Water Research 135 (2018) 112e121
Contents lists available at ScienceDirect
Water Research journal homepage: www.elsevier.com/locate/watres
Improving water quality using settleable microalga Ettlia sp. and the bacterial community in freshwater recirculating aquaculture system of Danio rerio Seong-Jun Chun a, b, Yingshun Cui a, Chi-Yong Ahn a, b, **, Hee-Mock Oh a, b, * a Cell Factory Research Center, Korea Research Institute of Bioscience and Biotechnology (KRIBB), 125 Gwahak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea b Department of Environmental Biotechnology, KRIBB School of Biotechnology, Korea University of Science and Technology (UST), 217 Gajeong-ro, Yuseonggu, Daejeon 34113, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 September 2017 Received in revised form 29 December 2017 Accepted 5 February 2018 Available online 8 February 2018
A highly settleable microalga, Ettlia sp., was applied to a freshwater recirculating aquaculture system (RAS) of Danio rerio to improve the treatment of nitrogenous compounds. The growth characteristics of the microalgae, water quality parameters, and bacterial communities were monitored for 73 days. In the treatment RAS, the inoculated Ettlia sp. grew up to 1.26 g/L and dominated (>99%) throughout the experiment, whereas naturally occurring microalgae grew to 0.57 g/L in the control RAS. The nitrate, nitrite, and ammonium concentrations in the treatment RAS were reduced by 50.1%, 73.3%, and 24.2%, respectively, compared to the control RAS. A bacterial community analysis showed that Rhodospirillales, Phycisphaerae, Chlorobiales, and Burkholderiales were the major bacterial groups in the later phase of the treatment RAS. Meanwhile, a network analysis among the Ettlia sp., bacterial groups, and environmental parameters, revealed that the bacterial groups played key roles in both water quality improvement and Ettlia sp. growth. In conclusion, the inoculation and growth of the Ettlia sp. and its associated bacteria in the RAS produced beneficial effects on the water quality by reducing the nitrogenous compounds and providing a favorable environment for certain bacterial groups to further improve water quality. © 2018 Elsevier Ltd. All rights reserved.
Keywords: Recirculating aquaculture systems Settleable microalgae Ettlia sp. Water quality Nitrate Algae-bacteria interaction
1. Introduction The estimated annual aquaculture production worldwide is 73.8 million tons and the contribution of aquaculture to the world's total fish production is 44.1% (FAO, 2016). Recirculating aquaculture systems (RAS) is generally designed to maintain clean water and provide a suitable environment for aquatic organisms. However, the operation of an RAS requires regular expert care and constant water exchange (5e10%/d) to maintain the water quality, including a tolerable range of dissolved oxygen, pH, carbon dioxide,
* Corresponding author. Cell Factory Research Center, Korea Research Institute of Bioscience and Biotechnology (KRIBB), 125 Gwahak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea. ** Corresponding author. Cell Factory Research Center, Korea Research Institute of Bioscience and Biotechnology (KRIBB), 125 Gwahak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea. E-mail addresses:
[email protected] (C.-Y. Ahn),
[email protected] (H.-M. Oh). https://doi.org/10.1016/j.watres.2018.02.007 0043-1354/© 2018 Elsevier Ltd. All rights reserved.
nitrogenous compounds, phosphorus, and solids concentration (Badiola et al., 2012). Among the various water quality parameters, excessive nitrog enous compounds (ammonium (NHþ 4 ), nitrite (NO2 ), and nitrate (NO )) need to be more carefully controlled for aquaculture sys3 tems. In nature, these compounds are controlled by photosynthetic organisms and denitrifying bacteria. However, nitrogenous compounds (mainly nitrate) tend to accumulate in aquaculture systems (~500 mg NO 3 -N/L) (Pierce et al., 1993), due to high feed loads and high fish densities. High concentrations of nitrate can reduce animal growth and decrease survival in aquacultures (Davidson et al., 2014; Pierce et al., 1993). Further, even low levels of nitrate can lead to chronic toxicity problems in sensitive aquatic organisms, particularly during the early stages of certain freshwater invertebrates, fish, and amphibians (Camargo et al., 2005). Traditionally, the nitrate concentration in aquaculture systems is controlled by water exchange, phytoremediation, biofilter treatment, and denitrification (Crab et al., 2007; Martins et al., 2010; Yang et al., 2001).
S.-J. Chun et al. / Water Research 135 (2018) 112e121
Many studies have previously demonstrated the success of using microalgae as a biological purifier for water treatment (Sutherland et al., 2015). Microalgae are primary oxygen-releasing photosynthetic microorganisms and have already been used in various applications. These microorganisms can improve the water quality by reducing nitrogenous compounds and producing various beneficial compounds (Richmond and Hu, 2013). Microalgae have also shown numerous positive effects on aquatic organisms, such as improving their stress response, physiological activity, starvation tolerance, and disease resistance (Borowitzka, 1997). Moreover, harvested microalgal biomass can be used as sustainable feed for aquaculture (Hemaiswarya et al., 2010). Several well-known algal species, such as Chaetoceros sp., Isochrysis sp., Chlorella sp., and Spirulina sp., have been applied in aquaculture systems (Borowitzka, 1997; Lananan et al., 2014; Sombatjinda et al., 2014). However, harvesting the microalgal biomass from water remains a major problem for the application of microalgae, due to their small cell size and floatability. Therefore, immobilization, periphyton, and bio-floc techniques have all been developed to overcome the harvesting problem of microalgae (Moreno-Garrido, 2008). In this study, an Ettlia sp., which has been assigned to the family Chlorococcaceae, was applied to a newly designed RAS to investigate its effects on water quality and bacterial community structures. This green alga, which is morphologically spherical and/or subspherical, was originally isolated from freshwater and later reported to also inhabit terrestrial environments (Pegg et al., 2015; Yoo et al., 2013). Ettlia sp. is known as a highly settleable microalga due to its autoflocculating property and is a suitable candidate for producing biodiesel and high-value products (e.g. lutein, b-carotenoid, ketocarotenoid, photoprotective reagent) (Lee et al., 2016; Salim et al., 2014; Yoo et al., 2013). Microorganisms form various ecological relationships, ranging from mutualism to competition, that reshape microbial community structures. Recently, association network techniques have frequently been applied to microbial abundance data to detect significant patterns of mutual exclusion between taxa and to represent them as a network (Faust and Raes, 2012). It is generally known that microalgae-bacteria exhibit relationships in the phycosphere may range from mutualism to parasitism (Ramanan et al., 2016). For example, microalgae can provide oxygen, microalgaederived substances, and a favorable environment for bacteria, while bacteria can provide bacteria-derived substances and nutrients for microalgae by decomposing organic polymers into small molecules that can be used by microalgae. Therefore, when applying microalgae to aquaculture systems, the effects of the microalgae on the bacterial community also need to be considered. Accordingly, this study investigated the water quality parameters, microalgae growth characteristics, and bacterial communities in a newly designed RAS to 1) uncover the effects of a highly settleable microalga Ettlia sp. on the water quality parameters, especially on nitrogenous compounds, 2) characterize the Ettlia sp. growth in the newly designed microalgal tank, and 3) reveal the effects of the Ettlia sp. on the bacterial communities in the RAS.
113
approximately 3 L/min. Commercial gravel (0.5e1 mm) was washed 3 times with tap water and 2 times with autoclaved distilled water and was then scattered to create an approximately 5 cm-deep layer at the bottom of the RT. To mimic a real aquaculture system, 50 Danio rerio were added to each RAS after filling the RT with approximately 50 L of tap water. The water temperature was adjusted and maintained at 25 C using a heater (Sobo, Korea). The MT was made of acrylic panels with two different colors: black panels on the sides to block light from outside and a transparent panel on the bottom to allow light to penetrate from a lighting system. The MT capacity and bottom area were 4 L and 0.12 m2, respectively. The lighting system (T5 8W 3) was attached to the outside of the MT transparent panel. The light intensity was measured using a quantum meter (LI-COR, USA), and set at 40e75 mmol/m2/s in the MT and 85 mmol/m2/s for the subsurface of the RT (12:12 h light-dark cycle). A water pump (Hyubshin, Korea) was used to circulate water between MT and RT. Flow rate between the MT and RT was 1 L/min and the dilution rate in the MT was 0.25 min1. To compensate for water loss due to evaporation in the RAS, distilled water was added using an automatic system. Commercial fish feed, TetraBits® Complete (Tetra GmbH, Germany), was supplied twice a day at 09:00 and 15:00 using an automatic feeder (KWzone, Malaysia), that supplied the total feed at a rate of 0.88 ± 0.26 g/d. 2.2. Microalgae preparation, sampling, and analysis The microalgae strain used in this study was Ettlia sp. KCTC 12109BP, which was obtained from the Korean Collection for Type Cultures (KCTC). The initial seed was cultivated using 4-L photobioreactors with a BG11 medium (Stanier et al., 1971). The harvested microalgae were washed 10 times with distilled water to remove any remaining medium, cell detritus, etc., and added to the treatment MT. After Ettlia sp. flocculated and stacked at the bottom of the MT, the pump was started at a low speed and gradually increased to a final dilution rate of 0.25/min. To determine the growth characteristics of the microalgae in the MT, the dry cell weight (DCW) and chlorophyll-a (chl-a) concentrations were measured in triplicate every 10e15 days using the following procedures. The water circulation was turned off, and following manual mixing, 50 ml of the microalgal suspension was collected from the MT. For comparison, the control microalgal biomass was collected using a scraper to detach the microalgal biofilm. The DCW was measured by filtering an aliquot of the harvested microalgal suspension through pre-weighed GF/C filters (Whatman, United Kingdom). After rinsing with distilled water, the filters were dried at 105 C for 24 h and reweighed. Chl-a was extracted using a chloroform and methanol mixture (2:1, v/v), and measured using a fluorometer (Turner 450, Barnstead/Thermolyne, Dubuque, IA). The biomass productivity was calculated based on the DCW and expressed as [mg DCW/L/d]. Furthermore, the particulate nitrogen (PN) in the MT was measured using a commercial kit (C-mac, Korea) after sonication at a resonance of 10 kHz for 5 min (Chisti and Moo-Young, 1986).
2. Materials and methods 2.3. Water quality analysis 2.1. System and experimental setup This study used a newly designed RAS, which was composed of a rearing tank (RT), biological sponge filters (BSFs), microalgal tank (MT), gravel, and heater (Fig. 1). A total of four RASs, two as controls and two for treatments, were established. The RT consisted of a commercial glass tank (40 40 40 cm) with two BSFs (SP-L4, Aquatech, Korea) connected to an air pump (SHD-60S, Shinhwa Hitech Co, Korea). The water flow rate for the BSFs was
The water quality parameters (temperature, pH, and dissolved oxygen (DO)) in the RT were measured using a portable instrument (Multi 3410, WTW GmbH, Germany). Nitrite, nitrate, and phosphate were determined using an ion-exchange chromatograph with suppressed conductivity (ICS 1600, Dionex, USA) and equipped with an IonPac AS22-HC (Dionex, USA) guard and analytical columns. Total ammonium nitrogen (TAN) was measured using a kit (C-mac, Korea). The turbidity was measured using a turbidity
114
S.-J. Chun et al. / Water Research 135 (2018) 112e121
Fig. 1. (A) Drawing of recirculating aquaculture system (RAS) used in this study, including description of system components. Photographs of microalgae on the surface of MT at day 24: (B) Control; (C) Treatment. Black arrows indicate water circulation direction.
meter (Lutron, Taiwan). The water quality analysis was performed in triplicate every 2e3 days. 2.4. DNA preparation and 16S rRNA gene amplification For the bacterial community analysis, samples were collected from the RT, BSFs, and MT at days 0, 29, and 60. One-liter water samples were collected from the RT and filtered using a sterilized 0.22 mm polycarbonate membrane filter (Millipore Corporation, USA). To analyze the BSF bacterial communities, a quarter of the sponges were detached from the BSFs and squeezed 10 times in autoclaved distilled water (approximately 500 ml); then, 50 ml was filtered to analyze the attached bacteria. After sampling, the sponges were re-attached to the BSFs. Ten milliliters of the microalgal suspension was collected from the MT and filtered as described above. All the membrane filters were stored at 80 C in a deep freezer until DNA extraction. Genomic DNA was extracted using a ChargeSwitch® Forensic DNA Purification Kit (Invitrogen, USA) according to the manufacturer's instructions. The bacterial 16S rRNA gene was amplified using a universal bacterial primer set, 341F/805R (341F:
CCTACGGGNGGCWGCAG; 805R: GACTACHVGGGTATCTAATCC), which targets the V3-V4 regions (Herlemann et al., 2011). The amplicons obtained were then purified using Agencourt AMPure XP beads (Beckman Coulter, USA) according to the manufacturer's instructions. Quantification of the DNA concentrations was performed using a Quant-iT dsDNA HS Assay Kit (Thermo Fisher Scientific, USA), then the purified amplicons were pooled in equimolar concentrations, and then sequenced using high-throughput pairedend Illumina sequencing (MiSeq, 2 250 bp reads) by the Macrogen Corporation (Seoul, South Korea). 2.5. Sequence analysis procedure The resulting sequences were processed using mothur (Schloss et al., 2009) according to the MiSeq standard operating procedure (http://www.mothur.org/wiki/MiSeq_SOP) (Kozich et al., 2013). The Silva database (release 123) was used to align and classify the sequences. All statistical analyses were performed using the R package (version 3.4.0). Briefly, low-quality sequences were removed from the analysis if they contained ambiguous characters, contained more than two mismatches to the forward primer or one
S.-J. Chun et al. / Water Research 135 (2018) 112e121
115
mismatch to the barcode, or were less than 300 bp or more than 500 bp in length. After removing doubletons, the pre-cluster method was applied to further reduce any sequencing errors produced by the Miseq Illumina sequencing platform (Huse et al., 2010). Chimeras were identified and removed using chimera.uchime (Edgar et al., 2011). The average read length was approximately 400 bp after trimming the barcode and primer sequences. A similarity cutoff of >99% was used to assign the same OTUs. The bacterial 16S rRNA gene sequences and accompanying metadata have already been deposited in the Sequence Read Archive (SRA) of NCBI under the project number PRJNA406893. 2.6. Diversity indices, association network and statistical analysis For a visual assessment of the differences between the controls and the treatment tanks, non-metric multidimensional scaling (NMDS) was performed using the R software (package: vegan) (Oksanen et al., 2007). The top 100 most abundant OTUs were selected to calculate the Bray-Curtis dissimilarities. The richness (Chao1 and ACE) and diversity (Shannon's and Simpson's indices) indices were calculated after subsampling using the mothur software package (Schloss et al., 2009). To examine the relationship among the microalgae, water quality parameters (temperature, pH, DO, ammonium, nitrite, nitrate, and phosphate), and bacteria, an association network was constructed based on Spearman's rank correlation coefficient (r), which was calculated using the R software (package: Hmisc) (Harrell et al., 2007). This study used the relative abundance of distinct bacterial orders with median values above 0.01%. The association network was constructed using highly correlated subsets of variables with significant correlations (jrj 0.7, P < 0.05). The network was visualized using open source software Cytoscape 3.5.1 (Shannon et al., 2003). The statistical significance of the clustering was tested using ANOSIM with the vegan package. Microalgal growth characteristics (DCW and chl-a) and water quality data (TAN, nitrite, nitrate, phosphate, temperature, pH, and DO) were analyzed by Student's ttest for comparison between the control and the treatment using Sigmaplot 12.0. Differences were considered significant at a P-value < 0.05. 3. Results 3.1. Growth characteristics of microalgae in microalgal tank (MT) The RASs were operated for a total of 73 days. The nutrient loading in the RAS was completely dependent on the daily added feed, where the feed supply was 0.88 ± 0.26 g/d that included approximately 46.6 ± 13.8 mg/d of nitrogen and 12.5 ± 3.7 mg/d of phosphorus. The inoculated Ettlia sp. became flocculated and stacked at the bottom of the MT (Fig. 1C). In the control, the naturally occurring microalgae (mainly diatoms) attached tightly and formed a microalgal biofilm on the bottom panel of the MT (Fig. 1B). The growth characteristics of the microalgae in both the control and treatment MT are shown in Fig. 2. In the treatment MT, the biomass (DCW) and chl-a concentration increased to 1.26 g/L and 13.85 mg/L, respectively, by the end of the experiment. The DCW and chlorophyll-a concentration in the control were 0.52 g/L and 5.46 mg/L, respectively, both of which were lower than those in the treatment MT. The biomass productivity of Ettlia sp. in the MT was 19.83 mg DCW/L/d. The chl-a concentration and biomass showed a significant correlation (r ¼ 0.99, P < 0.01), suggesting that almost all the DCW consisted of microalgal species. The microscopic observations revealed that the Ettlia sp. was the dominant microalgae species
Fig. 2. Growth characteristics of microalgae in the MT: DCW and Chl-a. Asterisks indicate significant differences in the control and the treatment.
(>99%) in the treatment MT throughout the experiment (Fig. S1B), while diatoms dominated in the control MT (Fig. S1A). At the end of the experiment (day 73), the PN contents were 1.7 ± 0.1 mg-N/g DCW and 22.6 ± 1.4 mg-N/g DCW in the control and treatment MT, respectively. 3.2. Changes in water quality The water temperature, pH, and DO were all monitored during the experiment (Fig. S2). The water temperature was regulated at 24e26 C. The DO and pH ranged from 7 to 8 mg/L and 8.0e8.3, respectively. The turbidity in the rearing tanks remained under 0.1 NTU throughout the experiment for both the control and treatment systems. The concentrations of inorganic compounds in both systems varied largely over the course of the experiment (Fig. 3). Therefore, the experimental data were separated into three different phases based on the fluctuations: phase 1 (0e29 days), phase 2 (30e60 days), and phase 3 (61e73 days). During phase 1, the ammonium concentrations increased rapidly within two days and reached maximum values of 0.44 mg/L and 0.32 mg/L in the control and treatment systems, respectively (Fig. 3A). These ammonium concentrations then decreased sharply, while the nitrite concentrations increased in both systems (Fig. 3B). The maximum nitrite concentration in the treatment system was approximately onefourth of that in the control system. The nitrate concentration in the control system then gradually increased from day 16 until the end of the experiment with a final value of 19.05 mg/L, whereas the nitrate concentration in the treatment system increased from day 16 to day 36 (8.95 mg/L) and was then maintained at the same level until the end of the experiment (Fig. 3C). Thus, the final nitrate concentration in the treatment system was only 50% of the control system. To evaluate whether Ettlia sp. consumed the nitrogenous compounds in the treatment system, we calculated both differences in the corresponding time points of the total dissolved nitrogen (TDN) and DCW between the control and the treatment and designated these differences as DTDN (TDNtreatment TDNcontrol) and DDCW (DCWtreatment MT DCWcontrol MT), respectively. These two values significantly correlated with each other (r ¼ 0.70, P < 0.05) (Fig. 4). At the end of the experiment (day 73), PN concentrations were 0.51 ± 0.04 mg-N/L and 6.69 ± 0.41 mg-N/L in the control and treatment MT, respectively. DPN (difference in PN between the control and treatment, i.e., PNtreatment MT PNcontrol MT) accounted for 65% of DTDN in RAS. The phosphate concentration increased steadily from phase 1 and peaked at an early stage of phase 2, with maximum values of
116
S.-J. Chun et al. / Water Research 135 (2018) 112e121
Fig. 3. Mean concentrations of dissolved nitrogenous compounds and phosphate in systems. (A) TAN, (B) Nitrite, (C) Nitrate, and (D) Phosphate. Water samples were collected from the RT of each RAS. Vertical arrows indicate sampling points for microbial community analysis. Asterisks indicate significant differences in the control and the treatment.
1.22 mg/L and 0.86 mg/L in the control and treatment systems, respectively (Fig. 3D). Similar to the nitrate concentrations, the phosphate concentrations in the treatment system were lower (approximately 30%) than those in the control system. The phosphate concentrations in both systems then declined gradually during phase 2, reaching final concentrations below 0.2 mg/L after 60 days.
3.3. Dynamics of bacterial community compositions The shifts of the bacterial community structure in the three
different components of the RAS were determined using the Illumina MiSeq sequencing approach. A total of 891,264 sequence reads of the 16S rRNA gene were obtained after trimming lowquality sequences, and removing chimeras and single/doubletons. These sequences were further divided into 4,801 OTUs (99% similarity cut-off). The bacterial community compositions showed a significant difference among the MT, BSFs, and RT and among the experimental phases (Fig. 5). Proteobacteria, mainly Xanthomonadales (19.3%), Caulobacterales (15.7%), Burkholderiales (12.0%), Rhizobiales (11.5%), and Oligoflexales (7.5%), dominated the bacterial
S.-J. Chun et al. / Water Research 135 (2018) 112e121
Fig. 4. Correlation between DDCW of MT and DTDN. DDCW ¼ DCWtreatment DTDN ¼ TDNtreatment TDNcontrol.
MT DCWcontrol MT,
community in the Ettlia sp. seed. Among these 6 dominant groups, Xanthomonadales, Caulobacterales, and Burkholderiales decreased to 1.9 ± 0.8%, 1.6 ± 0.4%, and 3.2 ± 0.1%, respectively, at day 60 in the treatment MT, while the relative abundance of some minor groups increased during the experiment. For instance, Rhodospirillales, Phycisphaerae, and Cytophagales represented 12.8 ± 1.5%, 14.3 ± 4.8%, and 12.5 ± 7.6%, respectively, of the bacterial community in the treatment MT at day 60. In addition to these major bacterial groups, Nitrosomonadales and Nitrospirales were observed
117
in the treatment MT at day 60 as minor groups (1% assigned sequences across all samples were used. Color scale represents log-transformed relative abundance of number of sequences. Classes with
