Coldwater RAS in an Arctic charr farm in Northern Norway

Aquacultural Engineering 41 (2009) 114–121 Contents lists available at ScienceDirect Aquacultural Engineering journal homepage: www.elsevier.com/locate/aqua-online Coldwater RAS in an Arctic charr farm in Northern Norway Steinar Skybakmoen a,*, Sten Ivar Siikavuopio b, Bjørn-Steinar Sæther b a b OppdrettsTeknologi (Fishfarming Technology), Nordslettveien 177, N-7038 Trondheim, Norway Nofima Marin, N-9291 Tromsø, Norway A R T I C L E I N F O A B S T R A C T Keywords: Coldwater Recirculating Water quality Water quality variations Use of coldwater recirculating aquaculture systems (RAS) are still very rare in Norway, and only two farms are producing Arctic charr. This project took place in one of these commercial Arctic charr farms; Villmarksfisk AS in Bardu, Northern Norway. The farm gets its make-up water from ground water that holds 5 8C year around. Temperature in the rearing water varies between 7.5 8C (‘‘low’’) to 12 8C (‘‘high’’) through the year. The biological filter in the RAS seems to work stable at both ‘‘low’’ and ‘‘high’’ temperatures, including after incidents when feeding has been stopped for a day and started on top again the day after. Such an extreme change in loading was measured as a 70% increase in TAN concentration, with only minor changes in nitrite levels recorded. The biofilter also kept the nitrite stable and low in spite of diurnal variation in TAN excretion at normal feeding regimes (every day in three periods at day-time). The farming concept is to stock the farm with wild caught juvenile fish for on-growing to market size fish (0.75–1.00 kg). A drop in growth rate during early autumn has been a main concern for the farm. This may reflect a seasonal shift in growth potential, sometimes referred to as ‘‘autumn depression’’. Interestingly, there is little sign of seasonal changes in the growth of hatchery-produced fish tested in the farm. Sampling of water quality through the seasons in tanks holding fish undergoing such growth depressions indicate that TAN excretion is much higher per kg feed used in the wintertime than in the springtime. This observation corresponds with the lack of weight gain during wintertime despite that the fish is feeding. Thus, feed conversion calculations indicate that feed utilization also varies with season reaching its nadir during this period. Both challenges concerning RAS in cold water and strongly reduced growth in autumn and winter time, have been investigated from 2005 to 2007 in a project financed by the Norwegian research council and the partners in the agricultural framework agreement. ß 2009 Elsevier B.V. All rights reserved. 1. Introduction Arctic charr (Salvelinus alpinus L.) is the northernmost distributed freshwater fish and is considered the most coldadapted species within the salmonid family (Johnson, 1980). The growth of Arctic charr populations is variable under natural conditions (Johnson, 1980; Sæther et al., 1996). Arctic charr tolerate high-density culture conditions, have an excellent fillet yield, are amenable to niche marketing, and are suitable for production within super-intensive recirculating systems (Jobling, 1987; Johnston, 2002; Summerfelt et al., 2004a,b). The company Villmarksfisk AS in Northern Norway (Bardu community in Troms county) operate an on growing (grow out) farm based on wild caught Arctic charr from lake Altevatn situated approximately 40 km from the farm (Siikavuopio et al., 2009a). The * Corresponding author. Tel.: +47 4804 7204. E-mail address: [email protected] (S. Skybakmoen). 0144-8609/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.aquaeng.2009.06.007 farm consists of a fully insulated building with two separate compartments each of 200 m3 rearing volume (four tanks each 50 m3; 45 m3 effective). Each compartment has their own recirculation system and both utilize the same source for makeup water; ground water from a borehole close to the building. Effluent water is discharged back to the ground in an infiltration system. A flow chart for the farm is shown in Figs. 1–3 show a simplified sketch of the farm and a picture inside the farm. The fish rearing installation and recirculation system were in place in 2002 and in operation from 2003. First regular year of operation was 2004. 2. Technical installations The rearing tanks are constructed as octagonal tanks where four tanks are put together in a square to share the inner walls. The walls are being built of vertical aluminium profiles on a concrete bottom. The inside of the tank walls are covered with fibre glass sheets which are glued together in the corners by sheets of S. Skybakmoen et al. / Aquacultural Engineering 41 (2009) 114–121 115 Fig. 1. Process flow chart for Arctic charr farm in Northern Norway. This flow sheet represents one of two identical production systems (Compartment A and B). polypropylene with a back-side of polyester. Each tank has a particle trap at the centre of the tank bottom with a sludge collector outside the tank that also allows inspection of the sludge production and feed loss (EcoTrap 250 with 70 L sludge collector; main outlet pipe dimension and outlet screen is Ø250 mm reduced to pipe Ø200 just after the trap; particle hose to the sludge collector is Ø50 mm). The water then passes through a mechanical filter with 60 mm openings (Hydrotech HDF 1202-1H, 0.37 kW). From the mechanical filter, water flows into a common sump for aeration, trickling biofilter and sump for both pumps lifting water to the top of the trickling biofilter and for the pumps lifting water back to the tanks. This sump is made of polyethylene plates welded together by extruder to a rectangular unit with a length of 9 m, water depth of 0.8 m and width of 1.4 m, total water volume of 10 m3. At the bottom of the common sump a number of 60 diffusors provide aeration/CO2-removal (Nopon MKP 600 Tube diffuser, Finland). These are fed by a blower (Venture HPB-200-XX-500T, 5.5 kW, Sweden). Gas to water ratio is 5:1 and the degassed CO2 is removed from the compartment through an outer wall ventilation channel of 0.1 m2 situated at floor level. Two pumps (Grunfos NB 80-160/177, 3 kW, Denmark) lift the water to the biofilter, installed for parallel operation and with stationary frequency (constant rpm). The trickling biofilter consist of 84 blocks each 0.55 m  0.55 m  0.55 m size (Bio-blok 200 from Exponet, Denmark). There are seven modules of blocks, in three layers and 2 by 2 Fig. 2. Simplified sketch of the farm, Villmarksfisk AS. 116 S. Skybakmoen et al. / Aquacultural Engineering 41 (2009) 114–121 Fig. 4. Flow chart for the spreadsheet. Make-up water is actually not introduced direct to the rearing tank, but in the common sump under the biofilter where CO2removal takes place. Fig. 3. From inside one of the departments. in each layer. The water is distributed from the top of the biofilter via seven rotating pipes, one for each module. These pipes turn around as water hits the top of the biofilter. When both pumps are working (normal operation), the hydraulic surface load is 472 L/ min/m2. The two recirculation pumps (Grunfos NB 100-200/100, 5.5 kW, Denmark) that lift the water back to the tanks, are installed at far end to the water inlet from the mechanical filter in the common sump. One of these pumps has a frequency controller, the other have stationary frequency. To this time, it has not been necessary to operate both pumps simultaneously. Before reaching the tanks, the water passes oxygen saturator units, one at each tank (Inline Down Flow Bubble Contactor, AquaOptima, Norway). Water is distributed into the tanks through vertical pipes (Ø160, located 25 cm from the tank wall) via 25 Ø18 mm holes. This provides a unidirectional circular water movement in the tanks enabling some control of the water velocity and thereby the fish’s swimming speed. Collected sludge from mechanical filters and sludge collectors is stored during winter time in two insulated under ground fibre glass tanks (35 m3 each). The sludge is utilized as fertiliser/soil improvement on grass fields during the summer season. as low as possible, and it is crucial that personnel operating this attend to keep it clean and effective. To this point the biofilter has been operationally very stable without any clogging problems. The diffusors at the bottom of the common sump has been in operation continuously without any cleaning and they still seems to work properly. Although there is a tendency for higher CO2-level which may be partly explained by clogged diffusors. 4. Biological experiences The farm’s production is based on wild caught fish from a nearby lake. Capture normally takes place from February until March–April, or as long as the ice is strong enough to support a safe operation. Some catching also takes place by boat during the summer. Reduced growth in late summer and during most of the autumn and winter has been a major problem. High mortality has also been experienced (Siikavuopio et al., 2009b). The most severe growth problem was experienced in a small scale highly controlled trial at the farm in 2005–2006. Specific growth rate declined dramatically during the period from July until late October. From November till January the growth increase, but is still lower than what can be expected for hatchery-produced fish (Sæther et al., 1996; Siikavuopio et al., 2009a). It has also been observed fish that apparently thrives well, based on appetite, still show poor growth rates combined with high mortality. 3. Technical experiences 5. Investigation plan 2005–2007 The main water flow through pumps (recirculation pumps) are dimensioned for 1000 L/min per tank. This flow is too high for the particle trap main outlet screen, creating a very high swirl velocity in the tank surface centre and standing waves in the tank. Normal main flow has been 350–550 L/min depending on loading. The particle trap system restrain only approximately 50% of feed loss and faeces from the fish, however, this is sufficient to get an impression of the appetite of the fish. The main problem with the particle trap is the fouling inside the trap and in the small hoses between the trap and the sludge collector. The mechanical filter plays a key role for controlling organic matter and keeping the BOD The recirculating systems at the farm showed good performance from the very beginning. A concern, however, was that there was lack of practical experience on water quality and water treatment efficiency in such systems with relatively cold water and high variation in organic loading from fish with un-known feeding capacity and excretion. It was also a concern that water quality calculations were complicated for people with mainly technical background and no comprehensive skills in water chemistry. In order to make some progress in this field, the project investigation plan was set up by a combination of: Table 1 Day and night measurements series; date and general situation at the time. Date/period Department Water temperature (8C) Feeding (kg/24 h) Light/darkness (h) Make-up water (L/min) Tank volume (m3) Total flow through (L/min) 2005 April 20–21a 2006 November 15–16 2007 September 17–18 Hall A Hall B Hall B 10.6 7.5 11.6 57.6 16.0 70.4 L/D 24/0 L/D 24/0 L/D 24/0 133 190 120 4  45 3  45 4  45 1800 1500 1800 a In 2005, the measurement was done the day after a complete feeding stop. In November 2006 and in September 2007 feeding was regular at day-time (3 or 4 feeding periods of 1–2 h duration). 117 S. Skybakmoen et al. / Aquacultural Engineering 41 (2009) 114–121 Table 2 Measured pH and alkalinity, both measured and calculated the CO2-content and calculated difference between calculated and measured CO2. Date 2006 2007 2007 2007 a b c d e f November 15 August 22 September 18 September 18 Water type Measured CO2 (mg/L) Measured pH Measured alkalinity (mg/L CaCO3) Calculated CO2 (mg/L) Groundwater Effluent water Groundwater Effluent water 13a 18b 13b 22b 7.48c 7.29d 7.42d 7.06d 180e 155e 165e 130e 11.9f 15.9f 12.5f 22.6f Difference between calculated and measured CO2 (% of measured value) 8.5 11.7 3.9 +2.7 Method: Oxyguard CO2 analyser. Method: LaMotte carbon dioxide test kit, model PCO-DR, code 7297-DR. Method: Oxyguard pH Manta. Method: Hanna instrument pH/ORP Meter HI 98150, sensor HI 1618D. Method: Merck alkalinity test 1.11109.0001. Formula: CO2 = alkalinity  10(6.3 pH).  Technical system input (as correct as possible by measurements and registrations).  Water quality measurements (as correct as possible with field test kits and methods).  Water quality calculations (as correct mass balances as possible for TAN and CO2 combined with simplified formulas for the carbonate system). All relevant information was summarized and controlled by means of a spreadsheet based on a flow chart (Fig. 4). The general idea was to use it as a calculation control system at the same time as the water quality measurement was done; establish an understanding of the importance of measured water quality parameters. Table 1 shows an overview of the intensive day and night measurements series performed at the farm. In April 2005, the measurement was done the day after a complete feeding stop. In November 2006 and in September 2007 feeding was regular at day-time (three or four feeding periods of 1–2 h duration). somewhat different: 06:00–08:00, 12:00–13:00 and 17:00– 18:00, see Fig. 6. In order to examine the TAN excretion, several calculations have been done based on measuring TAN, nitrite, nitrate, feeding and make-up water use. Not surprising, there are periods when TAN excretion is much higher than the theoretical medium value. Reasons for this can be feed loss or that the fish is not able to utilize the feed as expected. More surprising is the measurement from 2007, when the TAN excretion was much below 3.25% (2.74%). However, this can be related to the fact that the measurement was taken early autumn and that the fish still utilized the feed very well. Later in the autumn, e.g. as in November 2006, the TAN excretion recording was much higher (8.1%) than the expected value of 3.25%. It is only the measurement in springtime 2005 that indicate a TAN excretion close to the expected medium value of 3.25%. In order to decide CO2-excretion in a recirculation system, excretion from the bacterial activity in the biofilter should also be included. However, in a trickling filter the CO2-excretion in the 6. Water quality considerations and calculations In April 2005 the ground water was evaluated for total gas pressure (oxygen, CO2 and nitrogen). The water had very low oxygen saturation (47%), high content of CO2 (2200%) and was supersaturated with nitrogen (110%). The total gas pressure (97%) was close to barometric pressure. In February 2004 oxygen saturation was 57% and pH 7.24. Alkalinity was considered to be stable at 3.5 mmol/L (175 mg/L CaCO3) and CO2 was expected to be high, around 20 mg/L. Other measurements taken are set up in Table 2. From these measurements it was considered to be acceptable to use the simplified formula CO2 (mg/L) = alkalinity (mg/L CaCO3)  10(6.3 pH) for (Summerfelt, 1996) further calculations of the relationship between pH, alkalinity and CO2. 7. Ammonium and CO2-excretion As a starting point, it was assumed 32.5 g TAN excretion per kg feed (3.25%) for the feed used (Nutra Parr, protein content 50%, Skretting). Further, it was assumed that this excretion is a medium value over 24 h. More conservative estimations say 4.6% in TAN excretion (NRAC 2002), approximately 40% higher than the present start point. Our measurements in 2007 shows that the TAN content in the water was 37% higher at night-time as compared to day-time when feeding took place at day-time. Fig. 5 shows the variation in TAN content in treated water. The feeding periods were 07:00–08:00, 13:00–14:00, 16:00–17:00 and 19:00–20:00. From the measurement in 2005 the TAN content was 42% higher in the night than the medium level; however, it is important to recognise that at that day the feeding had been stopped the day before. The regular feeding times were also Fig. 5. Variation in TAN content in treated water through 24 h; from 2007. Fig. 6. Variation in TAN content in treated water through 24 h; from 2005. 118 S. Skybakmoen et al. / Aquacultural Engineering 41 (2009) 114–121 biofilter is likely to be released very quickly. For the purpose of calculations, the CO2-excretion is set to 1.3 g/g O2-consumed and the O2-consumption set to 0.35 kg oxygen per kg feed offered to the fish. 8. Water quality simulation The main reason for assessing water quality in a recirculation system is to assure that the most important water chemical parameters stays within acceptable levels, not compromising animal welfare (Colt, 2006). Also, it is important to use the measurements to supervise the functionality of the system enabling farmers to take correct action if one or more parameters have to be adjusted (Losordo et al., 2000; Losordo and Hobbs, 2000; Summerfelt, 2002; Summerfelt et al., 1993). By means of a simple spreadsheet it is possible to simulate the water quality in the rearing tank volume and in the treated water. For the actual farm, treated water is similar to the effluent water (see flow charts, Figs. 1 and 4). The accuracy of this simulation depends on knowledge of the following parameters: Main water flow (L/min). Make-up water flow (L/min). TAN excretion (% per kg feed, etc.). Removing efficiency of the water treatment processes. Make-up water quality (alkalinity, pH and CO2). Interactions between removing efficiency and water temperature.  All parameters involved in and influencing the carbonate system.       Fig. 8. Nitrification rate (TAN-removal rate) related to ammonium content in tank effluent water at Villmarksfisk AS for 2005, 2006 and 2007-series (dots). The curve in the diagram is a constructed line for preliminary dimensioning purposes. the water. At Villmarksfisk the TAN-removal rate is calculated by means of the earlier mentioned spreadsheet made for simulation of the systems functionality and water quality. Fig. 7 shows the TANremoval rate relative to temperature for the three investigation series performed in 2005, 2006 and 2007. Fig. 8 shows the TANremoval rate related to the TAN content in the outlet water from the tanks, i.e. before the biofilter. The TAN-removal rate is generally as expected, except for the series from September 2007, with the removal rate being less than expected. This may be due to high CO2 levels, but it can also be other reasons not yet examined or understood. For further information on biofilter and water quality factors affecting nitrification rate, see Chen et al. (2006). Examples:  By increasing main flow through, all excretion product concentration will be reduced as more water flow, e.g. to the CO2stripping unit hence more CO2 are being removed. This brings the pH up.  Reduced make-up water will reduce the alkalinity import to the system, and this will bring the pH down. 9. TAN-removal rate In a biological filter, bacteria convert ammonia (TAN) to nitrite (NO2) and nitrite is further converted to nitrate (NO3) (Timmons et al., 2002; Eding et al., 2006). The second step work faster than the first step and this is convenient because the acceptable limit of nitrite is much lower than for nitrate (Timmons et al., 2002). Looking at the TAN conversion separately, the TAN-removal rate is influenced strongly by the temperature and the TAN content of Fig. 7. Nitrification rate (TAN-removal rate) related to water temperature at Villmarksfisk AS for 2005, 2006 and 2007-series (dots). The lines in the diagram are a summary from different papers (Bovendeur, 1989; Eikebrokk, 1988; Rusten, 1986; Speece, 1973), see Gebauer et al. (2005) (textbook in Norwegian). 10. Operation and documentation from April 2005 Average water temperature was 10.6 8C and the feeding rate was approximately 60 kg per 24 h in compartment A where day and night measurements were performed. The purpose of the measurements was to examine how the biofilter reacts to variation in TAN excretion over a period of 24 h. In addition these measurement series should also reflect the biofilter reaction to the feeding stop the day before. The most important parameter in a recirculation system is the level of nitrite. Some have experienced sudden peaks in nitrite (e.g. in a pilot-scale farm at Villmarksfisk using submerged biofilter) resulting in instant and high mortality. The commercial scale system operating now does not give the same surprises (Fig. 9), despite sudden increases in organic load. Over a period of 12 h (3–15 h after start of feeding) the TAN content in the system increased 200%, whereas the nitrite increased only by 12%. The nitrate-producing bacteria in the biofilter had no problem to handle the increased production of nitrite as the nitrate content Fig. 9. Ammonia content strongly increased in treated water, but not the nitrite content. S. Skybakmoen et al. / Aquacultural Engineering 41 (2009) 114–121 Fig. 10. Approximately 24% increase in nitrate content the first 12 h period, next 12 h period a 15% slow down. 119 Fig. 11. Normal variation in TAN content in treated water, very little variation in nitrite. increased with 24% over the same 12 h period (Fig. 10). In the next period of 12 h (15–27 h after feeding start), the TAN content still increased but at a slower rate (67%); the nitrite content increased only 3%, and the nitrate content started to fall (15% reduction). 11. Operation and documentation from November 2006 During this period the farm struggled with very high mortality due to saprolegnia (fungal infection). Water temperature was dropped to 7.5 8C and feeding reduced. Fig. 11 shows the TAN variation during a 24 h period of the measurement series. Feeding time was 06:00–08:00, 12:00–13:00 and 17:00–18:00. The nitrite content was low and stable and apparently the biofilter handled the diurnal variation also at this low temperature. The content of nitrate in the outlet water from the tanks where on average only 6% higher than in the treated water. This indicate that the nitrification (TAN-removal) mainly occurs in the biofilter and also that there is very little suspended biofilm in the system (passive nitrification is Fig. 12. Weak increase in CO2 level in effluent water from the tanks in the afternoon, decreasing from midnight. Fig. 13. Spreadsheet for water quality simulation; situation from 2007-series. All bold figures are input/assumed values, all other figures are calculated. The assumed values for removal rates are adjusted in order to make the calculated values correspond with the measured values for TAN, CO2, alkalinity and pH. 120 S. Skybakmoen et al. / Aquacultural Engineering 41 (2009) 114–121 Fig. 17. CO2 variation day/night. Fig. 14. TAN variation day/night in effluent water from the tanks. Fig. 15. Alkalinity variation day/night. Fig. 18. TAN and nitrite variation day/night in treated water. as 0.3. On the contrary, during spring 2005 the CO2 removal rate was 0.71 and the nitrification rate was 0.64. Another reason for low nitrification rate can be high content of organic matter (BOD). This was not measured, only calculated, and can therefore be underestimated. Compartment B operates as the fish entrance area. This part of the farm has the most severe problems with fungus and periodically very low feed utilization. This situation can result in high organic load on the biofilter and, thus, reduced nitrification rate. Figs. 14–17 show the dynamics of TAN, alkalinity, pH and CO2. The CO2 level was calculated whereas the other parameters are measured. Despite some variation, the nitrification process is very stable, as indicated in Fig. 18. The nitrite content was under 0.5 mg/ L-N, which is considered to be acceptable. Fig. 16. pH variation day/night. low). The CO2 content varies over 24 h in the effluent water from the tanks (Fig. 12). The same picture appeared also in treated water, however, with less variation, indicating difficulties with removing CO2 when the CO2 content is low. 12. Operation and documentation from September 2007 During the autumn 2007, growth rates were good and mortalities low as compared to the previous year. The fish were treated with formaldehyde (1:3000) in order to reduce the saprolegnia spp. infection, and this is likely to have contributed to the improved performance. The previously mentioned simulation shows that CO2-level was high and this was attributed to reduced efficiency of the diffusors (Fig. 13). This can affect the nitrification rate in a negative way, which at this point was as low 13. Summary The biofilters at Villmarksfisk AS are traditional trickling filters; very robust filters that handle large variations in load. However, compared with other filter media trickling filters are sometimes considered as old fashioned and too space demanding. Because of the very short water retention time in the filter, a trickling filter needs a very high hydraulic flow through to maintain nitrification stability. Biomedia that is submerged all the time usually have a longer water retention time (contact time), and for these applications it is easier to obtain a lower TAN and nitrite content than in trickling filters. There are concerns, however, on how submerged biofilters responds to large variability in nutrition load, e.g. following a period with low load. Available literature dealing with, e.g. moving bed biofilters points to sudden increase in TAN load as critical events for these filters to handle (Rusten et al., 2006). The trickling biofilter at Villmarksfisk seems to react fast S. Skybakmoen et al. / Aquacultural Engineering 41 (2009) 114–121 enough at a relative high increase in TAN load, for instance such as represented by feeding stop, and the nitrification rate is satisfactory and stable at low temperature. Acknowledgements This paper is based upon work supported by the Norwegian research council, the partners in the agricultural framework agreement and Villmarksfisk AS. We want to thank Espen Haugland at Norwegian Institute for Agricultural and Environmental Research (Bioforsk Nord), Tromsø, who acted as project manager and his colleges Hallvard Jensen and Christian Uhlig. Special thanks to Dagfinn Lysne and Nils Steien at Villmarksfisk AS for all assistance and advice during the full duration of the project period. References Chen, S., Ling, J., Blancheton, J.-P., 2006. Nitrification kinetics of biofilm as affected by water quality factors. Aquaculture Engineering 34, 179–197 (Special Issue). Colt, J., 2006. Water quality requirements for reuse systems. Aquaculture Engineering 34, 143–156 (Special Issue). 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. Aquaculture Engineering 34, 234–260 (Special Issue). Gebauer, R., et al., 2005. Oppdrettsteknologi. Vannkvalitet og Vannbehandling i Lukkede Oppdretts-anlegg, 2nd edition. Tapir forlag, Trondheim, Norway, In Norwegian (Translated title: Fish farming Technology. Water quality and water treatment in closed fish farming systems). Jobling, M., 1987. Growth of Arctic charr (Salvelinus alpinus L.) under conditions of constant light and temperature. Aquaculture 60, 243–249. Johnson, L., 1980. The Arctic charr, Salvelinus alpinus. In: Balon, E.K. (Ed.), Charrs, Salmonid Fishes of the Genus Salvelinus. W. Junk, The Hague, Netherlands, 87 pp. Johnston, G., 2002. Arctic Charr Aquaculture. Blackwell Publishing, 272 pp. 121 Losordo, T.M., Hobbs, A.O., DeLong, D.P., 2000. The design and operational characteristics of the CP&L/EPRI fish barn: a demonstration of recirculating aquaculture technology. Aquaculture Engineering 22, 3–16. Losordo, T.M., Hobbs, A.O., 2000. Using computer spreadsheets for water flow and biofilter sizing in recirculating aquaculture production systems. Aquaculture Engineering 23, 95–102. Rusten, B., Eikebrokk, B., Ulgenes, Y., Lygren, E., 2006. Design and operation of the Kaldnes Moving Bed Biofilm Reactors. Aquaculture Engineering 34, 322–331 (Special Issue). Summerfelt, S.T., Wilton, G., Roberts, D., Rimmer, T., Fonkalsrud, K., 2004a. Developments in recirculating systems for Arctic char in North America. Aquaculture Engineering 30, 31–71. Summerfelt, S.T., Davidson, J.W., Waldrop, T.B., Tsukuda, S.M., Bebak-Williams, J., 2004b. A partial-reuse system for coldwater aquaculture. Aquaculture Engineering 31, 157–181. Summerfelt, S.T., 2002. Final Project Report for USDA/ARS Grant No. 59-1930-8-038, Technologies. Procedures and Economics of Cold-Water Fish Production and Effluent Treatment in Intensive Recycling Systems. The Conservation Fund’s Freshwater Institute, Shepherdstown, WV. Summerfelt, S.T., 1996. Engineering design of a water reuse system. In: Summerfelt, R.C. (Ed.), Walleye Culture Manual. NCRAC Culture Series 101. North Central Regional Aquaculture Center Publications Office, Iowa State University, Ames, pp. 277–309. Summerfelt, S.T., Hankins, J.A., Summerfelt, S.R., 1993. Modeling continuous culture with periodic stocking and selective harvesting to measure the effect on productivity and biomass capacity of fish culture systems. In: Wang, J.K. (Ed.), Techniques for Modern Aquaculture. American Society of Agricultural Engineers, Siant Joseph, Michigan, pp. 581–595. Siikavuopio, S.I., Sæther, B.-S., Skybakmoen, S., Uhlig, C., Haugland, E., 2009a. Effects of a simulated short winter period on growth in wild caught Arctic charr (Salvelinus alpinus L.) held in culture. Aquaculture 28, 431–434. Siikavuopio, S.I., Skybakmoen, S., Sæther, B.-S., 2009b. Comparative growth study of wild- and hatchery produced Arctic charr (Salvelinus alpinus L) in a coldwater recirculation system. Aquaculture Engineering 41, 122–126. Sæther, B.-S., Johnsen, H.K., Jobling, M., 1996. Seasonal changes in food consumption and growth of Arctic charr exposed to either simulated natural or a 12:12 LD Photoperiod at constant water temperatures. Journal of Fish Biology 48 (6), 1113–1122. Timmons, M.B., Ebling, J.M., Wheaton, F.W., Summerfelt, S.T., Vinci, B.J., 2002. Recirculating aquaculture systems, 2nd edition. NRAC Publication No.01-002.