrosis of tubular epithelial cells. Renal mineralization: mineralized material 177 intraepithelially or intraluminally. Number of melanomacrophage centres or pigmented macrophage 178 aggregates (PMAs): centres of mainly histiocytic macrophages that contain hemosiderin, melanin, 179 lipofuscin, and/or ceroid pigments and that serve as repositories for end-products of cell breakdown. 180 Liver: Hepatocellular cytoplasmic vacuolation: intracytoplasmic vacuoles containing glycogen or lipids. 181 Hepatitis/cholangiohepatitis: infiltration of acute or chronic inflammatory cells in liver tissue or around 182 bile ducts. 183 2.4. Water sampling and analysis 184 Total ammonia nitrogen, nitrite and nitrate were analysed once a week from the tank outlet water using 185 a spectrophotometer (Procedure 8038 Nessler, LCK341/342 and LCK340 respectively. DS 3900, Hach, 186 Loveland, USA). Alkalinity was analysed once a week with a standard method of titration (ISO 9963-187 1:1994) (TitraLab AT1000, Hach, Loveland, USA). 188 Particle size distribution (PSD) was analysed from the tank water and from the water taken from top of 189 the 2nd bioreactor at week 13 (S4031, PAMAS, Rutesheim, Germany). Optical analyses covered particle 190 sizes from 1 µm to 200 µm. A simple comparison of PSD between treatments was made by calculating 191 the β-values (slope of log10 frequency versus log10 particle size) according to Patterson et al. (1999). 192 Total particle surface area and volume were calculated by using the given particle size diameter 193 (assumed sphere) multiplied by the total number of particles. 194 Particle counts were also measured from tank water with a CASY cell counter with a capillary size of 45 195 µm (Model TT, OLS OMNI Life Science GmbH, Basel, Switzerland) at week 14. Measurement principal is 196 based on pulse area analysis, where low voltage field is cast through the samples. Measurement range 197 was between 0.8 µm to 30 µm. The 100 ml water samples were frozen before analysis. Triplicate 198 measurements per water sample were analysed using a sample size of 200 µl. 199 Total organic carbon (TOC) and UV254 (turbidity corrected) were monitored online at 6 minute intervals 200 in the fish tanks with a UV/VIS spectrometer (5 mm open path length, spectro::lyser, s::can, Vienna, 201 Austria). Carbon dioxide concentrations were monitored in the fish tanks at 6 minute intervals with a 202 carbon dioxide sensor (Franatech, Lüneburg, Germany). Two hour average values are presented for 203 these online measurements. 204 2.5. Solids sampling 205 Sludge was collected twice during the trial at weeks 7 and 11 for solids analysis. Sludge from swirl 206 separators was collected using a 0.31 litre tube placed at the bottom of the separators. The collection 207 period lasted six hours. Drum filter backwash water was collected for 16 hours, then weighed and 208 mixed, after which a subsample of 1 litre was collected. Fixed bed bioreactors were cleaned by vigorous 209 agitation with air, and one litre samples were collected from the top of the reactor and from the outlet 210 pipe. At week 11, water collected only from the top of the reactor was used because there was no 211 difference between these sampling points. All solids samples were put into a container and oven dried 212 (+ 80 °C) for two days. 213 Total solids (g kg-1) were calculated for FBBR and drum filter as: TS = (md - mt) / S x V / F, where md = 214 dried subsample mass (g), mt = container mass (g), S = sample size (l), V = total outflow volume (l d-1), F = 215 Feed intake (g d-1). Total solids (g kg-1) were calculated for the swirl separator as: TS = (md - mt) x 4 / F, 216 where md = dried subsample mass (g), mt = container mass (g), F = feed intake (g d-1). 217 2.6. Bioreactor nitrification rates 218 Bioreactor nitrification rates (g NOx h-1) were measured at the last week of the experiment, following 219 principles described by Jäntti et al. (2011). For the incubations, inlet water and carrier media were 220 collected from each bioreactor tank and transferred to the University of Jyväskylä. In the laboratory, 221 carrier media were divided into experimental vials (n = 30 per vial) with 360 ml inlet water, where 15NH4+ 222 was added (final concentration of 5 mg/L; 10—15 atm%). To ensure complete nitrification, the carrier 223 media was incubated for 3 hours at in situ temperature and under constant mixing by magnetic stirring 224 bars (150 rpm). To measure ammonium and nitrate concentrations and the stable isotope composition 225 of nitrite and nitrate, water samples were taken at the beginning of the experiment, and after 1.5 and 3 226 hours. Water samples were filtered with 0.2 um syringe filters and frozen immediately. Later, nitrate, 227 nitrite and ammonium concentrations were measured with a spectrophotometer (Lasa 100, Hach, 228 Loveland, USA). The stable isotope composition of nitrite and nitrate was measured using the denitrifier 229 method (Sigman et al., 2001). Briefly, 20 nmoles of sample NO2+3 were converted to N2O by cultured 230 denitrifying bacteria (Pseudomonas chlororaphis strain DSM 6698), which lack the enzyme responsible 231 for N2O reduction and the isotopic composition of N2O was measured using the IsoPrime 100 CF-IRMS 232 with a TraceGas preconcentrator interface. 233 2.7. Statistics 234 The effects of bioreactor design on nitrification efficiency, FCR, SGR, TGC, PSD and TS were analysed 235 using one-way ANOVA, and Tukey’s post hoc test was used for comparing the effects between 236 treatments, which takes the uneven sample sizes in the end of the experiment into account (Rusticus 237 and Lovato, 2014). A nonparametric Kruskall-Wallis test was used for total particle counts when 238 assumptions were not met for the parametric test. Effects of bioreactor design on water quality 239 parameters were analysed using Mixed ANOVA, where bioreactor design type (between subjects) and 240 measurement week (within subjects) were factors. The Bonferroni post hoc test was used for comparing 241 effects between treatments. For online measurements, daily average values were used. Statistical 242 analyses were done with SPSS (IBM SPSS Statistics, Armonk, USA) wherein 95 % confidence interval was 243 used. 244 3. Results 245 3.1. Fish growth and histopathology 246 No significant differences were found between treatments for FCR and SGR during the trial or TGC for 247 the whole experiment (Table 2). For the whole experiment, average feed loads were 30.09 kg (± 0.15 248 kg), 30.88 kg (± 0.35 kg) and 30.39 kg (± 0.28 kg) in the FF, FM and MM groups, respectively. 249 The most significant histopathological changes were noted in gill tissue (Table 1, supplementary 250 material). The severity scores for both lamellar epithelial cell hyperplasia (LEH) and focal branchitis were 251 slightly elevated at the beginning of the trial, for focal branchitis only in the FF group, and for LEH in all 252 groups including the flow-through system. 253 In kidney and liver tissue, no notable histopathological changes were seen during the experiment. The 254 PMAs noted during the experiment were mild to moderate, and no notable differences in their 255 occurrence over time, or differences between treatment groups or control group, were noted. The 256 inflammatory changes noted in this experiment were also minor and did not show any increase during 257 the course of the experiment. 258 3.2. Water quality 259 There was no difference in the TAN values between treatment groups, whereas nitrite values decreased 260 throughout the experiment in all groups. In the FF group, nitrite values were significantly higher than in 261 the FM and MM groups (P < 0.01). Nitrate values were higher in the MM group in comparison to the FF 262 (P < 0.01) and FM groups (P < 0.01) (Fig 2.). 263 Total organic carbon, UV254 and CO2 values were significantly different between the treatments (P < 264 0.01). TOC was lower in the FF group as compared to the FM and MM groups. The UV254 value was 265 lowest in the FF group, and highest in the MM group. CO2 concentration was highest in the FF group (P < 266 0.01), but there was no significant difference between the FM and MM groups (Fig. 3). 267 3.3. Total solids and PSD 268 The sum of total solids removed from the RAS units and solids removed from the swirl separators did 269 not differ between the treatments (Fig. 4). Solids removal by the drum filters was significantly affected 270 by the bioreactor systems (P < 0.01). In the RAS with two moving bed bioreactors, drum filters removed 271 solids the most, whereas in the RAS with two fixed bed bioreactors, solids removal by drum filters was 272 the lowest. In the FM group, drum filter solids removal was lower than in the MM group, but it was not 273 statistically significant (P = 0.051). 274 In fish tanks, particle size distribution values (β-value, total amounts, surface area and volume) were not 275 significantly different between the treatments, whereas differences were observed in water samples 276 taken after biofiltration. The β-values in water sampled after the second bioreactor was significantly 277 higher in the FF group compared to MM (P < 0.05), indicating that RAS with two fixed bed bioreactors 278 has a larger share of particles in small sizes (Table 3). Total particle amounts and surface area in the 279 biofiltered water were not affected by the treatments, whereas total particle volumes were significantly 280 higher in the MM group compared to the FF group (P < 0.05). Over 80% of the particles were below 3 281 µm in the FF and FM group and over 90% in the FF group (Fig. 1, supplementary material). 282 Although treatments with moving bed bioreactors had higher particle counts measured with the CASY 283 cell counter, the counts were not significantly different between the treatments due to high within-284 treatments variance (Kruskal-Wallis P = 0.24; Fig. 5). 285 3.4. Nitrification 286 The nitrification rate was significantly different between the treatments (P < 0.01). In the FF group, the 287 nitrification rate was lowest, but there was no difference between the MM group and FM group (P = 288 0.07). The nitrification rate did not differ between the first and second moving bed bioreactor, whereas 289 in the FF group, the second FBBR had a lower nitrification rate than the first FBBR (P <0.01) (Fig. 6). 290 4. Discussion 291 4.1. Fish performance 292 Wolf, J.C., Baumgartner, W.A., Blazer, V.S., Camus, A.C., Engelhardt, J.A., Fournie, J.W., Frasca, S.Jr, 506 Groman, D.B., Kent, M.L., Khoo, L.H., Law, J.M., Lombardini, E.D., Ruehl-Fehlert, C., Segner, H.E., Smith, 507 S. A., Spitsbergen, J.M., Weber, K., Wolfe, M.J. 2015. Nonlesions, Misdiagnoses, Missed Diagnoses, and 508 other interpretive challenges in fish histopathology studies: A guide for investigators, authors, 509 reviewers, and readers. Toxicol. Pathol. 43, 297–325 510 Ødegaard, H. 2006. Innovations in wastewater treatment: the moving bed biofilm process. Wat. Sci. 511 Technol. 53, 17–33 512 513 Figure 1. Schematic diagram of one RAS unit used in this experiment. BR = bioreactor, used as a fixed 514 bed (FBBR) or moving bed bioreactor (MBBR). 515 516 Figure 2. Mean total ammonium nitrogen (TAN) (A), nitrite-nitrogen (B) and nitrate-nitrogen (C) values 517 of the three RAS bioreactor designs ± SD. FF = Two consecutive fixed bed bioreactors (n=3 at weeks 1-8 518 and n=2 at weeks 9-14), FM = Fixed bed bioreactor followed by moving bed bioreactor (n=3) and MM = 519 Two consecutive moving bed bioreactors (n=3). 520 Table 1. RAS operational design and rainbow trout (Oncorhynchus mykiss) rearing conditions in the trial, 521 where different setups of fixed bed and moving bed bioreactors were studied. 522 Character istics Value Unit RAS unit (n=9) System volume 890 l Tank volume 500 l Relative water renewal rate 500 l kg - 1 feed Recirculation flow 15 l min - 1 Hydraulic retention time 5 – 8 d Tank hydraulic retention time 33 min Rearing conditions Fish density 19 – 82 kg m - 3 Feed quantity 0.22 – 0.45 kg d - 1 Average fish size 0.11 – 0.53 kg Bioreactor (n=2) Bioreactor water volume 125 l Bioreactor hydraulic retention time 8 min Carrier media volume 66 l Carried media area 49.5 m 2 Moving bed bioreactor air flow 15 l min - 1 Bioreactor hydraulic loading rate 436 l m - 2 d - 1 523 Table 2. Mean rainbow trout (Oncorhynchus mykiss) feed conversion ratio (FCR), specific growth rate 524 (SGR) (% bw d-1) and thermal growth coefficient (TGC) (± SD) during the trial (1 = days 0–27, 2 = days 28–525 55, 3 = days 56–92, 4 = 0–92) of the three RAS bioreactor designs. FF = Two consecutive fixed bed 526 bioreactors, FM = Fixed bed bioreactor followed by moving bed bioreactor and MM = Two consecutive 527 moving bed bioreactors (n=3, except when marked in asterisk, where n=2). 528 FCR SGR TGC Treatm ent 1 2 3 4 1 2 3 4 4 FF 0.85 ± 0.03 0.95 ± 0.04 *1.13 ± 0.03 *0.98 ± 0.01 2.36 ± 0.05 1.56 ± 0.02 *1.03 ± 0.02 *1.59 ± 0.00 *2.20 ± 0.01 FM 0.81 ± 0.01 0.91 ± 0.02 1.07 ± 0.03 0.95 ± 0.01 2.44 ± 0.04 1.58 ± 0.03 1.11 ± 0.03 1.64 ± 0.01 2.28 ± 0.03 MM 0.81 ± 0.02 0.95 ± 0.08 1.05 ± 0.04 0.95 ± 0.02 2.41 ± 0.03 1.54 ± 0.08 1.12 ± 0.02 1.62 ± 0.02 2.22 ± 0.05 529 530 Figure 3. Mean total organic carbon concentrations (TOC) (A), UV254 absorbance (B) and carbon dioxide 531 concentrations (C) measured online from the fish tank with UV/VIS spectrometer and CO2 probe of three 532 RAS bioreactor designs. FF = Two consecutive fixed bed bioreactors (n=3 at weeks 1-8 and n=2 at weeks 533 9-14), FM = Fixed bed bioreactor followed by moving bed bioreactor (n=3) and MM = Two consecutive 534 moving bed bioreactors (n=3). 535 536 Figure 4. Total solids removed from different water treatment steps proportioned into daily feed intake 537 of the three RAS bioreactor designs. FF = Two consecutive fixed bed bioreactors, FM = Fixed bed 538 bioreactor followed by moving bed bioreactor and MM = Two consecutive moving bed bioreactors 539 (n=3). FBBR = Fixed bed bioreactor. Mean values from two collection periods are presented (± SD). A 540 significant difference between treatments in drum filter backwash water is marked by different letters 541 (p < 0.01). 542 Table 3. Mean β-values, total particle counts, surfaces and volumes (± SD) at two sampling locations of 543 the three RAS bioreactor designs. FF = Two consecutive fixed bed bioreactors (n = 3), FM = Fixed bed 544 bioreactor followed by moving bed bioreactor (n = 3) and MM = Two consecutive moving bed 545 bioreactors (n=3). A significant difference between treatments is marked by different letters (p < 0.05). 546 Fish tank After 2nd bioreactor β Total counts (1.0x103 pcs ml-1) Total surface area (mm2 ml-1) Total volume (1.0x10-3 mm3 ml-1) β Total surface area (mm2 ml-1) Total counts (1.0x103 pcs ml-1) Total volume (1.0x10-3 mm3 ml-1) FF 3.7 ± 0.1 39.3 ± 8.3 0.7 ± 0.1 1.4 ± 0.1 4.1 ± 0.1 a 0.7 ± 0.2 40.5 ± 10.7 0.7 ± 0.2 a FM 3.7 ± 0.2 45.8 ± 14.1 1.2 ± 0.4 2.0 ± 0.7 3.8 ± 0.2ab 0.9 ± 0.5 35.6 ± 18.9 1.6 ± 0.5ab MM 3.6 ± 0.2 33.7 ± 13.9 1.0 ± 0.4 2.4 ± 1.0 3.6 ± 0.1b 1.3 ± 0.1 40.9 ± 4.2 3.2 ± 1.0b 547 548 549 Figure 5. Total particle counts of the three RAS bioreactor designs measured using the CASY cell counter. 550 FF = Two consecutive fixed bed bioreactors (n = 2), FM = Fixed bed bioreactor followed by moving bed 551 bioreactor (n = 3) and MM = Two consecutive moving bed bioreactors (n=3) ; ± SD of the most abundant 552 size classes, which were 1.0 µm for FF and 1.1 µm for FM and MM groups. There were no significant 553 differences between treatments. 554 555 Figure 6. Nitrification rate measured in the three RAS bioreactor designs measured using the stable 556 isotope labelling method. FF = Two consecutive fixed bed bioreactors (n = 2), FM = Fixed bed bioreactor 557 followed by moving bed bioreactor (n = 3) and MM = Two consecutive moving bed bioreactors (n=3). A 558 significant difference between treatments is marked by different letters and between different 559 bioreactors by different numbers (p < 0.01). 560