Earthy/musty odour bioaccumulation in Plecoglossus altivelis grown in a land-based aquaculture fed by reclaimed secondary effluent in wastewater treatment plant

As the gap between fish supply and demand widens, global land-based aquaculture production has grown rapidly and now accounts for more than half of total fisheries production, marking a shift from marine capture to aquaculture (FAO 2020). However, the reliance on local water sources has led to a slowdown in growth, mainly due to freshwater scarcity. To address this, reclaimed wastewater (RWW) from municipal wastewater treatment plants (WWTPs) is being explored as an alternative water source. Zaibel et al. (2021) demonstrated the feasibility of using RWW in aquaculture, showing positive effects on fish growth and confirming its safety for human consumption. Furthermore, utilizing RWW in aquaculture offers economic and environmental benefits. Despite being an ancient practice in various regions of the world, the reuse of RWW for aquaculture has remained unexplored.

On the other hand, emerging pollutants in wastewater effluent, such as organic micropollutants (OMPs) like pharmaceuticals and endocrine disruptors, can affect fish reproduction, causing developmental issues and reduced embryo production (Liney et al. 2006; Galus et al. 2013). While the effects of these pollutants are known, the impacts of volatile odour compounds (VOCs) like geosmin (GSM), 2-methylisoborneol (MIB), and 2,4,6-trichloroanisole (TCA) found in WWTP effluents have not been well studied. These VOCs are the primary compounds contributing to the odour profile of municipal wastewater effluent (Agus et al. 2011; Urase & Sasaki 2013). Additionally, the effects of VOCs (GSM and MIB) on fish quality are often discussed in conventional recirculating aquaculture systems (RASs) (Burr et al. 2012; Abd El-Hack et al. 2022; Moretto et al. 2022). However, studies on their effects in RWW-fed aquaculture, especially TCA, are still limited.

Earthy or musty odours in fish, especially aquaculture products, are a global problem. The previous study highlighted the main odour-causing compounds, GSM and MIB, which impart these odours to water and aquatic organisms. GSM and MIB are produced by bacterials (actinobacteria (mainly Streptomyces), myxobacteria and cyanobacteria) and fungi (Abd El-Hack et al. 2022). Although non-toxic, these VOCs are not completely removed in WWTPs and are readily transferred to organisms, accumulate in fish and can affect meat quality (Lin et al. 2002; Agus et al. 2011; Moretto et al. 2022). Bioaccumulation of these undesirable VOCs has been shown in Atlantic salmon and related species in aquaculture systems, resulting in disposal costs and loss of revenue (Burr et al. 2012). In conjunction with GSM and MIB, TCA is likely implicated in the development of musty odours in wastewater from treatment plants and water distribution systems receiving effluent from WWTPs (Agus et al. 2011; Takeuchi et al. 2013; Urase & Sasaki 2013; Zhang et al. 2016). TCA has also been detected in water supply reservoirs and drinking water treatment plants at concentrations at or below its odour threshold value (Bai et al. 2017). On the other hand, TCA is a potent compound responsible for the cork taint in wine, the musty odour of raisins, broiler chicken, coffee bean, and packaging materials (Land et al. 1975; Aung & Jenner 2004; Kato et al. 2011; Monteiro 2022). However, studies on the bioaccumulation of TCA and its impact on the olfactory quality of fish, particularly those cultured in RWW, are limited. The proliferation of WWTPs in developed cities may increase the presence of untreated odour compounds, such as TCA, in aquatic environments. This preliminary study will provide empirical data on TCA bioaccumulation and its effects on the quality of aquatic animals, specifically freshwater fish.

Plecoglossus altivelis, commonly known as Ayu sweetfish or Ayu in Japanese, is an amphidromous osmerid fish with a 1-year life cycle. Ayu has a life history in which juveniles and adults inhabit the middle reaches, while mature fish migrate downstream to spawn in summer and autumn (Pastene et al. 1991). In Japan, Ayu is said to be a symbol of a clear river and renowned for its unique flavour, often likened to watermelon (Nagayama et al. 2022). However, its natural population has been declining due to pollution of urban water (Takeuchi et al. 2021). In developed cities such as Tokyo, municipal wastewater treatment has grown significantly to ensure the environmental safety of water. In Tokyo, treated wastewater was reported to account for 32% of the river flow in the lower reaches of the Tama River and 71% of the river flow in the lower reaches of the Sumida River (Furumai 2008). Municipal WWTPs are therefore essential for sustainable management of urban water environments and provide opportunities for resource recovery, including water, heat, and nutrients. Advanced wastewater treatment technologies that focus on nutrient removal/recovery and sterilization, contribute to ecosystem conservation, particularly in aquatic environments, and support community development. Despite the importance of sensory perception in assessing the quality of treated water, research on odour removal technology for WWTP effluent has not garnered significant attention (Sado-Inanuma & Fukushi 2018).

This study investigates the feasibility and safety of using RWW for aquaculture, specifically focusing on its impact on fish growth and the bioaccumulation of VOCs such as TCA, MIB, and GSM in fish cultured with RWW. It also examines the effects of these VOCs on the olfactory quality of the fish. Notably, this is also the first study to propose using treated wastewater to farm Ayu sweetfish in a self-constructed aquaculture system at a WWTP in Tokyo. By utilizing water resources and treatment technology at the plant, this system offers continuous water flow, eliminating the need for water recirculation found in RASs. The system is also expected to improve energy efficiency compared with traditional land farming systems. Moreover, the cultivation of Ayu, a symbol of pure streams, with RWW contributes to expanding a positive image of wastewater reclamation in Japan.

Key parameters of RWW quality, including temperature, dissolved oxygen (DO), ammonium (NH₄-N), and nitrate (NO₃-N) concentrations, were investigated. The study also assessed the bioaccumulation of undesirable VOCs (TCA, GSM, and MIB) to evaluate their negative impacts on fish quality. Nutritional composition, desirable flavours (cucumber-like (CCO) and watermelon-like odours (WMO)), and heavy metals (chromium (Cr), arsenic (As), lead (Pb), cadmium (Cd), and total mercury (Hg)) were also evaluated to ensure fish safety. In addition, a pilot-scale land-based fish-farming system was developed using flow-through secondary effluent as the primary water source at a Tokyo WWTP.

The tank construction materials are fibreglass reinforced plastic (FRP), chosen for its durability and weather resistance. The receive tank is rectangular shape (126 cm × 105 cm with a height of 67 cm), while the moderation and culture tanks are circular (136 cm in diameter and 75 cm in depth). An automatic feeder is installed in the culture tank to provide fish feed. Key parameters such as temperature, DO, pH, NH₄-N, and NO₃-N concentrations of rearing water in the culture tank were regularly monitored. This monitoring was facilitated by a sophisticated remote management transmission system (Figure 1). This advanced system integrated water quality monitoring and control using cellular packet communications. In particular, the system was designed to alert managers via email in the event of any abnormal changes in water quality.

Two different culture trials were conducted in the summer and autumn seasons, lasting for 95 and 120 days, respectively. The specific culture conditions of two trials are detailed in Supplementary Table S2. In the summer trial (August to November 2021), 500 juvenile Ayu with an average body weight of 2.84 ± 0.6 g and an average body length of 6.53 ± 0.37 cm were cultured. The trial involved maintaining a flow rate of rearing water between 5.6 and 66.6 L min⁻¹, coupled with a feeding rate 0.2–0.8 g day⁻¹ fish⁻¹. Meanwhile, the autumn trial (November 2021–March 2022) included 250 juvenile Ayu with an average body weight of 0.72 ± 0.2 g and an average body length of 4.96 ± 0.3 cm. In this trial, a flow rate of 33.3–66.6 L min⁻¹ and a feeding rate of 0.1–0.7 g day⁻¹ fish⁻¹ were maintained.

A commercial granulated diet containing approximately 49–55% crude protein, 4–10% crude lipid, 4% crude fibre, 16% ash, 1.4–1.5% calcium, and 1.1–1.5% phosphorus was provided as the diet. The feeding rate in the culture tank was calculated on the basis of the number of fish, individual body weight, and water temperature. During the first few weeks of both culture trials, desalinated tap water was mixed with RWW to aid fish acclimatization. In the autumn trial, LED lighting was used during the night to delay fish maturation, as shown in Supplementary Table S2. In both trials, an automatic feeding device was used to adjust feeding time and amount according to fish growth.

Due to resource limitations, a control fish-farming system using river or underground water as a control treatment was not feasible. Instead, commercially cultured Ayu from local markets and wild Ayu caught in local rivers were used as a comparative benchmark in this study.

Water samples were collected weekly from the culture tank (after sterilization of secondary effluent using UV) and then refrigerated at 4 °C for subsequent analysis. Fish growth, including body weight and length, was assessed every 2 weeks. In addition, fish samples were collected at the end of each experiment. Fish flesh was carefully isolated, diced, and then frozen at −20 °C to facilitate subsequent analysis. A composite sample was taken from a minimum of three harvested fish to ensure homogeneity. In September 2021, samples of Ayu from commercial farms (cultured with spring water), and wild Ayu caught during the summer, were obtained. These samples were subjected to the same pre-treatment procedures as the RWW-cultured Ayu, allowing a comparative assessment of odour, nutritional composition, and heavy metal concentrations.

Information of the chemical reagents used in the analysis is given in Supplementary Table S3.

Measurement of the VOC concentrations was performed using a gas chromatograph-mass spectrometer (GC-MS, JMS-Q1500GC, JEOL, Japan) coupled with dynamic headspace extraction (HS:MS-62070 STRAP of JEOL build with autosampler 7890GC, Agilent, Japan). The HS extraction technique was selected due to its ability to detect VOC levels at ultra-trace levels (in the range of ng L⁻¹) while maintaining a simple sample preparation process (Vilar et al. 2022). In the extraction process, the sealed vial was heated at 80 °C for 30 min before extraction for 20 min by the solid phase extraction process (SPE). The volatile compounds were desorbed via GC-MS, and the analytes were separated and detected using a capillary column (INERT CAP1, 30 m × 0.25 mm × 0.4 μm). A split ratio of 7:1 was used for injection.

Specifically, the oven temperature was initially set at 40 °C for 3 min, then increased at a rate of 5 °C/min until it reached 150 °C, followed by a further increase at a rate of 30 °C/min until it reached 280 °C, and then held at this temperature for 5 min. The quantitative ions for MIB, GSM, and TCA were identified as m/z = 95, 107, 108, m/z = 111, 112, 125, and m/z = 195, 210, and 217, respectively. The molar ratio of peak areas between the analytes and the internal standard was used for quantification. Standard solutions with concentrations of 0, 1, 2, 5, 10, and 20 ng L⁻¹ were prepared for calibration purposes. The limits of detection (LOD) and limits of quantification (LOQ) for the undesirable VOCs are given in Supplementary Table S4. Calibration solutions for MIB, GSM, and TCA were prepared by adding known concentrations of the analytes (ranging from 0, 1, 2, 5, 10, and 20 ng L⁻¹) to a methanol solution (99.8% Fujifilm Wako Chemicals Co., Inc., Japan).

To determine the recovery rate and chemical purity of each compound, a standard mixture was prepared both spiked and unspiked with 5 ng L⁻¹ of MIB, GSM, or TCA as a quality control measure. Control samples for analysis were obtained from tuna flesh samples obtained from a local market and confirmed to be free of the targeted VOCs. Water samples spiked with 0 and 5 ng L⁻¹ of the targeted VOCs were used to determine the degree of purity of the chemicals and to calculate the recovery rate of the analysis. Evian bottled mineral water (Cachat-SAEME, Evian, France), which is known to be free of anthropogenic contaminants, was used as the VOC-free water (Tamtam et al. 2009). Supplementary Table S1 details the properties of GSM, MIB, and TCA.

Temperature, pH, and DO were measured using portable multimeters (HORIBA LAQUA PH210, Japan and Model KRK, DO-30N, Japan, respectively). Ion concentrations (Na⁺, K⁺, Mg²⁺, Ca²⁺, ⁠, Cl⁻, ⁠, ⁠, and ⁠) were determined by ion chromatography using a Dionex Integrion HPIC system (ThermoFisher Scientific, Japan). Chemical oxygen demand by dichromate (COD_Cr) was measured according to EPA method 410.4 (O'Dell 1993). Concentrations of SS and total phosphorus (TP) followed standard methods for water and wastewater treatment (Baird & Bridgewater 2017). For the enumeration of microbial indicators, the membrane filtration method (USEPA approved Hach Co.: Method 10029) was used to enumerate E. coli and total coliforms. It's worth noting that un-ionized ammonia (NH₃-N) is the major metabolite resulting from protein metabolism in fish and is excreted mainly through the gills. The toxicity of NH₃ to fish is determined by its concentration, which can be calculated using the total ammonia nitrogen and ionization fraction, which is highly dependent on the pH and temperature of the water (Colt 2006). The calculation of un-ionized NH₃-N concentration in rearing water is presented in Supplementary Table S10.

Ten millilitres of water samples were placed into a vial containing 4 g NaCl, to which 5 μL of 2,4,6-trichloroanisole-d3 was added as an internal standard for measuring the concentrations of the undesirable VOCs, including TCA, MIB, and GSM.

At least 20 fish were captured every 2 weeks for measurement of body weight and length. A 10,000-fold diluted anaesthetic solution (FA100, Sumitomo Pharma Animal Health, Japan) was used to induce short-term anaesthesia during the growth assessment process. The absolute growth rates of body length and weight, referred to as ALR and AWR, respectively, were calculated using the regularly collected Ayu body length and weight data (Lugert et al. 2014). The number of dead fish was recorded daily during farming system maintenance. The daily fish death rate (RFD) was calculated by dividing the daily number of fish that died by the total number of fish in the tank. The calculation of ALR, AWR, and RFD are presented in Supplementary Table S10.

The pre-treatment procedure, according to established protocols, was adapted from an analytical method typically used for the determination of releasable VOCs in food (Lindholm-Lehto 2022). Fish flesh samples of 0.05–0.1 g wet weight were carefully placed in a 20 mL headspace vial pre-filled with 4 g sodium chloride (NaCl). After the addition of 5 μL internal standard (2,4,6-trichloroanisole-d3, 10 ng L⁻¹ in methanol >99%) and 10 mL VOC-free water, the vial was tightly sealed with polytetrafluoroethylene septum caps. TCA, MIB, and GSM were determined using the headspace GC-MS method as described.

The measurements of desirable VOC concentrations, namely trans-2,cis-6-nonadienal (CCO) and trans,cis-3,6-nonadien-1-ol (WMO) in fish tissues were performed by an external laboratory belonging to Tokyo Metropolitan Industrial Technology Research Institute. The analytical set-up consisted of a GC-MS apparatus (GCMS-TQ8050 NX, Shimadzu, Japan) equipped with an automatic solid-phase microextraction device (SPME, DVB/PDMS, ø1.1 mm, Agilent build with a Shimadzu AOC-6000 autosampler). Meanwhile, 1,4-dichlorobenzene-d4 was used as an internal standard. For reference, Supplementary Table S4 provides details of the LOD and LOQ for these desired VOCs, while Supplementary Table S5 outlines the specific extraction conditions used.

The measurement of the nutritional composition and heavy metal content of the fish flesh samples was carried out by the Japan Food Research Laboratories according to established standard methods (Fujita et al. 2019). Specifically, the protein content was determined using the Kjeldahl method, with the amount calculated as 6.25 times the nitrogen (N) content. Energy content was determined using Atwater's coefficient, with a conversion factor of 4 for proteins and carbohydrates and 9 for fat. Total carbohydrate content was estimated by subtracting the measured moisture, protein, fat and ash from the total weight, while fibre content was obtained by subtracting each of the other components from the total weight. The salt equivalent content was estimated by multiplying the measured sodium content by 2.54. Vitamin and fatty acid (FA) contents in fish samples were analysed by high-performance liquid chromatography (HPLC) and GC-MS, respectively (Saito et al. 1999). Concentrations of arsenic (As), lead (Pb), cadmium (Cd), total mercury (Hg), and chromium (Cr) in Ayu flesh samples (on a dry weight basis) were determined by digestion methods approved by the Japanese Ministry of Environment, followed by analysis by atomic absorption spectroscopy (AAS) and inductively coupled plasma mass spectrometry (ICP-MS).

Microsoft Excel (version 16.67, Microsoft, Redmond, USA) and RStudio (version 2022.07.2 + 576) were used to calculate averages and standard deviations and to perform statistical analyses such as Student's t-test and one-way ANOVA. Pearson correlation and principal component analysis (PCA) were performed within RStudio to identify relationships between average values of water quality parameters, ALR, AWR, and RFD. PCA retained the first four principal components (PCs) that together accounted for at least 80% of the total variance to be discussed. To avoid misclassification of variables, a z-scale transformation was applied to standardize the data. Kaiser-Mayer-Olkin (KMO) and Bartlett's spherical tests were performed to validate the PCA. The complete extraction of VOCs in the aqueous phase of fish tissues was assumed to be achieved by the pre-treatment process in the GC-MS analysis. Therefore, the releasable concentrations of TCA, GSM, and MIB in fish tissues (expressed in μg kg⁻¹) were calculated by dividing the measured VOC concentrations by the wet weight of the fish tissues (Supplementary Table S10).

Table 1 shows the average values and standard deviations for the physicochemical parameters of the rearing water in the culture tank. Criteria values for Ayu culture were adopted, including pH, temperature (autumn trial), DO, SS, NH₄-N (summer trial), NO₃-N, and total coliform. The average pH in the summer trial was slightly higher than that of the autumn trial.

Table 1

Physicochemical parameters (average ± SD) of water samples in culture tank

Parameters .	n .	Summer trial .	Autumn trial .	Optimum condition .
pH	12–16	7.4 ± 0.4	7.1 ± 0.2	6.7–7.5(1)
Temperature (°C)	12–16	26.6 ± 1.7	19.6 ± 2.1	15–25(2)
DO (mg L−1)	12–16	7.6 ± 1.0	8.9 ± 1.8	>7(1)
SS (mg L−1)	3	9 ± 3.7	–	<25(1)
-N (mg L−1)	12–16	0.2 ± 0.3	1.2 ± 2.7	<1(3)
NH3-N (mg L−1)	12–16	0.008 ± 0.016	0.006 ± 0.011	<0.0125(3)
-N (mg L−1)	12–16	6.1 ± 3.2	6.5 ± 2.3	NA
CODCr (mg L−1)	12–16	14.8 ± 4.7	18.7 ± 6.0	NA
TP (mg L−1)	3	0.5 ± 0.1	0.7 ± 0.4	NA
E. coli (CFU/100 mL)	3	–	–	NA
Total coliform (CFU/100 mL)	3	19 ± 27	28 ± 4	<1,000(1)

Parameters .	n .	Summer trial .	Autumn trial .	Optimum condition .
pH	12–16	7.4 ± 0.4	7.1 ± 0.2	6.7–7.5(1)
Temperature (°C)	12–16	26.6 ± 1.7	19.6 ± 2.1	15–25(2)
DO (mg L−1)	12–16	7.6 ± 1.0	8.9 ± 1.8	>7(1)
SS (mg L−1)	3	9 ± 3.7	–	<25(1)
-N (mg L−1)	12–16	0.2 ± 0.3	1.2 ± 2.7	<1(3)
NH3-N (mg L−1)	12–16	0.008 ± 0.016	0.006 ± 0.011	<0.0125(3)
-N (mg L−1)	12–16	6.1 ± 3.2	6.5 ± 2.3	NA
CODCr (mg L−1)	12–16	14.8 ± 4.7	18.7 ± 6.0	NA
TP (mg L−1)	3	0.5 ± 0.1	0.7 ± 0.4	NA
E. coli (CFU/100 mL)	3	–	–	NA
Total coliform (CFU/100 mL)	3	19 ± 27	28 ± 4	<1,000(1)

SD, standard deviation; n, number of samples; DO, dissolved oxygen; SS, suspended solids; COD, chemical oxygen demand; TP, total phosphorus; NA, not available.

(1) Water quality standard for fisheries using water from river (JFRCA 2012), (2) Upper water temperature limit suitable for fish growth (Nagayama et al. 2022), (3) Permissible level for fish culture (Meade 1985).

View Large

Fish survival, reproduction, growth, and production are all affected by water quality. Temperature plays a critical role in the aquatic environment because it regulates many biological and physicochemical processes. Even a modest temperature change of 5 °C can cause stress or mortality in fish (Boyd & Tucker 1998). An ideal water temperature of 25 °C promotes Ayu growth, with increased territorial behaviour observed above 20 °C (Nagayama et al. 2022). Although average water temperatures in both trials were within the range suitable for Ayu rearing, elevated RWW temperatures during certain periods of the summer trials may have affected the chemical and biological responses of the fish. During 42–59 rearing days (data not shown), the maximum RWW temperature in the summer trial exceeded 30 °C, resulting in a greater temperature difference (3.1–8.1 °C) between day and night compared with other times. In contrast, the maximum water temperature in the autumn trial did not exceed 27°C, with the temperature difference between day and night generally less than 7°C for most of this culture time (data not shown). The autumn RWW temperature was more favourable for Ayu growth, as indicated by the higher body length and weight in the autumn trial than in the summer trial (p < 0.05, Figure 2).

The average water temperature and NH₃-N concentration in the summer trial were higher than those in the autumn trial. Meanwhile, the average DO, ⁠, and values in the autumn trial were higher than those in the summer trial (Figure 1). Other water quality parameters, such as SS, COD_Cr, TP, total coliforms, and other ions (shown in Supplementary Table S6), were also measured on a regular basis to ensure that there were no unexpected changes that would harm fish growth.

Table 2 shows the loadings and eigenvalues of the retained PCs from the PCA on rearing water and Ayu growth parameters during the summer and autumn trials. Variances and component matrices were calculated using FactoMineR and factoextra packages in RStudio. PCA showed valid results with KMO values > 0.5 and significant Bartlett's test (p < 0.001) in both trials. In the summer trial, PC1, PC2, PC3, and PC4 accounted for 38.4, 27, 11.2, and 10.3% of the total variance, respectively. Positive or negative loadings on PCs indicate relative correlations between variables. AWR-DO and RFD-pH-NH₄-NH₃ (absolute loading > 0.4) showed a positive correlation on PC1, implying that an increase in DO can increase the body weight of cultured fish, while an increase in pH and ammonia can cause an increase in fish mortality. Conversely, RWW temperature showed a negative correlation with DO, ALR, and AWR (absolute loading > 0.5) on PC2 (Table 2a), indicating a negative effect of water temperature on fish growth. In the autumn trial, PC1, PC2, PC3, and PC4 accounted for 36.9, 22.5, 14.3, and 9.6% of the total variance, respectively. In autumn, PC1 showed a positive correlation (absolute loading > 0.5) of ALR, AWR, NH₄, and NH₃, and a negative correlation of RWW temperature with ALR and AWR (absolute loading > 0.4, Table 2b). These results were inconsistent with those from the summer trials, where temperature and rearing water ammonia concentrations were not related to changes in fish growth. This suggests that the factors of temperature and ammonia were not significant in the autumn trial. Biplot graphs illustrating water quality parameters and Ayu growth in PC1 and PC2 for both seasons are shown in Supplementary Figure S2.

Supplementary Figure S3 illustrates the Pearson correlation coefficient between Ayu growth and water quality parameters. In the summer experiment, significant negative correlations (p < 0.01) were observed between AWR and pH, NH₄-N, and NH₃-N, and between ALR and NO₃-N (p < 0.01). AWR showed a significant positive correlation (p < 0.001) with DO. Daily rate of fish death (RFD) was significantly correlated with water pH (p < 0.05, Supplementary Figure S3(a)). In the autumn experiment, AWR was significantly correlated with NH₄-N and NH₃-N (p < 0.01), while ALR was significantly correlated with NO₃-N (p < 0.05). Water pH had a significant negative correlation with RFD (p < 0.05). No significant correlations were found between DO and growth parameters in either experiment (p > 0.05). Water temperature was not significantly correlated with growth parameters in both (p > 0.05, Supplementary Figure S3(b)). In summary, the significant correlations from the results of PCA and Pearson's correlation analysis confirm the notable impact of ammonia concentration and temperature of RWW water on fish growth, particularly during the summer.

Table 3a shows the average concentrations of earthy/musty VOC compounds in the rearing water from both trials. The recovery rates of TCA, MIB, and GSM were 110 ± 23, 90 ± 35, and 80 ± 24%, respectively. The average TCA, MIB, and GSM concentrations in the summer trial were 3.2, 6.2, and 9.9 ng L⁻¹, respectively. The autumn trial results were 18.3, 17.2, and 19.2 ng L⁻¹, respectively. Although the average concentrations of the target VOCs measured in the autumn trial were higher than those measured in the summer trial, only TCA showed a statistically significant difference (p < 0.05). The VOC concentrations in WWTP secondary effluent in our study were comparable to previous studies. Agus et al. (2011) reported the median concentration of TCA, MIB, and GSM in effluent from a WWTP using biological activated carbon to be 9.5, 11, and 27 ng L⁻¹, respectively. Meanwhile, Urase & Sasaki (2013) reported 4.3–37.7 ng L⁻¹ for TCA and 3.7–42.4 ng L⁻¹ for GSM in effluent from a WWTP using the activated sludge process. Furthermore, in a study on Nile tilapia cultured in net cages in Brazil, GSM and MIB concentrations in water ranged from 0.6 to 7.5 ng L⁻¹ and 1.5 to 24 ng L⁻¹, respectively (Podduturi et al. 2023). In another study of tilapia farms in tropical Brazil, GSM and MIB concentrations in water ranged from 50 to 200 ng L⁻¹ and 470 to 1,180 ng L⁻¹, respectively (Lopes et al. 2022). To our knowledge, very few studies have reported TCA in water used in aquaculture systems. Lindholm-Lehto (2022) investigated 14 off-flavour-inducing compounds in RAS-farmed fishes. Average concentrations of GSM and MIB concentrations in water of rearing tanks were detected at 15.4 and 74.9 ng L⁻¹, respectively, while TCA concentration were undetectable. Concentrations lower than 15 ng L⁻¹ GSM and MIB in water can induce undesirable odours in fish (Howgate 2004); however, the study indicated that off-flavours in fish can also be caused by compounds other than GSM and MIB (Lindholm-Lehto 2022).

Additionally, our findings support the notion that TCA and GSM are the primary VOCs responsible for the earthy/musty odour in WWTP effluent (Agus et al. 2011; Urase & Sasaki 2013). These results also suggest that UV treatment alone is not effective in reducing earthy/musty VOCs in secondary effluent below their odour threshold concentrations (OTCs). Ozonation followed by biological activated carbon and/or UV/H₂O₂ applied as an advanced wastewater treatment process were proposed to be more effective in removing the VOCs (Agus et al. 2011).

A normal person can detect an earthy/musty odour from wastewater effluent when the concentration of one of these VOCs (TCA, MIB, and GSM) in the water samples exceeds the OTCs. Relative odour intensity (ROI) quantifies which odour compounds pose the greatest sensory concern. ROI is calculated by dividing the measured concentrations by the lowest reported OTCs (Agus et al. 2011). ROI of RWW TCA, MIB, and GSM (ROI_water) were 32–183, 1.24–3.44, and 9.9–19.2, respectively (Table 3a). These findings suggest that TCA and GSM are of significant sensory concern in secondary effluent, consistent with previous reports (Agus et al. 2011; Urase & Sasaki 2013). Additionally, the ROI results indicated that TCA may have a greater impact than GSM in causing undesirable odours in wastewater effluent, irrespective of seasonal change.

TCA can be produced from 2,4,6-trichlorophenol (TCP), a widely used fungicide, herbicide, insecticide, and antiseptic. Numerous bacterium (e.g., Rhodococcus, Acinetobacter, and Pseudomonas strains) and fungi (Trichoderma strains) in biological reaction tanks can produce TCA (Agus et al. 2011; Urase & Sasaki 2013). To test the hypothesis that halophenols served as precursors for haloanisoles during biological wastewater treatment, Agus et al. (2011) conducted batch experiments using ¹³C-labeled TCP, showing a 5% molar yield of TCA after 24-h experiment. The O-methylation process in these microorganisms likely serves as a detoxification mechanism, as TCA is much less toxic than TCP (Zhang et al. 2016). However, TCP's formation in wastewater influent and its conversion to TCA during wastewater treatment are not well understood. On the other hand, TCP, banned in cosmetics in the US and Europe, can form from triclosan reacting with free chlorine (Matthew Fiss et al. 2007). Triclosan is an antimicrobial agent found in many common household hygienic products while free chlorine is used for microbial disinfection during water treatment. The WWTP in this study collected municipal wastewater with no effect of rainwater and agricultural runoff. Therefore, we hypothesized that TCP can be created from household activities such as laundry and/or dish washing. Consequently, the TCA formation from TCP during wastewater treatment is a primary source of TCA in WWTP effluent. In urban rivers like those in Tokyo, which receive mostly WWTP effluent, TCA can be a good indicator of WWTP impact on the aquatic environment.

The average concentrations of undesirable VOCs in Ayu tissues are presented in Table 3b. The concentrations of TCA, MIB, and GSM in the flesh tissues of RWW-cultured Ayu ranged from 4.7 to 5.7 μg kg⁻¹, 0.3 to 40.5 μg kg⁻¹, and 1.1 to 1.9 μg kg⁻¹, respectively. No significant difference was observed in the average concentrations of TCA and GSM in flesh tissues between the summer and autumn trials (p > 0.05). However, the concentration of MIB in flesh tissues was significantly higher in the summer trial than that in the autumn trial (p < 0.05). Compared with previous studies, the concentrations of MIB and GSM in RWW-cultured fish tissues in this study were remarkably higher than those reported in previous studies on the culture of other fish species. For example, in the studies on tilapia culture using river water, Lopes et al. (2022) reported 0.05–0.1 μg kg⁻¹ for GSM and 0.1–1.7 μg kg⁻¹ for MIB in fish flesh, while Podduturi et al. (2023) measured 0.05–0.8 μg kg⁻¹ for GSM and 0.1–0.8 ng kg⁻¹ for MIB in fish flesh. To our knowledge, there are no reports on TCA concentrations in cultured fish tissues. Lindholm-Lehto (2022) measured undesirable odour compounds in fish cultured in a RAS but TCA was not found.

Due to the lipophilic nature of earthy/musty VOCs, these compounds can be taken up by fish through dietary intake and passive diffusion through culture water (Abd El-Hack et al. 2022). The concentrations of TCA, MIB, and GSM in RWW-cultured Ayu tissues were respectively 23.5–28.5, 0.3–33.8, and 2.2–3.8 times higher than those in commercial Ayu. The average concentration of fish MIB in fish from the summer trial was above the expected threshold concentration suggested by Tucker (2000), while fish MIB in the autumn trial and GSM were not. However, this comparison is only relative because the OTCs may differ between fish species. If the concentration of VOCs in commercial Ayu is used as a reference in which there is no effect of selected VOCs, ROI for fish tissue can be calculated as shown in Table 3b. ROI of fish TCA, MIB, and GSM (ROI_flesh) were 23.5–28.5, 0.25–33.8, and 2.2–3.8, respectively (Table 3b). ROI_flesh of TCA was remarkably higher than that of GSM and autumn MIB. These results indicated that TCA may have the most pronounced effects on the odour of fish cultured with RWW regardless of seasonal change.

Supplementary Table S7 shows the average concentrations of desirable VOCs, namely CCO and WMO in fish tissues. The average concentration of CCO in flesh tissues of RWW-cultured Ayu was comparable to that found in wild and cultured Ayu purchased from a local market. Moreover, the average concentrations of flesh WMO (2.3–17.7 μg kg⁻¹) were lower than the average concentrations of CCO (20.1–68 μg kg⁻¹) in all samples analysed. Only freshwater euryhaline fish, such as salmon, trout, and Ayu, have been reported to possess CCO and WMO. However, these compounds were not found in river algae (diatoms) or in artificial diets for Ayu culture (Kawai & Sakaguchi 1998). The pleasant odours were most likely originated from three C9-compounds, namely 2E,6Z-nonadienal, 2E-nonenal, and 3,6-nonadienol, which are localized in the skin and viscera of fish (Hirano et al. 1992).

Supplementary Table S8-1 shows the nutritional composition of Ayu flesh tissues. The moisture, proteins, lipids, total fatty acids, ash, carbohydrates, salt equivalent, and total vitamin contents of RWW-cultured Ayu were 70–72 g/100 g, 17–18 g/100 g, 7–10 g/100 g, 6–9 g/100 g, 2–3 g/100 g, 0.1–0.7 g/100 g, 0.2–0.3 g/100 g, and 0.0004–0.0023 g/100 g fresh tissues, respectively. The levels of lipids, total fatty acids, carbohydrates, energy, and total vitamins in the Ayu flesh tissues were higher in autumn than in summer.

In terms of FA composition, RWW-cultured Ayu had a higher percentage of unsaturated FA, such as n-3 and n-6 polyunsaturated FA, than commonly cultured Ayu, while saturated FA and monoenoic unsaturated FA were lower or comparable to commonly wild and cultured Ayu (Supplementary Table S8-2). Some n-3 and n-6 polysaturated FA, such as eicosapentaenoic acid (EPA, 20:5 n-3), docosahexaenoic acid (DHA, 22:6 n-3), and linoleic acid (LA, 18:2 n-6), are essential FA that the human body cannot produce. These fatty acids must be obtained through diet. Surprisingly, the percentages of docosahexaenoic and linoleic acids in RWW-cultured Ayu were considerably higher than those in wild and cultured Ayu.

Supplementary Table S9 shows the concentrations of heavy metals (Cr, As, Pb, Cd, and Hg) in Ayu tissues. The Cr level in RWW-cultured Ayu and purchased Ayu were below the detection limit (≤0.5 mg kg⁻¹). Pb and Cd concentrations in flesh tissues of RWW-cultured Ayu and purchased Ayu samples were similarly undetectable (≤0.05 and ≤0.01 mg kg⁻¹, respectively). The metal concentrations in flesh tissue in the summer trial and intestine tissues in the autumn trial were 0.1–0.2 and 0.03–0.05 mg kg⁻¹, respectively. Meanwhile, As concentrations (0.4–0.8 mg kg⁻¹) in RWW-cultured Ayu tissues were slightly higher than those in purchased Ayu flesh tissues. Total Hg concentrations in RWW-cultured Ayu tissues and purchased Ayu samples ranged from 0.04 to 0.1 mg kg⁻¹ and 0.03 to 0.07 mg kg⁻¹, respectively. Heavy metal levels in Ayu tissues were lower than the Japanese and international standards for fish and foodstuffs (FAO 1983; EC 2008). These results suggest that eating RWW-cultured Ayu poses no risk to human health.

The study demonstrated that the quality of RWW, particularly temperature and ammonia levels, significantly impacts fish growth. Summer RWW negatively affected fish growth due to higher temperatures and ammonia concentrations, while autumn RWW was more favourable. Despite these challenges, heavy metals did not pose health risks to fish. The nutritional analysis of RWW-cultured Ayu indicates higher levels of lipids, total fatty acids, carbohydrates, energy, and total vitamins in autumn compared with summer. RWW-cultured Ayu exhibited lower SFA and higher PFA than wild and commonly cultured Ayu. Notably, essential fatty acids such as EPA, DHA, and LA were significantly higher in RWW-cultured Ayu. Significantly higher VOC levels in autumn RWW compared with summer RWW, with TCA showing the most pronounced seasonal variation. The concentrations of these VOCs in RWW-cultured Ayu were substantially higher than in commercial Ayu, with TCA exhibiting the highest ROI. Despite the presence of desirable odours like CCO and WMO, the study highlights the impact of earthy/musty VOCs, particularly TCA, on the olfactory quality of fish cultured in RWW. Ayu is known in Japan as a symbol of a clear river. The successful culture of Ayu using secondary effluent not only demonstrates the potential of reusing resources from WWTP for energy conservation but also contributes to the expansion of a positive image of wastewater reclamation in Japan. The findings underscore the necessity for effective treatment processes to mitigate VOCs in wastewater to improve the sensory quality of aquaculture products.

We are grateful for the collaboration and support from the Bureau of Sewerage Tokyo Metropolitan Government, Tokyo Metropolitan Sewerage Service Corporation and Meidensha Corporation, Dr Youhei Nomura from the Department of Environmental Engineering, Kyoto University, the professors and staff in the Department of Urban Engineering as well as The University of Tokyo who contributed to the successful completion of this study.

V-D.P.: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Visualization; Writing – original draft and editing. H.K.: Conceptualization; Funding acquisition; Project administration; Resources; Writing – review and editing. E.O.: Data curation; Formal analysis; Investigation; Methodology.

This work was supported by the Tokyo Metropolitan Sewerage Service Corporation and Meidensha Corporation.

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

The authors declare there is no conflict.