Direct resource recovery from sewage using a combined system of anaerobic-aerobic biological treatment and food production

A combined system of an anaerobic baffled reactor (ABR), a down-flow hanging sponge (DHS) reactor, an aquarium tank (AT), and a constructed wetland (CWL) was proposed as a new concept for sewage treatment. The ABR and DHS reactor, AT, and CWL were applied for biological sewage treatment, bioassay, and nutrient removal with food production, respectively. Killifishes and tomatoes were cultivated in the AT and CWL, respectively. In the ABR, 81.3% of total chemical oxygen demand and 76.5% of total biochemical oxygen demand were removed at 5.1 h of hydraulic retention time (HRT). Most remaining organic matter and 47.1% of ammonia were removed in the DHS reactor. In the CWL, 97.0% of total inorganic nitrogen and 78.6% of phosphate were removed with 3.87 kg/m of tomatoes producing yield at 4.4 days of HRT. In addition, anaerobic ammonium-oxidizing bacteria Candidatus Scalindua and ammonia-oxidizing bacteria Nitrospira and Nitorosococcus were considered as contributors to nitrogen removal in the CWL. The final effluent’s water can be utilized as recycled water by installation of sand filtration and disinfection processes. Therefore, the proposed system can be applied as a low-energy, low-cost sewage treatment system with direct resource recovery.

INTRODUCTION A sewage treatment system is a necessary infrastructure for improving public health. Activated sludge technology is a conventional sewage treatment process, which requires large consumption of energy, cost, and natural resources for both sewage and excess sludge treatment (Guo et al. 2013). In contrast, sewage can be also used to recover energy, nutrients, and water resources (Mo & Zhang 2013). Many studies on energy and resource recovery from sewage sludge have been conducted using anaerobic digestion, composting, and hydrothermal liquefaction (Wei et al. 2018;Du et al. 2019;Das et al. 2020). However, these technologies do not contribute to considerable reduction in power consumption and operating costs for sewage treatment. Recently, studies on direct recovery of energy and nutrients during sewage treatment have been conducted using anaerobic reactors, microbial fuel cells, algae, and aquatic plants (Liu et al. 2017;Nie et al. 2017;Aberysiriwardana-Arachchige et al. 2020;Iwano et al. 2020). Therefore, direct recovery of resources during sewage treatment can become essential for new-generation sewage treatment.
An anaerobic baffled reactor (ABR) and an up-flow anaerobic sludge blanket (UASB) reactor have been applied as anaerobic sewage treatment reactors in developing countries owing to their low cost of operation, construction, and maintenance (Yulistyorini et al. 2019;Passos et al. 2020). Trickling filters, which can be operated aerobically without aeration, were combined with these anaerobic reactors for sewage treatment (Bressani-Riveiro et al. 2019;Mazhar et al. 2021). In trickling filters, worms were grown and contributed to the reduction of excess sludge production in the food chain (Tamis et al. 2011). However, outbreak of worms becomes a sanitary problem during sewage treatment. In contrast, worms also become a natural feed for fishes when the system is combined with an aquarium tank (AT). In addition, fishes can be applied for bioassay (Berrebaan et al. 2020).
A constructed wetland (CWL) can be applied for direct nutrient recovery from sewage. In CWLs, nutrients in sewage can be harvested as food such as vegetables and fruits (Verhofstad et al. 2017). However, the nutrient removal efficiency of CWLs is still not as high in raw sewage treatment because of the high amount of organic matter ( Józẃiakowski et al. 2020).
Therefore, a new concept of sewage treatment based on the combination of biological treatment processes and food production was proposed in this study. The performance of the sewage treatment system, removal efficiency of organic matters, nitrogen and phosphate, methane recovery in the anaerobic reactor, food chain in the AT, food production in the CWL, and water quality of the final effluent were evaluated. Figure 1 shows a schematic diagram of the sewage treatment system used in this study. The sewage treatment system comprises an ABR, a DHS reactor, an AT, and a CWL. In the proposed system, organic matter in sewage was removed in the ABR-DHS system. Fishes in the AT were grown by consuming metazoans in the DHS effluent. Nutrients in sewage were recovered as vegetables in the CWL. The final effluent was considered as recycled water in the sewage treatment plant.

Experimental setup
The ABR was manufactured of polyvinyl chloride pipes (diameter: 30 mm). The ABR includes four compartments with an effective volume of 0.8 L. The DHS reactor was manufactured of a polyethylene terephthalate bottle (volume: 0.8 L) packed with polyurethane sponges as a biomass carrier. The AT and CWL were manufactured from 20 L plastic tanks. Killifishes (Oryzias) were bred in the AT. Tomatoes (Lycopersicum esculentum) were cultivated in the CWL. The landslide of the heavy rain disaster in Hiroshima was utilized as soil for the CWL. Polyurethane sponges were installed near the effluent of the CWL to prevent soil discharge. The granular sludge of the UASB reactor treating food processing wastewater was seeded in the ABR. The DHS reactor was started up without any seed sludge. The system was operated in a temperature-controlled room at 25°C. Synthetic sewage was obtained by mixing soy protein (as proteins), waste rice wine (as carbohydrates), waste oil (as lipids), toilet paper (as cellulose), and waste miso (as minerals) using a similar content ratio of carbohydrates, proteins, lipids, and cellulose in the sewage. The synthetic sewage had a total chemical oxygen demand (COD) of 364 + 93 mg/L, total biochemical oxygen demand (BOD) of 364 + 142 mg/L, ammonia of 21 + 8 mgN/L, phosphate of 1.1 + 0.6 mgP/L, and pH of 6.48 + 0.82. The flow rate of synthetic sewage incrementally increased from 1.9 to 3.7 L/day.

Chemical analysis
Wastewater samples were collected twice weekly from the sewage tank, ABR, DHS, and final effluents. The pH value was measured using a pH meter (Twin pH, Horiba, Japan). Total COD and total nitrogen were analyzed by a Hach apparatus (DR/2500, Hach, USA). Ammonia, nitrite, and nitrate concentrations were measured using an ion chromatograph equipped with a conductivity detector (CDD-10A VP, Shimadzu, Japan). Phosphate, color, and turbidity were analyzed by a digital packtest (DPM-MTSP, Kyoritsu Chemical-Check Lab., Japan). Methane concentrations in ABR biogas were analyzed using a gas chromatograph (GC-8A, Shimadzu, Japan) equipped with a thermal conductivity detector. BOD was analyzed using a standard method (APHA 2012). Coliforms were measured using petrifilm (6404EC, 3M company, USA).

Microbial community analysis
The analysis of microbial communities in DHS effluents and the AT (18S rRNA gene), and in the DHS reactor and final effluent (16S rRNA gene) was conducted. DNA extraction was performed using a MonoFas Bacterial Genomic Kit VII (Animos Corporation, Saitama, Japan). Polymerase chain reaction (PCR) amplification of the 18S rRNA gene V4 and V9 region (Hirakata et al. 2019) and 16S rRNA gene V3-V4 region (Caporaso et al. 2012) was performed using a Sapphire Amp Fast PCR Master Mix (Takara, Shiga, Japan). PCR products were purified using AMPure XP SPRI beads (Beckman). Massive parallel 16S rRNA gene sequencing and functional genes were conducted by the Miseq System (Illumina) using a Miseq Reagent Kit v.3 (2x 300 bp). Taxonomic classification and read count metrics were determined using the 16S Metagenomics Application v.1.0.0.1024. The algorithm used in this program is a high-performance implementation of the Ribosomal Database Project where reads reference an Illumina-curated version of the GreenGenes (May 2013) taxonomic database (Wang et al. 2007) for 16S rRNA gene, and QIIME 2 View for 18S rRNA gene (Bolyen et al. 2019).

RESULTS AND DISCUSSION
Sewage treatment performance of the ABR-DHS system Figure 2 shows the average COD and BOD concentrations of the influent, ABR, and DHS effluents. In the ABR, 81.3 + 12.5% of total COD and 76.5 + 14.6% of total BOD were removed. Finally, 91.0 + 5.5% of total COD and 85.5 + 12.4% of total BOD were removed by the ABR-DHS system. During the experimental period, the HRT of the ABR-DHS system was reduced from 12.2 to 6.1 h. This process performance was similar to that of activated sludge processes (Nakada et al. 2006;Tandukar et al. 2007), ABRs (Krishna et al. 2008;Hahn & Figueroa 2015), and combination systems of anaerobic-aerobic reactors (Nomoto et al. 2018;Mazhar et al. 2021) treating sewage. In the ABR, 0.02 m 3 /m 3 -sewage of biogas was produced at 5.1 h of HRT. By anaerobic digestion of excess sewage sludge, 0.5 m 3 /m 3 -sewage of biogas can be recovered (Cao & Pawtowski 2012;Jenicek et al. 2012;Li et al. 2013). Methane recovery of the ABR is only 4% of anaerobic digestion of excess sewage sludge. However, the ABR-DHS system can be operated without power consumption for aeration and excess sludge production. Therefore, the ABR-DHS system mainly contributes to reduction of power consumption for sewage treatment, not for energy recovery from sewage treatment. Figure 3 shows the average ammonia and nitrate concentrations of the ABR and DHS effluents. Nitrite was not confirmed in the DHS effluent during the entire experimental period. In the DHS reactor, 47.1 + 27.0% of ammonia was removed, and 56.8 + 26.9% of removed ammonia remained in the effluent as nitrate. Nitrogen removal efficiencies of DHS reactors treating sewage were in the range of 30-80% (Tawfik et al. 2006;Tandukar et al. 2007;Onodera et al. 2014). Compared with that in these previous studies, nitrogen removal efficiency of the DHS reactor in this study was lower. However, dissolved oxygen concentration in the AT was kept at 5.74 mg/L on average without aeration at 4.6-9.2 days of the HRT. In addition, killifishes were grown without any feed, and eutrophication was not confirmed in the AT during the entire experimental period. Therefore, the water quality of the DHS effluent was acceptable for autopurification and contributed to the food chain in the AT. Figure 4 shows predominant eukaryotic microorganisms in the DHS effluent and AT. The yield of 18S rRNA PCR product amplified from DNA extracted from the DHS effluent and AT was 12 ng/μL and 0.2 ng/μL, respectively. In the DHS effluent, Chlorophyta (predominant species: Desmodesmus cf. armatus), Apicomplexa, Cecozoa, Nematoda, and Rotifera (predominant species: Habrotrocha bidens) were the predominant phyla. Among these microorganisms, Nematoda, Apicomplexa, and Rotifera were highly reduced in the AT. In addition, Ciliphora (predominant species: Hypotrichidium paraconicum) and Gastrotricha (predominant species: Chaetonotus acanthodes) were increased in the AT. Nematoda and Rotifera were detected as major metazoans in the DHS reactor treating sewage and considered as contributors of excess sludge reduction (Onodera et al. 2013;Miyaoka et al. 2016). In contrast, Ciliphora and Gastrotricha were detected in the oxidation ditch process treating sewage, which has a longer HRT than have the DHS reactor and activated sludge process (Matsunaga et al. 2014). Furthermore, Gastrotricha increased in the sewage treatment process without detection of or with lower detection rate of Nematoda and Rotifera (Matsunaga et al. 2014;Miyaoka et al. 2016). In these metazoans, Gastrotricha has the smallest size (Segers 2008;Blaxter & Koutsovoulos 2015;Todaro et al. 2019). During the entire experimental period, worms were confirmed only in the DHS reactor, not in the AT. Therefore, Nematoda,  Rotifera, and worms were considered as main foods for killifishes in the AT. Furthermore, an outbreak of worms was also prevented by combining the DHS reactor and the AT. Figure 5 shows average total inorganic nitrogen (TIN) and phosphate concentrations of the DHS and final effluents. In the CWL, 97.0 + 4.9% of TIN and 78.6 + 6.9% of phosphate were removed at 4.4 days of HRT. Benvenuti et al. (2018) have applied a real-scale constructed floating wetland using Typha domingensis for domestic sewage treatment and achieved 41% of nitrogen removal and 37% of phosphorous removal at 11.5 days of the HRT. Ye et al. (2012) have applied real-scale ABR stage-CWL using Phragmites autralis, Canna indica, and Cyperus alternifolius and achieved 31% of nitrogen removal and 34% of phosphorous removal at 5.2 days of the HRT. Compared with the results of these studies, both nitrogen and phosphorous removal efficiencies in this study were much higher. The production yield of tomatoes in this study was 3.87 kg/m 2 . This value was approximately five times higher than that obtained in a pilot-scale CWL treating domestic wastewater (0.73 kg/m 2 ; Caselles-Osorio et al. 2018). Furthermore, the obtained value was similar to the conventional production yield (5-60 kg/m 2 ) for organic tomatoes (Bosona & Gebresenbet 2018;Liang et al. 2019;Ronga et al. 2019). Nitrogen and phosphate contents of tomato fruits were 892 mgN/kg-fruits and 392 mgP/kg-fruits, respectively. In terms of mass balance, 3.0% of input nitrogen and 5.1% of input phosphate were harvested as tomato fruits. In addition, 4.0% of input nitrogen and 1.4% of input phosphate were accumulated as sludge at the bottom of the AT. As the results of microbial community analysis (yield of 16S rRNA PCR product: 5.1 ng/μL), anaerobic ammoniumoxidizing bacteria Candidatus Scalindua (1.0%) and ammonia-oxidizing bacteria Nitrospira (0.6%) and Nitorosococcus (0.2%) were detected in the final effluent (detection rates of total sequence reads). In this study, most organic matter was degraded in the ABR-DHS system; then, the CWL was mainly used for nutrient removal. Therefore, in this study, low organic loading rate contributed to high nutrient removal efficiency by both microbes and cultivation of tomatoes in CWL.

Nutrient removal and food production in the CWL
Applicability of the final effluent for water recycling Table 1 shows the water quality of the final effluent and Japanese quality standards for discharge and recycled water. The final effluent achieved the Japanese quality standard for discharge. However, only pH and odor parameters achieved the Japanese quality standard for recycled water. The color of the final effluent was the color of organic matter (e.g., soy protein and miso) in synthetic sewage. The coliform count and turbidity of the final effluent had values similar to those of the secondary effluent of sewage treatment plants (Uemura et al. 2002;Qin et al. 2006). Figure 6 shows time course of turbidity and coliform of final effluent. In terms of analytical results, 80% of coliform count was below 5 CFU/mL, and 40% of turbidity was below 2. Thus, sporadically high values highly affected the average value of these parameters. The results of microbial community analysis showed that some pathogenic bacteria, such as Oligella ureolytica and Rickettsia (Weinert et al. 2009;Demir & Celenk 2014), were also detected in approximately 1% of total sequence reads in the final effluent. Therefore, post-treatment processes (e.g., sand filtration and disinfection) were required before using the final effluent as recycled water.

CONCLUSIONS
The results of this study indicated the high potential of the proposed sewage treatment system for pollutant removal and resource recovery from sewage. The ABR-DHS system achieved high performance of organic matter removal with biogas recovery at an HRT similar to that of activated sludge processes with a much lower power consumption. The water quality of DHS effluents was adequate for both autopurification and the food chain in the AT. In addition, the AT was used for bioassay and removal of worms, which occur in the DHS reactor. The CWL contributed to nutrient removal, food production, and water purification. The final effluent achieved the Japanese quality standards for discharge. However, additional post-treatments for removal of turbidity and pathogens, such as sand filtration and disinfection, were required to use the final effluent as recycled water.
The results of this study indicated an applicability of the proposed system for low-energy and low-cost sewage treatment with direct resource recovery. Furthermore, the proposed system can be applied by renovation of present sewage treatment systems such as activated sludge process. For practical use of the proposed system, some problems such as operation and maintenance of the system, climate property of the area, adaptable species for AT and CWL, and economical advantage, should be considered. As the next step, a pilot-scale experiment will be conducted to clarify these problems in future work.

ACKNOWLEDGEMENT
This work was supported by the Fuso innovative technology fund.

DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.