- Email: [email protected]
Characterising organic matter in recirculating aquaculture systems with fluorescence EEM spectroscopy
Water Research 83 (2015) 112e120
Contents lists available at ScienceDirect
Water Research journal homepage: www.elsevier.com/locate/watres
Characterising organic matter in recirculating aquaculture systems with fluorescence EEM spectroscopy ska-Sobecka b, A.C. Hambly a, *, E. Arvin b, L.-F. Pedersen c, P.B. Pedersen c, B. Seredyn C.A. Stedmon a a b c
National Institute for Aquatic Resources, Technical University of Denmark, Kavalergården 6, DK-2920 Charlottenlund, Denmark Technical University of Denmark, Department of Environmental Engineering, DTU Environment, Miljoevej, Building 113, DK-2800 Kgs. Lyngby, Denmark 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
a b s t r a c t
Article history: Received 13 February 2015 Received in revised form 19 June 2015 Accepted 22 June 2015 Available online 25 June 2015
The potential of recirculating aquaculture systems (RAS) in the aquaculture industry is increasingly being acknowledged. Along with intensified application, the need to better characterise and understand the accumulated dissolved organic matter (DOM) within these systems increases. Mature RASs, stocked with rainbow trout and operated at steady state at four feed loadings, were analysed by dissolved organic carbon (DOC) analysis and fluorescence excitation-emission matrix (EEM) spectroscopy. The fluorescence dataset was then decomposed by PARAFAC analysis using the drEEM toolbox. This revealed that the fluorescence character of the RAS water could be represented by five components, of which four have previously been identified in fresh water, coastal marine water, wetlands and drinking water. The fluorescence components as well as the DOC showed positive correlations with feed loading, however there was considerable variation between the five fluorescence components with respect to the degree of accumulation with feed loading. The five components were found to originate from three sources: the feed; the influent tap water (groundwater); and processes related to the fish and the water treatment system. This paper details the first application of fluorescence EEM spectroscopy to assess DOM in RAS, and highlights the potential applications of this technique within future RAS management strategies. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Recirculating aquaculture systems (RAS) Fluorescence excitation-emission matrix (EEM) spectroscopy Dissolved organic matter (DOM) PARAFAC analysis
1. Introduction Developing and advancing aquaculture technology is critical, as aquaculture is playing an increasingly vital role in sustainable food production (FAO, 2014). Intensive aquaculture systems with up to hundreds of metric tonnes of biomass generate large amounts of faecal matter daily. The organic waste in these systems is linked to the feed input (amount and quality) as well as feed digestibility and utilisation (Dalsgaard and Pedersen, 2011), and measures to remove dissolved and particulate matter are therefore often applied to
Abbreviations: BOD, Biological oxygen demand; C:N, Carbon to nitrogen ratio; CFB, Cumulative feed burden; COD, Chemical oxygen demand; COD:N, COD to nitrogen ratio; DOC, Dissolved organic carbon; DOC:N, DOC to nitrogen ratio; DOM, Dissolved organic matter; EEM, Excitation-emission matrix; FDOM, Fluorescent dissolved organic matter; PARAFAC, Parallel factor analysis; RAS, Recirculating aquaculture system; RU, Raman Units; TOC, Total organic carbon; UV, Ultraviolet; UVA, Ultraviolet A; UVB, Ultraviolet B. * Corresponding author. E-mail address: [email protected] (A.C. Hambly). http://dx.doi.org/10.1016/j.watres.2015.06.037 0043-1354/© 2015 Elsevier Ltd. All rights reserved.
mitigate negative impacts on water clarity and quality. Recirculating aquaculture systems (RASs) are land-based, closed-loop systems in which fish are grown and harvested. RAS benefits potentially include stable rearing conditions, reduced risk of escapees, options for pathogen control, as well as solids/particulate removal and hence decreased environmental impacts (Martins et al., 2010). Challenges to RAS include investment and operating costs, as well as the risk of operational failure. These failures may arise from poor water quality conditions and are often associated with disease encounters and/or insufficient pathogen control (Noble and Summerfelt, 1996; Pedersen et al., 2009). Attempts at mitigating such problematic conditions are typically carried out by severe biosecurity measures (Sharrer et al., 2005; Summerfelt et al., 2009) €kior by chemical treatment often with limited success (Rintama Kinnunen et al., 2005). Despite these challenges, farming RAS at commercial scales is becoming increasingly common with more species being cultured (Dalsgaard et al., 2013) and systems becoming increasingly large. The economic impacts of RAS
A.C. Hambly et al. / Water Research 83 (2015) 112e120
underperformance and failures are substantial, and minimising these events is important for the advancement of RAS and aquaculture in general. The full potential of RAS is yet to be realised, to become profitable and to accelerate development on land, as current water quality monitoring practices have been shown to be insufficient for reliable, failure-free operation. The management and mitigation of system failures is at present commonly focussed on maintaining chemical water quality by applying conventional wastewater treatment processes for removal of potentially toxic compounds such as ammonia, as well as by maintaining a balanced microbial community. The effectiveness of these mechanical and biological treatment processes is of utmost importance to RAS viability, and fundamental treatment processes such as particle/solids removal have been shown to be heavily impacted by inherent DOC concentrations (Hem et al., 1994; Summerfelt and Vinci, 2008; Timmons and Ebeling, 2010). For example, Ling and Chen (2005) observed such effects on nitrification efficiency, where an exponential decrease in nitrification performance was recorded with the addition of DOC into the water matrix. Hu et al. (2009) observed negative impacts on nitrification efficiency with increases in carbon to inorganic nitrogen ratios, and a number of studies have recommended that generally elevated C:N ratios are to be avoided (Guerdat et al., 2011; Michaud et al., 2006). Managing microbial water quality is also an important component of RAS management, as bacteria, parasites and algae can affect fish growth and survival, and ultimately RAS output (Blancheton et al., 2013). Recently, Moestrup et al. (2014) observed instances of fish kills within two Danish marine RAS where the cause was identified to be populations of two dinoflagellate species: Pfiesteria shumwayae and Luciella masanensis. Like many heterotrophic dinoflagellate species, Pfiesteria are reported to thrive in conditions where the water body contains high organic matter content (Burkholder and Glascow, 1997; Glibert et al., 2001; Lewitus et al., 1999), which is often the case in RAS utilising long retention times and insufficient solids/organic matter removal. This suggests that the efficient and successful management of a RAS is heavily reliant on controlling the accumulation and quality of dissolved organic matter (DOM) in the system. It follows that implementing more adequate monitoring and control of the organic content of RAS waters could therefore serve to detect deviations in treatment efficiency, and as a preventative measure for RAS failure and subsequent mortality events. Current RAS monitoring methods typically include the analysis of bulk organic matter in the water, including particulate matter, and inorganic nitrogen species such as ammonia, nitrate and nitrite. Additionally, gross indicators such as chemical oxygen demand (COD) and biological oxygen demand (BOD) can be applied, however, analysis of BOD typically requires 5e7 days which introduces an inherent delay in response to events. The organic matter within RASs is derived from a range of sources, each varying in relative importance with time, and so bulk organic matter measurements alone (e.g. DOC or absorbance at 254 nm, A254) may not provide an adequate estimate of the extent to which organic matter character may be fluctuating and potentially influencing RAS performance. The fraction of DOM that is fluorescent, FDOM, and its spectral characteristics of, have been successfully utilised as quantitative and qualitative measures of DOM across a range of natural and engineered systems including, lakes and rivers (Baker, 2001, 2002; Yamashita et al., 2010), wastewater (Bridgeman et al., 2013; Hambly et al., 2010a; Reynolds and Ahmad, 1997), drinking water (Shutova et al., 2014; Stedmon et al., 2011) and marine systems (Coble, 1996; Coble et al., 1990). Different components of FDOM have been shown to be tracers of DOM sub-fractions, including
113
biorefractory, labile and photochemically-active components (Stedmon and Cory, 2014). Fluorescence excitation-emission matrix (EEM) spectroscopy has emerged as the state-of-the-art in FDOM analysis and has become increasingly common as an analytical tool for water sciences (Coble et al., 2014). EEMs are generated by recording emission spectra over a range of excitation wavelengths and combining them to form a detailed contour map of the fluorescent properties of organic matter in a water sample. Multivariate data analysis techniques, in particular parallel factor analysis (PARAFAC), have proven their abilities to reliably decompose EEMs into independently varying fluorescent components, allowing more accurate identification of independent FDOM components (Murphy et al., 2013). As such, the ability of fluorescence spectroscopy to successfully characterise aquatic systems has become increasingly evident. Fluorescence analysis is a growing research area within natural and engineered water systems, and individual components of FDOM have been shown to correlate well with BOD (Hudson et al., 2008; Reynolds and Ahmad, 1997), COD (Bridgeman et al., 2013; Lee and Ahn, 2004), ammonia, nitrates and phosphates (Baker and Inverarity, 2004), as well as microbiological indicators such as total coliforms and E. coli (Cumberland et al., 2012). These are important parameters within RAS, though currently unable to be reliably monitored online and analysed within a reactive timeframe. Fluorescence spectroscopy, however, is a fast, sensitive and non-destructive analysis technique, and hence shows great potential in its application to realtime monitoring of RAS and aquaculture in general. More specifically, these properties could enable fluorescence analysis to become a valuable, real-time monitoring tool to optimise RAS management. This study provides the first application of fluorescence EEM spectroscopy and PARAFAC analysis to characterise DOM in aquaculture systems, more specifically within a RAS. Whether changes in FDOM are consistent across all types of RAS are yet to be determined, however this study evaluates the application of the EEM-PARAFAC technique itself. The aim was to test if the technique can identify characteristic organic matter fractions from the complex organic RAS matrix, and to therefore outline the potential of using fluorescence as a sensitive monitoring parameter of RAS water. 2. Materials and methods 2.1. Experimental design Four identical 1700 L freshwater RASs (Fig. 1) were stocked with juvenile rainbow trout (Oncorhynchus mykiss) with an average weight of 131 ± 6 g and brought to steady state conditions (in terms of nitrate equilibrium) over 2 months. Four daily feed inputs were applied within the study (125, 250, 375 or 500 g). The daily water renewal of each RAS was 80 L, corresponding to relative water renewal rates of between 160 and 640 L/kg feed, and the cumulative feed burden (CFB) ranged from 1.6 to 6.3 kg feed per m3 water renewal. The stocking density was adjusted to the feed given; initial biomasses were 14, 28, 42 and 56 kg/m3 at the four levels. As a consequence of the experimental design with fixed daily feed allocation the corresponding feeding ratio decreased from 1.8% to 1.1% of fish biomass per day during the experimental period due to biomass gain. An experimental fish feed consisting of 44% protein and 30% lipid (Pedersen et al., 2012) (Table 1) was utilised and allocated daily from 9:00 to 15:00 h by belt feeders. Forty litres of RAS water was drained daily from the swirl separator in each RAS, and replaced with 80 L of non-chlorinated tap water. The 40 L excess was discharged by overflow and evaporation. This corresponded to a water exchange
114
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Fig. 1. Schematic of the experimental recirculating aquaculture systems (RAS). Water circulation in the RAS is represented with thick arrows, dotted lines represent system inlet and water outlet; P indicates the position of the circulation pump.
Table 1 Ingredients and proximate composition of the experimental diet. Ingredient (as-fed basis)
g/kg
Fish meal LT 92 Spray blood red cells Wheat Wheat glluten Pea protein 75 Field beans Sunflower cake Soya cake Fish oil 60 Premixes Proximate composition
320 100 97 60 60 54 36 24 244 1 %
Proteina Fatb NFEc Ashd Moistured
44.0 29.9 11.3 6.1 8.7
a b c d
ISO 5983-2, 2005. Bligh & Dyer, 1959. NFE calculated as: dry mattereproteinelipideash. NMKL 23, 1991.
rate of 4.7% of system volume per day and a hydraulic retention time of 3 weeks. The water temperature was kept constant at 18.0 C by the use of heating elements within the RAS sump. Oxygen concentrations ranged from 7.2 to 8.5 mg/L, and were maintained by aeration and/or addition of oxygen with diffusers within the rearing tank. The pH was kept between 7.2 and 7.4, and was adjusted daily by the addition of sodium bicarbonate (approx. 20% w/w of added feed) to compensate for alkalinity loss from the nitrification processes of the biofilters. The systems were inspected daily and there were typically no uneaten pellets found. In the case of any uneaten pellets, these were removed by the swirl separator (Fig. 1). A feed extraction study was also performed by dissolving 10 g of the fish feed in 100 mL of MilliQ water. The mixture was then shaken at room temperature for 24 h before sampling.
2.2. Sampling Water samples for fluorescence EEM and DOC were collected hourly throughout the day and in the morning on the following day in order to study organic matter quantity and quality at different feed inputs. The water samples were taken from each RAS at the rearing tank, separator and after the biofilters. Feed extractions were sampled after 24 h and filtered through a glass filter. Serial dilutions were then carried out to enable sample analysis by DOC and fluorescence EEM spectroscopy. 2.3. Analyses All aquaculture water samples were filtered through 0.45 mm cellulose acetate filters (Millipore) and collected in acid washed and pre-combusted 40 mL glass vials with Teflon-lined caps. DOC analyses were performed with a Shimadzu TOC-V WP analyser with an ASI-V autosampler, calibrated with a six-point calibration curve using potassium hydrogen phthalate. Fluorescence EEMs were measured in a 1 cm quartz cuvette (4 mL volume) using a Varian Cary Eclipse Fluorescence Spectrophotometer (Varian, UK). Fluorescence EEMs were measured for excitation wavelengths of lex ¼ 200e400 nm at 5 nm increments across an emission range of lem ¼ 280e500 nm at 2 nm intervals. Excitation and emission slit widths were set to 5 nm, with a photomultiplier tube (PMT) voltage of 800 V. UV absorption measurements were obtained on a Varian Cary 50 Bio UVevisible Spectrophotometer (Varian, UK) in a 1 cm quartz cuvette and UVevisible spectra were recorded from 240 to 700 nm with slit width of 0.5 nm. 2.4. EEM post-processing All raw EEM processing was carried out using the drEEM toolbox (Murphy et al., 2013) within Matlab software (The MathWorks Inc.). The EEM correction process consisted of a blank EEM subtraction and scatter line removal, the application of excitation and emission correction factors, correction for inner filter effects, and normalisation to Raman Units (RU) before PARAFAC analysis was carried out.
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Blank EEMs were acquired from MilliQ water samples, and excitation and emission correction factors were obtained using Rhodamine B and a ground quartz diffuser. Raman and Rayleigh scatter peaks were removed from spectra within the drEEM toolbox, as well as wavelengths below lex ¼ 250 nm and lem ¼ 300 nm to limit spectrally noisy areas of the matrix. Inner filter effect correction factors were derived from the absorption measurements and Raman normalisation factors were calculated using MilliQ blank EEMs at lex ¼ 350 nm (Lawaetz and Stedmon, 2009). The resulting corrected and normalised EEM dataset was then subjected to parallel factor analysis (PARAFAC) with the drEEM toolbox (n ¼ 148). The least squares models for 3 to 7 component models were tested by running 30 iterations, with non-negativity and a convergence criteria of 1 108, and the most appropriate number of PARAFAC components was found using a combination of split half and residual analysis according to the recommendations in Murphy et al. (2013). 3. Results 3.1. Fluorescence characteristics and identified components Fig. 2 shows the general fluorescence characteristics of the DOM in the four types of RAS waters, the drinking water source used to fill the RAS, and the pure water extraction of DOM from the feed utilised in the experiment. The fluorescence signal of RAS samples were characterised by three regions: (1) a broad peak in the region of lex < 250 nm and lem ¼ 400e500 nm; (2) a clear peak at lex/ em ¼ 340/430 nm; and (3) a UV peak at lex/em 280/340 nm. A similar fluorescence signal was present in the source tap water but at much lower intensity, particularly in the UV emission region. The DOM extracted from the aquaculture feed was characterised by a high fluorescence intensity in the UV region, with a UVB peak at lex/ em ¼ 280/300 nm, overlapping with a broader UVA peak at lex/ em ¼ 280/340 nm. The PARAFAC analysis indicated that the fluorescence signal could be mathematically decomposed into five independently varying fractions (Fig. 3). Components 1 to 4 were dominated by visible emission region (>400 nm), whereas component 5 was dominated by a peak in the UVA emission region (300e380 nm). 3.2. Feed loading effects on DOC and FDOM components A linear relationship between DOC and feed input was observed, where every 125 g/day feed increase was accompanied by a DOC increase of approximately 5 mg/L (Fig. 4). Across each of the four feed input levels, samples displayed very similar fluorescence character, while differing in overall fluorescence intensities. Fluorescence intensities increased with increased feed input (Fig. 5), which is in general agreement with the feed relationship for DOC concentrations observed (Fig. 4). Although overall DOC was observed to be linear in relation to feed input, not all fluorescence components adhered to the same relationship (Fig. 5). Components F1, F2 and F4 showed a positive linear correlation (R > 0.99 in all cases) to feed input with fluorescence intercepts close to that of the tap water used to fill the RAS (Fig. 6). Component F5 differed by having a non-linear positive correlation with feed input, which could be modelled with a quadratic polynomial. Compared to the other components, F3 showed a marginal increase with increased feed input, with an intercept (analogous to a 0.0 g feed/day) of 1.5 RU (Raman Units). As DOC showed a strong linear correlation to feed input, the relationships between FDOM and DOC were comparable to that between FDOM and feed input (Fig. 5).
115
4. Discussion 4.1. Identified fluorescence components Similar PARAFAC components have been observed and described throughout the literature of DOM fluorescence, and a number of matching peaks were found from a review of literature and the OpenFluor database of PARAFAC aquatic fluorescence components (Murphy et al., 2014). The origins and properties of corresponding OpenFluor components are shown in Table 2. The peaks display similarities (minimum spectral similarity score of 0.95) to those within studies from a variety of aquatic sources, including: (1) marine water; (2) freshwater/drinking water; and (3) freshwater impacted by treated wastewater. This is consistent with the typical elements of the RASs investigated within this study: (1) the filling of the system with freshwater/drinking water; (2) a marine organism component; and (3) the system contains a biological treatment train through which the aquaculture water is recycled. These consistencies support the hypothesis that many organic matter fluorescence signals identified across contrasting aquatic systems essentially represent common ubiquitous degradation intermediates. Interestingly, component F5 contains peaks that have also been commonly associated with microbial activity (Hudson et al., 2008) as well as being correlated with wastewater and recycled water sources (Hambly et al., 2010a, 2010b; Hudson ska-Sobecka et al., 2011). Although no direct et al., 2007; Seredyn matches were found for component F1 within the OpenFluor database, the broad peak which dominates it (lex/em ¼ 360/450) shares distinct similarity with fluorophores commonly referred to as “humic-like” or “Peak C” (Coble, 1996). 4.2. Identification of fluorescence sources All of the PARAFAC components were observed to correlate to feed input, to varying degrees. This correlation could arise either: (a) directly e from dissolution of feed or undigested parts thereof; or (b) indirectly e from biological fish processes and/or processes in the biofilters related to increased feeding. To discern between the two, a feed dissolution study applying 10 g feed/100 mL (100 kg feed/m3) was undertaken to isolate the fluorescence signal arising from direct feed dissolution within the RAS. Fig. 6 shows the fluorescence loadings from the five components (F1eF5) for each sample type and it can be seen that by far the greatest contributor to the dissolved feed fluorescence is component F5, which accounts for 82% of the total FDOM within sample. This information, in combination with the known concentrations of the feed dissolution samples (100 kg feed/m3 water), was used to perform a conservative calculation for the direct contribution of feed dissolution to the FDOM of the RAS water. The grinding of the feed with a mortar and pestle likely enhanced dissolution compared to normal conditions, but the tests provide a first estimate of the effects of feed dissolution on the FDOM signal. If the observed fluorescence of a component is higher than the calculated dissolution value, there is a production of this fluorescence component within the RAS, whether it be from fish digestion or the treatment system. The calculation was also made under the conservative assumption that the five components were not decomposed in the RAS. The results are illustrated in Fig. 7 which reveals that the fluorescence components in the RAS water are related, to various degrees, to three sources: 1) the feed (F5); 2) the influent tap water (groundwater, F3); and 3) the fish and treatment processes (F1, F2, F4). The dissolved feed contributed a large proportion of component F5 to the calculated fluorescence loadings. In the RAS water, the observed F5 loadings were much lower than the calculated values, indicating degradation or removal of this fraction. Components F1
116
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Fig. 2. Typical EEMs of RAS waters at daily feed inputs of: (a) 125 g/day; (b) 250 g/day; (c) 375 g/day; and (d) 500 g/day; (e) feed dissolved in MilliQ water; and (f) tap water used to fill RAS. N.B. Z-axis (fluorescence intensity) scales differ between samples to allow comparison of fluorescence character.
A.C. Hambly et al. / Water Research 83 (2015) 112e120
117
Fig. 3. Contour plots of individual components from the 5-component PARAFAC model (generated from 148 samples from all four feed levels).
Fig. 4. DOC concentrations (mean ± s.d.) as a function of feed input.
and F2 can be explained by a mixture of the feed input, tap water and FDOM produced in the RAS, and component F4 can be explained predominantly by FDOM produced within the RAS. Finally, component F3 arises for the most part from tap water and the feed dissolution, with the tap water as the predominant source. Fig. 6 shows that the main constituent of tap water FDOM is component F3, which accounts for over 45% of fluorescence within the tap water samples and is consistent with the relationship observed between component F3 and feed input (Fig. 5). As the feed input has increased, the amount of tap water within the system has essentially stayed constant between RAS's (with a small relative increase due to daily fill of make-up water). It must be emphasised that the calculations above on the contribution of the feed to the fluorescence of the RAS water samples involves some uncertainty due to the aforementioned dissolution method. Another uncertainty is related to the assumption that the five components were not decomposed in RAS. Though this was unlikely to be the case for component F5, it may be
Fig. 5. Fluorescence component loadings (mean ± s.d.) at the four feeding levels, calculated from all samples in each group over the entire sampling period. Trendlines for F1eF4 were fitted by linear regression, whereas the trendline for F5 was fitted as a quadratic polynomial.
that some of the other components were partly decomposed (e.g. from uptake and decomposition by the fish), whilst being simultaneously produced by the treatment system. Whilst it is possible to ascertain the net result of such processes within this study, it is not possible to determine the extent to which components are both decomposed and produced. Despite the clear degradation of component F5 within the RAS (Fig. 7), the observed accumulation at higher feed loadings (Fig. 5) suggests its potential importance to the system. The fluorescence exhibited by component F5 is commonly referred to as “peak T” fluorescence (Coble, 1996) and often associated with elevated
118
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Fig. 7. Fluorescence loadings (mean ± s.d.) of each PARAFAC component for theoretical values (sum of contribution from dissolved feed and tap water signals) and observed values, at each of the four feed inputs.
Fig. 6. Fluorescence loadings (mean ± s.d.) of each PARAFAC component for RAS water (at feed inputs of 125, 250, 375 and 500 g/day), tap water and dissolved feed (displayed as the theoretical 125 g/day equivalent).
microbial activity (Hudson et al., 2008). Degradation studies have implied that this fluorescence component represents microbial bioavailable organic matter, which can be produced by, or consumed by bacteria (Stedmon and Cory, 2014). The accumulation of this signal at high feed loadings could therefore indicate the potential for a significant microbiological population change within the RAS brought about by the excess organic matter. In addition to excess feed input, increased DOM concentrations within a RAS could come from a number of other potential sources. Inadequate removal of unconsumed feed and faecal matter could result in a build-up of organic matter within a RAS, in addition to changed feed constituency, feed digestibility, recirculation rates and water exchange volumes. Unconsumed feed and faecal matter was manually removed by a swirl separator flushed every day during the current study, and recirculation rates, water exchange and feed constituency were kept constant. Overloading of the RAS treatment train may also lead to an increase in nutrients and organic matter. COD:N and DOC:N ratios have been directly linked to nitrification performance within a number of previous studies, where increases in these ratio have seen concurrent decreases in nitrification efficiency (Ling and Chen, 2005; Hu et al., 2009). Increased C:N ratio tend to favour growth of (opportunistic) heterotrophic bacteria over slower growing (stable) nitrifying bacteria (Attramadal et al., 2014). As aquaculture looks to
potentially employ more advanced water treatment processes such as UV and ozone within recirculating systems (Summerfelt and Hochheimer, 1997; Summerfelt et al., 2009) implementing a fast and reliable method for the monitoring of organic matter will become even more important. These processes have been shown to have varied treatment efficacy on organic matter fractions (HessErga et al., 2008; Tranvik and Bertilsson, 2001; Win, 2000). Recent RAS studies are suggesting that fish farms operated at high feed loadings are at risk of fish kills in cases of insufficient solids removal and organic material accumulation, potentially promoting toxic microalgae populations (Moestrup et al., 2014). Increased levels of organic matter in the system combined with reduced treatment process efficiencies can rapidly and detrimentally impact system and hence fish performance. As highlighted, the RAS in this study were operated under fixed conditions with a single fish species and feed type, and the DOM quantity and quality is likely to be affected by fish type, fish density, feed type, treatment train and operational conditions. However, the ability to of the EEM-PARAFAC method to identify and then monitor specific fluorescence components shows potential for application across a wide range of RAS operations. Fluorescence spectroscopy shows great potential in monitoring the quality and quantity of organic material in RASs, due to its high sensitivity and proven online monitoring ability. Ultimately, fluorescence monitoring could therefore give RAS operators the ability to react quickly to small changes in organic matter, to better manage the systems as a whole and potentially avert detrimental operational events.
Table 2 Fluorescence peak wavelength pairs for RAS 5-component PARAFAC model and literature comparisons. Component Peak(s)
lexcitation (nm)
lemission (nm)
Source(s) description from OpenFluor study matches
F1 F2 F3 F4 F5
360 325 250 270 (410) 285
450 410 440 480 330
(None over 0.95 spectral similarity) Freshwater lakesa; saline lakesb; rivers, estuaries and coastal marine watersc,d,e Drinking waterf; river plume and coastal marine watersc Freshwater lakesg Marine ballast waterh; agricultural-impacted subtropical wetlandsi; anthropogenic-impacted tropical riversj
a b c d e f g h i j
(Kothawala et al., 2012). (Osburn et al., 2011). (Kowalczuk et al., 2009). (Stedmon et al., 2003). (Cawley et al., 2012). (Shutova et al., 2014). (Kothawala et al., 2014). (Murphy et al., 2006). (Yamashita et al., 2010b). (Yamashita et al., 2010a).
A.C. Hambly et al. / Water Research 83 (2015) 112e120
5. Conclusions * PARAFAC analysis of EEMs shows a 5-component model to describe changing fluorescence components in an RAS stocked with rainbow trout: a) three components, F1, F2 and F4, which linearly correlated to feed input; (b) one component, F5, which correlated to feed input and showed accumulation with increased feed inputs; and (c) one component, F3, which correlated to tap water. * Four out of the five fluorescence components have previously been identified within the OpenFluor spectral database, from within river water, marine coastal water, drinking water and wetlands studies. * The five fluorescence components originate from three sources: 1) the feed input; 2) the influent tap water (groundwater); and 3) organic matter produced by the fish and treatment processes. * There is a considerable difference between the components as to the degree they originate from these three sources. F5 mainly originates from the fish feed, F4 mainly originates from organic matter produced in the RAS, whereas F1 and F2 originate both from the feed as well as being produced within the RAS. Finally, F3 predominantly represents the contribution of organic matter in the tap water used to fill the system. * Although DOC increases linearly with increased feed loading, PARAFAC analysis shows fluorescence components are not all linearly increasing. Measuring FDOM may offer a more specific approach to monitor the accumulation of bioavailable organic matter in a system, as opposed to bulk DOC. Specific FDOM components could potentially also be correlated to predict DOC values. * Future research should include the effect of fish species, fish density, feed composition, treatment types, salinity, and water exchange on FDOM. * Fluorescence is a promising parameter for monitoring the organic content of RASs, as it can be easily developed for online sensor measurements. Such a fluorescence-based early warning tool could contribute to the successful operation of RASs with higher feed loadings and lower water renewal requirements than is currently achievable, and ultimately advance the development and profitability of sustainable RAS industries. Acknowledgements This work was funded by the Danish Agency for Spatial and Environmental Planning, Ministry of the Environment (Ref. BLS403-00043 AquaFingerprint) and FP7-PEOPLE-2013-IIF, Project number 626147, FAMoRAS e Fluorescence Analysis and Monitoring of Recirculating Aquaculture Systems (Marie Curie Actions eInternational Incoming Fellowships. Adam Hambly). In addition the study was also financed by The Technical University of Denmark (DTU) through a strategic research programme between DTU Environment and DTU Aqua. References 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, 483e490. Baker, A., 2001. Fluorescence excitationemission matrix characterization of some sewage-impacted rivers. Environ. Sci. Technol. 35, 948e953. Baker, A., 2002. Fluorescence excitationemission matrix characterization of river waters impacted by a tissue mill effluent. Environ. Sci. Technol. 36, 1377e1382. Baker, A., Inverarity, R., 2004. Protein-like fluorescence intensity as a possible tool for determining river water quality. Hydrol. Process 18, 2927e2945. 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, 30e39. Bridgeman, J., Baker, A., Carliell-Marquet, C., Carstea, E., 2013. Determination of
119
changes in wastewater quality through a treatment works using fluorescence spectroscopy. Environ. Technol. 34, 3069e3077. Burkholder, J.M., Glascow, H.B., 1997. Pfiesteria piscicida and other Pfiesreria-like dinoflagellates: behavior, impacts, and environmental controls. Limnol. Oceanogr. 42, 1052e1075. Cawley, K.M., Butler, K.D., Aiken, G.R., Larsen, L.G., Huntington, T.G., McKnight, D.M., 2012. Identifying fluorescent pulp mill effluent in the Gulf of Maine and its watershed. Mar. Pollut. Bull. 64, 1678e1687. Coble, P.G., 1996. Characterization of marine and terrestrial DOM in seawater using excitation-emission matrix spectroscopy. Mar. Chem. 51, 325e346. Coble, P.G., Green, S.A., Blough, N.V., Gagosian, R.B., 1990. Characterization of dissolved organic matter in the Black Sea by fluorescence spectroscopy. Nature 348, 432e435. Coble, P.G., Lead, J., Baker, A., Reynolds, D.M., Spencer, R.M., 2014. Aquatic Organic Matter Fluorescence. Cambridge University Press, New York. Cumberland, S., Bridgeman, J., Baker, A., Sterling, M., Ward, D., 2012. Fluorescence spectroscopy as a tool for determining microbial quality in potable water applications. Environ. Technol. 33, 687e693. Dalsgaard, J., Pedersen, P.B., 2011. Solid and suspended/dissolved waste (N, P, O) from rainbow trout (Oncorynchus mykiss). Aquaculture 313, 92e99. 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, 2e13. FAO, 2014. The State of World Fisheries and Aquaculture 2014-Opportunities and Challenges. Food and Agriculture Organisation of the United Nations, Rome, Italy. Glibert, P.M., Magnien, R., Lomas, M.W., Alexander, J., Fan, C., Haramoto, E., Trice, M., Kana, T.M., 2001. Harmful algal blooms in the chesapeake and coastal Bays of Maryland, USA: comparison of 1997, 1998, and 1999 events. Estuaries 24, 875. 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, 10e18. Hambly, A.C., Henderson, R.K., Storey, M.V., Baker, A., Stuetz, R.M., Khan, S.J., 2010a. Fluorescence monitoring at a recycled water treatment plant and associated dual distribution system e implications for cross-connection detection. Water Res. 44, 5323e5333. Hambly, A.C., Henderson, R.K., Baker, A., Stuetz, R.M., Khan, S.J., 2010b. Fluorescence monitoring for cross-connection detection in water reuse systems: Australian case studies. Water Sci. Technol. 61, 155. Hem, L.J., Rusten, B., Ødegaard, H., 1994. Nitrification in a moving bed biofilm reactor. Water Res. 28, 1425e1433. Hess-Erga, O., Kihle Attramadal, K., Vadstein, O., 2008. Biotic and abiotic particles protect marine heterotrophic bacteria during UV and ozone disinfection. Aquat. Biol. 4, 147e154. Hu, J., Li, D., Liu, Q., Tao, Y., He, X., Wang, X., Li, X., Gao, P., 2009. Effect of organic carbon on nitrification efficiency and community composition of nitrifying biofilms. J. Environ. Sci. 21, 387e394. Hudson, N., Baker, A., Reynolds, D., 2007. Fluorescence analysis of dissolved organic matter in natural, waste and polluted watersda review. River Res. Appl. 23, 631e649. Hudson, N., Baker, A., Ward, D., Reynolds, D.M., Brunsdon, C., Carliell-Marquet, C., Browning, S., 2008. Can fluorescence spectrometry be used as a surrogate for the Biochemical Oxygen Demand (BOD) test in water quality assessment? An Ex. South West Engl. Sci. Total Environ 391, 149e158. €hler, S.J., Kothawala, D.N., Stedmon, C.A., Müller, R.A., Weyhenmeyer, G.A., Ko Tranvik, L.J., 2014. Controls of dissolved organic matter quality: evidence from a large-scale boreal lake survey. Glob. Change Biol. 20, 1101e1114. Kothawala, D.N., von Wachenfeldt, E., Koehler, B., Tranvik, L.J., 2012. Selective loss and preservation of lake water dissolved organic matter fluorescence during long-term dark incubations. Sci. Total Environ. 433, 238e246. Kowalczuk, P., Durako, M.J., Young, H., Kahn, A.E., Cooper, W.J., Gonsior, M., 2009. Characterization of dissolved organic matter fluorescence in the South Atlantic Bight with use of PARAFAC model: Interannual variability. Mar. Chem. 113, 182e196. Lawaetz, A.J., Stedmon, C.A., 2009. Fluorescence intensity calibration using the Raman scatter peak of water. Appl. Spectrosc. 63, 936e940. Lee, S., Ahn, K.H., 2004. Monitoring of COD as an organic indicator in waste water and treated effluent by fluorescence excitation-emission (FEEM) matrix characterization. Water Sci. Technol. J. Int. Assoc. Water Pollut. Res. 50, 57e63. Lewitus, A.J., Willis, B.M., Hayes, K.C., Burkholder, J.M., Glasgow Jr., H.B., Glibert, P.M., Burke, M.K., 1999. Mixotrophy and nitrogen uptake by pfiesteria piscicida (dinophyceae). J. Phycol. 35, 1430e1437. Ling, J., Chen, S., 2005. Impact of organic carbon on nitrification performance of different biofilters. Aquac. Eng. 33, 150e162. Martins, C.I.M., Eding, E.H., Verdegem, M.C.J., Heinsbroek, L.T.N., Schneider, O., Blancheton, J.P., d'Orbcastel, E.R., Verreth, J.A.J., 2010. New developments in recirculating aquaculture systems in Europe: a perspective on environmental sustainability. Aquac. Eng. 43, 83e93. 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, 224e233. Moestrup, Ø., Hansen, G., Daugbjerg, N., Lundholm, N., Overton, J., Vestergård, M., Steenfeldt, S.J., Calado, A.J., Hansen, P.J., 2014. The dinoflagellates Pfiesteria shumwayae and Luciella masanensis cause fish kills in recirculation fish farms in Denmark. Harmful Algae 32, 33e39.
120
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Murphy, K.R., Stedmon, C.A., Graeber, D., Bro, R., 2013. Fluorescence spectroscopy and multi-way techniques. PARAFAC. Anal. Methods 5, 6557. Murphy, K.R., Stedmon, C.A., Wenig, P., Bro, R., 2014. OpenFluorean online spectral library of auto-fluorescence by organic compounds in the environment. Anal. Methods 6, 658. Murphy, K.R., Ruiz, G.M., Dunsmuir, W.T.M., Waite, T.D., 2006. Optimized Parameters for Fluorescence-Based Verification of Ballast Water Exchange by Ships. Environ. Sci. Technol. 40, 2357e2362. Noble, A.C., Summerfelt, S.T., 1996. Diseases encountered in rainbow trout cultured in recirculating systems. Annu. Rev. Fish. Dis. 6, 65e92. Osburn, C.L., Wigdahl, C.R., Fritz, S.C., Saros, J.E., 2011. Dissolved organic matter composition and photoreactivity in prairie lakes of the U.S. Great Plains. Limnol. Oceanogr. 56, 2371e2390. 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, 246e254. 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 338e341, 237e245. Reynolds, D.M., Ahmad, S.R., 1997. Rapid and direct determination of wastewater BOD values using a fluorescence technique. Water Res. 31, 2012e2018. Rintam€ aki-Kinnunen, P., Rahkonen, M., Mannermaa-Ker€ anen, A., Suomalainen, L., €, H., Valtonen, E., 2005. Treatment of ichthyophthiriasis after malachite Mykra green. I. Concrete tanks at salmonid farms. Dis. Aquat. Organ 64, 69e76. ska-Sobecka, B., Stedmon, C.A., Boe-Hansen, R., Waul, C.K., Arvin, E., 2011. Seredyn Monitoring organic loading to swimming pools by fluorescence excitationeemission matrix with parallel factor analysis (PARAFAC). Water Res. 45, 2306e2314. 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, 135e149.
Shutova, Y., Baker, A., Bridgeman, J., Henderson, R.K., 2014. Spectroscopic characterisation of dissolved organic matter changes in drinking water treatment: from PARAFAC analysis to online monitoring wavelengths. Water Res. 54, 159e169. Stedmon, C.A., Cory, R.M., 2014. Biological origins and fate of fluorescent dissolved organic matter in aquatic environments. In: Coble, P.G. (Ed.), Aquatic Organic Matter Fluorescence. Cambridge University Press, New York. Stedmon, C.A., Markager, S., Bro, R., 2003. Tracing dissolved organic matter in aquatic environments using a new approach to fluorescence spectroscopy. Mar. Chem. 82, 239e254. ska-Sobecka, B., Boe-Hansen, R., Le Tallec, N., Waul, C.K., Stedmon, C.A., Seredyn Arvin, E., 2011. A potential approach for monitoring drinking water quality from groundwater systems using organic matter fluorescence as an early warning for contamination events. Water Res. 45, 6030e6038. Summerfelt, S.T., Hochheimer, J.N., 1997. Review of ozone processes and applications as an oxidizing agent in aquaculture. Prog. Fish-Cult 59, 94e105. Summerfelt, S.T., Vinci, B.J., 2008. Better management practices for recirculating systems. In: Tucker, C.S., Hargreaves, J.A. (Eds.), Environmental Best Management Practices for United States Aquaculture. Blackwell Publishing, Ames, Iowa. 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, 17e27. Timmons, M.B., Ebeling, J.M., 2010. Recirculating Aquaculture. Cayuga Aqua Ventures, Ithaca, NY. Tranvik, L.J., Bertilsson, S., 2001. Contrasting effects of solar UV radiation on dissolved organic sources for bacterial growth. Ecol. Lett. 4, 458e463. Win, Y., 2000. Influence of oxidation of dissolved organic matter (DOM) on subsequent water treatment processes. Water Res. 34, 2098e2104. ~ o, H., Jaffe , R., 2010. Optical characterization of disYamashita, Y., Maie, N., Bricen solved organic matter in tropical rivers of the Guayana Shield, Venezuela. J. Geophys. Res. 115.
Contents lists available at ScienceDirect
Water Research journal homepage: www.elsevier.com/locate/watres
Characterising organic matter in recirculating aquaculture systems with fluorescence EEM spectroscopy ska-Sobecka b, A.C. Hambly a, *, E. Arvin b, L.-F. Pedersen c, P.B. Pedersen c, B. Seredyn C.A. Stedmon a a b c
National Institute for Aquatic Resources, Technical University of Denmark, Kavalergården 6, DK-2920 Charlottenlund, Denmark Technical University of Denmark, Department of Environmental Engineering, DTU Environment, Miljoevej, Building 113, DK-2800 Kgs. Lyngby, Denmark 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
a b s t r a c t
Article history: Received 13 February 2015 Received in revised form 19 June 2015 Accepted 22 June 2015 Available online 25 June 2015
The potential of recirculating aquaculture systems (RAS) in the aquaculture industry is increasingly being acknowledged. Along with intensified application, the need to better characterise and understand the accumulated dissolved organic matter (DOM) within these systems increases. Mature RASs, stocked with rainbow trout and operated at steady state at four feed loadings, were analysed by dissolved organic carbon (DOC) analysis and fluorescence excitation-emission matrix (EEM) spectroscopy. The fluorescence dataset was then decomposed by PARAFAC analysis using the drEEM toolbox. This revealed that the fluorescence character of the RAS water could be represented by five components, of which four have previously been identified in fresh water, coastal marine water, wetlands and drinking water. The fluorescence components as well as the DOC showed positive correlations with feed loading, however there was considerable variation between the five fluorescence components with respect to the degree of accumulation with feed loading. The five components were found to originate from three sources: the feed; the influent tap water (groundwater); and processes related to the fish and the water treatment system. This paper details the first application of fluorescence EEM spectroscopy to assess DOM in RAS, and highlights the potential applications of this technique within future RAS management strategies. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Recirculating aquaculture systems (RAS) Fluorescence excitation-emission matrix (EEM) spectroscopy Dissolved organic matter (DOM) PARAFAC analysis
1. Introduction Developing and advancing aquaculture technology is critical, as aquaculture is playing an increasingly vital role in sustainable food production (FAO, 2014). Intensive aquaculture systems with up to hundreds of metric tonnes of biomass generate large amounts of faecal matter daily. The organic waste in these systems is linked to the feed input (amount and quality) as well as feed digestibility and utilisation (Dalsgaard and Pedersen, 2011), and measures to remove dissolved and particulate matter are therefore often applied to
Abbreviations: BOD, Biological oxygen demand; C:N, Carbon to nitrogen ratio; CFB, Cumulative feed burden; COD, Chemical oxygen demand; COD:N, COD to nitrogen ratio; DOC, Dissolved organic carbon; DOC:N, DOC to nitrogen ratio; DOM, Dissolved organic matter; EEM, Excitation-emission matrix; FDOM, Fluorescent dissolved organic matter; PARAFAC, Parallel factor analysis; RAS, Recirculating aquaculture system; RU, Raman Units; TOC, Total organic carbon; UV, Ultraviolet; UVA, Ultraviolet A; UVB, Ultraviolet B. * Corresponding author. E-mail address: [email protected] (A.C. Hambly). http://dx.doi.org/10.1016/j.watres.2015.06.037 0043-1354/© 2015 Elsevier Ltd. All rights reserved.
mitigate negative impacts on water clarity and quality. Recirculating aquaculture systems (RASs) are land-based, closed-loop systems in which fish are grown and harvested. RAS benefits potentially include stable rearing conditions, reduced risk of escapees, options for pathogen control, as well as solids/particulate removal and hence decreased environmental impacts (Martins et al., 2010). Challenges to RAS include investment and operating costs, as well as the risk of operational failure. These failures may arise from poor water quality conditions and are often associated with disease encounters and/or insufficient pathogen control (Noble and Summerfelt, 1996; Pedersen et al., 2009). Attempts at mitigating such problematic conditions are typically carried out by severe biosecurity measures (Sharrer et al., 2005; Summerfelt et al., 2009) €kior by chemical treatment often with limited success (Rintama Kinnunen et al., 2005). Despite these challenges, farming RAS at commercial scales is becoming increasingly common with more species being cultured (Dalsgaard et al., 2013) and systems becoming increasingly large. The economic impacts of RAS
A.C. Hambly et al. / Water Research 83 (2015) 112e120
underperformance and failures are substantial, and minimising these events is important for the advancement of RAS and aquaculture in general. The full potential of RAS is yet to be realised, to become profitable and to accelerate development on land, as current water quality monitoring practices have been shown to be insufficient for reliable, failure-free operation. The management and mitigation of system failures is at present commonly focussed on maintaining chemical water quality by applying conventional wastewater treatment processes for removal of potentially toxic compounds such as ammonia, as well as by maintaining a balanced microbial community. The effectiveness of these mechanical and biological treatment processes is of utmost importance to RAS viability, and fundamental treatment processes such as particle/solids removal have been shown to be heavily impacted by inherent DOC concentrations (Hem et al., 1994; Summerfelt and Vinci, 2008; Timmons and Ebeling, 2010). For example, Ling and Chen (2005) observed such effects on nitrification efficiency, where an exponential decrease in nitrification performance was recorded with the addition of DOC into the water matrix. Hu et al. (2009) observed negative impacts on nitrification efficiency with increases in carbon to inorganic nitrogen ratios, and a number of studies have recommended that generally elevated C:N ratios are to be avoided (Guerdat et al., 2011; Michaud et al., 2006). Managing microbial water quality is also an important component of RAS management, as bacteria, parasites and algae can affect fish growth and survival, and ultimately RAS output (Blancheton et al., 2013). Recently, Moestrup et al. (2014) observed instances of fish kills within two Danish marine RAS where the cause was identified to be populations of two dinoflagellate species: Pfiesteria shumwayae and Luciella masanensis. Like many heterotrophic dinoflagellate species, Pfiesteria are reported to thrive in conditions where the water body contains high organic matter content (Burkholder and Glascow, 1997; Glibert et al., 2001; Lewitus et al., 1999), which is often the case in RAS utilising long retention times and insufficient solids/organic matter removal. This suggests that the efficient and successful management of a RAS is heavily reliant on controlling the accumulation and quality of dissolved organic matter (DOM) in the system. It follows that implementing more adequate monitoring and control of the organic content of RAS waters could therefore serve to detect deviations in treatment efficiency, and as a preventative measure for RAS failure and subsequent mortality events. Current RAS monitoring methods typically include the analysis of bulk organic matter in the water, including particulate matter, and inorganic nitrogen species such as ammonia, nitrate and nitrite. Additionally, gross indicators such as chemical oxygen demand (COD) and biological oxygen demand (BOD) can be applied, however, analysis of BOD typically requires 5e7 days which introduces an inherent delay in response to events. The organic matter within RASs is derived from a range of sources, each varying in relative importance with time, and so bulk organic matter measurements alone (e.g. DOC or absorbance at 254 nm, A254) may not provide an adequate estimate of the extent to which organic matter character may be fluctuating and potentially influencing RAS performance. The fraction of DOM that is fluorescent, FDOM, and its spectral characteristics of, have been successfully utilised as quantitative and qualitative measures of DOM across a range of natural and engineered systems including, lakes and rivers (Baker, 2001, 2002; Yamashita et al., 2010), wastewater (Bridgeman et al., 2013; Hambly et al., 2010a; Reynolds and Ahmad, 1997), drinking water (Shutova et al., 2014; Stedmon et al., 2011) and marine systems (Coble, 1996; Coble et al., 1990). Different components of FDOM have been shown to be tracers of DOM sub-fractions, including
113
biorefractory, labile and photochemically-active components (Stedmon and Cory, 2014). Fluorescence excitation-emission matrix (EEM) spectroscopy has emerged as the state-of-the-art in FDOM analysis and has become increasingly common as an analytical tool for water sciences (Coble et al., 2014). EEMs are generated by recording emission spectra over a range of excitation wavelengths and combining them to form a detailed contour map of the fluorescent properties of organic matter in a water sample. Multivariate data analysis techniques, in particular parallel factor analysis (PARAFAC), have proven their abilities to reliably decompose EEMs into independently varying fluorescent components, allowing more accurate identification of independent FDOM components (Murphy et al., 2013). As such, the ability of fluorescence spectroscopy to successfully characterise aquatic systems has become increasingly evident. Fluorescence analysis is a growing research area within natural and engineered water systems, and individual components of FDOM have been shown to correlate well with BOD (Hudson et al., 2008; Reynolds and Ahmad, 1997), COD (Bridgeman et al., 2013; Lee and Ahn, 2004), ammonia, nitrates and phosphates (Baker and Inverarity, 2004), as well as microbiological indicators such as total coliforms and E. coli (Cumberland et al., 2012). These are important parameters within RAS, though currently unable to be reliably monitored online and analysed within a reactive timeframe. Fluorescence spectroscopy, however, is a fast, sensitive and non-destructive analysis technique, and hence shows great potential in its application to realtime monitoring of RAS and aquaculture in general. More specifically, these properties could enable fluorescence analysis to become a valuable, real-time monitoring tool to optimise RAS management. This study provides the first application of fluorescence EEM spectroscopy and PARAFAC analysis to characterise DOM in aquaculture systems, more specifically within a RAS. Whether changes in FDOM are consistent across all types of RAS are yet to be determined, however this study evaluates the application of the EEM-PARAFAC technique itself. The aim was to test if the technique can identify characteristic organic matter fractions from the complex organic RAS matrix, and to therefore outline the potential of using fluorescence as a sensitive monitoring parameter of RAS water. 2. Materials and methods 2.1. Experimental design Four identical 1700 L freshwater RASs (Fig. 1) were stocked with juvenile rainbow trout (Oncorhynchus mykiss) with an average weight of 131 ± 6 g and brought to steady state conditions (in terms of nitrate equilibrium) over 2 months. Four daily feed inputs were applied within the study (125, 250, 375 or 500 g). The daily water renewal of each RAS was 80 L, corresponding to relative water renewal rates of between 160 and 640 L/kg feed, and the cumulative feed burden (CFB) ranged from 1.6 to 6.3 kg feed per m3 water renewal. The stocking density was adjusted to the feed given; initial biomasses were 14, 28, 42 and 56 kg/m3 at the four levels. As a consequence of the experimental design with fixed daily feed allocation the corresponding feeding ratio decreased from 1.8% to 1.1% of fish biomass per day during the experimental period due to biomass gain. An experimental fish feed consisting of 44% protein and 30% lipid (Pedersen et al., 2012) (Table 1) was utilised and allocated daily from 9:00 to 15:00 h by belt feeders. Forty litres of RAS water was drained daily from the swirl separator in each RAS, and replaced with 80 L of non-chlorinated tap water. The 40 L excess was discharged by overflow and evaporation. This corresponded to a water exchange
114
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Fig. 1. Schematic of the experimental recirculating aquaculture systems (RAS). Water circulation in the RAS is represented with thick arrows, dotted lines represent system inlet and water outlet; P indicates the position of the circulation pump.
Table 1 Ingredients and proximate composition of the experimental diet. Ingredient (as-fed basis)
g/kg
Fish meal LT 92 Spray blood red cells Wheat Wheat glluten Pea protein 75 Field beans Sunflower cake Soya cake Fish oil 60 Premixes Proximate composition
320 100 97 60 60 54 36 24 244 1 %
Proteina Fatb NFEc Ashd Moistured
44.0 29.9 11.3 6.1 8.7
a b c d
ISO 5983-2, 2005. Bligh & Dyer, 1959. NFE calculated as: dry mattereproteinelipideash. NMKL 23, 1991.
rate of 4.7% of system volume per day and a hydraulic retention time of 3 weeks. The water temperature was kept constant at 18.0 C by the use of heating elements within the RAS sump. Oxygen concentrations ranged from 7.2 to 8.5 mg/L, and were maintained by aeration and/or addition of oxygen with diffusers within the rearing tank. The pH was kept between 7.2 and 7.4, and was adjusted daily by the addition of sodium bicarbonate (approx. 20% w/w of added feed) to compensate for alkalinity loss from the nitrification processes of the biofilters. The systems were inspected daily and there were typically no uneaten pellets found. In the case of any uneaten pellets, these were removed by the swirl separator (Fig. 1). A feed extraction study was also performed by dissolving 10 g of the fish feed in 100 mL of MilliQ water. The mixture was then shaken at room temperature for 24 h before sampling.
2.2. Sampling Water samples for fluorescence EEM and DOC were collected hourly throughout the day and in the morning on the following day in order to study organic matter quantity and quality at different feed inputs. The water samples were taken from each RAS at the rearing tank, separator and after the biofilters. Feed extractions were sampled after 24 h and filtered through a glass filter. Serial dilutions were then carried out to enable sample analysis by DOC and fluorescence EEM spectroscopy. 2.3. Analyses All aquaculture water samples were filtered through 0.45 mm cellulose acetate filters (Millipore) and collected in acid washed and pre-combusted 40 mL glass vials with Teflon-lined caps. DOC analyses were performed with a Shimadzu TOC-V WP analyser with an ASI-V autosampler, calibrated with a six-point calibration curve using potassium hydrogen phthalate. Fluorescence EEMs were measured in a 1 cm quartz cuvette (4 mL volume) using a Varian Cary Eclipse Fluorescence Spectrophotometer (Varian, UK). Fluorescence EEMs were measured for excitation wavelengths of lex ¼ 200e400 nm at 5 nm increments across an emission range of lem ¼ 280e500 nm at 2 nm intervals. Excitation and emission slit widths were set to 5 nm, with a photomultiplier tube (PMT) voltage of 800 V. UV absorption measurements were obtained on a Varian Cary 50 Bio UVevisible Spectrophotometer (Varian, UK) in a 1 cm quartz cuvette and UVevisible spectra were recorded from 240 to 700 nm with slit width of 0.5 nm. 2.4. EEM post-processing All raw EEM processing was carried out using the drEEM toolbox (Murphy et al., 2013) within Matlab software (The MathWorks Inc.). The EEM correction process consisted of a blank EEM subtraction and scatter line removal, the application of excitation and emission correction factors, correction for inner filter effects, and normalisation to Raman Units (RU) before PARAFAC analysis was carried out.
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Blank EEMs were acquired from MilliQ water samples, and excitation and emission correction factors were obtained using Rhodamine B and a ground quartz diffuser. Raman and Rayleigh scatter peaks were removed from spectra within the drEEM toolbox, as well as wavelengths below lex ¼ 250 nm and lem ¼ 300 nm to limit spectrally noisy areas of the matrix. Inner filter effect correction factors were derived from the absorption measurements and Raman normalisation factors were calculated using MilliQ blank EEMs at lex ¼ 350 nm (Lawaetz and Stedmon, 2009). The resulting corrected and normalised EEM dataset was then subjected to parallel factor analysis (PARAFAC) with the drEEM toolbox (n ¼ 148). The least squares models for 3 to 7 component models were tested by running 30 iterations, with non-negativity and a convergence criteria of 1 108, and the most appropriate number of PARAFAC components was found using a combination of split half and residual analysis according to the recommendations in Murphy et al. (2013). 3. Results 3.1. Fluorescence characteristics and identified components Fig. 2 shows the general fluorescence characteristics of the DOM in the four types of RAS waters, the drinking water source used to fill the RAS, and the pure water extraction of DOM from the feed utilised in the experiment. The fluorescence signal of RAS samples were characterised by three regions: (1) a broad peak in the region of lex < 250 nm and lem ¼ 400e500 nm; (2) a clear peak at lex/ em ¼ 340/430 nm; and (3) a UV peak at lex/em 280/340 nm. A similar fluorescence signal was present in the source tap water but at much lower intensity, particularly in the UV emission region. The DOM extracted from the aquaculture feed was characterised by a high fluorescence intensity in the UV region, with a UVB peak at lex/ em ¼ 280/300 nm, overlapping with a broader UVA peak at lex/ em ¼ 280/340 nm. The PARAFAC analysis indicated that the fluorescence signal could be mathematically decomposed into five independently varying fractions (Fig. 3). Components 1 to 4 were dominated by visible emission region (>400 nm), whereas component 5 was dominated by a peak in the UVA emission region (300e380 nm). 3.2. Feed loading effects on DOC and FDOM components A linear relationship between DOC and feed input was observed, where every 125 g/day feed increase was accompanied by a DOC increase of approximately 5 mg/L (Fig. 4). Across each of the four feed input levels, samples displayed very similar fluorescence character, while differing in overall fluorescence intensities. Fluorescence intensities increased with increased feed input (Fig. 5), which is in general agreement with the feed relationship for DOC concentrations observed (Fig. 4). Although overall DOC was observed to be linear in relation to feed input, not all fluorescence components adhered to the same relationship (Fig. 5). Components F1, F2 and F4 showed a positive linear correlation (R > 0.99 in all cases) to feed input with fluorescence intercepts close to that of the tap water used to fill the RAS (Fig. 6). Component F5 differed by having a non-linear positive correlation with feed input, which could be modelled with a quadratic polynomial. Compared to the other components, F3 showed a marginal increase with increased feed input, with an intercept (analogous to a 0.0 g feed/day) of 1.5 RU (Raman Units). As DOC showed a strong linear correlation to feed input, the relationships between FDOM and DOC were comparable to that between FDOM and feed input (Fig. 5).
115
4. Discussion 4.1. Identified fluorescence components Similar PARAFAC components have been observed and described throughout the literature of DOM fluorescence, and a number of matching peaks were found from a review of literature and the OpenFluor database of PARAFAC aquatic fluorescence components (Murphy et al., 2014). The origins and properties of corresponding OpenFluor components are shown in Table 2. The peaks display similarities (minimum spectral similarity score of 0.95) to those within studies from a variety of aquatic sources, including: (1) marine water; (2) freshwater/drinking water; and (3) freshwater impacted by treated wastewater. This is consistent with the typical elements of the RASs investigated within this study: (1) the filling of the system with freshwater/drinking water; (2) a marine organism component; and (3) the system contains a biological treatment train through which the aquaculture water is recycled. These consistencies support the hypothesis that many organic matter fluorescence signals identified across contrasting aquatic systems essentially represent common ubiquitous degradation intermediates. Interestingly, component F5 contains peaks that have also been commonly associated with microbial activity (Hudson et al., 2008) as well as being correlated with wastewater and recycled water sources (Hambly et al., 2010a, 2010b; Hudson ska-Sobecka et al., 2011). Although no direct et al., 2007; Seredyn matches were found for component F1 within the OpenFluor database, the broad peak which dominates it (lex/em ¼ 360/450) shares distinct similarity with fluorophores commonly referred to as “humic-like” or “Peak C” (Coble, 1996). 4.2. Identification of fluorescence sources All of the PARAFAC components were observed to correlate to feed input, to varying degrees. This correlation could arise either: (a) directly e from dissolution of feed or undigested parts thereof; or (b) indirectly e from biological fish processes and/or processes in the biofilters related to increased feeding. To discern between the two, a feed dissolution study applying 10 g feed/100 mL (100 kg feed/m3) was undertaken to isolate the fluorescence signal arising from direct feed dissolution within the RAS. Fig. 6 shows the fluorescence loadings from the five components (F1eF5) for each sample type and it can be seen that by far the greatest contributor to the dissolved feed fluorescence is component F5, which accounts for 82% of the total FDOM within sample. This information, in combination with the known concentrations of the feed dissolution samples (100 kg feed/m3 water), was used to perform a conservative calculation for the direct contribution of feed dissolution to the FDOM of the RAS water. The grinding of the feed with a mortar and pestle likely enhanced dissolution compared to normal conditions, but the tests provide a first estimate of the effects of feed dissolution on the FDOM signal. If the observed fluorescence of a component is higher than the calculated dissolution value, there is a production of this fluorescence component within the RAS, whether it be from fish digestion or the treatment system. The calculation was also made under the conservative assumption that the five components were not decomposed in the RAS. The results are illustrated in Fig. 7 which reveals that the fluorescence components in the RAS water are related, to various degrees, to three sources: 1) the feed (F5); 2) the influent tap water (groundwater, F3); and 3) the fish and treatment processes (F1, F2, F4). The dissolved feed contributed a large proportion of component F5 to the calculated fluorescence loadings. In the RAS water, the observed F5 loadings were much lower than the calculated values, indicating degradation or removal of this fraction. Components F1
116
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Fig. 2. Typical EEMs of RAS waters at daily feed inputs of: (a) 125 g/day; (b) 250 g/day; (c) 375 g/day; and (d) 500 g/day; (e) feed dissolved in MilliQ water; and (f) tap water used to fill RAS. N.B. Z-axis (fluorescence intensity) scales differ between samples to allow comparison of fluorescence character.
A.C. Hambly et al. / Water Research 83 (2015) 112e120
117
Fig. 3. Contour plots of individual components from the 5-component PARAFAC model (generated from 148 samples from all four feed levels).
Fig. 4. DOC concentrations (mean ± s.d.) as a function of feed input.
and F2 can be explained by a mixture of the feed input, tap water and FDOM produced in the RAS, and component F4 can be explained predominantly by FDOM produced within the RAS. Finally, component F3 arises for the most part from tap water and the feed dissolution, with the tap water as the predominant source. Fig. 6 shows that the main constituent of tap water FDOM is component F3, which accounts for over 45% of fluorescence within the tap water samples and is consistent with the relationship observed between component F3 and feed input (Fig. 5). As the feed input has increased, the amount of tap water within the system has essentially stayed constant between RAS's (with a small relative increase due to daily fill of make-up water). It must be emphasised that the calculations above on the contribution of the feed to the fluorescence of the RAS water samples involves some uncertainty due to the aforementioned dissolution method. Another uncertainty is related to the assumption that the five components were not decomposed in RAS. Though this was unlikely to be the case for component F5, it may be
Fig. 5. Fluorescence component loadings (mean ± s.d.) at the four feeding levels, calculated from all samples in each group over the entire sampling period. Trendlines for F1eF4 were fitted by linear regression, whereas the trendline for F5 was fitted as a quadratic polynomial.
that some of the other components were partly decomposed (e.g. from uptake and decomposition by the fish), whilst being simultaneously produced by the treatment system. Whilst it is possible to ascertain the net result of such processes within this study, it is not possible to determine the extent to which components are both decomposed and produced. Despite the clear degradation of component F5 within the RAS (Fig. 7), the observed accumulation at higher feed loadings (Fig. 5) suggests its potential importance to the system. The fluorescence exhibited by component F5 is commonly referred to as “peak T” fluorescence (Coble, 1996) and often associated with elevated
118
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Fig. 7. Fluorescence loadings (mean ± s.d.) of each PARAFAC component for theoretical values (sum of contribution from dissolved feed and tap water signals) and observed values, at each of the four feed inputs.
Fig. 6. Fluorescence loadings (mean ± s.d.) of each PARAFAC component for RAS water (at feed inputs of 125, 250, 375 and 500 g/day), tap water and dissolved feed (displayed as the theoretical 125 g/day equivalent).
microbial activity (Hudson et al., 2008). Degradation studies have implied that this fluorescence component represents microbial bioavailable organic matter, which can be produced by, or consumed by bacteria (Stedmon and Cory, 2014). The accumulation of this signal at high feed loadings could therefore indicate the potential for a significant microbiological population change within the RAS brought about by the excess organic matter. In addition to excess feed input, increased DOM concentrations within a RAS could come from a number of other potential sources. Inadequate removal of unconsumed feed and faecal matter could result in a build-up of organic matter within a RAS, in addition to changed feed constituency, feed digestibility, recirculation rates and water exchange volumes. Unconsumed feed and faecal matter was manually removed by a swirl separator flushed every day during the current study, and recirculation rates, water exchange and feed constituency were kept constant. Overloading of the RAS treatment train may also lead to an increase in nutrients and organic matter. COD:N and DOC:N ratios have been directly linked to nitrification performance within a number of previous studies, where increases in these ratio have seen concurrent decreases in nitrification efficiency (Ling and Chen, 2005; Hu et al., 2009). Increased C:N ratio tend to favour growth of (opportunistic) heterotrophic bacteria over slower growing (stable) nitrifying bacteria (Attramadal et al., 2014). As aquaculture looks to
potentially employ more advanced water treatment processes such as UV and ozone within recirculating systems (Summerfelt and Hochheimer, 1997; Summerfelt et al., 2009) implementing a fast and reliable method for the monitoring of organic matter will become even more important. These processes have been shown to have varied treatment efficacy on organic matter fractions (HessErga et al., 2008; Tranvik and Bertilsson, 2001; Win, 2000). Recent RAS studies are suggesting that fish farms operated at high feed loadings are at risk of fish kills in cases of insufficient solids removal and organic material accumulation, potentially promoting toxic microalgae populations (Moestrup et al., 2014). Increased levels of organic matter in the system combined with reduced treatment process efficiencies can rapidly and detrimentally impact system and hence fish performance. As highlighted, the RAS in this study were operated under fixed conditions with a single fish species and feed type, and the DOM quantity and quality is likely to be affected by fish type, fish density, feed type, treatment train and operational conditions. However, the ability to of the EEM-PARAFAC method to identify and then monitor specific fluorescence components shows potential for application across a wide range of RAS operations. Fluorescence spectroscopy shows great potential in monitoring the quality and quantity of organic material in RASs, due to its high sensitivity and proven online monitoring ability. Ultimately, fluorescence monitoring could therefore give RAS operators the ability to react quickly to small changes in organic matter, to better manage the systems as a whole and potentially avert detrimental operational events.
Table 2 Fluorescence peak wavelength pairs for RAS 5-component PARAFAC model and literature comparisons. Component Peak(s)
lexcitation (nm)
lemission (nm)
Source(s) description from OpenFluor study matches
F1 F2 F3 F4 F5
360 325 250 270 (410) 285
450 410 440 480 330
(None over 0.95 spectral similarity) Freshwater lakesa; saline lakesb; rivers, estuaries and coastal marine watersc,d,e Drinking waterf; river plume and coastal marine watersc Freshwater lakesg Marine ballast waterh; agricultural-impacted subtropical wetlandsi; anthropogenic-impacted tropical riversj
a b c d e f g h i j
(Kothawala et al., 2012). (Osburn et al., 2011). (Kowalczuk et al., 2009). (Stedmon et al., 2003). (Cawley et al., 2012). (Shutova et al., 2014). (Kothawala et al., 2014). (Murphy et al., 2006). (Yamashita et al., 2010b). (Yamashita et al., 2010a).
A.C. Hambly et al. / Water Research 83 (2015) 112e120
5. Conclusions * PARAFAC analysis of EEMs shows a 5-component model to describe changing fluorescence components in an RAS stocked with rainbow trout: a) three components, F1, F2 and F4, which linearly correlated to feed input; (b) one component, F5, which correlated to feed input and showed accumulation with increased feed inputs; and (c) one component, F3, which correlated to tap water. * Four out of the five fluorescence components have previously been identified within the OpenFluor spectral database, from within river water, marine coastal water, drinking water and wetlands studies. * The five fluorescence components originate from three sources: 1) the feed input; 2) the influent tap water (groundwater); and 3) organic matter produced by the fish and treatment processes. * There is a considerable difference between the components as to the degree they originate from these three sources. F5 mainly originates from the fish feed, F4 mainly originates from organic matter produced in the RAS, whereas F1 and F2 originate both from the feed as well as being produced within the RAS. Finally, F3 predominantly represents the contribution of organic matter in the tap water used to fill the system. * Although DOC increases linearly with increased feed loading, PARAFAC analysis shows fluorescence components are not all linearly increasing. Measuring FDOM may offer a more specific approach to monitor the accumulation of bioavailable organic matter in a system, as opposed to bulk DOC. Specific FDOM components could potentially also be correlated to predict DOC values. * Future research should include the effect of fish species, fish density, feed composition, treatment types, salinity, and water exchange on FDOM. * Fluorescence is a promising parameter for monitoring the organic content of RASs, as it can be easily developed for online sensor measurements. Such a fluorescence-based early warning tool could contribute to the successful operation of RASs with higher feed loadings and lower water renewal requirements than is currently achievable, and ultimately advance the development and profitability of sustainable RAS industries. Acknowledgements This work was funded by the Danish Agency for Spatial and Environmental Planning, Ministry of the Environment (Ref. BLS403-00043 AquaFingerprint) and FP7-PEOPLE-2013-IIF, Project number 626147, FAMoRAS e Fluorescence Analysis and Monitoring of Recirculating Aquaculture Systems (Marie Curie Actions eInternational Incoming Fellowships. Adam Hambly). In addition the study was also financed by The Technical University of Denmark (DTU) through a strategic research programme between DTU Environment and DTU Aqua. References 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, 483e490. Baker, A., 2001. Fluorescence excitationemission matrix characterization of some sewage-impacted rivers. Environ. Sci. Technol. 35, 948e953. Baker, A., 2002. Fluorescence excitationemission matrix characterization of river waters impacted by a tissue mill effluent. Environ. Sci. Technol. 36, 1377e1382. Baker, A., Inverarity, R., 2004. Protein-like fluorescence intensity as a possible tool for determining river water quality. Hydrol. Process 18, 2927e2945. 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, 30e39. Bridgeman, J., Baker, A., Carliell-Marquet, C., Carstea, E., 2013. Determination of
119
changes in wastewater quality through a treatment works using fluorescence spectroscopy. Environ. Technol. 34, 3069e3077. Burkholder, J.M., Glascow, H.B., 1997. Pfiesteria piscicida and other Pfiesreria-like dinoflagellates: behavior, impacts, and environmental controls. Limnol. Oceanogr. 42, 1052e1075. Cawley, K.M., Butler, K.D., Aiken, G.R., Larsen, L.G., Huntington, T.G., McKnight, D.M., 2012. Identifying fluorescent pulp mill effluent in the Gulf of Maine and its watershed. Mar. Pollut. Bull. 64, 1678e1687. Coble, P.G., 1996. Characterization of marine and terrestrial DOM in seawater using excitation-emission matrix spectroscopy. Mar. Chem. 51, 325e346. Coble, P.G., Green, S.A., Blough, N.V., Gagosian, R.B., 1990. Characterization of dissolved organic matter in the Black Sea by fluorescence spectroscopy. Nature 348, 432e435. Coble, P.G., Lead, J., Baker, A., Reynolds, D.M., Spencer, R.M., 2014. Aquatic Organic Matter Fluorescence. Cambridge University Press, New York. Cumberland, S., Bridgeman, J., Baker, A., Sterling, M., Ward, D., 2012. Fluorescence spectroscopy as a tool for determining microbial quality in potable water applications. Environ. Technol. 33, 687e693. Dalsgaard, J., Pedersen, P.B., 2011. Solid and suspended/dissolved waste (N, P, O) from rainbow trout (Oncorynchus mykiss). Aquaculture 313, 92e99. 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, 2e13. FAO, 2014. The State of World Fisheries and Aquaculture 2014-Opportunities and Challenges. Food and Agriculture Organisation of the United Nations, Rome, Italy. Glibert, P.M., Magnien, R., Lomas, M.W., Alexander, J., Fan, C., Haramoto, E., Trice, M., Kana, T.M., 2001. Harmful algal blooms in the chesapeake and coastal Bays of Maryland, USA: comparison of 1997, 1998, and 1999 events. Estuaries 24, 875. 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, 10e18. Hambly, A.C., Henderson, R.K., Storey, M.V., Baker, A., Stuetz, R.M., Khan, S.J., 2010a. Fluorescence monitoring at a recycled water treatment plant and associated dual distribution system e implications for cross-connection detection. Water Res. 44, 5323e5333. Hambly, A.C., Henderson, R.K., Baker, A., Stuetz, R.M., Khan, S.J., 2010b. Fluorescence monitoring for cross-connection detection in water reuse systems: Australian case studies. Water Sci. Technol. 61, 155. Hem, L.J., Rusten, B., Ødegaard, H., 1994. Nitrification in a moving bed biofilm reactor. Water Res. 28, 1425e1433. Hess-Erga, O., Kihle Attramadal, K., Vadstein, O., 2008. Biotic and abiotic particles protect marine heterotrophic bacteria during UV and ozone disinfection. Aquat. Biol. 4, 147e154. Hu, J., Li, D., Liu, Q., Tao, Y., He, X., Wang, X., Li, X., Gao, P., 2009. Effect of organic carbon on nitrification efficiency and community composition of nitrifying biofilms. J. Environ. Sci. 21, 387e394. Hudson, N., Baker, A., Reynolds, D., 2007. Fluorescence analysis of dissolved organic matter in natural, waste and polluted watersda review. River Res. Appl. 23, 631e649. Hudson, N., Baker, A., Ward, D., Reynolds, D.M., Brunsdon, C., Carliell-Marquet, C., Browning, S., 2008. Can fluorescence spectrometry be used as a surrogate for the Biochemical Oxygen Demand (BOD) test in water quality assessment? An Ex. South West Engl. Sci. Total Environ 391, 149e158. €hler, S.J., Kothawala, D.N., Stedmon, C.A., Müller, R.A., Weyhenmeyer, G.A., Ko Tranvik, L.J., 2014. Controls of dissolved organic matter quality: evidence from a large-scale boreal lake survey. Glob. Change Biol. 20, 1101e1114. Kothawala, D.N., von Wachenfeldt, E., Koehler, B., Tranvik, L.J., 2012. Selective loss and preservation of lake water dissolved organic matter fluorescence during long-term dark incubations. Sci. Total Environ. 433, 238e246. Kowalczuk, P., Durako, M.J., Young, H., Kahn, A.E., Cooper, W.J., Gonsior, M., 2009. Characterization of dissolved organic matter fluorescence in the South Atlantic Bight with use of PARAFAC model: Interannual variability. Mar. Chem. 113, 182e196. Lawaetz, A.J., Stedmon, C.A., 2009. Fluorescence intensity calibration using the Raman scatter peak of water. Appl. Spectrosc. 63, 936e940. Lee, S., Ahn, K.H., 2004. Monitoring of COD as an organic indicator in waste water and treated effluent by fluorescence excitation-emission (FEEM) matrix characterization. Water Sci. Technol. J. Int. Assoc. Water Pollut. Res. 50, 57e63. Lewitus, A.J., Willis, B.M., Hayes, K.C., Burkholder, J.M., Glasgow Jr., H.B., Glibert, P.M., Burke, M.K., 1999. Mixotrophy and nitrogen uptake by pfiesteria piscicida (dinophyceae). J. Phycol. 35, 1430e1437. Ling, J., Chen, S., 2005. Impact of organic carbon on nitrification performance of different biofilters. Aquac. Eng. 33, 150e162. Martins, C.I.M., Eding, E.H., Verdegem, M.C.J., Heinsbroek, L.T.N., Schneider, O., Blancheton, J.P., d'Orbcastel, E.R., Verreth, J.A.J., 2010. New developments in recirculating aquaculture systems in Europe: a perspective on environmental sustainability. Aquac. Eng. 43, 83e93. 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, 224e233. Moestrup, Ø., Hansen, G., Daugbjerg, N., Lundholm, N., Overton, J., Vestergård, M., Steenfeldt, S.J., Calado, A.J., Hansen, P.J., 2014. The dinoflagellates Pfiesteria shumwayae and Luciella masanensis cause fish kills in recirculation fish farms in Denmark. Harmful Algae 32, 33e39.
120
A.C. Hambly et al. / Water Research 83 (2015) 112e120
Murphy, K.R., Stedmon, C.A., Graeber, D., Bro, R., 2013. Fluorescence spectroscopy and multi-way techniques. PARAFAC. Anal. Methods 5, 6557. Murphy, K.R., Stedmon, C.A., Wenig, P., Bro, R., 2014. OpenFluorean online spectral library of auto-fluorescence by organic compounds in the environment. Anal. Methods 6, 658. Murphy, K.R., Ruiz, G.M., Dunsmuir, W.T.M., Waite, T.D., 2006. Optimized Parameters for Fluorescence-Based Verification of Ballast Water Exchange by Ships. Environ. Sci. Technol. 40, 2357e2362. Noble, A.C., Summerfelt, S.T., 1996. Diseases encountered in rainbow trout cultured in recirculating systems. Annu. Rev. Fish. Dis. 6, 65e92. Osburn, C.L., Wigdahl, C.R., Fritz, S.C., Saros, J.E., 2011. Dissolved organic matter composition and photoreactivity in prairie lakes of the U.S. Great Plains. Limnol. Oceanogr. 56, 2371e2390. 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, 246e254. 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 338e341, 237e245. Reynolds, D.M., Ahmad, S.R., 1997. Rapid and direct determination of wastewater BOD values using a fluorescence technique. Water Res. 31, 2012e2018. Rintam€ aki-Kinnunen, P., Rahkonen, M., Mannermaa-Ker€ anen, A., Suomalainen, L., €, H., Valtonen, E., 2005. Treatment of ichthyophthiriasis after malachite Mykra green. I. Concrete tanks at salmonid farms. Dis. Aquat. Organ 64, 69e76. ska-Sobecka, B., Stedmon, C.A., Boe-Hansen, R., Waul, C.K., Arvin, E., 2011. Seredyn Monitoring organic loading to swimming pools by fluorescence excitationeemission matrix with parallel factor analysis (PARAFAC). Water Res. 45, 2306e2314. 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, 135e149.
Shutova, Y., Baker, A., Bridgeman, J., Henderson, R.K., 2014. Spectroscopic characterisation of dissolved organic matter changes in drinking water treatment: from PARAFAC analysis to online monitoring wavelengths. Water Res. 54, 159e169. Stedmon, C.A., Cory, R.M., 2014. Biological origins and fate of fluorescent dissolved organic matter in aquatic environments. In: Coble, P.G. (Ed.), Aquatic Organic Matter Fluorescence. Cambridge University Press, New York. Stedmon, C.A., Markager, S., Bro, R., 2003. Tracing dissolved organic matter in aquatic environments using a new approach to fluorescence spectroscopy. Mar. Chem. 82, 239e254. ska-Sobecka, B., Boe-Hansen, R., Le Tallec, N., Waul, C.K., Stedmon, C.A., Seredyn Arvin, E., 2011. A potential approach for monitoring drinking water quality from groundwater systems using organic matter fluorescence as an early warning for contamination events. Water Res. 45, 6030e6038. Summerfelt, S.T., Hochheimer, J.N., 1997. Review of ozone processes and applications as an oxidizing agent in aquaculture. Prog. Fish-Cult 59, 94e105. Summerfelt, S.T., Vinci, B.J., 2008. Better management practices for recirculating systems. In: Tucker, C.S., Hargreaves, J.A. (Eds.), Environmental Best Management Practices for United States Aquaculture. Blackwell Publishing, Ames, Iowa. 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, 17e27. Timmons, M.B., Ebeling, J.M., 2010. Recirculating Aquaculture. Cayuga Aqua Ventures, Ithaca, NY. Tranvik, L.J., Bertilsson, S., 2001. Contrasting effects of solar UV radiation on dissolved organic sources for bacterial growth. Ecol. Lett. 4, 458e463. Win, Y., 2000. Influence of oxidation of dissolved organic matter (DOM) on subsequent water treatment processes. Water Res. 34, 2098e2104. ~ o, H., Jaffe , R., 2010. Optical characterization of disYamashita, Y., Maie, N., Bricen solved organic matter in tropical rivers of the Guayana Shield, Venezuela. J. Geophys. Res. 115.