INTERACTIONS BETWEEN MICRO-PARTICLES AND THE REARING ENVIRONMENT IN RECIRCULATING AQUACULTURE SYSTEMS Ph.D. Thesis by PAULO MIRA FERNANDES May 2015 Section for Aquaculture National Institute of Aquatic Resources, DTU Aqua Technical University of Denmark Hirtshals, Denmark Colophon Interactions between micro-particles and the rearing environment in recirculating aquaculture systems By Paulo Mira Fernandes Ph.D. thesis Defended on the 26th of June, 2015 DTU Aqua – National Institute of Aquatic Resources Reference: Fernandes, P.M. (2015). Interactions between micro-particles and the rearing environment in recirculating aquaculture systems. Ph.D. thesis. Section for Aquaculture, DTU Aqua, Technical University of Denmark, Hirtshals, Denmark. 122 pp. Cover photo: Paulo Mira Fernandes (2013) 2 1. PREFACE This Ph.D. dissertation is submitted in partial fulfilment to attain the Doctor of Philosophy degree (Ph.D.). The work shown herein was undertaken during my enrolment as a Ph.D. student at the Section for Aquaculture, National Institute of Aquatic Resources (DTU Aqua), Technical University of Denmark, in Hirtshals, Denmark. The research was funded by the Danish Ministry of Higher Education and Science through a Danish innovation consortium titled Recirculation Technology for Future Aquaculture (REFA, Renseteknologier til Fremtidens Akvakultur). These past few years have been truly overwhelming, shaped by all the people whom I met, and without whom I would not have reached this stage. I could not have found my way without the immeasurable help and inspiration from my two supervisors. My deepest appreciation goes both to Per Bovbjerg Pedersen and Dr. Lars-Flemming Pedersen, whose meticulous approach, valuable insights, and enlightened ideas helped shape some of the most interesting results contained herein. To Erik Poulsen, Ole M. Larsen, Rasmus F. Jensen, Remko Oosterveld, Dorthe Frandsen, Ulla Sproegel, Brian Møller, and Sara M. Nielsen, goes also a huge thank you: this piece of (my) history could not have been completed without your precious assistance. I would also like to thank the contribution of Dr. Peter V. Skov on the discussion and analysis of potential interactions between particles and fish; and of Carlos Letelier-Gordo on his comments in the first drafts of this thesis. To all my Hirtshals colleagues, heartfelt thanks for the companionship, the long nights discussing hot topics, helping me descend waterfalls, and everything else that cannot be remembered now or described herein. You kept my working insanity sane. Outside of Hirtshals, I would like to thank all the people that I have met and made me who I am today. A huge thank you goes to, as my father once put it, the other three out of the Fab Four (Tomé, Sara, and Catarina): I would be done to the beef without you guys! Nuref, Steffen, Sofie, and all the other 1000 Fryders, thank you all for taking part in my journey. Last but not least, I would like to express my gratitude to the people who kept me motivated and calm: to my family and friends back home, I am glad you always pushed me to keep growing personally and professionally; to Dorthe, thank you for putting up with me and my madness Hirtshals, 4th of May, 2015 “However, I continue to try and I continue, indefatigably, to reach out. There’s no way I can single-handedly save the world or, perhaps even make a perceptible difference – but how ashamed I would be to let a day pass without making one more effort.” Isaac Asimov, 1988 (The Relativity of Wrong) 3 4 TABLE OF CONTENTS 1. PREFACE ............................................................................................... 3 2. LIST OF PAPERS ................................................................................... 7 3. LIST OF ABBREVIATIONS .................................................................... 9 4. DANSK RESUMÉ ................................................................................. 11 5. ENGLISH ABSTRACT.......................................................................... 13 6. OBJECTIVES ....................................................................................... 15 7. RECIRCULATING AQUACULTURE SYSTEMS (RAS) ....................... 17 8. INTERACTIONS BETWEEN MICRO-PARTICLES AND THE REARING ENVIRONMENT IN RECIRCULATING AQUACULTURE SYSTEMS .. 19 8.1 Fish waste production ........................................................................................19 8.2 Solids removal .....................................................................................................20 8.2.1 Primary clarifiers ...................................................................................................... 21 8.2.2 Drum filter efficiency ................................................................................................ 22 8.2.3 Drum filter mesh size effect ..................................................................................... 23 8.3 Removal of dissolved nitrogen ..........................................................................24 8.3.1 Substrate removal in biofilms................................................................................... 25 8.3.2 Factors affecting nitrification .................................................................................... 26 8.3.3 Particle interactions with biofilms ............................................................................. 28 8.4 Other factors interacting with particles.............................................................30 8.5 Particle Size Distribution (PSD) in RAS.............................................................32 8.5.1 β-value for aquaculture operations .......................................................................... 33 8.5.2 PSD stabilization in RAS ......................................................................................... 34 8.6 Micro-particles in the fish tank ..........................................................................36 8.6.1 Micro-particles as microbial substrate ..................................................................... 36 8.6.2 Interactions between micro-particles and fish.......................................................... 38 8.7 Conclusions and future perspectives ...............................................................39 9. BIBLIOGRAPHY ................................................................................... 43 10. PAPER I .............................................................................................. 63 11. PAPER II ............................................................................................. 73 12. PAPER III ............................................................................................ 85 ANNEX I .................................................................................................. 117 5 6 2. LIST OF PAPERS Paper I: Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B. 2014. Daily micro particle size distribution of an experimental recirculating aquaculture system – A case study. Aquacultural Engineering 60: 28-34. doi:10.1016/j.aquaeng.2014.03.007 Paper II: Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B. 2015. Microscreen effects on water quality in replicated recirculating aquaculture systems. Aquacultural Engineering 65: 17-26. doi:10.1016/j.aquaeng.2014.10.007 Paper III: Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B. 2015. Influence of fixed and moving bed biofilters on micro particle dynamics in an experimental recirculating aquaculture system. Manuscript. 7 8 3. LIST OF ABBREVIATIONS Symbol Description Unit Amedia Total active surface area of media AOA Ammonia oxidizing Archaea - AOB Ammonia oxidizing bacteria - A:V Area to volume ratio, as the total surface area divided by the total volume of the fractionated distribution of a particle sample BFT Bio-floc technology BOD5 Biochemical oxygen demand, after 5 days incubation CFB Cumulative feed burden, as amount of daily feed delivered per daily volume of make-up water COD Chemical oxygen demand of a raw sample C:N Carbon to nitrogen ratio in the water - DBL Diffusive boundary layer - FBB Fixed bed biofilter - FCR Feed conversion ratio, unit of feed given per unit of weight gain FT Flow-through system - HRT Hydraulic retention time d Lhyd Hydraulic loading rate, as water flow rate per cross-sectional area of the filter vessel MBB Moving bed biofilter m2 1/m mg O2/L kg feed/m3 of make-up water mg O2/L g/g m3/m2/d 3 m /d MUW Make-up water NOB Nitrite oxidizing bacteria - PSD Particle size distribution - RAS Recirculating aquaculture system - SSA Specific surface area of biofilter media TAN Total ammonia-ammonium nitrogen (NH4+-N+NH3-N) TSS Total suspended solids WWTP Wastewater treatment plant - β-value Beta value, as the shape of the particle distribution after applying a double logarithmic transformation - m2/m3 mg N/L mg/L 9 10 4. DANSK RESUMÉ Fiskeopdræt i recirkulerede systemer (RAS) indebærer en række fordele, en af disse er muligheden for en konstant produktion under stabile forhold året rundt. I modsætning til åbne gennemstrømningsanlæg giver RAS mulighed for optimeret vækst og for reduceret miljøpåvirkning, ligesom f.eks. risikoen for udslip er elimineret. Disse fordele er et resultat af et lukket system med mulighed for kontrol af vandkvalitet på grund af de tilhørende renseforanstaltninger og -komponenter. Disse foranstaltninger er dog endnu ikke fuldt optimerede og især samspillet mellem de enkelte komponenter og deres funktion er af afgørende betydning for optimering af fiskeproduktionen. De bedste og mest effektive mekaniske rensekomponenter i RAS kan opnå en fjernelsesgrad på op mod 90-95% af det partikulære produktionsbidrag over 30 µm. På den anden side, skaber dette baggrund for en partikelfordeling i anlægsvandet hvor næsten alle partikler er under denne størrelse. Forøget vandskifte er ikke en reel mulighed for at reducere eller kontrollere mikropartiklernes antal, og de vil derfor typisk have en lang opholdstid i opdrætssystemet. Udover produktionsbidraget kan partikler også blive genereret i selve systemet, således som det er påvist i såvel adskillige RAS som i gennemstrømningsanlæg. Enhver komponent eller element som skaber turbulens, som f.eks. pumper eller faldende vand, producerer mikro-partikler via nedbrydning af større partikler. Fiskestørrelse og fodersammensætning samt flere andre elementer er ligeledes blevet påvist af have betydende indflydelse på partikler og disses størrelsesfordeling. På den anden side kan eksempelvis biofilter og døde zoner fjerne partikler via aflejring og sedimentation. Det er vigtigt fortsat at identificere komponenternes indflydelse på partikelstørrelsen, således at partikelfjernelsen kan blive yderligere forbedret. Akkumulering af mikro-partikler kan påvirke fisk og biofilter negativt i RAS. Når partikelstørrelsen mindskes, sker den en relativ øgning i overfladearealet af partiklerne og dermed af kontaktarealet mellem partikler og det omgivende vand. Dette medfører en forøget mulighed for bakterie-vedhæftning og -vækst og dermed også risiko for opformering af potentielt skadelige bakterier, tilstopning af gæller, pumper og filtre samt udvaskning af organisk stof og næringsstoffer fra partiklerne. Udbrud af sygdomme er blevet relateret til akkumulering af partikler i RAS ligesom biofilterfunktionen er påvist at blive reduceret når belastningen med organisk stof medfører, at C:N forholdet overstiger 1:1. Derved reduceres nitrifikationsprocessen i biofilteret idet de nitrificerende bakterier udkonkurreres af de heterotrofe. Dette er påvist for så vidt angår opløst organisk stof, mens betydningen af fin-partikulært materiale endnu ikke er fuld belyst. Formodentlig vil akkumulerede mikro-partikler i biofilteret primært være problematisk under særlige forhold eller driftsbetingelser, men der mangler generelt viden indenfor området. Denne PhD-afhandling omfatter 3 videnskabelige artikler (2 publicerede og 1 under indsendelse) samt en del ikke-publicerede data fra forskningsarbejdet gennem de sidste 3 år. De tre artikler omhandler: 1) døgnvariation af mikro-partikler på forskellige steder i RAS, 2) betydningen af mikrosigte og dennes maskevidde for partikelfordeling og generelle vandkvalitetsparametre i RAS, 3) produktion eller fjernelse af partikler og organisk stof via biofiltre med bevægeligt henholdsvis fast medie. Studierne blev alle gennemført i veletablerede, modne RAS som blev drevet under konstante betingelser vedrørende indfodring og vandskifte (0.1-3.1 kg/m3), svarende til normal drift på semi-intensive opdrætsanlæg i Danmark. I alle studier blev effekten af ændringer i system eller opsætning på partikler og partikelfordeling først gennemført når anlæggets baggrunds- eller basisniveauer var konstante og reproducérbare. 11 I den første artikel (I) blev der i et RAS under relativ lav belastning (0.1 kg foder/m3 vandskifte), undersøgt partikelfordeling (PSD) på en række steder i anlægget gennem 24 timer. Det blev påvist, at partikel-koncentration og -fordeling var stabile og ensartede gennem anlæggets komponenter og -tid. Anlægget var blevet kørt under konstante betingelser og belastning gennem 1 uge forud for prøvetagningen. Overordnet kunne det konstateres, at den relativt lave belastning og et rimeligt internt vandflow på 1,25 gange/time resulterede i en nærmest steady-state situation hvor hverken partikelkoncentrationen eller –fordelingen varierede signifikant mellem de forskellige udtagssteder i anlægget eller tidspunktet på døgnet. Denne steady-state vil være bestemt af drift og design af det enkelte RAS. I artikel II blev betydningen af maskevidden i mikrosigten (100, 60 eller 20 µm) for partikelstørrelsesfordelingen sammenlignet med anlæg uden mikrosigtedug i en triplicat opstilling (12 anlæg i alt). Alle anlæg blev kørt under konstant belastning (3.1 kg foder/m3 vandskifte) i 6 uger efter at anlæg og biofilter var modne og stabile, hvorefter sigtedug blev installeret. Efter 3 ugers forsøg, begyndte de partikulære faktioner at blive stabile i anlæg med mikrosigte mens de fortsatte med at stige/akkumulere i anlæg uden. Anlæg med 20 og 60 µm nåede ligevægt i uge 3 mens anlæg med 100 µm begyndte at blive stabile i uge 4. Efter 6 ugers drift var betydningen af mikrosigte åbenlys, idet koncentrationen af de partikulære parametre var ca. 30 % mindre end hvad der kunne noteres i anlæg uden mikrosigte. En konstant belastning og replicerede anlæg kørt under ensartede, konstante betingelser med et internt vandflow på 1,75 gange i timen producerede ensartet partikelfordeling i alle anlæg. Partiklerne var alt-overvejende små, mindre end 20 µm, hvilket medvirker til at forklare hvorfor der efter 6 ugers drift ikke var nævneværdig effekt af at anvende en dug på 20 µm i forhold til en på 100 µm. I den tredje artikel (III) blev effekten af fast (fixed-bed) og bevægeligt medie (moving-bed) biofiltre på partikelfordeling og -mængde undersøgt under kontrollerede betingelser og fast belastning (1 kg foder/m3 vandskifte). Den konstante belastning tillod, at den ene eller den anden type biofilter blev frakoblet systemet og det recirkulerede kredsløb således at effekten af biofiltertype kunne bestemmes, uden at den relative belastning på anlægget fra foder eller fisk blev ændret. Der blev påvist tilbageholdelse af partikler i fixed-bed filtre mens moving-bed filtre forøgede partikelbelastning i systemet, idet større partikeler blev ødelagt og nedbrudt til mindre. Netto-fjernelsen af organisk stof skete med samme rate i begge typer filter, men moving-bed fjernede netto mere af den partikulære fraktion, hvorimod fixed-bed netto fjernede mere af den opløste fraktion. Mekanisk påvirkning af partiklerne via beluftning og bevægelse i moving-bed filterene samt mediets struktur i fixed-bed filtrene er årsag til forskellene i partikler og disses fordeling. Forskellig mikrobiel struktur kan have haft betydning for fjernelsen af organisk stof. Under forsøgsgangene, som førte til artikel II og III, blev der også genereret data til undersøgelse af bl.a. 1) eventuel histo-patologisk effekt af mikro-partikler på regnbueørredernes gæller 2) sammenhæng mellem mikropartikler og biofilterkinetik og 3) sammenhæng mellem mikropartikler og mængde af frit-svømmende bakterier. Der blev ikke påvist nogen oplagte sammenhænge via disse undersøgelser, og derfor er data ikke publiceret her, men de potentielle interaktioner diskuteres desuagtet i kapitler i afhandlingen. Samlet set ser det således ud til, at partikelparametre opnår en steady-state i RAS under konstante betingelser med ens belasting og vandskifte og med relativt højt internt vandskifte (mindst 1,25 gange/time). Påvirkning af partikelfordelingen kan opnås gennem design og installering af forskellige rensekomponenter, men betydning og interaktion mellem partikler og fisk, partikler og bakterier samt partikler og biofilterfunktion bør undersøges nærmere i fremtidige studier. 12 5. ENGLISH ABSTRACT Recirculating aquaculture systems (RAS) have the advantage over other aquaculture systems in terms of stable year-round fish production. Contrary to inland flow through systems or net pen operation, RAS allow for better fish quality and growth, while minimizing the risk of fish escapees. This is derived from the enclosure of the rearing environment, and the installation of water quality control and waste treatment devices. However, waste removal processes are not fully optimized, and the interactions between several of the waste treatment units and their output are of paramount importance for optimal fish growth and performance. The most efficient solid removal devices in freshwater RAS remove about 90-95% of the solid waste above 30 µm in size. Conversely, this creates background particle distributions comprised mainly of solids with diameters below this range. Since increasing the water exchange rate in RAS may not be a possibility for micro-particle control, this type of particles will often have a long residence time within the system. Particles can also be produced within the system, as has been identified in several RAS and flow through systems. In general, any element that generates turbulence, such as pumps or waterfalls, produces micro-particles by disintegration of larger particles. Fish size, diet composition, and other farm components, such as degassing units and biofilters, have also been identified as change promoters in particle size or concentration. It is essential to continue identifying components that have an effect on mean particle size, so that solids removal can be further optimized. Micro-particle accumulation can impair fish and biofilter performance in RAS. As particle size decreases, there is a concomitant increase in the relative contact area between the particle and the water. Practically, this means that there is a greater potential for pathogen adhesion, clogging of fish gills, and leaching of associated nutrients, i.e. organic matter and nitrogen. Fish disease outbreaks have been associated with particle accumulation in RAS, while loading of organic matter on a carbon to nitrogen ratio (C:N) above 1:1 impairs nitrification due to out-competition of nitrifiers by heterotrophic bacteria. This is true for dissolved organic matter; yet, the interactions between biofilters and particulate organic matter in RAS are still not fully described. Presumably, accumulated micro-particles will become a problem only under specific conditions of biofilter size or mode of operation, and fish life stage, although more information is needed on the topic. The present thesis is accompanied by three scientific articles, and unpublished data acquired during the last three years. The three articles are related to 1) daily distribution of micro-particles in RAS; 2) the effects of microscreen mesh size on micro-particles and general water quality parameters in RAS; and 3) production or removal of particles and organic matter by fixed and moving bed biofilters in RAS. The studies were conducted in matured RAS operated under constant conditions of feeding and make-up water (0.1-3 kg/m3), reflecting normal operational conditions for semi-intensive RAS in Denmark. In all studies, the effects on particles related to changes in system components or configuration, were only assessed when system background levels demonstrated reproducibility over consecutive sampling dates. In paper I, in a RAS operated at a low cumulative feed burden (CFB) (0.1 kg feed/m3 make-up water), particle size distribution (PSD) measurements at several locations during a 24-h period, demonstrated the stabilization of particle concentration and distribution parameters. The system was operated under constant conditions of CFB for a week prior to the beginning of the sampling period. Overall, a relatively low feeding level and the internal water turnover rate (1.25 times/h), 13 steered the PSD towards a quasi-steady-state situation, where neither the concentration, nor the shape of the distribution varied significantly according to sampling location or time of the day. The effect of mesh size (100, 60 and 20 µm) on PSD, compared to a group without microscreen, was assessed in replicated RAS, as shown in paper II. Triplicate RAS for each group were operated under constant CFB conditions (3.1 kg feed/m3 make-up water) for 6 weeks after the biofilter was mature, as defined by the efficient and constant conversion of ammonia into nitrate. After 3 weeks of operation, solid waste parameters started to stabilize in groups with microscreens, but continued to accumulate in the group without microscreen. The 20 and 60 µm microscreen groups reached equilibrium at week 3, while the 100 µm group started to stabilize after week 4. After 6 weeks of operation, the effect of microscreen presence was apparent, as solid waste parameters were approximately 30 % lower than the waste amounts observed in the group without microscreens. A constant CFB, and replication of the same high internal water turnover rate (1.75 times/h), produced similar PSDs in all systems. These were mostly comprised of microparticles smaller than 20 µm, which helps explaining why the effect of a 20 µm microscreen was not different from the effect of a 100 µm microscreen after 6 weeks of operation. In paper III, the effects of fixed and moving bed biofilters on PSD were assessed in a RAS operated under controlled CFB (1 kg feed/m3 make-up water). Particle retention was observed in fixed beds, while moving beds increased the system particle load by disintegration of large particles. Net removal of organic matter occurred at the same rates in both modes of operation, although moving beds removed more of the particulate fraction, and fixed beds removed more of the dissolved fraction. Mechanical stress, induced by aeration in moving beds, and the distribution of the media in fixed beds, caused the observed trends in PSD. During the experiments related to paper II and paper III, data was also acquired in order to study the interrelationships between micro-particles and fish gill histopathology; between microparticles and biofilter kinetics; and between micro-particles and suspended bacteria abundance and activity. There were no clear correlations, and so, the data is not shown in this thesis. Nevertheless, the potential interactions are discussed in detail in specific chapters. In conclusion, it seems that under constant conditions of feed and make-up water, and an internal water circulation of minimum 1.25 times/h, particulate parameters reach a steady-state. This steady-state is related to system set-up and system operation. Hence, manipulation of the system PSD can be achieved through the installation of different components and devices, such as pumps, mechanical filters or biofilters. The scope of the interactions between particles and fish, bacteria in suspension, or biofilters still needs to be addressed further. 14 6. OBJECTIVES The main objective within this Ph.D. study was to assess and determine particle accumulation and removal in RAS. Focus was given to the chemical and physical description of particulate matter according to different RAS operational parameters and setups. Different RAS water treatment components were tested in relation to their effects on particulate waste, and the consequences on fish and system performance were monitored. Since particle accumulation in RAS is a recent topic of study, initial particle measurement results also served as a means to establish a new in-house protocol for particle determination. The research was divided into three parts related to: (1) sampling of PSD; (2) control and manipulation of PSD; and (3) interactions between the background PSD and biofilter, fish performance, or microbial abundance and activity. 1. On a first basis to this Ph.D., methods for particle sampling in RAS were tested (paper I). The assessment comprised a 24-h sampling campaign at several locations within an experimental RAS operated at a low CFB. The outcome from this study had implications on subsequent sampling protocols. 2. The second part of this Ph.D. was related to the influence of different RAS components on PSD, and remaining water quality parameters. The tested RAS components were chosen based on recent discussions in the Danish aquaculture sector, and reflected either what is currently being applied, or what the industry is considering for the future. Different microscreen mesh sizes effects on particulate matter were tested in triplicate RAS (paper II) and thereafter the interactions between biofilter mode of operation (fixed vs. moving bed) and particles were tested in an experimental RAS (paper III). 3. The interactions between micro particles and fish, bacterial activity, or biofilters, were assessed while conducting the experiments related to paper II and paper III. Although the data acquired in this section was not published, the interdependency and correlations between particles and biofilters, free-swimming bacteria, or fish are discussed in sections 8.3.3, 8.6.1, and 8.6.2, respectively. Three experimental questions, serving as the basis for the scientific articles contained in this thesis, were sought out during this Ph.D. study: Paper I: Do micro-particles reach equilibrium in RAS operated at stable experimental conditions? Paper II: Does microscreen mesh size influence the particle size distribution (PSD) in RAS? Paper III: Does biofilter mode of operation affect PSD in RAS? All experiments were conducted in smaller-scale RAS conditions similar to Danish industrial setups. Furthermore, all the studies were carefully designed and conducted at controlled operating conditions of water quality, feed loading or cumulative feed burden (CFB), and water renewal that included maturation and steady-state periods that allowed replication when necessary. 15 16 7. RECIRCULATING AQUACULTURE SYSTEMS (RAS) Recirculating aquaculture systems (RAS) are fish culturing systems where water is partially reused after being treated (Liao and Mayo, 1972; Bovendeur et al., 1987; Summerfelt et al., 2004). The principle behind RAS is to recycle water containing fish produced-waste through a series of treatment units designed for wastewater treatment plants, and refurbished for fish farming applications (Fig. 1). Due to the enclosure of the rearing environment and waste treatment possibilities (Cripps, 1994; Cripps and Bergheim, 2000; Piedrahita, 2003), RAS advantages over other aquaculture systems include reduced water consumption and nutrient discharge (Verdegem et al., 2006; Verdegem, 2013), improved fish welfare and conditioning due to enhanced water quality control, and better hygiene and disease management (Liltved and Cripps, 1999; Sharrer et al., 2005; Summerfelt et al., 2009; Park et al., 2011), and reduced risk of biological impact from, e.g., fish escapes or parasites (Liltved and Hansen, 1990; Naylor et al., 2000; Martins et al., 2010). Fig. 1. A conceptual process flow scheme for a traditional recirculating aquaculture system (RAS), with effluent control as end-of-pipe treatment. As the fish farming industry progresses and practices intensify, so do the technological advances. It is possible to apply different types of waste removal devices, and if necessary, effluent treatment devices. RAS intensification is usually determined by a so-called recirculation ratio (R-ratio, Leonard et al., 2002) or cumulative feed burden (CFB, Colt et al., 2006) defined by the amount of daily ingested feed by the fish, in relation to the daily replacement of water volume (kg feed/m3 make-up water). It follows that as the CFB increases, more waste needs to be treated to maintain fish water quality requirements. System categorization according to CFB operated is recognized as: flow-through systems (< 0.04 kg/m3); re-use (0.04-1 kg/m3); conventional RAS (1-5 kg/m3); and fully recirculating aquaculture systems (FREA, > 5 kg/m3) (Jokumsen and Svendsen, 2010; Martins et al., 2010). In addition to improved water quality control, it may be necessary to regulate the effluent levels, though this is usually only necessary in higher intensity systems (van Rijn, 1996, 2013). A general guideline for intensification of aquacultural practices and water treatment requirements based on Sindilariu (2007), Jokumsen and Svendsen (2010), and Timmons and Ebeling (2010) is shown in Fig. 2. 17 As part of an incomplete digestive process, fish excrete un-assimilated (dissolved) and undigested (solid) nutrients derived from the feed (Bureau and Hua, 2010; Dalsgaard and Pedersen, 2011). In order to prevent their accumulation in RAS and potential interference with fish performance, it is required that these are removed from the rearing environment (van Rijn, 1996, 2013; Schneider et al., 2005, 2007). It is also possible to regulate other parameters such as oxygen and pH (e.g. Colt and Watten, 1988; Loyless and Malone, 1997; Summerfelt et al., 2000; Colt, 2006), although control methods and requirements in RAS are beyond the scope of this thesis. Fig. 2. General guideline for intensification of practices in aquaculture, and water treatment requirements or strategies, dependent on the feed loading (CFB) operated. RAS – Recirculating Aquaculture Systems; FREA – Fully Recirculated Aquaculture Systems. Adapted from Sindilariu (2007), Jokumsen and Svendsen (2010), and Timmons and Ebeling (2010). The largest part of dissolved waste comprises nitrogenous compounds in the form of ammonia (NH3), ammonium (NH4), and urea (CO(NH2)2) (Bureau and Hua, 2010; Dalsgaard and Pedersen, 2011; Dalsgaard et al., 2015), and are removed from the water through the microbial N-cycle (Sharma and Ahlert, 1977; Hagopian and Riley, 1998). To overcome potential fish performance limitations due to the accumulation of NH3 and NH4, the growth of nitrifying organisms has to be supported in the system, i.e., in biofilters (Gutierrez-Wing and Malone, 2006; Malone and Pfeiffer, 2006). Biofiltration is a key process in RAS water quality management, and requires the maintenance and optimization of several water quality parameters, such as oxygen, alkalinity, and suspended solids concentration (Szwerinski et al., 1986; Eding et al., 2006). In freshwater RAS, efficient removal of solid particles above 30-60 µm can be relatively easily attained by installing a mechanical filter (e.g. a drum filter), while the rest of the solid waste produced within the system will be retained (e.g. Langer et al., 1996; Davidson and Summerfelt, 2005). Micro-particles hydraulic residence time in RAS is inversely proportional to the removal efficiency and water exchange rates (Chen et al., 1993; Patterson et al., 1999; Pfeiffer et al., 2008), which may be problematic for fish or biofilter performance. 18 8. INTERACTIONS BETWEEN MICRO-PARTICLES AND THE REARING ENVIRONMENT IN RECIRCULATING AQUACULTURE SYSTEMS The following chapter follows the scheme flow presented in Fig. 3, reflecting the processes by which solids in recirculating aquaculture systems (RAS) are (1) produced; (2) removed; and (3, 4) transformed. As solids progress through RAS, the particle size distribution (PSD) may change according to some specific parameters that mostly generate micro-particles (5), and may have a consequence on (3) biofilter, and (6) fish performance or microbial communities in suspension. Fig. 3. Pathway of particle production (1), removal (2), and transformation (3, 4) in a RAS, according to the main factors affecting particle size. The particle distributions after the recirculation loop (5) have potential interactions with fish and suspended bacteria when returned to the fish tank (6). 8.1 Fish waste production The bulk of the waste produced during aquacultural operations can be attributed to uneaten feed and fish excreta. Good management practices keep uneaten feed waste at a virtually negligible level (1-4 % of feed remains uneaten) (Reid et al., 2009; Bureau and Hua, 2010). Fish waste, on the other hand, is produced daily and needs to be treated in order to maintain good water quality within any aquacultural system, and to minimize effluent load (Cripps, 1994; Schneider et al., 2005; Wik et al., 2009). Fish produce waste at a rate equivalent to the feed conversion ratio, FCR (amount of feed given/amount of body mass gain, g/g). This ratio can be influenced by three groups of factors related to (1) feed and feed constituents (Heinen et al., 1996; Dalsgaard and Pedersen, 2011; Unger and Brinker, 2013); (2) fish species, genetics, and size (Clark et al., 1985; McKenzie et al., 2007); and (3) water quality (Roque d’Orbcastel et al., 2009; Davidson et al., 2013). FCR optimization is attempted through the inclusion of high-quality nutrients (or binders) at a balanced ratio between them that maximize nutrient uptake (Brinker, 2007, 2009; Dalsgaard et al., 2009; Meriac et al., 19 2014). At the same time, fish species, age, and genetics help define the FCR of a specific fish cohort (Summerfelt et al., 2004), while deteriorated water quality may severely affect FCR, due to poor fish conditioning and welfare (e.g. Martins et al., 2009a,b) At a good (low) FCR, more nutrients will be assimilated by the fish, thus leading to a smaller waste production (Heinen et al., 1996; Summerfelt et al., 2004; Davidson et al., 2013). The objective then is to optimize feeds that maximize fish growth, while minimizing waste production and associated treatment costs. As the industry progresses, improvements in feed formulation and composition have significantly lowered average FCRs in aquaculture to around 1 g/g. Even then, however, fish still produce waste, and this occurs in dissolved or solid form. The major fish waste sources that require treatment are dissolved nitrogen, and solid waste. - - In salmonids holding an FCR around 1g/g, solid waste is usually produced at a ratio of 250 g/kg feed given, as total suspended solids (TSS) (Bergheim et al., 1993; Chen et al., 1997; Timmons and Ebeling, 2010), or 290 g/kg feed consumed, as particulate chemical oxygen demand (COD) (Dalsgaard and Pedersen, 2011). Dissolved waste is mostly in nitrogen form, as total ammonia nitrogen (TAN = NH3+NH4+) or urea, while dissolved carbon and phosphorous waste account for less than 20% of the total dissolved excreta. In salmonids holding an FCR around 1g/g, urea comprises 10 % of the excreted dissolved N, while TAN excretion is around 80 % and occurs at a ratio between 30 and 40 g TAN/kg feed consumed (Bureau and Hua, 2010; Dalsgaard and Pedersen, 2011; Dalsgaard et al., 2015). Urea is not necessarily toxic at the concentrations observed in aquaculture productions, but ammonia can be toxic at very low levels (Thurston et al., 1981; Randall and Tsui, 2002). Although it is possible to predict the mass budget of produced dissolved and solid waste, it is difficult to predict the average size of the solids excreted by the fish (Chen et al., 1993; Brinker et al., 2005; Brinker, 2009). Regardless, solids produced by fish in aquaculture vary greatly in shape and size (Fig. 4) and the end distribution is highly determined by the consumption and digestibility of some feed constituents (Reid et al., 2009; Unger and Brinker, 2013; Meriac et al., 2014, 2015). Fig. 4. Size distribution and size class nomenclature of aquacultural solids. From Timmons and Ebeling (2010). 8.2 Solids removal To optimize water quality and to minimize fish performance limitations due to the accumulation of solids in RAS, particulate organic matter has to be removed quickly and gently from the fish culturing units. This is done through the bottom drain of the tank and can be optimized by installing a dual-drain e.g. a bottom outlet that handles a high sludge concentration in little flow (1520 % of the total flow) and a lateral outlet with a diluted waste flow (Timmons et al., 1998; Davidson and Summerfelt, 2004; Summerfelt et al., 2004; Summerfelt and Penne, 2005; Masaló 20 and Oca, 2013). Through the bottom flow, the bulk of the treatable solids can be quickly directed to solids removal devices. Solid removal devices vary greatly in shape, size and specificity, and their effectiveness is dependent on the effect each device has on a particular solid size and inherent characteristics of that size range (Fig. 5). The devices in practice are optimized based on weight, size, or chemical composition of the solids (Cripps and Bergheim, 2000; Piedrahita, 2003; Sindilariu, 2007; Couturier et al., 2009), but the most common primary treatment methods in RAS are rotating microscreens (e.g. drum filters) and sedimentation units. Fig. 5. Common solids removal devices applied, and the particle size range affected. Adapted from Cripps and Bergheim (2000). 8.2.1 Primary clarifiers On a solid weight-basis, sedimentation units (Henderson and Bromage, 1988; Merino et al., 2007; Bergheim and Brinker, 2003; Unger and Brinker, 2013), and hydrocyclones (Davidson and Summerfelt, 2005; Veerapen et al., 2005; Pfeiffer et al., 2008; Lee, 2014) have showed great efficiency in removing solids larger than 100 µm. Some settling of particles below 100 µm can occur, but sedimentation methods are not sufficiently efficient in removing this type of particles (Hussenot, 2003; Sindilariu, 2007). These methods are based on the natural density of the solids, and imply that their density-defined weight will generate drag forces that cause the particles to settle in proportion to their size, and according to Stokes’ law (Chen et al., 1993; Patterson et al., 2003; Unger and Brinker, 2013; Wong and Piedrahita, 2000). Stokes’ drag forces are stronger in bulkier, larger particles, thus, causing them to settle in a shorter period of time. On the other hand, smaller particles generate weaker drag forces in in laminar flow, which exponentially increases their settling velocity (Stokes, 1901). Micro screening is defined as the physical blockage of particles in a mesh with a specific pore size (Mäkinen et al., 1988; Bergheim et al., 1993; Ljunggren, 2006). Several filters are associated with the physical entrapment of particles in aquaculture, though drum filters are the most common option for solid waste removal. This can probably be attributable to the fact that drum filters require little space, minimal maintenance, and cause minimal water head loss (Sindilariu, 2007; Timmons and Ebeling, 2010). In drum filters, a rotational mesh configuration allows the operation of a secluded area inside the drum where the mesh can be cleaned by water (backwashing), and drive solids out of the system (Ali, 2013; Dolan et al., 2013). Drum filters usually remove 70-90% of solid waste as TSS (Fig. 6), and are most efficient with particles larger than 30 µm (Cripps, 1994, 1995; Kelly et al., 1997; Couturier et al., 2009). Particle removal by microscreens is a product of 21 loading conditions and particle size, though caking and clogging of the filter mesh can enhance the filtration of solid waste (Tien et al., 2001; Leiknes et al., 2006; Iritani et al., 2012). Fig. 6. Drum filter (mesh sizes 60-90 µm) removal efficiency according to TSS inlet concentration. From Timmons and Ebeling (2010). 8.2.2 Drum filter efficiency To facilitate waste removal calculations, particles are usually assumed spherical, though this is not always the case (Patterson and Watts, 2003a; Brinker et al., 2005c). In this sense, average removal is usually underestimated when selecting a specific mesh size. Furthermore, there is always around 15 % uncertainty related to average drum filter mesh size (Langer et al., 1996; Ljunggren, 2006; paper I). This means that a mesh with a theoretical pore size of 100 µm, in fact, ranges between 85-115µm, though the uncertainty can lead to even wider gaps. In turn, what is found after the drum filter is usually a combination of particles of different sizes, both larger and smaller than the operated pore size (Fig. 7). B 100 Cumulative volume size distribution (%) A Pore size (µm) 0 0 Particle size (µm) 150 Fig. 7. A) Theoretical particle separation efficiency as a function of microscreen pore size (from Ljunggren, 2006). B) Particle distribution after a 60 µm drum filter (adapted from paper I). Sub-optimal solids removal by drum filters at low inlet TSS concentrations (Fig. 6), is likely caused by the dependency of these concentrations to small particles (Bergheim et al., 1993; Chen et al., 1993; Brinker and Rösch, 2005; Brinker et al., 2005c). Although downsizing of a drum fil22 Mesh size (µm) 100 1 Air percentage vs. Mesh size Air percentage vs. Flow handling capacity 80 60 0.8 0.6 y = 8.950.054x R² = 0.99 0.4 40 0.040.066x y= R² = 0.95 20 0.2 Flow handling capacity (L/s) ter mesh size could be attempted to increase overall efficiency, it would exponentially raise the investment and running costs of the drum filter (Timmons and Ebeling, 2010). The capacity of a mesh is defined not only by the pore size itself but also by the thickness of the thread used and the air percentage that both factors categorize (Fig. 8). In this sense, a reduced pore size generally limits the flow handling capacity of a mesh, since it is normal to observe lower air percentages in smaller pore sizes (Ali, 2013; Dolan et al., 2013; paper II). Therefore, to handle the same flow, a drum filter with a reduced mesh air percentage would require a larger mesh surface area, and by that a larger drum would be required. 0 0 0 10 20 30 40 50 Mesh air percentage (% of surface area) Fig. 8. Microscreen mesh size and flow handling capacity according to mesh air percentage. Adapted from Dolan et al. (2013). 8.2.3 Drum filter mesh size effect As solids removal is a function of solids loading and particle size distribution, it would seem that utilizing the narrowest microscreen mesh size would render the lowest particle concentration (Cripps, 1994; Timmons and Ebeling, 2010). However, and despite the fact that different mesh sizes potentially have different effects on the removal of solid waste, there appears to be a mesh size threshold for drum filters in aquaculture (Cripps, 1995; Kelly et al., 1997; Patterson and Watts, 2003a). In operations with less than 25 % variation in daily CFB (Chen et al., 1997; Couturier et al., 2009; Timmons and Ebeling, 2010), the background solids concentration in a system will vary in a limited way. In this sense, the effects of recirculation will potentially shadow the effects of the solid removal devices, by generating PSDs mostly comprised of small particles (Patterson et al., 1999, 2003). Alternatively, removal could be stimulated by caking events that enhance the entrapment of particles by decreasing the mesh pore size (Tien et al., 2001; Leiknes et al., 2006; Iritani et al., 2012). Caking is defined as the formation of a visco-elastic layer of particles in the mesh that is caused by water-driven particle compression. Nevertheless, caking events usually lead to head loss and increased backwashing periodicity, which elevate operational costs and should generally be avoided (Ali, 2013; Dolan et al., 2013). However, the three factors, stable CFB, recirculation and caking, should help in predicting and improving drum filter removal efficiency. Other factors could also play a role in defining the overall solids removal efficiency. Studies on the characterization of unfiltered solids in aquaculture, have shown that solids distribution do not vary significantly below 100 µm (Cripps, 1995; Kelly et al., 1997). In this sense, applying a drum 23 filter with a narrower pore size may not significantly improve water quality. In intensive RAS operated with microscreens at a stable CFB (3.1 kg feed/m3 MUW), it seems that there is a threshold between 60-100 µm mesh size (Fig. 9), below which solid waste removal is not significantly improved (Patterson et al., 2003; paper II); in wastewater, the threshold stands between 250-500 µm (Rusten and Ødegaard, 2006). Fig. 9. Total suspended solids (TSS) concentration after 6 weeks of operation in triplicate RAS operated at a stable CFB (3.1 kg/m3 of make-up water). The RAS were operated with no microscreen (control), or with 100, 60, or 20 µm microscreens. * - Statistical significance (p < 0.05) compared to the control group. From paper II. If a 60 µm (or 100 µm) mesh is capable of handling the same waste as a 20 µm microscreen, then there is an attenuation of the need for a decreased pore size. This would significantly decrease investment and operational costs of commercial RAS, as related to the reduction in water head loss, and in energy consumption and backwashing requirements, leading also to improved dry matter content of the sludge (Ali, 2013; Dolan et al., 2013). However, the extrapolation of these results (paper II) should be cautiously made, as: - Experiments under stable and controlled conditions do not always reflect the daily variation that may occur in a farm (Couturier et al., 2009; Emparanza, 2009; Timmons and Ebeling, 2010); Finer mesh sizes may also allow improved system stability during fluctuations in particle load as often happens in commercial operations (Bergheim et al., 1993; Chen et al., 1997; Kelly et al., 1997); The capacity of a 10-40 µm mesh to remove parasites from natural water bodies should not be ignored, as it may be a requirement in farms extracting water from infested streams (Liltved and Hansen, 1990; de Kinkelin et al., 2002; Sitjà-Bobadilla et al., 2005; Sanz et al., 2009). Whether the limited or finite removal of solid waste by drum filters in freshwater RAS can or will be optimized by intrinsic or extrinsic methods, there is still cause for concern regarding accumulation of micro-particles in RAS, and their potential interactions with fish, the microbial community in suspension, and biofilters. 8.3 Removal of dissolved nitrogen To overcome potential fish health limitations related to the accumulation of ammonia, RAS are equipped with reactors filled with elements overgrown by bacteria that convert ammonia into less toxic products. Biofilters vary greatly in shape and format, and are equipped with different types of media that support the growth of heterotrophic and as nitrifying microorganisms (Gutierrez24 Wing and Malone, 2006; Malone and Pfeiffer, 2006). Biofilter media are generally characterized by the surface area they provide in relation to the volume they occupy, i.e. the specific surface area (SSA, m2/m3) (Colt et al., 2006). Media with a high SSA usually ensure large volumetric TAN removal rates, while guaranteeing a low footprint triggered by the downsizing of the reactor. However, and dependent on different configurations of the reactor and type of carrier, media with a large SSA may require additional maintenance, i.e., oxygen or air injection (Ødegaard et al., 1994; Rusten et al., 2006), increased backwashing intervals (Singh et al., 1999; Emparanza, 2009; Suhr and Pedersen, 2010), or periodic disinfection (Pedersen and Pedersen, 2012). Biological TAN removal is also termed nitrification, and results from the two-step microbial conversion of ammonia to nitrate (NO3-), via nitrite (NO2-) (Sharma and Ahlert, 1977; Rogers and Klemetson, 1985). The overall nitrification reaction (Eq. 1) requires 1.98 moles of bicarbonate (HCO3-) and 1.86 moles of oxygen (O2) to oxidize 1 mole of NH4 (Szwerinski et al., 1986; Loyless and Malone, 1997; Nogueira et al., 1998; Summerfelt et al., 2015). The reaction yields 0.02 moles of bacterial biomass (generic formula of C5H7NO2), and 0.98 moles of nitrate. Nitrate is not necessarily toxic to fish, so it is allowed to accumulate in the system (Colt, 2006; Pedersen et al., 2012; Davidson et al., 2014), and the only ways to control its accumulation are increased water exchange, or denitrification (van Rijn, 1996, 2013; Piedrahita, 2003). NH 1.86O 1.98HCO → 0.020C H NO 0.98NO 1.88H CO 1.04H O (1) There are three main groups of nitrifiers performing nitrification: ammonia oxidizing bacteria (AOB), ammonia oxidizing archaea (AOA), and nitrite oxidizing bacteria (NOB) (Hagopian and Riley, 1998; Tal et al., 2003; Schreier et al., 2010; van Kessel et al., 2010). The first two groups are responsible for the conversion of ammonia into nitrite, while the third group is responsible for the conversion of nitrite into nitrate. Nitrifying bacteria develop in clusters attached to the support media, and it is usual to find AOA, AOB and NOB in close proximity inside the biofilm (Zhang et al., 1995; De Beer et al., 1996; Rasmussen and Lewandowski, 1998; Picioreanu et al., 2000). As ammonia penetrates the biofilm, it supports the growth of the three groups of nitrifiers. However, uptake of substrate from the rearing environment is not a simple process and depends on bulk phase concentration, microbial population structure, and hydraulic conditions in the water phase. 8.3.1 Substrate removal in biofilms Substrate consumption or removal by attached biofilms (Fig. 10) is dependent on the processes occurring at three regions around the biofilm: (1) water phase; (2) biofilm-liquid interface and diffusive boundary layer (DBL); and (3) biofilm (Revsbech and Jørgensen, 1986; De Beer et al., 1994). Substrate bulk concentration defines the potential removal of an active biofilm, while the processes occurring at the liquid-biofilm interface, and inside the biofilm define the total substrate consumption. At larger water phase concentrations, the penetration depth of the substrate will be also larger (e.g. Rittmann, 1985; Siegrist and Gujer, 1985; Gonenc and Harremoës, 1990). Thereafter, the consumption of substrate can be described as a sequential dual step process of external substrate mass transfer, and internal mass diffusion. 1. At first, there is direct mass transfer of the substrate from the bulk phase into the biofilm. This occurs in a purely physical layer, the DBL, where substances are diffused from the water phase into the biofilm, or vice-versa (Revsbech and Jørgensen, 1986). When challenging the biofilm with increased hydraulic loading rates (Zhu and Chen, 2001a; Prehn et al., 2012), or aeration rates (Zhu and Chen, 2003; Rusten et al., 2006) the Reynolds number (Re) of the 25 flow increases. This creates turbulence in the biofilm-water interface, decreasing the DBL and the biofilm thicknesses, and enhancing substrate diffusion (De Beer et al., 1996; Rasmussen and Lewandowski, 1998; Beer and Kühl, 2001); 2. The second process is defined as the internal mass diffusion from the biofilm phase into the bacteria. Diffusion of substances may be hampered at this level as a result of biofilm tridimensional irregularities, random clustering of bacteria, existence of voids, etc. (Picioreanu et al., 2000; Dreszer et al., 2014). Increased water fluxes and turbulence in the water phase enhance the stratification and compactness of the biofilm, minimizing the heterogeneous structure of the biofilm (De Beer et al., 1994; Lopes et al., 2000; Yang et al., 2001; Liu et al., 2013), and overall improving substrate diffusion and utilization. Fig. 10. Conceptual model of substrate diffusion in a fixed growth biofilm. SW – substrate concentration in the water phase; SS – substrate concentration at the biofilm-DBL interface; S0 – substrate concentration at the maximum penetration depth. From Zhu and Chen (2001b). 8.3.2 Factors affecting nitrification Besides ammonia, oxygen, and alkalinity, other factors unrelated to nitrification reactions, can also affect nitrification (Fig. 11) (Szwerinski et al., 1986; Arvin and Harremoës, 1990; Colt et al., 2006; Eding et al., 2006). Carbon as organic matter can impair nitrification (Ohashi et al., 1995; Okabe et al., 1996; Satoh et al., 2000), since it stimulates the growth of heterotrophic bacteria. If the biofilter is loaded with carbon to nitrogen (C:N) ratios above 1:1, already the nitrification capacity of the biofilter can be reduced (Fig. 12) (Zhu and Chen, 1999, 2001; Carrera et al., 2004; Ling and Chen, 2005; Guerdat et al., 2011). As heterotrophic bacteria grow on the excess carbon, less oxygen is available for nitrifiers to perform nitrification (Wanner and Gujer, 1984; Zhang et al., 1995; Michaud et al., 2014). As previously mentioned, organic matter (COD) is produced at a ratio of ~280 g O2/kg feed consumed (Bureau and Hua, 2010; Dalsgaard and Pedersen, 2011; Pedersen et al., 2012). Loading of organic matter, is not easily controlled, and risks of extra loading are related to poor solids removal, peak loadings related to feeding strategies, feed spillage, reduced feed conversion ratio, and feed composition (Bovendeur et al., 1990a,b; Leonard et al., 2002; Emparanza, 2009; Guerdat et al., 2011). According to the amount of carbon in COD (1:2.68), fish waste excretion categorizes a constant C:N loading to the biofilter equivalent to at least 3:1in live fish operations (Zhu and Chen, 1999, 2001b). Consequently, the microbial communities adjusted to this loading will never purely represent a nitrifying community and there will always be heterotrophic bacteria in the media-attached biofilm (Rittmann, 1985; Amann and Kühl, 1998; Satoh et al., 2000; Beer and Kühl, 2001). 26 Fig. 11. Nitrification-rate limitations according to water-phase molar ratios between ammonia (NH3), bicarbonate (HCO3-) and oxygen (O2). From Szwerinski et al. (1986). Since heterotrophic bacteria grow much faster than nitrifiers (Hagopian and Riley, 1998; Schmidt et al., 2003; Tal et al., 2003; Paredes et al., 2007; Schreier et al., 2010), they can potentially place themselves in layers that better support their growth. Heterotrophic bacteria are thus usually found in the outer layers of the biofilm, where the oxygen concentration is higher (Wanner and Gujer, 1984; Verhagen and Laanbroek, 1991; Zhang et al., 1995; Fdz-Polanco et al., 2000; Satoh et al., 2000). This means that, if the conditions in the water body are favorable for heterotrophic growth, there may be a reduction in functionality of nitrifiers, due to decreased oxygen diffusion into the deeper layers of the biofilm (Särner and Marklund, 1984; Zhu and Chen, 2001a; Nogueira et al., 2002; Ling and Chen, 2005). Injecting air or pure oxygen into the reactor (Golz et al., 1999; Zhu and Chen, 2003; Vigne et al., 2011), or maintaining large hydraulic loading rates (Melin et al., 2005; Canziani et al., 2006; Prehn et al., 2012) can usually keep oxygen bulk concentrations at non-limiting levels and sustain both processes. Fig. 12. TAN removal rate versus TAN concentration of the reactor series, for different carbon to nitrogen (C:N) ratios. From Zhu and Chen (2001b). Other factors that potentially affect nitrification capacity and performance are sulphide (Æsøy et al., 1998), operational backwashing requirements (Suhr and Pedersen, 2010; Vigne et al., 2010; 27 Guerdat et al., 2011; Pfeiffer and Wills, 2011), and particle adsorption or deposition on top of the biofilm (Figueroa and Silverstein, 1992; Larsen and Harremoës, 1994; Janning et al., 1997, 1998). 8.3.3 Particle interactions with biofilms Most studies on the influence of C:N ratio on nitrification kinetics in aquaculture rely on the ratios between dissolved carbon and dissolved nitrogen (Zhu and Chen, 2001b; Ling and Chen, 2005; Guerdat et al., 2011). Although this is a good reference point on the interactions between heterotrophic bacteria and nitrifiers, the studies usually disregard the full spectrum of potential competition (Michaud et al., 2006). The largest part (~60-70% ) of carbon waste in aquaculture exists in the solid fraction and is organic (Heinen et al., 1996; Twarowska et al., 1997; Wik et al., 2009; Dalsgaard and Pedersen, 2011), so it may well play a defining role in the interactions between nitrifiers and heterotrophs. Studies on particle interactions with biofilters in RAS have focused mostly on particle production/removal dynamics through fixed bed or moving bed reactors (e.g. paper III). As solids pass through biofilter reactors, they can either be entrapped (Marquet et al., 1999, 2007; Franco-Nava et al., 2004b), disintegrated into smaller particles (Åhl et al., 2006; Ivanovic and Leiknes, 2008, 2012), or pass unaffected through the reactor. While the last process may not impair functionality of the biofilter, entrapment and disintegration of particles may affect biofilter performance (Arvin and Harremoës, 1990; Figueroa and Silverstein, 1992; Larsen and Harremoës, 1994). Removal of particulates in biofilters depends on a number of factors: (1) particle characteristics; (2) media and biofilm configuration; and (3) flow regime (Arvin and Harremoës, 1990; Larsen and Harremoës, 1994). The degradation potential of micro-particles in aquaculture is not wellknown, though it is recognized from wastewater that the rate of hydrolysis is slower than the rate of consumption of soluble substances (Laspidou and Rittman, 2002). Furthermore, it has also been suggested that the organic matter transported by smaller particles is more bio-available than the organic matter transported by larger particles (Balmat, 1957; Levine et al., 1985, 1991; Meriac et al., 2015). In this sense, entrapment and/or assimilation and hydrolysis of micro-particles in the biofilm have larger implications on biofilter kinetics than larger particles. Apart from a few observations (e.g. Michaud et al., 2006; paper III), the details on potential nitrification interactions due to particle dynamics through biofilters in RAS have not yet been quantified. As known from wastewater treatment observations, micro-particles can impair nitrification through one of two processes: (1) the leaching and hydrolyzation of organic matter; and (2) the potential adherence of particles to the biofilm surface (Fig. 13). (1) Some heterotrophic bacteria excrete polymeric substances or enzymes that help them hydrolyze particulate organic matter (Eliosov and Argaman, 1995; Nielsen et al., 1997; Barker and Stuckey, 1999; Guellil et al., 2001; Laspidou and Rittman, 2002). The same or other bacteria will then be able to consume the hydrolyzed organic matter, which will create an additional oxygen requirement in the biofilter. (2) Secondly, the adherence or entrapment of particles in the biofilm can block mass transfer of oxygen into the biofilm (Särner and Marklund, 1984; Särner, 1986; Figueroa and Silverstein, 1992; Larsen and Harremoës, 1994). Decreased local diffusion of oxygen into the biofilm will create areas with limited substrate consumption, thus decreasing the overall performance of the biofilter. 28 Fig. 13. Possible effect (localized oxygen shortage) of the attachment and hydrolyzation of an organic particle to the biofilm surface. From Särner (1986). The few studies that have dealt with particulate organic matter interactions with biofilters have also suggested the C:N ratio to be a controlling factor in heterotrophic-nitrifiers dynamics (Bovendeur et al., 1990b; Leonard et al., 2000; Carrera et al., 2004; Michaud et al., 2014; paper II). For instance, Bovendeur et al. (1990b) combined particulate and dissolved organic matter into one value to test nitrification performance of a fixed-film reactor. The authors observed that, although oxygen will be consumed first by heterotrophic bacteria, as long as oxygen in the bulk phase is maintained at non-limiting concentrations its penetration and diffusion into the biofilm is still sufficient to sustain nitrification (Table 1). In paper III, removal of particulate organic matter was observed in fixed bed and moving bed biofilters. While entrapment of particles was the likely cause for particle reduction in fixed beds, disintegration of particles and thereby removal of particulate organic matter was observed in moving beds. Although heterotrophic consumption of organic matter was observed in both reactor types, nitrification still occurred at normal rates. Abundant oxygen in the bulk phase coupled to a significant water flow speed (14.4 m/h) presumably ensured sufficient oxygen penetration into the biofilm to sustain the simultaneous removal of carbonaceous and nitrogenous compounds. Table 1. Relation between biofilm removal rates of organic matter and total ammonia (zero-order nitrification rate), and oxygen consumption rates involved; artificial biofilm monitored under standard conditions and varying COD loading rates in a series of independent experiments. From Bovendeur et al. (1990b) Removal rate (g/m2/d) Oxygen consumption rate (g/m2/d) COD NH4+-N Overall respiration NOx-N COD-O 0 9.89 10.74 11.36 0.575 0.475 0.446 0.536 2.75 2.54 2.96 3.11 2.49 1.89 1.82 1.99 0.268 0.655 1.14 1.12 It then seems that in aquaculture it is possible to operate biofilters as reactors that simultaneously remove ammonia and organic matter. Although interactions between heterotrophs and nitrifiers will occur due to elevated organic matter loadings, it is still possible to sustain nitrification at relatively high inlet C:N ratios. Possibly, there are two main interacting causes that sustain the simultaneous consumption of carbonaceous and nitrogenous compounds: (1) Keeping oxygen bulk concentrations close to saturation (Golz et al., 1999; Zhu and Chen, 2003; Vigne et al., 2011). Heterotrophic consumption of organic matter usually reflects 0’29 order kinetics, i.e. is independent of substrate concentration (Henze, 2002; Eding et al., 2006). Therefore, the scope for nitrification in terms of oxygen consumption can be maximized by knowing the oxygen requirements for organic matter degradation in the biofilter (Pano and Middlebrooks, 1983; Bovendeur et al., 1990a,b; Vollertsen and Hvitved-Jacobsen, 1999); (2) Rapid water fluxes will decrease biofilm and DBL thickness (Zhang and Bishop, 1994; De Beer et al., 1996; Rasmussen and Lewandowski, 1998; Prehn et al., 2012). Hence, the depth of substrate penetration can be maximized, which will help in sustaining nitrification by maintaining a high oxygen concentration in the bulk phase (Melin et al., 2005; Vigne et al., 2011). 8.4 Other factors interacting with particles A summary of the most common effects on particle size in flow-through farms (FT), RAS, or wastewater treatment plants (WWTP), according to factor and mechanism, but excluding primary solid removal devices, can be found in table 2. One of the main potential sources of fine particles in RAS – other than feed and feces – is biofilm, while specific system configurations can either remove or produce particles: - - - - Normal operation in moving bed bioreactors constantly shed off excess biofilm from the carrier media (Åhl et al., 2006; Ivanovic and Leiknes, 2008, 2012). Though not beneficial on a particle load basis, this keeps the biofilm in a nitrification-stable status (Hem et al., 1994; Ødegaard et al., 1994; Rusten et al., 2006). In static media beds without regular backwashing maintenance (Franco-Nava et al., 2004b; paper I), newly shed biofilm can be naturally released from the reactor. Furthermore, as the biofilm grows towards the outlet of the reactor, there is less space for density-dependent settling to occur (Wong and Piedrahita, 2000; Patterson et al., 2003; Merino et al., 2007) and, therefore, non-settled particles will be released from the reactor. When operated with suitable backwashing intervals, fixed beds have the potential to remove particles, since they will work as a physical barrier where particles attach and become entrapped (Bouwer, 1987; Larsen and Harremoës, 1994; Yang et al., 2001). Aeration sources or rate of aeration (Leiknes et al., 2006; paper III), turbulent flows (Maxey et al., 1996; Khan et al., 2011), degassing units (Patterson and Watts, 2003b), and pumps (McMillan et al., 2003; Sindilariu et al., 2009), can further increase particle load by disintegrating solids (feed, feces, or biofilm) into smaller particles. Long reactors or tanks with no re-suspension of water, may remove particles by out-settling (Brinker and Rösch, 2005; Masaló and Oca, 2013). Other factors, such as fish activity and behavior, standing biomass, and raceway height, may also affect particle size, as observed by Brinker & Rösch (2005) in a flow-through fish farm. Brinker and Rösch (2005) suggested that some guidelines should be followed in order to compensate for reduced mechanical treatment efficiency caused by decreased particle size. Theoretically, and although some of these are sometimes difficult to achieve in aquaculture, these guidelines might also be applied in RAS to reduce internal micro-particle production: 1. 2. 3. 4. 5. 30 Avoidance of units or processes that lead to violent water turbulence; Immediate effluent treatment after the fish production unit; Mixing fish of different sizes throughout the farm; Avoidance of dead-flow zones; Practice surface-oriented feeding to avoid disturbance of solids settling. Table 2. Factors affecting particles in aquacultural and wastewater systems. Data restricted to components or features whose main purpose is not solids removal. FT – Flow-through system; RAS – Recirculating aquaculture systems; WWTP – Wastewater treatment plant. Factor Mechanism Effect on particles System Source Time of day Fish more active during the day Size decreases during the day FT RAS Brinker and Rösch (2005) Patterson and Watts (2003) Aeration rate Destructive turbulence Size decreases with increased aeration RAS Brambilla et al. (2008) Shear forces coupled to destructive turbulence Size decreases after a propeller-wash bead filter RAS Pfeiffer et al. (2008) Size decreases and load increases with aeration rates in moving beds WWTP WWTP WWTP WWTP WWTP RAS Melin et al. (2005) Leiknes et al. (2006) Åhl et al. (2006) Ivanovic and Leiknes (2008) Ivanovic and Leiknes (2012) Paper III Fish size Constructive and suspending turbulence Size increases with decreasing fish size Distance from raceway bottom Differential sedimentation Size decreases with increasing distance FT RAS RAS FT Brinker and Rösch (2005) Franco-Nava et al. (2004) Merino et al. (2007) Brinker and Rösch (2005) Waterfalls Destructive fragmentation from the fall Size decreases after a waterfall FT Brinker and Rösch (2005) Tank cleaning Biofilm scrapping from tank walls Load increases shortly after routine cleaning FT Kelly et al. (1997) Pumps Centrifugal forces/Collision with impellers Size decreases within the pump RAS FT RAS McMillan et al. (2003) Sindilariu et al. (2009) Paper I Biofilm Biofilm detachment (shear/sloughing) Load and size increase after ”old” fixed bed biofilters RAS RAS RAS RAS Yang et al. (2001) Patterson and Watts (2003) Franco-Nava et al. (2004b) Paper I Load and size increase after a fluidized sand filter RAS Pfeiffer et al. (2008) Entrapment within the media Fixed media decreases particle load WWTP WWTP RAS RAS Bouwer (1987) Larsen and Harremoës (1994) Yang et al. (2001) Paper III Quiescent zones Differential sedimentation Particle size decreases due to sedimentation WWTP WWTP RAS RAS Marquet et al. (1999) Marquet et al. (2007) Merino et al. (2007) Paper I Degassing units Trickled fragmentation Size decreases during the fall RAS RAS Patterson and Watts (2003) Paper I Ozone Flocculation Size increases after contact RAS RAS RAS Rueter and Johnson (1995) Krumins et al (2001a) Krumins et al. (2001b) Flow rate Turbulent vs. laminar flow Turbulent flows disintegrate particles WWTP RAS WWTP Maxey et al. (1996) Franco-Nava et al. (2004) Khan et al. (2011) 31 8.5 Particle Size Distribution (PSD) in RAS Typical particle size distributions (PSD) in RAS are shown in Fig. 14. Due to the effects of aquacultural components (section 8.4), coupled with a high water re-use rate, PSD in RAS are usually dominated by micro-particles (Chen et al., 1993; Rueter and Johnson, 1995; Patterson et al., 1999). To better understand particle accumulation and interactions with the rearing environment, the physical and chemical properties of micro-particles must be fully understood. Cumulative size distribution (%) 100 paper III paper I 0 0 10 20 30 40 50 Particle size (µm) Fig. 14. Cumulative particle concentration of two experimental RAS. From paper I and paper III. Total suspended solids (mg/L) Solid loading is a function of particle concentration and volume (weight), while potential parasites and/or chemical compounds adhesion to solids is determined by particle surface area (Krumins et al., 2001a,b; Hill and Ng, 2002). Thus, singling out the potential effect of solids has to be performed in a way that comprises all three particle-defining parameters: concentration, surface area, and volume. This cannot be achieved through traditional TSS measurements , though it can be acquired through PSD determination (Patterson et al., 1999; Patterson and Watts, 2003b; Brinker et al., 2005c). As the PSD determination informs on particulate volume, area and concentration by size, it can be considered as a better predicting agent to particle effects than TSS. Furthermore, there is usually a good relationship between TSS and particle sample volume (Fig. 15). This means the latter may be used as a solids weight predictor (Chen et al., 1993; paper II), though, in itself, does not fully characterize particles in a specific solution. 50 45 40 35 30 25 20 15 10 5 0 0.0E+00 y = 8-07x R² = 0.77 1.0E+07 2.0E+07 3.0E+07 Particle volume concentration 4.0E+07 5.0E+07 (µm3/mL) Fig. 15. Relationship between particle concentration and solids weight (as TSS). Adapted from paper II. 32 8.5.1 β-value for aquaculture operations Particles in aqueous solutions follow a natural near-hyperbolic distribution (Rueter and Johnson, 1995; Patterson et al., 1999). Therefore, it is exponentially less probable to find a large particle than a much smaller one. To compare hyperbolic distributions, the skewness of the data, as the deviation from the normal distribution, is usually utilized (Park et al., 2011). For particle size distributions, the skewness can also be related to the slope of the distribution obtained after the double logarithmic transformation of the data set (Eq. 2 and Eq. 3), i.e. the β–value (Rueter and Johnson, 1995; Patterson et al., 1999). At first, the data sets obtained through PSD measurements may be divided into size classes, with the class boundaries being adjusted to fit an ascending geometric progression. Each size class (li) is then selected according to a ratio of li+1 = 1.26·li ( √2 ≈ 1.26), representing a regular progression where the volume of a median particle in any size class will be twice as large as the volume of a median particle in the consecutively smaller size class. ∗ ↔ (2) where N is the particle number; l is the particle size; and A and β are empirical constants; ΔNi is the number of counts, counts per unit volume, or percent of total counts in class i; Δli is the class width, represented by the nominal particle counts of size li*; li* is the median of the size class defined between li+1and li. To normalize the distribution, the data resulting from Eq. 1 should be log–log transformed (Eq. 2), from where the slope, β, can be extracted. log log log (3) The integration of the slope in particle distribution trends, allows the comparison of particle distributions from different systems. The slope in itself is generally in the range of 2-5 for aquaculture systems (Patterson et al., 1999; Brinker and Rösch, 2005; Brinker et al., 2005c). At lower β– values, more volume and area will be allocated in the larger size classes, while at larger β–values, more area and volume will be associated with smaller particles (Fig. 16). Fig. 16. Hypothetical distribution curves for PSD data defined by (A) β = 2 and (B) β = 4. From Patterson et al. (1999). 33 Another interesting characteristic of β–values is its dependency to aquaculture intensification, as observed through the comparison of reported β–values in flow-through (Brinker and Rösch, 2005; Brinker et al., 2005c) or RAS systems (Rueter and Johnson, 1995; Patterson et al., 1999; Krumins et al., 2001a,b; Patterson and Watts, 2003b; paper I; paper II). As previously suggested, retention time in the systems affects particle size, and this is visualized in the consistent report of large β–values for RAS operations. Furthermore, the use of varying elements that efficiently remove particles or promote turbulence significantly increase the observed β–values (Patterson et al., 1999; Patterson and Watts, 2003b). A typical example of this effect is the β–value derived from one of the systems in Chen et al. (1993) datasets. The system, equipped with a 9 m3 clarifier and a rotating biological contactor, ended up with a system-derived β–value of 6.29 and a distribution highly dominated with particles below 6.5 µm and out of the detection range of Patterson et al. (1999) particle counter. High β–values may either imply efficient screening of particles or simply accumulation of microparticles. The distinction is not easily determined, though processing of this information should inform on potential toxicity of the accumulated solids. Particles with smaller diameters have a high surface area to volume ratio (Fig. 17) (Witham et al., 2005; paper II), so they are said to be more efficient in carrying adsorbed nutrients and microbes (Balmat, 1957; Levine et al., 1985, 1991), and are as potential detrimental agents of fish performance (Chapman et al., 1987; Magor, 1988; Bullock et al., 1997). 7 Area:Volume ratio 6 5 4 3 2 1 0 1 10 100 1000 Particle diameter (µm) Fig. 17. Theoretical surface area to volume ratio of a perfect sphere, in relation to particle diameter. Asphere = 4∙π∙r2; Vsphere = 4/3∙π∙r3. 8.5.2 PSD stabilization in RAS Although PSD may be affected by several components or RAS characteristics, it has been suggested that at specific conditions of operation and maintenance the particle size distribution of a RAS may reach equilibrium. PSD steady-state (McMillan et al., 2003 Patterson and Watts, 2003; Meriac et al., 2015; paper I) is a function of (1) all the components that have an influence on mean particle size; (2) water exchange rates; and (3) internal water flows. While the first parameter is easily determined by the dynamics of PDS across a specific RAS unit, the last two factors have to be extrapolated from comparisons between several studies. - 34 Water exchange can dilute accumulated substances in RAS. A general description of PSD by Patterson and Watts (2003) in a farm with 100-190 % daily water renewal, showed that dilu- - tion may prevent excessive particle breakage, though dominance and resilience of microparticles is still apparent. More recently (paper I), in a system operated with 30 % daily water renewal, a stabilization of the β-value throughout the recirculation loop was observed (Table 3). It then seems that at large and varying water exchange rates the dynamics of PSD will be lost to overflow, while at low and stable water exchange rates, the mechanisms for PSD stabilization are grounded. High internal turnover rates define minimal substance removal by specific components (e.g. Zhu and Chen, 1999; Melin et al., 2005; Liu et al., 2013). If the loading rate determines consumption/removal, it can also be extrapolated that it will have an effect on removal/turnover of particles. The effect of specific units on PSD will thus be masked by high hydraulic turnover rates (minimum 1.25 times/h) (paper I). Table 3. β-values determined with 4 h intervals at several locations within a RAS, operated at a 28 % daily water renewal and a flow rate of 1.25 times/hour (paper I). BDF – Before the drum filter; ADF – After the drum filter; APR – After the pump reservoir; ASF – After the submerged biofilter; ATF – After the trickling filter. Sampling location Time of sampling 0h +4h +8h +12h +16h +20h +24 h Mean ± SD BDF β 3.12 3.41 3.44 3.08 3.29 3.17 3.15 3.23 ± 0.13 ADF β 3.25 3.67 3.21 3.60 3.57 3.46 3.33 3.44 ± 0.18 APR β 3.33 3.29 3.53 3.51 3.61 3.33 3.62 3.46 ± 0.14 ASF β 2.94 3.22 3.66 3.51 3.38 3.48 3.52 3.39 ± 0.24 ATF β 3.50 3.47 3.42 3.51 3.28 3.61 3.57 3.48 ± 0.11 An interesting point of PSD steady-state in RAS is the influence of mesh size on the time needed to reach equilibrium (Fig. 18). Experimental RAS operated with 20 or 60 µm microscreens reached steady-state 1-2 weeks earlier than RAS operated with 100 µm microscreens, while endpoint particle concentrations were similar between the different mesh sizes tested (paper II). At constant conditions of feed and make-up water, the background PSD seems to be highly dominated by micro-particles below 20 µm (Chen et al., 1993; Patterson et al., 1999). Consequently, removal by mesh sizes between 20-100 µm is similar (Cripps, 1995; Kelly et al., 1997; Patterson and Watts, 2003a; Rusten and Ødegaard, 2006), affecting other water quality parameters in an analogous manner. However, the increased filtration effort by narrower screens promotes equilibrium at a quicker pace (paper II). To my knowledge, this is the first time that timing to reach particle steady-state in RAS has been described, and it could lead to an interesting development for more intensive systems, where particle production and accumulation is higher. In this sense, the quicker removal of particles by narrower screens would more rapidly allow the prediction of the system performance. In paper II, other factors that could have enhanced the time to reach equilibrium include increased backwash frequency of the mesh, and backwashing of the biofilter and trickling filter. Although increased backwashing frequency in the last day did not improve removal by the different mesh sizes, their flow handling capacity may have been reduced during the experiment due to clogging and caking, thus minimizing their removal capacity (Tien et al., 2001; Brinker and Rösch, 2005; Leiknes et al., 2006; Iritani et al., 2012; Dolan et al., 2013). Secondly, if the carrying capacity of the fixed media reactors, in terms of particle removal, was achieved during the experiment, backwashing would enhance particle entrapment by removing the already attached particles (Singh et al., 1999; Emparanza, 2009). By increasing backwash frequency, the scope for removal would have been optimized in the microscreens and biofilters, and would have potentially increased solids removal by the narrower mesh sizes. 35 20 control Particle concentration (x103 Particles/mL) 18 100µm 60µm 20µm 16 14 12 10 8 6 4 2 0 0 7 14 21 28 35 42 Experimental days Fig. 18. Nominal particle concentration in RAS operated with no microscreen (control), or with a 100µm, 60µm, or 20 µm microscreen. Adapted from paper II. 8.6 Micro-particles in the fish tank As aquaculture practices continue their expansion towards more intensive RAS operations, a significantly larger proportion of waste is retained in the system (Martins et al., 2010; Dalsgaard et al., 2013; Verdegem, 2013). Some of the waste substances accumulate due to imperfect removal, while others accumulate because there is no real treatment in the recirculation loop. Among the substances that can potentially accumulate in RAS are, e.g., metals and minerals (Davidson et al., 2009, 2011; Martins et al., 2009a,b), phosphorus (Foy and Rosell, 1978, 1991; Garcia-Ruiz and Hall, 1996; Ebeling et al., 2003; Van Bussel et al., 2013), nitrate (Pedersen et al., 2012; Davidson et al., 2014), and dissolved (Bovendeur et al., 1990a,b; Guerdat et al., 2011) or particulate organic matter (Chen et al., 1993; Leonard et al., 2002; Patterson and Watts, 2003b). There could well be cause of concern for some of the wastes and substances accumulating in RAS as they might interact with and impair fish growth, welfare and quality. 8.6.1 Micro-particles as microbial substrate Accumulation of micro-particles in aquaculture sustains the development of microbial communities outside of the biofilter (Fig. 19) (Blancheton and Canaguier, 1995; Liltved and Cripps, 1999; Wold et al., 2014). Due to their physicochemical constitution and long residence time in the system, particles create extra space and nutrients for bacteria to proliferate (Levine et al., 1985, 1991; Liss et al., 1996; Droppo et al., 1997). At first, particles serve only as substrate for the formation of biofilm, though at some point, some bacteria may start releasing polymeric substances and enzymes that help them hydrolyze and consume the particle-bound organic matter (Nielsen et al., 1997; Barker and Stuckey, 1999; Guellil et al., 2001; Laspidou and Rittman, 2002; Morgenroth et al., 2002). Therefore, microbial concentration and diversity will vary according to the availability and composition of accumulated particles. In freshwater RAS, the dependency of bacteria to particles is somehow also dependent on feeding intensity (Table 4). When operating a RAS at a CFB of 1 kg/m3 of MUW, bacterial activity is proportional to amount of particles between 20-40 µm in size, and have a strong dependency towards particulate BOD5 trends. The relationship with particulate surface area is also quite strong at this CFB. However, as feeding intensity increases to a CFB of 3.1 kg/m3 of MUW, these rela36 tionships are lost. In more intensive systems, due to a heavier supply of nutrients, bacteria do not require extra physical substrate to grazer for nutrients, as their availability is probably universal and ubiquitous (Avnimelech et al., 1986; Avnimelech and Ritvo, 2003; De Schryver et al., 2008; De Schryver and Verstraete, 2009). Fig. 19. Relationship between bacterial concentration and particulate matter in a RAS during larval and fingerling production. From Blancheton and Canagier (1995). Table 4. Relationship between bacterial activity (measured as Bactiquant®) and organic matter and particle parameters, measured as coefficients of regression (R2) during the experiments related to Paper II and Paper III. CFB – Cumulative feed burden (kg feed/m3 make-up water); BOD5 – Biochemical Oxygen Demand after 5 days of incubation (mg O2/L). COD – Chemical Oxygen Demand (mg O2/L). Diss. – Dissolved fraction (filtered through GF/A filters 1.6µm Ø). Part. – Particulate fraction (Total – Dissolved = Particulate). N/A – Not available. Unpublished data. Parameter Paper II CFB = 3.1 Paper III CFB = 1 Total BOD5 0.917 0.370 Diss. BOD5 0.765 0.076 Part. BOD5 0.986 0.594 Total COD 0.417 0.520 Diss. COD 0.264 0.272 Part. COD 0.084 0.688 Particle counts 0.072 0.666 Particulate area 0.030 0.860 Particulate volume 0.004 0.685 β-value 0.044 0.222 Area to volume ratio 0.275 0.196 < 20 µm particles N/A 0.614 20 - 40 µm particles N/A 0.734 40 - 100 µm particles N/A 0.410 > 100 µm particles N/A 0.119 37 Most of the genera in association with particles are functionally heterotrophic (Leonard et al., 2000; Liltved and Cripps, 1999; Michaud et al., 2009). Among those are bacteria from the family Vibrionaceae, which are known to contain pathogenically active species (Bullock et al., 1994; Hess-Erga et al., 2008; Rurangwa and Verdegem, 2014). Although there are ways to try and control pathogen proliferation in aquaculture (Liltved and Hansen, 1990; Summerfelt, 2003; Sharrer et al., 2005; Summerfelt et al., 2009; Davidson et al., 2011b), it is increasingly more difficult to remove them from the system if they are in association with particles. Microbial maturation has been shown to improve fish larvae growth, while bio-floc technology (BFT) improves water quality while reducing feeding costs. Both techniques are quite interesting for controlling microbial communities in aquaculture due to their stabilizing effects on microbial communities, and established competitiveness with potentially pathogenic species. However, to better demonstrate their application and significant improvements in fish production, more studies in microbial maturation, BFT, and microbial dependency on CFB, are required. 8.6.2 Interactions between micro-particles and fish When challenged with specific environmental irritants or stressors, fish may exhibit a wide range of growth detrimental responses (Noble and Summerfelt, 1996; Ferguson, 2006; Vadstein et al., 2013). The responses widely vary from behavioral to physiological conditions, and may include increased hematocrit (Lake and Hinch, 1999) and blood cortisol levels (Martins et al., 2009a,b), skin bruises, inflammation, or erosion (Roque d’Orbcastel et al., 2009; Kolarevic et al., 2014), liver, kidney or pancreatic degeneration and failure (Bucher and Hofer, 1993; Michel et al., 2013), and gill lesions (Bernet et al., 1999; Chen et al., 2012; Magor, 1988). Besides reduced welfare and condition, the most common pathologies caused by interactions with micro-particles are related to gill abrasion and loss of functionality. Classic lesions related to abrasion of fish gills by particles can usually be observed as mechanical rupture of tissue, or as compensatory mechanisms to increase the gill gas-transfer surface area (Fig. 20). Among them, it is common to find lamellar fusion, or hyperplasia; lamellar edemas or aneurisms; or epithelial lifting (Mallat, 1985; Magor, 1988; Bernet et al., 1999; Ferguson, 2006). Direct damage to the gills can reduce fish performance due to decreased respiratory capacity, while pathogen transport by particles can cause disease outbreaks. Eventually, both pathologies may lead to the fish performing poorly, or even to death (Noble and Summerfelt, 1996). Related to particle loading and accumulation, fish are subject to encounter more particles in a RAS than in nature or a flow-through farm. Consequently, there is an increased risk of lesions in the gill rakes for fish growing in RAS (Magor, 1988; Michel et al., 2013). However, fish in RAS tend to perform better than fish in flow-through systems (Jokumsen and Svendsen, 2010; Attramadal et al., 2012a, 2014), possibly due to improved control over other water quality parameters. Though all solids potentially interact with fish, larger and denser solids are easily removed in RAS. This means that micro-particles (< 5 µm) are the most obvious antagonists of fish, in the sense that they can clog (Chapman et al., 1987; Lake and Hinch, 1999) and abrade fish gills (Magor, 1988; Bernet et al., 1999), or transport pathogens (Bullock et al., 1997; Hess-Erga et al., 2008; Vadstein et al., 2013). Most of the studies describing the interactions between micro-particles and fish, in specific the effects on fish gills, are still limited to natural environments such as rivers and other natural water bodies (e.g. Lake and Hinch, 1999), or have limited data from whence to conclude significant effects of accumulated particles (e.g. Chapman et al., 1987). In a side-study from paper II, the ef38 fect of RAS intensity or specific RAS components could not be observed on gill histopathology or fish growth between the treatments tested. Though specific pathologies were observed in both treatments, the effect seems to be unrelated to particles. When comparing these same gills to the gills of fish reared in a flow-through system, no significant effect of aquaculture system (and consequent particle encounters) could be observed (Table 5). Hence, more information is needed to determine the specific potential interactions between particle accumulation and fish performance and welfare in RAS. Fig. 20. Light microscope photographs of common structural pathologies of the fish gill. (A) Normal lamella at 10x40 magnification; (B) Aneurismatic lamella at 10x40 magnification; (C) Epithelial lifting (arrow-pointed) at 10x16 magnification; (D) Hyperplastic of fused lamella (inside the box) at 10x16 magnification. Unpublished data. Table 5. Comparative rank scores from a protocol comparing the gill structure, morphology, existence and degree of structural pathologies in fish reared in a flow-through system (FT), a RAS equipped with a swirl separator and 20 µm microscreens (RAS20) and a RAS equipped with a swirl separator (RASNO), both from paper II. Unpublished data. System Rearing time Particle concentration (#/mL) No. particle encounters Total score Rank FT 4 weeks 5020 0.13x1012 19.0 1st RASNO 6 weeks 14060 42.4x1012 24.4 3rd RAS20 6 weeks 5510 16.8x1012 21.7 2nd 8.7 Conclusions and future perspectives Only recently has the scientific community turned their attention to the toxicity of particles to fish or biofilter performance. As the industry keeps progressing towards more intense practices, it is important to continuously improve the knowledge on potentially detrimental aspects of substances 39 accumulating in RAS operations. The present thesis has documented several aspects of particle accumulation in RAS. - - - Several elements besides primary clarifiers can influence the particle size distribution in a RAS. Elements such as pumps, biofilters, waterfalls and quiescent zones have been shown previously to affect particle size and concentration, and were further documented in paper I and paper III. The major conclusion from these studies is that the overall β-value is a function of the several elements installed in any given RAS, and that other factors, sometimes independent from the elements in any system, such as biofilter backwashing periodicity, could further enhance or manipulate the system-derived β-value. At stable conditions of nutrient loading and hydraulic flows, a foundation for a system steady-state in particle size distribution is established. A minimum internal water turnover rate (1.25 times/h) is herein suggested as a driving force to achieve and maintain the PSD and β-value in equilibrium throughout a given system, independent of position or time of day.(paper I). The stability or equilibrium of PSD could be used as a tool for evaluating system operation, comparing different systems, and assessing the potential toxicity of accumulated particles within a system. Though type of clarifier affects PSD and solids removal in general, decreasing a microscreen mesh size does not significantly improve water quality (paper II). At long term RAS operations with stable CFB and long hydraulic residence time, the specific action of the several elements that influence PSD will be conserved within the system, and will determine distributions highly dominated by micro-particles (< 20 µm). Consequently, operating a microscreen with a pore size of 20 µm or a microscreen with a pore size of 100 µm does not significantly improve water quality. Although this may be influenced by cakeenhanced filtration and decreased flow handling capacity due to clogging, it indicates that operating a drum filter with a narrow mesh size might not be essential unless there are specific needs for parasite screening, for controlling fluctuating conditions and water quality, or if time to reach PSD equilibrium is a requirement. A direct consequence from this study (paper II), would be that the space required and the expenditures related to installation and operation of a drum filter could potentially be reduced. As aquacultural practices intensify in the future, so might particle accumulation. Recent studies have attempted to assess the importance of particle accumulation on biofilter and fish performance and welfare. There is, however, a large discrepancy in the literature regarding the influence of particulate organic matter on biofilter kinetics, or on fish performance and welfare: - - 40 While some studies suggest impaired nitrification at high organic matter inlet concentrations due to heterotrophic out-competition of nitrifiers (e.g. Zhu and Chen, 1999, 2001; Carrera et al., 2004; Guerdat et al., 2011), others suggest that this effect can be counteracted by proper management of flow and biofilm (e.g. Bovendeur et al., 1990a,b; Melin et al., 2005; Prehn et al., 2012). In paper II and paper III, no real evidence of impaired nitrification was observed when water flow into the biofilters were kept sufficiently high (elevation speed 4-14.4 m/h), regardless of the system C:N ratio (0.7-2.4:1), and biofilter type or mode of operation. The implications of particle interactions with fish in RAS are not yet fully described. Some studies suggest acceptable, and almost improved fish performance in RAS, even at high CFBs (e.g. Pedersen et al., 2012, paper II), while others suggest impaired fish performance due to particle accumulation (e.g. Chapman et al., 1987; Bullock et al., 1994). It seems that certain substances accumulating in RAS have a detrimental effect on fish growth and welfare (e.g. Martins et al., 2009a,b; Davidson et al., 2011a,b), though this does not seem to be related to the accumulation of micro-particles. The relationship between particles and microbiota has been shown in several natural or artificial water bodies, including RAS (e.g. Blancheton and Canaguier, 1995; Hargreaves, 2006; Vadstein et al., 2012). As bacteria search for substrate in the system, they can protect themselves within the tridimensional configuration of a particle, while also ensuring easy and direct access to organic matter (Avnimelech et al., 1986; Azim and Little, 2008; Hess-Erga et al., 2008; Blancheton et al., 2013). This relationship, however, seems to dissipate in more intensive systems, where nutrients are more available for bacteria (paper II and paper III). Although this relationship may affect disinfection activities (e.g. Liltved and Cripps, 1999), particles might also promote microbial maturation in any system with a long hydraulic retention time, such as in RAS (Attramadal et al., 2012a, 2014; Wold et al., 2014) or active suspension ponds (Avnimelech, 1999, 2006, 2007; Ebeling et al., 2006). A stable microbial community could prevent sudden outbursts of opportunistic, potentially pathogenic bacteria, thus safeguarding fish health and welfare. Microbial maturation (e.g. Skjermo et al., 1997; Skjermo and Vadstein, 1999) and bio-floc technology (Crab et al., 2007, 2009; De Schryver et al., 2008; Browdy et al., 2012) seem to be good alternatives for microbial control in aquaculture. Although micro-particles have been generally described as potentially harmful to fish and system performance in aquaculture, during these past 3 years, I found no indisputable evidence to support this theory. Though this final assessment may affect the way we view accumulated particles and their interactions in aquaculture systems, the consummated measurements were limited to a specific particle size range defined by the machine in use. Potentially, these interactions and implications can be further tested in other particle size ranges, also dealing with other particle chemical, physical, or electrostatic properties, such as the zeta potential (Tiller and O’Melia, 1993; Brezesinski and Möhwald, 2003; Leiknes et al., 2004; Nowack and Bucheli, 2007; Zhang et al., 2009). In turn, this may also lead to new and improved particle removal mechanisms, such as ultra-, nano-, and micro-filtration (e.g. Viadero Jr. and Noblet, 2002; Chiam and Sarbatly, 2011; Holan et al., 2013, 2014), foam fractionation (e.g. Weeks et al., 1992; Timmons et al., 1995; Brambilla et al., 2008; Barrut et al., 2013), or flocculation (e.g. Avnimelech, 1999; Crab et al., 2007; Ritvo et al., 2003). Otherwise, other particle control mechanisms could be derived from improvements in feed formulation or binder inclusion in the feed (e.g. Brinker et al., 2005a,b; Brinker, 2007, 2009; Unger and Brinker, 2013), thus improving the mechanical properties of the fish feces. Particles in RAS are usually overlooked, and they require further attention to assess how or whether their accumulation in RAS may pose a threat to fish or system performance. 41 42 9. BIBLIOGRAPHY Ali, S.A., 2013. Design and evaluate a drum screen filter driven by undershot waterwheel for aquaculture recirculating systems. Aquac. Eng. 54, 38–44. doi:10.1016/j.aquaeng.2012.10.006 Amann, R., Kühl, M., 1998. In situ methods for assessment of microorganisms and their activities. Curr. Opin. Microbiol. 1, 352–358. doi:10.1016/S1369-5274(98)80041-6 Arvin, E., Harremoës, P., 1990. Concepts and models for biofilm reactor performance. Wat. Sci. Tech. 22, 171–192. Attramadal, K.J.K., Salvesen, I., Xue, R., Øie, G., Størseth, T.R., Vadstein, O., Olsen, Y., 2012. Recirculation as a possible microbial control strategy in the production of marine larvae. Aquac. Eng. 46, 27–39. doi:10.1016/j.aquaeng.2011.10.003 Attramadal, K.J.K., Truong, T.M.H., Bakke, I., Skjermo, J., Olsen, Y., Vadstein, O., 2014. RAS and microbial maturation as tools for K-selection of microbial communities improve survival in cod larvae. Aquaculture 432, 483–490. doi:10.1016/j.aquaculture.2014.05.052 Avnimelech, Y., 1999. Carbon/nitrogen ratio as a control element in aquaculture systems. Aquaculture 176, 227–235. Avnimelech, Y., 2006. Bio-filters: The need for an new comprehensive approach. Aquac. Eng. 34, 172–178. doi:10.1016/j.aquaeng.2005.04.001 Avnimelech, Y., 2007. Feeding with microbial flocs by tilapia in minimal discharge bio-flocs technology ponds. Aquaculture 264, 140–147. doi:10.1016/j.aquaculture.2006.11.025 Avnimelech, Y., Ritvo, G., 2003. Shrimp and fish pond soils: Processes and management. Aquaculture 220, 549–567. doi:10.1016/S0044-8486(02)00641-5 Avnimelech, Y., Weber, B., Hepher, B., Milstein, A., Zorn, M., 1986. Studies in circulated fish ponds: organic matter recycling and nitrogen transformation. Aquac. Fish. Manag. 231–242. Azim, M.E., Little, D.C., 2008. The biofloc technology (BFT) in indoor tanks: Water quality, biofloc composition, and growth and welfare of Nile tilapia (Oreochromis niloticus). Aquaculture 283, 29–35. doi:10.1016/j.aquaculture.2008.06.036 Balmat, J.L., 1957. Biochemical oxidation of various particulate fractions of sewage. Sewage Ind. Waste. 29, 757–761. Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbial products (SMP) in wastewater treatment systems. Water Res. 33, 3063–3082. doi:10.1016/S0043-1354(99)00022-6 Barrut, B., Blancheton, J.P., Callier, M., Champagne, J.-Y., Grasmick, A., 2013. Foam fractionation efficiency of a vacuum airlift—Application to particulate matter removal in recirculating systems. Aquac. Eng. 54, 16–21. doi:10.1016/j.aquaeng.2012.10.003 43 Bergheim, A., Brinker, A., 2003. Effluent treatment for flow through systems and European Environmental Regulations. Aquac. Eng. 27, 61–77. Bergheim, A., Sanni, S., Indrevik, G., Hølland, P., 1993. Sludge removal from salmonid tank effluent using rotating microsieves. Aquac. Eng. 12, 97–109. doi:10.1016/0144-8609(93)90019-8 Bernet, D., Schmidt, H., Meier, W., Burkhardt-Holm, P., Wahli, T., 1999. Histopathology in fish: Proposal for a protocol to assess aquatic pollution. J. Fish Dis. 22, 25–34. doi:10.1046/j.13652761.1999.00134.x Blancheton, J.P., Attramadal, K.J.K., Michaud, L., D’Orbcastel, E.R., Vadstein, O., 2013. Insight into bacterial population in aquaculture systems and its implication. Aquac. Eng. 53, 30–39. doi:10.1016/j.aquaeng.2012.11.009 Blancheton, J.P., Canaguier, B., 1995. Bacteria and particulate materials in recirculating seabass (Dicentrarchus labrax) production system. Aquaculture 133, 215–224. doi:10.1016/00448486(94)00387-4 Bouwer, E.J., 1987. Theoretical investigation of particle deposition in biofilm systems. Water Res. 21, 1489–1498. doi:10.1016/0043-1354(87)90132-1 Bovendeur, J., Eding, E.H., Henken, A.M., 1987. Design and performance of a water recirculation system for high-density culture of the african catfish, Clarias gariepinus (Burchell 1822). Aquaculture 63, 329–353. Bovendeur, J., Spuij, F., Hermans, P.G.B., Klapwijk, A., 1990a. Mechanisms and kinetics of COD removal and simultaneous oxygen consumption in a rotating biological contactor biofilm. Wat. Sci. Tech. 22, 75–87. Bovendeur, J., Zwaga, A.B., Lobee, B.G.J., Blom, J.H., 1990b. Fixed-biofilm reactors in aquacultural water recycle systems: effect of organic matter elimination on nitrification kinetics. Water Res. 24, 207–213. Brambilla, F., Antonini, M., Ceccuzzi, P., Terova, G., Saroglia, M., 2008. Foam fractionation efficiency in particulate matter and heterotrophic bacteria removal from a recirculating seabass (Dicentrarchus labrax) system. Aquac. Eng. 39, 37–42. doi:10.1016/j.aquaeng.2008.05.004 Brezesinski, G., Möhwald, H., 2003. Advances in Colloid and Interface Science 100--102, 563– 584. doi:10.1016/j.cis.2012.08.001 Brinker, A., 2007. Guar gum in rainbow trout (Oncorhynchus mykiss) feed: The influence of quality and dose on stabilisation of faecal solids. Aquaculture 267, 315–327. doi:10.1016/j.aquaculture.2007.02.037 Brinker, A., 2009. Improving the mechanical characteristics of faecal waste in rainbow trout: the influence of fish size and treatment with a non-starch polysaccharide (guar gum). Aquac. Nutr. 15, 229–240. doi:10.1111/j.1365-2095.2008.00587.x Brinker, A., Koppe, W., Rösch, R., 2005a. Optimised effluent treatment by stabilised trout faeces. Aquaculture 249, 125–144. doi:10.1016/j.aquaculture.2004.12.029 44 Brinker, A., Koppe, W., Rösch, R., 2005b. Optimizing Trout Farm Effluent Treatment by Stabilizing Trout Feces: A Field Trial. N. Am. J. Aquac. 67, 244–258. doi:10.1577/A04-078.1 Brinker, A., Rösch, R., 2005. Factors determining the size of suspended solids in a flow-through fish farm. Aquac. Eng. 33, 1–19. doi:10.1016/j.aquaeng.2004.10.003 Brinker, A., Schröder, H.G., Rösch, R., 2005c. A high-resolution technique to size suspended solids in flow-through fish farms. Aquac. Eng. 32, 325–341. doi:10.1016/j.aquaeng.2004.07.002 Browdy, C.L., Ray, A.J., Leffler, J.W., Avnimelech, Y., 2012. Biofloc-based Aquaculture Systems. Aquac. Prod. Syst. 278–307. doi:10.1002/9781118250105.ch12 Bucher, F., Hofer, R., 1993. The effects of treated domestic sewage on three organs (gills, kidney, liver) of brown trout (Salmo trutta). Water Res. 27, 255–261. doi:10.1016/0043-1354(93)90083-T Bullock, G.L., Herman, R., Heinen, J.M., Nobel, A., Weber, A., Hankins, J.A., 1994. Observations on the occurrence of bacterial gill disease and amoebae gill infestation in rainbow trout culture in a water recirculation system. J. Aquat. Anim. Health 6, 310–317. Bullock, G.L., Summerfelt, S.T., Noble, A.C., Weber, A.L., Durant, M.D., Hankins, J.A., 1997. Ozonation of a recirculating rainbow trout culture system I. Effects on bacterial gill disease and heterotrophic bacteria. Aquaculture 158, 43–55. doi:10.1016/S0044-8486(97)00063-X Bureau, D.P., Hua, K., 2010. Towards effective nutritional management of waste outputs in aquaculture, with particular reference to salmonid aquaculture operations. Aquac. Res. 41, 777– 792. doi:10.1111/j.1365-2109.2009.02431.x Canziani, R., Emondi, V., Garavaglia, M., Malpei, F., Pasinetti, E., Buttiglieri, G., 2006. Effect of oxygen concentration on biological nitrification and microbial kinetics in a cross-flow membrane bioreactor (MBR) and moving-bed biofilm reactor (MBBR) treating old landfill leachate. J. Memb. Sci. 286, 202–212. doi:10.1016/j.memsci.2006.09.044 Carrera, J., Vicent, T., Lafuente, J., 2004. Effect of influent COD/N ratio on biological nitrogen removal (BNR) from high-strength ammonium industrial wastewater. Process Biochem. 39, 2035–2041. doi:10.1016/j.procbio.2003.10.005 Chapman, P.M., Popham, J.D., Griffin, J., Leslie, D., Michaelson, J., 1987. Differentiation of physical from chemical toxicity in solid waste fish bioassays. Water, Air Soil Pollut. 33, 295–308. Chen, Q.L., Luo, Z., Zheng, J.L., Li, X.D., Liu, C.X., Zhao, Y.H., Gong, Y., 2012. Protective effects of calcium on copper toxicity in Pelteobagrus fulvidraco: Copper accumulation, enzymatic activities, histology. Ecotoxicol. Environ. Saf. 76, 126–134. doi:10.1016/j.ecoenv.2011.10.007 Chen, S., Coffin, D.E., Malone, R.F., 1997. Sludge production and management for Recirculating Aquacultural Systems. J. World Aquac. Soc. 28, 303–315. doi:10.1111/j.17497345.1997.tb00278.x Chen, S., Timmons, M.B., Aneshansley, D.J., Bisogni Jr., J.J., 1993. Suspended solids characteristics from recirculating aquacultural systems and design implications. Aquaculture 112, 143–155. 45 Chiam, C., Sarbatly, R., 2011. Purification of Aquacultural Water: Conventional and New Membrane-based Techniques. Sep. Purif. Rev. 40, 126–160. doi:10.1080/15422119.2010.549766 Clark, E.R., Harman, J.P., Forster, J.R.M., 1985. Production of metabolic and waste products by intensively farmed rainbow trout, Salmo gairdneri Richardson. J. Fish Biol. 27, 381–393. doi:10.1111/j.1095-8649.1985.tb03187.x Colt, J., 2006. Water quality requirements for reuse systems. Aquac. Eng. 34, 143–156. doi:10.1016/j.aquaeng.2005.08.011 Colt, J., Lamoureux, J., Patterson, R.N., Rogers, G.L., 2006. Reporting standards for biofilter performance studies. Aquac. Eng. 34, 377–388. doi:10.1016/j.aquaeng.2005.09.002 Colt, J., Watten, B., 1988. Applications of pure oxygen in fish culture. Aquac. Eng. 7, 397–441. doi:10.1016/0144-8609(88)90003-9 Couturier, M., Trofimencoff, T., Buil, J.U., Conroy, J., 2009. Solids removal at a recirculating salmon-smolt farm. Aquac. Eng. 41, 71–77. doi:10.1016/j.aquaeng.2009.05.001 Crab, R., Avnimelech, Y., Defoirdt, T., Bossier, P., Verstraete, W., 2007. Nitrogen removal techniques in aquaculture for a sustainable production. Aquaculture 270, 1–14. doi:10.1016/j.aquaculture.2007.05.006 Crab, R., Kochva, M., Verstraete, W., Avnimelech, Y., 2009. Bio-flocs technology application in over-wintering of tilapia. Aquac. Eng. 40, 105–112. doi:10.1016/j.aquaeng.2008.12.004 Cripps, S.J., 1994. Minimizing outputs: treatment. J. Appl. Ichthyol. 10, 284–294. doi:10.1111/j.1439-0426.1994.tb00168.x Cripps, S.J., 1995. Serial particle size fractionation and characterisation of an aquacultural effluent. Aquaculture 133, 323–339. doi:10.1016/0044-8486(95)00021-S Cripps, S.J., Bergheim, A., 2000. Solids management and removal for intensive land-based aquaculture production systems. Aquac. Eng. 22, 33–56. doi:10.1016/S0144-8609(00)00031-5 Dalsgaard, J., Ekmann, K.S., Pedersen, P.B., Verlhac, V., 2009. Effect of supplemented fungal phytase on performance and phosphorus availability by phosphorus-depleted juvenile rainbow trout (Oncorhynchus mykiss), and on the magnitude and composition of phosphorus waste output. Aquaculture 286, 105–112. doi:10.1016/j.aquaculture.2008.09.007 Dalsgaard, J., Larsen, B.K., Pedersen, P.B., 2015. Nitrogen waste from rainbow trout (Oncorhynchus mykiss) with particular focus on urea. Aquac. Eng. 65, 2–9. doi:10.1016/j.aquaeng.2014.10.004 Dalsgaard, J., Lund, I., Thorarinsdottir, R., Drengstig, A., Arvonen, K., Pedersen, P.B., 2013. Farming different species in RAS in Nordic countries: Current status and future perspectives. Aquac. Eng. 53, 2–13. doi:10.1016/j.aquaeng.2012.11.008 Dalsgaard, J., Pedersen, P.B., 2011. Solid and suspended/dissolved waste (N, P, O) from rainbow trout (Oncorynchus mykiss). Aquaculture 313, 92–99. doi:10.1016/j.aquaculture.2011.01.037 46 Davidson, J., Good, C., Barrows, F.T., Welsh, C., Kenney, P.B., Summerfelt, S.T., 2013. Comparing the effects of feeding a grain- or a fish meal-based diet on water quality, waste production, and rainbow trout Oncorhynchus mykiss performance within low exchange water recirculating aquaculture systems. Aquac. Eng. 52, 45–57. doi:10.1016/j.aquaeng.2012.08.001 Davidson, J., Good, C., Welsh, C., Brazil, B., Summerfelt, S.T., 2009. Heavy metal and waste metabolite accumulation and their potential effect on rainbow trout performance in a replicated water reuse system operated at low or high system flushing rates. Aquac. Eng. 41, 136–145. doi:10.1016/j.aquaeng.2009.04.001 Davidson, J., Good, C., Welsh, C., Summerfelt, S.T., 2011a. Abnormal swimming behavior and increased deformities in rainbow trout Oncorhynchus mykiss cultured in low exchange water recirculating aquaculture systems. Aquac. Eng. 45, 109–117. doi:10.1016/j.aquaeng.2011.08.005 Davidson, J., Good, C., Welsh, C., Summerfelt, S.T., 2011b. The effects of ozone and water exchange rates on water quality and rainbow trout Oncorhynchus mykiss performance in replicated water recirculating systems. Aquac. Eng. 44, 80–96. doi:10.1016/j.aquaeng.2011.04.001 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, 30–40. doi:10.1016/j.aquaeng.2014.01.003 Davidson, J., Summerfelt, S.T., 2004. Solids flushing, mixing, and water velocity profiles within large (10 and 150 m3) circular “Cornell-type” dual-drain tanks. Aquac. Eng. 32, 245–271. doi:10.1016/j.aquaeng.2004.03.009 Davidson, J., Summerfelt, S.T., 2005. Solids removal from a coldwater recirculating system— comparison of a swirl separator and a radial-flow settler. Aquac. Eng. 33, 47–61. doi:10.1016/j.aquaeng.2004.11.002 De Beer, D., Kühl, M., 2001. Interfacial microbial mats and biofilms. Benthic Bound. Layer 374– 394. De Beer, D., Stoodley, P., Lewandowski, Z., 1996. Liquid flow and mass transport in heterogeneous biofilms. Water Res. 30, 2761–2765. doi:10.1016/S0043-1354(96)00141-8 De Beer, D., Stoodley, P., Roe, F., Lewandowski, Z., 1994. Effects of biofilm structures on oxygen distribution and mass transport. Biotechnol. Bioeng. 43, 1131–1138. doi:10.1002/bit.260431118 De Kinkelin, P., Gay, M., Forman, S., 2002. The persistence of infectivity of Tetracapsula bryosalmonae-infected water for rainbow trout, Oncorhynchus mykiss (Walbaum). J. Fish Dis. 25, 477–482. De Schryver, P., Crab, R., Defoirdt, T., Boon, N., Verstraete, W., 2008. The basics of bio-flocs technology: The added value for aquaculture. Aquaculture 277, 125–137. doi:10.1016/j.aquaculture.2008.02.019 47 De Schryver, P., Verstraete, W., 2009. Nitrogen removal from aquaculture pond water by heterotrophic nitrogen assimilation in lab-scale sequencing batch reactors. Bioresour. Technol. 100, 1162–1167. doi:10.1016/j.biortech.2008.08.043 Dolan, E., Murphy, N., O’Hehir, M., 2013. Factors influencing optimal micro-screen drum filter selection for recirculating aquaculture systems. Aquac. Eng. 56, 42–50. doi:10.1016/j.aquaeng.2013.04.005 Dreszer, C., Wexler, a. D., Drusová, S., Overdijk, T., Zwijnenburg, a., Flemming, H.-C., Kruithof, J.C., Vrouwenvelder, J.S., 2014. In-situ biofilm characterization in membrane systems using Optical Coherence Tomography: Formation, structure, detachment and impact of flux change. Water Res. 7. doi:10.1016/j.watres.2014.09.006 Droppo, I.G., Leppard, G.G., Flannigan, D.T., Liss, S.N., 1997. The freshwater floc: A functional relationship of water and organic and inorganic floc constituents affecting suspended sediment properties. Water, Air Soil Pollut. 99, 43–54. Ebeling, J.M., Sibrell, P.L., Ogden, S.R., Summerfelt, S.T., 2003. Evaluation of chemical coagulation–flocculation aids for the removal of suspended solids and phosphorus from intensive recirculating aquaculture effluent discharge. Aquac. Eng. 29, 23–42. doi:10.1016/S01448609(03)00029-3 Ebeling, J.M., Timmons, M.B., Bisogni Jr., J.J., 2006. Engineering analysis of the stoichiometry of photoautotrophic, autotrophic, and heterotrophic removal of ammonia–nitrogen in aquaculture systems. Aquaculture 257, 346–358. doi:10.1016/j.aquaculture.2006.03.019 Eding, E.H., Kamstra, A., Verreth, J.A.J., Huisman, E.A., Klapwijk, A., 2006. Design and operation of nitrifying trickling filters in recirculating aquaculture: A review. Aquac. Eng. 34, 234–260. doi:10.1016/j.aquaeng.2005.09.007 Eliosov, B., Argaman, Y., 1995. Hydrolysis of particulate organics in activated sludge systems. Water Res. 29, 155–163. Emparanza, E.J.M., 2009. Problems affecting nitrification in commercial RAS with fixed-bed biofilters for salmonids in Chile. Aquac. Eng. 41, 91–96. doi:10.1016/j.aquaeng.2009.06.010 Fdz-Polanco, F., Méndez, E., Urueña, M. a., Villaverde, S., García, P. a., 2000. Spatial distribution of heterotrophs and nitrifiers in a submerged biofilter for nitrification. Water Res. 34, 4081–4089. doi:10.1016/S0043-1354(00)00159-7 Ferguson, H.W., 2006. Systemic Pathology of Fish: a text and atlas of normal tissues in teleosts and their responses in disease, Second Edi. ed. Scotian Press, London. Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B., 2014. Daily micro particle distribution of an experimental recirculating aquaculture system-A case study. Aquac. Eng. 60, 28–34. doi:10.1016/j.aquaeng.2014.03.007 Fernandes, P.M., Pedersen, L.-F.L.-F., Pedersen, P.B.P.B., 2015. Microscreen effects on water quality in replicated recirculating aquaculture systems. Aquac. Eng. 65, 17–26. doi:10.1016/j.aquaeng.2014.10.007 48 Figueroa, L.A., Silverstein, J., 1992. The effect of particulate organic matter on biofilm nitrification. Water Environ. Res. 64, 728–733. Foy, R.H., Rosell, R., 1978. Fractionation of phosphorus and nitrogen loadings from a Northern Ireland fish farm. Aquaculture 96, 31–42. Foy, R.H., Rosell, R., 1991. Loadings of nitrogen and phosphorus from a Northern Ireland fish farm. Aquaculture 96, 17–30. Franco-Nava, M.A., Blancheton, J.P., Deviller, G., Charrier, A., Le-Gall, J.-Y., 2004a. Effect of fish size and hydraulic regime on particulate organic matter dynamics in a recirculating aquaculture system: elemental carbon and nitrogen approach. Aquaculture 239, 179–198. doi:10.1016/j.aquaculture.2004.05.040 Franco-Nava, M.A., Blancheton, J.P., Deviller, G., Le-Gall, J.-Y., 2004b. Particulate matter dynamics and transformations in a recirculating aquaculture system: application of stable isotope tracers in seabass rearing. Aquac. Eng. 31, 135–155. doi:10.1016/j.aquaeng.2004.01.003 Garcia-Ruiz, R., Hall, G.H., 1996. Phosphorus fractionation and mobility in the food and faeces of hatchery reared rainbow trout (Onchorhynchus mykiss). Aquaculture 145, 183–193. Golz, W.J., Rusch, K.A., Malone, R.F., 1999. Modeling the major limitations on nitrification in floating-bead filters. Aquac. Eng. 20, 43–61. doi:10.1016/S0144-8609(99)00008-4 Gonenc, E., Harremoës, P., 1990. Nitrification in rotating disc systems - II. Criteria for simultaneous mineralization and nitrification. Water Res. 24, 499–505. doi:10.1016/00431354(90)90235-X Guellil, A., Boualam, M., Quiquampoix, H., Ginestet, P., Audic, J.M., Block, J.C., 2001. Hydrolysis of wastewater colloidal organic matter by extra- cellular enzymes extracted from activated sludge flocs. Wat. Sci. Tech. 43, 33–40. Guerdat, T.C., Losordo, T.M., Classen, J.J., Osborne, J.A., DeLong, D., 2011. Evaluating the effects of organic carbon on biological filtration performance in a large scale recirculating aquaculture system. Aquac. Eng. 44, 10–18. doi:10.1016/j.aquaeng.2010.10.002 Gutierrez-Wing, M.T., Malone, R.F., 2006. Biological filters in aquaculture: Trends and research directions for freshwater and marine applications. Aquac. Eng. 34, 163–171. doi:10.1016/j.aquaeng.2005.08.003 Hagopian, D.S., Riley, J.G., 1998. A closer look at the bacteriology of nitrification. Aquac. Eng. 18, 223–244. doi:10.1016/S0144-8609(98)00032-6 Hargreaves, J. a., 2006. Photosynthetic suspended-growth systems in aquaculture. Aquac. Eng. 34, 344–363. doi:10.1016/j.aquaeng.2005.08.009 Heinen, J.M., Hankins, J.A., Adler, P.R., 1996. Water quality and waste production in a recirculating trout-culture system with feeding of a higher-energy or a lower-energy diet. Aquac. Res. 27, 699–710. doi:10.1111/j.1365-2109.1996.tb01305.x 49 Hem, L.J., Rusten, B., Ødegaard, H., 1994. Nitrification in a moving bed biofilm reactor. Water Res. 28, 1425–1433. Henderson, J.P., Bromage, N.R., 1988. Optimising the removal of suspended solids from aquacultural effluents in settlement lakes. Aquac. Eng. 7, 167–181. doi:10.1016/01448609(88)90019-2 Henze, M., 2002. Basic Biological Processes, in: Henze, M., Harremoës, P., Jes la Cour, J., Arvin, E. (Eds.), Wastewater Treatment: Biological and Chemical Processes. Springer-Verlag, Berlin, Heidelberg, p. 430. Hess-Erga, O.K., Attramadal, K.J.K., Vadstein, O., 2008. Biotic and abiotic particles protect marine heterotrophic bacteria during UV and ozone disinfection. Aquat. Biol. 4, 147–154. doi:10.3354/ab00105 Hill, P.J., Ng, K.M., 2002. Particle size distribution by design. Chem. Eng. Sci. 57, 2125–2138. doi:10.1016/S0009-2509(02)00106-9 Holan, A.B., Wold, P.-A., Leiknes, T.O., 2014. Membrane performance and fouling behavior of membrane bioreactors installed in marine recirculating aquaculture systems. Aquac. Eng. 58, 45– 51. doi:10.1016/j.aquaeng.2013.10.002 Holan, A.B., Wold, P.-A., Øie, G., Leiknes, T.O., 2013. Integrated Membrane Bioreactor for Water Quality Control in Marine Recirculating Aquaculture Systems. Sep. Sci. Technol. 48, 1758–1767. doi:10.1080/01496395.2013.774818 Hussenot, J., 2003. Emerging effluent management strategies in marine fish-culture farms located in European coastal wetlands. Aquaculture 226, 113–128. doi:10.1016/S0044-8486(03)00472-1 Iritani, E., Katagiri, N., Kanetake, S., 2012. Determination of cake filtration characteristics of dilute suspension of bentonite from various filtration tests. Sep. Purif. Technol. 92, 143–151. doi:10.1016/j.seppur.2011.05.011 Ivanovic, I., Leiknes, T.O., 2008. Impact of aeration rates on particle colloidal fraction in the biofilm membrane bioreactor (BF-MBR). Desalination 231, 182–190. doi:10.1016/j.desal.2007.11.046 Ivanovic, I., Leiknes, T.O., 2012. Particle Separation in Moving Bed Biofilm Reactor: Applications and Opportunities. Sep. Sci. Technol. 47, 647–653. doi:10.1080/01496395.2011.639590 Janning, K.F., Mesterton, K., Harremoës, P., 1997. Hydrolysis and degradation of filtrated organic particulates in a biofilm reactor under anoxic and aerobic con. Wat. Sci. Tech. 36, 279–286. Janning, K.F., Tallec, X. Le, Harremoës, P., 1998. Hydrolysis of organic wastewater particles in laboratory scale and pilot scale biofilm reactors under anoxic and aerobic conditions. Wat. Sci. Tech. Jokumsen, A., Svendsen, L.M., 2010. Farming of freshwater rainbow trout in Denmark. Charlottenlund, Copenhagen. 50 Kelly, L.A., Bergheim, A., Stellwagen, J., 1997. Particle size distribution of wastes from freshwater fish farms. Aquac. Int. 5, 65–78. Khan, M.M.T., Takizawa, S., Lewandowski, Z., Jones, W.L., Camper, A.K., Katayama, H., Kurisu, F., Ohgaki, S., 2011. Membrane fouling due to dynamic particle size changes in the aerated hybrid PAC-MF system. J. Memb. Sci. 371, 99–107. doi:10.1016/j.memsci.2011.01.017 Kolarevic, J., Baeverfjord, G., Takle, H., Ytteborg, E., Reiten, B.K.M., Nergård, S., Terjesen, B.F., 2014. Performance and welfare of Atlantic salmon smolt reared in recirculating or flow through aquaculture systems. Aquaculture 432, 15–25. doi:10.1016/j.aquaculture.2014.03.033 Krumins, V., Ebeling, J.M., Wheaton, F., 2001a. Part-day ozonation for nitrogen and organic carbon control in recirculating aquaculture systems. Aquac. Eng. 24, 231–241. doi:10.1016/S0144-8609(01)00061-9 Krumins, V., Ebeling, J.M., Wheaton, F., 2001b. Ozone’s effects on power-law particle size distribution in recirculating aquaculture systems. Aquac. Eng. 25, 13–24. doi:10.1016/S01448609(01)00064-4 Lake, R.G., Hinch, S.G., 1999. Acute effects of suspended sediment angularity on juvenile coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 56, 862–867. doi:10.1139/f99-024 Langer, J., Efthimiou, S., Rosenthal, H., Bronzi, P., 1996. Drum filter performance in a recirculating eel culture unit. J. Appl. Ichthyol. 12, 61–65. Larsen, T.A., Harremoës, P., 1994. Degradation mechanisms of colloidal organic matter in biofilm reactors. Water Res. 28, 1443–1452. Laspidou, C.S., Rittman, B.E., 2002. A unified theory for extracellular polymeric substances, soluble microbial products, and active and inert bi omass. Water Res. 36, 2711–2720. Lee, J., 2014. Separation of fine organic particles by a low-pressure hydrocyclone (LPH). Aquac. Eng. doi:10.1016/j.aquaeng.2014.07.002 Leiknes, T., Ødegaard, H., Myklebust, H., 2004. Removal of natural organic matter (NOM) in drinking water treatment by coagulation-microfiltration using metal membranes. J. Memb. Sci. 242, 47–55. doi:10.1016/j.memsci.2004.05.010 Leiknes, T.O., Ivanovic, I., Ødegaard, H., 2006. Investigating the effect of colloids on the performance of a biofilm membrane reactor (BF-MBR) for treatment of municipal wastewater, in: Water Institute of South Africa Biennial Conference. Durban, pp. 708–714. doi:10.4314/wsa.v33i2.49099 Leonard, N., Blancheton, J.P., Guiraud, J.., 2000. Populations of heterotrophic bacteria in an experimental recirculating aquaculture system. Aquac. Eng. 22, 109–120. doi:10.1016/S01448609(00)00035-2 Leonard, N., Guiraud, J.., Gasset, E., Cailleres, J.., Blancheton, J.P., 2002. Bacteria and nutrients—nitrogen and carbon—in a recirculating system for sea bass production. Aquac. Eng. 26, 111–127. doi:10.1016/S0144-8609(02)00008-0 51 Levine, A.D., Tchobanoglous, G., Asano, T., 1985. Characterization of the size distribution of contaminants in wastewater: Treatment and reuse implications. Water Pollut. Control Fed. 57, 805–816. doi:10.2307/25042701 Levine, A.D., Tchobanoglous, G., Asano, T., 1991. Size distributions of particulate contaminants in wastewater and their impact on treatability. Water Res. 25, 911–922. doi:10.1016/00431354(91)90138-G Liao, P.B., Mayo, R.D., 1972. Salmonid hatchery water reuse systems. Aquaculture 1, 317–335. Liltved, H., Cripps, S.J., 1999. Removal of particle-associated bacteria by pre®ltration and ultraviolet irradiation. Aquaculture 445–450. doi:10.1046/j.1365-2109.1999.00349.x Liltved, H., Hansen, B.R., 1990. Screening as a method for removal of parasites from inlet water to fish farms. Aquac. Eng. 9, 209–215. Ling, J., Chen, S., 2005. Impact of organic carbon on nitrification performance of different biofilters. Aquac. Eng. 33, 150–162. doi:10.1016/j.aquaeng.2004.12.002 Liss, S.N., Droppo, I.G., Flannigan, D.T., Leppard, G.G., 1996. Floc architecture in wastewater and natural riverine systems. Environ. Sci. Technol. 30, 680–686. Liu, H., Che, X., Zhang, Y., 2013. Performance of sequencing microbead biofilters in a recirculating aquaculture system. Aquac. Eng. 52, 80–86. doi:10.1016/j.aquaeng.2012.10.002 Ljunggren, M., 2006. Micro screening in wastewater treatment – an overview. Vatten 62, 171– 177. Loyless, J.C., Malone, R.F., 1997. A sodium bicarbonate dosing methodology for pH management in freshwater-recirculating aquaculture systems. Progress. Fish-Culturist 59, 198– 205. Magor, B.G., 1988. Gill Histopathology of Juvenile Oncorhynchus-Kisutch Exposed to Suspended Wood Debris. Can. J. Zool. Can. Zool. 66, 2164–2169. Mallat, J., 1985. Fish gill structural changes induced by toxicants and other irritants: A statistical review. Can. J. Fish. Aquat. Sci. 42, 630–648. Malone, R.F., Pfeiffer, T.J., 2006. Rating fixed film nitrifying biofilters used in recirculating aquaculture systems. Aquac. Eng. 34, 389–402. doi:10.1016/j.aquaeng.2005.08.007 Marquet, R., Mietton-Peuchot, M., Wheatley, A.D., 1999. Characterisation of trickling filter effluent by particle size distribution and high performance size exclusion chromatography. Water Res. 33, 1415–1424. doi:10.1016/S0043-1354(98)00337-6 Marquet, R., Muhammad, N., Vairavamoorthy, K., Wheatley, A.D., 2007. Particle size distribution to assess the performance of trickling filters. Process Saf. Environ. Prot. 85, 99–103. doi:10.1205/psep.05194 52 Martins, C.I.M., Eding, E.H., Verdegem, M.C.J., Heinsbroek, L.T.N., Schneider, O., Blancheton, J.P., Roque d’Orbcastel, E., Verreth, J.A.J., 2010. New developments in recirculating aquaculture systems in Europe: A perspective on environmental sustainability. Aquac. Eng. 43, 83–93. doi:10.1016/j.aquaeng.2010.09.002 Martins, C.I.M., Ochola, D., Ende, S.S.W., Eding, E.H., Verreth, J.A.J., 2009a. Is growth retardation present in Nile tilapia Oreochromis niloticus cultured in low water exchange recirculating aquaculture systems? Aquaculture 298, 43–50. doi:10.1016/j.aquaculture.2009.09.030 Martins, C.I.M., Pistrin, M.G., Ende, S.S.W., Eding, E.H., Verreth, J.A.J., 2009b. The accumulation of substances in Recirculating Aquaculture Systems (RAS) affects embryonic and larval development in common carp Cyprinus carpio. Aquaculture 291, 65–73. doi:10.1016/j.aquaculture.2009.03.001 Masaló, I., Oca, J., 2013. Hydrodynamics in a multivortex aquaculture tank: Effect of baffles and water inlet characteristics. Aquac. Eng. doi:10.1016/j.aquaeng.2013.11.001 Maxey, M.R., Chang, E.J., Wang, L.P., 1996. Interactions of Particles and Microbubbles with Turbulence. Exp. Therm. Fluid Sci. 12, 417–425. doi:10.1016/0894-1777(95)00130-1 McKenzie, D.J., Pedersen, P.B., Jokumsen, A., 2007. Aspects of respiratory physiology and energetics in rainbow trout (Oncorhynchus mykiss) families with different size-at-age and condition factor. Aquaculture 263, 280–294. doi:10.1016/j.aquaculture.2006.10.022 McMillan, J.., Wheaton, F., Hochheimer, J.., Soares, J., 2003. Pumping effect on particle sizes in a recirculating aquaculture system. Aquac. Eng. 27, 53–59. doi:10.1016/S0144-8609(02)00038-9 Melin, E., Leiknes, T.O., Helness, H., Rasmussen, V., Ødegaard, H., 2005. Effect of organic loading rate on a wastewater treatment process combining moving bed biofilm and membrane reactors. Wat. Sci. Tech. 51, 421–430. Meriac, A., Eding, E.H., Schrama, J., Kamstra, A., Verreth, J. a J., 2014. Dietary carbohydrate composition can change waste production and biofilter load in recirculating aquaculture systems. Aquaculture 420-421, 254–261. doi:10.1016/j.aquaculture.2013.11.018 Meriac, A., Tom, P.A., Eding, E.H., Kamstra, A., Schrama, J.W., Verreth, J.A.J., 2015. Effects of diet composition and ultrasound treatment on particle size distribution and carbon bioavailability in feces of rainbow trout. Aquac. Eng. 65, 10–16. doi:10.1016/j.aquaeng.2014.12.002 Merino, G.E., Piedrahita, R., Conklin, D.E., 2007. Settling characteristics of solids settled in a recirculating system for California halibut (Paralichthys californicus) culture. Aquac. Eng. 37, 79– 88. doi:10.1016/j.aquaeng.2007.01.006 Michaud, L., Blancheton, J.P., Bruni, V., Piedrahita, R., 2006. Effect of particulate organic carbon on heterotrophic bacterial populations and nitrification efficiency in biological filters. Aquac. Eng. 34, 224–233. doi:10.1016/j.aquaeng.2005.07.005 Michaud, L., Lo Giudice, a., Troussellier, M., Smedile, F., Bruni, V., Blancheton, J.P., 2009. Phylogenetic characterization of the heterotrophic bacterial communities inhabiting a marine 53 recirculating aquaculture system. J. Appl. Microbiol. 107, 1935–1946. doi:10.1111/j.13652672.2009.04378.x Michaud, L., Lo Giudice, A., Interdonato, F., Triplet, S., Ying, L., Blancheton, J.P., 2014. C/N ratio-induced structural shift of bacterial communities inside lab-scale aquaculture biofilters. Aquac. Eng. 58, 77–87. doi:10.1016/j.aquaeng.2013.11.002 Michel, C., Schmidt-Posthaus, H., Burkhardt-Holm, P., 2013. Suspended sediment pulse effects in rainbow trout (Oncorhynchus mykiss ) — relating apical and systemic responses. Can. J. Fish. Aquat. Sci. 70, 630–641. doi:dx.doi.org/10.1139/cjfas-2012-0376 Morgenroth, E., Kommedal, R., Harremoës, P., 2002. Processes and modeling of hydrolysis of particulate organic matter in aerobic wastewater treatment - A review. Water Sci. Technol. 45, 25–40. Mäkinen, T., Lindgren, S., Eskelinen, P., 1988. Sieving as an effluent treatment method for aquaculture. Aquac. Eng. 7, 367–377. Naylor, R.L., Goldburg, R.J., Primavera, J.H., Kautsky, N., Beveridge, M.C., Clay, J., Folke, C., Lubchenco, J., Mooney, H., Troell, M., 2000. Effect of aquaculture on world fish supplies. Nature 405, 1017–24. doi:10.1038/35016500 Nielsen, P., Jahn, a, Palmgren, R., 1997. Conceptual model for production and composition of exopolymers in biofilms. Water Sci. Technol. Noble, A.C., Summerfelt, S.T., 1996. Diseases encountered in rainbow trout cultured in recirculating systems. Annu. Rev. Fish Dis. 6, 65–92. Nogueira, R., Lazarova, V., Manem, J., Melo, L.F., 1998. Influence of dissolved oxygen on the nitrification kinetics in a circulating bed biofilm reactor. Bioprocess Eng. 19, 441–449. doi:10.1007/s004490050546 Nogueira, R., Melo, L.F., Purkhold, U., Wuertz, S., Wagner, M., 2002. Nitrifying and heterotrophic population dynamics in biofilm reactors: Effects of hydraulic retention time and the presence of organic carbon. Water Res. 36, 469–481. doi:10.1016/S0043-1354(01)00229-9 Nowack, B., Bucheli, T.D., 2007. Occurrence, behavior and effects of nanoparticles in the environment. Environ. Pollut. 150, 5–22. doi:10.1016/j.envpol.2007.06.006 Ohashi, A., Silva, D.G.V. de, Mobarry, B., Manem, J.A., Stahl, D.A., Rittmann, B.E., 1995. Influence of substrate C/N ratio on the structure of multi-species biofilms consisting of nitrifiers and heterotrophs. Wat. Sci. Tech. 32, 75–84. Okabe, S., Oozawa, Y., Hirata, K., Watanabe, Y., 1996. Relationship between population dynamics of nitrifiers in biofilms and reactor performance at various C:N ratios. Wat. Res. 30, 1563–1572. Pano, A., Middlebrooks, E.J., 1983. Kinetics of carbon and ammonia nitrogen removal in RBCs. Water Pollut. Control Fed. 55, 956–965. 54 Paredes, D., Kuschk, P., Mbwette, T.S.A., Stange, F., Müller, R.A., Köser, H., 2007. New aspects of microbial nitrogen transformations in the context of wastewater treatment - A review. Eng. Life Sci. 7, 13–25. doi:10.1002/elsc.200620170 Park, J., Kim, Y., Kim, P.K., Daniels, H. V., 2011. Effects of two different ozone doses on seawater recirculating systems for black sea bream Acanthopagrus schlegeli (Bleeker): Removal of solids and bacteria by foam fractionation. Aquac. Eng. 44, 19–24. doi:10.1016/j.aquaeng.2010.11.001 Patterson, R.N., Watts, K.C., 2003a. Micro-particles in recirculating aquaculture systems: microscopic examination of particles. Aquac. Eng. 28, 115–130. doi:10.1016/S01448609(03)00028-1 Patterson, R.N., Watts, K.C., 2003b. Micro-particles in recirculating aquaculture systems: particle size analysis of culture water from a commercial Atlantic salmon site. Aquac. Eng. 28, 99–113. doi:10.1016/S0144-8609(03)00003-7 Patterson, R.N., Watts, K.C., Gill, T.A., 2003. Micro-particles in recirculating aquaculture systems: determination of particle density by density gradient centrifugation. Aquac. Eng. 27, 105–115. doi:10.1016/S0144-8609(02)00042-0 Patterson, R.N., Watts, K.C., Timmons, M.B., 1999. The power law in particle size analysis for aquacultural facilities. Aquac. Eng. 19, 259–273. doi:10.1016/S0144-8609(98)00054-5 Pedersen, L.F., Pedersen, P.B., 2012. Hydrogen peroxide application to a commercial recirculating aquaculture system. Aquac. Eng. 46, 40–46. doi:10.1016/j.aquaeng.2011.11.001 Pedersen, L.-F., Suhr, K.I., Dalsgaard, J., Pedersen, P.B., Arvin, E., 2012. Effects of feed loading on nitrogen balances and fish performance in replicated recirculating aquaculture systems. Aquaculture 338-341, 237–245. doi:10.1016/j.aquaculture.2012.01.035 Pfeiffer, T.J., Osborn, A., Davis, M., 2008. Particle sieve analysis for determining solids removal efficiency of water treatment components in a recirculating aquaculture system. Aquac. Eng. 39, 24–29. doi:10.1016/j.aquaeng.2008.05.003 Pfeiffer, T.J., Wills, P.S., 2011. Evaluation of three types of structured floating plastic media in moving bed biofilters for total ammonia nitrogen removal in a low salinity hatchery recirculating aquaculture system. Aquac. Eng. 45, 51–59. doi:10.1016/j.aquaeng.2011.06.003 Picioreanu, C., van Loosdrecht, M.C., Heijnen, J.J., 2000. A theoretical study on the effect of surface roughness on mass transport and transformation in biofilms. Biotechnol. Bioeng. 68, 355– 369. Piedrahita, R., 2003. Reducing the potential environmental impact of tank aquaculture effluents through intensification and recirculation. Aquaculture 226, 35–44. doi:10.1016/S00448486(03)00465-4 Prehn, J., Waul, C.K., Pedersen, L.-F., Arvin, E., 2012. Impact of water boundary layer diffusion on the nitrification rate of submerged biofilter elements from a recirculating aquaculture system. Water Res. 46, 3516–24. doi:10.1016/j.watres.2012.03.053 55 Randall, D.J., Tsui, T.K.N., 2002. Ammonia toxicity in fish. Mar. Pollut. Bull. 45, 17–23. Rasmussen, K., Lewandowski, Z., 1998. Microelectrode measurements of local mass transport rates in heterogeneous biofilms. Biotechnol. Bioeng. 59, 302–309. doi:10.1002/(SICI)10970290(19980805)59:3<302::AID-BIT6>3.0.CO;2-F Reid, G.K., Liutkus, M., Robinson, S.M.C., Chopin, T.R., Blair, T., Lander, T., Mullen, J., Page, F., Moccia, R.D., 2009. A review of the biophysical properties of salmonid faeces: Implications for aquaculture waste dispersal models and integrated multi-trophic aquaculture. Aquac. Res. 40, 257–273. doi:10.1111/j.1365-2109.2008.02065.x Revsbech, N.P., Jørgensen, B.B., 1986. Microelectrodes: Their use in microbial ecology. Adv. Microb. Ecol. doi:10.1007/978-1-4757-0611-6_7 Rittmann, B.E., 1985. The effect of load fluctuations on the effluent concentration produced by fixed-film reactors. Water Sci. Technol. 17, 45–52. Ritvo, G., Dassa, O., Kochba, M., 2003. Salinity and pH effect on the colloidal properties of suspended particles in super intensive aquaculture systems. Aquaculture 218, 379–386. doi:10.1016/S0044-8486(02)00652-X Rogers, G.L., Klemetson, S.L., 1985. Ammonia removal in selected aquaculture water reuse biofilters. Aquac. Eng. 4, 135–154. doi:10.1016/0144-8609(85)90010-X Roque d’Orbcastel, E., Person-Le Ruyet, J., Le Bayon, N., Blancheton, J.P., 2009. Comparative growth and welfare in rainbow trout reared in recirculating and flow through rearing systems. Aquac. Eng. 40, 79–86. doi:10.1016/j.aquaeng.2008.11.005 Rueter, J., Johnson, R., 1995. The use of ozone to improve solids removal during disinfection. Aquac. Eng. 14, 123–141. Rurangwa, E., Verdegem, M.C.J., 2014. Microorganisms in recirculating aquaculture systems and their management. Rev. Aquac. 5, 1–14. doi:10.1111/raq.12057 Rusten, B., Eikebrokk, B., Ulgenes, Y., Lygren, E., 2006. Design and operations of the Kaldnes moving bed biofilm reactors. Aquac. Eng. 34, 322–331. doi:10.1016/j.aquaeng.2005.04.002 Rusten, B., Ødegaard, H., 2006. Evaluation and testing of fine mesh sieve technologies for primary treatment of municipal wastewater. Wat. Sci. Tech. 54, 31–38. doi:10.2166/wst.2006.710 Sanz, J., Strube, R., Quinzaños, S., Dahl, C.P., Montoliu, I., Gracenea, M., Mujeriego, R., 2009. Helminth eggs removal in water reclamation: Disc filtration as an effective barrier, in: IDA World Congress. pp. 1–14. Satoh, H., Okabe, S., Norimatsu, N., Watanabe, Y., 2000. Significance of substrate C/N ratio on structure and activity of nitrifying biofilms determined by in situ hybridization and the use of microelectrodes. Water Sci. Technol. 41, 317–321. doi:10.2965/jswe.22.763 56 Schmidt, I., Sliekers, O., Schmid, M., Bock, E., Fuerst, J., Kuenen, J.G., Jetten, M.S.M., Strous, M., 2003. New concepts of microbial treatment processes for the nitrogen removal in wastewater. FEMS Microbiol. Rev. 27, 481–492. doi:10.1016/S0168-6445(03)00039-1 Schneider, O., Chabrillon-Popelka, M., Smidt, H., Haenen, O., Sereti, V., Eding, E.H., Verreth, J.A.J., 2007. HRT and nutrients affect bacterial communities grown on recirculation aquaculture system effluents. FEMS Microbiol. Ecol. 60, 207–219. doi:10.1111/j.1574-6941.2007.00282.x Schneider, O., Sereti, V., Eding, E.H., Verreth, J.A.J., 2005. Analysis of nutrient flows in integrated intensive aquaculture systems. Aquac. Eng. 32, 379–401. doi:10.1016/j.aquaeng.2004.09.001 Schreier, H.J., Mirzoyan, N., Saito, K., 2010. Microbial diversity of biological filters in recirculating aquaculture systems. Curr. Opin. Biotechnol. 21, 318–25. doi:10.1016/j.copbio.2010.03.011 Sharma, B., Ahlert, R.C., 1977. Nitrification and nitrogen removal. Water Res. 11, 897–925. Sharrer, M.J., Summerfelt, S.T., Bullock, G.L., Gleason, L.E., Taeuber, J., 2005. Inactivation of bacteria using ultraviolet irradiation in a recirculating salmonid culture system. Aquac. Eng. 33, 135–149. doi:10.1016/j.aquaeng.2004.12.001 Siegrist, H., Gujer, W., 1985. Mass transfer mechanisms in a heterotrophic biofilm. Water Sci. Technol. 17, 1469–1471. doi:10.1016/0043-1354(85)90303-3 Sindilariu, P.D., 2007. Reduction in effluent nutrient loads from flow-through facilities for trout production: A review. Aquac. Res. 38, 1005–1036. doi:10.1111/j.1365-2109.2007.01751.x Sindilariu, P.-D., Brinker, A., Reiter, R., 2009. Waste and particle management in a commercial, partially recirculating trout farm. Aquac. Eng. 41, 127–135. doi:10.1016/j.aquaeng.2009.03.001 Singh, S., Ebeling, J.M., Wheaton, F., 1999. Water quality trials in four recirculating aquacultural system configurations. Aquac. Eng. 20, 75–84. doi:10.1016/S0144-8609(99)00003-5 Sitjà-Bobadilla, A., Padrós, F., Aguilera, C., Alvarez-Pellitero, P., 2005. Epidemiology of Cryptosporidium molnari in Spanish gilthead sea bream (Sparus aurata L.) and European sea bass (Dicentrarchus labrax L.) cultures: from hatchery to market size. Appl. Environ. Microbiol. 71, 131–139. doi:10.1128/AEM.71.1.131 Skjermo, J., Salvesen, I., Øie, G., Olsen, Y., Vadstein, O., 1997. Microbially matured water: a technique for selection of a non-opportunistic bacterial flora in water that may improve performance of marine larvae. Aquac. Int. 5, 13–28. doi:10.1007/BF02764784 Skjermo, J., Vadstein, O., 1999. Techniques for microbial control in the intensive rearing of marine larvae. Aquaculture 177, 333–343. doi:10.1016/S0044-8486(99)00096-4 Stokes, G.G., 1901. Mathematical and Physical Papers - Volume 3. Cambridge University Press, London. doi:http://dx.doi.org/10.1017/CBO9780511702266 57 Suhr, K.I., Pedersen, P.B., 2010. Nitrification in moving bed and fixed bed biofilters treating effluent water from a large commercial outdoor rainbow trout RAS. Aquac. Eng. 42, 31–37. doi:10.1016/j.aquaeng.2009.10.001 Summerfelt, R.C., Penne, C.R., 2005. Solids removal in a recirculating aquaculture system where the majority of flow bypasses the microscreen filter. Aquac. Eng. 33, 214–224. doi:10.1016/j.aquaeng.2005.02.003 Summerfelt, S.T., 2003. Ozonation and UV irradiation - An introduction and examples of current applications. Aquac. Eng. 28, 21–36. doi:10.1016/S0144-8609(02)00069-9 Summerfelt, S.T., Davidson, J., Waldrop, T.B., Tsukuda, S.M., Bebak-Williams, J., 2004. A partial-reuse system for coldwater aquaculture. Aquac. Eng. 31, 157–181. doi:10.1016/j.aquaeng.2004.03.005 Summerfelt, S.T., Sharrer, M.J., Tsukuda, S.M., Gearheart, M., 2009. Process requirements for achieving full-flow disinfection of recirculating water using ozonation and UV irradiation. Aquac. Eng. 40, 17–27. doi:10.1016/j.aquaeng.2008.10.002 Summerfelt, S.T., Vinci, B.J., Piedrahita, R.H., 2000. Oxygenation and carbon dioxide control in water reuse systems. Aquac. Eng. 22, 87–108. doi:10.1016/S0144-8609(00)00034-0 Summerfelt, S.T., Zühlke, A., Kolarevic, J., Reiten, B.K.M., Selset, R., Gutierrez, X., Terjesen, B.F., 2015. Effects of alkalinity on ammonia removal, carbon dioxide stripping, and system pH in semi-commercial scale water recirculating aquaculture systems operated with moving bed bioreactors. Aquac. Eng. 65, 46–54. doi:10.1016/j.aquaeng.2014.11.002 Szwerinski, H., Arvin, E., Harremoës, P., 1986. pH-decrease in nitrifying biofilms. Wat. Res. 20, 971–976. Särner, E., 1986. Removal of particulate and dissolved organics in aerobic fixed-film biological processes. Water Pollut. Control Fed. 58, 165–172. Särner, E., Marklund, S., 1984. Influence of particulate organics on the removal of dissolved organics in fixed-film biological reactors. Wat. Sci. Tech. 17, 15–26. Tal, Y., Watts, J.E.., Schreier, S.B., Sowers, K.R., Schreier, H.J., 2003. Characterization of the microbial community and nitrogen transformation processes associated with moving bed bioreactors in a closed recirculated mariculture system. Aquaculture 215, 187–202. doi:10.1016/S0044-8486(02)00372-1 Thurston, R. V., Russo, R.C., Vinogradov, G.A., 1981. Ammonia toxicity to fishes. Effect of pH on the toxicity of the un-ionized ammonia species. Environ. Sci. Technol. 15, 837–840. doi:10.1021/es00089a012 Tien, C., Teoh, S.K., Tan, R.B.H., 2001. Cake filtration analysis - The effect of the relationship between the pore liquid pressure and the cake compressive stress. Chem. Eng. Sci. 56, 5361– 5369. doi:10.1016/S0009-2509(01)00263-9 58 Tiller, C.L., O’Melia, C.R., 1993. Natural organic matter and colloidal stability: Models and measurements. Colloids Surfaces A Physicochem. Eng. Asp. 73, 89–102. doi:10.1016/09277757(93)80009-4 Timmons, M.B., Chen, S., Weeks, N.C., 1995. Mathematical Model of a Foam Fractionator Used in Aquaculture. J. World Aquac. Soc. 26, 225–233. Timmons, M.B., Ebeling, J.M., 2010. Recirculating Aquaculture, 10th ed, Aquaculture. Timmons, M.B., Summerfelt, S.T., Vinci, B.J., 1998. Review of circular tank technology and management. Aquac. Eng. 18, 51–69. doi:10.1016/S0144-8609(98)00023-5 Twarowska, J.G., Westerman, P.W., Losordo, T.M., 1997. Water treatment and waste characterization evaluation of an intensive recirculating fish production system. Aquac. Eng. 16, 133–147. Unger, J., Brinker, A., 2013. Feed and treat: What to expect from commercial diets. Aquac. Eng. 53, 19–29. doi:10.1016/j.aquaeng.2012.11.012 Vadstein, O., Andersen, T., Reinertsen, H.R., Olsen, Y., 2012. Carbon, nitrogen and phosphorus resource supply and utilisation for coastal planktonic heterotrophic bacteria in a gradient of nutrient loading. Mar. Ecol. Prog. Ser. 447, 55–75. doi:10.3354/meps09473 Vadstein, O., Bergh, Ø., Gatesoupe, F.-J., Galindo-Villegas, J., Mulero, V., Picchietti, S., Scapigliati, G., Makridis, P., Olsen, Y., Dierckens, K., Defoirdt, T., Boon, N., De Schryver, P., Bossier, P., 2013. Microbiology and immunology of fish larvae. Rev. Aquac. 5, S1–S25. doi:10.1111/j.1753-5131.2012.01082.x Van Bussel, C.G.J., Mahlmann, L., Kroeckel, S., Schroeder, J.P., Schulz, C., 2013. The effect of high ortho-phosphate water levels on growth, feed intake, nutrient utilization and health status of juvenile turbot (Psetta maxima) reared in intensive recirculating aquaculture systems (RAS). Aquac. Eng. 57, 63–70. doi:10.1016/j.aquaeng.2013.08.003 Van Kessel, M. a H.J., Harhangi, H.R., van de Pas-Schoonen, K., van de Vossenberg, J., Flik, G., Jetten, M.S.M., Klaren, P.H.M., Op den Camp, H.J.M., 2010. Biodiversity of N-cycle bacteria in nitrogen removing moving bed biofilters for freshwater recirculating aquaculture systems. Aquaculture 306, 177–184. doi:10.1016/j.aquaculture.2010.05.019 Van Rijn, J., 1996. The potential for integrated biological treatment systems in recirculating fish culture—A review. Aquaculture 139, 181–201. doi:10.1016/0044-8486(95)01151-X Van Rijn, J., 2013. Waste treatment in recirculating aquaculture systems. Aquac. Eng. 53, 49–56. doi:10.1016/j.aquaeng.2012.11.010 Veerapen, J.P., Lowry, B.J., Couturier, M.F., 2005. Design methodology for the swirl separator. Aquac. Eng. 33, 21–45. doi:10.1016/j.aquaeng.2004.11.001 Verdegem, M.C.J., 2013. Nutrient discharge from aquaculture operations in function of system design and production environment. Rev. Aquac. 5, 158–171. doi:10.1111/raq.12011 59 Verdegem, M.C.J., Bosma, R.H., Verreth, J.A.J., 2006. Reducing water use for animal production through aquaculture. Int. J. Water Resour. Dev 22, 101–113. Verhagen, F.J.M., Laanbroek, H.J., 1991. Competition for ammonium between nitrifying and heterotrophic bacteria in dual energy-limited chemostats. Appl. Environ. Microbiol. 57, 3255– 3263. Viadero Jr., R.C., Noblet, J.A., 2002. Membrane filtration for removal of fine solids from aquaculture process water. Aquac. Eng. 26, 151–169. doi:10.1016/S0144-8609(02)00011-0 Vigne, E., Choubert, J.-M., Canler, J.-P., Héduit, A., Sorensen, K., Lessard, P., 2010. A biofiltration model for tertiary nitrification of municipal wastewaters. Water Res. 44, 4399–4410. doi:10.1016/j.watres.2010.06.005 Vigne, E., Choubert, J.-M., Canler, J.-P., Heduit, A., Sørensen, K.H., Lessard, P., 2011. The role of loading rate, backwashing, water and air velocities in an up-flow nitrifying tertiary filter. Bioresour. Technol. 102, 904–912. doi:10.1016/j.biortech.2010.09.031 Vollertsen, J., Hvitved-Jacobsen, T., 1999. Stoichiometric and kinetic model parameters for microbial transformations of suspended solids in combined sewer systems. Water Res. 33, 3127– 3141. doi:10.1016/S0043-1354(99)00033-0 Wanner, O., Gujer, W., 1984. Competition in biofilms. Water Sci. Technol. 17, 27–44. Weeks, N.C., Timmons, M.B., Chen, S., 1992. Feasibility of using foam fractionation for the removal of dissolved and suspended solids from fish culture water. Aquac. Eng. 11, 251–265. doi:10.1016/0144-8609(92)90008-L Wik, T.E.I., Lindén, B.T., Wramner, P.I., 2009. Integrated dynamic aquaculture and wastewater treatment modelling for recirculating aquaculture systems. Aquaculture 287, 361–370. doi:10.1016/j.aquaculture.2008.10.056 Witham, C.S., Oppenheimer, C., Horwell, C.J., 2005. Volcanic ash-leachates: a review and recommendations for sampling methods. J. Volcanol. Geotherm. Res. 141, 299–326. doi:10.1016/j.jvolgeores.2004.11.010 Wold, P.-A., Holan, A.B., Øie, G., Attramadal, K.J.K., Bakke, I., Vadstein, O., Leiknes, T.O., 2014. Effects of membrane filtration on bacterial number and microbial diversity in marine recirculating aquaculture system (RAS) for Atlantic cod (Gadus morhua L.) production. Aquaculture 422-423, 69–77. doi:10.1016/j.aquaculture.2013.11.019 Wong, K.B., Piedrahita, R., 2000. Settling velocity characterization of aquacultural solids. Aquac. Eng. 21, 233–246. doi:10.1016/S0144-8609(99)00033-3 Yang, L., Chou, L.S., Shieh, W.K., 2001. Biofilter treatment of aquaculture water for reuse applications. Water Res. 35, 3097–108. Zhang, T.C., Bishop, P.L., 1994. Experimental determination of the dissolved oxygen boundary layer and mass transfer resistance near the fluid-biofilm interface. Water Sci. Technol. 60 Zhang, T.C., Fu, Y.-C., Bishop, P.L., 1995. Competitions for substrate and space in biofilms. Water Environ. Res. 67, 992–1003. Zhang, Y., Chen, Y., Westerhoff, P., Crittenden, J., 2009. Impact of natural organic matter and divalent cations on the stability of aqueous nanoparticles. Water Res. 43, 4249–4257. doi:10.1016/j.watres.2009.06.005 Zhu, S., Chen, S., 1999. An experimental study on nitrification biofilm performances using a series reactor system. Aquac. Eng. 20, 245–259. doi:10.1016/S0144-8609(99)00019-9 Zhu, S., Chen, S., 2001a. Impacts of Reynolds number on nitrification biofilm kinetics. Aquac. Eng. 24, 213–229. doi:10.1016/S0144-8609(00)00071-6 Zhu, S., Chen, S., 2001b. Effects of organic carbon on nitrification rate in fixed film biofilters. Aquac. Eng. 25, 1–11. doi:10.1016/S0144-8609(01)00071-1 Zhu, S., Chen, S., 2003. Effects of air diffusion turbulent flow on nitrification rate in fixed-film biofilters: A comparison study. N. Am. J. Aquac. 65, 240–247. doi:10.1577/C02-015 Æsøy, A., Ødegaard, H., Bentzen, G., 1998. The effect of sulphide and organic matter on the nitrification activity in a biofilm process. Wat. Sci. Tech. 37, 115–122. Ødegaard, H., Rusten, B., Westrum, T., 1994. A new moving bed biofilm reactor - applications and results. Water Sci. Technol. Åhl, R.M., Leiknes, T.O., Ødegaard, H., 2006. Tracking particle size distributions in a moving bed biofilm membrane reactor for treatment of municipal wastewater. Water Sci. Technol. 53, 33. doi:10.2166/wst.2006.205 61 62 10. PAPER I Fernandes, P.M. Pedersen, L.-F., Pedersen, P.B. 2014. Daily micro particle size distribution of an experimental recirculating aquaculture system – A case study. Aquacultural Engineering 60: 2834. doi:10.1016/j.aquaeng.2014.03.007 63 64 Aquacultural Engineering 60 (2014) 28–34 Contents lists available at ScienceDirect Aquacultural Engineering journal homepage: www.elsevier.com/locate/aqua-online Daily micro particle distribution of an experimental recirculating aquaculture system—A case study Paulo Mira Fernandes ∗ , Lars-Flemming Pedersen, Per Bovbjerg Pedersen Technical University of Denmark, DTU Aqua, Section for Aquaculture, The North Sea Research Centre, P.O. Box 101, DK-9850 Hirtshals, Denmark a r t i c l e i n f o Article history: Received 26 November 2013 Accepted 30 March 2014 Keywords: Micro-particles RAS Particle size distribution Filter Daily variation Water turnover a b s t r a c t The particle size distribution (PSD) in a recirculating aquaculture system (RAS) was investigated during a 24-h cycle. PSD was analyzed in water sampled at several locations in a recirculation loop containing a 60-m drum filter, a submerged fixed-bed biofilter and a trickling filter. In relation to total counts, the system was dominated by micro-particles with particles smaller than 20 m comprising >94% of the distribution in all samples. However, the system presented a substantial volumetric influence of larger particles, reflected by a PSD derivate ˇ-value of 3.40 ± 0.18. Overall ˇ-values throughout the compartments (p = 0.584) and experimental period (p = 0.217) were not significantly different, although specific components seemed to marginally affect the PSD. A high internal water turnover rate (one system passage every 50 min) promoted the rapid removal of large particles from the system. Permanent volumetric particle removal above 60 m (31% reduction in the relative contribution from each size by the drum filter) per passage, but marginal production and removal of particles throughout the rest of the system further support the ˇ-value stability and consequent PSD equilibrium. The results showed a stable ˇ-value in the mature RAS. The ˇ-value is influenced by the contained compartments and system configuration, and may be used as a system performance-predicting tool. Mechanisms of particle influence on system and fish performance should be addressed in future studies, and are herein discussed. © 2014 Elsevier B.V. All rights reserved. 1. Introduction Particulate matter and suspended solids represent an important characteristic of the water quality in aquaculture systems (Brinker and Rösch, 2005). Removal of solids in recirculation aquaculture systems (RAS) has attained considerable attention (Summerfelt et al., 1997; Patterson and Watts, 2003; Sindilariu et al., 2009) while the abundance and distribution of micro particles has received less attention (Cripps and Bergheim, 2000). Chen et al. (1993) observed that more than 95% of particle counts in RAS were in the <20 m fraction, a portion of which may have negative implications on fish health (Clark et al., 1985; Chapman et al., 1987; Bullock et al., 1997) or RAS performance (Michaud et al., 2006). Rueter and Johnson (1995) introduced a universal tool to size solids in aquaculture, termed particle size distribution (PSD), with Patterson et al. (1999) demonstrating its application in RAS. This ∗ Corresponding author. Tel.: +45 35 88 32 65; fax: +45 35 88 32 60. E-mail addresses: [email protected] (P.M. Fernandes), [email protected] (L.-F. Pedersen), [email protected] (P.B. Pedersen). http://dx.doi.org/10.1016/j.aquaeng.2014.03.007 0144-8609/© 2014 Elsevier B.V. All rights reserved. tool is based on the assumption that solids in an aqueous system fit the power law function, since they follow a near-hyperbolic size distribution characterized by continuous exponential decreases in particle counts along the abscissa (x-axis). The slope of the double logarithmic transformation of the data is termed the ˇ-value, which is a standpoint for assessing the scattering of particles within the distribution. The ˇ-value of aquaculture systems is generally in the range of 2–5 (Patterson et al., 1999). In practice, a ˇ-value of 2 will characterize a system largely dominated by relatively large particles, whereas a system with a ˇ-value of 5 will almost solely comprise fine solids. Assuming particle sphericity, the ratio of the distribution also extracts dominance of specific size classes in terms of available surface area and weight (volume), thus making the PSD comparable to classic solids characterization parameters, such as Total Suspended Solids. For engineering and biological purposes, the solids characterization is preferably demonstrated by available surface area or particle volume. Since the ˇ-value accounts for both – deriving this information from particle counts by size class (diameter) – it is an easier representation of the particle distribution in the system, and simplifies discrimination of potential effects of accumulated P.M. Fernandes et al. / Aquacultural Engineering 60 (2014) 28–34 29 the fish tanks, while a side-stream diverted part of the water into the pump reservoir. Total water volume in the system was 15.8 m3 . Make-up water was maintained at a steady level of 187 L h−1 (4.5 m3 d−1 ), correspondent to a feed loading of 95 g feed m−3 make-up water (10.5 m3 kg feed−1 ), and representing a complete system water change approximately every 3.5 days. Water flow through the system was maintained at 19.8 m3 h−1 giving a water velocity in the pipes of 42 cm s−1 , while water velocity in the fish tanks was steady at 3.8 cm s−1 . Water temperature in the rearing tanks was 14.5 ± 0.5 ◦ C. Experimental conditions were kept constant for weeks prior to the experimental period. 2.2. Particle size distribution sampling and analysis Fig. 1. Schematic representation of the experimental RAS system used in the experiment. Water movement was according to arrow directions. RT1-7 – rearing tanks; DF – drum filter; PR – pump water reservoir; X – pump; SF – submerged filter; TF – trickling filter. Samples were collected at the locations marked with a “+”. solids. Overall, the ˇ-value conveys a simple representation of several particle factors and provides important information on several engineering aspects of the system. By observing micro-particle distribution in RAS, Patterson and Watts (2003) concluded that the PSD-derived ˇ-value is a constant of the system, but is still dependent on specific operational factors that can stimulate variation in this index. Solid removal units, such as sedimentation basins, hydrocyclones, drum filters or foam fractionators are some of the possible components to produce a direct effect in the ˇ-value (e.g. Weeks et al., 1992; Langer et al., 1996; Summerfelt et al., 1997; Davidson and Summerfelt, 2005; Timmons and Ebeling, 2010; Wold et al., 2014), mostly through screening of unwanted particles. Employment of high flushing and water turnover rates (Patterson and Watts, 2003) can promote stabilization of particle concentration, while pumps and waterfalls (Langer et al., 1996; Kelly et al., 1997; Krumins et al., 2001; McMillan et al., 2003; Sindilariu et al., 2009) and feed-related factors (Patterson and Watts, 2003; Brinker and Rösch, 2005) have also been shown to qualitatively and quantitatively affect the PSD of closed aquaculture systems. The purpose of this study was to assess the daily PSD of an experimental RAS, presumed in steady-state, during a period of constant operation conditions. Potential components affecting the ˇ-value were examined, and a special focus was given to probable daily fluctuations of PSD in the specific RAS. 2. Materials and methods 2.1. Experimental rearing system The particle size distribution of an experimental RAS, located at the National Institute of Aquatic Resources, Hirtshals, Denmark, was analyzed over a 24 h-period. Tanks stocked with rainbow trout (Oncorhynchus mykiss) at an average density of 27 kg m−3 were fed a fixed feed amount of 0.43 kg d−1 (0.5% of stocking biomass for a 12-h period (8 am–8 pm), i.e., the first 12 h of the observation period). The experimental RAS (Fig. 1) comprised a series of compartments and rearing tanks, including seven 0.44 m3 culture tanks (Pedersen and Pedersen, 2006). From the outlet of the tanks, water flowed through a 60 m drum filter, from where it entered a 1 m3 pump reservoir and was pumped into the bottom of an up-flow, fixed-bed biofilter. The water then entered the top of a trickling filter, and flowed, via a reservoir beneath the trickling filter, back to Sampling started at 8:00 am and occurred every fourth hour in the subsequent 24-h period, at five locations within the RAS (N = 35). In each sampling event, three independent grab samples were collected at the same time (within 10 min) in 100 mL plastic tubes with screwable caps at the different sampling locations (marked by “+” in Fig. 1): before the drum filter (BDF); after the drum filter (ADF); after the pump reservoir (APR); after the submerged filter (ASF); and after the trickling filter (ATF). Each sampling point was placed immediately after a component (or combination of components) that might affect the PSD: the fish and fish tanks before the drum filter (BDF), the drum filter (ADF), the pump reservoir, the pump and the make-up water inlet (APR), the submerged filter (ASF) and the trickling filter (ATF). Immediately after collection (<30 min), the samples were transported to and randomly analyzed in an optical particle counter (OPC): the AccuSizerTM 780 SIS (Particle Sizing Systems, Santa Barbara, CA, USA). Any potential out-settling of particles was not specifically accounted for, but occurred in similar fashion throughout all samples. The measuring method of the counter is based on the Single Particle Optical Sensing (SPOS) method, or the light scattering profiles of single particles passing through a narrow tube that leads to an illuminated area. The shadow size generated by each particle in stated zone creates a correspondent change in voltage measured by a sensor, which then relates the electrical pulse to particle size. The PSD of a given sample is then generated by the program using a standard calibration curve constructed with a set of uniform particles of known diameters. The AccuSizerTM 780 SIS filter was able to detect particles ranging between 2 and 1000 m in diameter. For this study, cut-off points were defined for the PSD analysis, and only the particles observed within the 5–300 m range were considered for the ˇvalue and statistical tests. The statistical representation of the particles <5 m in all data sets was not sufficiently accurate to demonstrate reproducibility (Brinker and Rösch, 2005), while no particles were detected above 300 m in any sample. Therefore, cutoff points were determined at these marks. Before any analysis took place, the particle counter collecting tube was washed by running the machine twice: once with a solution of soap and milli-Q water, and a subsequent run with only milli-Q water. Every sample was gently agitated with a glass rod just before measuring. The program read each sample three times, presenting an average of the last two readings. Between each measurement, the sampler tube was externally and internally rinsed with milli-Q water. Data obtained from the particle counter was then transferred into EXCEL spreadsheets for posterior calculations and analysis of ˇ-value, particle counts, surface area and volume of particle size distribution from each sample. The ˇ-value analysis was conducted according to the log–log fitted regression of the PSD power-law function, as fundamentally described in Patterson et al. (1999). 30 P.M. Fernandes et al. / Aquacultural Engineering 60 (2014) 28–34 Since the PSD of an aquaculture system follows the power-law distribution (Eqs. (1) and (2)) (Patterson et al., 1999), the data set may be divided into size classes. The class boundaries are adjusted to fit an ascending geometric progression, as defined by Kavanaugh et al. (1980) and Patterson et al. (1999). Each √ size class (li ) was then 3 selected according to a ratio of li+1 = 1.26·li ( 2 ≈ 1.26, representing a geometric progression where the volume of an average sized particle in a given size class will be twice as large as the volume of an average particle in the consecutively smaller size class (Rueter and Johnson, 1995; Patterson et al., 1999; Krumins et al., 2001). Furthermore, the size class boundaries are set so that the ratio between the variation of sizes in a size class (li ) and the mean diameter of a size class (li *) is constant (li /li * = 0.23). dN Ni ∗−ˇ = A · l−ˇ ↔ = A · li dl li (1) where N is the particle number density; l is the particle size parameter; and A and ˇ are empirical constants; Ni is the number of counts, counts per unit volume or percent of total counts in class I; li * is the median of the size class defined between li+1 and li ; li is the class width, represented by the nominal particle counts of size li * . To normalize the distribution, the data resulting from equation (1) should be log–log transformed (Eq. (2)). log dN = log A − ˇ · log l dl (2) which is a straight line with slope – ˇ. Once ˇ is computed for a particular system, information regarding the contribution of each size class to total particle counts, surface area and total volume of the system is readily available. 2.3. Statistical analysis Standardized PSD data were examined by regression analysis. The suitability of the power-law function to explain the data was analyzed by recording the regression coefficient (R2 ) and the correlation coefficient (R) (Brinker and Rösch, 2005; Krumins et al., 2001; Patterson and Watts, 2003; Patterson et al., 1999). One-way ANOVAs were employed to compare the daily pattern of the ˇ-value in each component, and two-way repeated measures ANOVAs were applied to compare the daily patterns and compartment effect in the ˇ-value and total counts, within the same model. Post hoc comparisons were performed on the data set and comprised comparisons between the effects of the different compartments as well as the effect of time in each compartment. Tukey HSD was utilized for group mean comparisons. Differences at p < 0.05 were considered statistically significant. Values are presented as mean ± standard deviation (SD), or mean ± standard error of the mean (SEM). Fig. 2. Mean particle counts per location during the experimental period. BDF – before the drum filter; ADF – after the drum filter; APR – after the pump reservoir; ASF – after the submerged filter; ATF – after the trickling filter. differences in fractionated surface area and total volume (Fig. 3). PSD was not significantly different at the several locations. However, both the drum filter and the trickling filter seemed to affect mean particle size by volumetrically removing large particles, while the submerged filter appeared to produce particles. Samples collected before the drum filter (BDF) presented higher volumetric percentages of the larger size classes. The drum filter, with a nominal mesh size of 60 m, reduced the volumetric distribution at the 60 m threshold by 31%. In numbers, the distribution shifted 28% on average at the same threshold (data not shown). The surface area distribution patterns were equivalent to the volumetric distributions. However, there was a larger dominance of the small-sized particles before the drum filter. In this case, the drum filter decreased the area contribution by 10% regarding particles above 60 m. After the submerged and trickling filters the distributions also presented increased contribution from larger particles. Although minor variations were observed throughout the system, these were non-significant at the 95% confidence level. 3.2. Daily variation of particle size distribution Particle size distribution followed similar daily patterns at each sampling location (one-way ANOVA, p = 0.080), as observed in the BDF samples (Fig. 4). While feeding time did not induce statistically significant effects on PSD, non-significant time-lag production of large particles apparently occurred during the day. Similar to effects of specific components on PSD, feeding time also seemed to have an effect on the distributions, but was not sufficiently strong to yield significant results. 3. Results 3.3. Overall system ˇ-value All particles measured in the 35 water samples from the system were less than 300 m in diameter. In all of the sampling events, skewed distributions were observed with the majority of particles being smaller than 20 m. However, in terms of volume, the input to the drum filter presented a significant contribution (50%) from particles larger than 60 m. After the drum filter, and at all other locations in the system, the contribution from particles above 60 m was reduced to 25% at maximum. Average particle counts per sample (mean ± SD) were 1903 ± 771 particles mL−1 (Fig. 2). 3.1. Particle size distribution throughout the system Particles <20 m were in a high number in all samples collected (>94% of total distribution). Therefore, comparisons were based on The particle size distribution of each of the 35 water samples collected from the experimental RAS was evaluated according to the power law equation (cf. example in Fig. 5). In all cases, regression coefficients (R2 ) ≥ 0.90 were observed in log-transformed data (N = 35). Overall, the ˇ-values were not dependent on sampling time or location (repeated measures ANOVA, p = 0.238). Table 1 shows the PSD regression slopes obtained in each location throughout the 24-h time frame. The compartment-specific ˇ-values did not differ significantly between locations (one-way ANOVA, p = 0.584), and remained constant during the observational period (one-way ANOVA, p = 0.217). Based on pooled log-transformed data, an overall system ˇ-value of 3.40 ± 0.18 (mean ± SD) was obtained. Cumulative area contribution per size class (%) P.M. Fernandes et al. / Aquacultural Engineering 60 (2014) 28–34 31 a 100 80 60 Before the drum filter 40 After the drum filter After the pump reservoir 20 After the submerged filter After the trickling filter 0 50 0 100 150 200 250 300 Cumulative volume contribution per size class (%) Mean diameter of particle size class (μm) b 100 80 31% 60 Before the drum filter 40 After the drum filter After the pump reservoir 20 After the submerged filter After the trickling filter 0 0 50 100 150 200 250 300 Mean diameter of particle size class (μm) Fig. 3. Cumulative contribution of the particle distribution in individual size classes to: (a) – available surface area; (b) – total volume. Fig. 4. Histogram of the relative volumetric contribution by individual size classes to the particle distribution at each sampling time, before the drum filter. Table 1 Derived ˇ-values (R2 > 0.90, N = 35) for each sampling time of the 24-h period (starting at 8 am ∼t = 0) and average values obtained at the five sampling locations. Sampling location BDF ADF APR ASF ATF Mean ± SD Time of sampling ˇ ˇ ˇ ˇ ˇ 0h +4 h +8 h +12 h +16 h +20 h +24 h 3.12 3.25 3.33 2.94 3.50 3.41 3.67 3.29 3.22 3.47 3.44 3.21 3.53 3.66 3.42 3.08 3.60 3.51 3.51 3.51 3.29 3.57 3.61 3.38 3.28 3.17 3.46 3.33 3.48 3.61 3.15 3.33 3.62 3.52 3.57 3.23 3.44 3.46 3.39 3.48 ± ± ± ± ± 0.13 0.18 0.14 0.24 0.11 32 P.M. Fernandes et al. / Aquacultural Engineering 60 (2014) 28–34 0 Log l* 1 0.5 1.5 2 5 900 a 4 700 3 600 2 1 500 y = -3,08x + 3,04 R² = 0,97 400 300 0 -1 Log (ΔN/Δl) ΔN/Δl 800 -2 200 y = 101183x-3,08 R² = 0,97 100 -3 -4 0 50 0 100 150 200 250 300 350 400 l* (μm) 0 Log l* 1 0.5 1.5 2 900 5 b 4 700 3 600 2 500 1 y = -3,66x + 3,30 R² = 0,97 400 0 300 -1 200 -2 y = 431063x-3,66 R² = 0,97 100 Log (ΔN/Δl) ΔN/Δl 800 -3 0 -4 0 50 100 150 200 l* (μm) 250 300 350 400 Fig. 5. Power law graphical representation of samples collected during the trial: (a) BDF sample at +12 h; (b) ASF sample at +8 h. Observed data with () power law modeling and () linear regression of log–log transformed data and correspondent regression coefficients (dashed line for the power law and full line for the linear logarithmic regression) are illustrated. Derived ˇ-values in the example are 3.08 for (a), and 3.66 for (b). Data shown is expressed as change in particle counts (N/l) per size class, against the middle point of each size class (l*). 4. Discussion RAS are largely dominated by smaller particles (Chen et al., 1993; Langer et al., 1996; Patterson et al., 1999; Sindilariu et al., 2009) as opposed to what may be observed in flow-through systems (Brinker and Rösch, 2005; Brinker et al., 2005). Occurrence of large solids is sporadic and may be patchily distributed in time. The same observation was found in the present study, with distributions greatly dominated by micro-particles. Consequently, only non-significant differences in PSD-derived ˇ-values were observed, both in time and location. The data obtained in this trial fitted well into the power law model (Table 1), which is in agreement to Patterson et al. (1999) stating that aquaculture particles follow a hyperbolic distribution. Furthermore, ˇ-value seems to be a stable, system-dependent parameter, being only marginally influenced by operational procedures and system configurations, as also observed by Patterson and Watts (2003). 4.1. Particle size distribution throughout the system The present study revealed equal PSD throughout the recirculation loop, even though some compartments seemingly influenced the particle size distribution. The overall PSD is in accordance with Patterson and Watts (2003) study in RAS, where a stable ˇ-value was observed, regardless of sampling location. The main difference, however, is that these authors replaced all system water on a daily basis (daily water exchange 100–190% of total volume), while in the present study, the system was operated at a hydraulic retention time of 3.5 days (29% of total system volume was exchanged daily) causing particles to accumulate and be conserved within the system. A reduction of 31% in the cumulative volumetric contribution of particles larger than 60 m was observed per water passage through the drum filter, demonstrating the specific removal of such particles by the used screen (Langer et al., 1996). Nevertheless, around 18% of the distributions after the drum filter were related to particles larger than the mesh size. Variation in average drum filter mesh size might explain this phenomenon, as mesh lesions were not found and backwashing was witnessed. The AccuSizer® assumes particle sphericity to simplify calculations and engineering determinations. However, particles are not always spherical and can also easily break down. Their non-sphericity and deformability characteristics may consequently promote the passage of particles with an average diameter larger than the rated drum filter mesh size. Thus, both the particle characteristics and the variation in actual pore size may cause the distribution after the drum filter to contain some particles larger than average mesh opening. Sindilariu et al. (2009) also observed particles larger than the nominal mesh size after the drum filter. These authors attributed their low treatment efficiency to constructional defects or installation P.M. Fernandes et al. / Aquacultural Engineering 60 (2014) 28–34 problems. Moreover, practical experience of the manufacturer (H. Mortensen, pers. comm.) relates 15–20% of the distributions after the drum filter to variations in average filter pore size. The minor PSD variation induced by the pump, the submerged biofilter and the trickling filter might be related to the varying hydraulic conditions at each location. Pumps decrease particle size (McMillan et al., 2003; Sindilariu et al., 2009), and a combined effect of the pump, the hydraulic retention time and fast water fluxes through the pump reservoir, may have contributed to a breakdown of particles, although non-significant, at the APR sample sets. It has been demonstrated, that flow rate induces PSD variation within a biofilter, mainly by out-settling of particles and biofilm sloughed off by water turbulence and resistance, which generates thinner biofilms at increasing water velocities (Sandu et al., 2002; FrancoNava et al., 2004a,b; Brinker and Rösch, 2005). On average, the ASF sample set presented 1.3 and 1.5 times more particles than the APR and ATF subsets, respectively, particularly at the larger end of the distribution (>60 m). This categorizes the submerged filter as a net particle producer and the trickling filter as a potential secondary particle removal component. Considering the operational water fluxes and the 1.2 m cascade through biofilter material, breakdown and disaggregation of large particles into smaller ones may have occurred at this location (Brinker and Rösch, 2005; Sindilariu et al., 2009). Overall, and although the tested components did not significantly affect the distribution, the findings in this study suggest that the PSD of any RAS in steady-state can be estimated any time anywhere within the recirculation loop. 4.2. Daily variation of particle size distribution Feeding regime has been shown to have no direct effect on PSD throughout a flow-through farm (Brinker and Rösch, 2005). However, feed-related factors, such as pellet integrity, dust content and physical characteristic (Patterson and Watts, 2003), or operational maintenance, incl. cleaning of tanks (Kelly et al., 1997), have been identified to affect PSD. No effects of fish circadian rhythm were detected in the present study and the daily PSD variation was not significantly different (one-way ANOVA, p = 0.377). As shown by Brinker and Rösch (2005), the ˇ-value may be directly affected by fish movement; however, in our case, fish movement had little to no effect on ˇ-value. Divergent tank configurations might be the basis for this difference, but more importantly, the high internal water turnover practiced – opposed to a flow-through system in Brinker and Rösch (2005) – may have played a more defining role on PSD. 4.3. Overall system ˇ-value The present study demonstrated that the RAS tested was dominated by micro-particles as also previously described (Chen et al., 1993; Patterson and Watts, 2003). Water turnover rates through the system promoted a total system turnover rate of 1.2 times per hour and, except for the drum filter, no component was specifically installed for the removal of fine solids. For these reasons, the drum filter effect on particle size distribution was quick, although particles smaller than 60 m were retained in the system. This in turn promoted the development of a particle distribution steady-state through the system. By tracing particulate carbon through a RAS, Franco-Nava et al. (2004a,b) suggested that the same particles may be observed passing across several RAS compartments. Furthermore, McMillan et al. (2003) suggested a steady-state stabilization of the particle distribution in RAS, which was later observed by Patterson and Watts (2003) within their own setup. A stabilization 33 of the ˇ-value through time and location thus suggests the hypothesized particle steady-state through our setup and experimental period. The feed loading applied in this trial (95 g feed m−3 ) represents a system with a low accumulation of compounds. Also including a high internal water turnover rate and relatively short hydraulic retention time, basis is provided for low accumulation of particles, and with constant distributions at several locations throughout the recirculation loop. Hence, in mature RAS, the ˇ-value seems to be a resilient property of the system and, overall, a stable parameter. Consequently, and from a technical point of view, samples for PSD analysis may be collected anywhere and anytime within the recirculation loop. It should be noted, however, that this is only true for systems operated at very stable and controlled conditions of feed, make-up water and pump speed. At increasing ˇ-values, the total amount of fine particles present in a RAS may be a potential threat to fish health and system performance. Smaller particles present a larger area to volume ratio, which may lead to secondary leaching of nutrients and dispersal of heterotrophic bacteria and pathogenic microorganisms (Chapman et al., 1987; Bullock et al., 1997; Michaud et al., 2006). Therefore, from a management and scientific point of view, it is important to quantify and qualify particle accumulation and distribution because of the latent harmful potential that fine particles may encompass. 5. Conclusion The present study demonstrated that the PSD-derived ˇ-value was constant throughout the system, and that only minor fluctuations were observed during the day. In this specific study, ˇ-value may be stated a very stable and robust parameter, holding the potential to be generalized into a solids descripting parameter of mature and constant-operations RAS. The particular RAS used for this study, having a mean ˇ-value of 3.40, was dominated by fine particles. Overall, the ˇ-value of the examined system remained stable throughout the day wherever sampling took place. By applying the power law to the data gathered, it could be observed that the system-characteristic ˇvalue was a stable parameter. It was not directly influenced by the individual compartments but seems descriptive for a given system configuration of RAS, namely systems having high internal water turnover rates. Potential consequences of systems dominated by high numbers of micro-particles are still not fully addressed, and future studies are necessary to assess the specific action and importance of micro-particles at the system level (e.g. biofilter nitrification) and on fish performance (e.g. gill-related diseases) in RAS. Furthermore, the specific effect of different versions within the same component category (e.g. moving bed vs. fixed bed biofilters; 100 m vs. 40 m drum filter mesh size; centrifugal vs. airlift pumps) on RAS particle distribution is not well understood, and should be further explored. Acknowledgements The authors appreciate Alexander Brinker, Fisheries Research Station of Baden-Württemburg (Langenargen, Germany), for his critical and constructive comments to the development of this manuscript. The care taking of the fish and of the experimental facilities by Erik Poulsen, and the analytical assistance in the laboratory by Brian Møller, both from the Technical University of Denmark, Section for Aquaculture (Hirtshals, Denmark), are highly appreciated too. 34 P.M. Fernandes et al. / Aquacultural Engineering 60 (2014) 28–34 References Brinker, A., Rösch, R., 2005. Factors determining the size of suspended solids in a flow-through fish farm. Aquac. Eng. 33, 1–19. Brinker, A., Schröder, H.G., Rösch, R., 2005. A high-resolution technique to size suspended solids in flow-through fish farms. Aquac. Eng. 32, 325–341. Bullock, G.L., Summerfelt, S.T., Noble, A.C., Weber, A.L., Durant, M.D., Hankins, J.A., 1997. Ozonation of a recirculating rainbow trout culture system. I. Effects on bacterial gill disease and heterotrophic bacteria. Aquaculture 158, 43–55. Chapman, P.M., Popham, J.D., Griffin, J., Leslie, D., Michaelson, J., 1987. Differentiation of physical from chemical toxicity in solid waste fish bioassays. Water Air Soil Pollut. 33, 295–308. Chen, S., Timmons, M.B., Aneshansley, D.J., James, J., 1993. Suspended solids characteristics from recirculating aquacultural systems and design implications. Aquaculture 112, 143–155. Clark, E.R., Harman, J.P., Forster, J.R.M., 1985. Production of metabolic and waste products by intensively farmed rainbow trout, Salmo gairdneri Richardson. J. Fish Biol. 27, 381–393. Cripps, S.J., Bergheim, A., 2000. Solids management and removal for intensive landbased aquaculture production systems. Aquac. Eng. 22, 33–56. Davidson, J., Summerfelt, S.T., 2005. Solids removal from a coldwater recirculating system—comparison of a swirl separator and a radial-flow settler. Aquac. Eng. 33, 47–61. Franco-Nava, M.A., Blancheton, J.P., Deviller, G., Le-Gall, J.-Y., 2004a. Particulate matter dynamics and transformations in a recirculating aquaculture system: application of stable isotope tracers in seabass rearing. Aquac. Eng. 31, 135– 155. Franco-Nava, M.A., Blancheton, J.P., Deviller, G., Charrier, A., Le-Gall, J.Y., 2004b. Effect of fish size and hydraulic regime on particulate organic matter dynamics in a recirculating aquaculture system: elemental carbon and nitrogen approach. Aquaculture 239, 179–198. Kavanaugh, M.C., Tate, C.H., Trussell, A.R., Trussell, R.R., Treweek, G., 1980. Use of particle size distribution measurements for selection and control of solid: liquid separation processes. Particulates in Water: Characterization, Fate Effects and Removal, vol. 189. American Chemical Society Advances in Chemistry Series, Washington, DC, USA, pp. 305–328. Kelly, L.A., Bergheim, A., Stellwagen, J., 1997. Particle size distribution of wastes from freshwater fish farms. Aquac. Int. 5, 65–78. Krumins, V., Ebeling, J.M., Wheaton, F., 2001. Ozone’s effects on power-law particle size distribution in recirculating aquaculture systems. Aquac. Eng. 25, 13–24. Langer, J., Efthimiou, S., Rosenthal, H., Bronzi, P., 1996. Drum filter performance in a recirculating eel culture unit. J. Appl. Ichthyol. 12, 61–65. McMillan, J., Wheaton, F., Hochheimer, J., Soares, J., 2003. Pumping effect on particle sizes in a recirculating aquaculture system. Aquac. Eng. 27, 53–59. Michaud, L., Blancheton, J.P., Bruni, V., Piedrahita, R., 2006. Effect of particulate organic carbon on heterotrophic bacterial populations and nitrification efficiency in biological filters. Aquac. Eng. 34, 224–233. Patterson, R.N., Watts, K.C., 2003. Micro-particles in recirculating aquaculture systems: particle size analysis of culture water from a commercial Atlantic salmon site. Aquac. Eng. 28, 99–113. Patterson, R.N., Watts, K.C., Timmons, M.B., 1999. The power law in particle size analysis for aquacultural facilities. Aquac. Eng. 19, 259–273. Pedersen, L.-F., Pedersen, P.B., 2006. Temperature-dependent formaldehyde degradation in trickling filters. N. Am. J. Aquac. 68 (3), 230–234. Rueter, J., Johnson, R., 1995. The use of ozone to improve solids removal during disinfection. Aquac. Eng. 14, 123–141. Sandu, S., Boardman, G.D., Watten, B.J., Brazil, B., 2002. Factors influencing the nitrification efficiency of fluidized bed filter with a plastic bead medium. Aquac. Eng. 26, 41–59. Sindilariu, P.-D., Brinker, A., Reiter, R., 2009. Waste and particle management in a commercial, partially recirculating trout farm. Aquac. Eng. 41, 127–135. Summerfelt, S.T., Hankins, J.A., Weber, A.L., Durant, M.D., 1997. Ozonation of a recirculating rainbow trout culture system II. Effects on microscreen filtration and water quality. Aquaculture 158, 57–67. Timmons, M.B., Ebeling, J.M., 2010. Recirculating Aquaculture, 10th ed. Cayuga Aqua Ventures, Ithaca, NY, USA, NRAC Publication No. 401-2010. Weeks, N.C., Timmons, M.B., Chen, S., 1992. Feasibility of using foam fractionation for the removal of dissolved and suspended solids from fish culture water. Aquac. Eng. 11, 251–265. Wold, P.-A., Holan, A.B., Øie, G., Attramadal, K.J.K., Bakke, I., Vadstein, O., Leiknes, T.O., 2014. Effects of membrane filtration on bacterial number and microbial diversity in marine recirculating aquaculture system (RAS) for Atlantic cod (Gadus morhua L.) production. Aquaculture 422–423, 69–77. Paulo Fernandes is a PhD student in the National Institute of Aquatic Resources, in Technical University of Denmark, Section for Aquaculture (Hirtshals, Denmark). His project topic is on the “Interactions between colloidal particles and biofilters in Recirculating Aquaculture Systems”. Lars-Flemming Pedersen is a research scientist in the National Institute of Aquatic Resources, in Technical University of Denmark, Section for Aquaculture (Hirtshals, Denmark). His research at the section includes water quality in recirculating aquaculture systems. Per Bovbjerg Pedersen is a senior research scientist and head of section in the National Institute of Aquatic Resources, in Technical University of Denmark, Section for Aquaculture (Hirtshals, Denmark). He is involved in aquaculture through research, development and counseling mostly regarding recirculating aquaculture systems and including model trout farms. 72 11. PAPER II Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B. 2015. Microscreen effects on water quality in replicated recirculating aquaculture systems. Aquacultural Engineering 65: 17-26. doi:10.1016/j.aquaeng.2014.10.007 73 74 Aquacultural Engineering 65 (2015) 17–26 Contents lists available at ScienceDirect Aquacultural Engineering journal homepage: www.elsevier.com/locate/aqua-online Microscreen effects on water quality in replicated recirculating aquaculture systems Paulo Fernandes ∗ , Lars-Flemming Pedersen, Per Bovbjerg Pedersen Technical University of Denmark, DTU Aqua, Section for Aquaculture, The North Sea Research Center, P.O. Box 101, DK-9850 Hirtshals, Denmark a r t i c l e i n f o Article history: Available online 3 November 2014 Keywords: COD Mesh size Microscreen Nitrification kinetics Particle size distribution Water quality a b s t r a c t This study investigated the effects of three microscreen mesh sizes (100, 60 and 20 m) on water quality and rainbow trout (Oncorhynchus mykiss) performance compared to a control group without microscreens, in triplicated recirculating aquaculture systems (RAS). Operational conditions were kept constant during a 6-week period where the microscreens were manually rinsed three times a day. The effects of microscreen cleaning frequency and nitrification performance were subsequently assessed. Compared to the control group, microscreens removed particles, reduced particulate organic matter, and increased ˇ-values. Particulate parameters reached steady-state in all treatment groups having microscreens at the end of the trial. The time to reach equilibrium seemingly increased with increasing mesh size but the three treatment groups (100, 60 and 20 m) did not significantly differ at the end of the trial. Increased backwashing frequency over a 24-h period had no further significant effects on the parameters measured. The results demonstrated the role and importance of a microscreen, and showed that mesh size, within the range tested, is less important at long operations under constant conditions. Fish performed similarly in all treatments. Preliminary screening of trout gills did not reveal any pathological changes related to microscreen filtration and the resulting water quality. Biofilter performance was also unaffected, with 0 -order nitrification rates (k0a ) being equivalent for all twelve systems (0.148 ± 0.022 g N m−2 d−1 ). Mechanisms for RAS equilibrium establishment, within and between systems with different mesh sizes, are discussed. © 2014 Elsevier B.V. All rights reserved. 1. Introduction Relatively few studies have investigated the effect of microscreen mesh size on particle distribution in recirculating aquaculture systems (Langer et al., 1996; Patterson and Watts, 2003; Davidson et al., 2011a; Dolan et al., 2013). Filter efficiency is commonly tested on a mass balance basis by analyzing total suspended solids (TSS) before and after the filter applied (Bergheim et al., 1993; Twarowska et al., 1997; Davidson and Summerfelt, 2005; Roque d’Orbcastel et al., 2009). However, the nature and composition of micro-solids is highly variable (Brinker and Rösch, 2005), and TSS depends on fractionation of the particle distribution, as well as patchiness due to sampling. The interaction between both factors ∗ Corresponding author. Tel.: +45 35 88 32 65; fax: +45 35 88 32 60. E-mail addresses: [email protected] (P. Fernandes), [email protected] (L.-F. Pedersen), [email protected] (P.B. Pedersen). http://dx.doi.org/10.1016/j.aquaeng.2014.10.007 0144-8609/© 2014 Elsevier B.V. All rights reserved. reduces the strength of TSS as a deterministic tool for solids control and other tools should, therefore, be used in association with TSS. An alternative measure for TSS is the particle size distribution (PSD) (Cripps, 1995; Rueter and Johnson, 1995; Brinker et al., 2005), which is based on the quantification of particles by size class in aqueous solutions (Patterson et al., 1999). Defined as spheres, particles present characteristics by which they are measured and represented (Rueter and Johnson, 1995; Langer et al., 1996; Patterson et al., 1999; Krumins et al., 2001a,b), and this information can be acquired through PSD measurements. Firstly, by plotting the distribution, the ratio of particles per size class (ˇvalue) will inform on evenness of the distribution. Since particles in aqueous solutions present a hyperbolic relation between density (concentration per size class) and size, applying a logarithmic transformation to the data allows the visualization of such evenness. Secondly, information about the distribution by surface area and volumetric weight of each particle can also be extrapolated from PSD measurements. A good relationship between the volumetric distribution and TSS can also be observed (Chen et al., 1993), which 18 P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 allows a certain degree of comparison between former (Langer et al., 1996) and more recent experimental approaches (Dolan et al., 2013). This allows particles to be represented by volume or weight, which may allow the calculations of solid waste accumulation on a mass balance basis. However, because large particles are normally heavier, weight-based measurements almost do not consider the potential impact of finer particles. Bacteria, pathogens and nutrients can adhere to the particle surface, and at long residence times, as observed in RAS, they may end up leaching nutrients into the water, increasing the system load. Fine particles are categorized by a higher surface area per particle weight (Witham et al., 2005), and are, therefore, better accounted for by surface area measurements. Lastly, accounting the total distribution surface area against the total distribution volume (A/V ratio) is speculated to define the potential toxicity of the particles that remain in suspension in aquaculture (Patterson et al., 1999; Brinker et al., 2005), in a similar fashion to the ˇ-value. In aquaculture, it is important to quickly and gently remove solid waste from the system. If their hydraulic residence time (HRT) is too high, nutrients associated with accumulated solid waste may end up leaching into the water, potentially impairing fish and system performance. Current technology efficiently screens the majority (by weight) of the particulate waste accumulated within RAS. Specifically, drum filters effectively remove solid waste by size in RAS, but the real effect on particle size is usually unknown (Chen et al., 1993; Rueter and Johnson, 1995). In terms of particle size removal, drum filter might have efficiencies close to 100%, resulting in distributions dominated (>95% of total counts) by fine particles below 20 m in diameter (Chen et al., 1993; Cripps, 1995; Langer et al., 1996; Patterson et al., 1999; Fernandes et al., 2014). However, to remove increasingly smaller particles, narrower screens need to be installed in the drum filters. As Cripps (1995) and Kelly et al. (1997) demonstrated, the percentage of waste comprised in the small-sized solid waste is not easily removed by mesh sizes down to 20 m, and removal efficiencies decline when nutrient loading is low (Cripps, 1995; Timmons and Ebeling, 2010). Therefore, decreasing the drum filter mesh size is probably not an optimal solution due to almost insignificant differences in the removal of solid waste. Moreover, decreasing the installed mesh size reduces the drum filter hydraulic capacity per m2 of mesh size, which leads to the enlargement of the drum filter, and larger backwashing water and frequency requirements (Dolan et al., 2013). Both factors lead to increased investment and operational costs, which is not ideal in smaller operations. Observed and expected final particle distributions after passage through a specific mesh are also sub-optimal and these may be attributable to variations in particle shape or to accurate mesh size (H. Mortensen (Hydrotech, Veolia Water Technologies, Sweden), pers. comm.; Fernandes et al., 2014). As particle size decreases, it is relatively more difficult to remove them from the systems (Rueter and Johnson, 1995; Summerfelt et al., 1997; Cripps and Bergheim, 2000; Krumins et al., 2001a,b; Davidson et al., 2011b; Barrut et al., 2013), and so they accumulate within RAS. Comparative studies on the effects of drum filters operated with different mesh sizes are, however, scarce. Likewise, existing studies usually do not focus on comparisons using the same relative feed loading and make-up water balances in the system. In this sense, we assessed the influence of microscreen presence and mesh size on RAS water quality. Particle size, concentration, and distribution, and the resulting water quality were measured in 12 replicated RAS in triplicate treatment groups during a 6-week period of constant operating conditions. Potential backwashing frequency improvements on water quality were also assessed over a 24-h period, as well as nitrification performance of biofilters associated with each of the treatment groups. 2. Materials and methods 2.1. Experimental setup Twelve replicated pilot-scale RAS (Fig. 1) were randomly assigned to one of the four experimental treatments dealing with solids removal (mesh size of 100, 60 or 20 m or no mesh). One of the three systems in the 60 m treatment was, however, excluded from the analysis due to unforeseen issues regarding temperature control. Total system volume was approximately 1.7 m3 , and comprised a 500 L fish rearing unit, and a 760 L submerged fixed-bed biofilter containing BIOBLOK 150® (Exponet, Denmark), with a specific surface area (SSA) of 150 m2 m−3 . Each system was also equipped with a swirl separator, a pump sump, and a trickling filter as described by Pedersen et al. (2012). The utilization of the swirl separator promoted equal settling of large solids in each system, and minimized potential limitations and problems due to solids accumulation in the control group. Before the main trial was started, all systems went through a three month maturation period without microscreens. Initially, each tank was stocked with 25 kg m−3 of 50 g rainbow trout (Oncorhynchus mykiss). Throughout the maturation period, biomass control was performed at regular intervals to keep fish density below 60 kg m−3 . Fish were fed a constant commercial feed (EFICO Enviro 3 mm, BIOMAR, Denmark) ration of 250 g feed day−1 during a 6-h period (9:30–15:30). Make-up water (MUW) was kept at 80 L d−1 . The systems were considered mature when stable nitrate concentrations were observed for consecutive days, demonstrating net removal of the daily N loading. After the maturation period, fish biomass was reduced to obtain initial densities of 25 kg m−3 of 150 g fish in each tank. Through the duration of the trial, the fish were fed a constant ration of 250 g day−1 during 6 h (9:30–15:30), and MUW was kept at 80 L d−1 . These conditions represent a relative water renewal rate of 320 L MUW kg−1 feed, and correspond to a feed loading of 3.1 kg feed m−3 MUW. As a consequence of the fixed daily feed loading and fish biomass increase, the feeding ratio decreased during the trial from 1.00 to 0.70% of fish biomass per day. Fish waste production was calculated based on the findings of Dalsgaard and Pedersen (2011), and corrected for observed apparent Feed Conversion Ratio (FCRa ) during the pre-experimental period. Specifically, expected daily Total Ammonia Nitrogen (TAN) production was calculated assuming a daily production of 38.1 mg TAN g−1 feed, whereas daily organic matter (measured as Chemical Oxygen Demand, COD) production was calculated based on a ratio of 111 mg O2 g−1 feed. The loading of the system thus corresponded to 5.5 mg L−1 and 14.9 mg L−1 for TAN and COD, respectively. 2.2. Sampling procedures during steady state The main experiment was conducted over a period of six weeks during steady-state RAS operation and included weekly sampling. Two subsequent trials, consisting of intensive backwashing and nitrification kinetics assessment, were also conducted. i. Effect of mesh size on water quality At day 0, the microscreens were installed after the swirl separator in each system. Backwashing of the microscreens was performed three times a day for the duration of the trial, and the systems were operated in these conditions for a period of six weeks. From day 0 and every 7th day for the duration of the trial, 10 L grab samples were collected at 8:00 am (before feeding) at each individual pump sump. P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 19 Fig. 1. The 1.7 m3 experimental recirculating aquaculture system (RAS, N = 12). Water recirculation through the RAS is represented by thick arrows. Water exchange (make-up water, MUW) occurred by adding 80 L non-chlorinated tap water to the pump sump, after partial (40 L) removal at the swirl separator. Surplus system water overflowed to a collecting pipe from the pump sump. X – Pump. Modified after Pedersen et al. (2009). ii. Effect of filter backwash frequency The second trial involved increasing the microscreen backwashing frequency from three times a day to once every hour during 24 h. Samples collected at the end of the 6-week period (main trial, day 42) were used as a reference for comparison with the samples collected after the 24 h intensive backwashing campaign (i.e., day 43). iii. Microscreen effect on nitrification kinetics The third trial involved the study of the nitrification performance of the submerged, fixed-bed biofilter in each system. TAN degradation was assessed by adding NH4 Cl into the biofilter and pump sump loop followed by successive sampling over time, while, bypassing the fish tank. Each system was spiked with 11.35 g NH4 Cl, resulting in an initial TAN concentration of 2.5 mg N L−1 . 15 mL samples were collected at times −0.25, 0.25, 0.5, 1.0, 1.5, 2, 3, 4, 6 and 8 h after NH4 Cl addition. Each sample was immediately forced through 0.20 m sterile filters (SARSTEDT, Germany) and kept refrigerated at 4 ◦ C until analysis. 2.3. Microscreens The different treatment groups were established by installing microscreens of varying mesh size on a fixed rack construction (Fig. 2). Each rack contained six perforated trays (570 mm × 370 mm with a depth of 70 mm, 0.33 m2 surface area) mounted at different heights, with a 2◦ inclination, leading any overflowing water into the lower trays. After passing the bottom tray the water was collected in a 175 L PE-container, with a level pump automatically returning the water into the corresponding RAS. Overall, the rack filter caused a 1 m head loss that was eased by the setup of the plastic racks, which also ensured minimal particle disturbance. The rack device was installed in all systems, including the control group. This ensured a standardized process for providing similar conditions to all treatments. The number of filters per system, and corresponding mesh size, is noted in Table 1. Fig. 2. Experimental solids filter. Disposition of the meshes were as described in Table 1. Each rack was sized at 570 mm × 370 mm with a depth of 70 mm, and presented 0.33 m2 of surface area. Water flow is marked with white arrows. A 175 Lwater collecting tank was installed at the bottom of the racks, where a return pump was operated automatically through a water level sensor. Table 1 Four experimental treatment groups (N = 3) based on micro screen filters of various mesh sizes. Mesh level Treatment groups Control I II III IV V VI 100 m 60 m 20 m 100 m 100 m 60 m 60 m 100 m 60 m 60 m 20 m 20 m 20 m 20 P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 All microscreens were manually cleaned using of a high pressure washer three times a day (08:00–14:00–18:00) and by transiently removing the microscreen from the racks. 2.6. Calculations and statistical analysis “Particulate” fractions (CODPART , BOD5 PART , and PPART ) were calculated by subtracting the “Dissolved” from the “Total” fraction. A biodegradability index of organic matter, modified after Zhu and Chen (2001), Marquet et al. (2007), and Dalsgaard and Pedersen (2011) was calculated as BOD5 COD−1 for the particulate, dissolved and total fractions. Particle size distribution data was utilized to gather data on total counts, volume, and surface area, to calculate the ˇ-value (Patterson et al., 1999; Fernandes et al., 2014), and to extract the relative contribution of specific size fractions (<20, 20–100 and >100 m). A numerical ratio of the area and volume data was calculated for each sample, and the outcome was named the area to volume ratio (A/V ratio). The statistical significance of the individual ˇ-values was assessed by applying least squares regression to individual samples. Goodness-of-fit was set at regression coefficients (R2 ) >0.90. FCRa was calculated for each RAS in accordance to the cumulative feed intake (feed administered) and the biomass increase over the experimental period. Specific growth rate (SGR, % d−1 ) was calculated according to Hopkins (1992) as SGR = (ln(Wti /Wti−1 ))/(ti − ti−1 ), where Wti and Wti−1 being fish average weight per system (g) at end (ti ) and start (ti−1 ) of the six weeks period, and ti −ti−1 the duration of the experimental period (days). 0 -order nitrification rates (k0a ) from trial iii were calculated according to Henze (2002) as r0a = k0a , where r0a is the surface TAN removal rate (g N m−2 d−1 ). The r0a was calculated by extracting the slope of the linear TAN degradation rate and dividing it by the specific surface area of the biofilter media. k0a was defined from the time of NH4 Cl addition until bulk TAN concentration reached 1.0 mg N L−1 (Eding et al., 2006; Prehn et al., 2012). Least squares linear regression models were used to test k0a in each system, regarding R2 > 0.90 as proper goodness-of-fit. Data processing was carried out in Microsoft Excel (Microsoft Office, United States of America), while statistical tests were carried out in R (The R project for Statistical Computing, Bell Laboratories, United States of America). Two-way ANOVAs were utilized to assess statistical differences between treatments. Student t-tests were carried out to analyze individual system trends over time. A probability level (˛) of 0.05 was utilized to determine the statistical significance of each test. 2.4. Operating conditions and daily monitoring Water exchange was conducted as described in Pedersen et al. (2012). A fixed volume of 80 L, or 4.7% of total system volume, was replaced every day, representing a hydraulic retention time of 21 days in the system. The main flow to the biofilter was 3 m3 h−1 , while the flow to the fish tank was 1.5 m3 h−1 . Water temperature was kept constant at 18 ± 0.4 ◦ C by using feedback regulation by automated porcelain heaters placed at the top of each biofilter. Dissolved oxygen (DO) concentration was monitored daily (Hach Lange HQd40, Germany), and was maintained above 8 mg L−1 by aeration and additional oxygen diffusion into the rearing tank. The pH was maintained between 7.2 and 7.4, and was adjusted daily by addition of sodium bicarbonate (50 g NaHCO3 kg feed−1 ) to compensate for alkalinity consumption during the nitrification processes. Alkalinity check measurements were made weekly by using test strip kits (Hach® AquaChek for Total Alkalinity, United States of America). Relative N-compounds (TAN, NO2 -N and NO3 -N) levels were monitored daily using test strip kits (Merck, United States of America) as described in Pedersen et al. (2007). Fish were inspected daily and any moribund or dead fish were removed from the systems. 2.5. Water analysis Weekly grab samples were analyzed for COD, total-N, TAN (NH4 + -N+NH3 -N), NO2 -N, NO3 -N, o-PO4 3− , total-P, alkalinity, UV transmittance, optical density, and particle size distribution (PSD). Additional samples were collected at days 0 and 42 for analysis of TSS, and biochemical oxygen demand after five days (BOD5 ). After collection, samples were stored at 4 ◦ C or frozen for subsequent analyses according to standard protocols, as described in Suhr and Pedersen (2010) and in Dalsgaard and Pedersen (2011). UV transmission and absorbance were measured at 254 nm in quartz glass cuvettes, and optical density was measured at 650 nm (Klausen and Grønborg, 2010). COD, BOD5 , UV transmittance, and optical density were analyzed in raw (total) and filtered (dissolved) samples. “Total” refers to analysis on homogeneous unfiltered samples, whereas “Dissolved” refers to samples filtered through GF/A filters (1.6 m, 47 mm Ø, WhatmanTM , GE Healthcare Europe GmbH, Denmark). The PSD was measured using an Optical Particle Counter (OPC, AccuSizerTM 780 SIS, Particle Sizing Systems, United States of America) with 128 measuring channels, according to the method described in Fernandes et al. (2014). The OPC was operated with a filter able to detect particles ranging between 0.5 and 400 m. For this study, a low-level threshold was defined at 5 m, as the statistical representation of the particles <5 m in all data sets was not sufficiently accurate to demonstrate reproducibility (Brinker and Rösch, 2005). 3. Results 3.1. Effect of mesh size on water quality 3.1.1. Organic matter content The particulate COD fraction (CODPART ) accumulated in the control systems, while being reduced or stabilizing in the systems with microscreens (p = 0.004, Table 2). A significant decrease in CODDISS (Table 2) over time was observed in all groups: 25–23 mg O2 L−1 , 37–18 mg O2 L−1 , 34–22 mg O2 L−1 , and 35–25 mg O2 L−1 , for the 20, Table 2 Pooled samples (mean ± SD, N = 3) of dissolved Chemical Oxygen Demand (CODDISS ) and particulate Chemical Oxygen Demand (CODPART ) per treatment, throughout the main experiment. Treatment CODDISS (mg O2 L−1 ) Day 0 Control 100 m 60 m 20 m 35 34b 37b 25b ± ± ± ± CODPART (mg O2 L−1 ) Day 14 13.3 3.1 8.9 3.7 37 32b 34b 31b ± ± ± ± 6.6 3.8 4.2 2.1 Day 21 26 24a 28a 23a ± ± ± ± 7.7 3.3 2.9 5.3 Day 35 24 22a 18a 24a ± ± ± ± 1.3 3.8 8.6 3.1 Day 42 25 22a 18a 23a ± ± ± ± 7.8 3.8 8.6 1.3 Day 0 19 11 18c 10b ± ± ± ± Day 14 5.2 5.6 4.2 3.0 11 9 8b ab 7 ± ± ± ± 2.5 1.8 1.4 4.6 Day 21 152 101 91,b 101,b ± ± ± ± 2.6 2.3 3.2 1.0 Day 35 202 91 61,a 41,a ± ± ± ± 4.9 4.1 0.0 2.0 Day 42 272 81 71,ab 71,ab Values in the same COD fraction (DISS or PART) from the same treatment with different superscript letters are statistically different at several dates (p < 0.05). Values of the same date in the same COD category (DISS or PART) with different superscript numbers are statistically different between treatments (p < 0.05). ± ± ± ± 12.2 6.1 1.1 1.1 P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 21 In addition, a concurrent increase of the A/V ratio (p < 0.001) occurred throughout the experiment, representing a relative decrease in larger particles. Toward the end of the trial, the measured particulate parameters reached a steady-state in all filtered systems, and no significant differences could be observed between the 100, 60 and 20 m treatments (p = 0.972). Although being equal after 42 days, the equilibrium for each treatment was reached at different stages. Particle counts were statistically insignificant throughout the experiment in the 100 m treatment (p-value). However, significant reductions occurred in the 60 (p-value), and 20 m treatments (p-value). Stable particle concentrations could be observed from days 28 and 21 onwards in the 60, and 20 m treatments, respectively. Fig. 3. Pooled samples (mean ± SD, N = 3) per treatment of BOD5 PART CODPART −1 ratio at the beginning and at the end of the experiment. a,b,c Columns in the same treatment with different superscript numbers are significantly different between dates at p < 0.05. *Statistical significance (p < 0.05) compared to the control group of the same day. 3.1.3. Chemical water quality parameters Background N-compound levels (Table 5) were similar between treatments with insignificant differences in total-N (p = 0.380, average 107 mg N L−1 , range 104–110), not related to mesh size. Nitrate levels were identical between and within treatments (p = 0.626) and did not change during the experimental period (105 ± 3 mg N L−1 ). During the experiment, the alkalinity measured before daily NaHCO3 addition decreased from 1.5 to 0.6 meq L−1 over time. Observed variations were in the same range through all systems and treatments (p = 0.963). Total-P was higher in the control group compared to the three groups with microscreens (p = 0.032) at the end of the experiment. Ortho-phosphate remained constant throughout the experiment (p = 0.108, Table 6). PPART concentrations were significantly reduced by the three mesh sizes on day 42 (p = 0.014), but no differences were observed between meshes. The water UV transmittance increased as mesh size decreased (p = 0.006), while optical density decreased with increasing mesh size in unfiltered samples (p = 0.192). No differences between treatments were observed for unfiltered UV absorbance (p = 0.068) or unfiltered optical density (p = 0.193). 60, 100 m and control treatments. However, presence of microscreen or mesh size did not significantly affect the decrease in CODDISS observed (p = 0.783). CODPART and CODDISS trends affected CODTOTAL (p = 0.032), and at day 42, average CODTOTAL was 52 mg O2 L−1 for the control group, and 30, 25 and 30 mg O2 L−1 for the 100 m, 60 m and 20 m groups, respectively. BOD5 DISS (p = 0.044) decreased in all systems except the 20 m treatment between days 0 and 42 (Table 3). On day 42, BOD5 DISS was significantly different in the 100 (p < 0.001) and 60 m (p < 0.001) treatments compared to initial conditions. BOD5 PART (p = 0.337) was unaffected by the presence of or microscreen mesh size (p = 0.604, and p = 0.735 for days 0 and 42, respectively). The ratio of particulate matter biodegradability (BOD5 PART CODPART −1 , Fig. 3) increased with decreasing filter mesh size over time (p = 0.001). Initially, BOD5 PART was close to 26 ± 11.1% of CODPART. At the end of the trial, BOD5 PART CODPART −1 ratios of 16 ± 5.0%, 39 ± 12.8%, 48 ± 9.7%, and 51 ± 6.5%, were observed in the control, 100 m, 60 m, and 20 m treatment groups, respectively. Biodegradability of dissolved organic matter was not affected by mesh size (p = 0.363). 3.1.4. Fish performance Final fish densities (70 ± 3.5 kg m−3 , p = 0.733), FCR (1.0 ± 0.09 g g−1 , p = 0.809) and SGR (0.9 ± 0.07% d−1 , p = 0.878) were even in all systems. Some mortality was observed in all systems, though not related to mesh size (p = 0.234). Fish biomass mortality in each tank occurred between 0 and 2% in the control treatment, 0 and 3% in the treatment group with 100 m filters, 0 and 4% in the 60 m treatment, and 2 and 5% in the 20 m group. 3.1.2. Particle size distribution analysis Throughout the experimental period, the three types of filter (100, 60 and 20 m treatments) caused a reduction in total particle counts compared to the control group (p < 0.001, Table 4). The reduction was more apparent in the relative contribution from particles in the range of 20–100 m (p < 0.001), thereby causing a relative increase of the <20 m fraction (p < 0.001). On a total counts basis, particles larger than 100 m were almost non-existent (<0.5% of the total distribution) in all of the 11 RAS. A significant increase in ˇ-value toward the end of the experiment was observed in the 60 (p = 0.002) and the 20 m (p = 0.004) systems, but not in the100 m systems (p = 0.753), representing a decrease in the presence of particles larger than 20 m. 3.2. Effect of filter backwashing frequency Increased backwashing frequency did not change the resulting water quality in the 11 RAS. The CODPART fractions (Fig. 4, p = 0.705) and TSS (Fig. 5, p = 0.595) remained unchanged when sampled before and after the intensive cleaning period. Particle counts (p = 0.774), ˇ-values (p = 0.725) and A/V ratios (p = 0.952) also did not change between day 42 and 43. PPART (Fig. 6, p = 0.521) Table 3 Pooled samples (mean ± SD, N = 3) of dissolved BOD5 and calculated (total-dissolved) solid BOD5 per treatment at the start and end of the experimental period. BOD5 dissolved Biochemical Oxygen Demand after 5 days; BOD5 PART – calculated (total-dissolved) particulate Biochemical Oxygen Demand after 5 days. Treatment BOD5 DISS (mg O2 L−1 ) Day 0 Control 100 m 60 m 20 m 3.0 2.0b 2.6b 1.6 ± ± ± ± BOD5 Day 42 1.86 0.21 0.08 0.26 12 1.6 1.41,a 2.12,a 1.81 ± ± ± ± PART (mg O2 L−1 ) Day 0 0.47 0.17 0.48 0.15 3.4 3.4 3.6 2.8 ± ± ± ± DISS Day 42 1.10 0.26 0.33 0.83 Values in the same BOD5 fraction (DISS or PART) from the same treatment with different superscript letters are statistically different at several dates (p < 0.05). Values of the same date in the same BOD5 category (DISS or PART) with different superscript numbers are statistically different between treatments (p < 0.05). 3.9 2.5 3.5 3.8 ± ± ± ± 0.80 1.08 1.22 0.97 – 3.0 3.4 3.3a 3.3a a ± ± ± ± Day 0 0.23 0.56 0.21 0.17 ˇ-valuea 3.6 3.5 3.8b 3.4b b ± ± ± ± 0.17 0.28 0.10 0.16 Day 21 3.6 3.4 3.9b 3.7b b ± ± ± ± 0.16 0.18 0.14 0.19 Day 35 3.5 3.5 3.9b 3.7b b ± ± ± ± 0.13 0.10 0.20 0.14 Day 42 11.4 12.0 14.6c 8.9b ± ± ± ± Day 0 3.94 2.02 0.15 2.41 14.6 9.02 8.82,b 5.51,a 3 ± ± ± ± Day 21 0.64 4.47 1.11 1.94 Counts (no. particles mL−1 ) 16.3 6.71 6.21,a 6.81,a ± ± ± ± 0.53 3.47 3.53 1.88 ± ± ± ± 16.3 8.12 4.712,a 5.21,a 2 Day 42 3 Day 35 0.93 3.89 2.32 2.72 0.15 0.19 0.19a 0.16a a ± ± ± ± Day 0 0.029 0.012 0.028 0.011 A/V ratio (m−1 ) 0.23 0.25 0.29b 0.18a b ± ± ± ± Day 21 Values in the same category (ˇ-value, Counts, or A/V ratio) from the same treatment, with different superscript letters, are significantly different on different dates at p < 0.05. Values from the same date in the same category (ˇ-value, Counts, or A/V ratio) , with different superscript numbers, are significantly different between treatments, at p < 0.05. a See section 2.6 for explanation of ˇ-value calculations. Control 100 m 60 m 20 m Treatment 0.049 0.072 0.002 0.040 0.23 0.181 0.302,b 0.312,b 1,b ± ± ± ± Day 35 0.014 0.045 0.013 0.053 0.21b 0.23 0.25b 0.28b ± ± ± ± Day 42 0.017 0.030 0.047 0.064 Table 4 Particle size distribution and characteristics of micro-particles during the main experiment (mean ± SD, N = 3). ˇ-Valuea – slope of the log-log transformation of the particle size distribution (N = 44, R2 > 0.90); counts – nominal particle count concentration per sample; A/V – numerical transformation of the particle size distribution calculated as available surface area/total volume of each sample, based on average particle diameter of each size class. 22 P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 Fig. 4. Pooled samples (mean ± SD, N = 3) per treatment of CODPART before and after the intensive backwashing period. *Statistical significance (p < 0.05) compared to the control group of the same day. Fig. 5. Pooled samples (mean ± SD, N = 3) per treatment of TSS before and after the intensive backwashing period. *Statistical significance (p < 0.05) compared to the control group of the same day. Fig. 6. Pooled samples (mean ± SD, N = 3) per treatment of particulate phosphorus (PPART ) before and after the intensive backwashing period. *Statistical significance (p < 0.05) compared to the control group of the same day. also remained unaffected after the hourly backwash of the microscreens. 3.3. Microscreen effect on nitrification kinetics Surface specific TAN removal measured as 0 order nitrification rates (k0a ) were similar (0.14–0.18 g N m−2 d−1 , p = 0.052) in all systems when measured for 8 h after spiking. In all systems, ammonia concentrations decreased from the added 2.5 mg TAN L−1 to at least 0.5 mg TAN L−1 at the end of the sampling period. Most systems removed TAN fast enough to reach 1 mg N L−1 approximately 4 h into the 8 h sampling period (p = 0.112). Corresponding volumetric TAN removal rates ranged from 20.7 to 27.5 g N m−3 d−1 (p = 0.052). P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 23 Table 5 Pooled samples (mean ± SD, N = 3) of background N-compounds concentration of the whole main experiment period. Total-N – total Nitrogen; TAN – Total Ammonia Nitrogen (NH4 + -N+NH3 -N); NO2 -N – Nitrite-N; NO3 -N – Nitrate-N; Alk – Alkalinity. Treatment Total-N (mg N L−1 ) Control 100 m 60 m 20 m 110 104 110 104 ± ± ± ± NH4 -N (mg N L−1 ) 4.8 7.3 6.8 4.9 0.13 0.13 0.14 0.13 ± ± ± ± NO2 -N (mg N L−1 ) 0.018 0.039 0.031 0.018 0.07 0.07 0.09 0.07 ± ± ± ± 0.037 0.038 0.069 0.058 NO3 -N (mg N L−1 ) 107 102 108 102 ± ± ± ± 4.8 7.2 6.7 4.9 Alk (meq L−1 ) 0.8 0.9 0.8 0.8 ± ± ± ± 0.36 0.35 0.39 0.36 Table 6 Pooled samples (mean ± SD, N = 3) of other system water quality parameters. O-PO4 3− – ortho-phosphate; solid-P – particulate phosphorus fraction. Treatment o-PO4 (mg P L−1 ) Control 100 m 60 m 20 m 4.0 3.7 3.8 3.3 Solid-P (mg P L−1 ) Unfiltered ± ± ± ± 0.45 0.51 0.26 0.59 0.5b 0.3a 0.4a 0.3a ± ± ± ± 0.13 0.10 0.09 0.06 Optical density (cm−1 )B UV absorbance (%T, 1 cm)A 40.4a 44.7b 43.3b 45.6b ± ± ± ± 4.19 2.76 3.79 3.07 Filtered 47.3 49.4 49.0 50.1 ± ± ± ± Unfiltered 4.18 3.22 3.25 2.89 0.018b 0.012a 0.014ab 0.010a ± ± ± ± Filtered 0.0070 0.0045 0.0063 0.0058 0.005 0.004 0.004 0.004 ± ± ± ± 0.0019 0.0017 0.0017 0.0018 Values in the same column, with different superscript lowercase letters, are significantly different at p < 0.05. A UV absorbance (%T, 1 cm) measured at 254 nm. B Optical density (cm−1 ) measured at 650 nm. Alkalinity averaged 1.5 meq L−1 at the start of the spiking trial, and decreased by 0.23 meq L−1 at the end of the sampling period. The pH in all systems averaged 7.60 and ranged from 7.3 to 7.8 during the trial. Both alkalinity (p = 0.776) and pH (p = 0.367) did not differ between treatment groups during the spiking trials. 4. Discussion In this study, balances of particles and organic matter in RAS operated under identical arrangements with varying filtration conditions were assessed. Filtration effort (mesh size) was thus evaluated on RAS in stable operation, and it was observed that after ca. 3 weeks solid waste parameters started to stabilize, with narrower mesh sizes establishing steady-state faster. It was observed that, during prolonged operations under stable conditions (constant loading) with live fish, it is important to operate a microscreen, as confirmed by the significant improvements in water quality promoted by all mesh sizes compared to the control group. However, it seems that operating a mesh size of 20 m does not significantly improve water quality compared to a 100 m mesh size. Kelly et al. (1997) first suggested similar removal efficiencies by narrow (20 m) and large (100 m) mesh sizes, when comparing the composition of different size fractions of aquaculture wastes. Furthermore, Timmons and Ebeling (2010) noted that removal efficiencies decreased at lower loadings, interconnecting both phenomena. As size decreases, particles become increasingly difficult to remove from the system (Chen et al., 1993; Timmons and Ebeling, 2010) mainly because their constitution influences their deformability characteristics (Balmat, 1957; Kelly et al., 1997; Dolan et al., 2013; Fernandes et al., 2014;), breaking them apart more easily. The steady-state observed in all screen treatments is in accordance to what has previously been suggested (McMillan et al., 2003; Patterson and Watts, 2003) concerning RAS solids management during long operations of stable and constant conditions. 4.1. Microscreen effect on water quality 4.1.1. Organic matter content COD production, and thus the mass balances of production and removal, was calculated according to Dalsgaard and Pedersen (2011) and corrected for the FCRa obtained. CODTOT daily production was estimated to be around 111 mg O2 g−1 feed, corresponding to 14.9 mg O2 L−1 in the system water. From this baseline, and excluding removal, CODDISS should theoretically have reached 263 mg O2 L−1 at steady-state. However, CODDISS levels were maintained around 20 mg O2 L−1 , which documents substantial CODDISS removal within each system. Calculations, based on the weekly measurements of CODDISS (Table 2), show that 10.6 mg O2 L−1 were removed/consumed daily in the control treatment. Similarly, the 100, 60, and 20 m treatment groups presented daily removal/consumption rates of 85, 83 and 82 mg O2 g−1 feed, respectively. At the system level, consumption was 11.4, 11.2 and 11.1 mg O2 L−1 , for the 100, 60 and 20 m treatments, respectively. The final CODTOTAL in the control group (ca. 52 mg O2 L−1 ) corresponded well to the ones previously observed by Pedersen et al. (2012) under similar conditions (ca. 57 mg O2 L−1 ) (Table 2). The net CODPART removal by the microscreens was around 22 mg O2 g−1 feed, or 3 mg O2 L−1 per week, corresponding to a total removal of ca. 149 mg O2 g−1 feed, or 20 mg O2 L−1 by each microscreen during the 6 weeks. CODPART remained constant in the control treatment, but the reduction and balance that occurred in the three filtered treatment groups, respectively suggest the effect of microscreen and the equilibrium of solid waste in the RAS operated with microscreens. Steady-state establishment of particulate compounds has not been previously described, although McMillan et al. (2003) and Patterson and Watts (2003) have suggested this phenomenon could happen in well-controlled and stable RAS. Net solid waste removal by the microscreens was not directly measured but solids generation conditions were equal in all systems probably creating very similar loadings. At prolonged residence times within the system, solids may eventually break down, creating particle size distributions dominated by fine particles. Initially, a 20 m screen will remove more particulate matter than a 60 m or 100 m microscreen, but during prolonged operation under constant conditions, the effect of mesh size is weakened. Cake formation, especially in older meshes, is hypothesized to improve solid waste removal by dedicated filters, although Dolan et al. (2013) showed that under specific conditions, cake formation on drum filter meshes does not significantly improve waste removal. Therefore, the similar removal by the various mesh sizes in our setup seems to be attributable to other factors. Operational characteristics, such as feed and make-up water, were similar in all systems, and this may have generated very similar solid waste composition in each system. Furthermore, aquaculture waste is mostly comprised in the fine solids range (<30 m), as observed by Kelly et al. (1997), so the potential removal by a 20 m and a 100 m screens is virtually the same. 24 P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 A higher biodegradability index (Zhu and Chen, 2001; Marquet et al., 2007; Dalsgaard and Pedersen, 2011) was observed in the filtered systems, indicating a relatively higher removal of COD than BOD5 in the systems with microscreens. The accumulation of CODPART in the control group resulted in stable BOD5 COD−1 ratios at days 0 and 42, further supporting the microscreen effect on particulate compounds in RAS. The particulate BOD5 COD−1 ratio tended to increase with decreasing mesh size. Likewise to trends of CODPART , BOD5 PART differences were masked by the feed loading, and more strongly related to the presence/absence of microscreens, rather than to specific mesh size. 4.1.2. Particle size distribution analysis At the end of the main trial, systems with microscreens presented 3.5 times less particles than the control systems, highlighting the importance of screening the water with microscreens to reduce particle and solid waste concentration. Particle concentration in the filtered groups followed CODPART trends, with stable distributions at the end of the experiment, being similar in all the systems containing microscreens. In this experiment, all ˇ-values derived were as expected for RAS (2.9–3.9), as suggested by Patterson et al. (1999) and Patterson and Watts (2003). The ˇ-values increased in all the systems throughout the experimental period, concurrent with a high turnover of removal over production rate of the larger particles. Furthermore, the increase in the relative contribution of fine particles (as documented by the increasing ˇ-value) suggests an equivalent effect in all systems, rather than just a microscreen effect in the 100, 60, and 20 m treatments. Furthermore, the increase in ˇ-value observed in the control group might have been caused by the breakdown of larger particles into smaller solids (Langer et al., 1996; McMillan et al., 2003; Sindilariu et al., 2009), typical of RAS operated for long periods (Chen et al., 1993), as aided by the cascading effect of waterfalls within every system (Brinker and Rösch, 2005). Although all ˇ-values increased to the same extent, there was an obvious microscreen effect on the A/V ratio. Smaller particles represents a larger surface area on a weight basis (McMillan et al., 2003; Brinker and Rösch, 2005), which implies that higher A/V ratios represent size distributions having increased contribution from fine particles. This is consistent with distributions dominated by micro-particles, as observed in systems with microscreens. Generally, the longer a RAS is operated the more the particle size distribution will be dominated by fine particles (Chen et al., 1993; McMillan et al., 2003). Specific aquaculture components affect the PSD (Brinker and Rösch, 2005), and units such as the biofilter and the trickling filter may further affect the particle distribution (McMillan et al., 2003; Patterson and Watts, 2003; Franco-Nava et al., 2004; Brinker and Rösch, 2005; Pfeiffer et al., 2008; Sindilariu et al., 2009). We hypothesize, that the similarity in particulate characteristics of all microscreen filtered systems were caused by the specific, constant conditions in our setup. Constant input (feed loading) and constant operating conditions and flows progressively resulted in the establishment of a balance between particle production and particle removal during the experiment. At the end of the trial, the effect on water quality of having microscreens was apparent, whereas effect of different mesh size was not. 4.1.3. Chemical water quality parameters Stable NO3 -N concentrations of 106 mg N L−1 corresponded well to a daily feeding rate of 250 g, an average FCRa of 1.0 ± 0.09 g g−1 , and 4.7% daily water renewal (Colt et al., 2006; Dalsgaard and Pedersen, 2011; Pedersen et al., 2012). TAN and NO2 -N were at the same levels throughout all operated systems suggesting similar and adequate nitrification rates in all systems (Eding et al., 2006). Interestingly, a CODDISS concentration above 10 mg O2 L−1 did not seem to affect nitrification. Zhu and Chen (2001) suggested that at C/N ratios >0.5, nitrification may be hampered. Assuming a carbon in COD ratio of 1/2.68 (Zhu and Chen, 2001), the observed C/N ratio of approximately 0.7 in this experiment did not significantly affect nitrification. Phosphorus adheres to particles (Cripps, 1995; Kelly et al., 1997), and therefore accumulates in RAS when particulate filtration is not properly ensured. In this study, phosphorus accumulation followed particle size distribution trends. The differences observed in PPART levels between the Control group and the filtered treatments were caused by the presence of a microscreen in the 100, 60 and 20 m groups. Water color and optical density were improved by the presence of microscreens, as previously observed by Klausen and Grønborg (2010), and Davidson et al. (2011a). 4.1.4. Fish performance Fish performance was not significantly affected by filtering or different mesh size. Likewise, the resulting water quality did not influence performance or explain observed variation. Screening of representative trout gill tissue from the 20 m mesh treatment and Control treatments (data not shown) did not indicate significant lamellae pathologies between operating a RAS with or without a microscreen. It has previously been shown that fish reared in intensive RAS may perform sub-optimally compared to fish in less intensive systems with lower feed loadings (Davidson et al., 2009, 2014; Martins et al., 2009). As suggested by several authors (Clark et al., 1985; Chapman et al., 1987; Bullock et al., 1997), fine particles may adhere to fish gill structure, consequently leading to hampered fish growth due to decreased capacity to osmorregulate and breathe. This could not be confirmed in the present study, perhaps due to non-stressing, stable conditions and permanent optimal oxygen concentrations in the rearing environment. Nevertheless, irritation and accumulation of fine particles in fish gills could still be a problem to consider in commercial systems, where feed loading, water quality, system design, and particle control may be less optimal and vary on a daily basis. 4.2. Effect of filter backwashing frequency Results from the intensive backwashing trial confirms the hypothesis of having achieved a steady-state. Standard/routine backwashing method (three times a day) compared to the intensive method (once per hour during a 24-h period) did not affect particle counts, CODPART , or TSS in any of the treatment groups. It seems that all systems had reached equilibrium by the time the main experiment was over, indicating that backwashing the microscreens three times a day in this set-up might have been enough to cope with diurnal particle production. 4.3. Microscreen effect on nitrification kinetics Effect of mesh size and presence of microscreen on the specific TAN removal rate (k0a ) in the biofilters were not statistically significant (all-systems average of 0.148 ± 0.022 g m−2 d−1 ), and the minor variations observed were independent of mesh size applied. Some inter-system variability was observed, with k0a ranging from 0.125 to 0.207 g N m−2 d−1 at a 9 m h−1 biofilter elevation speed. These results are similar to the previous findings by Pedersen et al. (2012), and Prehn et al. (2012) using the same type of media (k0a = 0.148–0.174 g m−2 d−1 ), and comparable elevation velocities (2.5–40 m h−1 ). The loading of biofilters with varying levels of COD N−1 ratios may affect nitrification kinetics (Bovendeur et al., 1990; Ling and Chen, 2005; Zhu and Chen, 2001; Michaud et al., 2014). Specifically, the impact of particulate COD N−1 ratios is not fully understood, P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 albeit Michaud et al. (2006) observed a shift in bacterial communities in the biofilter at C/N ratios of 0.5. At higher C/N ratios, heterotrophic bacteria may out-compete nitrifiers for space and nutrients, which results in a loss of nitrification efficiency (Leonard et al., 2000; Zhu and Chen, 2001; Ling and Chen, 2005; Chen et al., 2006; Michaud et al., 2006, 2014). Although the systems were loaded with potentially limiting C/N ratios, hydrolysis and consumption of organic matter did not hamper nitrification in our systems. According to Ling and Chen (2005), the COD N−1 ratios operated might reduce nitrification by 60–70% of total submerged biofilter capacity. The retroactive effect of COD on nitrification rates could not be detected in this study. A possible explanation may be related to the relative loading of the biofilter unit, as well as sufficient oxygen concentrations (O2 /TAN ratio of ∼2.5) at the start of the spiking trials (Nogueira et al., 1998; Suhr and Pedersen, 2010). 5. Conclusions This study proved balances of particle size distribution, COD, and BOD5 in stable replicated RAS with and without microscreens. By using microscreens, the resulting particulate waste reached an equilibrium condition; fine mesh (20 m) treatment groups reaching steady state faster than systems with 60 or 100 m. The documented steady-state conditions were a result of the stable operating conditions and a constant input (fixed feed loading). Installation of microscreens did not affect the removal of dissolved substances. Although particle counts were reduced 3.5 times, neither nitrification kinetics nor fish performance was improved by the utilization of microscreens between 20 and 100 m. Acknowledgements The authors would like to thank the help and guidance of Peter Vilhelm Skov, the National Institute of Aquatic Resources, Section for Aquaculture (Hirtshals, Denmark) in the development of a protocol to assess fish gill histopathology. We also appreciate the assistance of Ole Madvig Larsen and Rasmus Frydenlund Jensen, and their helpfulness to the care-taking of the fish and experimental facilities. We would also like to thank Dorthe Frandsen, Ulla Sprogel and Brian Møller for their analytical assistance in the laboratory. References Balmat, J.L., 1957. Biochemical oxidation of various particulate fractions of sewage. Sew. Ind. Waters 29, 757–761. Barrut, B., Blancheton, J.P., Callier, M., Champagne, J.-Y., Grasmick, A., 2013. Foam fractionation efficiency of a vacuum airlift – application to particulate matter removal in recirculating systems. Aquac. Eng. 54, 16–21, http://dx.doi.org/10.1016/j.aquaeng.2012.10.003. Bergheim, A., Sanni, S., Indrevik, G., Hølland, P., 1993. Sludge removal from salmonid tank effluent using rotating microsieves. Aquac. Eng. 12, 97–109, http://dx.doi.org/10.1016/0144-8609(93)90019-8. Bovendeur, J., Zwaga, A.B., Lobee, B.G.J., Blom, J.H., 1990. Fixed-biofilm reactors in aquacultural wate recycle systems: effect of organic matter elimination on nitrification kinetics. Water Res. 24, 207–213. Brinker, A., Rösch, R., 2005. Factors determining the size of suspended solids in a flow-through fish farm. Aquac. Eng. 33, 1–19, http://dx.doi.org/10.1016/ j.aquaeng.2004.10.003. Brinker, A., Schröder, H.G., Rösch, R., 2005. A high-resolution technique to size suspended solids in flow-through fish farms. Aquac. Eng. 32, 325–341, http://dx.doi.org/10.1016/j.aquaeng.2004.07.002. Bullock, G.L., Summerfelt, S.T., Noble, A.C., Weber, A.L., Durant, M.D., Hankins, J.A., 1997. Ozonation of a recirculating rainbow trout culture system. I. Effects on bacterial gill disease and heterotrophic bacteria. Aquaculture 158, 43–55. Chapman, P.M., Popham, J.D., Griffin, J., Leslie, D., Michaelson, J., 1987. Differentiation of physical from chemical toxicity in solid waste fish bioassays. Water Air Soil Pollut. 33, 295–308. Chen, S., Ling, J., Blancheton, J.P., 2006. Nitrification kinetics of biofilm as affected by water quality factors. Aquac. Eng. 34, 179–197, http://dx.doi.org/ 10.1016/j.aquaeng.2005.09.004. Chen, S., Timmons, M.B., Aneshansley, D.J., Bisogni Jr., J.J., 1993. Suspended solids characteristics from recirculating aquacultural systems and design implications. Aquaculture 112, 143–155. 25 Clark, E.R., Harman, J.P., Forster, J.R.M., 1985. Production of metabolic and waste products by intensively farmed rainbow trout, Salmo gairdneri Richardson. J. Fish Biol. 27, 381–393, http://dx.doi.org/10.1111/j.1095-8649.1985. tb03187.x. Colt, J., Lamoureux, J., Patterson, R.N., Rogers, G., 2006. Reporting standards for biofilter performance studies. Aquac. Eng. 34, 377–388, http://dx.doi.org/10.1016/ j.aquaeng.2005.09.002. Cripps, S.J., 1995. Serial particle size fractionation and characterisation of an aquacultural effluent. Aquaculture 133, 323–339. Cripps, S.J., Bergheim, A., 2000. Solids management and removal for intensive land-based aquaculture production systems. Aquac. Eng. 22, 33–56, http://dx.doi.org/10.1016/S0144-8609(00)00031-5. Dalsgaard, J., Pedersen, P.B., 2011. Solid and suspended/dissolved waste (N, P, O) from rainbow trout (Oncorynchus mykiss). Aquaculture 313, 92–99, http://dx.doi.org/10.1016/j.aquaculture.2011.01.037. Davidson, J., Good, C., Welsh, C., Brazil, B., Summerfelt, S.T., 2009. Heavy metal and waste metabolite accumulation and their potential effect on rainbow trout performance in a replicated water reuse system operated at low or high system flushing rates. Aquac. Eng. 41, 136–145, http://dx.doi.org/10.1016/ j.aquaeng.2009.04.001. Davidson, J., Good, C., Welsh, C., Summerfelt, S.T., 2011a. The effects of ozone and water exchange rates on water quality and rainbow trout Oncorhynchus mykiss performance in replicated water recirculating systems. Aquac. Eng. 44, 80–96, http://dx.doi.org/10.1016/j.aquaeng.2011.04.001. Davidson, J., Good, C., Welsh, C., Summerfelt, S.T., 2011b. Abnormal swimming behavior and increased deformities in rainbow trout Oncorhynchus mykiss cultured in low exchange water recirculating aquaculture systems. Aquac. Eng. 45, 109–117, http://dx.doi.org/10.1016/j.aquaeng.2011.08.005. 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 Onchorhynchus mykiss within water recirculating aquaculture systems. Aquac. Eng. 59, 30–40, http://dx.doi.org/10.1016/j.aquaeng.2014.01.003. Davidson, J., Summerfelt, S.T., 2005. Solids removal from a coldwater recirculating system – comparison of a swirl separator and a radial-flow settler. Aquac. Eng. 33, 47–61, http://dx.doi.org/10.1016/j.aquaeng.2004.11.002. Dolan, E., Murphy, N., O’Hehir, M., 2013. Factors influencing optimal micro-screen drum filter selection for recirculating aquaculture systems. Aquac. Eng. 56, 42–50, http://dx.doi.org/10.1016/j.aquaeng.2013.04.005. Eding, E.H., Kamstra, A., Verreth, J.A.J., Huisman, E.A., Klapwijk, A., 2006. Design and operation of nitrifying trickling filters in recirculating aquaculture: a review. Aquac. Eng. 34, 234–260. Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B., 2014. Daily micro particle distribution of an experimental recirculating aquaculture system – a case study. Aquac. Eng. 60, 28–34, http://dx.doi.org/10.1016/j.aquaeng.2014.03.007. Franco-Nava, M.-A., Blancheton, J.P., Deviller, G., Le-Gall, J.-Y., 2004. Particulate matter dynamics and transformations in a recirculating aquaculture system: application of stable isotope tracers in seabass rearing. Aquac. Eng. 31, 135–155, http://dx.doi.org/10.1016/j.aquaeng.2004.01.003. Henze, M., 2002. Basic biological processes. Chapter 3. In: Henze, M., Harremoës, P., Jansen la Cour, J., Arvin, E. (Eds.), Wastewater Treatment: Biological and Chemical Processes. , third ed. Springer Verlag, Berlin, Heidelberg. Hopkins, K.D., 1992. Reporting fish growth: a review of the basics. J. World Aquac. Soc. 23, 173–179. Kelly, L.A., Bergheim, A., Stellwagen, J., 1997. Particle size distribution of wastes from freshwater fish farms. Aquac. Int. 5, 65–78. Klausen, M.M., Grønborg, O., 2010. Pilot scale testing of advanced oxidation processes for degradation of geosmin and MIB in recirculated aquaculture. Water Sci. Technol. Water Supply 10, 217, http://dx.doi.org/10.2166/ws.2010.246. Krumins, V., Ebeling, J.M., Wheaton, F.W., 2001a. Ozone’s effects on power-law particle size distribution in recirculating aquaculture systems. Aquac. Eng. 25, 13–24, http://dx.doi.org/10.1016/S0144-8609(01)00064-4. Krumins, V., Ebeling, J.M., Wheaton, F.W., 2001b. Part-day ozonation for nitrogen and organic carbon control in recirculating aquaculture systems. Aquac. Eng. 24, 231–241, http://dx.doi.org/10.1016/S0144-8609(01)00061-9. Langer, J., Efthimiou, S., Rosenthal, H., Bronzi, P., 1996. Drum filter performance in a recirculating eel culture unit. J. Appl. Ichthyol. 12, 61–65. Leonard, N., Blancheton, J.P., Guiraud, J.P., 2000. Populations of heterotrophic bacteria in an experimental recirculating aquaculture system. Aquac. Eng. 22, 109–120, http://dx.doi.org/10.1016/S0144-8609(00)00035-2. Ling, J., Chen, S., 2005. Impact of organic carbon on nitrification performance of different biofilters. Aquac. Eng. 33, 150–162, http://dx.doi.org/10.1016/ j.aquaeng.2004.12.002. Marquet, R., Muhammad, N., Vairavamoorthy, K., Wheatley, A., 2007. Particle size distribution to assess the performance of trickling filters. Process Saf. Environ. Prot. 85, 99–103, http://dx.doi.org/10.1205/psep.05194. Martins, C.I.M., Ochola, D., Ende, S.S.W., Eding, E.H., Verreth, J.A.J., 2009. Is growth retardation present in Nile tilapia Oreochromis niloticus cultured in low water exchange recirculating aquaculture systems? Aquaculture 298, 43–50, http://dx.doi.org/10.1016/j.aquaculture.2009.09.030. McMillan, J.D., Wheaton, F.W., Hochheimer, J.N., Soares, J., 2003. Pumping effect on particle sizes in a recirculating aquaculture system. Aquac. Eng. 27, 53–59, http://dx.doi.org/10.1016/S0144-8609(02)00038-9. Michaud, L., Blancheton, J.P., Bruni, V., Piedrahita, R.H., 2006. Effect of particulate organic carbon on heterotrophic bacterial populations and nitrification efficiency in biological filters. Aquac. Eng. 34, 224–233, http://dx.doi.org/10.1016/ j.aquaeng.2005.07.005. 26 P. Fernandes et al. / Aquacultural Engineering 65 (2015) 17–26 Michaud, L., Lo Giudice, A., Interdonato, F., Triplet, S., Ying, L., Blancheton, J.P., 2014. C/N ratio-induced structural shift of bacterial communities inside lab-scale aquaculture biofilters. Aquac. Eng. 58, 77–87, http://dx.doi.org/ 10.1016/j.aquaeng.2013.11.002. Nogueira, R., Lazarova, V., Manem, J., Melo, L.F., 1998. Influence of dissolved oxygen on the nitrification kinetics in a circulating bed biofilm reactor. Bioprocess Eng. 19, 441–449. Patterson, R.N., Watts, K.C., 2003. Micro-particles in recirculating aquaculture systems: particle size analysis of culture water from a commercial Atlantic salmon site. Aquac. Eng. 28, 99–113, http://dx.doi.org/10.1016/ S0144-8609(03)00003-7. Patterson, R.N., Watts, K.C., Timmons, M.B., 1999. The power law in particle size analysis for aquacultural facilities. Aquac. Eng. 19, 259–273. Pedersen, L.-F., Pedersen, P.B., Nielsen, J.L., Nielsen, P.H., 2009. Peracetic acid degradation and effects on nitrification in recirculating aquaculture systems. Aquaculture 296, 246–254, http://dx.doi.org/10.1016/j.aquaculture.2009.08.021. Pedersen, L.-F., Pedersen, P.B., Sortkjær, O., 2007. Temperature-dependent and surface specific formaldehyde degradation in submerged biofilters. Aquac. Eng. 36, 127–136, http://dx.doi.org/10.1016/j.aquaeng.2006.09.004. Pedersen, L.-F., Suhr, K.I., Dalsgaard, J., Pedersen, P.B., Arvin, E., 2012. Effects of feed loading on nitrogen balances and fish performance in replicated recirculating aquaculture systems. Aquaculture 338 (341), 237–245, http://dx.doi.org/10.1016/j.aquaculture.2012.01.035. Pfeiffer, T.J., Osborn, A., Davis, M., 2008. Particle sieve analysis for determining solids removal efficiency of water treatment components in a recirculating aquaculture system. Aquac. Eng. 39, 24–29, http://dx.doi.org/10.1016/j.aquaeng.2008.05.003. Prehn, J., Waul, C.K., Pedersen, L.-F., Arvin, E., 2012. Impact of water boundary layer diffusion on the nitrification rate of submerged biofilter elements from a recirculating aquaculture system. Water Res. 46, 3516–3524, http://dx.doi.org/10.1016/j.watres.2012.03.053. Roque d’Orbcastel, E., Blancheton, J.P., Belaud, A., 2009. Water quality and rainbow trout performance in a Danish Model Farm recirculating system: comparison with a flow through system. Aquac. Eng. 40, 135–143, http://dx.doi.org/10.1016/j.aquaeng.2009.02.002. Rueter, J., Johnson, R., 1995. The use of ozone to improve solids removal during disinfection. Aquac. Eng. 14, 123–141. Sindilariu, P.-D., Brinker, A., Reiter, R., 2009. Waste and particle management in a commercial, partially recirculating trout farm. Aquac. Eng. 41, 127–135, http://dx.doi.org/10.1016/j.aquaeng.2009.03.001. Suhr, K.I., Pedersen, P.B., 2010. Nitrification in moving bed and fixed bed biofilters treating effluent water from a large commercial outdoor rainbow trout RAS. Aquac. Eng. 42, 31–37, http://dx.doi.org/10.1016/j.aquaeng.2009.10.001. Summerfelt, S.T., Hankins, J.A., Weber, A.L., Durant, M.D., 1997. Ozonation of a recirculating rainbow trout culture system: II. Effects on microscreen filtration and water quality. Aquaculture 158, 57–67. Timmons, M.B., Ebeling, J.M., 2010. Recirculating Aquaculture, NRAC Publi. USDA Cayuga Aqua Ventures. Twarowska, J.G., Westerman, P.W., Losordo, T.M., 1997. Water treatment and waste characterization evaluation of an intensive recirculating fish production system. Aquac. Eng. 16, 133–147. Witham, C.S., Oppenheimer, C., Horwell, C.J., 2005. Volcanic ash-leachates: a review and recommendations for sampling methods. J. Volcanol. Geotherm. Res. 141, 299–326, http://dx.doi.org/10.1016/j.jvolgeores.2004.11.010. Zhu, S., Chen, S., 2001. Effects of organic carbon on nitrification rate in fixed film biofilters. Aquac. Eng. 25, 1–11, http://dx.doi.org/10.1016/ S0144-8609(01)00071-1. Paulo Mira Fernandes is a PhD student in the National Institute of Aquatic Resources, in Technical University of Denmark, section for Aquaculture (Hirtshals, Denmark). His project topic is on the “Interactions between colloidal particles and biofilters in Recirculating Aquaculture Systems”. Lars-Flemming Pedersen is a research scientist in the National Institute of Aquatic Resources, in Technical University of Denmark, Section for Aquaculture (Hirtshals, Denmark). His research at the section includes water quality in recirculating aquaculture systems. Per Bovbjerg Pedersen is a senior research scientist and head of section in the National Institute of Aquatic Resources, in Technical University of Denmark, section for Aquaculture (Hirtshals, Denmark). He is involved in aquaculture through research, development and counseling mostly regarding recirculating aquaculture systems and including model trout farms. 12. PAPER III Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B. Influence of fixed and moving bed biofilters on micro particle dynamics in a recirculating aquaculture system (in prep.) 85 86 2 Influence of fixed and moving bed biofilters on micro particle dynamics in a recirculating aquaculture system 3 4 5 Paulo Mira Fernandes ǂ ([email protected]) Lars-Flemming Pedersen ([email protected]) Per Bovbjerg Pedersen ([email protected]) 6 7 8 ǂ 1 – Corresponding author – Tel.: +45 35 88 32 65. Fax: +45 35 88 32 60. E-mail: [email protected] Technical University of Denmark, DTU Aqua, Section for Aquaculture, The North Sea Research Center, P.O. Box 101, DK-9850 Hirtshals, Denmark 87 9 Abstract 10 11 12 13 Accumulation of fine particulate organic matter in recirculating aquaculture systems (RAS) is a balance between system input (from feed to waste), internal transformation, removal and dilution. The mechanisms leading to fine particle accumulation in RAS are not fully understood, and neither is the potential influence of biofilters in this respect. 14 15 16 17 18 19 This study describes the effect of fixed bed biofilters (FBB) and moving bed biofilters (MBB), on particle size distribution and organic matter. It was assessed in an 8.7 m3 RAS with four equal biofilters – two FFB and two MBB. The RAS was stocked with rainbow trout (Oncorhynchus mykiss), and operated under constant feed loading conditions (1 kg feed/m3 of make-up water) for more than three months. Production or removal of micro particles according to biofilter mode of operation (FBB vs. MBB) was assessed by operating all biofilters simultaneously as well as separately. 20 21 22 23 24 25 26 27 28 In periods where each mode of operation was assessed separately, FBB reduced the particle concentration by approximately 200 particles/mL; MBB increased particle concentration by an average of 250 particles/mL. In FBB, a 10 % reduction in particle concentration also represented a 10 % reduction in total particle surface area and particle volume. In MBB, a 10 % increase in particle concentration also represented a 10 % increase in total particle surface area, but had no effect on total particle volume. A volumetric reduction of particles > 100 µm, and an equivalent volumetric increase of particles < 40 µm, showed that MBB produced fine particles by disintegration of larger particles. A constant volumetric removal of particulate matter through all size classes by FBB demonstrates their function as secondary particle removal units. 29 30 31 32 33 34 Net removal of organic matter occurred at the same rates in both modes of operation. FBB removed a higher amount of dissolved BOD5 than MBB, while MBB removed a higher amount of particulate BOD5 than FBB. All filters performed with stable nitrification rates when operated together or separately, as observed by low ammonia and nitrite concentrations, and a stable, elevated nitrate concentration throughout the experiment. Generally, removal of ammonia and nitrite was larger in FBB than in MBB. 35 36 37 38 The trends observed when only FBB or MBB were in operation were also observed when all filters were operated simultaneously. Differences in biofilm formation, development, and maintenance, coupled to reactor flow characteristics are discussed in relation to the fate of micro particles and organic matter when operating fixed or moving bed biofilters. 39 40 Keywords: biofilter performance, fixed bed, moving bed, organic matter, particle size distribution, RAS 88 41 1. Introduction 42 43 44 45 46 47 48 49 50 51 52 53 54 Quick and gentle removal of solid waste is of paramount importance in recirculating aquaculture systems (RAS). Current technologies efficiently screen the majority of the particulate waste generated within such systems. However, the removal efficiency decreases significantly with decreasing particle size (Cripps, 1995; Kelly et al., 1997; Timmons and Ebeling, 2010). This is reflected by particle size distributions dominated (> 95% of total counts) by particles smaller than 20 µm in diameter (Chen et al., 1993; Patterson et al., 1999). Furthermore, particle dynamics in RAS are not only associated with solid removal devices. For instance, fixed bed biofilters have been found to either remove (Bouwer, 1987; Marquet et al., 1999, 2007), or produce particles (Franco-Nava et al., 2004a,b), while moving bed reactors (Åhl et al., 2006; Ivanovic and Leiknes, 2008, 2012), pumps (McMillan et al., 2003; Sindilariu et al., 2009), and waterfalls (Brinker and Rösch, 2005) potentially increase the number of particles. To better understand particle presence, dynamic, accumulation and potential effects on RAS and fish performance, it is important to identify all locations within RAS where particles can be affected. 55 56 57 58 59 60 61 62 Fine particle accumulation may affect RAS operation and fish performance, as these particles tend to have a larger surface to volume ratio than larger particles (Witham et al., 2005). In practice this means that there is a relatively larger area to volume ratio per single particle for the attachment of microbial organisms (Liltved and Cripps, 1999; Wold et al., 2014), or transport of substances (Balmat, 1957; Levine et al., 1985, 1991). Fine particles might impair fish health by clogging fish gills (Chapman et al., 1987; Bullock et al., 1997), and might affect biofilter performance by stimulating heterotrophic bacteria growth and consequent competition with nitrifiers for space and oxygen (e.g. Särner, 1986; Larsen and Harremoës, 1994). 63 64 65 66 67 68 69 70 71 72 73 74 75 In aquaculture, biofilters are primarily used for the two-step oxidation of ammonia and nitrite into nitrate (Sharma and Ahlert, 1977; Hagopian and Riley, 1998; Schreier et al., 2010). Nitrifying organisms naturally colonize specific media in the bioreactors, designed to maintain a stable and large bacterial community that can effectively remove ammonia from the rearing water. Recent studies have characterized fixed bed (FBB) and moving bed (MBB) biofilters, however, mainly in terms of N-removal capacity (Suhr and Pedersen, 2010; Pedersen et al., 2015). The main differences between fixed bed and moving bed biofilters are the induced motion of the media – mechanically or by aeration – and the self-cleaning ability of MBB (Hem et al., 1994; Ødegaard et al., 1994; Rusten et al., 2006). Shear forces act on the outer layers of the biofilm growing in the carrier material, keeping it at a steady level by controlling excessive biofilm formation, and consequently risking an increased system particle load (Åhl et al., 2006; Ivanovic and Leiknes, 2008, 2012). On the other hand, fixed bed biofilters need to be regularly backwashed to remove accumulated matter or excess biofilm (Singh et al., 1999; Malone and Pfeiffer, 2006; Emparanza, 2009; Pfeiffer and Wills, 2011). 76 77 78 79 80 81 82 83 84 The bacterial development and colonization in fixed and moving bed has been shown to differ (Michaud et al., 2006, 2014). The constant movement of the media in MBB presumably promotes the establishment of a homogeneous microbial community in the carriers, and of a uniform substrate concentration from inlet to outlet of the reactor (Ødegaard et al., 1994; Rusten et al., 2006). In FBB a substrate gradient (C and N) concentration from inlet to outlet potentially supports the establishment of biofilms with different bacterial communities along that gradient. Consequently, heterogeneous biofilm structures and bacterial micro niches are formed, which can routinely affect the hydrodynamics and water dispersion in the reactor (Zhang et al., 1995; Rasmussen and Lewandowski, 1998; Fdz-Polanco et al., 2000; Picioreanu et al., 2000), and promote the competi- 89 85 86 tion between nitrifiers and heterotrophic bacteria (e.g. Wanner and Gujer, 1984; Zhang et al., 1995; Okabe et al., 1996; Leonard et al., 2000; Nogueira et al., 2002). 87 88 89 90 91 92 93 Few studies have compared particle dynamics (formation and removal) in different types of biofilters. This is further complicated since the existing studies have been conducted under varying experimental conditions, using carrier media with different reactor filling rates, shapes or specific surface area (SSA, m2/m3). To assess how specific changes to system configuration affect general water quality parameters, it is important to operate the system in a stable, controlled manner (Colt et al., 2006). This way, all the observations can be attributed to the configurational changes, and not to external factors. 94 95 96 97 98 99 100 In this study we assessed the specific effect of moving and fixed bed biofilters on organic matter and particle dynamics in an experimental RAS. The experiment was carried out under constant operating conditions of feed loading, reflecting realistic water quality in commercial RAS. The media for FBB and MBB was similar in shape and SSA (750 m2/m3), differing only in the relative density of the material used. The study was a single factor experimental design in terms of system configuration: effects of fixed vs. moving biofilter media on particle size distribution and organic matter dynamics. 90 101 2. Materials and Methods 102 103 104 105 106 107 108 109 110 111 112 The effect of biofilter mode of operation on water quality parameters was assessed with two fixed bed and two moving bed biofilters in a single experimental RAS. All biofilters were connected in parallel to the same system, where there was the option to bypass individual biofilters for experimental purposes. The RAS was operated to meet a number of criteria in order to test and compare biofilter performance (Colt et al., 2006). Attention was given to reaching system maturation and steady state conditions achieved during controlled operating conditions in terms of fish density, feed input, and make-up water. Throughout the study, these criteria generated a constant feed loading (1 kg feed/m3 make-up water) and water quality resembling commercial RAS conditions. Constant feed loading was ensured throughout the experiment by adjusting feed and make-up water according to the amount of biofilters in operation: when half of the total active biofilter surface area was bypassed, feeding and MUW were reduced proportionally. 113 2.1 Experimental setup 114 115 116 The experiment was conducted in an 8.5 m3 experimental RAS (Fig. 1). The system included a 5.5 m3 fish holding tank, four individual biofilter loops, a trickling filter, and a 40 µm drum filter (Hydrotech, Veolia Water Technologies, Sweden). 117 118 119 120 121 122 123 124 The four biofilter loops were connected in parallel to a common pump sump. Each of the 0.40 m3 up-flow biofilters was filled with 0.20 m3 of carrier media (RK BioElements, RK Plast A/S, Denmark), sharing the same shape, and specific surface area (750 m2/m3). Active surface area (Amedia) in each reactor was, thus, 150 m2. Two of the biofilters were filled with RK BioElements Heavy (density 1.2 g/cm3, embedded with BaSO4), thus functioning as a static fixed bed biofilter (FBB). The other two biofilters were moving bed biofilters (MBB) filled with RK BioElements Medium (density 1.0 g/cm3). The media in MBB was kept in constant motion by delivery of 10 L/min compressed air at the bottom of each reactor. 125 126 127 128 The flow to each biofilter was measured with a portable clamp-on flowmeter (PortaflowTM204, Micronics Ltd, United Kingdom) and kept constant at 7.2 m3/h. This ensured identical hydraulic surface loading rates (Lhyd) at 345 m3/m2/d, and elevation speeds of 4 mm/s. Hydraulic residence time in the biofilters was close to 3.5 min. 129 130 131 132 From the top of each biofilter, the water overflowed to the trickling filter (BIOBLOK 150®, ExpoNet, Denmark) from where it re-entered the fish tank. The drum filter received a constant 3.0 m3/h water flow from the bottom outlet of the fish tank, which permitted the screening of the system water once every 2.8 h. 133 2.2 Management and operating conditions 134 135 136 137 138 139 The tank was initially stocked with 130 kg of 45 g rainbow trout (Oncorhynchus mykiss). A commercial diet (EFICO Enviro 3mm, BioMar, Denmark) was given to the fish and gradually increased until a maximum of 2.0 kg feed/d was reached. Make-up water was also gradually increased until 2.0 m3/d, while maintaining a feed loading of 1 kg feed/m3 make-up water at all times. The system was allowed to mature for 8 weeks until nitrate levels were constant, representing a stable balance between daily N removal (as nitrate) equivalent to the daily N addition (as ammonia + urea). 91 140 141 142 During the experiment, biomass removal was conducted regularly to keep the fish density between 35-50 kg/m3. The feed conversion ratio (FCR, g feed given/g biomass gain) was close to 1.1 for the whole experimental period, and total fish mortality was below 0.1 %. 143 144 145 146 147 148 149 150 Dissolved oxygen (DO) concentration was monitored daily (Hach Lange HQd40, Germany), and was maintained above 8 mg/L by aeration and additional oxygen diffusion into the rearing tank. The pH was also monitored daily (Hach Lange HQ40, Germany) and was maintained between 7.3 - 7.5 by the daily addition of sodium bicarbonate (~50 g NaHCO3/kg feed) to compensate for alkalinity consumption in the nitrification process. Alkalinity measurements were conducted twice a week using test strip kits (Hach® AquaCheck for Total Alkalinity, USA). Total ammonia nitrogen (TAN), NO2-N, and NO3-N levels were monitored daily using Merck KGaA (Germany) test kits (MColortestTM 111117, MQuantTM 110057, and MQuantTM 110020, respectively). 151 2.3 Sampling procedures 152 153 154 155 156 157 158 159 The experiment (Table 1) included an initial 8-week acclimatization and maturation period, followed by a 6-week experimental period. The experimental period was divided into four phases defined by the type of biofilter connected to the system. These included two phases where all filters were in operation and two phases where each biofilter mode of operation was exclusively connected to the RAS. Between each of the latter phases, all biofilters were connected to the system for reacclimatization. In all phases, water quality was measured three times a week to assess the individual or combined effect of biofilter mode of operation on particle, organic matter, and dissolved nitrogen dynamics. 160 161 162 163 164 Water quality assessment was conducted by collecting 24-h pooled samples in 10-L containers (1 sample of 400 mL per hour) using refrigerated automatic samplers (ISCO Inc., USA) at five different locations: in the system (outlet of the fish tank and common inlet to all biofilters), and at the outlet of each of the biofilters. Supplementary grab samples for particle size distribution (PSD) analysis were collected in the morning (before feeding) at the same locations. 165 2.4 Water analysis 166 167 168 169 170 171 Samples for PSD determination were analyzed within 30 min of collection using an optical particle counter (OPC), the AccuSizer™ 780 SIS (Particle Sizing Systems, USA) according to the method described in Fernandes et al. (2014). The measuring range of the OPC was 0.5-400 µm, but a lower limit threshold was defined at 5 µm, as the statistical representation of the particles below this size was not sufficiently accurate to demonstrate reproducibility (Brinker and Rösch, 2005; Fernandes et al., 2015). 172 173 174 175 Water samples collected automatically were kept refrigerated (4 ˚C) from the moment of collection until processing or analysis in the laboratory. They were analyzed for total and dissolved Chemical Oxygen Demand (COD), Biochemical Oxygen Demand, (BOD5), TAN (NH4+-N+NH3-N), nitrite-N and nitrate-N. 176 177 178 179 Laboratory procedures were according to standard protocols, as described in Fernandes et al., (2015). Total (CODTOT, BOD5 TOT) refers to analysis on homogeneous unfiltered samples, while Dissolved (CODDISS, BOD5 DISS) refers to filtered samples using GF/A filters (1.6 µm, 47 mm Ø; WhatmanTM, UK). 180 92 181 2.5 Calculations and statistical analyses 182 183 184 Particle size distribution results were utilized to gather data on total counts, volume and surface area; and to extract the relative contribution of specific size fractions (< 20, 20 - 40, 40 - 100 and > 100 µm), in the counts, volume and surface area categories. 185 186 187 188 189 Particulate fractions (CODPART, BOD5 PART) were calculated by subtracting the Dissolved from the Total fraction for each parameter. A biodegradability index of the organic matter (Zhu and Chen, 2001; Dalsgaard and Pedersen, 2011) was calculated as BOD5/COD for the Particulate, Dissolved and Total fractions. The biodegradability indexes were arcsine transformed before assessment of statistical significance. 190 191 192 193 194 195 Data processing and statistical tests were carried out in R (The R Project for Statistical Computing, Bell Laboratories, USA). Two-way repeated measures ANOVA were utilized to assess statistical differences between biofilters and mode of operation. Orthogonal contrasts (Yossa and Verdegem, 2015) were performed post hoc on raw data to assess differences in the means of the ANOVA groups. A probability level (α) of 0.05 was utilized to determine the statistical significance of each test. 93 196 3. Results 197 3.1 Effects on particle size distribution 198 199 200 201 202 203 Measured particle concentrations in the system ranged from 1117±114 particles/mL to 2161±431 particles/mL throughout the experiment (Fig. 2a). Particle concentration out of the FBB was always lower than out of the MBB (Fig. 2a). An approximate 10 % particle concentration decrease by FBB corresponded to a 10 % reduction in particle surface area (Fig. 2b) and particle volume (Fig. 2c). MBB released approximately 10 % more particles and particulate surface area (Fig. 2b), though these increases had virtually no effect on particle volume (Fig. 2c). 204 205 206 207 208 209 210 211 212 213 MBB and FBB differed in their removal of particle counts, area and volume in the < 20, and the 2040 µm fractions (Fig. 3). At the < 20 µm range (Fig. 3a), MBB generated roughly 250 particles/mL, while FBB removed approximately 200 particles/mL, both compared to the same system concentration. This difference between modes of operation, at this size class, was also apparent in terms of particulate surface area (Fig. 3b) and particle volume (Fig. 3c). At the 20-40 µm range, on average, MBB produced 14 particles/mL while FBB removed 12 particles/mL, which was also proportionately reflected in terms of particulate surface area and volume. At the 40-100 µm range, MBB had no effect on particle concentration, surface area, or volume, while all three parameters were marginally reduced in FBB. For the size fraction larger than 100 µm, both modes of operation removed particles and reduced surface area and volume at similar levels. 214 3.2 Effects on organic matter 215 216 217 218 A constant level of dissolved COD was observed in the RAS throughout the study (Fig. 4a, 18.8±1.2 mg O2/L). Particulate COD in the system water increased when only MBB were in operation: from 3.9±0.46 mg O2/L in the first three phases, to 6.0±1.50 mg O2/L in the last phase (Fig. 4a). COD removal occurred at similar levels for FBB and MBB. 219 220 221 222 223 224 225 Dissolved BOD5 levels were constant at 1.8±0.17 mg O2/L throughout the experiment (Fig. 4b). Particulate BOD5 values increased exclusively when only MBBs were in operation: from 2.0±0.16 mg O2/L in the first three phases, against 3.7±0.43 mg O2/L in the last phase (Fig. 4b). Total BOD5 removal was similar for both FBB and MBB. Particulate BOD5 removal per 24 h was larger in MBB than FBB (avg. 1.5±0.23 mg O2/L vs. 0.9±0.33 mg O2/L, respectively), while the opposite was observed in terms of dissolved BOD5 removal (0.5±0.29 mg O2/L for FBB and 0.0±0.15 mg O2/L for MBB). 226 227 228 229 230 Throughout the experiment, system COD concentration was 5 times larger than system BOD5 concentration. System biodegradability indexes (Table 2) were constant for total (avg. 0.18 BOD5/COD), dissolved (avg. 0.08 BOD5/COD) and particulate (avg. 0.58 BOD5/COD) organic matter fractions. Removal of biodegradable organic matter was larger in the particulate than in the dissolved fraction, for both modes of operation. 231 3.3 Effects on nitrogenous compounds 232 233 234 235 TAN and nitrite concentrations in the RAS (Fig. 5) were low throughout the study, ranging from 0.12±0.02 mg TAN/L to 0.27±0.03 mg TAN/L throughout the study, while nitrite-N concentrations ranging from 0.10±0.02 mg N/L to 0.22±0.02 mg N/L (Fig. 5b). Nitrate-N concentration in the RAS was elevated and ranged from 40±1.0 mg N/L to 45±1.2 mg N/L (Fig. 5c) throughout the study. 94 236 237 238 239 Average TAN removed in FBB and MBB was 0.06±0.04 mg N/L and 0.04±0.01 mg N/L, respectively. Nitrite-N was consistently lower at the outlet of FBB, compared to MBB outlet levels. FBB and MBB produced 0.38±0.12 mg N/L and 0.35±0.16 mg N/L of nitrate-N per each 24-h sampling period. 95 240 4. Discussion 241 4.1 Effects on particle size distribution 242 243 244 245 246 More than 93 % of the particles measured were below 20 µm in size. Particle concentration increased after passage through the MBB while the opposite was observed in the FBB. At high particle concentrations, the OPC operated in this study, showed decreased performance in measuring particles below 5 µm (Brinker and Rösch, 2005; Fernandes et al., 2015). Due to this statistical limitation, all data below this threshold were excluded from the analysis. 247 248 249 250 251 252 253 254 255 256 The effects of MBBs on PSD observed in this study were similar to the results obtained by Åhl et al. (2006) and Ivanovic and Leiknes (2012). The balance between particle volume production and removal through all size classes in MBB was equal to zero. This means that the volume removed from the large particle fraction ended up in the fine particle fraction. Mechanical stress, or shear forces (Droppo et al., 1997; Ivanovic and Leiknes, 2008) within the reactor, induced by vigorous aeration and mixing, is an important factor in the process of particle disintegration. Furthermore, the shear forces induced to the media by aeration sustain the self-cleaning mechanisms of MBBs, by constantly shedding off excess biofilm (Ødegaard et al., 1994; Rusten et al., 2006). Thus, MBB generated small particles predominantly through disintegration and transformation of larger particles, though biofilm shear and tear probably added to the effect. 257 258 259 260 261 262 263 264 265 266 267 A significant particle volume reduction (Fig. 3c) within FBB implies that particles entering the filters were therein trapped. Particle retention could be caused by density dependent settling when transport occurred through the water column (Wong and Piedrahita, 2000; Patterson et al., 2003). Alternatively, the settled media and natural growth of the biofilm may have generated a biofilm barrier that entrapped particles at the bottom of the biofilter (Arvin and Harremoës, 1990; Larsen and Harremoës, 1994). Either way, FBB in this study performed as a secondary particle removal unit, as has previously been shown (Bouwer, 1987; Franco-Nava et al., 2004b). Backwashing is important for the maintenance of quasi-optimal conditions in FBB (Singh et al., 1999; Malone and Pfeiffer, 2006; Emparanza, 2009; Pfeiffer and Wills, 2011). Future studies regarding backwash periodicity could further describe particle dynamics through FBB, and how to maximize or optimize the removal of particles. 268 269 270 271 272 273 274 275 At stable conditions of system maintenance, feed loading, system turnover rates, and hydraulic retention time in the system, PSD can become a stable parameter in a given RAS (McMillan et al., 2003; Patterson and Watts, 2003; Fernandes et al., 2014, 2015). In the present study, and because feed loading was maintained constant, all changes to background PSD could be attributed to changes in biofilter operation. System levels of particle concentration where thus dependent on type and number of biofilters utilized: lower concentrations were measured with FBB in operation only, higher with only MBB; and intermediate when all filters were connected. Therefore, it seems that PSD in RAS can be influenced by biofilter mode of operation. 276 4.2 Effects on organic matter 277 278 279 280 281 282 The net removal of total organic matter (CODTOT and BOD5 TOT) was similar in both modes of operation throughout the study. MBB removed more particulate BOD5, while FBB removed more dissolved BOD5. Accumulation of dissolved compounds in RAS, is mostly determined by the feed loading (Leonard et al., 2000; Martins et al., 2009; Pedersen et al., 2012; Verdegem, 2013). If the conditions are optimized (Bergheim et al., 1993; Chen et al., 1997), particulate organic matter can also be a stable parameter in RAS (McMillan et al., 2003; Patterson and Watts, 2003; Fernandes et 96 283 284 285 286 287 288 289 290 291 al., 2014, 2015). Since feed loading was constant throughout the experiment, variations in system water quality would likely have been caused by changes in biofilter operation only. The higher removal of particulate and dissolved organic matter by the FBB then explains the lower background levels when only FBB was operated and, likewise, the increased background levels when only MBB was operated. Though micro particle concentration and distribution are related to feed loading and RAS components, biofilter mode of operation also affects type and amount of dissolved and particulate substances. These variations can be attributed to differences in hydraulics (Singh et al., 1999; Pfeiffer and Wills, 2011; Vigne et al., 2011; Prehn et al., 2012) and microbial communities (Hagopian and Riley, 1998; Fdz-Polanco et al., 2000; Michaud et al., 2014). 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 Bacteria assimilate nutrients only in dissolved form, either directly from the environment, or after hydrolyzation of particulate matter (Särner, 1986; Larsen and Harremoës, 1994; Yang et al., 2001; Leonard et al., 2002). The net consumption of BOD5 DISS in FBB, as opposed to MBB, might represent such differences in microbial communities between the biofilters (Chen et al., 2006; Michaud et al., 2006, 2014; Blancheton et al., 2013). Heterotrophic bacteria in FBB may have consumed dissolved organic matter directly, while hydrolyzation of particulate organic matter masked the consumption of dissolved organic matter in MBB. As particle size decreases, adhered organic matter becomes more bio-available for bacteria (Balmat, 1957; Levine et al., 1985, 1991), and so the disintegration of particles may have helped sustain heterotrophic communities in MBB. Lastly, due to the operating mechanisms, biofilm shear and tear occurs more notably in MBB. Bacterial biofilm accounts for slowly biodegradable organic matter (COD) and not for easily biodegradable organic matter (BOD5) (Kwok et al., 1998; Elenter et al., 2007; Morgenroth, 2008). Therefore, the discrepancies in the consumption of both fractions through MBB can probably be attributed to the consumption of readily available particulate organic matter (as particles), and the release of slowly biodegradable particulate organic matter (as biofilm). 307 4.3 Effects on nitrogenous compounds 308 309 310 311 312 313 All filters performed with stable nitrification rates when operated together or separately, as observed by low TAN and nitrite concentrations, and stable, elevated nitrate concentrations throughout the study (Colt et al., 2006; Pedersen et al., 2012). Stable nitrate concentrations (40-45 mg N/L) in the RAS throughout the study reflect constant experimental conditions, at a feeding level of 1 kg/m3 of make-up water, and that the system was operated at steady state. This is an important prerequisite that can justify the comparison between filter types in the present experimental setup. 314 315 316 317 318 319 Generally, MBB presents lower nitrification rates than FBB in the same experimental conditions (Suhr and Pedersen, 2010; Michaud et al., 2014; Pedersen et al., 2015), which may help explain the increase of both compounds, and the reaching of a new system level when only MBB was in operation (Fig. 5a,b). However, since TAN and nitrite-N did not build-up significantly during the experiment, it can be concluded that the changes caused by number and type of biofilter in operation did not affect system values and performance. 320 321 322 323 324 325 326 327 Some studies have identified nitrification to be impaired at system dissolved carbon to nitrogen (C:N) ratios above 1:1 (Zhu and Chen, 2001; Carrera et al., 2004; Eding et al., 2006; Michaud et al., 2006, 2014; Guerdat et al., 2011). However, these studies usually disregard the contribution of particulate organic matter. In our case, at system average C:N ratios close to 1.8:1 (dissolved) or 2.4:1 (dissolved + particulate) nitrification occurred at similar rates to what has been published with the same type of reactors (Suhr and Pedersen, 2010; Pedersen et al., 2012, 2015; Prehn et al., 2012). These high C:N ratios may not have affected the nitrification performance of any reactor type due to optimal hydraulics, that kept oxygen bulk concentrations close to saturation. At 4 mm/s water eleva- 97 328 329 330 tion speed, oxygen and ammonia can penetrate deeper into the biofilm (Henze, 2002; Mašić et al., 2010; Vigne et al., 2010; Prehn et al., 2012), thus, potentially nourishing the simultaneous removal of organic matter and nitrogenous compounds (Bovendeur et al., 1990a,b; Rusten et al., 2006). 98 331 5. Conclusion 332 333 334 335 336 337 338 339 340 In this study, particle dynamics and balances were compared through two types of biofilter. Static FBB removed particles of all sizes from the system water by entrapment. Mechanical shear caused by aeration and associated filter movement caused large particles to disintegrate into smaller ones in MBB, also releasing slowly biodegradable biofilm in the process. Both of the two biofilter types tested removed dissolved BOD5 (highest in FBB), and particulate BOD5 (highest in MBB). The trends observed when only FBB or MBB were in operation were conserved when all filters were operated simultaneously. Consumption of organic matter, or the resulting micro particle distribution, did not seem to affect the normal nitrification performance of each mode of operation, as compared to other studies. 99 341 6. Acknowledgements 342 343 344 345 The technical support and caretaking of the fish and the system by Erik Poulsen, Ole Madvig Larsen, and Remko Oosterveld, as well as the help during the laboratory analysis by Dorthe Frandsen, Ulla Sproegel, Brian Møller, and Sara Møller Nielsen, all staff at the Technical University of Denmark, DTU Aqua, Section for Aquaculture (Hirtshals, Denmark), is highly appreciated. 100 346 Bibliography 347 348 Arvin, E., Harremoës, P., 1990. Concepts and models for biofilm reactor performance. Wat. Sci. Tech. 22, 171–192. 349 350 Balmat, J.L., 1957. Biochemical oxidation of various particulate fractions of sewage. Sewage Ind. Waste. 29, 757–761. 351 352 353 Bergheim, A., Sanni, S., Indrevik, G., Hølland, P., 1993. Sludge removal from salmonid tank effluent using rotating microsieves. Aquac. Eng. 12, 97–109. doi:10.1016/01448609(93)90019-8 354 355 356 Blancheton, J.P., Attramadal, K.J.K., Michaud, L., D’Orbcastel, E.R., Vadstein, O., 2013. Insight into bacterial population in aquaculture systems and its implication. Aquac. Eng. 53, 30–39. doi:10.1016/j.aquaeng.2012.11.009 357 358 Bouwer, E.J., 1987. Theoretical investigation of particle deposition in biofilm systems. Water Res. 21, 1489–1498. doi:10.1016/0043-1354(87)90132-1 359 360 361 Bovendeur, J., Spuij, F., Hermans, P.G.B., Klapwijk, A., 1990a. Mechanisms and kinetics of COD removal and simultaneous oxygen consumption in a rotating biological contactor biofilm. Wat. Sci. Tech. 22, 75–87. 362 363 364 Bovendeur, J., Zwaga, A.B., Lobee, B.G.J., Blom, J.H., 1990b. Fixed-biofilm reactors in aquacultural water recycle systems: effect of organic matter elimination on nitrification kinetics. Water Res. 24, 207–213. 365 366 Brinker, A., Rösch, R., 2005. Factors determining the size of suspended solids in a flow-through fish farm. Aquac. Eng. 33, 1–19. doi:10.1016/j.aquaeng.2004.10.003 367 368 369 Bullock, G.L., Summerfelt, S.T., Noble, A.C., Weber, A.L., Durant, M.D., Hankins, J.A., 1997. Ozonation of a recirculating rainbow trout culture system I. Effects on bacterial gill disease and heterotrophic bacteria. Aquaculture 158, 43–55. doi:10.1016/S0044-8486(97)00063-X 370 371 372 Carrera, J., Vicent, T., Lafuente, J., 2004. Effect of influent COD/N ratio on biological nitrogen removal (BNR) from high-strength ammonium industrial wastewater. Process Biochem. 39, 2035–2041. doi:10.1016/j.procbio.2003.10.005 373 374 375 Chapman, P.M., Popham, J.D., Griffin, J., Leslie, D., Michaelson, J., 1987. Differentiation of physical from chemical toxicity in solid waste fish bioassays. Water, Air Soil Pollut. 33, 295– 308. 376 377 378 Chen, S., Coffin, D.E., Malone, R.F., 1997. Sludge production and management for Recirculating Aquacultural Systems. J. World Aquac. Soc. 28, 303–315. doi:10.1111/j.17497345.1997.tb00278.x 379 380 Chen, S., Ling, J., Blancheton, J.P., 2006. Nitrification kinetics of biofilm as affected by water quality factors. Aquac. Eng. 34, 179–197. doi:10.1016/j.aquaeng.2005.09.004 101 381 382 383 Chen, S., Timmons, M.B., Aneshansley, D.J., Bisogni Jr., J.J., 1993. Suspended solids characteristics from recirculating aquacultural systems and design implications. Aquaculture 112, 143–155. 384 385 Colt, J., Lamoureux, J., Patterson, R.N., Rogers, G.L., 2006. Reporting standards for biofilter performance studies. Aquac. Eng. 34, 377–388. doi:10.1016/j.aquaeng.2005.09.002 386 387 Cripps, S.J., 1995. Serial particle size fractionation and characterisation of an aquacultural effluent. Aquaculture 133, 323–339. doi:10.1016/0044-8486(95)00021-S 388 389 Dalsgaard, J., Pedersen, P.B., 2011. Solid and suspended/dissolved waste (N, P, O) from rainbow trout (Oncorynchus mykiss). Aquaculture 313, 92–99. doi:10.1016/j.aquaculture.2011.01.037 390 391 392 Droppo, I.G., Leppard, G.G., Flannigan, D.T., Liss, S.N., 1997. The freshwater floc: A functional relationship of water and organic and inorganic floc constituents affecting suspended sediment properties. Water, Air Soil Pollut. 99, 43–54. 393 394 395 Eding, E.H., Kamstra, A., Verreth, J.A.J., Huisman, E.A., Klapwijk, A., 2006. Design and operation of nitrifying trickling filters in recirculating aquaculture: A review. Aquac. Eng. 34, 234–260. doi:10.1016/j.aquaeng.2005.09.007 396 397 398 Elenter, D., Milferstedt, K., Zhang, W., Hausner, M., Morgenroth, E., 2007. Influence of detachment on substrate removal and microbial ecology in a heterotrophic/autotrophic biofilm. Water Res. 41, 4657–4671. doi:10.1016/j.watres.2007.06.050 399 400 Emparanza, E.J.M., 2009. Problems affecting nitrification in commercial RAS with fixed-bed biofilters for salmonids in Chile. Aquac. Eng. 41, 91–96. doi:10.1016/j.aquaeng.2009.06.010 401 402 403 Fdz-Polanco, F., Méndez, E., Urueña, M. a., Villaverde, S., García, P. a., 2000. Spatial distribution of heterotrophs and nitrifiers in a submerged biofilter for nitrification. Water Res. 34, 4081– 4089. doi:10.1016/S0043-1354(00)00159-7 404 405 406 Fernandes, P.M., Pedersen, L.-F., Pedersen, P.B., 2014. Daily micro particle distribution of an experimental recirculating aquaculture system-A case study. Aquac. Eng. 60, 28–34. doi:10.1016/j.aquaeng.2014.03.007 407 408 409 Fernandes, P.M., Pedersen, L.-F.L.-F., Pedersen, P.B.P.B., 2015. Microscreen effects on water quality in replicated recirculating aquaculture systems. Aquac. Eng. 65, 17–26. doi:10.1016/j.aquaeng.2014.10.007 410 411 412 413 Franco-Nava, M.A., Blancheton, J.P., Deviller, G., Charrier, A., Le-Gall, J.-Y., 2004a. Effect of fish size and hydraulic regime on particulate organic matter dynamics in a recirculating aquaculture system: elemental carbon and nitrogen approach. Aquaculture 239, 179–198. doi:10.1016/j.aquaculture.2004.05.040 414 415 Franco-Nava, M.A., Blancheton, J.P., Deviller, G., Le-Gall, J.-Y., 2004b. Particulate matter dynamics and transformations in a recirculating aquaculture system: application of stable 102 416 417 isotope tracers in seabass rearing. Aquac. Eng. 31, 135–155. doi:10.1016/j.aquaeng.2004.01.003 418 419 420 Guerdat, T.C., Losordo, T.M., Classen, J.J., Osborne, J.A., DeLong, D., 2011. Evaluating the effects of organic carbon on biological filtration performance in a large scale recirculating aquaculture system. Aquac. Eng. 44, 10–18. doi:10.1016/j.aquaeng.2010.10.002 421 422 Hagopian, D.S., Riley, J.G., 1998. A closer look at the bacteriology of nitrification. Aquac. Eng. 18, 223–244. doi:10.1016/S0144-8609(98)00032-6 423 424 Hem, L.J., Rusten, B., Ødegaard, H., 1994. Nitrification in a moving bed biofilm reactor. Water Res. 28, 1425–1433. 425 426 427 Henze, M., 2002. Basic Biological Processes, in: Henze, M., Harremoës, P., Jes la Cour, J., Arvin, E. (Eds.), Wastewater Treatment: Biological and Chemical Processes. Springer-Verlag, Berlin, Heidelberg, p. 430. 428 429 430 Ivanovic, I., Leiknes, T.O., 2008. Impact of aeration rates on particle colloidal fraction in the biofilm membrane bioreactor (BF-MBR). Desalination 231, 182–190. doi:10.1016/j.desal.2007.11.046 431 432 Ivanovic, I., Leiknes, T.O., 2012. Particle Separation in Moving Bed Biofilm Reactor: Applications and Opportunities. Sep. Sci. Technol. 47, 647–653. doi:10.1080/01496395.2011.639590 433 434 Kelly, L.A., Bergheim, A., Stellwagen, J., 1997. Particle size distribution of wastes from freshwater fish farms. Aquac. Int. 5, 65–78. 435 436 437 438 Kwok, W.K., Picioreanu, C., Ong, S.L., Van Loosdrecht, M.C.M., Ng, W.J., Heijnen, J.J., 1998. Influence of biomass production and detachment forces on biofilm structures in a biofilm airlift suspension reactor. Biotechnol. Bioeng. 58, 400–407. doi:10.1002/(SICI)10970290(19980520)58:4<400::AID-BIT7>3.0.CO;2-N 439 440 Larsen, T.A., Harremoës, P., 1994. Degradation mechanisms of colloidal organic matter in biofilm reactors. Water Res. 28, 1443–1452. 441 442 443 Leonard, N., Blancheton, J.P., Guiraud, J.., 2000. Populations of heterotrophic bacteria in an experimental recirculating aquaculture system. Aquac. Eng. 22, 109–120. doi:10.1016/S01448609(00)00035-2 444 445 446 Leonard, N., Guiraud, J.., Gasset, E., Cailleres, J.., Blancheton, J.P., 2002. Bacteria and nutrients— nitrogen and carbon—in a recirculating system for sea bass production. Aquac. Eng. 26, 111– 127. doi:10.1016/S0144-8609(02)00008-0 447 448 449 Levine, A.D., Tchobanoglous, G., Asano, T., 1985. Characterization of the size distribution of contaminants in wastewater: Treatment and reuse implications. Water Pollut. Control Fed. 57, 805–816. doi:10.2307/25042701 103 450 451 452 Levine, A.D., Tchobanoglous, G., Asano, T., 1991. Size distributions of particulate contaminants in wastewater and their impact on treatability. Water Res. 25, 911–922. doi:10.1016/00431354(91)90138-G 453 454 Liltved, H., Cripps, S.J., 1999. Removal of particle-associated bacteria by pre®ltration and ultraviolet irradiation. Aquaculture 445–450. doi:10.1046/j.1365-2109.1999.00349.x 455 456 Malone, R.F., Pfeiffer, T.J., 2006. Rating fixed film nitrifying biofilters used in recirculating aquaculture systems. Aquac. Eng. 34, 389–402. doi:10.1016/j.aquaeng.2005.08.007 457 458 459 Marquet, R., Mietton-Peuchot, M., Wheatley, A.D., 1999. Characterisation of trickling filter effluent by particle size distribution and high performance size exclusion chromatography. Water Res. 33, 1415–1424. doi:10.1016/S0043-1354(98)00337-6 460 461 462 Marquet, R., Muhammad, N., Vairavamoorthy, K., Wheatley, A.D., 2007. Particle size distribution to assess the performance of trickling filters. Process Saf. Environ. Prot. 85, 99–103. doi:10.1205/psep.05194 463 464 465 466 Martins, C.I.M., Pistrin, M.G., Ende, S.S.W., Eding, E.H., Verreth, J.A.J., 2009. The accumulation of substances in Recirculating Aquaculture Systems (RAS) affects embryonic and larval development in common carp Cyprinus carpio. Aquaculture 291, 65–73. doi:10.1016/j.aquaculture.2009.03.001 467 468 469 Mašić, A., Bengtsson, J., Christensson, M., 2010. Measuring and modeling the oxygen profile in a nitrifying Moving Bed Biofilm Reactor. Math. Biosci. 227, 1–11. doi:10.1016/j.mbs.2010.05.004 470 471 472 McMillan, J.., Wheaton, F., Hochheimer, J.., Soares, J., 2003. Pumping effect on particle sizes in a recirculating aquaculture system. Aquac. Eng. 27, 53–59. doi:10.1016/S0144-8609(02)000389 473 474 475 Michaud, L., Blancheton, J.P., Bruni, V., Piedrahita, R., 2006. Effect of particulate organic carbon on heterotrophic bacterial populations and nitrification efficiency in biological filters. Aquac. Eng. 34, 224–233. doi:10.1016/j.aquaeng.2005.07.005 476 477 478 Michaud, L., Lo Giudice, A., Interdonato, F., Triplet, S., Ying, L., Blancheton, J.P., 2014. C/N ratio-induced structural shift of bacterial communities inside lab-scale aquaculture biofilters. Aquac. Eng. 58, 77–87. doi:10.1016/j.aquaeng.2013.11.002 479 480 Morgenroth, E., 2008. Modelling Biofilms, Biological Wastewater Treatment - Principles, Modelling and Design. 481 482 483 Nogueira, R., Melo, L.F., Purkhold, U., Wuertz, S., Wagner, M., 2002. Nitrifying and heterotrophic population dynamics in biofilm reactors: Effects of hydraulic retention time and the presence of organic carbon. Water Res. 36, 469–481. doi:10.1016/S0043-1354(01)00229-9 104 484 485 486 Okabe, S., Oozawa, Y., Hirata, K., Watanabe, Y., 1996. Relationship between population dynamics of nitrifiers in biofilms and reactor performance at various C:N ratios. Wat. Res. 30, 1563– 1572. 487 488 489 Patterson, R.N., Watts, K.C., 2003. Micro-particles in recirculating aquaculture systems: particle size analysis of culture water from a commercial Atlantic salmon site. Aquac. Eng. 28, 99– 113. doi:10.1016/S0144-8609(03)00003-7 490 491 492 Patterson, R.N., Watts, K.C., Gill, T.A., 2003. Micro-particles in recirculating aquaculture systems: determination of particle density by density gradient centrifugation. Aquac. Eng. 27, 105–115. doi:10.1016/S0144-8609(02)00042-0 493 494 Patterson, R.N., Watts, K.C., Timmons, M.B., 1999. The power law in particle size analysis for aquacultural facilities. Aquac. Eng. 19, 259–273. doi:10.1016/S0144-8609(98)00054-5 495 496 497 Pedersen, L.-F., Oosterveld, R., Pedersen, P.B., 2015. Nitrification performance and robustness of fixed and moving bed biofilters having identical carrier elements. Aquac. Eng. 65, 37–45. doi:10.1016/j.aquaeng.2014.10.005 498 499 500 Pedersen, L.-F., Suhr, K.I., Dalsgaard, J., Pedersen, P.B., Arvin, E., 2012. Effects of feed loading on nitrogen balances and fish performance in replicated recirculating aquaculture systems. Aquaculture 338-341, 237–245. doi:10.1016/j.aquaculture.2012.01.035 501 502 503 Pfeiffer, T.J., Wills, P.S., 2011. Evaluation of three types of structured floating plastic media in moving bed biofilters for total ammonia nitrogen removal in a low salinity hatchery recirculating aquaculture system. Aquac. Eng. 45, 51–59. doi:10.1016/j.aquaeng.2011.06.003 504 505 506 Picioreanu, C., van Loosdrecht, M.C., Heijnen, J.J., 2000. A theoretical study on the effect of surface roughness on mass transport and transformation in biofilms. Biotechnol. Bioeng. 68, 355–369. 507 508 509 Prehn, J., Waul, C.K., Pedersen, L.-F., Arvin, E., 2012. Impact of water boundary layer diffusion on the nitrification rate of submerged biofilter elements from a recirculating aquaculture system. Water Res. 46, 3516–24. doi:10.1016/j.watres.2012.03.053 510 511 512 Rasmussen, K., Lewandowski, Z., 1998. Microelectrode measurements of local mass transport rates in heterogeneous biofilms. Biotechnol. Bioeng. 59, 302–309. doi:10.1002/(SICI)10970290(19980805)59:3<302::AID-BIT6>3.0.CO;2-F 513 514 Rusten, B., Eikebrokk, B., Ulgenes, Y., Lygren, E., 2006. Design and operations of the Kaldnes moving bed biofilm reactors. Aquac. Eng. 34, 322–331. doi:10.1016/j.aquaeng.2005.04.002 515 516 517 Schreier, H.J., Mirzoyan, N., Saito, K., 2010. Microbial diversity of biological filters in recirculating aquaculture systems. Curr. Opin. Biotechnol. 21, 318–25. doi:10.1016/j.copbio.2010.03.011 518 Sharma, B., Ahlert, R.C., 1977. Nitrification and nitrogen removal. Water Res. 11, 897–925. 105 519 520 521 Sindilariu, P.-D., Brinker, A., Reiter, R., 2009. Waste and particle management in a commercial, partially recirculating trout farm. Aquac. Eng. 41, 127–135. doi:10.1016/j.aquaeng.2009.03.001 522 523 Singh, S., Ebeling, J.M., Wheaton, F., 1999. Water quality trials in four recirculating aquacultural system configurations. Aquac. Eng. 20, 75–84. doi:10.1016/S0144-8609(99)00003-5 524 525 526 Suhr, K.I., Pedersen, P.B., 2010. Nitrification in moving bed and fixed bed biofilters treating effluent water from a large commercial outdoor rainbow trout RAS. Aquac. Eng. 42, 31–37. doi:10.1016/j.aquaeng.2009.10.001 527 528 Särner, E., 1986. Removal of particulate and dissolved organics in aerobic fixed-film biological processes. Water Pollut. Control Fed. 58, 165–172. 529 Timmons, M.B., Ebeling, J.M., 2010. Recirculating Aquaculture, 10th ed, Aquaculture. 530 531 Verdegem, M.C.J., 2013. Nutrient discharge from aquaculture operations in function of system design and production environment. Rev. Aquac. 5, 158–171. doi:10.1111/raq.12011 532 533 534 Vigne, E., Choubert, J.-M., Canler, J.-P., Héduit, A., Sorensen, K., Lessard, P., 2010. A biofiltration model for tertiary nitrification of municipal wastewaters. Water Res. 44, 4399–4410. doi:10.1016/j.watres.2010.06.005 535 536 537 Vigne, E., Choubert, J.-M., Canler, J.-P., Heduit, A., Sørensen, K.H., Lessard, P., 2011. The role of loading rate, backwashing, water and air velocities in an up-flow nitrifying tertiary filter. Bioresour. Technol. 102, 904–912. doi:10.1016/j.biortech.2010.09.031 538 Wanner, O., Gujer, W., 1984. Competition in biofilms. Water Sci. Technol. 17, 27–44. 539 540 541 Witham, C.S., Oppenheimer, C., Horwell, C.J., 2005. Volcanic ash-leachates: a review and recommendations for sampling methods. J. Volcanol. Geotherm. Res. 141, 299–326. doi:10.1016/j.jvolgeores.2004.11.010 542 543 544 545 Wold, P.-A., Holan, A.B., Øie, G., Attramadal, K.J.K., Bakke, I., Vadstein, O., Leiknes, T.O., 2014. Effects of membrane filtration on bacterial number and microbial diversity in marine recirculating aquaculture system (RAS) for Atlantic cod (Gadus morhua L.) production. Aquaculture 422-423, 69–77. doi:10.1016/j.aquaculture.2013.11.019 546 547 Wong, K.B., Piedrahita, R., 2000. Settling velocity characterization of aquacultural solids. Aquac. Eng. 21, 233–246. doi:10.1016/S0144-8609(99)00033-3 548 549 Yang, L., Chou, L.S., Shieh, W.K., 2001. Biofilter treatment of aquaculture water for reuse applications. Water Res. 35, 3097–108. 550 551 552 Yossa, R., Verdegem, M.C.J., 2015. Misuse of multiple comparison tests and underuse of contrast procedures in aquaculture publications. Aquaculture 437, 344–350. doi:10.1016/j.aquaculture.2014.12.023 106 553 554 Zhang, T.C., Fu, Y.-C., Bishop, P.L., 1995. Competitions for substrate and space in biofilms. Water Environ. Res. 67, 992–1003. 555 556 Zhu, S., Chen, S., 2001. Effects of organic carbon on nitrification rate in fixed film biofilters. Aquac. Eng. 25, 1–11. doi:10.1016/S0144-8609(01)00071-1 557 558 Ødegaard, H., Rusten, B., Westrum, T., 1994. A new moving bed biofilm reactor - applications and results. Water Sci. Technol. 559 560 561 Åhl, R.M., Leiknes, T.O., Ødegaard, H., 2006. Tracking particle size distributions in a moving bed biofilm membrane reactor for treatment of municipal wastewater. Water Sci. Technol. 53, 33. doi:10.2166/wst.2006.205 562 107 563 564 Authors - Paulo Mira Fernandes 565 566 567 568 569 Paulo Fernandes is a PhD student from the Section for Aquaculture, National Institute of Aquatic Resources, in the Technical University of Denmark (Hirtshals, Denmark). His project topic is on the “Interactions between colloidal particles and biofilters in Recirculating Aquaculture Systems”. - Lars-Flemming Pedersen 570 571 572 573 574 Lars-Flemming Pedersen is a senior research scientist in the Section for Aquaculture, National Institute of Aquatic Resources, in the Technical University of Denmark (Hirtshals, Denmark). His research at the section includes water quality in recirculating aquaculture systems. - Per Bovbjerg Pedersen 575 576 577 578 579 Per Bovbjerg Pedersen is a senior research scientist and head of section Section for Aquaculture, National Institute of Aquatic Resources, in the Technical University of Denmark (Hirtshals, Denmark). He is involved in aquaculture through research, development and counseling mostly regarding recirculating aquaculture systems and including model trout farms. 108 580 Tables (2) 581 582 583 Table 1. Timeline and protocol for the experimental RAS with four biofilters, two fixed bed biofilters (FBB) and two moving bed biofilters (MBB). A constant feed loading of 1 kg/m3 make-up water was ensured from start to end. Amedia – Total active surface area of media (m2). 584 585 586 Timeline Biofilters weeks 1-8 All filters Amedia (m2) 600 Feed (kg/d) 0.5-2 MUW (m3/d) 0.5-2 Feed loading (kg/m3) 1 Amedia loading (g /m2/d) 0.8-3.3 weeks 9-10 All filters 600 2 2 1 3.3 Water quality assessment ω week 11 FBB only χ 300 1 1 1 3.3 Water quality assessment ω weeks 12-13 All filters 600 2 2 1 3.3 Water quality assessment ω week 14 1 1 1 3.3 MBB only ψ 300 χ – MBB bypassed ψ – FBB bypassed ω – Measurements of COD, BOD5, PSD, and N compounds, as described in sections 2.3 and 2.4. Description RAS build-up Biofilter maturation Fish acclimatization Water quality assessment ω 109 587 588 589 Table 2. Biodegradability indexes in the system (N=6) and at the outlet (N=6) of each biofilter mode of operation. BOD5 – Biochemical Oxygen Demand after 5 days of incubation; COD – Chemical Oxygen Demand; FBB – Fixed bed biofilters; MBB – Moving bed biofilters. System Outlet 590 591 592 χ Location Total BOD5/CODχ Dissolved FBB MBB FBB MBB 0.15 ± 0.06 0.21 ± 0.00 0.12a± 0.02 0.17b± 0.00 0.08 ± 0.03 0.09 ± 0.00 0.08 ± 0.02 0.09 ± 0.01 Particulate 0.54 ± 0.13 0.62 ± 0.07 0.49 ± 0.07 0.46 ± 0.06 – Biodegradability indexes calculated based on system and outlet BOD5 and COD levels, as described in section 2.5. – values in the same column with a different superscript letter are significantly different (P<0.05) between modes of operation. a,b 110 593 Figures (5) 594 595 596 597 598 599 600 Fig. 1. The experimental 8.5 m3 recirculating aquaculture system with a 40-µm drum filter and four equal biofilters mounted in parallel, connected to one trickling filter above the fish tank. Fixed bed biofilters containing heavy carrier elements are denoted FBB; moving bed biofilters containing neutral carrier elements are denoted MBB. Arrows represent water flow direction: black represents water inlet and outflow; white represents flow with all filters in operation; and yellow represents water flows during biofilter bypass. 111 4000 a Particle concentration (no. particles/mL) 3500 System *** Outlet FBB Outlet MBB b 3000 a,b 2500 2000 * a,b 1500 a 1000 500 0 All filters 601 FBB only All filters MBB only Particle surface area (mm2/mL) 4 3.5 b System Outlet FBB Outlet MBB 3 *** 2.5 b 2 1.5 a,b 1 * a,b a 0.5 0 All filters 602 Particle volume (mm3/mL) 0.04 0.035 FBB only c System All filters Outlet FBB MBB only Outlet MBB 0.03 0.025 0.02 0.015 N.S. b 0.01 a,b N.S. 0.005 a,b a 0 All filters 603 604 605 606 607 608 FBB only All filters Experimental Phase MBB only Fig. 2. Total particle (mean ±SD) concentration (a), area (b), and volume (c) at the system level and biofilter outlets for each experimental phase (as described in Table 1), for particles sized between 5400 µm. System values with different superscript letters are significantly different between phases at P < 0.05. Significant differences between the two modes of biofilter operation in the same phase are marked with asterisks. *** – P < 0.001; * – P < 0.05; N.S. – P > 0.05. 112 Changes (OUT-IN) in particle concentration per specified size class (no. particles/mL) 400 300 a *** Fixed Bed Moving Bed 200 100 ** N.S. N.S. 20-40 40-100 >100 0 -100 -200 -300 -400 <20 609 Changes (OUT-IN) in equivalent particle area per specified size class (mm2/mL) 0.15 b *** Fixed Bed Moving Bed 0.1 *** 0.05 * N.S. 40-100 >100 0 -0.05 -0.1 -0.15 <20 Changes (OUT-IN) in equivalent particle volume per specified size class (mm3/mL) 610 c Fixed Bed Moving Bed 4.0E-04 *** 2.0E-04 ** * N.S. 0.0E+00 -2.0E-04 -4.0E-04 -6.0E-04 <20 611 612 613 614 615 616 20-40 6.0E-04 20-40 40-100 Particle size class (µm) >100 Fig. 3. Changes (Out-In; mean±SD) in particle counts (a), surface area (b), and volume (c), when passing each biofilter mode of operation (N=6 per mode of operation), subdivided into specified size classes (< 20, 20 - 40, 40 - 100, > 100 µm). Significant differences between the two modes of biofilter operation are marked with asterisks. *** – P < 0.001; ** – P < 0.01; * – P < 0.05; N.S. – P > 0.05. 113 Chemical Oxygen Demand (mg O2/L) 35 a 25 a b a 20 15 10 5 0 617 Biochemical Oxygen Demand (mg O2/L) 7 System Outlet FBB Outlet MBB System Outlet FBB 5 4 System Outlet FBB Outlet MBB particulate BOD5 dissolved BOD5 b 6 a System Outlet MBB a Outlet MBB a b a a 3 2 1 0 618 619 620 621 622 623 624 625 particulate COD dissolved COD a 30 System Outlet FBB Outlet MBB System Outlet FBB System Outlet FBB Outlet MBB System Experimental phases Fig. 4. Concentration (mean ±SD) of particulate and dissolved a) chemical oxygen demand (COD); and b) biochemical oxygen demand (BOD5) in the system and at the biofilter outlets per experimental phase (as described in Table 1). Data on white background represent data collected during operation with all filters in operation; data on dark grey represents data collected when only FBB was connected to the system; data on light grey background represents data collected when only MBB was connected to the system. System columns (particulate + dissolved = total) with different superscript letters are significant at p<0.05. 114 Total Ammonia Nitrogen (mg N/L) 0.35 0.30 System a Outlet FBB 0.25 Outlet MBB b a,b 0.20 a 0.15 a 0.10 0.05 0.00 All filters 626 0.30 System b FBB only Outlet FBB All filters Outlet MBB 0.25 Nitrite-N (mg N/L) MBB only b 0.20 0.15 a a All filters FBB only a 0.10 0.05 0.00 627 All filters MBB only 60 c Nitrate-N (mg N/L) 50 System a,b 40 Outlet FBB a Outlet MBB a,b b 30 20 10 0 All filters 628 629 630 631 FBB only All filters Experimental Phase MBB only Fig. 5. Concentration (mean ±SD) of a) Total ammonia nitrogen (TAN), b) nitrite-N, and c) nitrateN at the system level and biofilter outlets per experimental phase (as described in Table 1). System values with different superscript letters are significantly different between phases at P < 0.05. 115 116 ANNEX I Single Particle Optical Sensing (SPOS) measuring method and statistical examination 117 118 Single Particle Optical Sensing (SPOS) measuring method The AccuSizer TM780 SIS filter can be operated with a 0.5 – 400 µm or a 2 – 1000 µm filters. The statistical representation of the particles below 2 – 5 µm in all data sets is not sufficiently accurate to demonstrate reproducibility. Therefore, a cutoff point is usually determined at this mark. The measuring method of the AccuSizer is based on the Single Particle Optical Sensing (SPOS) method, or the light scattering pro-files of single particles passing through a narrow tube that leads to an illuminated area (Fig. A1). The shadow size generated by each particle in stated zone creates a correspondent change in current voltage measured by the sensor, which then relates the electrical pulse to particle size. The PSD of a given sample is then generated by the program using a standard calibration curve constructed with a set of uniform particles of known diameters. Fig. A1. Single particle optical sensing (SPOS) measuring method and auto-dilution system of the AccuSizer TM780 SIS. Single particles pass through a narrow tube that leads to an illuminated area, where they create a shadow. The shadow generates a current voltage change in the sensor, that related the intensity to the size of the particle. When particle concentration is too high at specific diameter intervals, the auto-dilution system dilutes the sample with a non-conductive solution. Before any analysis takes place, the particle counter collecting tube needs to be washed by running the machine twice: once with a solution of soap and milli-Q water, and a subsequent run with only milli-Q water. Every sample needs to be gently agitated (with a glass rod or shaking) just before measuring. The program reads each sample three times, presenting an average of the last two readings. Between each measurement, the sampler tube needs to be externally and internally rinsed with milli-Q water. It is recommended that no longer than 30 min should pass between sample collection and sample measurement with the Accusizer. The data may be processed and analyzed in Microsoft Office Excel. Further explanation on processing of the data acquired through measurements with the AccuSizer, can be found in Fernandes et al. (2014) and Fernandes et al. (2015) – paper I and paper II, respectively. 119 Reproducibility and statistical examination of the SPOS method To produce data with the AccuSizer, it is generally recommended that standard curves be plotted. This is achieved through the use of Certified Size Standard Polymer microspheres in water (Duke Scientific Corporation, Palo Alto, USA) with diameters ranging 1 – 650 µm. Table A1and fig. A2 show the theoretical and the observed details of certain standards (5, 10, 50 and 200 µm) used for the calibration of the machine. The data shows a high reproducibility between observed vs. theoretical mean diameter of the microspheres measured with the AccuSizer. Table A1. Labelled characteristics of the standards (Certified Size Standard Polymer microspheres in water) utilized to produce standard measuring curves in the AccuSizer), against the observed mean and standard deviation. Lot No. Cat. No. Nom. Diam. (µm) Mean Diam. (µm) Mean Std. Dev. (µm) Size Dist. CV (µm) (%) Obs. Mean (µm) Obs. Std. Dev. (µm) 23352 23341 23344 28342 4205 4210 4250 4320A 5.0 10 50 200 5.130 10.15 49.8 200 0.034 0.06 0.8 4.0 0.05 0.10 0.7 9.0 5.11 10.28 50.11 198.8 0.205 0.068 1.33 2.7 1.0% 1.0% 1.4% 4.5% Frequency of microspheres meashued at each size detector channel (%) 70% milli-Q 60% 5µm 50% 10µm 20µm 40% 50µm 30% 20% 10% 0% 0 10 20 30 40 50 60 70 80 90 100 Particle size (µm) Fig. A2 – Frequency of standard microspheres measured in each size category with the aid of four size standards (5, 10, 20, 50 µm) and milli-Q water, in the AccuSizer. Below 5 µm in size, the data is not reproducible due to a low capacity of the machine (failure with the auto-dilution system) in dealing with high particle concentrations at narrow diameters. The output (left-side of fig. A3) shows a high discrepancy in particle concentration below 5 µm for true triplicate samples. The same type of discrepancy occurs above 150-200 µm, where patchiness and concentration of particles in this size range decrease the reproducibility of true triplicates. The latter is a characteristic of the low sampling volume inability to detect specific particles that are sparsely distributed in a water sample. 120 Particle concentration (no. particles/mL) 1000 111 100 112 113 10 1 0.1 0 50 100 150 200 Particle size (µm) Fig. A3. Average (blue line) profile of true triplicate samples measured in the AccuSizer. The AccuSizer is not able to distinguish particles below 2 – 5 µm when particle concentration is too high. Moreover, due to the scarcity of large particles in small volumes, there is a large error fraction in the measurement of particles with diameters above 150 – 200 µm. 121 122
